Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 Nonlinear Seismic Analysis of U Thick-Walled Concrete Canyon Structure 135 Bob V, Winkel, Gary R

Natural CONF-8910192 - _ Phenomena '* C> if..' Hazards Mitigation Conference Proceedings October 3-5, 1989 Holiday Inn, World's Fair Knoxville, Tennessee

HIHHIIHUHi Organized by mmwm• • Lawrence Livermore iuiiiLkiiiil National Laboratory •iimni

Sponsored by U.S. Department of Energy Headquarters Office of Nuclear Safety MASTER fUSTWflUTrOH OF THIS nOCUUTN: IS dBr'W Organizing Committee James R. Hill U.S. Department of Energy Office of Safely Appraisals Washington, DC Robert C. Murray Conference Chairman Lawrence Livcrmorc National Laboratory Livcrmore, CA Karen L. Anderson Conference Administrator Lawrence Livermore National Laboratory Livcrmore, CA Lilian S. Deem an Proceedings Administrator Lawrence Livermorc National Laboratory Livermorc, CA Lynn M. Costa Proceedings Layout Coordinator Lawrence Livcrmore National Laboratory Livcrmore, CA Janet Crampton-Pipcs Layout Support Lawrence Livcrmorc National Laboratory Livcrmore, CA L. Carole Austin Photographer Impell Corporation Mission Viejo, CA

Technical Committee Robert C. Murray Lawrence Livermore National Laborr.tory Livermore, CA James R. Hill U.S. Department of Energy Office of Safety Appraisals Washington, DC Jean B. Savy Lawrence Livermore National Laboratory Livcrmore, CA Stephen A. Short Impell Corporation Mission Vicjo, CA James R. McDonald Institute of Disaster Research Texas Tech University Lubbock, TX Frank E. McClurc Lawrence Berkeley Laboratory Berkeley, CA CONP-8910192— DE90 006928 Table of Contents

Foreword v Conference at a Glance vi Session 1: DOE Natural Phenomena Guidelines 1 Earthquake Design and Evaluation 2 Stephen A. Short, Robert C. Murray, Roben P. Kennedy Wind and Tornado Guidelines 12 James R. McDonald Flood Design and Evaluation 18 Martin W. McCann, Jean B. Savy

Session 2: Seismic Analysis 27 Seismic and Cask Drop Excitation Evaluation of the Tower Shielding Reactor 28 Steven P. Harris, Roger L. Stover, James J. Johnson, Basiiio N. Sumodobila Seismic Evaluation of Safety Systems at the Savannah River Reactors 35 Gregory S. Hardy, James J. Johnson, Stephen J. Eder, Thomas Monahon, Darrel Ketcham

Session 3: Seismic Design 45 Base Isolation for Nuclear Power and Nuclear Material Facilities - Report On the Status of the Practice , 46 John M, Fidinger, F. F. Tajirian, C. A, Kirctier, N. Vaidya, M. Constantinou, J, M. Kelly, D. Ovadia, R. Seidensiicker Cost-Benefit Assessment of the Seismic Design of the TUFF Repository Waste Handling Facilities 58 Chiller V. Subramanian Seismic Procurement Requirements at the FPR Facility at INEL 66 Greg S. Hardy, Michael J. Griffin, Gait E. Bingham Seismic Qualification of Safety Class Components in Existing Non-Reactor Facilities at Hanford Site 73 Ernesto C. Ocor.ia

Session 4: Modification of Existing Facilities 79 Wind/Seismic Comparison for Upgrading Existing Structures 80 Richard A. Ciller Poslearthquake Safety Evaluation of Buildings at DOE Facilities 89 Ronald Gallagher Study to Evaluate the Feasibility of Constructing a Retrofit Containment for the 105-L Reactor at the Savannah River Plant 99 Robert D. Quinn Practical Approaches to Implementing Facility-Wide Equipment Strengthening Programs 109 Raymond II. Kincaid, Elwood A. Smielana

Session 5: Poster Session 119 Comparison of Proposed Seismic Design Criteria for the INEL 120 Hans J. Dahlke, Robert J. Secondo Consistent Application of Codes and Standards 121 Mark A. Scott Design of Buried Concrete Encasements 122 Richard M. Drake Failure Probability Estimate of Type 304 Stainless Steel Piping 129 William L. Dougherty, Nabil G. Awadalla, Robert L. Sindelar, llardayal S. Mehta, Santath Ranganaih

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 Nonlinear Seismic Analysis of u Thick-Walled Concrete Canyon Structure 135 Bob V, Winkel, Gary R. WagenbUist Seismic Analysis Procedures for the Plutonium Processing Building of the Special Isotope Separation Plant 142 Carl P. Chen, Fredrick F. Tajirian, Ricardo A. A. Todeschini, ItansJ. Datilke Natural Hazard Losses: A DOE Perspective 151 James R. Hill Seismic Analysis and Testing of Clay Tile Walls at the Oak Ridge Y-12 Plant 162 Kenneth E. Fricke, W. Dale Jones Seismic, High Wind, Tornado, and Probabilistic Risk Assessments of the High Flux Isotope Reactor 172 Steven P. Harris, Roger L. Stover, Philip S. Hashimoto, John O. Dizon Variability of Relative Seismic Site Response at Los Alamos, NM 179 Leigh House. W. Scott Phillips

Session 6: DOE Orders/Codes/Standards 185 Application of Project Design Peer Review to Improve Quality Assurance 186 Frank E. McClure Equivalent Static Seismic Analysis Approach for Process Equipment in Moderate and High Hazard Facilities 192 C. Richard Hammond Seismic Design Criteria at (he Idaho National Engineering Laboratory 199 Brent G, Harris

Session 7: DOE Orders/Codes/Standards 205 Comparison of Evaluation Guidelines for Life-Safety Seismic Hazards 206 Loring A. Wyllie, Richard Jay Love Probability-Based Justification for Reduction of Seismic Demand for Short Durations 214 Masoud Moghtaderi-Zadeh, Sohrab Esfandiari Seismic Design Criteria for Special Isotope Separation Plant Structures 220 Matthew W. Wrona, Steven J. Wulhrich. David L. Rose, Jedd G. Starkey Simplified Seismic Analysis Applied to Structures Systems and Components With Limited Radioactive Inventories 226 John D. Stevenson

Session 8: Seismic Hazard 237 Probabilistic Assessment of the Seismic Hazard for Eastern United Slates Nuclear Power Plants 238 Jean B. Savy, Don Bernreuter, Richard Mensing Keeping Pace With the Science: Seismic Hazard Analysis in the Central and Eastern United Slates 252 Kevin J. Coppersmith, Robert R. Youngs Keeping Pace With the Science: Seismic Hazard Analysis in the Western United Stales 262 Robert R. Youngs, Kevin J. Coppersmith

Session 9: Seismic Hazard 271 Earthquake Recurrence Rate Estimates for Eastern Washington and the Hanford Site 272 Alan C. Rohay Geologic Aspects of Seismic Hazards Assessment at the Idaho National Engineering Laboratory, Southeastern Idaho 282 Richard P. Smith, William R. Hacked, David W. Rodger.? Variations of Earthquake Ground Motions with Depth and Its Effect on Soil-Structure Interaction 290 Ching Y. Chang, Wen S. Tseng, Y.K. Tang, Maurice S. Power Natural Phenomena Analyses, Hanford Site, Washington 299 Ann M. Tollman

Second DOE Nat viral Phenomena Hazards Mitigation Conference - 1989 Session 10,' Po*stiT Session A Portable Backup Power Supply to Assure Extended Decay Heal Removal During Natural Phenomena -Induced Station Blackout .'. 310 Lurry D. Proctor, Larry D. Merryman, William /:. Sallce Earthquake Strong Ground Motion Studies ai the Idaho National Engineering Laboratory 317 iva/fG, Wong, Waller J. Silva, Robert B. Darragh, Cathy L. Stark, Douglas II. Writ-lit, Suzetle M. Jackson, Glen S. Carpenter, Richard Smith, Dennis M. Anderson, llollie K. Gilbert, Don L. Scott Probabilistic Seismic Hazards: Guidelines and Constraints in Evaluating Results 329 Ross K. Sadigh. Maurice S. Power Site-Specific Response Spectral Attenuation Relationships: Significance on Ground Motion Predictions 330 Maurice S. Power, Khosrow Sadigh Structural Evaluation of Safely Class Components to Natural Phenomena Loading 336 Thomas J. Conrads Transmission of Low-Magnitude Seismic Excitation Into Hanford Site Structures 337 Evan O. Weiner Advances in Seismic Criteria to Qualify Structures, Systems and Components in Operating Reactors 343 Miguel A. Manrique, Walter R. Bak

Session 11: Probabilistic Risk Assessment „ 351 Overview of Seismic Probabilistic Risk Assessment for Structural Analysis in Nuclear Facilities 352 John W. Reed Probabilistic Evaluation of Main Coolant Pipe Break Indirectly Induced by Earthquakes Savannah River Project L & P Reactors 365 Stephen A. Short, Donald A. Wesley, Nahil G. Awadalla, Robert P. Kennedy Large-Break Frequency for the SRS Production Reactor Process Water System 375 William L. Dougherty, Nabil G. Awadalla, Robert L. Sindelar, Spencer II. Bush

Session 12: Probabilistic Risk Assessment 381 External Event Probabilistic Risk Assessment for the High Flux Isotope Reactor (HFIR) 382 George F. Flanagan, David II. Johnson, David R. Buttemer, Harold R. Perla, Shan II. Chien N Reactor External Events PRA 3% John T. Baxter Use of a Phased Approach to Evaluate the Seismic Probabilistic Risk Assessment for Production Reactors at ihc Savannah River Site 405 //. Elwyn Wingo Probabilistic Risk Assessment of Earthquakes at the Rocky Flats Plant and Subsequent Upgrade to Reduce Risk 414 Sandra A. Day

Invited Speakers 419 The Performance of the Armenia Nuclear Power Plant and Power Facilities in the 1988 Armenia Earthquake 420 Peter I. Yanev International Decade for Natural Disaster Reduction 427 Walter W. Hays

Natural Phenomena Hazards Bibliography 430 List of Attendees 431 Index of Authors 436

Sconcl DOE Natural Phenomena Hazards Mitigation Conference - 1989

iii Foreword

I would like to take this opportunity to thank you for attending the Second U.S. Department of Energy Natural Phenomena Hazards Mitigation Conference in Knoxvillc, Tennessee. I hope this meeting was productive for you in sharing ideas, hearing about ongoing programs, and taking results of previous studies back to yfHir office for further dissemination.

This conference has been organized into ten presentation sessions which include an overview of the DOE Natural Phenomena Guidelines, Seismic Analysis, Seismic Design, Modifying Existing Facilities, DOE Orders, Codes, and Standards (2 sessions), Seismic Hazard (2 sessions), and Probabilistic Risk Assessment (2 sessions). Two poster sessions were also included in the program to provide a different forum for communication of ideas.

Over the past fourteen years, Lawrence Livcrmorc National Laboratory, Nuclear Systems Safety Program, has been working with the U. S. Department of Energy, Office of Safety Appraisals and their predecessors in the area of natural phenomena hazards. During this time we have developed seismic, extreme wind/iomado, and flood hazard models for DOE sites in the United Slates. Guidelines for designing and evaluating DOE facilities for natural phenomena have been developed and arc in interim use throughout the DOE community. A scries of state-of-ihe practice manuals have also been developed to aid the designers. All of this material is listed in the Natural Phenomena Hazards Bibliography included in these proceedings. This conference provides a mechanism to disseminate current information on natural phenomena hazards and their mitigation. It provides an opportunity to bring together members of the DOE community to discuss current projects, to share information, and to hear practicing members of the structural engineering community discuss their experiences from past natural phenomena, future trends, and any changes to building codes. Each paper or poster presented is included in these proceedings. We have also included material related lo the luncheon and dinner talks.

Finally, I would like to thank the conference participants, sessions chairmen, organizing committee, conference staff, and paper and poster presenters for their efforts in making the conference a success.

Robert C. Murray Conference Chairman Lawrence Livcrmorc National Laboratory

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Refer- ence herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise docs not necessarily constitute or imply its endorsement, recommendation, or favoring by the United Slates Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 iv Conference at a Glance

Tuesday, October 3,1989 8:30-10:00 A.M. DOE Natural Phenomena Guidelines 10:00-10:30 A.M. Break 10:30-12:00 Noon Seismic Analysis 12:00-1:30 P.M. Lunch 1:30-3:00 P.M. Seismic Design 3:00-3:30 P.M. Break 3:30 - 5:00 P.M. Modification of Existing Facilities 5:00-6:00 P.M. Poster Session 6:30-8:00 P.M. Reception

Wednesday, October 4,1989 8:30 - 10:00 A.M. DOE Ordcrs/Codes/Standards

10:00-10:30 A.M. Break 10:30-12:00 Noon DOE Orders/Codes/Standards 12:00-1:30 P.M. Lunch 1:30 -3:00 P.M. Seismic Hazard 3:00 -3:30 P.M. Break 3:30-5:00 P.M. Seismic Hazard 5:00 6:00 P.M. Poster Session 7:30-9:30 P.M. Dinner

Thursday, October 5,1989 8:30 - 10:00 A.M. Probabilistic Risk Assessment 10:00-10:30 A.M. Break 10:30 - 12:00 Noon Probabilistic Risk Assessment 12:00 Noon Adjourn

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 Second DOIi Natural Phrinurirna Hazards Mitigation 198*) Scronrl IXJE Natural Phenomena Hn/anls Mitigation Confcn^nrr - ]\)H9 Jim Hill, Chairman of the Conference Organizing Committee, and Jim knight, Director of the Office of Safety Appraisals, welcome participants to the Second DOE Natural Phenomena Hazards Mitigation Conference.

Second DOE Natural Phenomena Hazards Mitigations Conference - 1989

VIII Session! DOE Natural Phehoiiiena Guidelines

Second DOE Natural Phenomena Hazards Mitigations Conference- 1989 DOE NATURAL PHENOMENA GUIDELINES EARTHQUAKE DESIGN AND EVALUATION

Stephen A. Short Impell Corporation Mission Vlejo, California Robert C. Murray Lawrence Livermore National Laboratory Livermore, California Robert P. Kennedy RPK Structural Mechanics Consulting Yorba Unda, California

ABSTRACT Design and evaluation guidelines for DOE facilities subjected to earthquake, wind/tornado, and flood have been developed. This paper describes the philosophy and procedures for the design or evaluation of facilities for earthquake ground shaking. The guidelines are intended to meet probabilistic-based performance goals expressed in terms of annual probability of exceedance of some level of structural damage. Meeting performance goals can be accomplished by specifying hazard probabilities of exceedance along with seismic behavior evaluation procedures in which the level of conservatism introduced is controlled such that desired performance can be achieved. Limited inelastic behavior is permitted by permitting demand determined from elastic response spectrum analyses to exceed capacity by an allowable inelastic demand-capacity ratio specified in the guidelines for different materials and construction.

INTRODUCTION they include deterministic procedures even The design and evaluation guidelines for though performance goals and hazard descrip- Department of Energy facilities subjected to tions are probabilistically based. natural phenomena hazards (Ref. 1) are intended The guidelines have been developed as part to assure adequate facility designs for the effects of an overall natural phenomena hazard project of earthquake, extreme wind or tornado, and as illustrated in Figure 1. Phase 1 of this program flood. These guidelines consist of: 1) facility-use identified critical facilities throughout the DOE categories each with probabilistic-based per- complex. This phase identified 25 sites across formance goals; 2) specified hazard probability the U.S. where there exists a variety of structures to develop loads; and 3) deterministic ranging from ordinary buildings such as offices procedures for response evaluation and per- or warehouses to facilities containing significant missible behavior criteria. The guidelines apply amounts of hazardous materials. In Phase 2, to all natural phenomena hazards and all DOE hazard curves were developed at each DOE site facilities across the United States in a consistent relating annual probability of exceedance to manner. In addition, the guidelines represent a hazard parameters such as earthquake ground simplified approach to facility design and evalu- acceleration, wind speed, or flood water depth. ation as they are based on existing codes and Phase 3 consists of the development of the

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 2 guidelines. Phase 4 includes supporting studies Performance goals are probabilistic-based on specific natural phenomena hazard problem and expressed in terms of annual probability of areas. exceeding a level of structural damage. Per- t Performance goals have been assigned formance goals are given for four facility use based on facility characteristics such as: ^vul- categories including: 1) General Use; 2) nerability of occupants; 2) cost of replacement of Important or Low Hazard; 3) Moderate Hazard; facility and contents; 3) mission dependence; 4) and 4) High Hazard. Facility use categories and hazardous materials contained; and 5) potential the corresponding performance goals are shown for off-site release. For ordinary facilities, design in Table 1. which is consistent with conventional building codes such as the Uniform Building Code (Ref. Table 1 2) is acceptable. For facilities containing haz- Performance Goals For Each ardous materials, there should be very low Facility-Use Category probability of damage due to natural phenomena Facility Use Performance Goal Annual hazards. Category Probability of Exceedance General 10"3 of the onset of major structural damage

PHASE 1 Use to the extent that occupants are endangered DEFINE EXISTING CRITICAL FACILITIES AT EACH SITE WITH RCIP Or SITE PERSONNEL Important 5x10*4 of facility damage to the extent that or Low the facility cannot perform its function Hazard Moderate 10'4 of facility damage to the extent that the CRITICAL FACILITY (PROPOSED OR EXISTING) Hazard facility cannot perform its function High 10'5 of facility damage to the extent that the Hazard facility cannot perform its function PHASE I DEVELOP HAZARD MODELS FOR KACH SITE General Use and Important or Low Hazard categories correspond to facilities whose design or evaluation would normally be governed by conventional building codes. The General Use category includes normal use facilities for which PHASES 3 AND 4 no extra conservatism against natural phenom- EVALUATE KACH PACILITY ON A UNIFORM AND RATIONAL BASIS ena hazards is required beyond that in conven- SPECirY RESPONSE EVALUATION PROCEDURES tional building codes that include earthquake, wind, and flood considerations. Important or Low Hazard facilities are those where it is very DETERMINE APPROPRIATE PERMISSIBLE RESPONSE HAZARD LEVEL (PHASE 3) CRITERIA (PHASE 3) important to maintain the capacity to function and

GOOD DESIGN DETAOINC AND PRACTICES (PHASE 4) to keep the facility operational in the event of natural phenomena hazards. Conventional Figure 1 building codes would treat hospitals, fire and Natural Phenomena Hazards Program police stations, and other emergency handling facilities in a similar manner to the requirements Performance goal is the combined function of of these guidelines for Important or Low Hazard the likelihood of hazard occurrence strong facilities. enough to produce damage and the strength of Moderate and High Hazard categories apply facility to resist natural phenomena hazards. to facilities which deal with significant amounts of Therefore, performance goals may be achieved hazardous materials. Damage to these types of by specification of hazard probability along with facilities could potentially endanger worker and evaluation procedures with controlled level of public safety and the environment. As a result, it conservatism. Hazard probability and conser- is very important for these facilities to continue to vatism in the evaluation process may be different function in the event of natural phenomena haz- for earthquake, wind, and flood. However, the ards, such that the hazardous materials may be combined effects should lead to achievement of controlled and confined. For both of these the same performance goal for all hazards. categories, there must be a very small likelihood

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 of damage due to natural phenomena hazards. the extent that significant amounts of hazardous Guideline requirements for Moderate Hazard materials cannot be controlled and confined, facilities are more conservative than require- occupants are endangered, and functioning of ments found in conventional building codes. the facility is interrupted. The performance goal Requirements for High Hazard facilities are even for Moderate Hazard facilities is to limit damage more conservative. such that confinement of hazardous materials is For Genera/ Use facilities, the primary concern maintained. The performance goal for High is preventing major structural damage or facility Hazard facilities is to provide very high confi- collapse that would endanger personnel within dence that hazardous materials are confined the facility. A performance goal annual proba- during and following a natural phenomena bility of exceedance of about 10"3 of the onset of hazard occurrence. Maintaining confinement of significant facility damage is appropriate for this hazardous materials requires that damage be category. This performance is considered to be limited in confinement barriers. Structural consistent with conventional building codes, at members and components should not be dam- least for earthquake and wind considerations. aged to the extent that breach of the confinement The primary concern of conventional building or containment envelope is significant. codes is preventing major structural failure and Furthermore, ventilation filtering and containers maintaining life safety under major or severe of hazardous materials within the facility should earthquakes or winds. This primary concern for not be damaged to the extent that they are not preventing structural failure does not consider functional. In addition, confinement may depend repair or replacement of the facility or the ability on maintaining safety-related functions, so that of the facility to continue to function after the monitoring and control equipment should remain occurrence of the hazard. operational following, and possibly during, the Important or Low Hazard Use facilities are of occurrence of severe earthquakes, winds, or greater importance due to mission-dependent floods. considerations. In addition, these facilities may For High Hazard facilities, a performance goal pose a greater danger to on-site personnel than of an annual probability of exceedance of about general use facilities because of operations or 10-5 of damage, to the extent that confinement materials within the facility. The performance functions are impaired, is judged to be reason- goal is to maintain both capacity to function and able. This performance goal approaches, at least occupant safety. Important or Low Hazard faci- for earthquake considerations, the performance lities should be allowed relatively minor structural goal for seismic-induced core damage asso- damage in the event of natural phenomena ciated with design of commercial nuclear power hazards. This is damage that results in minimal plants. For Moderate Hazard facilities, a interruption to facility operations and that can be performance goal of an annual probability of easily and readily repaired following the event. A exceedance of about 10~4 of damage, to the performance goal annual probability of excee- extent that confinement functions are impaired, dance of between 10*3 and 10"4 of structure or is judged appropriate. equipment damage, to the extent that the capacity of the facility is able to continue to EARTHQUAKE GUIDELINES AND OVERALL function with minimal interruption, is judged to be RISK ASSESSMENT reasonable. This performance goal is believed Seismic design or evaluation of a hazardous to be consistent with the design criteria for facility is only part of the process of assessing essential facilities (e.g., hospitals, fire and police overall risk as shown in Figure 2. The seismic stations, centers for emergency operations) in design or evaluation process encompasses the accordance with conventional building codes. first two steps of risk assessment; characteriza- Moderate or High Hazard Use facilities pose a tion of the earthquake hazard and structural potential hazard to the safety of the general public evaluation. The overall risk is a combined and of the environment due to the presence of function of the performance goal to which the radioactive or toxic materials within these facili- facility has been designed and the factors ties. Concerns about natural phenomena hazards for these categories are facility damage to

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 4 associated with the source term and dose cal- response and in the specification of acceptable culation which affect whether hazardous mate- seismic response, in addition to the specified rials might reach the public or the environment in design/evaluation shaking level. Potential the event of significant seismic structural sources of conservatism in the seismic design or damage. Performance goals are expressed in evaluation process can be broken down into the terms of annual probability of exceedance of following three categories. structural or equipment damage (to the extent that hazardous materials are not confined). 1. Specification of the seismic input. These goals facilitate the seismic design or 2. Criteria for performing the seismic evaluation process by separating the assess- response analysis. ment of risk into two parts: (1) earthquake performance of buildings and equipment and (2) 3. Specifications of material strengths and dissemination of hazardous materials. estimations of structural capacities.

Sufficient conservatism must be combined into CHARACTERIZATION OF HAZARD these three categories in order to achieve a desired performance goal, which can be E<1 ISO Flood expressed in terms of an annual probability of exceedance of a desired performance level (such STRUCTURAL EvAl UATlON as continued operation with minimal interrup- rling and ["quipm^nt Kt's tion). Such conservatism may be introduced in ISO many different ways. One way to achieve a desired performance goal is to place all of the conservatism into the specification of the seismic input and use median response analyses and median structural DOSE. CALCUl ATiON capacity estimates (median approaches intro- Meteorology duce no bias, conservative or unconservative). Dernography Thus, if the performance goal is to achieve less i co'ogy than a 10"* annual probability of damage to confinement of hazardous materials, one way to ESTIMATE achieve this goal would be to specify for design or evaluation a seismic input with less than 10"5 annual probability of being exceeded, and then specifying median-centered response analyses •'' C I" and capacity evaluations. However, this Figure 2 approach has two primary deficiencies. Flow Diagram for Assessment of Risk from Natural Phenomena Hazards 1. The available seismic record in the United States is less than 400 years long. ACHIEVEMENT OF PERFORMANCE GOALS Extrapolating this seismic record to annual IN SEISMIC DESIGN OR EVALUATION probabilities of exceedance of less than The performance goal for seismic design or about 2 x 10*3 requires considerable trad- evaluation is a combined function of the likeli- ing of spatial for temporal considerations. hood of earthquake occurrence of sufficient size Generally, the existing seismic record and to produce damage and the strength of the facility seismological and geophysical consider- to withstand earthquake shaking. The likelihood ations permit a reasonable extrapolation of of earthquakes is expressed by a seismic hazard ground motion levels down to annual curve which relates annual probability of excee- probabilities of about 2 x 10"4. Ground dance to maximum ground acceleration. The motions associated with annual probabili- strength of a facility to withstand earthquake ties less than about 2 x 10"4 become hiyhly shaking depends on sources of conservatism controversial, and they have seldom been introduced in the evaluation of earthquake used for seismic design or evaluation.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 2. Existing codes and standards are geared corresponding to this base shear level are to specifying material strengths and struc- compared with code specified ultimate or tural capacities very conservatively. Even allowable levels. ultimate strength codes are intended to Moderate And High Hazard Facilities achieve less than a 2% probability of failure The guidelines for Moderate and High Hazard if the code capacities are reached. Median facilities generally follow the DOD Manual, strength capacity estimates are controver- "Seismic Design Guidelines for Essential Build- sial and can seldom be used in a regulatory ings" (Ref. 3). For these facilities, the guidelines environment. are to perform dynamic analysis of structures to obtain seismic response. These analyses utilize By the seismic design and evaluation guidelines median ground response specira, post-yield for DOE facilities, a conservative evaluation damping, and unity load & importance factors. approach is employed along with specified The resulting seismic response is compared to hazard exceedance probabilities which are yield or ultimate level capacity. The guidelines greater than the performance goal exceedance approach is to perform an elastic response probabilities. spectrum analysis but to permit limited inelastic behavior by permitting response to exceed yield EARTHQUAKE GUIDELINES level by the permissible inelastic demand- General Use & Important Or Low Hazard capacity ratio, Fu Faciiities Hazard Exceedance Probabilities The guidelines for General Use and Important Historically, non-Federal Government Gen- or Low Hazard facilities are strongly based on the eral Use and Essential or Low Hazard facilities seismic provisions of the 1988 Uniform Building located in California, Nevada, and Washington Code (Ref. 2). For these facilities, the guidelines have been designed for the seismic hazard follow UBC provisions except ground motion is defined in the Uniform Building Code. Other determined from site-specific data instead of the regions of the U.S. have generally used either code seismic zone map or the general site some version of the UBC seismic hazard defini- amplification factor from the code. By the tion or else have ignored seismic design. Past guidelines, seismic response is determined from UBC seismic provisions (1985 and earlier) are equivalent static analysis for regular, uniform based upon the largest earthquake intensity structures or from dynamic analysis for irregular which has occurred in a given region during the structures. In either case, the base shear is given past couple of hundred years. These provisions by the following code equation: do net consider the probability of occurrence of such an earthquake and thus do not make any explicit use of a probabilistic seismic hazard V analysis. However, within the lastten years, many codes including ATC-3, the Canadian Building where: Z = a seismic zone factor equivalent to Code, and the 1988 Uniform Building Code have peak ground acceleration used probabilistic-based seismic zone maps I = a factor accounting for the which have consistent uniform annual probability importance of the facility of exceedance for all regions of the U.S. The C = a spectral amplification factor suggested annual frequency of exceedance for W = the total weight of the facility the design seismic hazard level differs somewhat Rw= a reduction factor to account for between these codes, but all lie in the range of energy absorption capability 10-2 to 10"3. The 1988 UBC seismic zone factors corresponds to the design seismic hazard level All code provisions are included in the guidelines having about a 10 percent frequency of excee- except that the Z factor is determined from dance level in 50 years which corresponds to an 3 site-specific seismic hazard curves and the C annual exceedance frequency of about 2x10" . factor is determined from the 5 percent damped median (or mean) site-specific ground response spectra. Response determined from analyses

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

6 On the other hand, nuclear power plants are safe shutdown earthquake (SSE) for nuclear designed so that safety systems do not fail if power plants, The influence of the specified subjected to a safe shutdown earthquake (SSE). hazard exceedance probabilities on peak ground The SSE generally represents the expected acceleration is illustrated in Figure 3. Although it ground motion at the site either from the largest depends on the shape of individual seismic historic earthquake within the tectonic province hazard curves, on the average, the hazard within which the site is located or from an probability for Important or Low Hazard facilities assessment of the maximum earthquake poten- results in ground motion about 25 percent higher tial of the appropriate tectonic structure or than that for General Use facilities and the hazard capable fault closest to the site. The key point is probability for High Hazard facilities results in that this is a deterministic definition of the design ground motion about twice that for General Use SSE. Recent probabilistic hazard studies have facilities. indicated that for nuclear plants in the eastern U.S., the design SSE level generally corresponds to an estimated mean annual frequency of exceedance of between 10*3 and 10"4. Also, during the last ten years, considerable interest Seismic Hazard Cu has developed in attempting to estimate the

Exceeda n 2x1 seismic risk of these nuclear power plants in - I terms of annual probability of seismic-induced "o 1x1 024>- • core melt or risk of early fatalities and latent 2x1 j | - 7 cancer to the public. Many studies have been °" i - I I I . 1 conducted on seismic risk of individual nuclear r-abi l I power plants. Because those plants are very o conservatively designed to withstand the SSE, fL. i !.. f. V... • these studies have indicated that the seismic risk 08.10 .17 is acceptably low (generally less than about 10*5 annual probability of seismic-induced core Acceleration (g) damage) when such plants are designed for SSE Figure 3 levels with a mean annual frequency of excee- Influence of Hazard Probabilities dance between 10"3 and 10"4. With this comparative basis for other facilities, Conservatism in Seismic Evaluation the recommended hazard exceedance proba- By the seismic design/evaluation guidelines, bilities (as shown in Table 2) are judged to be the recommended performance goals and haz- consistent and appropriate to define the seismic ard exceedance probabilities are different. This hazard for DOE facilities. difference indicates that conservatism must be introduced in the seismic evaluation approach. Table 2 The differences for the various categories are Hazard Exceedance Probabilities illustrated in Table 3. HAZARD The specified hazard exceedance probabili- FACILITY EXCEEDANCE ties are appropriate only if the seismic design or CATEGORY PROBABILITY evaluation is conservatively performed. The level General Use 2x10-3 UBCZ of conservatism should increase from the Gen- Important or Low Hazard 1x10"3 - eral Use to the High Hazard category. For General Use and Important or Low Hazard Moderate Hazard 1x10"3 4 category facilities, conservatism introduced High Hazard 2x10" SSE should be consistent with that which exists in the Uniform Building Code. For High Hazard cate- As discussed previously, the hazard exceedance gory facilities, conservatism should approach probability for General Use facilities is similar to that used for nuclear power plants. The that corresponding to the UBC Z factor and the guidelines follow the philosophy of: 1) gradual hazard exceedance probability for High Hazard facilities is similar to that corresponding to the

Second DOIC Natural Phenomena Ha/.ards Mitigation Conference - reduction in hazard annual exceedance proba- absorption capacity are used over the entire bility; and 2) gradual increase in conservatism of frequency range and because stiff structures evaluation procedure as one goes from a General may not be as stiff as idealized for dynamic Use to a High Hazard facility. analysis due to soil-structure interaction, concrete cracking, base mat flexibility, and softening Table 3 during inelastic response, the maximum spectral Performance Goals & Recommended acceleration is used for building response eval- Hazard Probabilities uation if the fundamental frequency is larger than Hazard Ratio of Hazard the frequency range at which the maximum Facility Performance Exceedance to Performance spectral acceleration occurs. Category Goal Probability Probability General Table 4 3 3 Use 1x10- 2x10- 2 Potential Sources Important of Conservatism or Low Hazard 5x1 CH 1x10-3 2 Potential Sources Conservatism Possible of Conservatism Intentionally Additional Moderate in Guidelines Conservatism Hazard 1X10-4 1x10-3 10 Maximum ground acceleration High Hazard 1x10-5 2X10-4 20 Response spectra amplification Damping Potential sources of conservatism in the Analysis methods seismic design/evaluation process are listed in Specification of material X Table 4. In order to control the level of conser- strengths vatism such that the performance goals can be Estimation of structural cap^ity X approximately achieved, many aspects of the seismic design/evaluation process are per- Load factors X* formed using median-centered properties and Importance factor* X* methods such that no conservatism is Limits on inelastic behavior X introduced. Conservatism is only intentionally Soil-structure interaction X introduced in the limits on inelastic behavior and Effective peak ground motion X in the use of code-specified material strengths Effects of a large foundation X and capacity equations as shown in Table 4. Effects c* foundation embed- X Intentional conservatism is controlled to ment approximately achieve performance goal. It is recognized that the simple seismic evaluation General Use & Important or Low Hazard only. methods employed by the guidelines may also Damping include some unintentional conservatism in the Damping is another part of the seismic areas shown in Table 4. It is permissible to design/evaluation process where it is intended perform more sophisticated evaluations to that no conservatism is introduced. Damping remove these sources of unintentional conser- values corresponding to response just beyond vatism. the yield level are intended for dynamic analyses Earthquake Ground Response Spectra of Moderate and High Hazard facilities. Because Median (i.e., no conservative bias) spectral there is a great deal of uncertainty in damping at amplification is used by these guidelines for all this response level, conservative estimates of categories. For General Use & Important or Low median post-yield damping values (Table 5) are Hazard facilities, the C factor in the UBC base recommended for usage. shear equation comes from the median spectra Acceptable Analysis Approaches at the fundamental frequency and 5 percent For General Use & Important or Low Hazard damping. For Moderate & High Hazard facilities, facilities, seismic response is evaluated by median spectral amplification at post-yield equivalent static analysis (Figure 4) for regular, damping levels is used. Because constant uniform structures or by dynamic analysis for reduction factors accounting for inelastic energy irregular structures. In either case, the results are

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

8 scaled such that the base shear is given by the limited amount as a means to account for energy code equation including site-specific information absorption in the inelastic range. Based on for peak ground acceleration and spectral observations during past earthquakes and con- amplification. For Moderate and High Hazard siderable dynamic test data, it is known that facilities, seismic response is evaluated by structures can undergo limited inelastic dynamic analysis (Figure 5), using median deformations without unacceptable damage ground response spectra, post-yield damping when subjected to transient earthquake ground values, unity load and importance factors, and motion. inelastic demand-capacity ratios, Fu, to permit limited inelastic behavior. Table 5 Recommended Damping Values DAMPING TYPE OF STRUCTURE (% OF CRITICAL) EQUIPMENT AND PIPING 5 WELDED STEEL & PRESTRESSED CONCRETE 7 BOLTED STEEL & REINFORCEO CONCRETE 10 MASONRY SHEAR WALLS 12 WOOD 15

ZICW V = R Ul Frtqutncy (hz) Figure 5 Dynamic Analysis for Moderate & High Hazard Facilities 7777777777777 l •• V ' The inelastic demand-capacity ratios are Figure 4 evaluated based upon the permissible inelastic Static Analysis for General Use behavior level which depends on the materials & Important or Low Hazard Facilities and type of construction. For more hazardous facilities, lower demand-capacity ratios may be Limits On Inelastic Behavior used to add conservatism to the design or Energy absorption in the inelastic range of evaluation process, such that increased proba- response of structures and equipment to earth- bility of surviving any given earthquake motion quake motions can be very significant. Generally, may be achieved. The guidelines employ the an accurate determination of inelastic behavior inelastic demand-capacity ratio approach with necessitates dynamic nonlinear analyses per- added conservatism for High Hazard facilities. formed on a time-history basis. However, non- Example values of inelastic demand-capacity linear structural response may be approximated ratios, Fu, are shown in Table 6 along with the through the use of inelastic demand-capacity corresponding code Rw values. ratios. If the permissible demand-capacity ratios are not exceeded, it would be concluded that the EARTHQUAKE DESIGN DETAILING & structure was adequate for earthquake loading. QUALITY ASSURANCE PROVISIONS Inelastic demand-capacity ratios permit struc- Both the 1988 UBC provisions and these DOE tural response to exceed yield stress levels a guidelines consist of two parts: 1) lateral force

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

9 provisions used to proportion the structure; and high quality materials and construction which 2) detailing provisions to assure energy provide redundancy and energy absorption absorption capacity and to avoid failure due to capacity) can withstand earthquake motion well local instability or brittle behavior. Each part is in excess of design levels. However, if details equally important. In addition, the DOE guide- which provide redundancy or energy absorbing lines have quality assurance provisions and peer capacity are not provided, there is little real review provisions similar to that recommended in margin of safety built into the facility. It would be the "Blue Book" by the Structural Engineers possible for significant earthquake damage to Association of California for Important or Low occur at ground shaking levels only marginally Hazard, Moderate Hazard, and High Hazard above the design lateral force level. Poor facilities. materials or construction could potentially lead to damage at well below the design lateral force Table 6 level. Furthermore, poor design details, materi- Example Inelastic als, or construction increase the possibility that a Demand-Capacity Ratios, Fu dramatic failure of a facility may occur. STRUCTURAL UBC MODERATE HIGH For important or hazardous facilities, there are SYSTEM/ELEMENT Rw HAZARD HAZARD special quality assurance measures recom- Steal Moment Frame 12 3.0 2.5 mended beyond that which is standard checking and inspection practice of design offices. The Steel Braced Frame Diagonal Braces 8 1.7 1.4 recommended program includes: 1) indepen- Concrete Shear Wall dent review of the design basis and approach; Shear 8 1.7 1.4 2) specification of an appropriate testing and Moment 2.75 2.25 inspection program by the engineer of record to assure that his design is implemented as The guidelines briefly describe general design intended; and 3) construction observation by the considerations which enable structures or engineer of record including review of the testing equipment to perform during an earthquake in and inspection reports of his program and peri- the manner intended by the designer. These odic site visits to observe general compliance design considerations attempt to avoid prema- with the plans and specifications. ture, unexpected failures and encourage ductile behavior during earthquakes. As discussed SUMMARY above, characteristics of the lateral force- The seismic design and evaluation guidelines resisting systems are as important or more so for DOE facilities are summarized in Table 7. The than the earthquake load level used for design goals of the lateral force, detailing and quality or evaluation. These characteristics include assurance provisions are: redundancy; ductility; tying elements together to behave as a unit; adequate equipment anchor- 1. Provide sufficient strength and stiffness to age; understanding behavior of non-uniform; resist potential lateral loads non-symmetrical structures or equipment; detailing of connections and reinforced concrete 2. Avoid failures due to brittle behavior or elements; and the quality of design, materials, instability and construction. The level of earthquake 3. Understand behavior such that some ground shaking to be experienced by any facility "weak link" is not overlooked. in the future is highly uncertain. As a result, it is important for facilities to be tough enough to 4. Assure high quality materials and con- withstand ground motion in excess of their struction in the manner intended by the design ground motion level. There can be high designer confidence in the earthquake safety of facilities designed in this manner. Earthquakes produce transient, limited energy loading on facilities. Because of these earthquake characteristics, well-designed and constructed facilities (i.e., those with good earthquake design details and

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 10 Status Of Guidelines REFERENCES The guidelines document is available in draft 1. Kennedy, R.P., Et Al, "Design and Evaluation form from DOE headquarters. The guidelines are Guidelines for Department of Energy Facili- currently being reviewed by DOE headquarters. ties Subjected to Natural Phenomena Haz- However, the guidelines are being used in the ards," (draft) UCRL-15910, Lawrence DOE community. They are required for design Livermore National Laboratory, Livermore, of new facilities by DOE General Design Criteria, CA, May 1989. Order 6430.1A (Ref. 4). 2. Uniform Building Code, 1988 Edition, Inter- Training sessions or workshops on the use of national Conference of Building Officials, these guidelines are being conducted. Thus far, Whittier, California, 1988. workshops attended by about 50 participants each have been held in Albuquerque, New 3. Seismic Design Guidelines for Essential Mexico in May 1989 and in Seattle, Washington Buildings (A Supplement to "Seismic in September 1989. Additional workshops will be Design for Buildings"), Army TM5-809- conducted in Charleston, South Carolina in 10.1, Navy NAVFAC P-355.1, Air Force AFM February 1990 and Chicago, Illinois in May or 88-3, Chapter 13.1, Departments of the June 1990. Army, Navy and Air Force, Washington, D.C., February, 1986. Table 7 4. U.S. Department of Energy, General Summary Of Design Criteria, DOE Order 6430.1 A, Earthquake Guidelines Washington, D.C., 1987 (draft). FACILITY-USE CATEGORY General Important Moderate High Use or Low Hazard Hazard Hazard HAZARD EXCEEDANCE 2x10-3 1x10-3 1x10-3 2X10-4 PROBABILITY RESPONSE Mbdian (or Mean) Amplification SPECTRA DAMPING 5% Post-Yieid ACCEPTABLE Static or Dynamic ANALYSIS Analysis Normalized to Dynamic Analysis* APPROACHES Code Level Base Shear IMPORTANCE 1=1.0 1=1.25 Not Used* FACTOR LOAD Code Specified Load Load Factors FACTORS Factors Appropriate for of Unity Structural Material

INELASTIC Accounted for by Ryy Fu from Guidelines DEMAND in UBC Base Shear Applied to Dead Load CAPACITY Equation Plus Live Load RATIOS Plus Earthquake MATERIAL Minimum Specified or Known In-situ Values STRENGTH STRUCTURAL Code Ultimate or Y'-ild Level CAPACITY Allowable Level PEER REVIEW. QA, SPECIAL — Required INSPECTION

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 WIND AND TORNADO GUIDELINES

James R. McDonald Texas Tech University Civil Engineering Department P.O. Box 4089 Lubbock, TX 79409-1023

ABSTRACT

The objective of the Department of Energy Natural Phenomena Hazards Project is to provide guidance and criteria for design of new facilities and for evaluation of existing ones subjected to extreme winds, earthquakes, and floods. This paper describes the treatment of wind and tornado hazards. Four facility-use categories are defined which represent increasing levels of risk to personnel or the environment in the event of a high wind event. Facilities are assigned to a particular category, depending on their mission, value, or toxic material content. The assigned facility-use category determines the design and evaluation criteria. The criteria are based on probabilistic hazard assessment. Performance goals are also specified for each facility-use category. A uniform approach to design wind loads, based on the ANSI A58.1-1982 standard, allows treatment of high winds and hurricane and tornado winds in a similar manner. Based on the wind hazard models, some sites must account for the possibility of tornadoes while others do not. Atmospheric pressure changes and missiles must be taken into account when considering tornadoes. The design and evaluation guidelines are designed to establish consistent levels of risk for different natural phenomena hazards and for facilities at different geographical locations.

INTRODUCTION design and evaluation guidelines, and 4) The objective of the Department of produce manuals that describe detailing Energy Natural Phenomena Hazards Project practices that lead to desirable is to provide guidance and criteria for structural behavior. The purpose of this design of new facilities and for paper is to describe briefly the evaluation, modification, or upgrade of procedures in and the philosophy behind existing facilities such that DOE the wind and tornado guidelines. facilities are adequately constructed to The design and evaluation guidelines safely withstand the effects of [1] define facility-use categories, earthquakes, extreme winds, and flooding. facility performance goals, and methods The program has progressed in four for evaluating structural response. phases: 1) select DOE sites to be Hazard probability levels are established included in the project, 2) develop from which loadings on buildings or hazard probability models, 3) prepare facilities can be determined. Procedures Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

12 are recommended for evaluating facility amounts of hazardous materials. It is response to wind hazard loads and important for these facilities to criteria are specified for assessing continue to function in the event of whether the computed structural response natural phenomena hazards, such that the is acceptable. hazardous materials can be controlled and confined. FACILITY-USE CATEGORIES Facility-use categories are suggested PERFORMANCE GOALS for design/evaluation of DOE facilities. Table 2 presents performance goals for It is the responsibility of DOE each facility-use category. The management to assign facilities to these performance goal for General Use categories and select acceptable facilities is to prevent collapse which performance goals. The four facility-use would endanger persons within the categories are briefly described in Table facility. Repair or replacement costs or 1. General Use and Important or Low the ability of the facility to continue Hazard categories correspond to to function after a wind event that facilities whose design or evaluation significantly exceeds design values is would normally be governed by not considered critical. Performance is conventional building codes. Moderate consistent with the goals of conventional and High Hazard categories apply to building codes for ordinary buildings. facilities which deal with significant Important or Low Hazard Facilities are of

TABLE 1. FACILITY-USE CATEGORIES

Facility-Use Category Description Qanarat U»a Facilitiee which have a non-mlaaion dependent purpoae, euch aa administration Fecilrtlee buildings, cafetariaa, storage, maintenance and repair facilKlee which are plant or grounds oriented. Important or Low Facilitiee which have mission dependent use (e.g., laboratories, production facilitiea, Hazard Facllitiaa and computer centers) and emergency handling or hazard recovery facilitiee (e.g., hospitals, fire stations). Moderate Hazard Facilltlee where confinement of contents Is necessary for public or employee pro- Facilitiee tection. Examples would be uranium enrichment plants, or other facilitiee involving the handling or storage of significant quantities of radioactive or toxic materials. High Hazard FacilKlee where confinement of contents and public and environment protection are Facilitiee of paramount importance (e.g., facilitiee handling substantial quantitlee of In-procees Plutonium or fuel reprocessing facilities). Facilities in thle category represent hazarda with potential long term and widespread effects.

TABLE 2. PERFORMANCE GOALS FOR EACH FACILITY-USE CATEGORY

Facility Use Parformanca Qoai Performance Goal Annual Category Description Probability of Exceedance General Maintain Occupant Safety 10~3 of the onset of major structural damage to the extent Use that occupants are endangered Important or Occupant Safety, Continued Operation 5x10"4 of facility damage to the extent that the facility Low Hazard with Minimal Interruption cannot perform its function Moderate Occupant Safety, Continued Function, 10*4 of facility damage to the extent that the facility Hazard Hazard Confinement cannot perform its function High Occupant Safety, Continued Function, Very 10~s of facility damage to the extent that the facility Hazard High Confidence of Hazard Confinement cannot perform its function

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

13 greater importance to mission dependence. facilities. Safety related functions They pose a greater danger to on-site require that monitoring and control Feraonnel because of operations or equipment must remain operational during Materials within the facility. The and following a severe wind event that oerformance goal is to maintain capacity exceeds design criteria. The performance to function and provide occupant safety goal is to provide a very high confidence in '•.he event that design wind speeds are that hazardous materials are confined exceeded. The goal is consistent with during and following a high wind or design criteria for essential facilities tornado event. in conventional building codes. Moderate Hazard facilities pose a WIND/TORNADO HAZARD CURVES potential hazard to safety of the general Site specific wind hazard curves were public and the environment owing to the developed for all DOE sites included in presence of radioactive or toxic the project [2]. Straight winds and materials within the facilities. The tornadoes were considered for each site. performance goal is to limit damage such A hurricane hazard assessment was that confinement of hazardous materials performed for the Pinellas Plant in is maintained. Ventilation filtering Florida. must remain functional. Performance Figure 1 shows two typical wind/tornado goals are not met by conventional codes hazard curves. The curve for Oak Ridge or standards. National Laboratory (ORNL) is typical of Confinement of contents, as well as a site with high tornado risk, whereas public and environment protection, is of the curve for Standford Linear paramount importance for High Hazard Accelerator Center (SLAC) is

Tornado Rtqton

6 10 ORHL ID"6

Tornado Region

2 4 I10 SLAC

SITOKJM Wind Rcoion

-2 ° 10 -

10 3 to *

ORNL Design Criteria Connoefs ElUcls or Tornadoes ~i r SLAC Ctuan CriMno Dot) Not -,

10 _3 10 100 ISC 200 250 J00

Wine Sote* [mo*

FIGURE 1. STRAIGHT WIND AND TORNADO HAZARD CURVES FOR TWO SITES Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 14 representative of a site with very low Criteria for Wind tornado risk. The probability associated The hazard probability for General Use with the wind-tornado transition point and Important or Low Hazard facilities is (see Fig. 1) was used to determine if set at an annual probability of design criteria should include effects of exceedance of 2xlO"2 . An importance tornadoes or not. factor greater than unity, in effect, The straight wind hazard curves are increases the probability for Important based on annual extreme wind data from or Low Hazard facilities to lxlO"2. The the site or from nearby National Weather importance factor is multiplied times the Service Stations. A Type I Fisher- wind speed associated with the hazard Tippett extreme value distribution is probability. fitted to the set of annual extreme Minimum missile criteria are provided fastest-mile wind speeds. The tornado for Moderate and High Hazard facilities hazard assessment is based on records to account for windborne debri3 that maintained by the University of Chicago could be picked up by high wind3 or weak [3]. The DAPPLE method of hazard tornadoes. assessment was performed by Dr. Ted Fujita [4], Tornado intensity is based Criteria for Tornadoes on the Fujita-scale [5]. The winds are Because tornado hazard probability is interpreted as 2-3 second gusts. low even in regions where tornadoes are considered common, a relatively low WIND DESIGN CRITERIA FOR DOE SITES annual probability of exceedance is The design criteria judged to meet the specified. This low value is necessary performance goals for each facility-use to be sure that the criteria fall on the category are summarized in Table 3. The tornado hazard curve. The same value of criteria are divided into two parts: 2xlO"5 is specified for both Moderate and criteria for sites that need not consider High Hazard facilities. However, an tornadoes and criteria for those that do. importance factor of 1.35 for High Hazard

TABLE 3. SUMMARY OF MINIMUM DESIGN CRITERIA Building Category General U«e Important or Moderate Hazard High Hazard Low Hazard

Annual Probability 2x10-2 2x10-2 1x10-3 1x10"* of Exceedance w 1 Importance 1.0 1.07 1.0 1.0 n Factor* d

Mlscile Criteria 2x4 timber plank 15 Ib @ 50 2x4 timber plank 15 Ib @ 50 mph (horiz.); max. height 30 mph (horiz.); max. height 50 ft. ft.

Annual Hazard Probability 2x10-5 2x10-5 of Exceedance Importance Factor* I = 1.0 1 = 1.35 APC 40 psl @ 20 p»f/sec 125 p»t @ 50 psf/tec t 2x4timber plank 15 lb@ 100 2x4 timber plank 15 Ib @ 150 o mph (horiz.); max. height mph (horlz.), max, height r 150 tt; 70 mph (vert.) 200 tt; 100 mph (vert.) n • MiMlle Criteria 3 in. dia. ltd. steel pipe, 75 3 in. dia. »td. steel pipe, 75 d Ib @ 50 mph (horlz.); max. Ib @ 75 mph (horiz.); max. 0 height 75 ft, 35 mph (vert.) height 100 tt, 50 mph (vert.) 3,000 Ib automobile @ 25 mph, roll* and tumble*

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

15 facilities increases the wind speed and, TORNADO MISSILES in effect, lowers the exceedance The issue of design criteria for probability. This approach is possible tornado missiles has been a controversial because the slopes of the tornado hazard one for many years, especially in the curves are essentially the same. commercial nuclear power industry. The Tornado resistant design must account Nuclear Regulatory Commission (NRC) for atmospheric pressure change and requires a spectrum of missiles to be missiles. The criteria specify both the considered (including planks, pipes, total pressure drop and a rate of utility poles, and automobiles) in their pressure change. Values appropriate for Safety Analysis Reports. While these tornado design wind speeds at DOE sites missiles and their associated impact are specified. speeds may be consistent with the NRC Tornado missiles include a 2x4 in. criteria (Reg guide 1.76), generic timber plank weighing 15 lb, a 3 in. dia studies by various contractors suggest steel pipe weighing 75 lb, and a 3000 lb that these missiles are very rare events, automobile (High Hazard only). Both a of the order of less than 10"7 probability maximum effective height above ground and per year. The missiles specified in a horizontal missile speed are specified. these guidelines are based on field Computer simulation of tornado missile observations of missiles in tornado trajectories confirm that the maximum damage paths and computer simulations of horizontal missile speed occurs at the the missile trajectories. maximum specified height of the missile. The missiles specified in the design The vertical and horizontal components of and evaluation guidelines (UCRL 15910) the missiles are decoupled. The maximum require certain types of construction to vertical missile speed occurs upon impact resist their perforations. For example, with the ground. The horizontal and the 2x4 timber plank will perforate vertical missile components need not be residential timber construction, metal combined vectorially. cladding, and unreinforced concrete masonry. It will not perforate or spall UNIFORM APPROACH TO DESIGN WIND LOADS concrete masonry walls 8 in. thick or A uniform approach to design for wind more, if all vertical cells are grouted loads has been developed that lets the and reinforced. It will not perforate or designer use essentially the same spall reinforced concrete walls 6 in. procedure regardless of the type of wind thick or more. The 3 in. diameter steel storm being considered. The procedures pipe perforates unreinforced concrete and of ANSI A58.1-1982 [6] can be applied for brick veneer walls. Effects on straight winds, hurricanes, and reinforced concrete masonry walls 8 in. tornadoes. The additional effects of thick or more have not been tested to atmospheric pressure charge and missiles date, although a program is underway to must be taken into account for tornadoes. do this at Texas Tech University. Use of the uniform approach is possible Previous tests show that the pipe missile for the following reasons: 1) wind will not perforate or spall a 9 in. thick pressure characteristics addressed in reinforced concrete wall [7], ANSI A58.1 can be measured in the wind Reasonable estimates of tornado tunnel and on full-scale buildings, and missile impact can be obtained using the 2) damage observed from extreme winds, BRL equation for steel plates and the hurricanes, and tornadoes has similar Rotz equation or the Modified NDRC characteristics. Thus, the wind- equation for reinforced concrete. Large structure interaction, at least to the tornado missiles such as the automobile extent needed for design purposes, for can produce a structural response failure all three types of wind storms is such as a collapsed column or wall panel. similar. Methods for these calculations are found in the literature [8].

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

16 CONCLUSION [7] J.R. McDonald, "Impact Resistance of The guidelines present a uniform Common Building Materials to Tornado approach for determining wind loads for Missiles, " Proceedings of the 6th design and evaluation of DOG facilities. U.S. Conference on Wind Engineering. Once the facility-use category is A. Kareem, Ed. (Houston, TX, March 8- determined, the criteria are determined 10, 1989), University of Houston, for the DOE site under consideration. 1989. The guidelines provide consistent levels of safety from one site to another [8] J.R. McDonald, "Extreme Winds and regardless of geographic location and, to Tornadoes: Design and Evaluation of some extent, between the three natural Buildings and Structures," UCRL- hazard phenomena. 15747, LLNL, Livermore, CA, 1985.

REFERENCES [1] R.P. Kennedy, S.S. Short, J.R. McDonald, M.W. McCann, and R.C. Murray, ''Design and Evaluation Guidelines for Department of Energy Facilities Subjected to Natural Phenomena Hazards," UCRL-15910, U.S. Department of Energy, 1989.

[2] D.W. Coats and R.C. Murray, "Natural Phenomena Hazards Modeling Project: Esitreme Wind/Tornado Hazard Models for DOE Sites, " UCRL-53526, Rev. 1, LLNL, Livermore, CA, 1985.

[3] T.T. Fujita, J.J. Tecson, and R.F. Abbey, "Statistics of U.S. Tornadoes Based on the DAPPLE Tornado Tape," Preprints of the 11th Conf. on Severe Local Storms (Kansas City, MO, Oct. 2-5, 1972) AMS, Boston, MA.

[4] T.T. Fujita, "Workbook of Tornadoes and High Winds for Engineering Applications," SMRP No. 165, Dept. of Geophysical Sciences, the University of Chicago, 1978.

[5] T.T. Fujita, "Proposed Characterization of Tornadoes and Hurricanes by Area and Intensity," SMRP No. 91, Dept. of Geophysical Sciences, the University of Chicago, 1971.

[6] ANSI, "Minimum Design Loads for Buildings and Other Structures," ANSI A58.1-1982, American National Standards Institute, New York, NY.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

17 DOE NATURAL PHENOMENA GUIDELINES FLOOD DESIGN AND EVALUATION

Martin W. MeCann Jack R. Benjamin & Associates Mountain Vi«w, California Jtan B. Savy Lawrencs Uvermort National Laboratory LJv«rmor«, California

ABSTRACT Design and evaluation guidelines for DOE facilities subjected to earthquake, wind/tornado, and flood have been developed and presented in UCRL-15910 (Ref.1). This paper summarizes Chapter 6 of UCRL-15910 describing the philosophy and procedures for the design or evaluation of facilities for flood, The flood design and evaluation guidelines seek to ensure that DOE facilities satisfy the performance goals described in UCRL-15910. The guidelines are applicable to new and existing construction; however, in the evaluation of existing facilities, fewer design options may be available to satisfy the performance goals. Evaluation of the flood design for a facility consists of: 1) defining the design basis flood (DBFL), 2) evaluating site conditions (e.g., facility location, location of openings and doorways), and 3) assessing flood design strategies (e.g., build above DBFL levels, harden the site).

DESIGN BASIS FLOOD (DBFL) The first two items are determined as part of the Use of the term DBFL should be understood site hazard assessment. Flood loads must be to mean that multiple flood hazards may be assessed on a facility-by-facility basis. included in the design. For example, a site Table 1 defines the design basis events that located along a river may have to consider the must be considered. For each hazard, the worst potential for river flooding as well as the possible combination of events defines the DBFL. These hazards associated with rainfall that could cause events apply for all facility categories, subject to onsite flooding (e.g., roofs, streets). Factors the constraint that the probability of exceedance contributing to potential river flooding such as is equal to or greater than the design basis. For example, if the design basis flood probability for spring snowmett or upstream dam failure must 3 be considered. The DBFL for each flood type General Use facilities is 2x10* per year, failure of an upstream dam need not be considered if (e.g., river flooding, rainfall, snow) is defined in 3 terms of: the frequency of failure is less than 2x10~ . For purposes of design, the event combinations in 1. peak flood level (e.g., flow rate, volume, Table 1 are assumed to be perfectly correlated. elevation, depth of water) corresponding to In other words, the combinations of events listed the mean hazard annual probability of are assumed to occur with certainty if the con- exceedance, ditions stated are met. 2. combinations of events (e.g., storm surge, wave action); and EVALUATION OF SITE CONDITIONS The steps in the flood evaluation process as 3. evaluation of flood loads (e.g., hydrostatic illustrated in Figure 1 include: and/or hydrodynamic forces, debris loads).

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

18 TABLE 1 DESIGN BASIS FLOOD EVENTS

Primary Hazard Event River Flooding 1. Tide Effect* (If applicable) 2. Wind wave activity and Event 1. (above). 3. Coincident upstream dam failure, if for the design basis flood, (1) the reservoir elevation is greater than or equal to an elevation which is 90% of available free-board; or (2) spillway is structurally unable to pass the design basis flood; and Events 1. and 2. above. 4. Ice forces and Event 1. above. Dam Failure All modes must be considered (e.g., seismicalty induced, random structural failures, upstream) Local Precipitation Roof drains clogging, and storm sewers blocked Tsunami Tide effects. Storm Surge (due to, e.g., hurricane, Tide effects and wind wave activity (If not Included in the hazard analysis). seiche) Levee or Dike Failure Consider failure for events less than the deeign basis (i.e., failure during a flood, less than the design basis).

1. Determine the facility category (see UCRL- 8. If the facility is located below the DBFL 15910). level (even if the facility has been hard- 2. From the results of a site screening analy- ened), emergency procedures should be sis or flood hazard study, identify the provided to evacuate personnel and to sources of flooding at the facility. secure the facility when the flood arrives. 3. Based on the flood design guidelines in In principle, buildings that fit into one category UCRL-15910, determine the DBFL for each or another should be designed for different flood hazard. The design basis flood hazard levels because of the importance should include possible combinations of assigned to each. However, because floods hazards and the assessment of flood loads have a common-cause impact on all buildings at (e.g., hydrostatic and hydrodynamic loads) or below the design basis flood level, the design or other effects (e.g., scour, erosion). basis for the most critical structure may govern 4. Determine whether the site or facility is the design for all buildings onsite when it is more situated above the DBFL flood level. If not, feasible to harden a site, rather than an individual alternative design strategies must be con- building. Exceptions to this case exist when sidered such as hardening the facility or buildings are at significantly different elevations developing emergency operation plans. or there are large spatial separations, or in the case when individual buildings are hardened to 5. Evaluate whether roof drainage is ade- resist the expected flood loads (i.e., addition of quate to convey design level precipitation watertight doors to a High Hazard facility build- to prevent ponding or excessive roof loads. ing). The structural design of the roof system It is important to consider possible interaction should also be evaluated. between buildings or building functions. For 6. Evaluate the site stormwater management example, if a High Hazard facility requires system to determine whether applicable emergency electric power in order to maintain design regulations (i.e., DOE 6430.1 A [Ref. safety levels, buildings which house emergency 2] or local regulations) are satisfied. Site generators and fuel should be designed to a High drainage should also be adequate to sat- Hazard category flood level. In general, a sys- isfy the DBFL (e.g .precipitation). tematic review of a site for possible common- cause dependencies is required. 7. For existing construction, review whether the building and/or the site are hardened by adequate flood protection devices.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

19 II i a e n \ • y s ource 6 0 > Fiooa 1 0 e 31 g n Bas 3 F 10 00 Fo' A I

Yes

Yes I R Ov Roof Design An d Si te Sto r mwaief Drainage

/ N 0 Rev i 9 a

Crllftria f s^ Oes l g n

Yes

FIGURE 1. FLOOD EVALUATION PROCEDURE FLOOD DESIGN STRATEGIES stand the effects of flood forces such that The basis for the flood evaluation procedure the performance goals are satisfied. is defined according to a hierarchy of design 3. For the DBFL, if adequate warning is avail- strategies. They are: able, emergency operation plans can be developed to safely evacuate employees 1, Situate facilities above the DBFL level. and secure areas with hazardous, mission- 2. Harden a site or individual facility to with- dependent, or valuable materials.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

20 If a DOE facility is situated above the DBFL, the design decisions may have resulted in all build- performance goals are readily satisfied. An ings being sited above possible flood levels. option to satisfy the performance goals is to Consequently, in some cases it may be apparent harden a building or site against the effects of that floods do not pose a substantial hazard to floods such that the chance of damage is facility operations. For these so called "dry sites," acceptably low and to provide emergency it may be possible to demonstrate, without per- operation plans. This dual strategy is secondary forming a detailed hazard assessment, that the to siting facilities above the DBFL level because flood design guidelines described in this some probability of damage does exist and document are satisfied. facility operations may be interrupted. The concept of a dry site as used here does Whether or not a facility is situated above the not imply that a site is free of all sources of DBFL should be assessed on the basis of the flooding (e.g., all sites are exposed at least to critical flood elevation. The critical elevation precipitation). Rather, a dry site is interpreted to represents the flood level at which, if flooding mean that facilities (new or existing) are located were to occur beyond this depth, the perform- high enough above potential flood sources such ance level specified as part of the performance that a minimum level of analysis demonstrates goals would be exceeded. Typically, the first floor that design guidelines are satisfied. For example, elevation or a below-grade elevation (i.e., foun- for the flooding source of local precipitation, the dation level) is assumed to be the critical eleva- adequacy of the stormwater management sys- tion. However, based on a review of a facility, it tem can be readily demonstrated (e.g., roof may be determined that only greater flood depths drainage, storm sewers, local topography). would cause damage (e.g., critical equipment or To consider flood hazards at DOE sites, a materials may be housed above the first floor). two-phase evaluation process is used. In the first The critical elevation will depend on the flood phase, flood screening analyses are performed. hazard (e.g., hydrostatic, hydrodynamic loads), These studies provide an initial evaluation of the the building structure, and the facility category. potential for flooding at a site. As part of the screening analysis, available hydrologic data DOE FLOOD HAZARD ASSESSMENTS and results of previous studies are gathered, and While probabilistic hazard evaluations for a preliminary assessment of the probability of seismic and wind phenomena have been per- extreme floods is performed. Results of the formed for all of the DOE sites, comparable screening analysis can be used to assess evaluations for flood hazards have been whether flood hazards can occur at a site. In performed at only 9 of these sites. Flood some cases, these studies may demonstrate that screening evaluations have been performed for flood hazards are extremely rare and, therefore, eight sites in the jurisdiction of the Albuquerque performance goals are satisfied. For those sites Operations Office. Also, a flood hazard with high potential for flooding and which have assessment has been performed for the Hanford Moderate Hazard and High Hazard facilities, the Project Site. The results of these evaluations second phase will be undertaken. This consists have been summarized in Reference 3. of detailed probabilistic flood hazard assess- All sites are exposed to the potential effects of ment. flooding. For example, localized flooding due to A number of methods have been developed rainfall can occur in streets, in depressed areas, to assess the probability of extreme floods. and on roofs. In addition, flooding can occur on These include: a nearby river, lake, or ocean. The objective of probabilistic hazard evaluations is to assess the 1. extrapolation of frequency distributions, probability of extreme events that have a low probability of being exceeded. In the case of 2. joint probability techniques, floods, facilities at DOE sites may not be exposed 3. regional analysis methods, to extreme flood hazards. Because of topogra- 4. paleohydrologic evaluation of floods, and phy, regional climate, or the location of sources of flooding in relation to a site, extreme flooding 5. Bayesian techniques. on-site may be precluded. For existing facilities,

Second DOE Natural FMienomena Hazards Mitigation Conference - 1989 21 There is no general agreement in the literature For Moderate and High hazard facilities, a regarding the appropriateness of these methods comprehensive flood hazard assessment should to estimate the probability of extreme floods. be performed, unless the results of the screening Each approach has its advantages and disad- analysis demonstrate that the performance goals vantages and thus no single technique is well- are satisfied. established. General Use Facilities In estimating the probability of extreme floods The performance goal for General Use facili- it is important that uncertainty analysis be per- ties specifies that occupant safety be maintained formed. The uncertainty analysis should con- and that the probability of severe structural sider statistical uncertainty due to limited data damage be less than or about a 10*3 per year. and the uncertainty in the flood evaluation For General Use facilities, the DBFL corresponds models used (e.g., choice of different statistical to the hazard level whose mean annual proba- models, uncertainty in flood routing). bility of exceedance is 2x10"3. In addition, event combinations that must be considered are listed FLOOD DESIGN GUIDELINES FOR EACH in Table 1. FACILITY-USE CATEGORY To meet the performance goal for this cate- Table 2 shows guidelines recommended for gory, two requirements must be met: (1) the each facility category in terms of the hazard input, facility structural system must be capable of hazard annual probability, design requirements, withstanding the forces associated with the DBFL and emergency operation plan requirements. and (2) adequate flood warning time must be Unlike design strategies for seismic and wind available to ensure that building occupants can hazards, it is not always possible to provide be evacuated (1 to 2 hours). If the facility is margin in the flood design of a facility. For located above the DBFL, then structural and example, the simple fact that a site is inundated occupant safety requirements are met. (forgetting for a moment the possible structural For structural loads applied to roofs, exterior damage that might occur), may cause significant walls, etc., applicable building code require- disruption (clean-up) and downtime at a facility; ments (e.g., DOE 6430.1 A, Uniform Building this may prove an unacceptable risk in terms of Code (UBC) References 2,4) provide standards economic impact and disruption of the mission- for design that meet the performance goal for dependent function of the site. In this case, there General Use facilities. is no margin, as used in the structural sense, that For existing construction, or at new sites can be provided in the facility design. Therefore, where the facility cannot be above the DBFL level, the facility must be kept dry and operations must an acceptable design can be achieved by: be unimpeded. As a result, the annual probability of the DBFL corresponds to the performance 1. Providing flood protection for the site or for goal probability of damage, since any excee- specific General Use facilities, such that dance of the DBFL results in consequences that severe structural damage does not occur, exceed the performance goal. and The DBFL for General Use and Important or 2. Developing emergency procedures in Low Hazard facilities can generally be estimated order to secure facility contents above the from available flood hazard assessment studies. design flood elevations in order to limit These include: the results of flood screening damage to the building to within accept- studies, flood insurance analyses, or other able levels and to provide adequate warn- comparable evaluations. For these facility types, ing to building occupants. it is not necessary that a full-scope hazard evaluation be performed, if the results of other recent studies are available and, if uncertainty in the hazard estimate is accounted for.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

22 TABLE 2 FLOOD GUIDELINES SUMMARY

Facility Us* Category Flood Design General Use Important or Moderate Hazard High Hazard Step Low Hazard Flood Hazard Flood Insurance studies or Flood insurance studies or Site probabilistic hazard Site probabilistic hazard Input equivalent input and Table equivalent Input and Table analysis and Table 1 analysis and Table 1 1 combinations 1 combinations combinations combinations Hazard Annual 2x10'3 5x10"4 1X10-4 1x10-5 Probability Structural UBC or applicable criteria UBC or applicable criteria Flood hazard Flood hazard Evaluation for roof and site drainage, for roof and site drainage, analysis, strength design analysis, strength design (Roofs, etc.) building load factors, and building load factors, and design criteria design criteria Warning and Required to evacuate on- Required to evacuate on- Required if buildings are Required if building* ant Emergency site personnel if site is site personnel and to below DBFL below DBFL Procedures below DBFL secure vulnerable areas if site is below DBFL level

Important or Low Hazard Facilities As in the case of General Use facilities, UBC The performance goal for Important or Low design standards or local ordinances should be Hazard facilities is to limit damage and interrup- used to determine design requirements and site tion of facility operations while also maintaining drainage. Site drainage should be adequate for occupant safety. For these facilities, the DBFL is roofs and walls to prevent flooding that would equal to the flood whose probability of excee- interrupt facility operations. dance is 5x10"4 per year plus the event combi- Moderate Hazard Facilities nations listed in Table 1. The results of flood The performance goal for Moderate Hazard insurance studies routinely report the flood level facilities is continued function of the facility, corresponding to the 2x10"3 probability level. For including confinement of hazardous materials purposes of establishing the DBFL for Important and occupant safety. Facilities in this category or Low Hazard facilities, the results of these should be located above flood levels whose studies can be extrapolated to obtain the flood annual probability of exceedance is 10"4, with a probability of 5.0x10"4 of being exceeded including the combinations of events shown !n (if this result is not reported). A range of Table 1. extrapolations should be considered, with a Emergency operation procedures must be weighted average being used as the design developed to secure hazardous materials, pre- basis. pare Moderate Hazard facilities for possible For new construction, facilities in this category extreme flooding and loss of power, and for an should be located above the DBFL. For existing extended stay on-site. Emergency procedures construction, or at new sites where the above should be coordinated with the results of the siting criteria cannot be met, an acceptable flood hazard analysis, which provides input on design can be achieved by the same measures the time variation of flooding, type of hazards to described for General Use facilities. For Impor- be expected, and their duration. The use of tant or Low Hazard facilities whose critical ele- emergency operation plans is not an alternative vation is below the DBFL, emergency procedures to hardening a facility to provide adequate con- must be developed to mitigate the damage to finement unless all hazardous materials can be mission-dependent components and systems. completely removed from the site. These procedures may include installation of Roofs should be designed in accordance with temporary flood barriers, removal of equipment UBC standards in order to drain rainfall whose to protected areas, anchoring vulnerable items, probability of exceedance is 10"4. The amount or installing sumps or emergency pumps. of ponding that can occur on building roofs should be controlled by adding scuppers (openings in parapet walls) and/or limiting parapet wall heights. If ponding on-site is expected

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 23 to occur, drainage should be provided to convey depth on a roof specified by DOE 6430.1 A or the stormwater away from the facility. Alterna- applicable local regulations apply. The roof tively, doors and openings should be made should be designed to consider the maximum watertight. depth of water that could accumulate if the High Hazard Facilities primary drainage system is blocked. The performance goals for High Hazard faci- Design Requirements lities are basically the same as for Moderate Design criteria (i.e., for allowable stress or Hazard facilities. However, a higher confidence strength design, load factors, and load combi- is required that the performance goals are met. nations) for loads on exterior walls or roofs due Facilities in this category should be located to rain, snow, and ice accumulation should follow above flood levels whose annual probability of applicable code standardsfor the materials being exceedance is 10"5, including combinations of used. events listed in Table 1. Required emergency General Use and Important or Low Hazard operation procedures are the same as those for Facilities Moderate Hazard facilities. Roofs should be Facilities that are subject to flood loads should designed in accordance with UBC standards in be designed according to provisions of UBC or order to drain the rainfall whose probability of local ordinances and specified flood load exceedance is 10"5. The control of ponding is combinations (e.g., ponding, hydrostatic). the same as that recommended for Moderate Moderate and High Hazard Facilities Hazard facilities. Buildings and related structures that are directly impacted by flood hazards should be FLOOD DESIGN PRACTICE FOR FACILITIES constructed of reinforced concrete and designed BELOW THE DBFL ELEVATION according to strength methods as required by For structures located below the design basis ACI 349-85. Load factors and combinations flood level, mitigation measures other than siting specified in Reference 5 should be used. at a higher elevation can provide an acceptable Design of Other Civil Engineering Facilities margin of safety. In general, structural measures In addition to the design of buildings to with- are considered next, followed by non-structural stand the effects of flood hazards, other civil actions (i.e., flood warning and emergency works must be designed for flood conditions. operations plans). In practice, for sites located These include components of the stormwater below the design basis flood level, a combination management system such as street drainage, of structural and non-structural measures are storm sewers, stormwater conveyance systems used. Guidelines for structural flood mitigation such as open channels, and roof drainage. measures are described in this section. Applicable procedures and design criteria spe- Flood Loads cified in DOE 6430.1 A and/or local regulations To evaluate the effects of flood hazards, cor- should be used in the design of stormwater responding forces on structures must be evalu- systems. However, the design of individual ated. Force evaluations must consider facilities to resist the effects of local, onsite hydrostatic and hydrodynamic effects, including flooding (e.g., local ponding, street flooding) the impact associated with wave action. In should be evaluated to ensure that the perform- addition, the potential for erosion and scour and ance goals are satisfied. debris loads must be considered. Good engi- Flood Protection Structures neering practice should be used to evaluate flood Facilities can be hardened to withstand the loads. The forces due to ice formation on bodies effects (e.g., loads, erosion, scour) of flood of water should be considered in accordance hazards. Typical hardening systems are: with DOE 6430.1 A. Building roof design should provide adequate 1. structural barriers (e.g., building, watertight drainage as specified by DOE 6430.1 A and in doors), accordance with local plumbing regulations. 2. waterproofing (e.g., waterproofing exterior Secondary drainage (overflow) should be pro- walls, watertight doors), vided at a higher level and have a capacity at least that of the primary drain. Limitations of water 3. levees, dikes, seawalls, revetments, and

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 24 4. diversion dams and retention basins. REFERENCES 1. Kennedy, R,P,, Et Al, "Design and Evaluation Applicable design guides for levees, dikes, Guidelines for Department of Energy Facili- small dams, etc. can be found in U.S. Army Corps ties Subjected to Natural Phenomena Haz- of Engineers, U.S. Bureau of Reclamation, Soil ards," (draft) UCRL-15910, Lawrence Conservation Service reference documents. Livermore National Laboratory, Livermore, Design of structural systems such as exterior CA, May 1989. waits, roof systems, doors, etc. should be 2. U.S. Department of Energy, General designed according to applicable criteria for the Design Criteria, DOE Order 6430.1 A, facility category considered. Washington, D.C., 1987 (draft). FLOOD RISK ASSESSMENT 3. Savy, J.B. and R.C. Murray, "Natural Phe- In some cases the need may arise for DOE or nomena Hazards Modeling Project: Flood the DOE site manager to perform a quantitative Hazard Models for Department of Energy flood risk assessment. There may be a variety of Sites," UCRL-53851, Lawrence Livermore reasons requiring a comprehensive risk evalu- National Laboratory, Livermore, CA, 1988. ation of a site. These considerations include: 4. Uniform Building Code, 1988 Edition, Inter- national Conference of Building Officials, 1. Demonstration that the performance goals Whittier, California, 1988. are satisfied. 5. Federal Insurance Administration, Federal 2. Evaluation of alternative design strategies Emergency Management Agency, "Exhibit 1, to meet the performance goals. Flood Insurance Manual Revision," Washing- 3. Detailed consideration of conditions at a ton, D.C., October 1980. site that may be complex, such as varying hydraulic loads (e.g., scour, high velocity flows), system interactions, secondary failures, or a potential for extraordinary health consequences. 4. A building is not reasonably incorporated in the four facility categories.

A quantitative evaluation of the risk due to flooding can be assessed by performing a probabilistic safety analysis (PSA). The objective of a flood PSA is to evaluate the risk of damage to systems important for maintaining safety and operating a critical facility. Risk calculations can be performed to evaluate the likelihood of damage to onsite systems and of public health consequence.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

25 Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

26 Session 2 Seismic Analysis

Second DOE Natural Phenomena Hazards Mitigations Conference - 1989

27 SEISMIC AND CASK DROP EXCITATION EVALUATION OF THE TOWER SHIELDING REACTOR

S. P. Harris^ R. L. Stover ^ J. J. Johnson # B. N. Sumodobila

ABSTRACT

During the current shutdown of the Tower Shielding Reactor II (TSR-II), analyses were performed to determine the effect of nearby cask drops on the structural and mechanical integrity of the reactor. This evaluation was then extended to include the effects of earthquakes. Several analytic models were developed to simulate the effects of earthquake and cask drop excitation. A coupled soil-structure model was developed. As a result of the analyses, several hardware modifications and enhancements were implemented to ensure reactor integrity during future operations.

"The submitted manuscript hu been •uttered by • contractor of the U.S. Government under contract No. DB- AC05-MOR21400. Accordingly, the US. Government reuini • nonudutivc, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for U.S. Government Purposes.'

INTRODUCTION PLANT DESCRIPTION The Tower Shielding Reactor facility was built in 1954 The TSR facility is located at the Oak Ridge National and is now used as a low power radiation source for Laboratory in Oak Ridge, Tennessee. It is a pressurized, radiation shielding experiments. In the past, the TSR site light water moderated and cooled reactor operating at a was also used as a test facility for dropping spent fuel casks maximum of 1 MWt. The core is spherical in shape and weighing up to 50,000 pounds. consists of curved uranium-aluminum fuel plates with After the TSR-II was shut down in 1987 following safety control plates inside the fuel annulus. The reactor technical concerns at the High Flux Isotope Reactor, vessel sits inside of a concrete shielding structure (Big independent review teams recommended that several Beam Shield) consisting of 5 feet of concrete on three assessments be performed before restart. The first of sides and a movable neutron beam shutter on the fourth these was to determine the effects of past cask drops at the side (Figure 1). TSR facility on the structural and mechanical integrity of the TSR-H. The second was to evaluate the response of ANALYTIC MODELS the reactor to a seismic event. EQE was requested to The reactor vessel is composed of two major perform both of these evaluations. Several analytic structural elements: the reactor pressure vessel and the models were developed to simulate the effects of cask reactor internals. The pressure vessel is basically a drop and earthquake excitation. The analyses were cylindrical aluminum tank with a hemispherical bottom performed in the time domain. The cask drop analysis and a circular cover plate (head). The tank is utilized accelerometer data from 5,000- and 12,900-pound approximately 8 feet long and 3/4 inch thick; its inside cask drops. The seismic analysis used an acceleration time diameter at the hemispherical (lower) end is 37 inches, history matched to a 0.15g Newark-Hall response and 40 inches at the upper end. The reactor vessel is spectrum. supported in the vertical direction only at the bottom

* EOE Engineering ** Oak Ridge National Laboratory Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

Research sponsored by the Office of Energy Research and 28 Development, VS. Department of Energy, under contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc. surface of the Hanged connection lo the vessel head. were obtained from analyses using finite element While the reactor internals are considerably more programs. The SSI response calculations were performed complex than the pressure vessel, the two major structural using Fourier analysis techniques. components consist of the central cylinder assembly and (he ioni/alion chamber/lead shield (lead shield), The REACTOR MODEL central cylinder is a 17-inch diameter (inside) lube The shield structure model utilizes a shear beam suspended from the reactor head and co-linear with the representation of the structure. The reactor vessel is vertical axis of the vessel. It travels the full length of the coupled with the shield model at the reactor support reactor and supports ihe reactor core fuel elements. The sleeve. The reactor vessel model includes the reactor central cylinder also provides the majority of the lateral internals composed of ihe two major substructures, the load resistance for the reactor internals. Four labs internals central cylinder, and the ionization chamber around ihe bottom edge of the central cylinder fit between guide assembly and lead shield. lugs fastened lo ihe 3/4-inch thick hemispherical The model includes a 3-dimcnsional representation of aluminum shell surrounding the lower fuel elements with the reactor vessel and the central cylinder as two a small gap. If the bottom of the central cylinder cantilever beams coupled at the reactor vessel head, which displaces laterally and closes this gap, the lugs provide is connected at ihc support sleeve to ihe shield structure. support for the cylinder, changing its behavior from a free The vessel head is assumed to be a rigid coupling to a propped cantilever. member. The guide tube-lead shield assembly model The control support tube assembly is located within includes the vertical support members and the asymmetric ihe central cylinder and supports the control mechanism affects of the guide tube attachment of the lead shield. and the upper fuel elements. The lead shield fits within Node locations are chosen to coincide with locations of ihe central cylinder and is also suspended from the mechanical connections, critical stress locations, or where reactor head. There is a 1/8-inch gap between the lead displacements are deemed important for determination of shield and Ihe central cylinder, allowing independent gap closures. The single finiic element model (Figure 3) structural response as long as this gap is not closed. developed lo represent the major structural components Figure 2 shows a simplified view of the reactor illustrating of the TSR-II: the shield structure and foundation, the the gaps between the central cylinder and reactor vessel, reactor vessel, (he central cylinder, and the lead shield is as well as between the central cylinder and lead shield, shown. The half-space stiffness matrix at the base of the COUPLED SOIL STRUCTURE MODEL model was calculated using a rigid foundation mat with a A coupled soil structure model was developed for the size of 15 by 20 feel and represented the soil stiffness seismic evaluation. The earthquake motions arc broad properties in translation, vertical and rocking directions. banded dynamic excitation, and coupled soil structure This matrix was developed for cases which envelope the response is an important contributor to overall system shallow clay soil site conditions. Limited soil boring data performance. The beam shield structure was modeled lo indicate a predominantly medium clay material underlying predict the frequency of vibration for the combined soil the shield beam structure. Values of soil shear wave and shield structure. The soil structure interaction (SSI) velocities of 600, 900,1350, and 3500 fps were selected problem was solved in the frequency domain by the which envelope a range from medium clay lo rock-like soil substructure approach using the CLASSI computer code. stiffness. Time history analyses of the coupled soil, The key elements of this method were: specification of support structure, and reactor vessel model were run for ihe free-field ground motion, representation of the each soil property value. foundation impedance, and representation of the dynamic Nonlinear dynamic effects of the response, interaction, characteristics of the structures. These elements were and possible coupling forces between two reactor internal combined in the SSI analysis to calculate the response of structures were recognized. These arc the response of the the coupled soil-structure system. The free-field motion central lead shield structure within the small (1/8 inch) was represented by an orthogonal triad of acceleration annular space formed by the internal central cylinder time histories at the ground surface. The foundation (Figure 2). When a structural component vibrates in a impedances were represented by frequency-dependent, viscous fluid, fiuid-struclure interaction effects gives rise complex-valued matrices; the impedance matrix for a to a fluid force which can be characterized as an added single rigid foundation was a (> x 6 matrix relating six force mass and damping contribution to the dynamic response components to six displacements. The structures were of the component. represented by their fixed-base cigen-syslcms and modal Analytic and correlated test studies have been damping factors, and pseudo-static modes and foundation performed by several investigators of this effect. These coupling stiffness matrices. These structural input dala effects were quantified to determine ihe added effective

DO!.'. Natural Phnxnncna lla/anis Mitigation Conference - 1989 TANK ASWIMLY

COVER HATE

SHUTTER ASSEMBLY

Figure 1: Section Through Beam Shield Looking South

•^--FUEL ELEMENTS

NOT TO SCUlt Figure 2: TvSR-II Reactor Schematic View

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

30 mass and damping effects, The calculated added mass SEISMIC ANALYSES and dumping effects on the lend shield were 81 times the Response time history analyses were performed for nominal mass and 12% of critical, respectively. This the combined soil, beam shield structure, reactor-coupled model is called the "uncoupled" internals model. A lead shield and central cylinder model. Response second "coupled" model was also utilized which assumed parameters versus time were calculated for each node of the lead shield and central cylinder interaction results in the model. The response of this coupled soil and completely coupled response. structure system envelope the effect of assumptions on soil shear wave velocity by computation of response for CASK DROP soil shear wave velocities of 600, 900, 1350, and 3500 fps. Two drop tests of 30-foot height were performed and As the soil stiffness increases, the coupled soil-structure measurements of the response of the TSR-II were made. frequency varies from 7 Hz to 15 Hz. For the 3500 fps The casks differed in weight (5,000 and 12,900 pounds) case, rock-like soil conditions, no significant soil-structure and provided information on the TSR-IJ response as a response frequency was exhibited (Figure 4). function of load level. The top head of the vessel was The reactor internals, the central cylinder, and lead instrumented with two 3-componenl acceleromcters. The shield respond to the input motions as a cantilever. The time history recordings were processed as follows: envelope displacement response of the uncoupled (1) The two vertical records arc averaged lime-step by internals model indicated no impact of the central cylinder time-step to obtain the time history of vertical with its bottom stops. The displacement response of the acceleration. lead shield was 1.8 inches, however, much greater than the (2) The iwo vertical records are subtracted time-step by 1/8-inch clearance to the central cylinder and therefore lime-step and the resultant scaled by the reciprocal of the impact would occur. With impact, the response of the distance between the instruments to obtain the rocking system cannot be characterized as linear clastic and lime history. harmonic, but is nonlinear, forced dynamic response.

CASK DROP ANALYSES NONLINEAR RESPONSE STUDY The dynamic response of the TSR-II was predicted A one degree of freedom model of the lead shield using the processed acceleration time histories as input was developed and its response to the time history of motion at the top of the reactor model without the beam motions at the reactor vessel head over the range of soil shield structure portion of the model. The structural parameters was solved. The solution was performed using system was analyzed with the computer code SSIN, which a direct step wise integration procedure of the equations uses the mode-superposition method of dynamic analysis of motion developed by N. C. Nigam and P. C. Jennings of in the lime domain, and assumes a linear multiple degree the California Institute of Technology. of freedom model. A computer code was used to perform ihe Nigam- Using the time histories from the 12,900-pound Jennings procedure. A modification was made to check (12.()K) drop, the model was run using the coupled and for each time step "t" that the clearance between the uncoupled model described above. Displacements and shield and central cylinder is less than 1/8 inch. When velocity time histories at the critical clearances between contact was made, the velocity V of the lead shield was sel reactor internal structures were obtained. to -V and the direct solution was continued. This solution The central cylinder responds to the input motions as represents forced dynamic motion, impact, and fully a cantilever with a fundamental frequency of 20.7 Hz, and elastic rebound of the lead shield off of the central its maximum tip displacement just exceeds ihc clearance cylinder. to ihe bottom guides. Energy transfer upon impact with Acceptable values of impact velocity were developed the bottom guides is negligible, since the tip velocity is using energy balance techniques. The nonlinear gap approaching zero. The lead shield responds to the input conditions were approximated by using an energy balance motion as a cantilever, with a maximum displacement approach at the point and time of impact (gap closure) to much less than 1/8 inch, insufficient to cause impact with distribute loads to the structural elements. This method the central cylinder. Physically, the internal behavior can used the velocity at closure to define the kinetic energy of be thought of as the central cylinder raltling between the the system, and equates this to the strain energy of the bottom guides at a frequency near 21 Hz and the lead resisting members. The equivalent static loads were then shield oscillating independently around 3 Hz. calculated. Allowable component member stresses were The analyses of both the lead shield and central computed using simplified analytical techniques to cylinder were extrapolated to a 50K drop lest by assuming distribute the forces resulting from the dynamic analysis. a linear increase in ihe amplitude of the input acceleration lime histories. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

31 "T t^Jr

Ructoa CENTRAL (ONtZATlONOtMEASSEIWLV VESSEL CYLINDER •HO LEAO JM1CLD

H*lF5P»Ci SOK.H0MI

(•»na«*i«'*M«

Figure 3: Coupled Soil Structure Reactor Model

X 10 J 0.S

0.'

.2 0.3-

io- 10' Frequency (Hz)

Legend: Notes: 7s > 600 fps Spectra for 51 Damping Vs - 900 Jps Acceleration - i.n/sec**2 Vs - :350 fps Location - NoQe 18 Vs - 3500 fps

Figure 4: Acceleration Time History Response Spectra of the Reactor Vessel Head for Varied Soil Properties

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

32 RESULTS The conclusion reached was that given the 1/8-inch The acceptable values for both cask drop excitation gap, impact of the lead shield and central cylinder may and seismic were found to be controlled by the strength of exceed design allowables of the central cylinder joint for the central cylinder screw joint. Thirty-two stainless steel the 0.15g evaluation earthquake (Figure 5). Impact screws connect the upper and lower halves of the central velocities of the central cylinder against its bottom stops cylinder. The lower portion of the central cylinder is due to cask drop excitation arc shown in Figure 6. fabricated of 6061 T6 aluminum alloy and the upper section is 5052 aluminum.

5.0 10.0 15.0 X 10 TlM-Stcond*

Figure 5: Velocity Time History for Nonlinear Forced Dynamic Excitation

X 10,-1 •: 0.01M

fir HP \i \L}L yvvv '^vr 1/32" gap to bottom Impact, v«loclty> 2.0 ln/a*c. fluid*

0.0 0.2 0.4 O.f 0.* 1.0 1.2 X 10 Tiat-MeoMs

Figure 6: Displacement Response Time History Scaled for 50K Drop, X-Direction, Central Cylinder Assembly

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

33 CONCLUSIONS AND RECOMMENDATIONS For cask drop loads, the reactor vessel, internal structure, support components, and connections were found to be within design stress allowables except for two components. The central cylinder joint screws and the bottom plate weld on the fission chamber well were both predicted to exceed the design allowable value. As a result, cask drops were suspended pending further assessments. In addition, an assessment of the potential weld failure concluded that there was no impact on safe operation of the TSR-II. From the seismic analysis, it was concluded that seismic excitation would not prevent the safety-control plates from functioning and terminating reactor operation on demand. However, the central cylinder joint screws were found to exceed the design allowable value. As a result, spacers were installed between the central cylinder and lead shield to reduce the gap and impact velocities to acceptable values. Tensile tests were also performed on the screws to establish allowable values. Completion of these actions allowed the design allowable value to be met.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

34 SEISMIC EVALUATION OF SAFETY SYSTEMS AT THE SAVANNAH RIVER REACTORS

Gregory S. Hardy Thomas Monahon James J. Johnson and and Darrel Ketcham Stephen J. Edcr of of Westinghousc Savannah River Co EQE Engineering, Inc. Savannah River Site Costa Mesa, CA 92626 Aiken, S.C. 29801 San Francisco, CA 94105

ABSTRACT

A thorough review of all safety related systems in commercial nuclear power plants was prompted by the accident at the Three Mile Island Nuclear Power Plant. As a consequence of this review, the Nuclear Regulatory Commission (NRC) focused its attention on the environmental and seismic qualification of the industry's electrical and mechanical equipment. In 1980, the NRC issued Unresolved Safety Issue (USI) A-46 to verify the seismic adequacy of the equipment required to safely shut down a plant and maintain a stable condition for 72 hours. After extensive research by the NRC, it became apparent that traditional analysis and testing methods would not be a feasible mechanism to address this USI A-46 issue. The costs associated with utilizing the standard analytical and testing qualification approaches were exorbitant and could not be justified. In addition, the only equipment available to be shake table tested which is similar to the item being qualified is typically the nuclear plant component itself. After 8 years of studies and data collection, the NRC issued its "Generic Safety Evaluation Report" approving an alternate seismic qualification approach based on the use of seismic experience data. This experience-based seismic assessment approach will be the basis for evaluating each of the 70 prc-1972 commercial nuclear power units in (he United States and for an undetermined number of nuclear plants located in foreign countries. This same cost-effective approach developed for the commercial nuclear power industry is currently being applied to the Savannah River Production Reactors to address similar seismic adequacy issues. This paper documents the results of the Savannah River Plant seismic evaluation program. This effort marks the first complete (non-trial) application of this stalc-of-lhe-arl USI A-46 resolution methodology.

INTRODUCTION

The Savannah River Site (SRS) material production Because these arc nuclear material production reactors reactors (K, L and P) were built in the early 1950's for which do not generate electricity, their design is the Atomic Energy Commission and arc currently significantly different from a commercial nuclear owned by the Department of Energy (DOE). The power plant. The K, P and L Reactors arc low-pressure reactors were operated by E.I. du Pont dc Nemours & reactors moderated and cooled by heavy water. Company until early 1989 and arc currently operated Reactor control is provided by a conventional control by the Westinghousc Savannah River Company. rod system, and an emergency shutdown capability is

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

35 provided by a gravity-driven safety rod system backed SRS SEISMIC PROGRAM FORMAT up by an independent, liquid-poison (gadolinium nitrate) injection system. Ultimate heal rejection is by The seismic evaluation program at Savannah River a once-lhrough cooling system which, under shutdown closely paralleled the (Jeneric Implementation conditions, can supply water by gravity feed. Procedure (GIP) generated for the Seismic Qualification Utilities Group (SQUG). The GJP The SRS reactors were designed to both the 1946 (Reference 1) documents the experience based Uniform Building Code (UBC) and to a 1.951 UBC methodology developed to resolve USI A-46 and has supplement specifically generated for SRS. This been reviewed and approved by the NRC (Reference .supplement includes a 0.1 g seismic requirement and a 2). While the use of earthquake experience data is the 1000 psf blast overpressure requirement. These design primary thrust of the SRS seismic evaluation program, criteria generally applied only to the civil structures there are several other innovations which also improve with very few of the mechanical and electrical systems traditional seismic qualification techniques. These receiving an initial seismic or blast design. Subsequent innovations include the assimilation and use of seismic to the original plant design, some of the piping and shake table data in a generic manner; the development equipment were evaluated for seismic loading (and of simplified analytical tools and realistic criteria for upgraded if required), but the majority of the equipment anchorage; and the development of realistic equipment remained without a seismic evaluation. criteria for the generation of seismic demand (i.e. floor response spectra). To address the impact that current seismic criteria might have on the SRS reactors, a seismic evaluation The Savannah River Seismic Evaluation Program program for these three reactors was initiated in 1988. represents an expansion of the SQUG GJP in several The scope of work covered in this evaluation was based areas. Programalic changes were incorporated to add on NRC criteria as outlined in NRC NUREG 1211 credibility and defcndability to the SRS program. In "Regulatory Analysis for Resolution of Unresolved addition, several technical changes were added to the Safety Issue A-46, Seismic Qualification of Equipment SQUG procedure to address conditions unique to in Operating Plants" (Reference 12). These criteria Savannah River. These changes included: include all piping and equipment necessary to safely shut down the reactor for 72 hours following a Design o Site specific procedures were developed to be Basis Earthquake (DBE) occurring at 100% reactor used in conjunction with the SQUG procedures. power. Systems included in the initial study were selected portions of the Process Water (PW), Cooling o A Systems Engineer was made a member of each Water (CW), Supplementary Safety (shutdown) System walkdown team. (SSS), the D.C. Power System that drives the PW circulating pumps, and associated control and o More detailed documentation was provided, instrumentation components. Also included were facilities to provide a means of returning spilled o Overview by a Technical Review Team (TRT). process water to the system and pumps and piping for flood control in the event of failure of non safety o Independent reviews of procedures and related piping. The seismic characterization of applications. equipment and raceways at SRS was accomplished using the methodology which the NRC has endorsed to o Additional verification for shell type expansion resolve Unresolved Safety Issue A-46 for commercial anchors. nuclear power facilities. The major portion of this A-46 resolution methodology is based on experience o A special investigative program was initialed to data that was developed by EQE Engineering (EQE) develop allowable loads for lead shield and the Seismic Qualification Utilities Group (SQUG). expansion anchors. A similar experience-based seismic evaluation methodology was adopted in this SRS seismic The TRT consisted of senior specialists from evaluation to address the piping within the systems Westinghouse Savannah River Co. (formerly DuPont), being addressed. EQE, and United Engineers and Conslructors (UE&C) whose function was to develop ihc implementation and acceptance criteria for the project and to resolve

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

36 technical issues. The TRT also conducted a program identified the safe shutdown path, system boundaries review and a walkdown of the safe shutdown and piping and/or equipment that musl function equipment to ensure that the procedures were being during or after a DBE in order to shut down the applied correctly and to verify uniformity between reactor and maintain it in a safe shutdown condition walkdown groups. The two independent reviews were for 72 hours, conducted by SQUG steering group members who pioneered the A-46 resolution methodology, SQUG In identifying the safe shutdown path and equipment, consultants, and by the Senior Seismic Review and the following conditions were assumed: Advisory Panel (SSRAP) who serve as technical advisors to SQUG (Reference 3). These two reviews o Offsitc power may not be available for up to provided valuable insight relative to the SRS 72 hours following the earthquake. procedures and to the judgments being made by the SRS walkdown teams. o No other extraordinary events or accidents (e.g., LOCAs, fires, floods, extreme winds, The SRS seismic evaluation procedure is documented sabotage) are postulated to occur other than in SRS reports (References 4 & 11). These procedures the earthquake itself and loss of offsite include the floor response spectrum and instructions to power. walkdown personnel regarding documentation requirements and deviations from the G1P (such as o Redundant systems and/or equipment must be reduction of anchor bolt capacity because of lower qualified and be available to ensure that concrete strength). The procedure requires a one-week failure of a single item docs not compromise training program for each participant in the seismic the ability to meet the shutdown criteria. evaluation project coupled with independent study in each of the technical areas. This training is considered o Instrumentation and controls necessary to to be necessary due to the high degree of judgment and monitor critical conditions and ensure experience required when applying the subject seismic appropriate corrective action is taken was evaluation methodology. The primary sections of the included. procedures arc: o Where operator actions arc relied upon to o Identification of Safe Shutdown Equipment achieve and maintain safe shutdown, the time required for the action was considered. o Plant Walkdown PLANT WALKDOWN o Seismic Demand The responsibilities and qualifications of those o Equipment Review individuals participating in the SRS walkdowns follow those outlined in the GIP. Each walkdown team o Anchorage Evaluation includes at least two seismic capability engineers. It may also include systems/plant operations engineers, o Cable Tray, Conduit and Piping Review plant maintenance personnel and a relay engineer if the relay walkdown is done concurrently. The seismic o Seismic Spacial System Interaction capability engineers arc expected to exercise Assessment engineering judgment during the walkdown and to apply the methodology developed to resolve USI A-46. A brief description of each of these areas is given Collectively the walkdown learn must have knowledge below together with a description of the results found of the performance of equipment and structures in past at Savannah River. earthquakes, nuclear plant walkdown experience, knowledge of nuclear design standards, and expertise IDENTIFICATION OF SAFE SHUTDOWN in the seismic design, seismic analysis and vibration EQUIPMENT test qualification practices relative to nuclear plant equipment. Each individual seismic capability engineer SRS personnel that were knowledgeable of the system must possess a portion of the collective experience functions and requirements developed documents (hat discussed above, they musl be a degrced engineer or

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

37 equivalent, and they must have at least five years of applicable to equipment located in the applicable nuclear experience. At least one member of facility wilhin 40 feet above grade each walkdown team must also be a registered elevation. professional engineer. The SRS project hud four walkdown teams (one per major system) normally with o Floor response spectra at equipment four to five members per team. locations arc shown to be enveloped by a spectrum equal to 1.5 limes the SQUG The purpose of the plant walkdown is to examine Bounding Spectrum. each component within the safe-shutdown list and to evaluate parameters specific to the as-built condition SOUG Bounding Spectrum. In order to verify the of the item. There are four basic criteria that must be applicability of the earthquake experience data base satisfied in order to verify the seismic adequacy of for seismic qualification of equipment, SSRAP has equipment during the walkdown. The first criterion is developed and SQUG has adopted a generic seismic that the seismic capacity response spectra envelops the motion bounding spectrum (Reference 3). The purpose seismic demand spectrum curve over the frequency of this SQUG Bounding Spectrum is to compare the range of interest. The second criterion is that the potential seismic demand on equipment in the facility equipment being reviewed is similar to equipment in being evaluated with the estimated seismic demand the experience data base and that all specific caveats that similar equipment experienced in data base and inclusion rules associated with that equipment facilities subjected to earthquakes. For convenience, class arc satisfied. The third criterion is that the the SQUG Bounding Spectrum is expressed in terms of anchorage capacity, installation, and rigidity arc free-field ground motion at a facility rather than floor adequate for the seismic demand loads. The last response or equipment response. The SQUG Bounding criterion is that seismic interactions must not cause Spectrum represents approximately two-thirds of the equipment to fail to perform its safely-related estimated average free-field ground motion to which function. Each of these 4 areas which arc addressed on the data base equipment was actually subjected. The the walkdown arc described in the following sections. SQUG Bounding Spectrum is based on the frcc-ficld ground response spectra from four data base sites: SEISMIC DEMAND Sylmar Converter Station (1971 San Fernando), El Ccntro Steam Plant (1979 Imperial Valley), Pleasant Several aspects of seismic demand arc necessary to Valley Pumping Plant (1983 Coalinga), and Llollco show applicability of and implement the SQUG Pumping Plant (1985 Chile). All had average peak methodology: ground accelerations greater than 0.4g. These earthquake response spectra were selected based on o Earthquake experience data base applicability earthquake characteristics (highest ground motion, duration, and frequency content), and presence and o Applicability of the Generic Equipment performance of representative equipment. The SQUG Ruggedncss Spectra (GERS) Bounding Spectrum is defined in terms of a 5% damped horizontal ground response spectrum. This spectrum o Anchorage demand bound is intended for comparison with the 5% damped horizontal design ground response spectrum for the To use the earthquake experience data base to facility to be evaluated. Hence, the earthquake demonstrate the seismic adequacy of equipment, the experience data base is demonstrated to be applicable design basis earthquake motions for the facility being with respect to seismic demand when the horizontal evaluated must be shown to be enveloped by estimates design ground response spectrum for the facility is less of the motion experienced by facilities in the than the SQUG Bounding Spectrum at the approximate earthquake experience data base. Two options exist to frequency of vibration of the equipment and all higher satisfy this criteria: frequencies. This comparison of ground response spectra is judged to be applicable for equipment o The facility design free-field ground mounted less than about 40 feet above grade and for motion as specified by design ground reasonably stiff structures. If equipment frequencies response spectra are enveloped by the are less than about 8 Hz, floor response spectra must be SQUG Bounding Spectrum. The earthquake compared with 1.5 times the SQUG Bounding Spectrum experience data base approach is then as discussed below. Reference 3 contains an expanded

Second DOE Natural Phenomena Hazards Mitigation Contcrvnvv - 1089

38 discussion of aspects of the derivation of the bounding the earthquake experience data base up to 40 feel spectrum. above grade. Floor response spectra at higher elevations are being reviewed and will extend Floor Response Spectra. An alternative to comparing applicability of the earthquake experience data base. the SQUG Bounding Spectrum with the horizontal design ground motion is to compare horizontal floor EQUIPMENT EVALUATION response spectra to 1.5 times the SQUG Bounding Spectrum. This alternative may always be invoked and The guidelines used for verification of the seismic must be invoked when the equipment item is located adequacy of electrical and mechanical equipment at greater than 40 feet above grade, has a nalural frequency less than about 8 Hz, or the ground response SRS closely followed the GIP developed by SQUG spectra criteria fails. Floor response spectra to be (Reference 1). Slight deviations from the GIP compared should be realistic, i.e. median-centered equipment review sections by the SRS specific conditional on the occurrence of the design ground procedure (Reference 4) were necessary mainly due to motion. Conservatively calculated in-structurc spectra site specific quality documentation requirements. The may be compared but arc not required and this extra GIP provides the technical approach and generic conservatism should be recognized as unnecessary. procedures for operating nuclear plants to evaluate seismic adequacy of mechanical and electrical Generic Equipment Ruggcdness Spectra (GERS). The equipment needed to achieve a safe shutdown condilion applicability of GERS from the excitation standpoint is following a safe shutdown earthquake. related to floor response spectra at the equipment support location (Reference 6), Conservatively These equipment evaluation guidelines present an calculated floor response spectra may be directly alternative qualification approach, based primarily on compared with GERS, If median-centered or realistic the performance of equipment in past earthquakes. floor response spectra are calculated, they need to be Reviews of equipment experience data from multiplied by 1.5 for comparison purposes which conventional power plants and industrial facilities ensures conservatism in the demand specification. subject to past earthquakes shows that with established inclusion rules and caveats, and below certain seismic Anchorage Demand. Anchorage evaluations arc motion bounds, it is unnecessary to perform explicit performed by comparing seismic demand on the anchor seismic qualification in order to demonstrate to its seismic capacity. To evaluate the structural functionality of many classes of equipment following integrity of the equipment anchorage and its load path, an earthquake. The guidelines address twenty classes applied loads arc derived from response spectral of equipment, established by their representation in the accelerations at the equipment support location. If seismic experience data base and also by their median-centered floor response spectra are the basis of similarity to nuclear plant equipment including the evaluation, a load factor of 1.25 is applied to construction, operation, capacity, and application introduce conservatism. (Reference 5). Where higher equipment seismic capacity or function during an earthquake needs to be Vertical Components of Motion. Although inclusion demonstrated, the review guidelines include generic criteria for the earthquake experience data base and criteria based on shake-table test results (Reference 6). GERS arc cast in the form of horizontal ground and floor response spectra, vertical motions need to be The scope of equipment covered by the procedure included in all quantitative assessments such as includes active mechanical and electrical equipment anchorage evaluations. such as motor control centers, swilchgcar, transformers, distribution panels, pumps, valves, HVAC equipment, SRS Reactors Evaluation. Evaluation of the SRS K- batteries and their racks, engine and motor generators, and L- reactors was performed for design ground and instrumentation and control panels, cabinets, and response spectra defined by US NRC Regulatory Guide racks. Relays are also reviewed to determine if plant R.G. 1.60 anchored to 0.2g PGA. A comparison of the safe shutdown systems could be adversely affected by SQUG Bounding Spectrum to R. G. 1.60 is shown in relay (contact) chatter as a result of an earthquake. Figure 1. Figure 2 shows selected floor response The equipment review guidelines evaluate the seismic spectra comparisons with 1.5 limes the SQUG Bounding capacity versus demand of a component item, and Spectrum. These comparisons show the applicability of include a detailed in-plant evaluation of the component

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

39 structural integrity, anchorage load path, and anchorage capacity evaluation guidelines include an functionality parameters. inspection checklist for in-planl review and assessment of an anchor's as-installed condition and other critical The results from the SRS equipment evaluations anchorage parameters, to assist in fastener strength were in general confirmatory in nature, demonstrating determination. Simple equivalent static analysis seismic ruggedness for most equipment. The isolated methods are then used to determine the equipment cases of outliers were generally associated with unique, anchorage lateral acceleration level capacity. This is plant specific mounting details such as customized defined as the horizontal load that when applied at the frame supports and vibration isolation support details equipment component center of mass will cause the which required further detailed analysis. There was anchorage to reach its assigned strength (which one case of an outlier due to a deep well casing on a includes appropriate factors of safety). vertical pump which exceeded the length of those in the experience data base; additional study verified its For the SRS seismic review implementations, plant seismic adequacy. Further evaluation was also specific anchorage evaluation guidelines had to be required to assess the seismic ruggedness of the cast developed. The SRS guidelines differed from the EPRI iron material of yokes of isolated air operated valves methodology (Reference 7) due to unique plant-specific in the review scope. The data packages assembled conditions, and also included an in-plant inspection during the evaluations proved to be extremely helpful guideline refinement whose necessity became evident as throughout the review process due to their the seismic reviews progressed. The SRS specific completeness and clarity. Although the documentation conditions not covered by the EPRI anchorage procedure at SRS exceeded the minimum requirements evaluation guidelines include lower strength concrete outlined in the GIP, the added effort proved beneficial than considered by the EPRI study, and lead sleeve in the long run. expansion-type concrete anchor bolts.

ANCHORAGE Anchor bolt capacities for cast-in-placc bolts (covered by the EPRI guidelines) were estimated by Seismic experience data and shake table tests have reducing the EPRI capacities which were based on demonstrated that adequate anchorage of equipment higher strength concrete. For expansion anchor bolts and distribution system installations is a critical of the types covered by the EPRI study, capacities parameter for component survivability during strong were based on manufacturers' test-data for lower motion earthquakes. The Electric Power Research strength concrete as stated in the original SRS Institute (EPRI) conducted a detailed program to (DuPont) design standards, with factors of safety develop simple and effective anchorage evaluation consistent with the EPRI study recommendations. For guidelines (Reference 7) to support the SQUG efforts the lead sleeve expansion anchors, the SRS embarked towards resolution of USIA-46. The EPRI anchorage upon a testing program of abandoned lead sleeve evaluation guidelines were developed to provide a anchors to establish ultimate capacities. Allowable consistent and cost effective manner for assessing the "generic" capacities for the seismic reviews utilized seismic capacity of equipment anchorage, in appropriate factors of safely consistent with the EPRI conjunction with developing technical justification for recommendations. Due to the generally low capacities elimination of unnecessary sources of conservatism in that resulted from the large variance in ultimate the anchorage evaluation procedure. These anchorage capacities from the SRS lead shell anchor testing criteria arc considered applicable to evaluation of program, another test phase was conducted to develop existing anchorages as well as for upgrading or bolt-specific proof torque test relationships versus designing new anchorages. allowable capacity on an as-nccded basis. Using this relationship, bolt-specific torque proof load tests were The EPRI anchorage guidelines address several types conducted where seismic demand exceeded the generic of fasteners including expansion anchor bolts, casl-in- lead shell anchor capacities. Bolts failing the proof place bolts, and welding to embedded or exposed steel. torque load test were replaced, typically with The fastener strength criteria were developed by conventional non-shell type expansion anchors. compiling and analyzing a vast quantity of lest data. Criteria for other fastener types were adopted from existing codes and standards with appropriate elimination of unnecessary conservatism. The

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

40 The EPRI guidelines utilize dcluilud in-phtnl The SRS evaluation guidelines for electrical cable inspection requirements fur verifying expansion anchor and conduit raceway systems included screening installation adequacy. The inspection guidelines criteria and procedures for verification of seismic include a lightness check that ensures expansion anchor adequacy. Seismic ruggedncss of raceway systems was sel. During the SRS seismic reviews, it was determined defined as protecting electrical cable function and that ihe tightness checks for shell-type anchors were at maintaining overhead support. The evaluation times meaningless as the bolt tightening was simply guidelines address seismic ruggedness by walkdowa forcing the concrete insert shell up against the guidelines and limited analytical review guidelines. equipment component base plate. As a refinement to The walkdown guidelines are for in-planl seismic the EPRI procedure, shell type expansion anchor ruggedness reviews which have two purposes. First, the inspection guidelines for ihc SRS adopted a check in-plant review screens raceway systems to check that requiring removal of hold-down bolts from the they arc representative of the experience data base that concrete anchor expansion sleeve (after initial forms the basis for the guidelines. The screening lightening) to verify that a gap existed between the top guidelines also check for certain details that may lead of the anchor shell and the bottom of the component to undesirable seismic performance as shown by past base plate. experience. Second, the in-plant review selects worst- case, bounding samples of as-installed raceway system Of the hundreds of expansion anchors reviewed at supports for limited analytical review. The limited Savannah River, seismic adequacy was verified for analytical review guidelines check that the bounding about 80-90 percent using the SRS specific guidelines. sample supports arc as rugged as those thai have been The majority of outlier anchors identified were due to shown to perform well by past earthquake experience. improperly installed shell type anchors and low The checks assess the raceway support dead load capacity lead shell expansion anchors. Bolt integrity, ductility, vertical capacity, and lateral replacement for ihc outlier conditions typically was not capacity. difficult and considerably increased component seismic margin. Based on the SRS reviews, recommendations The raceway evaluations al SRS generally were made to the SOUG program to adopt the demonstrated acceptability of the as-installed developed revisions to the EPRI bolt in-plant configurations. Identified outliers were associated inspection procedure. with isolated conditions of unique anchorage details requiring additional study, and isolated cases of CONDUIT AND CABLE TRAYS EVALUATION expansion anchors nol fully set. The raceway systems at SRS were observed to be well constructed, lightly Plant specific evaluation guidelines were developed loaded with short spans, and wilh high capacity for seismic evaluation of conduit and cable trays at support systems for seismic load. The rigid support SRS, based on a conservative interpretation of the systems were evaluated lo have high seismic margins current efforts of the SQUG raceway evaluation inherent with their design. The many flexible support guideline development program al the lime the SRS systems were evaluated as acceptable due to their high reviews commenced (Reference 8). The approach is vertical capacity and ductility for lateral seismic based on seismic experience data, shake table test data, loading, consistent with the experience based criteria. component test dala, and bounding analyses. Seismic experience data have shown that cable tray and PIPING EVALUATIONS conduit systems consistently perform well at conventional power and industrial facilities subject lo Seismic evaluation of piping systems at SRS included past strong-motion earthquakes, even though ihe review of systems with and without available dynamic systems are typically not designed for earthquakes. A analyses. Certain piping systems had been previously number of shake table lests on portions of cable tray dynamically analyzed to evaluate their ability to and conduit systems confirm the observations from past survive the design basis earthquake. For these systems, earthquakes and demonstrate that typical the scope of this evaluation was limited to assuring configurations perform well under repeated high level thai the piping dynamic analyses were available, and seismic input test motion on Ihc order of lg zero period thai the analysis configurations were representative of acceleration the as-installed conditions. The piping system boundary included the piping configuration from component to component or to other anchor points.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

41 Other interconnections to the subject piping were SYSTEM INTERACTIONS included from the connection to the first anchor point or component. Anchors were denned as the point at System interaction is characterized by the intcr- which the pipe is restrained in three directions and system dependencies that may result in potential three rotations, or through a scries of supports which harmful effects of initialing events to required safety provide equivalent restraint of the pipe, systems. The seismic system interaction issue is addressed in three NRC documents: Regulatory Guide For piping systems lacking existing available 1.29, and Unresolved Safety Issues (USI) A-17 and dynamic analysis, the reviews were based primarily on A-46. NRC Regulatory Guide 1.29 (Reference 10) seismic experience data and in part on engineering stipulates "Those portions of structures, systems, or judgment and simple calculations. Piping systems in components whose continued function is not required this category were reviewed for identified favorable but whose failure could reduce the functioning of any seismic performance installation attributes which are plant Category I component to an unacceptable safely typical of piping systems found in conventional plants level should be designed and constructed so that the (i.e., process, industrial and power plants). A Safe Shutdown Earthquake would not cause such significant number of piping systems in conventional failure". NRC USI A-17 defines the seismic system plants, which have been subject to earthquakes with interaction concern and describes five separate areas peak ground motions in excess of the SRS design basis relative to seismic spacial interactions: Category II earthquake, have been reviewed by utility sponsored over Category I (II/I) failure and falling conditions, organizations such as SQUG and EPRI (Reference 9). seismic deflection/impact, differential motion induced Also, piping configurations representative of those failures, seismic induced spray and flooding, and found in conventional plants have been tested as part seismic induced fires. Some of these seismic system of seismic qualification activities in commercial interaction issues within USI A-17 arc being addressed nuclear power plants. and resolved as a part of the USI A-46 program resolution. The SRS program has committed lo follow The comprehensive experience and test data indicate the USI A-46 scope and will address the first three of that welded steel piping can withstand earthquakes these areas: II/I failure and fulling (up to the first with ground motions exceeding O.Sg peak ground support), seismic deflection/impact, and differential acceleration. The few reported failures are attributed displacement failures. to conditions that are considered inadequate in well designed conventional systems, and the effect of large The II/I seismic system interaction issue relates to seismic induced displacements on the same inadequate the effects of a non-seism ically designed component features. Also, failures have been due to certain other (Category II) failing during an earthquake and design aspects which arc particularly sensitive to subsequently falling or sliding into a Category I safely differential building motions. Evaluation of the SRS related component. The seismic deflection interaction piping without dynamic analysis was performed by issue relates to potential impacts between Category II assuring that the piping systems do not contain design and Category I components due lo relative details otherwise considered inadequate in conventional displacements during the earthquake. The differential plants, and critical attributes that arc sensitive to large displacement interaction issue relates lo distortion- seismic induced movements. The piping system reviews controlled behavior which occurs when a componcnl is at SRS identified a few configurations requiring constrained and thereby forced lo undergo the same additional study due to lack of sufficient lateral movement as that of a major structure (e.g. damage support strength, and also limited cases requiring resulting from components crossing seismic separations bracing to preclude potential problems associated with between buildings or systems, such as piping connected large seismic induced motion. Other isolated problems to equipment lhat experience large seismic movements). were observed due to as-instalicd conditions of expansion anchor bolts to the building structure. The primary methods utilized on ihc SRS project to Seismic adequacy was verified for several of the as- evaluate seismic spacial interaction concerns included installed configurations and no "generic" SRS issues engineering judgment, earthquake experience data and related to seismic adequacy of the piping systems were simple calculations. Engineering judgment is useful for identified. obvious situations which arc judged rot to be a credible concern. Simple seismic calculations can be useful to calibrate an engineer's judgment. The

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

42 earthquake experience database developed by EQE reviews concluded that the SRS program was under the sponsorship of SQUG contains detailed comprehensive, technically correct and conducted in ;i seismic data on the performance of a wide variety of professional manner. Comments and suggestions from components in past earthquakes. This database is very these two reviews were incorporated where deemed useful in evaluating the credibility of postulated appropriate and have resulted in a stronger program at potential interactions through a comparison to the SRS. Due to the fact that the SRS program commenced performance of similar configurations in past before the SQUG methodology was completed, many of earthquakes. Specific objectives of the experience daty the open issues to the SQUG program were resolved as approach include the documentation of the most a result of the SRS study. In addition, the SRS study common sources of seismic damage, identification of provided new direction to the requirements of SQUG the threshold of seismic motion corresponding to guidelines in areas such as documentation, anchorage various types of damage, and determination of inspection, independent review and piping evaluation. installations that arc typically undamaged by Many of these SRS procedures will prove helpful to all earthquakes in facilities that arc representative of of the older reactors (DOE and Commercial) which will critical nuclear power plant systems. be undergoing seismic reviews in the future.

There were relatively few credible interactions REFERENCES identified during the SRS walkdowns, and many of those identified were resolved with minimal effort. 1. Bishop, Cook, Purccll and Reynolds; EQE The majority of the interactions identified related to Incorporated; MPR Associates, Incorporated; unanchorcd equipment being in close proximity to Stevenson and Associates; URS Corporation/John safety related equipment. A. Blumc and Associates, Engineers; SQUG Report, "Generic Implementation Procedures (GIP) for CONCLUSIONS Seismic Verification of Nuclear Plant Equipment," Rev., 1 December 1988. The Savannah River Seismic Characterization Program reviewed the seismic adequacy of the safe 2. NRC "Generic Safely Evaluation Report on SQUG shutdown systems for the K and L production reactors. Generic Implementation Procedure for The methodology utilized in that review is described in Implementation of USI A-46," forwarded to SQUG this paper and was found to be cost-effective, by NRC letter dated July 29,1988. expedient, thorough and defendablc. In general, most of the components within the safe-shutdown systems 3. Senior Seismic Review and Advisory Panel were found to possess adequate inherent capacity to (SSRAP), "Seismic Ruggcdncss of Equipment," survive the 0.2 g design basis earthquake at SRS. prepared for SQUG, in collaboration with the US Retrofits were recommended and are currently being NRC. implemented in those cases where seismic adequacy could not be established by more detailed analyses. 4. Savannah River Report RTR 2582 "Technical The principal areas where retrofits were required Basis, Procedures, and Guidelines for Seismic included equipment anchorage (particularly where the Characterization of SRS Reactors" 1988. lead shield expansion anchors existed), relays, relocation of proximate non-safety components (system 5. EQE Report, "Summary of the Seismic Adequacy interaction concerns) and structural load path concerns. of Twenty Classes of Equipment Required for These findings arc consistent with the types of Safe Shutdown of Nuclear Plants," EQE Inc., San situations which have been found on SQUG trial plant Francisco, California August 1988 (Draft). walkdowns and on previous seismic margin studies and seismic risk assessments for commercial nuclear power 6. EPRI NP-5223, "Generic Seismic Ruggcdncss of plants. Nuclear Plant Equipment" ANCO Engineers, Inc., Culver City, California, May 1987. The SRS seismic evaluation program had two independent reviews in addition to DOE's review. 7. EPRI NP-5228, "Seismic Verification of Nuclear These independent reviews were conducted by the Plant Equipment Anchorage, Volumes 1 and 2," Seismic Qualifications Utility Group and by the Senior URS Corporation/John A. Blumc & Associates, Seismic Review and Advisory Panel. Both of these Engineers, San Francisco, California, May 1987.

Second DDK \ntural Phrrmmrnrt Ha'nrds Mitigation Confrrvnrr

43 8. EQE Report, "Cable Tray and Conduit System 11. Savannah River Special Procedure, SP 2446, Seismic Evaluation Guidelines," July 21,1988 "Seismic Characterization of Selected K, L (Draft). Reactors Safety Systems", August 29,1988.

9. EPRINP-5617, "Recommended Piping Seismic- 12. United States Nuclear Regulatory Commission, Adcquacy Criteria Based on Performance NUREG 1211, "Regulatory Analysis for During and After Earthquakes," Volumes 1 and Resolution of Unresolved Safety Issue A-46, 2, EQE Incorporated, San Francisco, California, Seismic Qualification of Equipment in January 1988. Operating Plants, February, 1987.

10. United States Nuclear Regulatory Commission, Acknowledgement Regulatory Guide 1.29, "Seismic Design This document was prepared for the U.S. Department of Energy under Contract No. DE-ACO9-88SR18035.

Tttquincy IHzl Satti: «issue Scunlir. Acctitf*t;:ns

Figure 1 COMPARISON OF SQLJG BOUNDING SPECTRUM WITH R. G. 1.60

/ . \ 1 / 1 \ <\ i ' t - ,i #^ 8 i S -* / ^\ * 1 / 0.*+

O.J-t-

a.o. .

Petquincy Frequency IHzl

tlmtion -;0.0 !!><4:»n 3.9 i..»«tion -.5.0

Figure 2 COMPARISON OF FLOOR RESPONSE SPECTRA Figure 2 (D) LONGITUDINAL WITH l.5xSQUG BOUNDING SPECTRUM (A) TRANSVERSE

Second DOE Natural Phenomena Hazards Mitigation Conference - 19S9

44 Session 3 Seismic Design

Second DOE Natural Phenomena Hazards Mitigations Conference - 1989 BASE ISOLATION FOR NUCLEAR POWER AND NUCLEAR MATERIAL FACILITIES

Report on the Status of the Practice

J.M. Eidinger Impell Corporation

F. F. Tajirian Bechtel National Incorporated

C.A. Kircher J.R. Benjamin & Associates, Inc.

N. Vaidya P.C. Rizzo Associates

M. Constantinou S.U.N.Y. Buffalo

J. M. Kelly U.C. Berkeley

D. Ovadia Pacific Gas and Electric

R. Seidensticker Argonne National Laboratory

ABSTRACT

This report serves to document the status of the practice for the use of base isolation systems in the design and construction of nuclear power and nuclear material facilities.

The report first describes past and current (1989) applications of base isolation in nuclear facilities. The report then provides a brief discussion of non- nuclear applications. Finally, the report summarizes the status of known base-isolation codes and standards.

INTRODUCTION Europe, U.S.A., and Japan, and current code efforts The primary purpose of this report is to bring to related to base isolation. the attention of the Department of Energy (DOE) To put the topic into perspective.the following list community the breadth of base-isolation efforts in presents the past and current efforts towards base the nuclear industry. isolation of nuclear facilities. Seismic isolation of nuclear structures is a design approach which has been receiving increased Koeberg. Two unit PWR (2x900 MWe) plant in attention in recent years. This paper reviews the South Africa. Operational since 1983. Uses rein- background of base isolated structures, current pro- forced elastomer bearing pads with friction plates grams which are considering seismic isolation in for horizontal isolation.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

46 S- Two twin unit PWRs in France. Opera- years. For example, the 1983 SMIRT had one ses- tional since 1985. Uses reinforced elastomer bear- sion dedicated to seismic isolation, with a special ing pads for horizontal isolation. volume of the Journal of Nuclear Engineering and Superphenix g. 1500 MWe LMFBR in France. Fi- Design [Vol. 84, No. 3,1985] published which pre- nal design completed. Uses elastomer bearing pads sented a set of invitational papers on themes devel- and viscous dampers for horizontal isolation, and oped at this SMIRT session. Each subsequent steel springs and dampers for partial vertical isola- SMIRT has included further base isolation papers, tion. and the 1989 SMIRT devoted four sessions in the Karun River. Two unit PWR (2x900 MWe) de- main conference, and four additional sessions in a signed for Iran. Construction suspended in 1978. special Post-SMIRT Seminar to this topic. Uses reinforced elastomer bearings with friction The annual Pressure Vessel and Piping confer- plates for horizontal isolation. ences organized by ASME have also included sym- Le Carnet- 1400 MWe PWR in France. Prelimi- posia on seismic isolation. The first and second nary design completed; final design not initiated. Symposia on Seismic, Shock, and Vibration Isola- This concept is relegated to future program. Uses re- tion were held in 1987 and 1988. Sixteen and four- inforced elastomer bearings for horizontal isola- teen papers were presented at these two conferences, tion. respectively *,2. A third symposium held in 1989 Laguna Verde. 1300 MWe PWR proposed by Elec- had twenty four papers presented on this topic^. tricite de France for Mexico. Uses reinforced elas- To this date, no licensee in the United States has tomer bearings for horizontal isolation. Project approached the Nuclear Regulatory Commission cancelled. with a request to operate a complete nuclear plant LSPB. Large Scale Prototype Breeder. Conceptual built on a seismic isolation system. The reasons study, funded by EPRI and DOE, in 1984-85. that this is so are several, and among the primary PRISM. Power Reactor Inherently Safe Module. reasons are the following: Liquid Metal Reactor. 465 MWe. Ongoing study, • The base isolation technology has blossomed led by General Electric, Bechtel National, funded by only in the late 1970s and 1980s - and since 1978 no DOE. new reactor orders have been placed in the United £A£fi. Sodium Advanced Fast Reactor. 450 MWe. States Recent study, base isolation considered as an alter- • There are no NRC-approved guidelines, or nate design option, led by Rockwell International, codes, that define the design basis needs for nuclear Bechtel National, funded by DOE. SAFR project has plants on base isolation systems. Perceived diffi- been stopped in 1989. culties and delays in licensing may deter applica- MHTGR. Modular High Temperature Gas Cooled tion Reactor. Cooperative design, by General Atomic, • The cost savings (or dis-savings) of a base- Combustion Engineering, Bechtel National, funded isolated nuclear plant have not yet been quantified by DOE. 135MWe. or proven in practice for different sites; and such FBR. 1000 MWe advanced demonstration Fast cost savings (or dis-savings) are not widely recog- Breeder Reactor, to be designed in Japan. nized by design practitioners. STPP. Seismic Technology Program Plan, A pro- As of 1989, the need for new nuclear facilities gram (developed by ETEC for the DOE) to study presents the opportunity to examine the seismic iso- seismic issues for LMRs. Not yet funded. lation option. The potential for new nuclear facili- USSIRP. United States Seismic Isolation Research ties is evident for the following reasons: Program. A program (sponsored by the National • The Department of Energy (DOE) has deter- Science Foundation) to lead to implementation of mined the need for a new tritium producing reactor seismic isolation as a standard economic strategy. Not yet funded. • The DOE has an ongoing program to investigate new types of reactors, both for possible military EPRI/CRIEPI/CEGB. A joint program between or energy production purposes, which are considered EPRI (USA), CRIEPI (Japan) and CEGB (England) "safer" than those presently in operation in the to study seismic isolation systems for LMRs. United States CRIEPI A seven year research program by CRIEPI • The move toward standardization and a one- in seismic isolation, begun in 1987, Japan. stop licensing process improves the potential for Base isolation topics have become prevalent in seismic isolation as a means to make standardiza- nuclear plant oriented conferences over recent tion possible

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

47 • Utility owners in the United States are seeing sible for structural damage, resulting in signifi- their overcapacity rapidly eroding, due to the con- cant reductions in seismic loads. Thus the design tinued expansion of the economy; and losing exist- and qualification of equipment and piping becomes ing capacity due to retirement of older facilities, or a simpler task than today. Since the response of iso- excessive polluting facilities. Thus, they may in the lated structures is highly predictable, the risk of ac- near future begin to order new plants. cidents due to uncertainties in the input motions is For these reasons, an early review, evaluation reduced, safety margin is increased, and plant in- and development of seismic base isolation in the vestment protection is enhanced. If seismic design context of the US nuclear environment is important. criteria are upgraded, for example due to the discov- This report describes the background of base ery of new geotectonic conditions, the standard plant isolation for nuclear plants; describes many of the design would probably not have to be altered and nuclear facility base isolation designs either al- only the isolation system would need to be updated. ready implemented or in current design; presents a Due to the high cost of development of LMR de- brief status of non-nuclear applications; and dis- signs and limited available resources, interna- cusses current codes and guidelines. tional cooperation could be highly desirable, partic- Topics such as detailed design issues; experi- ularly for the nuclear steam supply system (NSSS) ence of actual base-isolated structures, or areas components. Seismic isolation would facilitate the needed for further development are beyond the scope development and application of an international of this status report. These important topics have standard design of the NSSS and would allow the been, and will continue to be described in the techni- decoupling of the NSSS design development, which cal literature, probably at an ever-increasing pace. is global in nature, from the balance of plant (BOP) design and licensing which is regional in nature. NUCLEAR PLANT BACKGROUND BOP designs can be developed by each country in To reduce capital costs so that future nuclear accordance with its own local requirements without plants are competitive with those using alternate impacting the standard NSSS design. This would sources of energy, large portions of the plant should enhance the opportunities of international collabo- be standardized. Furthermore, to gain public accep- ration in the development of the NSSS for LMR tance, these plants must be reliable and should have plants. several passive inherent features to provide public Several advancements in recent years are re- safety and plant investment protection. Seismic de- sponsible, for making seismic isolation a practical sign can play a major role in achieving a standard- alternative. These include the development of ized design which could accommodate varying highly reliable elastomeric compounds used in seismic conditions. One approach to standardiza- seismic bearings which are capable of supporting tion would be to design a plant using traditional large vertical loads and can accommodate large methods for a safe shutdown earthquake (SSE) horizontal deformations during the earthquake which envelopes the responses of 90% of existing without becoming unstable. Additionally, the devel- U.S. nuclear sites. This would lead to high seismic opment of high damping elastomers and other me- loads especially in components and equipment and chanical energy disgipaters to control resulting dis- would still exclude California sites and limit ex- placements in the isolators, has provided the oppor- porting potential of these plants to high seismic tunity to keep the response to manageable levels. countries. Liquid Metal Reactor (LMR) designs The development of verified computer programs, the which consist of thin walled vessels designed to ac- compilation of reliable test results of individual commodate large thermal transients under low op- seismic isolators under extreme loads, shake table erating pressures are more sensitive to seismic tests for evaluating system response, and validation loads and thus would be particularly penalized by of computer programs all add to our confidence in this approach. being able to predict the response. An alternative would be to seismically isolate the plant. Seismic isolation is a recent development CURRENT PROGRAMS - FRANCE that is gaining rapid worldwide acceptance in the France has succeeded in establishing an eco- commercial field4 and is being implemented in ad- nomically competitive nuclear program due to the vanced nuclear designs of the future ^. This ap- successful implementation of a standardized 900 proach decouples the structure from the components MWe PWR plant. This design is suitable for the of ground motion which are predominantly respon- majority of sites in France where seismic accelera-

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

48 tions are less than 0.2 g. When a plant is to be lo- upper plate which is embedded in the upper mat is cated at a site with higher seismicity, seismic isola- stainless steel. The plate combination provides a tion is specified to limit seismic loads in the plant to friction coefficient of 0.2. When the ground accel- the levels that the standard design can accommo- erations exceed the friction coefficient, slip occurs, date. The system was developed by Spie-Batignolles thus limiting the level of shear strains in the pads and Electricite de France**. The reactor, fuel, elec- and the forces in the building to the same level as trical, maintenance, and auxiliary buildings are that for moderate sites. A total of 2000 pads 70x70x10 placed on a common upper mat which is in turn sup- cm were used. ported on pedestals cast integrally with the lower The construction costs for this system were justi- mat with the isolators placed in between. The type of fied in that it allowed a standardized plant to be built isolator used depends on the site acceleration level. at a site with no (or little) additional costs for re- For sites with moderate seismicity, relatively thin design, strengthening and requalification of com- steel laminated elastomeric pads are used. ponents.

CRUAS KAR UN RIVER, IRAN The operational two twin-unit Cruas-Meysse The design of the Karun River nuclear power plants (total 4x900 MWe) are located in the Rhone plant incorporates the standard 2x900 MWe reactor Valley of France. The site SSE acceleration is 0.2 g. units. The plant layout is very similar to that at The plants are supported on 1800 pads, each measur- Koeberg. Construction permit for the plant was ing 50x50x6.5 cm '. The pads are similar to stan- granted in 1978. Construction activities at this site dard neoprene bridge bearing pads. They have were suspended in late 1978 after completion of the three layers of elastomer, each 13.5 mm thick rein- lower raft. The isolation bearings for this plant had forced with 3 mm-thick steel plates and 10 mm thick been fabricated and shipped to the site. top and bottom plates. An isolation system is used The seismic design criterion at the Karun River for this site since there is a probability of shallow site was defined by USNRC R.G. 1.60 response earthquakes of low magnitude (Richter 4 to 4.5) oc- spectra anchored to a horizontal peak ground accel- curring close to the site producing higher accelera- eration of 0.3g. The bearings are about 65 cm square tions and high frequency motion. The fixed base in plan and comprise six layers of elastomer. The frequency of the reactor building is roughly 4.5 Hz interior layers are 16 mm thick and the exterior and corresponds to the peak frequency of the antici- layers are 8 mm thick. The bearings are reinforced pated spectrum. With the pads, the frequency is re- by 5 mm thick steel shims. The total height of a duced to 1 Hz which significantly reduces the forces bearing is about 13 cm. Similar to the Koeberg de- on the structure and internal equipment. The max- sign the isolation system at Karun River includes a imum displacement capacity of the pads is only 5 friction interface to help dissipate the seismic en- cm, but due to the high frequency input, the antici- ergy and limit the response of the isolated nuclear pated displacement is only 2.6 cm. island. Including the lower raft, retaining wall, pedestals and bearings, the cost of the isolation sys- SUPERPHENIX-2 tem at Cruas is about 2% to 3% of the total civil works Seismic isolation has been incorporated in the cost. design of a large 1500 MWe liquid metal fast breeder reactor (LMFBR), the Superphenix 2. The KOEBERG entire nuclear island is isolated in the horizontal A nuclear power plant in South Africa has been direction using elastomeric pads and viscous built on an isolation system by the French construc- dampers and additionally, the reactor cavity is iso- tion company Framatome^. The Koeberg plant is lated in the vertical direction using steel springs based upon the French 2x900 MWe design. In South and viscous dampers. Special guides are imple- Africa, it is located at a site with higher seismicity mented between the reactor cavity and the reactor building to minimize rocking of the reactor. than that at Cruas, with a SSE level of 0.3g. For this higher seismicity, the base isolation The final design of Superphenix 2 has been completed. The construction of this plant awaits feed- system includes both elastomeric bearings and back and operating experience from Superphenix 1 friction plates. The friction plates are placed be- and an improvement in energy demand projec- tween the top of the pads and the upper mat. The tions. lower plate is made of a lead-bronze alloy and the

Second DOE Natural Phenomena Hazards Mitigation Conferenre - 1989 The PRISM LMR concept incorporates seismic iso- OTHER FRENCH PWR DESIGNS lation in the reference design to enhance plant Designs were also developed for a single 1300 safety margins, to support plant standardization, MWe PWR at Laguna Verde Mexico, and a two unit and to potentially reduce plant costs. The SAFR 1400 MWe PWR at Le Carnet in western France. LMR concept incorporated seismic isolation as an Two types of isolation systems were considered: alternate design: recently (1989), the SAFR project elastomeric pads with sliders, and ordinary pads was stopped. For MHTGR seismic isolation is be- with German GERB type viscous dampers in paral- ing studied for adapting the non-isolated reference lel to limit horizontal displacements. These de- design, developed for moderate seismic zones, to signs were not implemented. high seismic sites. Both the PRISM and MHTGR projects continue to be funded as of late 1989. OTHER FRENCH NUCLEAR FACILITY DE- SIGNS PRISM Other isolated nuclear structures include a three PRISM is a compact standardized LMR reactor story reinforced concrete radioactive waste process- installed in blocks consisting of three reactors per ing building supported on GAPEC type laminated 465 MWe power block***, it incorporates a horizontal elastomeric bearings *0 and three spent fuel storage isolation system to isolate the reactor module with pools at the La Hague reprocessing plant*!. its key safety functions of reactor shutdown, shutdown heat removal, and containment systems. The OTHER EUROPEAN DESIGNS small diameter of the PRISM vessel provides suffi- The current status of seismic isolation of cient intrinsic resistance in the vertical direction to LMFBRs in Europe was discussed in a Specialists' minimize amplifications in vertical ground mo- meeting on Fast Breeder Reactors. The results of tions and makes vertical isolation unnecessary. this meeting are summarized by Martelli^, The The total weight of these structures is approximately main conclusion was that for LMFBRs, seismic iso- 4500 tons. The isolation system selected consists of lation offers sufficient advantages to warrant fur- 20 high damping steel laminated elastomeric bear- ther development to resolve some of the outstanding ings. The bearings have a diameter of 132 cm and a technical problems. total height of 58.7 cm and consist of thirty layers of In England, British Nuclear Fuels have de- rubber 1.27 cm thick and 29 steel plates 0.32 cm signed a nuclear fuel reprocessing facility as a base thick. The bearing dimensions were selected to isolated structure. give a horizontal frequency of 0.75 Hz and a vertical frequency over 20 Hz. The entire isolated structure UNITED STATES is housed in an underground silo. Studies have been performed to review available The design SSE for PRISM is a maximum hori- isolation devices and to examine their applicability zontal and vertical acceleration of 0.3 g anchored to a design earthquake that envelopes the NRC Regu- to U.S. nuclear design*3j14 Interest in adapting seismic isolation to LMFBRs has gained interest: latory Guide 1.60 spectra. Options for siting in under the sponsorship of DOE and Electric Power higher seismic zones have been retained. Analyti- Research Institute (EPRI), studies were performed to cal results show that horizontal accelerations are apply isolation to the Large Scale Prototype Breeder substantially reduced in all the reactor components (LSPB) reactor plant. Several types of isolation with isolation *9. The peak spectral acceleration at the core support plate for 2 percent damping was re- systemsl°,16,17 were reviewed. These studies concluded that there were significant advantages to iso- duced from 16.5 g to 0.25 g. Furthermore, horizontal late LSPB. spectral peaks above 2 Hz are eliminated. Since completion of the LSPB studies, DOE has A series of quasi-static tests were performed at been supporting the development of three compact the University of California Earthquake Engineer- advanced reactor concepts, the first two are LMRs: ing Center (EERC) in Richmond on four half-scale the Power Reactor Inherently Safe Module (PRISM); bearings. Each bearing had a diameter of 66 cm the Sodium Advanced Fact Reactor (SAFR); and a and consisted of thirty alternating layers of rubber Modular High Temperature Gas Cooled Reactor 0.63 cm thick and 29 steel plates. The bearings were (MHTGR). The PRISM design has been selected as first tested to design conditions expected during an the U.S. reference Advanced Liquid Metal Reactor. SSE. The tests verified that the bearings are capable of undergoing several cycles of varying shear

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 strain without appreciable change to their stiffness Argonne National Laboratory (AND23 is work- or damping. Furthermore, the bearings have 10% ing with Shimizu of Japan on a joint LJ.S./Japanese damping or more for all applicable strains^. The program for testing PRISM type bearings in a bearings were then subjected to extreme loads re- unique facility built by Shimizu at Tohoku Univer- sulting in a maximum horizontal displacement of sity in Sendai, Japan. The facility consists of two 36.6 cm which corresponds to 200% shear strain in full-size three story buildings built side by side the rubber and which is three times larger than the identical in every way except one structure is seis- displacement computed for the SSE. The high per- mically isolated and the other is not^. ANL has formance margins of the bearings were thus furnished the elastomeric bearings to test the bear- demonstrated by the fact that the bearings were ca- ing performance under actual earthquakes and pable of sustaining displacements triple the SSE- during static and dynamic forced vibration tests. displacements without failure or damage^*. One of Comprehensive analysis of the building response is the bearings was subjected to a vertical load of 1800 also planned. The resulting information will pro- tons with no failure. vide correlations between laboratory test data ob- Similar half-scale bearings with bolted type tained at EERC and ETEC and field data. ANL's connection instead of dowelled connection have effort is funded by the National Science Foundation; been tested to compare the failure mechanism of the the program is expected to last two years. two systems. Tests in Japan have shown that bolted bearings can accommodate larger horizontal dis- SAFR placements than dowelled ones. The PRISM de- (The DOE is no longer funding further work on sign,which currently uses dowelled connections the SAFR design, The following description de- will be reassessed depending on the outcome of these scribes recently completed work). tests. The test series are repeated for quarter-scale The SAFR plant concept employs a 450 MWe pool and full-scale bearings to determine the effects of type LMR as its basic module. The reactor assembly scaling on bearing performance characteristics module is a standardized shop-fabricated unit that and failure modes. can be shipped to the plant site by barge for installa- The Energy Technology Engineering Center tion^. The SAFR plant ensures both public safety (ETEC) is in the process of assembling a large bear- and investment protection by means of a variety of ing test machine under DOE funding. The fixture passive and localized features. It incorporates in- will be ready for testing PRISM half-scale bearings herent capability for reactor shutdown and adequate similar to the ones tested at EERC. The main objec- core and spent fuel cooling under all circum- tive of these tests is to compare and quantify the ef- stances. The seismic design basis is a SSE level of fects of dynamic loading on bearing response. The 0.3 g. To enhance seismic margins, and to permit machine capacity was selected based on testing full- siting in regions of high seismicity, studies were scale PRISM bearings dynamically. It will be ca- performed to investigate the feasibility of isolating pable of testing a single bearing up to 152 cm in di- the SAFR reactor building. ameter and 91 cm high under a maximum vertical Three isolation concepts were considered: hori- load of 750 tons and a maximum horizontal load of zontal isolation of the entire nuclear island; a hy- 150 tons with a horizontal stroke of ±38 cm up to a brid system similar to the one proposed for Super- frequency of 0.75 Hz and is also capable of applying phenix 2 in which the building is isolated in the hor- a maximum static horizontal displacement of 63 izontal direction and the reactor cavity is isolated in cm. the vertical direction; and full three dimensional Other programs planned to qualify the PRISM isolation whereby the building is simultaneously seismic isolation design include: shake table per- isolated in both horizontal and vertical directions. formance tests modelling a large weight, which rep- After reviewing the merits of each concept the third resents PRISM's mass distribution, on a set of bear- concept was incorporated in an alternate to the ref- ings to evaluate system performance during real erence design. This uses a new type of isolation earthquakes and tc evaluate the effects of input with system which consists of low shape factor steel lam- strong long period energy. large scale shake table inated elastomeric bearings with high damping test, seismic hazard assessment, and the develop- which provide flexibility in both the horizontal and ment of design guidelines^ vertical directions. The total isolated weight is 39,000 tons. Approximately 100 bearings with a diameter of 107 cm and a total height of 42.6 cm are uti-

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

51 Uzed. The horiaontal frequency is 0.5 Hz and the vertical frequency between 4 and 5 Hz, or steel vertical frequency is around 3 Hz. In general, a springs and viscous dampers will also be studied. large amount of rocking will result in buildings supported on such bearings. However, the SAFR SEISMIC TECHNOLOGY PROGRAM PLAN building has a sufficiently low center of gravity and The Seismic Technology Program Plan (STPP) a wide base to limit rocking to acceptable levels. has been coordinated by ETEC27 for the DOE to min- This type of isolation system has not been previ- imize the impact of seismic design on advanced ously tested. Dynamic tests of prototype quarter- LMR development and to avoid costly seismic de- scale bearings have been performed at EERC to ex- sign problems associated with LWR experience. amine the feasibility of using the proposed bearings Specific goals of the STPP are to identify and to pri- for three-dimensional isolation of nuclear build- oritize research and development needs, and to pro- ings and to verify the applicability of existing de- vide the basis for a LMR seismic design guide. Five sign formulas to low shape factor bearings. Two research and development needs are identified, types of bearing connections were tested: a dowel with seismic isolation verification given the highest type, and a rigidly bolted type. Stability and failure priority. For seismic isolation, a six year program limits of the two designs are being assessed. Results costing $9.0 million is defined. The objectives of of these tests should be available in late 1989. this program would be to test several types of isolation systems, to develop and verify the necessary MHTGR tools to analyze isolated systems, to develop appro- The MHTGR advanced reactor concept is being priate seismic inputs including long period motion developed under a cooperative program involving effects and accounting for beyond design basis DOE, the utilities, and the nuclear industry. The events, collecting and analyzing performance data reactor capacity which is 135 MWe and the configu- of existing isolated structures and evaluating the ration selected provides a higher margin of safety integrated effects of earthquake characteristics on and investment protection than current generation the seismic risk to isolated plants. The results of reactors2**. The standard MHTGR reactor and this program will produce a validated seismic de- steam generator are enclosed in a concrete silo sign and analysis technology. which is fully embedded to minimize seismic loads. Depending upon whether additional improve- The standard design which does not include seis- ments are made to other areas of seismic technol- mic isolation was developed to envelope the seismic ogy, a base isolated plant is projected to save , versus conditions at 85 percent of U.S. nuclear sites using a a non-base-isolated plant, in a high seismic zone, maximum SSE of 0.3 g. An initial investigation about 5% to 10% of the total plant cost. into the feasibility of seismically isolating the MHTGR to extend available sites into areas of UNITED STATES SEISMIC ISOLATION higher seismicity was recently completed by Bechtel RESEARCH PROGRAM National Inc. For high seismic sites, such as along The USSIRP is a coordinated National Science the California coast, it is anticipated that the plant Foundation research program that will lead to the would be designed for a SSE of approximately 0.7 g. implementation of seismic isolation in the United A concept for horizontally isolating the reactor ves- States as a standard economic strategy, compatible sel and steam generator while minimizing the Im- with existing design codes, in five years time. It is pact on the reference plant layout was developed. intended to apply to non-nuclear construction, but The isolated structure is supported on 42 high damp- will likely produce results useful to the nuclear in- ing elastomeric bearings. The design was based on dustry. The program is not yet funded. a horizontal isolation frequency of 0.60 Hz. Unre- In principal, the program will be led by a techni- solved issues which need further investigation in- cal coordinator who will be assisted by four advisory clude radiation resistance and shielding of the panels, as follows: bearings, inspectability and replaceability of the • Executive Panel. Includes researchers within bearings, and design of systems such as the main the program, one from each research area. steam and feedwater pipes passive cooling ducts to • Consulting Engineers Panel. Includes prac- accommodate the large relative movements between ticing structural engincars actively engaged in the the isolated structure and the fixed silo. The feasi- design of base isolated structures. bility of three dimensional isolation using either elastomeric bearings as proposed in SAFR with a

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

52 • Users Group Panel, Includes experts from As demonstrated by the designs for Koeberg and companies and organizations that are identified as Cruas, one can save the existing design basis for a most probable users of the technology. plant, should a new source of seismicity be discov- • Materials Panel, Includes representatives of ered after the plant has been designed. The San base isolation hardware suppliers. Onofre Unit 1 plant is an example of the costs asso- The USSIRP is organized to have five research ciated with new-found "seismicity"at an operating groups that work as a coordinated program to pro- nuclear plant. This plant was originally designed duce the required products. The groups are: Ground to a 0.50g motion, using the design rules of the late Motion; Modeling and Dynamic Response; Design 1960s. Later, Units 2 and 3 were built alongside Unit Criteria; Materials; Economics. 1, and the new geotechnical investigations led to the increased seismic design basis on Unit 1 to 0.67g. EPRI/CRIEPI/CEGB SEISMIC ISOLATION PRO- Approximately $180 million was spent on redesign GRAM and construction to bring Unit 1 up to 0.67g; this EPRI and CRIEPI have been collaborating in a amount excludes any lost generating capacity costs joint program to evaluate the technical feasibility of of a forced 23 month outage while new hardware was selected seismic isolation systems and their appli- designed and installed. If this plant had used a cability in the design of LMRs^S. Available seis- seismic isolation system, then the increase in seismic isolation devices have been evaluated and can- micity could have been adjusted for by altering the didate systems have been selected. Reduced and structural characteristics of the isolation system. full scale tests of these isolators and shake table tests Other recent examples of high seismic design are planned. These will provide confirmation of costs include the Comanche Peak station - whose final design input for an isolated plant design. SSE motion is 0.12 g, yet has had to undergo a multi- The first phase of this program consisted of test- year seismic re-verification process costing in the ing half scale elastomeric bearings with lead many hundreds of millions of dollars, to satisfy the plugs^9 to confirm performance characteristics to plant's design basis. More examples are the seven investigate failure modes and to verify restoring Units owned by TVA (Browns Ferry 1,2 and 3, Watts capabilities. The bearings tested measured Bar 1 and 2 and Sequoyah 1 and 2), each of which has 25.4x25.4x6.5 cm and were manufactured in the recently undergone seismic upgrade programs to U.S. according to specifications provided by Burns satisfy design basis issues. Multi-million dollar and Roe, Inc., the contractor assisting EPRI in the seismic upgrade programs have occurred at almost program. U.K's Central Electric Generating every operating plant in the United States. For all Board (CEGB) has joined the program in its second these existing plants, seismic upgrades have proved phase. A second isolation system consisting of lam- to be expensive, chiefly due to the large amounts of inated elastomeric pads 15x15x7 cm and GERB type commodities (piping, raceways, anchorages, etc.) within these facilities which need to be checked un- viscous dampers provided by CEGB are being tested der new or altered seismic loads. Little of this by CRIEPI. After completion of these tests, full scale would be needed in a base-isolated plant, as these elements will be tested. commodities see very small seismic loads in the first place, and any increase in site seismicity, or NUCLEAR PLANT COSTS FOR SEISMIC DESIGN validation needs of design bases, would use up little AND UPGRADE of the large margin inherent in these commodities. One driving force to consider base isolation is the excessive costs and delays associated with the seismic design of nuclear plant facilities. These JAPAN costs and delays have been attributed to overly con- Interest in seismic isolation in Japan has in- servative seismic design requirements and proce- creased significantly in recent years. Research dures that have been adopted in a rapidly evolving and development in isolation is being carried out by and increasingly stringent regulatory environ- several of the major construction companies which ment. Resulting plant cost increases in excess of have to date implemented isolation in over 30 build- S300 million have been documented*^, and approx- ings. These companies have decided that seismic isolation is superior to conventional seismic design imately 40% of current License Event Reports filed and can give a company a competitive edge in the with the NRC are related to seismic design issues**!. construction industry. Several universities and

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 government agencies are also participating in dif- Tadotsu shake table; construction and observation ferent research programs. of the response of a large scale isolated nuclear One of the reasons for the rapid development of building model; observation and cataloging of the this technology is the availability of state of the art response of all isolated buildings in Japan during bearing test machines at several of the companies earthquakes; identification of a set of design involved in seismic isolation development. Most earthquakes with long period components; and the notable among them is Kajima's machine which development of design guidelines for isolated nu- was brought to operation in 1988. This unique ma- clear systems. chine is capable of applying independent horizontal Different concepts of isolation are being consid- loads along two perpendicular axes in real time ered. This includes total base isolation of the nu- while simultaneously maintaining a vertical load. clear island building (NIB) or isolation of the reac- The availability of several large shake tables has tor cavity in the horizontal direction (2-D) or in the allowed the promotion of seismic isolation to the pub- vertical plus horizontal directions (3-D), or hybrid lic by comparing the performance of non-isolated isolation where the NIB is isolated in the horizontal structural models with isolated models. The largest direction and the reactor cavity in the vertical direc- shake table in the world is at Tadotsu and is 15x15 m tion. It is recognized, however, that vertical isola- in plan dimensions and has a maximum loading tion is more difficult and since all the buildings iso- capacity of 1000 tons. A maximum horizontal accel- lated in Japan use horizontal isolation only, it eration of 1.84 g and a maximum vertical accelera- would be more difficult to collect performance data tion of 0.92 g can be applied. The table has been used on vertically isolated buildings in the near future. for testing several large scale nuclear components Thus, it is expected that if seismic isolation is se- for LWRs. Scale factors ranged from 1 to 3.7 with lected for the first DFBR, only horizontal isolation model weights from 290 to 750 tons. will be adapted, while keeping vertical isolation an Seismic isolation is also gaining support in the option for future applications. Several types of de- Japanese nuclear industry. This is especially true vices are being considered including steel lami- among companies and utilities participating in the nated high damping elastomeric bearings, elas- development of the advanced demonstration FBR tomeric bearings with various energy absorbers in- plant (DFBR) whose construction is expected to start cluding lead plugs, steel dampers, viscous dampers in the mid to late 1990s. At present, an important and friction dampers and coil springs with viscous factor inhibiting the commercialization of FBRs is dampers. the construction cost. The cost of Monju, a demon- The lead reactor manufacturer and construction stration plant with a capacity of 280 MWe, is nearly companies participating in this program are Hi- twice that of commercial light water plants of the tachi and Kajima, respectively. In a joint study it 1000 MWe class^2. Seismic isolation is expected to was shown that the design of FBRs could be greatly reduce significantly the cost of seismic design in simplified if seismic isolation is used^. For ex- FBRs and to facilitate standardized FBR designs ample, the total reactor building weight could be re- for a wide range of siting conditions. duced by more than fifty percent. The two other ma- Research and development has been required to jor reactor manufacturers, Toshiba and Mitsubishi develop isolation systems with the required relia- Heavy Industries and the remaining top five con- bility for nuclear applications. Performance expe- struction companies are also participants. Another rience of isolated Ptructures must be gained in less program*^ investigated the effect of damping type critical structures as a precondition for accepting on the performance of equipment in isolated struc- seismic isolation for nuclear plants. Thus, re- tures. Four types: friction, lead, oil, and viscous search in this field is receiving large support from were tested. It was concluded that viscous damper the government and electric companies. The Cen- gave the best attenuation of accelerations in equip- tral Research Institute of Electric Power Industry ment. The study recommended that the choice of (CRIEPI) is managing a seven year research pro- damping should be selected based on the application. gram in seismic isolation, which started in 1987. Shake table tests performed at EERC have also The objectives of this program are: establishment of demonstrated that accelerations in equipment in a seismic isolation design for the LMFBR; selection isolated structures are minimized when high damp- of appropriate seismic isolation devices; perfor- ing elastomeric bearings with no add-on damping mance of large scale element tests of selected de- elements are used. However, when other elements vices; performance of large scale system tests on the are added to provide additional damping, they

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

54 inevitably cause high frequency response and in- high damages/outages post-earthquake. In this re- creased accelerations in equipment^,*^. rpnea( j. spect, the important "cost" issue, as seen by the vantages of using high damping rubber bearings in owner of such a facility, will include both initial nuclear applications and a summary of these test construction as well as lifetime operations costs. results have been studied"*?. The feasibility of using lead filled elastomeric bearings for horizontal iso- CODE EFFORTS lation was investigated experimentally*^. Feasi- This section of the report summarizes the status bility of 3-D isolation using steel coil springs and of various groups throughout the world to develop codes (or guides) for design of base isolated struc- viscous dampers was studied experimentally and tures. The technical aspects of the design codes or analytically*^, A hybrid system in which the build- guidelines are not provided herein, partially since ing is isolated horizontally using elastomeric bear- they are lengthy, and partially since some are not ings and the reactor floor is isolated vertically on yet published. steel springs is under investigation^. UNITED STATES NON-NUCLEAR APPLICATIONS The Seismology Committee of the Structural This report has emphasized the current status of Engineers Association of California (SEAOC) has base isolation for nuclear facilities. As of 1989, recently completed the development of seismic regu- there are more than 100 applications of base isola- lations for the design of base isolated structures41. tion in non-nuclear facilities, mostly for bridges These regulations are an update of the seismic re- and low rise structures. Most applications are in quirements developed by the Northern Section of New Zealand (>40); there are several in Japan 42 (>30), and United States (>15); Canada, China, SEAOC and published in 1986 . The SEAOC regu- England, France, Greece, Iceland, Iran, Italy, Mex- lations have been submitted to the International ico, Romania, USSR, and Yugoslavia all have at Conference of Building Officials (ICBO) for con- least one application. Almost all applications have sideration as Division III of Chapter 23 (General been built in the 1980s. The 1990s may very well be a Design Requirements) of the Uniform Building decade where base isolation in the non-nuclear Code43. After review and adoption by ICBO, these world "comes into its own", with many hundreds of regulations will be published as an appendix to sup- applications. plement existing seismic design requirements for As of the date of writing this paper, no existing conventional fixed-base buildings and are intended base isolated structure has undergone a truly for use with non-nuclear base isolation applica- "major" earthquake. Several buildings have per- tions. formed well through earthquake motions on the order of O.Olg to 0.05g, and one (the Okumura build- EUROPEAN COMMUNITY ing in Tsukuba, Japan) has performed well in a The design of structures in member countries of 0.20g earthquake. The Te Teko bridge in New the European Community will be governed by Eu- Zealand has experienced an approximately 0.3g rocode No. 8, which will become effective in 1992. ground motion. Post-earthquake studies have Part 5 of this code is concerned with foundations. A shown that these structures have performed as pre- proposed section, Annex 5A, addresses design of dicted in design. base isolated structures. This annex is preliminary Perhaps the most promising areas for non-nu- and will undoubtedly be influenced by ongoing code clear application is in the area of isolating facilities work in member countries, such as France and whose functionality following a very large Italy. earthquake is important to society. Candidate structures are power generating facilities; hospi- FRANCE tals; etc. Similarly, facilities which house very ex- A French base-isolation design code is currently be- pensive, or sensitive, equipment, like computers, ing developed. When complete, this code will be are also candidates. adapted as a chapter in the French earthquake code, Some key aspects as to whether to or not to base- "Regies Parasismiques P.S. 1986." The responsi- isolate such structures are whether the facilities ble agency for the French seismic code is the Asso- need to be operable after the earthquake; or whether ciation Francaise Parasismiques, located in Saint the facilities cannot afford to incur unacceptable Remy. An initial draft of the base isolation code is

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 expected by the end of 1989, The code is intended for rapid worldwide acceptance in the commercial field the regulation of non-nuclear structures, but will and is being implemented in advanced nuclear de- also establish minimum requirements for nuclear signs of the future. Since the seismic response of applications. isolated nuclear structures is of a predictably lower amplitude, the risk of accidents due to uncertainties ITALY in the input motions can be reduced, public safety Since 1988, the Italian National Commission for margins can be increased, plant investment protec- Atomic and Alternative Energy Sources (ENEA) in tion can be enhanced, and standard plant design cooperation with the Italian institute, ISMES, has can be achievable. Progress in codes and standards been developing design guidelines for seismic iso- should soon make the base isolation option a viable lation of nuclear reactors. A preliminary draft has alternative for the design of nuclear facilities in the already been prepared by ENEA and an updated United States in the next few years. version should be available in late 1989. 1Chung, H.., editor, "Seismic, Shock and Vibration Isola- JAPAN tion," Seismic Engineering, Recent Advances in Design, In Japan, two groups are developing guidelines Analysis, Testing and Qualification Method, ASME PVP- for the use in non-nuclear applications. First, the 127,1987. Architects Institute of Japan (AIJ) is developing de- ^Chung, H., and Mostaghel, N., editors, "Seismic, Shock and sign guidelines for use by the building designer. A Vibration Isolation," ASME PVP-147,1988. first draft of this document is expected by late 1989. ^Chung, H., and Fujita, T., editors, "Seismic, Shock and Vi- Separately, the Building Center of Japan is develop- bration Isolation," ASME PVP-181,1989. ing guidelines for use by oversight committees in 4Kelly, J. M., "Aseismic Base Isolation: Review and Bibli- the review of isolated buildings. The Ministry of ography," Soil Dynamics and Earthquake Engineering, Vol. Construction of Japan requires all base-isolated 5, No. 3,1986. buildings to have a special construction permit that ^Tajirian, F.F., and Kelly, J.M., "Seismic Isolation of Nuclear involves extensive review by the Building Center. Plants: a World Overview," Seismic Engineering, Research A draft of this document is also expected in late 1989. and Practice, ASCE, 1989. With respect to nuclear construction applica- ^Plichon, C, et al, "Protection of Nuclear Power Plants tions, the CRIEPI is pursuing an extensive base iso- Against Seism," Nuclear Technology, Vol. 49,1980. lation research program for isolated nuclear struc- 'Postollec, J-C, "Les Foundations Antisismiques de la Cen- tures including the development of design guide- trale Nucleare de Cruas-Meysse," Notes du Service Etude lines. The research program is long term and final Geni Civil d'EDF-REAM, 1983. design guidelines are not expected until 1992. °Jolivet, J., and Richli, M.H., "Aseismic Foundation System for Nuclear Power Stations," Proc. SMIRT-4, Paper K.9/2, NEW ZEALAND San Francisco, 1977. In New Zealand, where the application of seis- ^Plichon, C, Gueraud, R., Richli, M.H., and Casagrande, mic isolation to bridges is quite common, the Min- J.F., "Protection of Nuclear Power Plants Against Seism," istry on Construction has developed internal regu- Nuclear Technology, Vol. 49, 1980. 10 lations for the design of isolated bridges. Isolated Delfosse, G. C, and Delfosse, P.G., "Earthquake Protection buildings have been regulated on a case-by-case ba- of a Building Containing Radioactive Waste by Means of sis. Base Isolation System," 8th World Conference on Earthquake Engineering, San Francisco, August, 1984. CHINA ^Bouchon, Marc, "Nuclear Spent Fuel Storage Pools on In 1987 the Seismic Isolation Committee was Aseismic Bearing Pads," 9th World Conference on Earthquake Engineering, Tokyo, August, 1988. established under the Architectural and Structural ^Martelli, A., "Some Remarks on the Use and Perspectives Society of China, and activities in code development of Seismic Isolation for Fast Reactors," Seismic, Shock and have started. Each implementation of base isolation Vibration Isolation, PVP-147, 1988. in China to date has happened individually, with 13 advance approval by authorities. Kunar, R.R., and Maine, T., "A Review of Seismic Iso- lation for Nuclear Structures," Electric Power Research In- stitute, NP-1220-SR, October, 1979. CONCLUSIONS l^Vaidya, N.R., and Eggenberger, A.J., "Feasibility Eval- Seismic isolation is a significant recent devel- uation of Base Isolation for Aseismic Design of Structures," opment in earthquake engineering that is gaining

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

56 D'Appalonia Consulting Engineers, NSP Project No, 82- 31U,S. NRC Engineering Evaluation Report No. 1365,1984. AE0D/E707, "Design and Construction Problem! at Operat- 15Burns and Roe, Inc., "LSPB, Large Scale Prototype ing Nuclear Plants," March 31,1987. Breeder, Base Isolation System Conceptual Design," Pre- 32Fujita, T., "Earthquake Isolation Technology for Indus- pared for U.S. Dept. of Energy and EPRI, September, 1984. trial Facilities - Research, Development and Application in 16Preskakis, G.N., and Sigal, G.B., "Seismic Isolation Sys- Japan," Bulletin of the New Zealand National Society for tems for LMFBR Plants," 8th SMIRT Conference, Vol K(b), Earthquake Engineering, Vol. 18, No. 3, September, 1985. Brussels, August, 1985. 33Mizukoshi, K., et al, "Conceptual Framework Design and 17Ikonomu, A.S., "Alexisismon Isolation Engineering for Shaking Table Verification Test on Seismically Isolated Nuclear Power Plants," Nuclear Engineering and Design, LMFBR Reactor Building," 9th World Conference on 1985. Earthquake Engineering, Tokyo, August, 1988. 18Berglund, R.C., Tippets, P.E., Salerno, L.N., "PRISM, A •"Fujita, S., et al, "Earthquake Isolation Systems for Sade, Economic and Testable Liquid Metal Fast Breeder Buildings of Industrial Facilities Using Various Types of Reactor Plant," ANS Topical Meeting on Safety of Next Damper," 9th World Conference on Earthquake Engineer- Generation Power Reactors, Seattle, Washington, May, ing, Tokyo, August, 1988. 1988. 35Kelly, J.M., "The Influence of Base Isolation on the 1*Tajirian, F.F., and Schrag, R.R., "Conceptual Design of Seismic Response of Light Secondary Equipment", Poc. of Seismic Isolation for the PRISM Liquid Metal Reactor," the Internationa] Conference on Natural Rubber for SMIRT 9, Vol K2, Lausanne, 1987. Earthquake Protection of Buildings and Vibration Isolation, 20Tajirian, F.F., and Kelly, J.M., "Testing of Seismic Isola- Malaysia, 1982. tion Bearings for Advanced Liquid Metal Reactors 36Kelly, J.M., and Tsai, H.C., "Seismic Response of Light (PRISM)," Seismic, Shock and Vibration Isolation, PVP-147, Internal Equipment in Base Isolated Structures," Report No. 1988. UCB/EERC-84/17, Berkeley, 1984. 2lTajirian, F.F., Kelly, J.M., and Glueckler, E.L., "Testing of 37Tajirian, F.F., and Kelly, J.M., "Seismic and Shock Iso- Seismic Isolation for the PRISM Advanced Liquid Metal lation System for Modular Power Plants," Chapter 7, Seis- Reactor Under Extreme Loads," SMIRT 10, August, 1989. mic Engineering, Recent Advances in Design, Analysis, ^lueckler, E.L., Tajirian, F.F., and Kelly, J.M., "Seismic Testing and Qualification Method, ASME PVP-127,1987. Isolation for a Modular Liquid Metal Reactor Concept ^Yasaka, A., et al, "Feasibility Study on the Seismic Iso- (PRISM)," ANS Topical Meeting on Safety of Next Genera- lation of Pool-Type LMFBR 3. Experiments of Lead and tion Power Reactors, Seattle, Washington, May, 1988. Elastomeric Bearings, "SMIRT 9, Vol. K2, Lausanne, 1987. 23Seidensticker, R.W., Letter to F. Tajirian, dated December 39Sonoda, Y., et al, "A Study on the Seismic Isolation of Fast 6,1988. Breeder Reactor Plant 3-D Seismic Isolation of Reactor 24Tamura, K., Yamahara, H., and Izumi, M., "Proof Tests of Structure," SMIRT 9, Vol. K2, Lausanne, 1987. Base-Isolated Building Using Full Sized Model," Seismic, ^"Sakurai, A. et al, "Seismic Isolation Structure for Pool- Shock and Vibration Isolation, ASME PVP-147,1988. Type LMFBR - Isolation Building with Vertically Isolated 25 Floor for NSSS," SMIRT 9, Vol. K2, Lausanne, 1987. Oldenkamp, R.D., Brunings, J.E., Guenther, E., and Hren, 41 R., "Update - Sodium Fast Reactor (SAFR) Concept," Proc. "Tentative General Requirements for the Design and of American Power Conference, Vol. 50,1988. Construction os Seismic Isolated Structures," Ad Hoc Base 26Neylan, A., Ng., R., and Dilling, D., "Designing a Reactor Isolation Subcommittee of the Seismological Committee, for the Next Generation," Proc. 23rd Intersociety Energy Structural Engineers Association of California, August 15, Conversion Engineering Conference, Vol.1, ASME, 1988. 1989, (to be published in Building Standards, International 27ETEC, "Seismic Technology Program Plan," Prepared for Conference of Building Officials, as proposed code change the U.S. Dept. of Energy by the Energy Technology proposal, Division III of Chapter III, "Earthquake Regula- Engineering Center, 1988. tions for Seismically-Isolated Structures," January, 1990). 28 'tentative Seismic Isolation Design Requirements, Base Gray, S., Rodwell, E., and Hottori, S., "EPRI/CRIEPI Joint Isolation Subcommittee of the Seismology Committee, Study Program in Support of the Advancement of Liquid Structural Engineers Association of Northern California, Metal Reactor (LMR)," Proc. 23rd Intersociety Energy San Francisco, California, September, 1986. Conversion Engineering Conference, Vol. 1, ASME, 1988. 43 Uniform Building Code, International Conference of ^Mitsubishi Heavy Industries, Ltd., "Results of Reduced Building Officials, Whittier, California. Size Element Elastomeric Bearing with Lead Plug Test," a Report to CRIEPI, March, 1987. 30NUREG/CR-1508, "Evaluation on the Cost Effects on Nuclear Power Plant Construction Resulting From the In- crease in Seismic Design Level," April, 1981.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

57 COST-BENEFIT ASSESSMENT OF THE SEISMIC DESIGN OF THE TUFF REPOSITORY WASTE HANDLING FACILITIES

C. V. Subramanian Sandia National Laboratories, Division 8511 Livermore, CA 94551

This paper summarizes a cost-benefit assessment of the seismic design of the waste-handling facilities associated with the prospective high-level waste repository at Yucca Mountain, Nevada. It provides a very brief description of the methodology used and the costs and benefits of varying design levels for vibratory ground motions and surface fault displacements for structures, components, and equipment that are important to safety in the waste-handling facilities.

COST-BENEFIT ASSESSMENT METHODOLOGY The total cost is divided into two elements: accident-related costs and nonaccident-related A cost-benefit study determines the optimum costs. An estimate of accident-related costs is solution for the problem under consideration. The associated with the probabilities of both earthquake independent variable in a cost-benefit analysis occurrences and system, structure, and component could be a continuous function, such as ground failures; thus, this estimate requires the motion acceleration or a set of discrete calculation of expected rather than direct costs. alternatives, such as specific fault rupture dis- An estimate of nonaccident-related costs is direct placements. The optimum seismic design level and straightforward. for a given structure can be obtained by simply setting to zero the first derivative of the total cost Accident-related costs resulting from a seismic objective function, Cry, with respect to the design event are difficult to quantify in dollars; hence, acceleration, a: such a quantification is made with built-in uncertainties in the estimates. Accident-related costs resulting from a seismic event are evaluated for the following attributes: da (l) • offsite public exposures; In this case, the objective function is the total cost of • short-term occupational exposure; the initial investments and consequences, • offsite property damage/cleanup; expressed in terms of the design acceleration. • onsite damage, repair, and/or Obviously, other cost parameters could be used: decontamination; and e.g., total cost per health effect (deaths) or the • mission delays. incremental cost per reduction in health effect could be optimized. These concepts are in common Nonaccident-related cost attributes considered use and could be employed to arrive at a decision include: regarding the design level. In this paper, total cost is used.

Note: This work was supported by the U.S. Department of Energy (Office of Civilian Radioactive Waste Management, Yucca Mountain Projects Office) under Contract No. DE-AC04-76DPOO789

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

58 associated dose for each of the damage states • engineering construction for was calculated. both structures and equipment, • licensing, Because of uncertainties in modeling • site characterization, and parameter values, the damage states were • nonregulatory delays. represented as a conditional probability of the event level. For each damage state, this can be The steps needed to obtain the optimum design shown as a probability density function and a level are summarized below. cumulative probability function (the graphical representation of which is often referred to as ACCIDENT-RELATED COSTS fragility curves). A fragility curve is defined as the useful limit of the prescribed damage 1. An earthquake could cause a spectrum of state. A set of fragility curves for each design different damage states. However, different level was developed so that all damage state earthquake events would cause different fragilities were known. damage states on a given structure. Each damage state could result in different offsite The seismic hazard data for the Yucca M'ii onsite consequences. Further, events that Mountain repository site was determined by are within the design basis are assumed to evaluating the relationship between the cause no damage. Therefore, the damage state annual probability of exceedance and the peak of the structure is determined for the "beyond- ground acceleration or fault displacement. design-basis" conditions for each of the specific design levels. The seismic hazard data for the site was convoluted with the fragility curves for the For critical facilities designed for realistic structure to determine the damage state ground motions and expected to withstand any probabilities. Because each damage state is ground notion without collapsing, four related to a radioactive release, compu- damage states are deemed sufficient and tational results can be summarized for each meaningful. Based on this, the following four design level. damage states were defined in terms of the complete structural response: The accident-related costs were quantified in dollars for each attribute, and for each damage • Light (L), state, the total cost was summed and presented • Moderate (M), as a function of the associated radioactive • Heavy (H), and release. • Total (Tj. 7. Given the annual probability of release and the The damage state L is associated with an cost of the release, the expected cost, E(c), ca i be earthquake occurrence slightly beyond the computed as design basis earthquake. The other states correspond to increasing levels of beyond- design-basis events. E(c) = c(x)f(x)dx , (2)

2. For each specific design level, the levels of the beyond-design-basis events that would cause where c(x) and fTx) are functions of cost and the given damage states to occur was release. determined. For each damage state, a list of structural failures, such as potential falling The relationship shown above can be concrete and concrete crack widths and algebraically approximated by lengths, was also identified. Given the falling concrete, concrete crack widths and lengths, and associated radioactive material c(x ) + c(x. ,) E(c) = inventories, the offsite radioactive release and —i-^—H!-P(X.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 thick. This building is used for the receiving, The terms of this equation are shown for a preparing, and packaging in containers of light- water reactor wastes and defense high-level wastes release resulting in x rems between Xj and before being transported underground for storage, XJ+J. The summation of incremental expected The seismic design basis for this facility is 0.4 g costs for each accident-related attribute ground motion with no specific requirements for resulted in the annual expected cost of the fault displacement (Reference 3). The design level accident-related effects. for this building was varied, and each design was evaluated from ground motion levels between 0.2 8. These steps were repeated for all of the and 1.0 g and fault displacement between 0 and 100 accident-related attributes listed earlier for cm. each design level considered, and the results were plotted. Seismic hazard curves for both ground motion and fault displacement at the repository site are shown NONACCIDENT-RELATED COSTS in Figures 3, 4, and 5. Details of the methodology used to develop these curves are given in 9. The calculation of nonaccident-related costs Reference 1. These curves indicate that the does not involve the probability of the release annual probability of exceedance for horizontal of radioactive material. These costs are acceleration of 0.4 g is about 5 x 10"^ and the incurred regardless of whether an accident annual probability of a fault rupture >1 cm under occurs. These are directly calculated costs for the waste-handling building is about 10'^. Hence, each of the nonaccident-related attributes although the ground rupture hazard is more listed earlier. uncertain than the acceleration hazard, the very low probabilities of the former make it 10. The individual cost elements were summed to insignificant in the cost-benefit evaluation. give total nonaccident-related cost. Spalling and cracking of concrete were the 1.1. The accident-related and nonaccident- primary structural damage types investigated. related costs were summed to obtain total Massive or total structural collapse was ruled out present cost, C™, as a function of design because of the inherent strength of the waste- handling building design. The basic accident level. This relationship was plotted scenarios were identified for different degrees of graphically to obtain the optimum design damage. In these scenarios, spalling of concrete level. pieces was assumed to damage spent fuel assemblies or containers and generate airborne This methodology was applied to evaluate the radioactive particles within the hot-cell structures. waste-handling facilities of the proposed Yucca Such particles can escape through the cracks of the Mountain repository for both ground motion and damaged structures into the atmosphere. The hypothetical fault displacements. More details of quantity of the radioactive release depends on the methodology may be found in References 1 and various factors, including the amount of spalling 2. and cracking. Therefore, the probability of a radioactive release is coupled to the probability that APPLICATION AND RESULTS a specific amount of spalling and cracking will occur. In the study reported here, four damage In this study, the methodology described above states for the structure were defined and was focused primarily on the seismically-induced quantified •>. terms of structure deformations, damages to the main waste-handling structures, crack sizes, and spalling concrete pieces. These systems, and components that result in are light, moderate, heavy, and total damage radiological releases. Waste-handling building states. Cracks and spalling pieces were estimated used in this study is based on the configuration for these damage states. Using the damage states shown in Reference 3. A plan view and cross- as limit states, fragility curves with probabilities section of the same are shown in Figures 1 and 2. expressed as a function of peak ground The structural system of this building consists of shear walls and slabs ranging from 2.0 to 5.5 ft.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

60 >»i —

(J)n n o r •c* Q. O O

3 O 3 1 _HJfi"^Sr®LS

03 0a3 era I—• 5' n o §

Figure 1. General Layout of Waste Ifandling PLAN-EL, CH«Ot_f AjjOve Building - Yucca Mountain Project n a 3 a. D O

v n 3 O

N 3.

o 3

r WASTE HANDLING BUILDING 2 n

OP c Figure 2. Cross-Section of Waste Handling Building - Yucca Mountain Project IP a 6 a O o PI z S3 c 3 a o 3 n S5

03 3.

0.1 1 10 S3 PEAK ACCELERATION VERTICAL RUPTURE (en) RUPTURE DISPLACEICNT 5" 3 o Figure 3. Horizontal Ground Acceleration Figure 4. Ground Rupture Hazard Curve Figure 5. Median Acceleration Associated Seismic Hazard Curve for the for the Yucca Mountain Site with Ground Rupture at the o Yucca Mountain Site Yucca Mountain Site 3 3 h

CD 00 'Si acceleration or fault displacements wore nonaccident-related costs in Figure 7 does not developed for each structural element. alter the basic shape of Figure 7. This indicates that for this repository site and for the seismic Utilizing the information on crack sizes and the design criteria, consequences of seismic events number of concrete pieces that would potentially are irrelevant to the selection of the design level. spall for each of the damage states and the information on the waste inventory inside the As a last step, an evaluation was done to determine waste-handling building, the quantities of the effects of the uncertainties in the various radioactive materials released inside the waste- parameters on the results discussed above. It was handling building and to the outside environment concluded that the overall uncertainty in these for each damage state were evaluated. These parameters needs to be very large (on the order of results were used to estimate various accident 10^ or greater) before they would affect these consequences, including offsite public exposures, results. short-term occupational exposures, damage to offsite properties, damage to onsite structures and Complete details of this study can be found in equipment, and mission delays caused by Reference 1. disruption to repository operation. The dollar costs related to these consequences were estimated and CONCLUSIONS then summed to obtain the total costs for each of the four damage states. These total costs were then Some of the important conclusions derived combined with the probability of exceedance of the from this study are given below. These damage states to give the expected values of conclusions can be drawn notwithstanding accident costs for the different design levels. The uncertainties and approximations used in the probability of exceedance of a damage state (or the study. corresponding offsite dose) was obtained by integrating the product of the probability density • The expected cost and risk to the public at function of the seismic hazard and the fragility all design levels are very low. This probability function corresponding to the damage implies that this facility is a low seismic state. Probable cost of accident as a function of the risk facility. seismic design level is shown in Figure 6. The most striking feature of Figure 6 is the extremely low expected accident-related costs for all possible • The total nonaccident-related cost is fairly design levels considered in the study. constant for design levels between 0.2 and 0.6 g.

The next step in the evaluation was to determine • The increase in nonaccident-related costs the nonaccident-related costs for the attributes if the design level is changed from 0.4 g to listed earlier. These are design and construction 1.0 g is on the order of $150 million. activities for structures and equipment, licensing activities, site characterization activities, and • The waste-handling building appears to be mission delays. As indicated earlier, these costs quite resistant to potential fault are incurred regardless of whether an accident displacement. Hence, specifying a fault- occurs. These are direct costs and are offset design with a goal of no damage is straightforward to compute except for the mission not necessary. It would be costly to achieve delays. The individual cost elements for these and would not readily gain acceptance attributes were computed for each design level and owing to the lack of established design and then summed to give the total nonaccident-related construction code requirements. cost as shown in Figure 7. This figure indicates that the nonaccident-related costs are rather REFERENCES insensitive in the mid-range (0.2-0.6 g) of design levels. A comparison of Figure 7 with Figure 6 1. C.V. Subramanian, N. Abrahamson, A.H. indicates that the accident-related costs are Hadjian, L.J. Jardine, J.B. Kemp, O.K. extremely small compared to the direct or Kiciman, C.W. Ma, J. King, W. Andrews, nonaccident-related costs. Further, summation of and K.P. Kennedy, "Preliminary Seismic the accident-related costs from Figure 6 to the

Second DOE Natural Phenomena Hazards MM Ration Conference - 1989 Design Cost-Benefit Assessment of the Tuff Repository Waste-Handling Facilities," SAND88-1600, Sandia National Laboratories, Albuquerque, NM, February 1989. C.V. Subramanian and A.H. Hadjian, "Cost- Benefit Assessment Methodology for Seismic Design of High-Level Waste Repository Facilities," SAND88-1931, Sandia National Laboratories, Albuquerque, NM, March 1989 (published in SMiRT 10 Conference Proceedings, Paper #RO6/3, Anaheim, CA, August 1989).

H.R. MacDougall, L.W. Scully, J.R. Tillerson (Editors), "Site Characterization Plan Conceptual Design Report," SAND84- 2641, Sandia National Laboratories, Albuquerque, NM, September 1987.

£500 T

2000 ' 300 ••

250 • a 1500 • S § 200 • c 1000- 3 1 so - •

° 100- 500-

50 ••

-i 1 0.2 0.4 0.6 0.8 o.z 0.4 O.S O.I

Design Level Evthquak* (g)

Figure 6. Probable Cost of Accident as a Figure 7. Total Nonaccident-Related Costs Function of the Seismic Design Level as a Function of Design Acceleration

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

(S3 SEISMIC PROCUREMENT REQUIREMENTS AT THE FPR FACILITY AT INEL

GREG S. HARDY & MICHAEL J. GRIFFIN EQE ENGINEERING INCORPORATED GAIL E. BINGHAM WESTINGHOUSE IDAHO NUCLEAR COMPANY

ABSTRACT

Traditional methods used to seismically qualify equipment for new facilities has been either by testing or analysis. Testing programs are generally expensive and their input loadings are conservative. It is also generally recognized that standard seismic analysis techniques produce overly conservative results. Seismic loads and response levels for equipment are typically calculated that far exceed the values actually experienced in earthquakes. A more efficient method for demonstrating the seismic adequacy of equipment has been developed which is based on conclusions derived from studying the performance of equipment that has been subjected to actual earthquake excitations. This earthquake experience data concludes that damage or malfunction to most types of equipment subjected to earthquakes is far less than that predicted by traditional testing and analysis techniques. The use of conclusions derived from experience data provides a more realistic approach in assessing the seismic ruggedncss of equipment. By recognizing this inherently higher capacity that exists in specific classes of equipment, vendors can often supply "off the shelf equipment without the need to perform expensive modifications to meet requirements imposed by conservative qualification analyses. This paper will describe the development of the experienced based method for equipment seismic qualification and its application at the FPR facility.

BACKGROUND

The traditional approach to seismically qualifying 2. Reduced costs both for procurement and equipment has been to write into the equipment engineering of the equipment vendor specification the seismic design criteria of the facility where the equipment is to be installed. Design specifications for new facilities have traditionally The vendor then provides the seismic qualification included special requirements to insure that equipment documentation for the equipment item. This will survive the facility design basis earthquake (DBE). To approach is time consuming and typically produces insure that their products will survive the DBE, equipment overly conservative results. The earthquake suppl'crs are required to perform a seismic analysis or a experience database approach introduced in this vibration test as specified in (he procurement paper is a more efficient method that incorporates specifications which must be recovered in the purchase two fundamental advantages over the more price of the equipment. It is generally recognized that traditional seismic qualification methods: standard seismic analysis procedures produce overly conservative results. Seismic loads and response levels for 1. Reduced equipment seismic specification equipment are typically calculated that far exceed the requirements values actually experienced in earthquakes. Experience in the performance of most types of equipment in

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

66 earthquakes shows that damage or malfunction occurs EARTHQUAKE EXPERIENCE DATABASE far less often than conventional analyses predict. The conclusion reached is thai equipment generally has a Very few components of nuclear plant systems arc high tolerance to seismic loads. Utilization of the unique to nuclear facilities. Nuclear plant systems experience database approach gives a much more include electrical swilchgcar, control panels, motor- realistic assessment of the true seismic ruggedness of operated valves, pumps, piping, ducts, conduit, cable equipment and includes effects such as higher modal trays, and many other items that are common damping, nonlinear response, local plastic yielding and a components of conventional power plants and industrial capability to absorb high stresses for short time periods. facilities. The seismic experience database was Each of these effecls is difficult to accurately account for developed to address the problem of equipment analytically. By recognizing the inherently higher qualification in nuclear plants which were built before capacities that exist in specific classes of equipment, specific seismic requirements existed. By reviewing the vendors can often supply "off the shelf equipment performance of facilities (commercial as well as without undergoing expensive modifications to meet the industrial) thai contain equipment similar to that found requirements imposed by conservative qualification in nuclear power plants, conclusions can be drawn about analyses. The end result is that specification the performance of nuclear plant equipment during and requirements for seismic qualification of equipment can after a design basis earthquake. be significantly reduced. Strong motion earthquakes frequently occur in The direct result of reduced design specification California and throughout the world; power plants or requirements for the equipment vendor is reduced cost industrial facilities arc often located in these affected to the purchasing organization. Analysis and testing areas. By studying the performance of these qualification can be very costly, and the costs must be earthquake-affected (or database) facilities, a large recovered in the purchase price. This higher cost is also inventory of various types of equipment that have attributed to the modifications that may be necessary as experienced substantial seismic motion can be compiled. a result of conservative seismic analyses used in The ground acceleration experienced at most of these demonstrating equipment seismic ruggedness. The end database sites, measured by nearby ground motion result of using the experience based approach is a records, is comparable to, or in excess of, the seismic significant cost savings. design basis for practically all United States nuclear plant sites. A significant cost savings also results for the facility Architect/Engineer associated with the project as he The primary purposes of the seismic experience must dedicate engineering time to reviewing and cataloging the traditional, and sometimes quite database arc: extensive, seismic qualification reports. Since the seismic qualification requirements arc reduced the o To determine the most common sources of seismic documentation is reduced yielding a substantial seismic damage, or adverse effects to facilities savings in this area as well. that contain equipment representative of safety- related nuclear plant systems. This paper presents the development and industry acceptance of the experience based approach for the o To determine the thresholds of seismic motion scisn.ic qualification of equipment. The application of corresponding to various types of seismic the experience uala based approach to process systems, damage. equipment, HVAC and structures at the Fuel Processing Restoration (FPR) facility that is currently being o To determine the performance of equipment constructed at the Idaho National Engineering during earthquakes, regardless of the levels of Laboratory (INEL) is presented. This equipment must seismic motion. be designed to withstand external events such as earthquakes, in order to ensure a minimal risk to the o To determine seismic integrity criteria with public safety. regard to minimum standards in equipment construction and installation, based on past experience, to ensure the ability to withstand anticipated seismic loads.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

67 The database was compiled through surveys of the EXPERIENCE DATABASE METHODOLOGY following types of facilities: The experience database methodology has been o Fossil-fueled power plants formalized in (he Generic Implementation Procedure (GIP) [1], which has been developed to resolve o Hydroelectric power plants Unresolved Safety Issue (USI) A-4612|. USIA-46 was identified by the NRC to address the seismic adequacy o Electric distribution stations of older equipment in operating nuclear power plants which had been qualified using methods different from o Petrochemical facilities those which are currently being required. The GIP procedure incorporates the earthquake experience o Water treatment and pumping stations approach to seismic qualification and couples that with simplified anchorage evaluation methods. The basic o Natural gas processing and pumping stations methods involved in the GIP can be summarized as:

o Manufacturing facilities 1. Comparing the site earthquake response spectrum to the experience database bounding o Large industrial facilities spectrum.

o Commercial facilities (focusing on their HVAC 2. Reviewing the equipment to assure plants) similarity /representation to equipment contained in the database. In general, data collection efforts focused on facilities located in the areas of strongest ground motion for each 3. Verifying the database seismic restrictions arc earthquake investigated. Facilities were sought that satisfied. contained substantial inventories of mechanical or electrical equipment or control and distribution systems. 4. Reviewing the equipment installation for unusual Because of the number of earthquake-affected areas and or nontypical component arrangements, cither types of facilities investigated, there is a wide diversity in internal or external. the types of installations included in the database. For the types of equipment of focus, this includes a wide 5. Reviewing the equipment anchorage adequacy. diversity in age, size, configuration, application, operating conditions, manufacturer, type of building, 6. Reviewing the area around the installed location within building, local soil conditions, quality of equipment to assure no seismic interaction maintenance, and quality of construction. hazards (such as unsecured lighting fixtures) exist. The seismic experience database is founded on studies of over 100 facilities located in the strong motion areas The GIP procedure applies to existing equipment of 19 earthquakes that occurred in the United States and installations and consists predominantly of a visual other parts of the world since 1971. The earthquakes inspection (steps 2-6) of the equipment item of interest investigated range in Richter magnitude from 5.2 to 8.1. by an engineer experienced with the experience Measured or estimated peak ground accelerations for database and equipment seismic response. database sites range from O.lOg to 0.85g. The bracketed duration of strong motion (on the order of O.lOg or In applying the experience database methodology to greater) ranges from 5 seconds to about 50 seconds. new or future equipment installations, the GIP Local soil conditions range from deep and soft alluvia to requirements must be re-written in terms of hard rock. The sites range from the epicentral area to specification requirements. These requirements arc great distances from the epicenter. The buildings divided into two separate areas of responsibility: housing the equipment of interest have a wide range in size and type of construction. As a result, the database 1. Equipment Vendor includes a wide diversity of seismic input to equipment in terms of seismic motion amplitude, duration, and 2. Architect/Engineer (AE) frequency content.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

68 The vendor seismic specification requirement are o The seismic adequacy review of items located associated with the construction of the equipment around the component, such that they do not component, while the installation requirements are represent a potential interaction hazard. typically not the responsibility of the equipment vendor and fall to the plant engineers or to an AE, The above requirements are directly related to Requirements must be written into the vendor instances of damage that have occurred to components specifications which will demonstrate similarity to the contained in the earthquake experience database and experience database components. EQE's earthquake arc most easily implemented during the design stage experience database includes a tremendous volume of before the equipment has been installed. After equipment which allows for the generation of a broad installation the majority of these requirements must be range of similarity parameters such as equipment type, verified via a walkdown by engineers qualified in using manufacture, construction, material, vintage and seismic the experience database. response. There are several equipment classes where specific construction requirements are necessary based EXPERIENCE DATABASE ACCEPTANCE on the results of this type of configuration found in the experience database. Examples of typical construction In December of 1980, the NRC initiated USI A-46 to requirements include: address the question of the seismic adequacy of equipment in 49 operating nuclear plants (72 units) that o Limitations on the cutout size in electrical panel were not licensed to current criteria. The experience side walls, which relates to cabinet stiffness and database approach was developed through the thus response. sponsorship of the Seismic Qualification Utilities Group (SQUG) which was formed in 1981 in conjunction with o Limitations on the size and weight of external an agreement with the NRC to develop alternate attachments to electrical cabinets. methods to resolve seismic safely issues for critical systems and components in operating nuclear power o Limitations on sheet metal thickness for plants. The NRC subsequently adopted the experience electrical cabinet enclosure construction. database approach as the most cost effective method in resolving USI A-46 over the more traditional types of o Specifying core/coil assemblies of transformers equipment qualification methods. shall be positively anchored to the base enclosure structure. Initial acceptance of the experience database approach came in September 1983 when the NRC issued The majority of the "off-the-shelf equipment NUREG-1018 [3]. NUREG-1018 endorses a pilot incorporale these requirements already, thus, the vendor program [4] which was tasked to demonstrate the can typically provide his commercial grade component feasibility of using earthquake experience data in lieu of without having to provide the lime consuming seismic formal qualification of equipment in operating plants. design and potentially major structural modifications to Shortly thereafter SQUG and the NRC jointly agreed meet the conservative seismic analysis results. upon a panel of independent recognized seismic experts to evaluate the experience data approach to seismically The majority of the seismic experience database qualify equipment. This panel, the Senior Seismic requirements arc addressed by the architect/engineer. Review and Advisory Panel (SSRAP), was assigned the These requirements consist of specifying minimum tasks of (1) reviewing the experience database equipment anchorage and identifying critical installation methodology, (2) determining the limits to which requirements that the design architect/engineer experience data could be applied in the seismic considers in order to insure the components seismic qualification of equipment, and (3) recommending ruggedness as demonstrated by earthquake experience additional areas where the methodology should be data. Examples of installation requirements include: expanded. o Impact considerations between unattached The members of the SSRAP reviewed several cabinet sections located adjacent to one another. database facilities and nuclear plants to judge similarity between the equipment in nuclear power plants and in o Flexibility considerations for connecting items to the conventional plants from which past earthquake the equipment, such as electrical conduit and experience data were collected. The NRC, SQUG, and piping.

Second DOP^ Natural Phenomena Hazards Mitigation Conference - 1989 SSRAP also had discussions relative to the similarity o Low Voltage Swilchgear issue with representatives from vendors and experts in the areas of equipment construction and potential o Medium Voltage Switchgoar failure modes. o Transformers The SSRAP completed its review [5] and endorsed the concept of using earthquake experience dala for seismic o Horizontal Pumps qualification purposes. In addition to SSRAP's endorsement, the major seismic design and review o Direct Drive Vertical Pumps regulatory documents in the United Stales now include the experience data methodology as an alternative o Fluid-Operated Valves approach to the seismic qualification of equipment. Examples of regulatory documents endorsing the o Motor-Operated Valves application of the experience data approach include: o Centrifugal Fans o U.S. NRC NUREG-1211, "Regulatory Analysis for Resolution of Unresolved Safety Issue A-46, o Automatic and Manual Transfer Switches Seismic Qualification of Equipment in Operating Plants" [6] o Distribution Panels

o U.S. NRC Regulatory Guide 1.100, Revision 2, o Chillers "Seismic Qualification of Electrical and Mechanical Equipment for Nuclear Power o Air Handling Units Plants" [7] o Uninterruptible Power Supplies o U.S. Standard IEEE Sid 344-1987, "IEEE Recommended Practice for Seismic Qualification o Instruments on Racks of Class IE Equipment for Nuclear Power Generating Stations" [8] o Sensors

o U.S. NRC "Generic Safety Evaluation Report on The application of experience data to these equipment SQUG Generic Implementation Procedure for categories at the FPR facility consisted of three stages: Implementation of USIA-46" [9] 1. Development of seismic requirements for o DOE 6430.1A, "General Design Criteria" [10] inclusion to the procurement specifications for each class of equipment. APPLICATION AT FPR 2. The review of vendor submiltals to assure the seismic requirements arc satisfied. The Fuel Processing Restoration (FPR) Facility is part of the Idaho National Engineering Laboratory (INEL) 3. The performance of a final walkdown review for in Idaho Falls, Idaho. The FPR Facility is being the installed equipment. designed to recover uranium from product solutions produced by different processes at the INEL Chemical The first stage consisted of reviewing the FPR design Processing Plant. The facility will contain processes and functional requirements for each category of involving highly radioactive materials. Process systems, equipment to establish similarity/representation to equipment, HVAC and structures must be designed to equipment in the experience database. Two separate withstand loads due to natural phenomenon such as groups of seismic specification requirements were earthquakes. The project scope involved developing developed for FPR as described above. seismic qualification criteria for sixteen different equipment classes at FPR based on earthquake 1. The equipment vendor experience data. The equipment classes addressed arc: 2. The architect/engineer o Motor Control Centers

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

70 Specification requirements for the equipment vendor equivalent to that of the anchor boll option were minimal since EQE performed a similarity above, The minimum effective weld area assessment for the manufacturers and models which shall be 0.20 square inches. were on the bidders list. Each equipment class had several equipment vendors specified for review against 2. Adjacent equipment to the MCC's should be the developed specification requirements where it was reviewed for positive anchorage such that determined that the equipment vendor could simply potential failure will not occur that could result supply his standard off-the-shelf unit with no in the impacting of the MCC sections. modifications required. Specific requirements had to be specified for the electrical equipment since minimum 3. The area beside and above the MCC's should be construction standards arc necessary in order to assure reviewed for potential II/I interactions. an adequate seismic capacity level. 4. Attached conduit or cable tray should be The architect /engineer requirements consisted of reviewed for adequate flexibility. specifying minimum anchorage and installation criteria. Various anchorage schemes were specified in most RESULTS AND CONCLUSIONS cases. The installation requirements arc predominantly associated with the flexibility of connecting items The seismic procurement project at the Fuel (conduit, cabling, piping, etc.) to the equipment and Processing Restoration Facility at INEL resulted in an reviewing the area surrounding the component for efficient and cost-effective procedure to produce interaction hazard potentials. These installation procurement documentation during the design stages of uquircments should be implemented during the design a new construction project. The cost reductions stem SK.;C and verified during a post-installation walkdown. from both the reduced costs in qualification (testing and analysis), a reduction in documentation and review, and An example of the two separate requirements also a potential reduction by allowing procurement of developed for the FPR Motor Control Centers arc: commercial grade equipment as opposed to specially- designed seismic grade equipment. The subject EQUIPMENT VENDOR procurement specifications for FPR were reviewed and approved by DOE and have been sent out to potential 1. MCC cabinets should not be customized or vendors in lieu of seismic design criteria. The results to modified from that provided by the equipment dale have proven this methodology to be a much easier vendor. process to implement, both for the vendors and for the plant design personnel. 2. MCC construction should conform to NEMA ICS 6 standards, or equivalent. REFERENCES 3. MCC sections should be manufactured such that [1] URS/BLUME. Generic Implementation adjacent vertical sections can be securely bolted Procedure (GIP1 for Seismic Verification of together. Nuclear Plant Equipment. Prepared for SQUG, December 1988. ARCHITECT-ENGINEER [2] T.Y. Chang. Seismic Qualification of Equipment 1. MCC should be installed with the following in Operating Nuclear Power Plants. Unresolved minimum anchorage: Safety Issue A-46, Washington, DC, December 1980. o (2) 1/2" diameter anchor bolts along the exterior edges of each vertical section, one [3| T.Y. Chang. Seismic Qualification of Equipment front, one rear. Single section MCC's must in Operating Nuclear Power Plants. A Status be installed with (4) 1/2" diameter anchor Report on Unresolved Safely Issue A-46. bolls; or NUREG-1018, Washington, DC, September 1983. o Welding the base of each vertical section to embedded steel with 2 fillet welds located

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 (4J P,l. Yancv, S.W. Swan, Program for the [8] American National Standard IEEE Std. 344- Development of an Alternative Approach lo 1987, IEEE Recommended Practice for Seismic Seismic Equipment Qualification. San Francisco, Qualification of Class IE Equipment for Nuclear Ca., September 1982. Power generating Stations.

[5] Senior Seismic Review and Advisory Panel [9J NRC. Generic Evaluation Report on SQUG (SSRAP), Use of Seismic Experience Data to Generic Implementation Procedure for Show Ruggedncss of Equipment in Nuclear Implementation of USI A-46. forward to SQUG Power Plants. Washington, DC, August 1988. by NRC letter dated July 29,1988.

[6] T.Y. Chang. Regulatory Analysis for Resolution [10] U.S. Department of Energy. General Design of Unresolved Safety Issue A-46, Seismic Criteria. DOE 6430.1A, April 6,1989. Qualification of Equipment in Operating Plants. NUREG-1211, Washington, DC, February 1987.

[7] NRC Regulatory Guide 1.100, Revision 2, Seismic Qualification of Electric and Mechanical Equipment for Nuclear Power Plants. Office of Nuclear Regulatory Research, June 1988.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

72 SEISMIC QUALIFICATION OF SAFETY CLASS COMPONENTS IN NON-REACTOR NUCLEAR FACILITIES AT HANFORD SITE

Ernesto C. Ocoma, Principal Engineer Westinghouse Hanford Company P.O. Box 1970 Rich!and, Washington 99352

ABSTRACT This paper presents the methods used during the walkdowns to compile as-built structural information to seismically qualify or verify the seismic adequacy of safety class components in the Plutonium Finishing Plant complex. The Plutonium Finishing Plant is a non-reactor nuclear facility built during the 1950's and was designed to the Uniform Building Code criteria for both seismic and wind events. This facility is located at the U.S. Department of Energy Hanford Site near Richland, Washington.

INTRODUCTION used and the obstacles encountered Initially it was necessary to define during the walkdowns. the specific boundaries of safety class systems and the necessary components to DEFINITION OF SAFETY CLASS COMPONENTS be qualified. This was accomplished AND SYSTEMS through the interaction of Project, Definition of safety class Safety, and Structural Analysis components, systems, and the appropriate engineering personnel (Seismic Task system boundaries was accomplished Force) addressing safety scenarios through a series of discussions among peculiar to the operation of the members of the Seismic Task Force. Plutonium Finishing Plant (PFP) complex. The meetings addressed safety scenarios The components identified to be of major peculiar to the process and operation importance were certain gloveboxes and of the PFP complex. Some of these filter boxes. The exhaust duct system discussions yielded concerns about connecting the gloveboxes to the filter the stability of the PFP components boxes also turned out. tc be an important during a DBE. safety class system. As these components are necessary for continued operation and The list of safety class components new missions, current safety criteria included 85 gloveboxes and 44 filter must be satisfied, including boxes. The potential collapse of the qualification to the Hanford Site design exhaust duct systems connecting the basis earthquake (DBE). Spot checks gloveboxes to the filter boxes was between drawings and in situ also of concern because of the configurations of components indicated accumulation nf radioactive contaminants extensive field walkdowns would be over the years. The specific concerns required to compile as-built dimensional associated with the safety class items data before initiating an analytical during a seismic event were the effort. The objective of this paper is followiny: to describe preparations and methods

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

73 Will the gloveboxes and filter WALKDOWN PREPARATION boxes slide or topple and possibly Throughout the process to determine initiate a fire, unacceptable powder the safety class components, equipment dispersal, or an explosion? location drawings showing item identification numbers were used to 2. Will the exhaust duct systems expedite the selection process. Using collapse and result in a radioactive these location drawings, a preliminary release? walkdown was conducted to verify the room numbers, in situ equipment numbers, A simple diagram depicting a filter and room accessibility. Room accessor box, exhaust duct, and glovebox entry requirements demanded by facility arrangement is shown in Figure 1. Non- operation or training requirements safety class components or systems dictated the walkdown schedules. The within the seismic zoneof influence of following room entry requirements a safety class item are also a potential dictated the walkdown schedule and threat during a seismic event and are make-up of the team: required to be structurally qualified using the seismic interaction criteria. 1. Protective clothing 2. Radiation protection technician Filter Filter Box 3. Security guard Room 4. "On mask" Walkdown of rooms requiring items 1 and 2 were scheduled during day shift while rooms requiring a security guard were scheduled during other shifts when foot traffic was minimal. Rooms requiring personnel to be on mask were scheduled only after the on-mask requirement was cleared and were then given priority to take advantage of the short time of open access. A separate •Exhaust Duct grouping was also used to identify only the rooms having gloveboxes with lead • Duct Level Floor shielding. The shielding precludes visual inspection of the attachment between the glovebox and the support frame. Walkdowns of these items were scheduled with the concurrence of the maintenance departments so that personnel would be available to remove Glove Box and reinstall the shielding.

Documentation packages of the field walkdown information were prepared for each component, checked, duplicated, Support and filed with the PFP Design Services group. Detail and assembly drawings and associated documents showing Ground Floor dimensions, weights, and material of the internal items for each glovebox and filter box were included in the packages. One set of the files was used Figure 1. 78909105.1 in the seismic evaluation. The other Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 74 set remained in PFP Design Services Horizontal Olst. (ft) files to ensure that data were not 2 3 4 5 6 7 lost during the evaluation process. J ' I 1 M I ' I • I ' I ' I

As the other purpose of these , Maximum Strike Path for Missiles" walkdowns was to identify and resolve Starting from the Duct Level potential seismic interaction concerns, and Below In RMC Line seismic interaction criteria were -5 developed for the non-safety class piping and conduits that are normally supported for deadweight effects. A missile zone of influence curve (Figure 2) was developed to evaluate the potential seismic generated missiles -10 in the room. The seismic walkdown teams (SWT) were made up of design engineers and led by structural analysis engineers. The SWT members were qualified for -15 facility entry by undergoing Radiation Safety Training, Criticality Training, and the PFP Building Orientation Training. Proper badging, security clearance, and permits to carry £-20 prohibited articles (cameras) in secured 8 areas were also needed for each member of the SWT. I Before the walkdown, each design engineer of the SWT was given -25 instructions on plant conduct, acquisition of and verificaton of analytical data required by the seismic analyst, use of the seismic interaction acceptance criteria, and generic walkdown documentation to record and -30 verify the as-built configuration of each component.

WALKDOWNS OF SAFETY CLASS ITEMS •35 The PFP walkdown was accomplished by SWTs subdivided into groups of two people. Selection of the design engineers was based on their experience with various components. The groups were equipped with one of the two sets -40 of files per component, measuring instruments, Polaroid cameras, and generic documentation sheets. One member recorded the dimensions, material, shape, and orientation of the structural -45 i I . I i I j_ member, and documented the in situ Figure 2. 78909105.2 Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

75 structural configuration of the diameter. The bolt length was component. If the drawing callout was found to be the only dimension that correct, it was highlighted in yellow; could not be accurately verified. if there was a discrepancy, the drawing Conservatively the shortest length was updated in red. If an item was for each bolt diameter was employed found to exist in the glovebox or filter in performing the structural box but had no data file in the package, evaluation. only the location and approximate weight were entered. Detailed sketches and Visual inspection was also Polaroid pictures were provided if the performed to ensure that the anchor discrepancy pertained to the support bolts were in good condition, that is, structure of the component. Polaroid no excessive concrete spall ing, bolts pictures were also taken to show all were not loose, nuts were against the sides of the component. Global base plates, and base plates were orientation was also marked on all solid. These conditions were documented photos. A more detailed verification during the walkdowns. and inspection was performed on the following: Some gloveboxes were found to have thicker than normal grout under the 1. Method of attachment between the component base plate. This condition qlovebox/filter box and the support may affect the capacity of the anchor structure. A random sampling bolt, especially when the shortest revealed 50% discrepancies between anchor bolt is used. A pull test was drawings and in situ attachments, performed to qualify these anchor which necessitated the detailed bolts. The applied load was 125% of the inspection. The gloveboxes with calculated tension load required for lead shielding around the base that qualification. precludes visual inspection without removal were inspected during the During the walkdown each of the off shifts with the help of SWT members wis also instructed to maintenance personnel to remove the complete a seismic analysis enveloping shielding. chart (SAEC). This information was used to determine how many unique 2. Orientation of the structural shapes analyses were required and which item The major and minor axes orientation could be qualified by analyses of other of the structural members is critical configurations. The chart requests the to the stress analysis and was following information for each component documented. and support structure. 3. Concrete anchor bolt size. Most of Component the original drawings specifying anchor bolts showed only the • Overall dimensions of the glovebox diameter. Some referred to the or filter box brand name, but none specified the overall length of bolts. A survey • Direction of major axis in the spare parts/storeroom and conversations with personnel who • Approximate total weight of the were involved during the installation component of the components revealed that for each bolt diameter, varying lengths Component Support Structure were used. A random sampling using a nondestructive examination (NDE) • Two elevation views technique verified that different lengths existed for the same • Member size and shape Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

76 • Height • supports and location • restrained direction at each • Type of anchorage support location • wall, ceil ing, and floor Ch he 85 gloveboxes to be penetration detail seismically qualified, 54 required major • support/hanger detail. detailed analysis and 31 required only minor analyses when the envelope Other Items approach was used. Of the 44 filter boxes, 17 required major analyses and a. Problem identification number 27 required minor analyses. The minor analysis requires only one-half the b. Item description time of the major analysis. In order to maximize the effectiveness of the c. Photograph of the item enveloping approach, similar components were segregated into groups. These d. Description of anchorage groups were assigned to analysts having previous analysis experience with that e. Dimensional location and direction type of item. of the item from the safety class component IDENTIFICATION OF SEISMIC INTERACTION ISSUES f. Plan and elevation room sketches The seismic interaction walkdowns showing safety class item and item were conducted in parallel with the representing a potential threat component as-building walkdowns. The interaction walkdowns required almost g. Approximate weight of the item. twice the amount of time to complete as the as-building effort. All of the items that might represent a potential threat, whether Equipped with the seismic acceptable or not, were identified and interaction acceptance criteria and the documented in the walkdown packages. missile zone-of-influence chart, members This was necessary so that future of the SWT identified items that were changes or relocation of equipment in potential threats to the safety class the room would not generate additional components during a seismic event. seismic concerns. These were either documented as acceptable or classified as requiring The seismic interaction items that a structural upgrade. The documentation are not acceptable were listed on a resulting from the walkdowns to identify special Open Action Item List. These the seismic interaction items was items required detailed stress analysis compiled in a separate file and contains and/or structural upgrades for seismic the following: qualification. Piping and Conduit Several seismic interaction concerns included items attached to the walls a. Problem identification number or ceiling (e.g., fire extinguishers, security cameras, motion detectors, b. Item description overhead mirrors, CAM alarms, and loudspeakers. In most cases these items c. Photograph of the item we.e adequately anchored. For items located overhead, verification of the d. Isometric sketch for each item anchorage is time consuming; therefore, showing the following: it was assumed that these items were

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 77 not tied-down. If the item was located CONCLUSIONS outside the trajectory envelope (Figure The primary objective of the 2), it was acceptable. The trajectory walkdowns was to provide as-built of some items fell within the enveloping structural information of the PFP safety curve. To eliminate detailed class components so that accurate inspections of the anchorage of these seismic evaluations could be performed items, simple calculations were to verify seismic adequacy of the performed to estimate the impact energy components during a DBE. The documents and the corresponding mass necessary generated have been filed in PFP Design to cause the glovebox window panels to Services Group. The majority of the fail. This calculated mass was then resulting discrepancies were related used to screen the item of concern. to the component anchorage, whether it All data and documents generated by be to the floor or support structure. the first member of the SWT were Most of the support structures required verified and approved by the second upgrades to the anchorage system. For member for completeness and accuracy. the most part these structural upgrades were generic in nature. The drawing OBSTACLES ENCOUNTERED numbers used in the walkdowns and Very few equipment obstacles were analyses were listed and distributed to encountered during the walkdowns. PFP Design Services, and potential During the period of time in which the changes to the components listed on walkdowns took place, the metal these drawings must be approved by a production line was in a maintenance structural analysis group. outage, while the product handling line was still in operation. Maintenance Another objective of the walkdowns work on some of the components was in was to identify and resolve potential progress when the walkdown was seismic interaction or II/I items. initiated. There were 149 potential problems identified. Of the 149, simple One problem pertained to support screening criteria developed ahead of members and lead shielding removed by time allowed 144 of these to be disposed the maintenance crew for ease of access of as acceptable. The remaining five in performing their own work. There problems required additional detailed were no warning signs to suggest that calculations. Of these, three were these items were temporarily removed, qualified by analyses and two required which resulted in walkdown discrepancies minor anchor upgrades. and inconsistent analysis results.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

78 Session 4 Modification of Existing Facilities

Second DOE Natural Phenomena Hazards Mitigations Conference - 1989

79 WIND/SEISMIC COMPARISON FOR UPGRADING EXISTING STRUCTURES

Richard A Giller Westinghouse Hanford Company P.O.Box 1970 Richland, WA 99352 ABSTRACT This paper depicts the analysis procedures and methods used to evaluate three existing building structures for extreme wind loads. The three structures involved in this evaluation are located at the U.S.Department of Energy's Hanford Site near Richland, Washington. This site is characterized by open flat grassland with few surrounding obstructions and has extreme winds in lieu of tornados as a design basis accident condition. This group of buildings represents a variety of construction types, including a concrete stack, a concrete load-bearing wall structure, and a rigid steel-frame building. The three structures included in this group have recently been evaluated for response to the design basis earthquake that included non-linear time history effects. The resulting loads and stresses from the wind analyses were compared to the loads and stresses resulting from seismic analyses. This approach eliminated the need to prepare additional capacity calculations that were already contained in the seismic evaluations.

INTRODUCTION test information already completed for The intent of this evaluation was tha seismic evaluation. to qualify three structures for the site design basis wind loads by directly 291Z STRUCTURE comparing building member reactions induced by the wind to those induced by Stack Description the seismic excitation for which the The 291Z exhaust stack is 200-ft building had already been qualified. high and made of reinforced concrete. The wind analyses resulted in It has a base diameter of 18 ft with a component-demand values for moments, 9-inch thick wall that decreases linearly shears, displacements, and stresses. to the top which has a diameter of 14.5 These were compared to the seismically ft with a 6-inch thick wall. The stack generated demand values. As the loads is supported by a deep block foundation imposed on the structure by the seismic resting on a sandy-gravel soil. The excitation were larger than those from stack itself is a low-hazard building the wind, and the structure responds in but is close enough to a high-hazard approximately the same fashion to both facility to cause a seismic II over I phenomena; this procedure eliminated concern. As such the stack was analyzed the need to perform another capacity using the design criteria for a high- calculation. There is also a savings hazard structure. realized by the use of thewalkdown an'1

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

80 Seismic Analysis Procedure The seismic analysis procedure consisted of a simultaneous application of three, free field, acceleration time histories: two horizontal and one vertical. The free field horizontal safe shutdown earthquake (SSE) is a site-specific smooth response spectra curve with a zero-period acceleration of 0.25 g [1]. The free field vertical component SSE is 2/3 of the horizontal SSE spectra. The soil-structure interaction effects on the responses of the stack were modeled using soil springs. Two models were used to calculate the dynamic response; one was a stick model and the other a more refined solid model. The reactions from this seismic excitation were 200 ft compared to the calculated capacities of the stack at reinforcement yield initiation and at ultimate capacity. Performance of the Wind Analysis The wind analysis consisted of simple calculations of cantilever moments and shears at 22.5-ft intervals throughout the height of the stack. These elevations coincide with those used in the seismic analysis. The wind loads were calculated using the American National Standards (ANSI) A58.1 [2] wind formula coefficients with a 90-mi/h extreme wind. These loads were increased by the appropriate load factors as set forth in the UCRL- 15910 [3] guidelines. The results of the wind analysis are plotted on the seismic demand-capacity curves shown ts: in Figure 2. By inspection, the demand from the wind loading is less than both that from the seismic excitation and also the stack cross-section capacity envelope. 236Z PLUTONIUM RECLAMATION FACILTY Building Description The second structure to be evaluated was a 79-ft-square, concrete, bearing wall-type structure that is 48-ft high. It has a 20- by 29-ft Figure 1 penthouse that adds an additional 22 291Z Stack ft to the building height. The

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

81 I' ''.-• ! ij(..'. .'Urin;. • ,;

5. J fin oV r;r ^ i 1 s ?'•: . i j n ; 'i. i £ "0 :.•""; I i 0 H ] T J ' M -. 1 it'' .. 1; 'j,"! r, •:r'u?;:n:?• i rt

O) a3

Reinforcement yield initiation capacity

T 2 4 6 8 10 12 14 16 18 20 Moment (thousands of ft-kips)

CD Figure 2 Concrete Stack Demand-Capacity Curves -a Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

82 structure has 8- to 12-inch thick The detailed evaluation of the concrete walls and 6- to 10-inch thick concrete walls and metal roof panels to concrete floor slabs. It also has an wind loads required additional interior containment cell structure evaluation since the reactions from integral with the building that contains wind pressures in some of these small both 24-inch walls and ceiling slab. panel areas are different than those The roof is a steel/ concrete deck from the seismic response. The dead laminate supported by steel beams and and live load floor moments were open web steel joists. The foundation combined with those calculated from consists of wall footings that are 3 the wind loads using the appropriate to 8 feet below grade and rest on sandy- load factors. The wall and roof gravel soil. reactions were then compared to code strength allowables, and were found to Seismic Analysis Procedure be with in acceptable limits. The seismic analysis was completed by employing the modal superposition response spectra method, using the site specific 0.25 g zero-period ground acceleration spectra in the horizontal directions and 2/3 of this spectra in the vertical direction. Both dead weight and live loads were verified by field walkdowns and used in this evaluation. Two seismic models wera used and the results were enveloped for the final qualification. The first model assumed flexible cross walls, and the second assumed rigid cross walls. Tables 1 and 2 show the results of the flexible cross wall model only, as this model was demonstrated to have the lower capacity. Conservatively the shear forces and moments resulting from the wind loads were compared to those from the seismic model yielding the lowest moments and shear forces. Performance of the Wind Analysis From the wind analysis, simple building cantilever moments, shears, and Figure 3 torsion were calculated for the building 236Z Building Plan using the ANSI A58.1 wind formulas, and UCRL-15910 load factors. The building torsion comes from the wind pressure on the penthouse structure, which is offset 234-5Z PLUTONIUM FINISHING PLANT from the center of building resistance. These moments, shears, and torsion loads Building Description are listed in Tables 3 and 4 and are The third building to be evaluated compared to the appropriate seismic is a three-story, steel-framed building values in Tables 1 and 2. Because the with braced sheet metal walls and roof. resultant wind loads are only 10% to 55% It has overall dimensions of 150 ft in of the seismically induced loads, the the north-south direction, 440 ft in the structure is acceptable. east-west direction and is 47 ft high. There are three internal 8-inch concrete

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

S3 Table 1. N-S Seismic Forces. Elevation Shear force Torsion Bending moment (ft) (kips) (k-ft) (k-ft) 56 39 57 452 46 155 291 1,996 34 323 ,463 5,584 24 1 107 ,096 16,843 12 1,506 ,656 34,871 0 1,763 6,757 55,978

Table 2. E-W Seismic Forces. Elevation Shear Force Torsion Bending moment (ft) (kips) (k-ft) (k-ft) 56 47 288 543 46 186 291 2,401 34 355 11,237 6,645 24 964 17,771 16,144 12 1,273 23,704 31,338 0 1,456 25,305 48,764

Table 3. N-S Wind Forces. Elevation Shear force Torsion Bending moment (ft) (kips) (k-ft) (k-ft) 56 21 0 150 46 33 0 386 34 86 1,589 1,196 24 123 887 2,235 12 162 981 3,153 0 200 1,149 5,662

Table 4. E-W Wind Forces. Elevation Shear force Torsion Bending moment (ft) (kips) (k-ft) (k-ft) 56 15 49 106 46 24 32 273 34 76 936 948 24 112 ,737 1,883 12 151 ,457 2,899 0 189 2,835 5,052

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

84 walls that run east-west tor a The foundation for this building consists substantial length of the building. of square column pedestals and footings These encase some of the columns and that are 4 to 9 feet below grade and rest extend from the ground floor to the on sandy-gravel soil. second floor. These concrete walls add substantial stiffness in the east-west Seismic Analysis Procedure direction but are not connected by The seismic evaluation of this reinforcement to the second floor slab. structure required a rigorous analysis The external sheet metal wall panels that included three-dimensional, dynamic, are fabricated using two metal panels nonlinear techniques. It also with internal insulation and bolted to necessitated cyclic testing of beam to channel girts. The roof deck is a column clip-angle type connections, single layer of sheet metal spot-welded obtaining linear and nonlinear building to the channel roof members, and covered responses, and accounting for the P-A with a built-up tar roofing. The "duct effects for large building level" floor at the 18-ft elevation is displacements. Preliminary linear two- a corrugated sheet metal deck supported dimensional analysis was not adequate by open web steel joists and is not to demonstrate structural adequacy. The considered a competent diaphragm. The results of the two- and three-dimensional second floor, located at the 30-ft seismic and wind analyses are shown in elevation, is a 4-inch concrete slab Table 5. Included in Table 5 are values supported by steel beams. This slab for shear, bending stress, and ties all the columns together and acts displacement resulting from both seismic as a diaphragm for lateral load transfer. and wind evaluations.

•440 ft-

220 ft-

Expansion joint

148 ft -N- 180 ft

10ft r T 42 ft

80 ft -240 ft- -120 ft-

Flgure 4 234-5Z Building Plan

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 85 Performance of the Wind Analysis The evaluation of the wall and roof The evaluation consisted of a two- panel sections required detailed dimensional linear analysis of the calculations based on local wind overall building structure in the weakest pressures. Because of the flexible or north-south direction only. By nature of the sandwich construction of inspection of the seismic analysis the wall panel sections, having spans results and the building structure, the up to 16 ft, the calculations regarding north-south direction was determined to the ability of the panel to remain on be the controlling direction. Building the building during high winds were member stresses, moments, forces, and inconclusive. A testing program was displacements were calculated with the completed to qualify these wall panels. aid of the finite element computer program. This model included the moment Exterior Wall Panel Testing Program rotation capacity of the building The testing program included three connections. The resulting member separate tests scenarios. The panel stresses were compared to American drift tests were completed to determine Institute of Steel Construction (AISC) the effects of building drift on the [4] allowables as well as the stresses integrity of the attachments of the calculated in the seismic analyses, and wall panels. The second series of found to be within acceptable limits. tests were enacted to qualify the The resulting beam-column connection external panels for normal pressure moments were found to be less than the stemming from wind loads. The third actual tested capacities. The building series of tests was to determine the displacements from the wind loads were adequacy of the panels to withstand approximately 1/2 of the displacements the penetration by extreme-wind missiles. resulting from seismic response. As no spare building panels were

Table 5. Plutonium Finishing Plant Seismic and Wind Results Structural Response

Analysis Base Beam Column Type Shear Stress Stress Disp. (lb/in2) (lb/in2) (in.) 2D wind analysis 3lK 10,100 23,700 3..00 D + L + W (static) 2D seismic analysis 58K 15,500 49,400 7,.62 D + L + SSE (static) 2D seismic analysis 40K 11,000 36,800 5..07 D + L + SSE (dynamic) 3D seismic analysis 44K not not 7..20 D + L + SSE (time-history) calculated calculated D - dead load L - live load W - wind load

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 86 available, new panels were fabricated and deformation of the panel (bowing), and tested. bolt hole enlargement was the only damage incurred, but with no loss of function, The panel drift test results showed i.e., the panel remained attached to the the panel to be very stiff in the in- framework. plane shear direction with bolt hole enlargement being the only damage Finally, the missile test qualified experienced when the structure was the panels for an extreme wind missile cyclically tested to 150% of the having 160% of the energy required by the predicted seismic displacement. criteria. Some ripping of the first sheet metal layer and denting of the The tests to simulate external second as well as bolt hole enlargement pressure effects qualified the external were observed, but no penetration was wall panels for 150% of the design basis experienced during the test series. (90 mi/h) wind load. Some plastic

Outside Face

Panel Section

Sheet- metal screws

Channel spacer Detail Section Section A-A

Figure 5 234-5Z Building Wall Panel

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 87 REFERENCES [1] U.S. Department of Energy-Rich!and Operations Office, Hanford Plant Standards, DOE-RL Order 6430.IB, SOC-4.1, Rev. 10, June 1988. [2] American National Standards Institute, Minimum design Loads for Buildings and Other Structures. ANSI A58.1-1982, March 10, 1982. [3] R.P. Kennedy, Design and Evaluation Guidelines for Department of Energy Facilities Subjected to Natural Phenomena Hazards. UCRL-15910, University of California Research Laboratory, Livermore, California, May 1989. [4] American Institute of Steel Construction (AISC), Manual of Steel Construction, 8th Edition, November 1978.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

88 POSTEARTHQUAKE SAFETY EVALUATION OP BUILDINGS AT DOE FACILITIES

Ronald Gallagher R. P. Gallagher Associates, Inc. 116 New Montgomery St., Suite 206 San Francisco, CA 94116

ABSTRACT

New postearthquake building safety evaluation procedures have been developed. The procedures cover inspection and safety assessment of the principal types of building construction found in the U.S., including wood, masonry, tilt-up, concrete, and steel frame structures. Guidelines are also provided for appraising the structural safety significance of ground movements resulting from geologic hazards and for the inspection of nonstructural elements for falling and other hazards.

INTRODUCTION from material production reactors and high- After a damaging earthquake, there is a security plants and laboratories to conventional need to inspect of buildings to assess damage and office buildings, warehouses, and shops. As can be determine if they are safe to occupy and use. The seen from Table 2, a number of DOE sites have Applied Technology Council recently completed a high seismic exposures. Motions in the 0.10 to project to develop procedures for postearthquake 0.20g range and above are normally damaging to building safety evaluation (Refs. 1 and 2). The conventional construction. Sites with high project, designated ATC-20, developed safety exposure, as well as critical facilities at sites with evaluation methodology for the common types of lower exposure, need plans for postearthquake building construction found in the U.S. This is inspections. briefly summarized here, with emphasis on its application to DOE facilities. KNOWN HAZARDOUS FORMS OF CONSTRUCTION The ATC-20 methodology is designed to be Based on past earthquake performance, used by building officials and structural both in the U.S. and abroad, certain types of engineers in the assessment of conventional building construction have had a poor record. building construction. Because of the great number Given below is a brief list of some of the most of structures to inspect and an anticipated shortage hazardous categories: of skilled manpower needed to make the assessments in the immediate aftermath of a Unreinforced masonry (URM) damaging earthquake, the ATC-20 procedures call Nonductile concrete frames for a coarse screening, designated the Rapid Early tilt-up buildings Evaluation, followed, if necessary, by a thorough Early precast structures visual examination, designated the Detailed Concrete frame with URM infill walls Evaluation, by structural engineers. In the event Concrete frames with soft stories that further evaluation is required, the owner must hire a consultant and an Engineering Evaluation In making postevent inspections, knowledge of must be performed. The evaluation techniques are past performance characteristics of the type of summarized in Table 1. building under consideration can be extremely useful. SEISMIC HAZARD AT DOE FACILITIES DOE has a large inventory of structures at many locations around the country. These range

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

89 TABLE 1

ATC-20 Building Evaluation Techniques

Required Example Time Technique Personnel Objective per Building

Rapid Building Rapid assessment of safety. Used to 10-20 Evaluation inspectors quickly post obviously unsafe and minutes Civil/structural apparently safe structures, and to identify engineers buildings requiring a Detailed Evaluation. Architects

Detailed Structural Thorough visual evaluation of 1-4 Evaluation engineers damaged buildings and questionable hours situations. Used to identify buildings requiring an Engineering Evaluation.

Engineering Structural Detailed engineering investigation 1-7 Evaluation engineering of damaged buildings, involving use of days consultant construction drawings, damage data and or more new structural calculations.

TABLE 2

Seismic Hazard at DOE Facilities (Ref. 3)

500 yr. 500 yr Site event Site event

Bendix • 08g FMPC • 10g Los Alamos .18 Oak Ridge .15 Mound .12 Paducah .33 Pantex .08 Portsmouth .08 Rocky Flats .13 Nevada Test Site .21 Sandia/Albuquerque .17 Han ford .09 Sandia/Livermore .41 LBL .55 Pinellas .04 LLNL .41 Argon n,e/East .10 LLNL, Site 300-854 .32 Argonne/West .12 LLNL, Site 300 -834/836 .28 Brookhaven .12 ETEC .53 Princeton .13 SLAC .45 INEL .10 Savannah River .08

Second DOE Natural Phenomena Ha-ards Mitigation Conference - 1989

90 OVERVIEW OF ATC-20 METHODOLOGY necessarily coarse. Inspectors are to look for There are three safety-evaluation readily observable, gross kinds of structural procedures within the ATC-20 methodology: distress, and geotechnical conditions, that threaten building safety. Ordinarily, only the exterior of 1. Rapid Evaluation the building is inspected unless there is a reported 2. Detailed Evaluation or suspected problem. This is done primarily to 3. Engineering Evaluation maximize the number of inspections in the immediate postevent period. The entire procedure Each of these is used for a specific purpose and is summarized in Table 5. must be carried out by qualified individuals (see Table 1). The safety evaluation procedure for Examples of the use of Rapid Evaluation to ordinary buildings is shown diagramatically in determine safety posting and barricading Figure 1. This figure details the usual sequence of requirements are given in Table 6. Significant damage assessment and posting. aftershocks will ordinarily require reinspection and reposting. After undergoing safety-evaluation, buildings are posted with placards as DETAILED EVALUATION INSPECTED, LIMITED ENTRY, or UNSAFE. This method is used to evaluate the safety of This is done to let owners, occupants, and the public damaged buildings and doubtful situations. It is know whether buildings are safe for use. A intended to be used to provide reasonable description of the posting classifications is given assurance that a building, as well as elements of in Table 3. There is also a special posting the building that could cause falling hazards, are category, AREA UNSAFE, used to designate sufficiently safe before it is put back into service. unsafe areas. These may be either inside or Ideally, a Detailed Evaluation should be carried outside a building. For instance, if a fall hazard is out by a team of at least two structural engineers. observed, the area within potential striking In the aftermath of a large quake, however, this distance must be roped off or otherwise barricaded may not be possible. One alternative is the use of a to prevent entry. team consisting of a structural engineer and a building inspector. At DOE facilities, the latter The goal of safety evaluation is to may be a representative of the on-site facilities eventually post every building reviewed either staff. INSPECTED (i.e., safe) or UNSAFE. Buildings posted UNSAFE will require repair or demolition. Because this paper is a great simplification of the ATC-20 methodology, the discussion RAPID EVALUATION presented here is in summary form. Given below Rapid Evaluation is performed on a are general damage evaluation guidelines. These building by evaluating it against six basic criteria are applicable to common types of buildings. The (Table 4). These are externally observable inspection guidelines given require the use of conditions or items of damage that, individually judgment. In occasional instances, a different or collectively, are sufficient to warrant use of the action from that recommended may be warranted. UNSAFE or, in the ca ;e of falling or other hazards, the AREA UNSAFE posting category. 1. Overall Level of Damage. Examine for severely racked walls, whole stories out of If a building is found to have none of the plumb, leaning building, broken foundations, conditions listed in Table 4, and no other hazard is and other massive distress. present, it is apparently safe and posted INSPECTED. If the situation is doubtful and Collapse or Partial Collapse UNSAFE requires more review, the building should be posted LIMITED ENTRY and given a Detailed Building or Individual Story Noticeably Evaluation. Leaning UNSAFE

Because the Rapid Evaluation process is designed to conserve anticipated limited manpower resources, the damage assessments are

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

91 Structure Identified For Evaluation

RAPID EVALUATION

Apparently OK Obviously Unsafe

Questionable

Post Post Post Inspected Limited Entry Unsafe

DETAILED EVALUATION |

Safe, but may need repairs. Unsafe, must be repaired or removed.

Questionable

Post Post Post Inspected Limited Entry Unsafe

ENGINEERING EVALUATION

Safe, but needs repairs. Unsafe, must be repaired or removed.

Post Post Inspected Unsafe

Flowchart of normal building safety evaluation and posting process.

FIGURE 1

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

92 TABLE 3

Building Safety Evaluation Classifications

Posting Classification Description

INSPECTED No apparent hazard found, although repairs may be required. Original lateral load capacity not significantly decreased. No restriction on use or occupancy.

LIMITED ENTRY Dangerous condition believed to exist. Entry by owner permitted only for emergency purposes and only at own risk. No usage on a continuous basis. Entry by public not permitted. Possible major aftershock hazard.

UNSAFE Extreme hazard, may collapse. Imminent danger of collapse from an aftershock. Unsafe for occupancy or entry, except by authorities.

AREA UNSAFE Designated area is unsafe. It must not be entered, except by authorities.

TABLE 4

Rapid Evaluation Criteria

1. Building has collapsed, partially collapsed, or moved off its foundation.

2. Building or any story is s,Tiificantly out of plumb.

3. Obvious severe damage to primary structural members, severe racking of walls, or other signs of severe distress.

4. Obvious parapet, chimney, or other falling hazard present.

5. Large fissures in ground, massive ground movement, or slope displacement present.

6. Other hazard present (e.g., toxic spill, radioactive contamination, broken gas line).

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

93 TABLE 5 Rapid Evaluation Inspection Procedure

1. Examine the entire outside of the structure. 2. Examine the ground in the general area of the structure for fissures, bulged ground, or signs of slope movement. 3. Ordinarily enter a building only when there is a suspected problem or when the structure cannot be viewed sufficiently from the outside. Do not enter obviously unsafe structures. 4. Evaluate the structure using the six criteria (Table 4). Make sure exitways are clear. 5. Post the structure according to the results of the evaluation (e.g., UNSAFE), and place a placard at every entrance to a building classified LIMITED ENTRY or UNSAFE. 6. Explain the significance of UNSAFE or LIMITED ENTRY postings to building occupants, and advise them to leave immediately.

TABLE 6 Examples of Posting and Barricading

Condition Present Post Buildings * In danger of collapse UNSAFE * In danger from collapse of adjacent structure * In danger from geotechnical hazard (e.g., slope failure) * Structurally safe, but other hazard prevents its use (e.g., toxic spill) * Doubtful safety; further LIMITED ENTRY evaluation required. * No apparent hazard present INSPECTED Other Hazards * Falling hazard present AREA UNSAFE * Ruptured gas line * Downed power lines

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

94 Building Off Foundation UNSAFE residual drift in one or more stories should be considered unsafe until shown otherwise. Fractured Foundations UNSAFE Multistory Frame Building with Residual 2. Vertical Load System. Inspect the vertical load Drift UNSAFE system. No structure must be allowed to remain in service if there are any real doubts 5. Degradation of the Structural System. about its ability to safely carry vertical loads. Examine the structural system to determine if Failure of the vertical load system either its strength and stiffness have been reduced by globally (e.g., partial coUapse) or locally (e.g., degradation. This is a particular concern for buckled columns, failed corbel) is generally concrete and masonry structures. Concrete considered grounds for posting the entire frames that experience cracking, spalling, structure unsafe. and local crushing of concrete may have their overall strength greatly degraded even without Columns Noticeably out of Plumb UNSAFE the presence of other failure symptoms (e.g., buckled or out-of-plumb columns, diaphragm Buckled or Failed Columns UNSAFE failure, etc.).

Roof or Floor Framing Separation fromWalls or Seriously Degraded Structural Other Vertical Supports UNSAFE System UNSAFE

Bearing Wall, Pilaster, or Corbel Cracking That 6. Falling Hazards. Examine parapets, Jeopardizes Vertical Support UNSAFE cladding, ornamentation, signs, joists and beams ending on ledgers, interior partitions, Other Failure or Incipient Failure of Significant ceilings, and light fixtures for falling Vertical-Load-Carrying Element UNSAFE hazards.

3. Lateral Load System. First identify and then Falling Hazard AREA UNSAFE inspect the lateral-load-resisting system. Broken X-bracing, severely cracked shear 7. Slope and Foundation Distress. Examine the walls, hinging and massive spalling in ground in the immediate area of the building concrete columns, badly damaged for evidence of mass ground displacements. If diaphragms, and the like are strong evidence foundation distress is observed or is suspected, that the lateral load system may not be viable. or if there are new 1-inch or wider cracks in the Under these conditions, the structure should be foundation, or new differential settlements in considered unsafe until shown otherwise. excess of 1-inch, or if there are fissures more than several inches wide in the vicinity of Broken, Leaning, or Seriously Degraded Moment buildings, a geotechnical engineer or Frames UNSAFE engineering geologist should examine the site and assist with the safety evaluation. Severely Cracked Shear Walls UNSAFE Geotechnical hazards may cover a large area.

Broken or Buckled Vertical Braces ....UNSAFE Base of Building Pulled Apart or Differentially Settled, with Fractured Foundations, Walls, Broken or Seriously Damaged Diaphragms or Floors, or Roof. UNSAFE Horizontal Bracing UNSAFE Building in Zone of Faulting or Suspected Major Other Failure or Incipient Failure of Significant Slope Movement UNSAFE Lateral-Load-Carrying Element UNSAFE Building in Danger of Being Impacted by 4. P-Delta Effects. For tall frame structures, Sliding or Falling Landslide Debris from particularly high-rise buildings, any residual Upslope UNSAFE story drift is generally quite serious, and any frame structure showing a significant

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

95 8- Other Hazards. Examine the structure for drawing review and walk-down inspection. A unsafe conditions such as asbestos release, written inspection plan can be developed which broken fuel lines, chemical spills, etc. coordinates and relates a post-event inspection Operability and safety concerns for fixed with knowledge and insight gained from equipment are given in Table 7. individual building seismic studies and/or inspections. Spill of Known or Suspected Dangerous Materials , AREA UNSAFE Essential facilities such as critical material areas, high security areas, and security Figure 2 illustrates inspection points for and fire stations should be given Detailed some of the building types found at DOE facilities. Evaluations by structural engineers as soon as It is suggested that the interested reader consult possible after the event. Reference 1 for more detailed information and guidance on inspecting and evaluating damaged ACKNOWLEDGEMENTS structures. ATC-20 was prepared for the Applied Technology Council by R. P. Gallagher ENGINEERING EVALUATION Associates, Inc. The work was sponsored by tha After a Detailed Evaluation, any further California Governor's Office of Emergency evaluations would normally be done by a Services, the California Office of Statewide Health structural engineering consultant retained by the Planning and Development, and the Federal owner to prepare an Engineering Evaluation of the Emergency Management Agency. structure. Such a study will typically include detailed reconnaissance and mapping of the REFERENCES damage, preparation of structural calculations, [1] "Procedures for Postearthquake Safety and a quantitative assessment of the strength of the Evaluation of Buildings," Applied Technology damaged structure. It may also involve Council Report ATC-20, Redwood City, CA, 1989. preparation of plans for emergency repairs (e.g., shoring) to enable the structure to be placed back in [2] "Field Manual: Postearthquake Safety- service during the immediate postevent period. Evaluation of Buildings," Applied Technology Council Report ATC-20-1, Redwood City, CA, 1989. RECOMMENDED PRE-EV^N?1 ACTION As part of the overall earthquake [3] "Design and Evaluation Guidelines for preparedness program at DOE facilities, it is Department of Energy Facilities Subjected to desirable to have a pre-arrangement for an Natural Phenomena Hazards," Draft UCLR- engineering team to inspect the facility as soon as 15910, Department of Energy, May 1989. possible after the event. Such an arrangement can begin, after selection of a team, with a review of structural drawings and available seismic design calculations. This should be followed by a tour of the facility so that a high degree of familiarity is achieved. If the design team of record is available, and in close proximity to the site, they would generally make an excellent choice. On-site facilities staff can also be trained to make inspections.

Facilities with hazardous materials, high security requirements, and other characteristics which make them critical facilities should have detailed, facility- specific inspection plans drawn up in advance. Weaker parts of buildings and those areas prone to damage and degradation can often be identified through detailed seismic evaluation, or alternately by a combination of

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

96 TABLE 7 Qperability and Safety Concerns for Fixed Equipment

Item Principal Concerns Main boilers Sliding, broken gas/fuel lines, broken exhaust lines, broken/bent steam and relief lines Chillers Sliding, loss of function, leaking refrigerant Emergency generators Failed vibration isolation mounts; sliding; broken fuel, signal, and power lines leading to loss of function; broken exhaust lines Fuel tanks Sliding or overturning, leaks, broken fuel lines Battery racks Damaged rack, dislodged batteries, acid spill Fire pumps Anchorage failure, misalignment between pump and motor, broken piping On-site water storage Tank or vessel rupture, pipe break Communications equipment Sliding, overturning, or toppling leading to loss of function Main transformers Sliding, oil leak, loss of function Main electrical panels Sliding or overturning, broken or damaged conduit or electrical bus Elevators (traction) Counterweights out of guide rails, cables out of sheaves, dislodged equipment Other fixed equipment Sliding or overturning leading to loss of function (or damage to adjacent equipment)

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

97 Separation ol floorelement ! #AlwcheeK for In-plane Roof doming wpofaflon fiom Okjprvoom warp wapmo of diaphragm vwrtlecMiupport (m Imitonla* plane) Sp

Racking o< piocosi panels (flue to dil't)

Oulwaxl l*anma pontl

Corbel dotnooeleoclng to Story out of promt Column talur* Conw Clocking at op*mngi Wol pond Moaranan from pouMeloaorverticoJvjcport (olowleveO

Precast Tilt-up

•roMnorbucWtcf h b#i

Bfok«n horizontal Separation DfocJng or connecNom wdistresj

Oktmt ot moment connection!, Ooprvogm dkttMt partlculaiVwneie bdti uwd In temkxi

Panel tone ciocung CoKjmm out of plumb •uHdng leaning Beomcor (otonyltvtf) evpomre or remforcemeni Column concrtt* spoino, Sroken or buckled brace, wpoKJt* of («lnforcom»nt broken connecrkx*

Concrete Frame Light-steel Frame

Inspection points for representative structures (Ref. 1). FIGURE 2

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

98 STUDY TO EVALUATE THE FEASIBILITY OF CONSTRUCTING A RETROFIT CONTAINMENT FOR THE 105-L REACTOR AT THE SAVANNAH RIVER PLANT

Robert D. Quinn NUTECH Engineers, Inc. 14S Martinvale Lane San Jose, CA 95119

ABSTRACT

This paper presents a summary of a study performed to determine the feasibility of constructing a retrofit containment dome meeting the requirements of the ASME Boiler and Pressure Vessel Code for nuclear containment vessels over the existing Savannah River 105-L reactor. Using existing large dome structures as a guide, design concepts were developed and analyses performed to evaluate the structural feasibility of containment dome structures. Construction schedules and costs were estimated to assess financial feasibility as well. It was concluded that such a retrofit containment dome was structurally feasible and within the capabilities of present day construction technology.

INTRODUCTION

The 105-L reactor is one of three, heavy water moderated, production nuclear reactors which have been operating at the Savannah River Plant. The primary purpose of these reactors is to produce special nuclear materials for use in the nation's defense program. Because of the role of the 105-L reactor, its design and basic operating conditions are much different and much less impactive than the more numerous commercial nuclear power plants in the United States. The basic operating conditions of a commercial boiling water reactor (BWR) are 1000 psi and 500*F, and those of a pressurized water reactor (PWR) are 2000 psi and 575'F. The 105-L reactor operates at a maximum pressure of 150 psi and a maximum temperature of 140'C (284*F). The effects, therefore, of a loss of coolant accident (LOCA), which is the design basis accident (DBA) condition for containment design and is defined as an instantaneous break in a reactor coolant line, 7 are much less severe. In 1987, an investigation was performed to determine the feasibility of constructing a containment structure over (he Figure 1 - Plot Plan of 105-L Containment Dome existing 105-L reactor to mitigate the small probability of a release of any radioactive material should a LOCA occur. ancc activities. A plot plan and section of the 105-L reactor The containment structure was envisioned as a free standing and the placement of the 700 fool diameter containment dome approximately 700 feet in diameter and approximately structure is shown in Figures 1 and 2. An isometric view of 200 feet high. The 700 foot diameter permits the the 105-L facility enclosed in the dome is shown in Figure 3. containment to completely encircle the existing 105-L The scope of the feasibility study only included reactor structure and permit normal operations and maintcn- evaluation of structural design and construction feasibility

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

99 Title 10, Code of Federal Regulations, Part 50 (10CFR5Q), Appendix A [1], Criterion 16 governs the design of containments for commercial nuclear reactors. It defines the term containment or containment structure as a specific plant feature which consists of "an essentially leak- * tight barrier against the uncontrolled release of radioactivity • to the environment...". The U.S. Nuclear Regulatory Commission (NRC) is miS=== A *e3W* 1| charged with administering these requirements and has 700' 0- DIAMETER J developed Regulatory Guides to describe and document acceptable methods to implement the 64 criteria contained in 10CFR50, Appendix A. In addition, the NRC has published a series of Standard Review Plans to document how Figure 2 - Sectional View of 105-L Containment Dome applications to construct and operate a commercial nuclear power plant will be reviewed. These two series of for an ASME containment vessel dome structure. Other documents form the framework for commercial nuclear items required for a fully functional containment system, power plant design, including containment design. such as the basemat, ventilation, security, lighting, and In addition, the American Society of Mechanical instrumentation, as well as operational considerations such Engineers (ASME) has developed detailed standards for the as surveillance and maintenance, were not included as part analysis, design, and fabrication of nuclear power plant of the feasibility study and are not discussed herein. components. This standard, the ASME Boiler and Pressure Vessel Code (ASME Code), Section III, "Nuclear Power DESIGN CRITERIA Plant Components" [2], addresses both steel and concrete containments, and has been incorporated into the Regulatory BASIS Guides and Standard Review Plans by reference. It is the The initial task in performing the study was to define the basic document used to analyze, design and build contain- criteria to be used for the feasibility analysis and design. ments, but because of its application to commercial nuclear Maximum use was made of information and methods power plants many of its provisions have specific references utilized by commercial nuclear power plants. to features unique to commercial nuclear power plants.

Figure 3 - Isometric View of 105-L Facility Enclosed in Containment Dome

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

100 DEFINITIONS any containment design alternative for the 105-L reactor In general, the criieria for the analysis and design of a must be retrofitted to the existing facility. Since it is containment can he subdivided into three categories as desirable to minimize changes to the existing facility, the follows: containment dome concept was conceived. This structure would completely enclose the 105-L reactor and the 1. The effects and requirements for operating the immediately adjacent supporting facilities. The dome would reactor, be approximately 700 feet in diameter and 200 feet high. 2. The effects of natural phenomena, and While the design of a structure of this size may present 3. The effects an accident would have on the structural and economic problems, it has a benefit in having containment. a volume of almost 49 million cubic feet. This reduces the design pressure and temperature requirements for the The first of these categories includes live loads, dead postulated accidental energy release from a LOCA to very loads, the configurational requirements such as doors and small levels. These levels are much less than the design hatches needed to operate and service the reactor, the effects requirements for contemporary commercial nuclear power of normal reactor operation on the containment, and testing plant containments. requirements. Natural phenomena will effee1. all structures including DESIGN CODES the containment. Criteria development for containments As previously stated, Section HI of the ASME Code considers the occurrence of very low probability natural includes requirements for both steel and concrete phenomena including large earthquakes, tornados, flooding, containments. The materials which were considered in the and tidal waves. feasibility study for the construction of a containment The main purpose of a containment is to protect the included steel and reinforced concrete. The rules applied to public from the consequences of an accident. The structure the design of the containment dome for each of these should contain the energy and radioactivity inside were it to materials are discussed below. be suddenly released from the reactor as a result of an The ASME Code, Section III, Division I, Subsection NE accident. This is referred to as the design basis accident specifically addresses the design of steel containments. The (DBA) and is postulated by 10CFR50 Appendix A [1] as a foundation and other concrete structures which are not a part sudden lack of cooling water resulting from a "double-ended of the containment boundary are to be designed to the rupture of the largest pipe of the reactor coolant system." requirements of ACI 349 [3], which is the American This is referred to as a loss-of-coolant accident (LOCA). In Concrete Institute's design and construction code for nuclear a LOCA, energy is released as superheated steam. The safety related concrete structures. This code was selected as pressure retaining capacity of the containment required to being an appropriate compromise between the importance contain this steam will depend upon the volume of the and safety significance of these structures and the non- containment. Containing the steam in a small volume will pressure retaining nature of their function. The foundation require a containment structure capable of withstanding high and other appurtenance structures, below the level of the pressures, whereas containing the same amount of steam in a containment dome, will be similar for each construction very large volume will require a structure capable of concept. The applicable codes and standards used for their withstanding low pressures. design will be identical for each concept studied. The containments for commercial nuclear power plants The ASME Code, Section III, Division 2, Subsection CC have evolved to a balance between volume and strength on specifically addresses the design of reinforced concrete an economic basis. The present generation of commercial containments. boiling water reactors (BWRs) have a containment structure with a volume of approximately 1.8 million cubic feet. In DESIGN LOADS addition, BWR containments include a pressure suppression As indicated previously, the loads for the analysis and system to reduce the pressure and temperature effects which design of the containment come from three general the release of the superheated steam has on this containment. conditions and include loads with a very small probability of The containment is designed for a pressures and occurrence. All loads that have been addressed in the temperatures on the order of 50 psig and 185T as a result of Regulatory Guides and Standard Review Plans for the a LOCA. analysis and design of containments are discussed below. In a commercial nuclear power plant, the configuration Some of these loads are not applicable to the analysis and of the containment and the design of the reactor system (the design of this containment but are included below for Nuclear Steam Supply System or NSSS) proceed completeness. concurrently, and the final design is the most economic Dead loads are the weight of the structure and any fixed combination of NSSS and containment designs. By contrast, equipment. A tabulation of the weights of various building

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

101 materials is contained in Appendix A of American National Department of Energy Sites" [8] which is referenced in Site Standard, ANSI A58,1-1982, "Minimum Design Loads for Specification No. 70%. Building and Other Structures" [4], There is no fixed Tornado missile criteria was taken from Site equipment that is presently anticipated to be attached to the Specification No. 7096 without modification. containment dome structure. The effects of a design basis accident (DBA) or loss of Live loads consist of nonpermanent loads associated coolant accident (LOCA) were defined as an increase in with the use or occupancy of the structure. Because of the temperature of 3*F and an increase in pressure of 1.0 psig. unique size, configuration, and use of the containment dome In commercial nuclear power plants the containment structure, conventional live loads arc not applicable, A 500 structure is in close proximity to reactor coolant system 1b. roof (dome) live load is assumed for any maintenance or piping. Since this is not the case for the Savannah River inspection requirements. For the purpose of load containment design, it was assumed there would be no other combinations, snow loads are considered live loads. Snow accident related loads. These loads which are not loads are in accordance with ANSI A58.1-1982 [4]. considered include jet impingement, pipe whip impact, and Containments as leak tight structures are normally missile impact. maintained at a negative pressure with respect to the It was assumed that there would be no post accident atmosphere in order to reduce or eliminate the possible flooding of the containment due to the size and volume of release of radioactive material to the environment. This the containment in comparison to the 105-L reactor. negative internal pressure is assumed as 0.25 psig. In For the design of rigid, subsurface structures it was addition, the containment will be tested to ensure its leak assumed that an equivalent fluid pressure of 30 lbs/cubic lightness at a positive intcnal pressure of 1.25 psig. foot would account for any lateral soil pressures. The Design temperatures were obtained from ASHRAE available site soil boring logs did not indicate any ground Handbook [5] and arc the 99th pcrccntile values for the water near the foundation, therefore it was assumed there Savannah River Plant area. would be no hydrostatic pressure or buoyancy. Seismic criteria was obtained from Site Specification There are no piping or equipment loads because none of No. 7096, "Structural Specification for Building Code the reactor operating piping or equipment is attached to the Requirements", Revision 3 [6], In order to adapt the containment structure. It is all contained within the 10S-L information contained in Site Specification No, 7096 to the reactor building. The weight of non-reactor operational format of the ASME Code, some changes in terminology piping and equipment, if applicable, would be considered as were required. The ASME Code and supporting material a dead load. refer to two seismic criteria. As defined in 10CFR100 For feasibility study purposes, it was assumed that Appendix A [7], an operating basis earthquake (OBE) is that specific hurricane loads were not applicable and that any seismic event that "could reasonably be expected to affect hurricane severe wind phenomena would be less than that the plant site during the operating life of the plant", and for assumed for tornado loads. "which those features of the nuclear power plant necessary As the Savannah River Plant is at least 75 miles from the for the continued operation without undue risk to the health Atlantic Ocean it was assumed that no tsunami (tidal wave) and safety of the public are designed to remain functional". was applicable. A safe shutdown earthquake (SSE) is the largest possible Aircraft impact loads are considered in commercial acceleration that could occur at the site due to any postulated nuclear power plant design if the distance from a plant site fault activity, and for "which structures, systems and to the end of a runway is less than 10 miles. This condition components important to safety are designed to remain does not exist at the 105-L reactor, so no aircraft impact load functional". were considered to be applicable. For the purpose of this study the SSE was assumed to be Because the proposed containment dome will not be the Design Basis Earthquake (DBE) specified by the Site enclosed by another structure there arc no additional Specification and the OBE was assumed to be one half that external pressure or temperature loads not considered in value. This is consistent with 10CFR100 Appendix A and previous load definitions. the previous revision to the site specification. Wind loads are determined by the analytical method LOAD COMBINATIONS AND ACCEPTANCE indicated in ANSI A58.1-1982 [4] for a maximum wind CRITERIA speed of 83 mph. For the purposes of this study, separate Load combinations were developed from Standard tornado parameters were calculated using 210 mph as the Review Plan, Section 3.8. i "Concrete Containment" [9], and maximum horizontal wind speed. Specific tornado data was Section 3.8.2, "Steel Containment" |1()|. They arc divided developed from "Natural Phenomena Hazards Modeling into various "service limits" depending upon the probability Project: Extreme Wind/Tornado Hazard Models for of occurrence of ihc load or load combinations. These service limits are defined in the ASME Code. The Code

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

102 also defines the allowable stress limits for each service limit. Costs of the three major large dome structures The combination of the probability of occurrence of a load constructed in the United States are listed below. These or load combination and allowable stress is intended to costs have been escalated to 1987 dollars and arc adjusted define a constant level of risk. This has been identified as for cost in the Savannah River area. 10-7 for nuclear power plants as compared to 10-3 for conventional structures. Year 1987 Cost at Structure Location Completed Savannah River FEASIBILITY EVALUATION Astrodome Houston, TX 1966 $ 110 million EXISTING LARGE DOME STRUCTURES Superdome New Orleans, LA 1975 $ 320 million Several large domes have been constructed recently in Kingdome Seattle, W A 1976 $ 94 million the United States which are similar in size to the proposed containment dome for the 105-L reactor. These domes were The completed costs of these structures include the costs constructed as weather protection for large sports for seating, mechanical systems, and other facilities complexes. They include the Kingdome in Seattle, associated with a sports arena. The costs of the dome Washington; the Astrodome in Houston, Texas; and the structure and foundation run about 35-50% of the total cost. Supcrdome in New Orleans, Louisiana. However, these construction costs are for normal commercial construction and do not account for the Structural Feasibility of Large Dome Structures additional costs associated with nuclear facilities. A factor Each of the three large dome structures listed above is of 3 on commercial construction costs in estimating the cost comparable in size to the containment dome being for a nuclear facility is not unreasonable. Therefore, for considered for the Savannah River 105-L reactor. The fact purposes of comparison, the costs listed above are thai these dome structures have been completed and are considered fairly representative of what a nuclear successfully fulfilling their design functions indicates that it containment dome structure may cost. is indeed structurally feasible to construct a containment dome of the size under consideration. DESIGN CONCEPTS Of the three dome structures mentioned above, the Kingdomc is most similar in size to the containment dome Design Approach being considered for Savannah River, and is discussed in This section discusses the analysis methods and design more detail below. approach used in the development of a conceptual design for The Kingdome is a reinforced concrete dome structure the 700 foot diameter containment dome. The analytical which is 670 feet in diameter and 250 feet high. The vertical methods and design approach used are similar for the steel sidewalls aie 140 feet high and contain ramps for spectator and concrete containment dome concepts. access. The dome roof is 110 feet high and consists of 40 The geometry of a dome structure is such that it exhibits upstanding concrete ribs approximately two feet wide by six a characteristic known as axisymmetry. The analysis of feet high with a 5 inch thick concrete slab in a hyperbolic such structures can be greatly simplified as a result. Several paraboloid shape between each rib. The curvature of the 5 structural analysis programs are available for axisymmetric inch slab between ribs provides stability against buckling analysis. For purposes of this feasibility study, the and allows the use of the economical thin slab. axisymmetric finile difference program BOSOR4 [11] was Mr. Jack V. Christiansen of Seattle, the designer of the used to perform major portions of the analysis. Other Kingdome, was contacted as part of this feasibility study. analytical work was performed using conventional hand After discussions with him regarding the requirements for analysis techniques. the 105-L reactor dome, it is fell lhat structurally and Each of the applicable loadings and load combinations configurationally ihe domes were similar and thai ihe same was investigated for the containment dome analysis. Those concepts used for the design and construction of the loads which were determined to be most important to ihe Kingdome could be adapted for the successful construction design of the containment dome were identified. The effects of a containment dome over ihe 105-L reactor. of these loadings were evaluated in greater detail. Due to the large span of the containment dome structure, Cost of Large Dome Structures the governing criteria for structural adequacy is the stability Another factor in the feasibility of constructing a large of the dome. Stability refers to the ability of the structure to dome structure is the cost of ihe structure. Prohibitively carry loads without sudden extreme deformations or high costs would make the construction of such a structure collapse. The phenomenon of a structure transitioning from unattractive. a stable condition to an unstable condition is referred to as buckling. In general, Ihc type of loading which is important

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

103 when assessing structural stability is a comprcssivc load, dome geometry was used as the basic design concept for the such as external pressure on a sphere. As a result, the loads steel dome, and spherical dome geometry was used for the which proved to be most important for the design of the concrete dome. dome were those which caused comprcssive stresses. Those The next step in the design development considered a loads included dead load, live loads (including snow loads), shell with meridional stiffening only. The meridional negative internal pressure loads, seismic loads, and wind stiffencrs increased the buckling capacity, however their loads. effectiveness was limited and further increases in the size of The computer program BOSOR4 was used to perform the stiffeners within practical size limits had little effect on both axisymmctric bifurcation and snap-through buckling the buckling capacity. Circumferential stiffening was then analyses. Snap-through buckling refers to the axisymmetric added to the dome structure. The circumferential stiffening collapse of a dome structure, while bifurcation buckling provided a substantial increase in the buckling capacity of refers to a condition in which the shell goes from a smooth the structure. surface to a wave shape. The buckling analyses were done Various combinations of meridional and circumferential by applying a uniform external pressure load to the stiffeners were evaluated to develop a balanced design analytical model and then incrementing the load until the which provided the required compressive stress capacity structure buckled. A number of buckled waveforms were while minimizing the number of stiffeners required. investigated to determine the minimum buckling load and In accordance with the objectives of this study, designs the stress at which the buckling occurred. were developed only far enough to demonstrate structural The analytical model predicts a buckling load and stress feasibility. Additional efforts to provide an optimized for a mathematically perfect model, i.e. a model with no design were not performed. As such, the design concepts material or geometric imperfections. In addition, no factor documented in the following sections do not necessarily of safety is assumed in the BOSOR4 analysis. To develop represent an optimum design, but are designs which meet an allowable comprcssive shell stress which accounts for the the functional and operating requirements for the effects of material and geometric imperfections and provides containment dome for the Savannah River 105-L reactor as a factor of safety, knock-down factors were developed defined above. which are applied to the analytical results. The knock-down factor for the steel dome was developed by evaluating the Reinforced Concrete Design allowable external pressure for a sphere the diameter of the The reinforced concrete containment dome design containment dome in accordance with the ASME Code, developed in this feasibility study for the Savannah River Section III, Division 1, Subsection NE. The theoretical 105-L reactor consists of a 700 foot diameter reinforced critical buckling pressure for the same sphere was also concrete shell, meridional stiffeners (ribs), and calculated using classical solution techniques. The ratio circumferential stiffeners (rings). A thin steel plate lines the between the classical buckling pressure and the ASME Code inside of the concrete shell for pressure retention as required allowable external pressure was used as the knock-down by the ASME Boiler and Pressure Vessel Code, Section III factor. [2], Division 2, Subsection CC. For the concrete containment, the knock-down factor The shell and liner are spherical in shape. The depth of was developed using the criteria in ACI publication SP-67, the concrete shell varies from 2 feet at the tension ring to 1 "Concrete Shell Buckling" [12]. foot at the top of the dome, while the steel liner is a constant The dome design must meet the requirements for 1/4 inch thick. Thirty-six ribs nominally spaced 10* apart allowable stress as determined from the applicable codes stiffen the shell from the tension ring at the base of the dome identified above. As discussed previously, the governing to the 20 foot diameter compression ring at the top. These factor for the design of the dome is stability. The design rectangular ribs vary from 6 feet deep to 3 feet deep from approach developed a containment dome concept for which bottom to top. The rectangular rings are nominally spaced the maximum compressive stress, caused by the loadings 50 feet apart and have the same depth as the ribs at each defined in the design criteria document, was less than the meridional location. The rings are non-continuous with allowable compressive stress. segments spanning from rib to rib around the circumference. An unstiffened "clean shell" dome was used as the basic This design assumes 4,000 psi concrete and 60,000 psi steel design model for developing the conceptual design details. for the reinforced concrete sections and type 304L stainless Both spherical and torispherical dome geometries were steel for the liner. Details of the reinforced concrete evaluated. The buckling capacity of the unstiffened dome containment dome arc shown in Figure 4. was less than required for the compressivc design loads, but As can be seen from Figure 4, the spherical shell and ribs provided a measure of how much stiffening was required to frame into a 8 foot deep tension ring for lateral stability. provide adequate compressivc load carrying capability. The tension ring is supported on a 2 foot thick concrete ring Based on this initial clean shell evaluation, the torispherical wall augmented by 6 foot deep pilasters at each stiffener

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

104 constructed from 1-1/4 inch plate with a total depth of 3'-6" are nominally spaced every 10 feel. The stringers, which arc also 1-1/4 inch thick T-scctions, are 3'-0" deep and are nominally spaced 8 feet apart. The steel material assumed in this design is ASME SAS16, Grade 70 carbon steel with a minimum yield strength of 38,000 psi and an ultimate tensile strength of 70,000 psi. Details of the steel containment dome are shown in Figure 5. As can be seen from Figure 5, a torispherical dome is vertical where it meets with the concrete ring wall. The CIRCUMFERENTIAL RING advantage of this configuration is that nearly all the load CONCRETE RIB being transferred from the dome to the wall is vertical, CONCRETE SHELL resulting in minimal overturning and thrust loads on the wall and simplifying the wall and foundation design. TENSION RING The ring wall has been conceptualized as a 35 foot high, 2 foot thick wall with reinforcing pilasters. The wall and 2' CONCRETE WALL foundation design concept for the steel dome is the same as 61 CONCRETE PILASTER that shown for the concrete dome shown in Figure 4. GRADE 4' THK. FOUNDATION 18" PILINGS

RING STIFFENER 14'-0 MERIDIONAL STIFFENER -J Figure 4 - Concrete Containment Dome Detail /r- STIFFENER FLANGE /— STIFFENER WEB location. The wall and foundation design is also shown in e Figure 4. yz ^— SHELL r The conceptual design described in the previous section /- S.S. LINER E \ and analyzed as part of this study is a simple, conservative © *•• 'i i i' concept chosen primarily to demonstrate feasibility. i ii • Evaluating the design of the Seattle Kingdome [13][14], it — REINFORCED was detennined that the concrete dome conceptualized in CONCRETE WALL this study has many similarities with the Kingdome. For example, stability proved to be a significant factor in both V" designs. The circumferential rings provide stability against 6'-0" buckling for the dome analyzed in this study, while the 1 hyperbolic paraboloid shape between each rib on the Kingdome provides similar stability for that structure. This further confirms the feasibility of a reinforced concrete Figure 5 - Steel Containment Dome Detail dome, and suggests potential for development of a more economical design concept during future design phases. CONSTRUCTION SCHEDULES AND COSTS The construction schedule estimates were developed Structural Steel Design based on standard construction equipment and techniques. The structural steel containment dome design developed The actual construction schedule will vary based on final in this feasibility study for the Savannah River 105-L reactor design configuration, available construction equipment and consists of a 700 foot diameter steel shell reinforced with techniques, owner scheduling requirements, etc. meridional stiffeners (stringers) and circumferential The construction cost estimates were made using stiffeners (rings). The shell is torisphcrical in shape, and is generally accepted estimating techniques. The actual cost to fabricated from 1-1/4 inch thick shell plate. T-shaped rings construct the final design containment dome may vary.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

105 depending upon the final design configuration, scope of Once sufficient tec sections have been erected, placing work and owner needs, etc. the precast circumferential beams and roof planks can begin. The liner plate scams between adjacent panels are welded, Reinforced Concrete and the remaining slab rcbar placed. Concrete is then The general sequence of events for the construction of a pumped from the base to complete the roof and stiffening concrete containment dome for the Savannah River 105-L rings. This operation must be performed in a symmetrical reactor is described below. A perspective view of a concrete pattern to assure a uniform load increase on the structure and containment dome under construction is shown in Figure 6. prevent failure due to nonsymmetrical load conditions. The first steps in construction will be to establish general On completion of concreting the temporary supports and on-site working areas including a precasting yard and batch erection cranes are removed, and the surface of the dome plant facilities. Construction will proceed with the covered with an appropriate single ply roofing material for excavation of foundations, pile driving, and pouring of the weather protection. The liner plate is painted to provide foundation. This construction includes the concrete ring protection against corrosion. wall with a tension ring at the top with appropriate The dome should then be tested for pressure and leak connections for the containment dome and stiffeners, as.well tightness. as airlocks and hatches. During this time work can begin on The schedule for construction of the concrete precasting of the dome tee sections which consist of a rib containment dome is estimated to be 4.5 years, including (meridional stiffener) and a portion of the slab, and the dome final design and preparation of construction specifications. roof planks to be- installed between the tee sections. The It is assumed in the schedule that no NRC approval cycle is dome liner plate is cast integrally into both the tee sections required. and roof planks. The estimated cost for the concrete containment dome Once the ring wall is complete and the compression ring concept includes the material costs of the ring wall and is supported in position by an appropriate support structure, foundation, the concrete dome structure, and the airlocks placement of the precast concrete tee sections can proceed. and hatches. Estimated costs for establishing on site Once four tee sections have been placed, the central facilities, erection equipment, painting, final engineering supporting tower should be lowered to avoid transferring design and an allowance for site manager costs have also large loads to the temporary support structure. The been included. The estimated cost for a concrete dome is remaining tee sections are erected to maintain a symmetrical $270 million. pattern of load on the compression and tension rings.

Figure 6 - Construction Perspective of Concrete Dome

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

106 Figure 7 - Construction Perspective of Steel Dome

Structural Steel and preparation of construction specifications. It is assumed The general sequence of events for the construction of a in the schedule that no NRC approval cycle is required. structural steel containment dome for the Savannah River The estimated cost for the structural steel containment 105-L reactor is described below. A perspective view of a dome concept includes the material costs of the ring wall partially completed steel containment dome is shown in and foundation, the steel dome structure, and the airlocks Figure 7. and hatches. Estimated costs have also been included for The first steps in construction will be to establish a items such as establishing on-site facilities, temporary subassembly fabrication facility on-site, and to proceed with fixtures and erection equipment, painting, fabricator design fabrication of subassemblies for later lifts. Also, a concrete costs, freight, final design, and an allowance for site batch plant facility or arrangements for procurement of manager costs. The final estimated cost is $320 million. ASME qualified concrete from an off-site supplier must be completed for construction of the concrete foundation and CONCLUSIONS ring wall as discussed previously. Once the ring wall including airlocks and hatches is The conclusion of this study is that a containment dome completed, the knuckle (toroidal) section of the dome can be meeting NRC regulations and requirements, can feasibly be erected. While this work is in progress, the circumferential constructed over the 105-L reactor at the Savannah River dome subassemblies will be assembled and welded. A set of Plant. The results of the study indicate that either reinforced temporary moveable interior supports are required to support concrete or structural steel could be used for the construction the circumferential subassemblies as they are erected and of the dome, with reinforced concrete offering the greater welded. Each complete circumferential ring must be economy. These conclusions are supported by preliminary completed prior to starting on the next ring. Subsequent structural calculations and experience with existing dome subassemblies will be lifted over the knuckle with a comparable large dome structures. A perspective view of crane and moved up the dome on a carriage. Continue to the completed concrete containment dome concept is shown construct dome rings until the dome is complete. in Figure 8. The surface of the dome can then be covered with an appropriate single ply roofing material or painted for It is further concluded that the dome could be erected weather protection. with minimal impact on continuing operations of the 105-L The dome should then be tested for pressure and leak reactor. The construction methods envisioned by this study tightness. postulated no significant effect or modifications required on The schedule for construction of the steel containment existing 105-L reactor facilities. dome is estimated to be 4.7S years, including final design

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

107 Figure 8 - Perspective of Completed Containment Dome

REFERENCES

[1] Title 10, Code of Federal Regulations, Part. 50, [8] D. W. Coats and R. C. Murray, "Natural Appendix A, "General Design Criteria for Nuclear Phenomena Hazards Modeling Project: Extreme Power Plants", Office of the Federal Register, Wind/Tornado Hazard Models for Department of National Archives and Records Administration, Energy Sites," Lawrence Livermore National 1986. Laboratory, August 1985. [2] American Society of Mechanical Engineers, [9] Standard Review Plan, Section 3.8.1, "Concrete "Boiler and Pressure Vessel Code", Section III, Containment", U.S. Nuclear Regulatory 1986 Edition. Commission, Office of Nuclear Reactor [3] American Concrete Institute, "Code Requirements Regulation. for Nuclear Safety Related Concrete", ACI-349, [10] Standard Review Plan, Section 3.8.2, "Steel 1985. Containment", U.S. Nuclear Regulatory [4] American National Standards Institute, "Minimum Commission, Office of Nuclear Reactor Design Loads for Buildings and Other Structures," Regulation. ANSI A58.1,1982. [11] D. Bushnell, "BOSOR4: Program for Stress, [5] American Society of Heating, Refrigerating and Buck'ing, and Vibration of Complex Shells of Air-Condiuoning Engineers, ASHRAE Handbook, Revolution", Structural Mechanics Software Series. 1981. Volume 1, pp. 11-143. [6] E. I. du Pont de Nemours and Company, "Structural [12] American Concrete Institute, "Concrete Shell Specification for Building Code Requirements, All Buckling", SP-67. Projects, Savannah River Plant", Specification [13] J. V. Christiansen, "The Kingdome," presented at 70%, Revision 3. ASCE Fall Convention, San Francisco, California, [7] Title 10, Code of Federal Regulations, Part 100, October 1977. Appendix A, "Reactor Site Criteria", Office of the [14] D. P. Billington, Thin Shell Concrete Structures. Federal Register, National Archives and Records Second Edition, McGraw Hill Book Company, Administration, 1986. 1982.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

108 PRACTICAL APPROACHES TO IMPLEMENTING FACILITY WIDE EQUIPMENT STRENGTHENING PROGRAMS

Raymond H. Kincaid Elwood A. Smietana EQE Engineering 3150 Bristol Street, Ste. 350 Costa Mesa, California 92626

ABSTRACT

Equipment strengthening programs typically focus on components required to ensure operability of safety related equipment or to prevent the release of toxic substances. Survival of non-safety related equipment may also be crucial to ensure rapid recovery and minimize business interruption losses. Implementing a strengthening program for non-safety related equipment can be difficult due to the large amounts of equipment involved and limited budget availability.

EQE has successfully implemented comprehensive equipment strengthening programs for a number of California corporations. Many of the lessons learned from these projects are applicable to DOE facilities. These include techniques for prioritizing equipment and three general methodologies for anchoring equipment. Pros and cons of each anchorage approach are presented along with typical equipment strengthening costs.

INTRODUCTION

Equipment seismic retrofit has historically 1) to help facility engineers understand focussed on strengthening of components typical earthquake damage to equipment, and essential to operability of safety related 2) present conceptual sketches showing equipment or to prevent the release of approaches which can be used to help protect radioactive materials. Usually these items equipment systems against earthquake damage. comprise only a small percentage of the It is primarily intended for non-safety related equipment contained within a facility. equipment at General Use or Important or Survival of non-safety related equipment may Low Hazard facilities. also be essential for continued facility operation following a major earthquake. Strengthening of non-safety related Often, inadequate attention has been paid to equipment (e.g. switchgear, transformers, air anchoring or bracing these components. handlers, unique production equipment, etc.) can significantly minimize losses and A previous report [1] entitled, "Practical downtime following a major earthquake and Equipment Seismic Upgrade and reduce the risk of an extended business Strengthening Guidelines", was developed for interruption. The major problem in Lawrence Livermore National Laboratory to implementing this type of a program is how to aid DOE facility engineers in addressing this efficiently manage the large amounts of issue. The purpose of this report was twofold:

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

109 equipment which can be strengthened However, unanchorcd components arc throughout a facility. very vulnerable. Typical observed failure modes include tipping or sliding This paper addresses these issues and and tearing of attached cabling. presents three different techniques for implementing a comprehensive facility wide Piping. Damage normally occurs due strengthening program. It is based upon work to differential displacements imposed EQE has performed for a number of on the piping system by the support California corporations. Many of these structure. Failure usually initiates at approaches are directly applicable to DOE joints. Welded steel pipe lines facilities. generally perform well while threaded or PVC pipe is vulnerable to damage. EQUIPMENT SUSCEPTIBILITY TO EARTHQUAKE DAMAGE Tanks. Tanks can tip over or slide due to inadequate bracing or anchorage. California normally comes to mind as the Tank movement can fail attached highest earthquake risk area in the United piping resulting in release of toxic or States. However, there are many other regions flammable liquids. which are also at risk. These include Puget Sound, Salt Lake City, the New Madrid area Vibration Isolated Equipment. Simple in the Central United States, and Charleston, vertical spring vibration isolators used South Carolina. Many DOE sites arc located on air handlers, pumps, emergency in these areas of moderate to high seismic risk generators, etc. arc very susceptible to and are susceptible to earthquake damage. failure. Equipment misalignment and damage to attached piping or cabling Seismic experience data have shown that may result when the equipment falls properly anchored and braced industrial-grade off the vertical isolator. equipment generally has an inherent seismic ruggedness and demonstrated capability to Strengthening equipment to mitigate these withstand significant seismic motion without earthquake damage modes is straightforward. damage. However, in the normal Seismic retrofit usually involves adding manufacturing environment, equipment is additional anchorage or bracing and/or usually unanchored and very susceptible to providing adequate flexibility in attached damage. Typical observed damage modes cabling or piping. Snubbers can be used to include: limit displacement of vibration isolated components. The trick is to do it cost Computer Facilities. Data centers and effectively. local networks often are very susceptible to earthquake damage. CONSIDERATIONS IN DEVELOPING AN Losses include both direct damage as EQUIPMENT STRENGTHENING PROGRAM well as business interruption resulting from facility downtime. Computer Practical difficulties in implementing a equipment tends to be particularly comprehensive equipment strengthening vulnerable to damage since it is program can keep it from ever getting off the relatively tall and heavy and may tip ground. Typical issues include: How do you over or slide during ground shaking. handle the large volumes of equipment Large, raised computer floors may potentially requiring strengthening with only- collapse due to a lack of lateral very limited funding? How do you increase bracing. awareness and interest in the subject of non- safety related equipment retrofit? How do Electrical Equipment. Properly braced you set objectives and prioritize equipment and anchored electrical equipment has requiring strengthening? The potential list of an excellent performance record.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

110 issues to be addressed may be long and 4. Revise priorities based upon senior complex. management review.

Objectives or strengthening will depend Priorities are assigned based upon a on organizational goals. Protecting the life number of factors including hazard posed to and safety of employees and the public is nearby personnel, component value, usually of foremost importance. Minimizing importance to overall operations, and direct losses and downtime may also be adequacy of existing anchorage and bracing. critical. In many cases, downtime and Summary tables (e.g. Table 1) can be used to resulting business interruption losses may pose organize key information for upper the greatest economic risk to a facility. management review. Preliminary priorities assigned by the project engineer are either Not all equipment requires strengthening. confirmed or modified (up or down) during Given the large amounts of equipment present senior management review by adjusting the in typical manufacturing environments, priority assigned to the final priority column. strengthening all or even most items may not Equipment is then strengthened according to be practical given limited resource priority as funding becomes available. availability. Key to determining the scope of the program is to perform a needs analysis EQUIPMENT STRENGTHENING and determine what is important to ongoing operations and will be needed immediately Equipment strengthening measures are following a major earthquake. Comprehensive conceptually more straightforward than interviews with product, process, and building seismic retrofit. The basic idea is to operations management personnel are a good secure the item against earthquake forces place to start. Facilities and maintenance through additional anchorage and/or bracing. staff should also be interviewed. Equipment strengthening approaches A thorough review of operational usually can be grouped into one of three requirements assures that all bases are general categories. These include: covered. Involvement of different organizational groups also insures key people 1. Generic Fixes are knowledgeable regarding the issue and generates internal support for the program. 2. Specific Fixes

Once overall program objectives have 3. Specific/Generic Fixes been established, equipment prioritization can begin. Typical steps include: Each approach has significant advantages and disadvantages when compared to each 1. Walkdown each piece of equipment other. The best approach for a facility will to determine seismic performance depend upon budget constraints, type and mix issues and assign a preliminary of prioritized equipment, whether internal priority. personnel or a contractor will be used to install retrofit hardware, and experience of 2. Gather critical data including the engineering staff. A discussion of the overall dimensions, estimated pros and cons of each approach follows. weights, existence/adequacy of existing anchorage, installation SPECIFIC EQUIPMENT STRENGTHENING interferences, etc. Photo document the item and key issues. Specific equipment strengthening is just what it sounds like - specifically engineered 3. Assign preliminary priority. anchorage for each individual piece of

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

ill Table 1 Equipment Prioritlzatlon

Number of Preliminary Final Priority Location/Item Photo ID # Similar Items Priority I ) Comments Welded Bellows Area

Portable Welding 18/10:47,48 12 (1) Machines Heat Treat Area

Furnace 18/10:51,52,53 1 1 (1) B Horizontal Tank 18/10:52,56 1 2 (3) C Priority: 1. Life safety and/or business interruption concern; or high value equipment subject to damage. Proceed with anchorage/restraint upgrade. 2. Medium value equipment subject to damage or sliding. Anchorage of selected items may be desired to increase safety or reduce damage. 3. No further action required. Comments: A. Table-top mounted equipment; no anchorage/restraint provided. B. Some anchorage/-estraint provided; need to confirm adequacy. C. Marginal or no anchorage/restraint evident, although provisions for anchorage appear to be available. equipment requiring seismic retrofit. Typical drawings suitable for obtaining steps include: contractor bids. 1. Review each item to specifically This approach is the technique normally determine problem areas and used to anchor equipment. It results in a well required fixes. engineered and highly reliable seismic retrofit. However, it may not be cost effective when 2. Field measurement of key dealing with large volumes of equipment for dimensions and estimate weight. the following reasons. 3. Perform necessary calculations and First, it requires engineering of the finalize design in accordance with anchorage for every item without any design criteria. consideration of component similarity. This may result in additional costs due to 4. Make final field check to assure duplication of engineering and drafting constructibility. efforts. 5. Generate engineering sketches or

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

112 Second, because component similarity is 7. Organize information in manual not explicitly accounted Tor the design format for use by personnel process, each fix will be unique. This may installing retrofits. result in higher construction costs since special fabrication may be required for each While conceptually easy to visualize, this anchorage detail. Application of the specific approach can be quite difficult to implement anchorage approach normally should be in practice. The most difficult part is limited to unique, highly hazardous, or costly organizing data for presentation in manual items. format.

GENERIC EQUIPMENT STRENGTHENING On one recent project, equipment was categorized according to major types or classes The second approach is generic equipment found throughout the facility. Major strengthening. It is based upon the following equipment classes included communications, hypotheses: 1) identical anchorage details can electrical, emergency, mechanical, HVAC, be used to retrofit many different types of office, piping and conduit, process, shop, equipment so long as the items are similar in storage rack and cabinets, and tanks and size, weight, and have similar installation vessels. Within each category, equipment clearance requirements, and 2) equipment subclasses were defined to provide additional retrofits can be logically presented in an easy breakdown and refinement. Tables were to read manual format. A good description of developed for each subcatcgory showing this approach is prc-cngincercd anchorage for required anchorage configurations, bolt sizes, major equipment categories contained in a and generic anchorage details. tabular look-up format. To demonstrate this approach, suppose a Basic steps involved with generic small, floor mounted chiller requires seismic equipment strengthening include the anchorage. It weighs 7,000 lbs., with following: dimensions of 8' long by 4' wide by 5' high. The steps required to determine anchorage 1. Review plant equipment and requirements are: develop categories and subcategories which generally encompass the 1. Go to the appropriate equipment majority of plant equipment. class (Table 2 - Mechanical Equipment). 2. Walkdown a representative sample to determine typical issues and 2. Within this equipment class, problems, and develop required available subcategories include skid retrofit concepts. mounted equipment without vibration isolators and skid 3. Take necessary measurements for mounted equipment with vibration each required category or isolators. For this item, the subcategory. appropriate subcategory is skid mounted equipment without 4. Perform bounding calculations to vibration isolators (Table 3). size generic details and anchorage requirements. 3. Determine the Height/Width (1.25) and Height/Length (0.62) ratios. 5. Make spot checks to help assure constructibility. 4. Using Table 3 in conjunction with a review of potential installation 6. Generate sketches showing generic interferences, the only allowable anchorage details. anchorage category is III (seismic stops on all four sides). Using the

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

113 appropriate weight, height/length, Table 2 height/width, and anchorage Mechanical Equipment Groups configuration requirements, the only allowable anchorage Applicable Table configuration is C. (Two seismic stops along each side of the chiller). Skid Mounted Equipment Without Vibration Isolators 5. Reviewing possible base anchorage details (7, 8 or 9), detail 7 (Figure o Chillers 1) appears to be the best choice for o Fans this application. Therefore, use o Motors detail 7 with 5/8 inch expansion o Packaged A/C Units anchors. Skid Mounted Equipment N/A If correctly implemented, this approach on Vibration Isolators can yield significant savings in engineering time. In addition, installation costs can be o Boilers reduced due to similarity of anchorage details o Chillers resulting in lower fabrication costs. o Compressors o Fans

Table 3 Skid Mounted Equipment Without Vibration Isolators Applicable Details: Anchorage Configuration

D G j' Category I (Existing Bolt Holes): 11 Category II (Seismic Stops at Corners): 5, 6 Category III (Seismic Stops Along Four Sides): 7-9 E K Category IV (Anchors Along Two Sides Only): 12-16 Category V (Anchors Along Four Sides): 12-16 F 1 Expansion(l) Weight Height/ Height/ Anchorage Base mchor Anchor (lbs.) Width ingth Category Configuration ietails Diameter

<5000 <4.0 <2.0 I B or D 11 1/2 <2.0 <2.0 II D 5,6 1/2 <2.0 <2.0 III C 7-9 1/2 <4.0 <2.0 IV B 12-16 1/2 <4.0 <2.0 V C 12-16 1/2

5001-10000 <2.0 <2.0 I A, B, or D 11 1/2 <2.0 <2.0 II D 5,6 5/8 <2.0 <2.0 HI C 7-9 5/8 <2.0 <2.0 IV B 12-16 1/2 <4.0 <2.0 I B or C 11 3/4 <4.0 <2.0 V C 12-16 5/8 Notes: (1) Diameters are in inches. Bolt embedment is given in Appendix A under General Notes. (2) Internal components on vibration isolators shall be anchored using Detail 69.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

114 However, there are a number of potential f4"CAP disadvantages. First, the real world does not lend itself to generic approaches. Often, STIFFENER R. generic details can easily be installed at three SEE SCHEDULE FOR WE/GMT places while the fourth location has an >5000 LBS ONLY interference. Developing procedures to accommodate special anchorage requirements due to interferences is difficult.

Second, personnel installing the retrofits may misinterpret the manual and use improper anchorage in some cases. * . • * Finally, this approach will normally ANGLE SEE handle 70% to 80% of equipment requiring NOTES: SCHEDULE anchorage. Remaining items will require development of specific anchorage details as /. Hs* HEIGHT OF EQUIPMENT ABOVE CONCRETE discussed previously. 2. Hs* 2" UiXMfUM FOR %", %*. AND #> BOLTS Figure 1. Detail 7 - Seismic Stop

SEE SCHEDULE

EXPANSION EXPANSION ANCHOR ANCHOR

v . i u .• i " i • I DFTML 7B DFTA/l 7A EQUIPMENT WEIGHT tOOOO LBS, MEMBER SIZE SCHEDULE SnFTFNER Hs (IN} ANGLE SIZE TH/CKENESS ftN.J

O L2'4x2'4x'/4 y< <2 L6x4x}4 '4 J <4 L8x6x% 4 Figure 1. (Cont.) Detail 7 - Seismic Stop

COMBINATION OF SPECIFIC AND 1. Walkdown each item to specifically GENERIC ANCHORAGE determine the required retrofit.

This approach combines the best features 2. Develop family of generic details of specific and generic anchorage approaches. based on component similarity. It can be described as a specific call out of prc-enginccrcd anchorage details. Basic steps 3. Determine specific anchorage include: requirements at each location using generic details.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

115 4. If required, design a specific fix As an example, assume there are some for a particular anchorage location chilled water pumps located in a mechanical or piece of equipment. equipment penthouse which require seismic restraint. The basic steps are as follows: 5. Perform constructibility reviews. 1. Using Table 4, locate the 6. Develop construction documents required equipment item (drawings/sketches and tabulated (penthouse, chilled water pumps, data). 4 total, located in Chiller Room). This approach is similar in concept to generic equipment strengthening. However, 2. Use detail 1.3C (Figure 2) to there arc two major differences. First, all anchor the base of the unit. equipment has been reviewed rather than a Note that top anchorage is not representative sample. This insures retrofit required for this item. approaches will be installable and require minimum field modification. Second, generic 3. Install details per anchorage anchorage look-up tables such as Table 3 have configuration F (Figure 3). been eliminated to avoid problems with installation personnel misinterpreting how to 4. Review the comments column to use them. Instead, a master call-out table is note any special anchorage provided which fully defines the required requirements or installation anchorage for a particular item. concerns.

Table 4 Equipment Anchorage Specification Table

Location/ Equipment Base Anchor Top Anchor Anchor Description Quantity Location Detail Detail Config. Comments

Penthouse

Heat Exchanger 1 Chiller Room 5/8" Exp. U) 3 anchors total. Anchor

Zinsco Switchgear 3 Chiller Room 3/8" Exp. (2) Install two exp. Panel Anchor anchors per panel through front base channel.

Wayne Receiver 1 Chiller Room 3/8" Exp. H Tank Anchor

Chilled Water Chiller Room 1.3C Place details on Pumps the inside of support channels with 1/4" gap.

Notes: (1) Install anchors in existing holes in equipment base supports. (2) Anchor configuration applies to each panel section.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

116 SEE SCHEDULE EXIST. EQUIPMENT BASE EXPANSION ANCHOR

ANGLE SEE E

EXPANSION ANCHOR FRONT MEW S££ SCHEDULE

DETA/l SCHEDULE SIDE D£TA/L NO. ANGLE S/ZE ANCHOR DM. {IN.J^ t.JA JxJx% '4 t.JB JxJx% % t.JC 6x6x% % J t.JD 6x6x% 4

Figure 2. Detail 1.3 - Seismic Stop effective manner. Though the cost of equipment strengthening is highly variable, costs often range from $1 to $5 per square foot for strengthening of high priority equipment in medium to large facilities.

Costs are on the order of $1 to $2 per square foot for facilities which require only nominal anchorage and bracing of building support equipment such as transformers, fans, and HVAC systems. For light assembly and Figure 3. Anchorage Configuration "F" manufacturing operations, equipment strengthening costs can range from about $1 This approach combines the best aspects to $3 per square foot. Raised computer floors of specific and generic anchorage. It takes can be retrofit for about $3 to $5 per square full advantage of component similarity in foot. Strengthening high technology generating anchorage details and specifically equipment, such as those used in some checks for interference problems prior to electronics and aerospace environments, installation of retrofit hardware. Anchorage typically costs about $3 to $6 per square foot. fabrication costs are minimized since the Clean room environments can cost $5 per identical details are used for many similar square foot or more due to the high cost of items. However, more engineering time is special preparation, installation procedures, required to develop equipment strengthening and clean-up. manuals than in the generic approach. CONCLUSION TYPICAL COSTS Risk of an extended business interruption Equipment strengthening based on a due to earthquake damage to critical, non- combination of generic and specific details safety related equipment can be as great if can be carried out in an efficient and cost- not greater than risks to buildings.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

117 Equipment is often unanchorcd and can be vulnerable to tipping or sliding damage. Fortunately, it is often practical and cost- effective to survey facility equipment, select certain critical items to be scismically strengthened based on safety and operational criteria, and implement an equipment strengthening program.

Three general approaches for implementing an equipment strengthening program have been discussed. Each approach has advantages and disadvantages. Based upon recent experience, a combination of generic and specific equipment strengthening is recommended as generally being the most cost effective technique for implementing a facility wide strengthening program. However, the best approach for your facility will depend upon staff experience, type and number of items being retrofit, budget and schedule.

Strengthening equipment is cost-effective when savings and benefits realized from increased safety, reduced potential damage and downtimes, and reduced risk of an extended business interruption are considered. Coupled with a building strengthening program, equipment retrofit can play a vital part in assuring an organization's ability to survive and quickly return to business following a major earthquake.

REFERENCES

[1] EQE Engineering, Inc., "Practical Equipment Seismic Upgrade and Strengthening Guidelines," prepared for Lawrence Livermore National Laboratory, Livermore, California, November 1985

1332nb\knoxpap.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

118 Sessions Poster Session

Second DOE Natural Phenomena Hazards Mitigations Conference - 1989

119 COMPARISON OF PROPOSED SEISMIC DESIGN CRITERIA FOR THE INEL

H. J. Dahlke, Staff Engineer Westinghouse Idaho Nuclear Company R. J. Secondo, Director of Engineering & Construction Management Division U.S. Department of Energy, Idaho Operations Office

ABSTRACT

The U.S. Department of Energy (DOE) has selected the Idaho National Engineering Laboratory (INEL) as the preferred site for the Special Isotope Separation (SIS) Project. The Plutonium Processing Building (PPB) of the SIS facility if the only safety class (Category I) structure required to be designed to withstand the effects of a Design Basis Earthquake (DBE).

Several seismic design criteria may govern the design and analysis of this structure. The latest (October 1986) revision of the DOE-ID Architectural Engineering Standards is based on DOE Order 6430.1, "General Design Criteria" and specifies a single free-field horizontal peak bedrock acceleration for various locations within the INEL site and response spectra from the U.S. Nuclear Regulatory Commission (NRC) Regulatory Guide 1.60, "Design Response Spectra for Seismic Design of Nuclear Power Plants." The revised DOE Order 6430.1 A (Interim Final 1/9/89) specifics that site-specific seismic parameters shall be determined for a DBE for safety class (Category I) structures of nonreactor nuclear facilities. Further it specifies that site-specific earthquake hazard models and response spectra given in UCRL-53582, Rev. 1, shall be used to select the appropriate seismic ground acceleration and that the design guidance given in UCRL-15910 shall be used in applying UCRL-53582, Rev. 1. It further allows site-specific studies to be substituted for the UCRL-53582, Rev. 1, data if a higher level of detail is required. Separately developed criteria for various INEL locations based on a strong ground-motion evaluation from a seismograph array will not only include local attenuation effects but also the effects of the 1983 M7.3 Borah Peak earthquake which was not included in the UCRL-53582, Rev. l.data.

A direct comparison or evaluation of these criteria is difficult since bedrock acceleration and 84th perccntile response spectra are specified by one source and surface acceleration and mean response spectra by another. Furthermore, some criteria are derived using deterministic techniques while others are based on probabilistic techniques. The paper will present the results of a parameter sensitivity study using soil column investigations with the computer program SHAKE to develop the mans of comparing these different criteria. The parameter sensitivity study will evaluate effects of the variation of soil properties (shear wave velocity and modulus, density, damping coefficient), soil depth, acceleration time histories, and the point of application of the control motion. The results will then be used to compare

• bedrock accelerations and 84th percentile Regulatory Guide 1.60 site independent spectra • surface accelerations and mean (50th pcrccntilc) UCRL-53582, Rev. 1, site dependent spectra • surface accelerations and mean spectra developed from INEL measurements

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

120 CONSISTENT APPLICATION OF CODES AND STANDARDS

M. A. Scott Wcstinghousc Hanford Company P.O. Box 1970 Richland, Washington 99352

ABSTRACT

The guidelines presented in the U.S. Department of Energy, "General Design Criteria" (DOE 6430.1 A), and the "Design and Evaluation Guidelines for Department of Energy Facilities Subject to Natural Phenomena Hazards" (UCRL-15910) provide a consistent and well defined approach to determine the natural phenomena hazards loads for U.S. Department of Energy site facilities. The guidelines for the application of loads combinations and allowables criteria are not as well defined and are more flexible in interpretation. This flexibility in the interpretation of load combinations can lead to conflict between the designer and overseer. The establishment of an efficient set of acceptable design criteria, based on U.S. Department of Energy guidelines, provides a consistent baseline for analysis, design, and review. Additionally, the proposed method should not limit the design and analytical innovation necessary to analyze or qualify the unique structure. This paper investigates the consistent application of load combinations, analytical methods, and load allowables and suggests a reference path consistent with the U.S. Department of Energy guidelines.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

121 DESIGN OF BURIED CONCRETE ENCASEMENTS

Richard M. Drake Fluor Daniel, Inc. 3333 Michelson Drive Irvine, California 92730

ABSTRACT

The operation of many Department of Energy (DOE) sites requires the transfer of radioactive liquid products from one location to another. DOE Order 6430.1A requires that the transfer pipelines be designed and constructed so that any leakage can be detected and contained before it reaches the environment. One design option often considered to meet this requirement is to place the pipeline in a stainless steel- lined, buried concrete encasement. This provides the ergi- neer with the design challenge to integrate standard structural design principles with unique DOE requirements. The complete design of a buried concrete encasement must consider seismic effects, leak detection, leak confinement, radiation shielding, thermal effects, pipe supports, and constructability. This paper contains a brief discussion of each of these design considerations, based on experience gained during the design of concrete encasements for the Process Facilities Modifications (PFM) project at Hanford.

INTRODUCTION • The roof of the encasement pro- The operation of many Department of vides some shielding benefit; Energy (DOE) sites requires the transfer of radioactive products from one location • The actual transfer pipelines to another. The design of these transfer are more accessible during cons- pipelines must reflect the unique re- truction, when pressure testing, quirements of each application at each or when leaks must be located; site. DOE Order 6430.1A[1] provides and general performance requirements, but detailed design criteria must be devel- • Construction of the outer (con- oped on a project-by-project basis. crete) containment can be nearly completed prior to pressure When only one transfer pipeline is testing of the inner (pipeline) involved, a double-walled pipe is usually containment. the most economical option. When multiple transfer pipelines are involved, a The design of a buried concrete stainless steel-lined concrete encasement encasement will require the integration is usually the most economical option. of DOE requirements, project require- The use of a concrete encasement has ments, and code requirements. This paper several advantages: is based on the design of buried concrete

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

122 encasements for the Process Facility Section 1305-6.3.2 indicates that the Modifications (PFM) project at Hanford design of transfer lines for liquid and recently adopted DOE Order 6430.1A decontamination wastes in plutonium requirements. These requirements are storage facilities shall consider the use summarized and design suggestions are of encasements as specified in Section offered by the author, 1300-7.4.

REQUIREMENTS SUMMARY Section 1323-5.2 indicates that The requirements driving the use of similar encasements be used to establish buried concrete encasements can be found "primary and secondary confinement bound- in DOE Order 6430.1A. The encasement aries in underground portions of high- designs must consider Design Basis Acci- level liquid waste systems." dent (DBA) conditions and limit the release of radioactive and hazardous For other special facilities, the materials, general requirements noted above may drive the decision to use a buried con- DOE Order 6430.1A crete encasement.

Section 1300-1.3 indicates that the DESIGN CONSIDERATIONS design of nuclear facilities shall pro- The complete design of a transfer tect the public from hazards associated pipeline involves many considerations. with radioactive materials as a result of DOE requirements for leak detection, normal operations including the effects environmental protection, and radiation of natural phenomena pertinent to the safety must be satisfied. DBA and site. operating load conditions must be resisted with acceptable stresses and deforma- Section 1300-7.1 indicates that tions. Practical considerations of confinement systems shall accomplish the constructability and cost must also be following: included. When multiple transfer pipelines are routed together, a common con- • Minimize the spread of radioac- crete encasement best meets the criteria. tive and other hazardous materials within the unoccupied pro- Concrete Encasement Routing cess areas and to occupied Early in the design, process consid- areas. erations identify the need for pipelines to transfer radioactive materials from • Minimize the release of radioac- one location to another. Pipeline dia- tive and other hazardous materi- meters, temperatures, and pressures are als in facility effluents during identified and the elevations at the normal operation and anticipated pipeline termination points are also operational occurrences. usually fixed at this time. Then a routing must be selected. • Limit the release of radioactive and other hazardous materials Some of the design considerations resulting from DBAs including that must be addressed in routing trans- severe natural phenomena. fer pipelines include:

Section 1300-7.4 indicates that • Existing and future underground double-walled pipes or pipes within a construction must be secondary confinement structure encase- considered. ment shall be used in all areas where the primary pipe leaves the facility.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

123 • Long straight runs should be avoided (or pipe loops added) to allow for pipeline thermal movements.

• Construction access should be available along the entire route.

• Transfer pipelines from/to dif- fering locations should be combined to the maximum extent.

• Burial depth has impact on radiation shielding requirements.

Concrete Encasement Form Concrete encasements are usually buried for radiation safety reasons. The side walls must be thick enough to withstand lateral soil pressures. The top and bottom slabs must be thick enough to withstand vertical soil pressures. Many Figure 1. Typical PFM Project of the load conditions are resisted by Concrete Encasement the rigid frame interaction of the walls of the encasement and its bottom slab. The top slab of the encasement could The walls are utilized to provide support be cast in-place utilizing form deck. If and anchorage for the transfer pipelines. access is desired to the encasement for operations or maintenance purposes, the Figure 1 indicates a typical top slab could include precast access concrete encasement designed for the PFM hatches located over valve or instrumen- Project at Hanford. tation locations. If a long time period is expected between encasement construc- Constructability tion and transfer pipeline installation, The construction of the concrete the entire length of the encasement could encasement can not be separated from the be built with precast roof panels. installation of the transfer pipelines within it. The two walls and the bottom Leak Detection slab of the encasement must be cons- Ideally the transfer pipelines and tructed first. Next, the pipe supports concrete encasement can be designed to and transfer pipelines must be installed. slope from the point of origin to the Only after the pipelines are inspected receiving location. If the slope is and pressure tested can the roof slab of steep enough, and the distance short the encasement be poured or placed. enough, leaks from the pipeline will flow within the containment to the receiving The two walls and the bottom slab of location. the encasement are candidates for jointly precasting into U-shaped sections. The In practice, the distance and eleva- shipping pieces should be as large as tion changes between transfer points are possible within '.he constraints of high- not likely to allow such a simple method way weight limitations, trailer lengths, of leak detection. In addition, the con- and construction crane capacity. Precast struction tolerances on the floor slab, sections can be joined in the field with with or without liner plate, will often short make-up sections. result in short uphill sections or Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

124 pockets in a nominally sloping alignment. Crack control criteria specified in While large volume leaks in transfer ACI 224R[4) for water retaining struc- pipeline will eventually reach the tures (.004 in.) nay be appropriate for receiving location» a means must be concrete encasements that may be required developed to detect and locate smaller to contain large volumes of liquids. The leaks along the length of the encasement. design result of a more stringent crack control criteria will be more steel One option for this along-the-length reinforcement along the length of the leak detection would be the use of tele- concrete encasements. operated devices to patrol the floor slab of the encasement. The encasement design DESIGN DETAILS must allow sufficient clearance to allow clear passage for such a device and its Stainless Steel Liner Plate trailing cables. A similar option would Stainless steel liner plate is often used be to provide TV port manholes at regular to protect the concrete. It protects the spacing along the encasement to allow secondary confinement to facilitate remote viewing via a TV camera lowered cleanup if leaks are detected in the into the encasement at the ports. This transfer pipelines, the primary confine- TV would view from fixed locations, and ment. It is desirable that the liner would be both less costly and less effi- plate be leak resistant. On some pro- cient than the traveling TV device. A jects liner plate leak resistance may be third option is the use of leak detection required. If the liner plate serves no ribbons placed on the floor of the confinement function, it is not necessary encasement. to vacuum test weld surfaces for micro- scope leaks. Visible cracks and holes in The supporting instrumentation could the liner plate should always be regarded be designed to notify a panel operator as unacceptable because they would be when and where a leak has been detected. large enough to let leaking fluids attack the concrete. Leak Confinement Once a leak has developed in the The design of the liner plate will primary confinement (a transfer pipe- probably be governed by thermal strain, line) , the leak must be contained by the although seismic strain should also be secondary confinement (the concrete checked. Stainless steel has an expan- encasement) until the leak is detected, sion coefficient 50% larger than the con- located, and repaired. crete it is anchored to. Therefore, as the encasement inner surfaces heat up In most situations, the concrete during operations, the concrete will side walls and floor slab will be suffi- restrain the liner plate's thermal expan- cient to provide the required contain- sion. If the resulting compressive ment. While normal concrete is regarded stresses reach the liner plate's buckling as relatively impervious to liquids, limit, the plate will elastically buckle. concrete subject to high fluid pressure This buckling is not a concern as it will will allow passage of liquids through it. relieve any thermal strains, and deforma- If the transfer pipelines carry large tions out-of-plane will be small. The volumes of fluids, and a delay might be maximum load transfer to the anchorage anticipated between leak initiation and angles embedded in the concrete will be shutdown of the leaking pipeline, con- the plate's buckling load. crete design should involve crack control criteria more stringent than normally The liner plate installation should required by ACI 318[2] or ACI 349[3] be by the "wallpaper" method, welding the (.013 in.). liner plate to angles or plates embedded in the concrete. If the walls and floor slab are precast, the attached liner

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

125 plate could also be installed off-site. of lead or steel plate is a very expen- The inner top slab surface should also sive option and should only be considered have liner plate if any of the transfer after other alternatives are found to be pipeline operate under pressure. If the unacceptable. top slab is to be cast- in-place, the liner plate can be stiffened with angles Transfer Pipeline Supports and serve as a permanent form. If the Proper support and anchorage of the top slab is part of an access cover pipelines must be provided. Supports system, the liner plate should be inte- must be designed with two primary con- grally cast with the concrete cover. sideration; they must transfer all pipeline loads to the anchorage embedded in Radiation Shielding the concrete; and they must provide a The transfer pipelines contain fastening capability for the pipelines. radioactive liquid products. These pro- Sliding supports must allow attachment of ducts emit gamma radiation which can harm U-bolts or similar fastenings. Anchor personnel. Shielding materials must be supports must be of sufficient thickness placed between the radiation source and to allow welding, and to transfer dead, any area which could be occupied by per- seismic, thermal, and test loads to the sonnel. When quantifying the radiation concrete. source(s), it is important to consider the future use of all transfer pipelines Figure 2 indicates typical transfer in the encasement, including any spares. pipelines supports designed for the PFM The required mass of shielding material project at Hanford. is calculated considering the maximum cumulative source and the maximum allow- NATURAL HAZARDS PHENOMENA MITIGATION able radiation level at the ground sur- As previously indicated, the con- face. Computer programs for shielding crete encasement serves as the secondary design are available from the Radiation confinement. To meet DOE 6430.1A Shielding Information Center (RSIC) requirements, the concrete encasement is located at the Oak Ridge National usually required to resist Design Basis Laboratory. Accidents (DBA). DBA events are defined in DOE 6430.1A, Section 0111-99.0. Of The required shielding is compared most interest to the design of buried with the shielding provided by the con- concrete encasements are earthquake, crete roof slab of the encasement and any flood and tornado natural phenomena earth material cover. If the shielding hazards. provided is insufficient, additional material is required. Earth materials Design Basis Earthquake are usually the most economical material Design basis earthquake loadings are to provide additional shielding. If pro- defined for DOE sites in UCRL- 53582[5]. cess considerations preclude lowering the Analysis methods to account for seismic encasement elevation, the additional effects on long underground elements are earth cover will take the form of a pro- well documented in literature[6][7]. tective berm. These r-tatic methods assume that the underground element is more flexible than This type of berm must be protected the soil and that it will assume the from erosion to preclude deterioration in shape of the soil as seismic waves travel shielding provided. If the earth berm is through it. Axial strains and curvatures high enough, it may be unacceptable at are developed for the soil and the con- road crossings. In these cases, lead or forming element. These methods are steel plate may be laid across the reasonable for flexible elements, and are encasement roof slab to provide equiv- conservative for stiff elements. The alent radiation shielding in less thick- effects of bends, tees, and buildirg ness than the earth materials. The use

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

226 connections are accounted for utilizing beam on elastic foundation theories.

These simple static analysis methods should be utilized for the analysis of seismic effects on buried concrete encasements. It is recognized "•Vs. that these methods might be conservative for the stiff encasements, but the required axial reinforcement utilizing &** these methods is not considered excessive. While it is possible to spend •I \ more effort performing rigorous time it history analyses on buried encasements restrained by soil springs, it is not we* t"» likely to pay for itself in reduced reinforcement, especially if seismic effects do not govern.

Ground failure potential due to TYPE I-A earthquake should also be evaluated. If earthquake activity could cause the soil along the encasement route to liquify, settle, or laterally move, design measures must be taken or a new route selected.

Design Basis Flood For the DOE sites and for new facilities at existing sites, the Design Basis Flood (DBFL) should ideally be THIS or*;,— / \ / addressed by siting a new facility above the DBFL level for the appropriate \ annual probability of exceedance. , , i ! • ' Evaluation procedures to address DBFL concerns are indicated in : *! l o * \ UCRL-15910[8]. Containment structures 1 t 1 should be watertight. This is more a design detailing problem than an I "' ^ „ ft 1 * * engineering problem. All entries to the i j concrete encasement should be designed to resist water pressures of moving and •Ultf t ponding flood waters. Allowing for the possibility of in-leakage to the encasement, provisions must be made to remove TYPE II-A water by gravity drainage or pumping to a retention location where the water can be evaluated for radiation or hazardous Figure 2. Typical PFM Project contamination. Transfer Pipeline Supports Design Basis Tornado Design Basis Tornado (DBT) loadings include wind pressures, atmospheric pressure changes, and missiles. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

127 Site specific LBT loadings are defined techniques are available to evaLuutf the for DOE sites in UCRL-53526[9]. Only Che effects of natural phenomena hazards. DBT-generated missiles impacting the roof Other design concerns must also be con- slab are of interest to the design of sidered Including constructability and buried concrete encasements. leak detection.

If the encasement roof slab is at or REFERENCES near the ground surface, tornado missiles can cause several local effects to the of [1] DOE Order 6430.1A, General Design Criteria, 1987. the encasement: [2] American Concrete Institute, ACI Penetration - The displacement of a 318-83, "Building Code Requirements missile into the target. for Reinforced Concrete." [3j American Concrete Institute, ACI Perforation - The displacement of a 349-85, "Code Requirements for Nuclear Safety Related Concrete missile through the target, with or Structures." without missile exit velocity. [4J American Concrete Institute, ACI Spalling - The ejection of target 224K-80, "Control of Cracking in Concrete Structures." material from the front face. [5] UCRL - 53582: "Natural Phenomena Scabbing - The peeling-off of target Hazards Modeling Project: Seismic Hazard Models for Department oi material from the back face. Energy Sits," Rev. 1, 1984.

Perforation and scabbing are [6] American Society of Civil Engineers (ASCF-), 1984, Guidelines for the unacceptable local effects for concrete Design of Oil and Gas Pipeline encasements serving as containments. Systems. Local effects can be evaluated using empirical methods summarized in ASCE [7) Shah, H., and S.L. Chu, "Seismic Analysis of Underground Structural Manual No. 58[10]. Elements, Journal of the Power Division, ASCE, July 1974, pp. Overall structural response of the 53-62. encasement roof slab to tornado generated [8] UCRL - 15910: "Design and Evalua- missiles must also be accounted for. tion Guidelines for Department of Structural response can be evaluated Energy Facilities Subjected to using dynamic methods prescribed by Natural Phenomenon Hazards," 198B. Biggstll]. [9] UCRL - 53526: "Natural Phenomena Hazards Modeling Project: Extreme The roof slab minimum thickness will Wind/Tornado Hazard Models for Department of Energy Sites," Rev. 1, be determined to meet radiation shielding 1985. requirements. This thickness should be [10] American Society Civil Engineers evaluated for local effects and overall (ASCE;, 1980, Structural Analysis structural response, conservatively and Design of Nuclear Plant neglecting the resistance of any thin Facilities, Manuals and Reports on Engineering Practice - No. 58. soil layers covering the encasement roof slab. [11] Biggs, J.M., Introduction to Structural Dynamics, McGraw-Hill, SUMMARY 1964. Buried concrete encasements must be designed in accordance with the requirements outlined in DOE 6430.1A. Of primary engineering concern are the provisions for radiation safety. Analysis

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

128 FAILURE PROBABILITY ESTIMATE OF TYPE 304 STAINLESS STEEL PIPING

W L. Daugherty, N. G. Awadalla. R. L Sindelar H. S. Mehta, S. Ranganath Westinghouse Savannah River Company General Electric Company Savannah River Laboratory San Jose, CA Aiken, SC 29808

ABSTRACT The primary source of in-service degradation of the SRS production reactor process water piping is intergranular stress corrosion cracking (IGSCC). IGSCC has occurred in a limited number of weld heat affected zones, areas known to be susceptible to IGSCC. A model has been developed to combine crack growth rates, crack size distributions, in-service examination reliability estimates and other considerations to estimate the pipe large-break frequency. This frequency estimates the probability that an IGSCC crack will initiate, escape detection by ultrasonic (UT) examination, and grow to instability prior to extending through-wall and being detected by the sensitive leak detection system. These events are combined as the product of four factors: 1. The probability that a given weld heat affected zone contains IGSCC. 2. The conditional probability, given the presence of IGSCC, that the cracking will escape detection during UT examination. 3. The conditional probability, given a crack escapes detection by UT, that it will not grow through-wall and be detected by leakage. 4. The conditional probability, given a crack is not detected by leakage, that it grows to instability prior to the next UT exam. These four factors estimate the occurrence of several conditions that must coexist in order for a crack to lead to a large break of the process water piping. When evaluated for the SRS production reactors, they produce an extremely low break frequency. The objective of this paper is to present the assumptions, methodology, results and conclusions of a probabilistic evaluation for the direct failure of the primary coolant piping resulting from normal operation and seismic loads. This evaluation was performed to support the ongoing PRA effort and to complement deterministic analyses addressing the credibility of a double-ended guillotine break.

INTRODUCTION Vw«v, u. ing for the primary coolant system. The material of con- The Savannah River Site production reactors op- struction for the primary pressure boundary is Type 304 erate at low temperature and pressure, permitting the use stainless steel. Due to low applied stresses and the in- of relatively thin-walled piping for the primary coolant herent toughness and ductility of the piping material, system. The material of construction for the primary pres- the probability of a DEGB is extremely low. The objec- sure boundary is Type 304 stainless steel. ITiese reactors live of this paper is to present the results and conclusions were built in the 1950's, and have undergone various mod- of a probabilistic evaluation for the direct failure of the ifications and upgrades since that time. The maximum primary coolant piping resulting from normal operation rate loss-of-coolant accident (LOCA) for the Savannah and seismic loads. The failure by indirect (seismic) means River production reactors is the hypothetical double- is addressed in a separate paper. This evaluation supports ended guillotine break (DEGB) of a large process water the ongoing PRA effort and to complements deterministic pipe f 1). These reactors operate at low temperature and analyses addressing the credibility of a double-ended guil- prcssurc oermitting the use of relatively thin-walled pip- lotine break. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

129 DISCUSSION WELD CRACKING PROBABILITY, IV The SRvS production reactor process water piping Experience in the commercial nuclear industry has undergone limited degradation from intergranular shows that 6 to 8% of sensitized stainless weldments expe- stress corrosion cracking (IGSCC). IGSCC has occurred rience IGSCC. In the large process water piping, 48 weld- in a limited number of weld heat affected zones, areas ments have been identified as containing IGSCC since UT susceptible to IGSCC. This evaluation combines crack inspection began in 1984 [2]. This same piping in the growth rates, crack size distributions, in-service examina- three operating reactors contains a total of 781 circumfer- tion reliability estimates, and other considerations to esti- ential welds which were inspected. This gives an inci- mate the pipe large break frequency. This frequency, dence rate of 6%. Hence the probability that a wcldmcnt PBreak' estimates the probability that an IGSCC crack will contains IGSCC is taken as 0.08 which envelopes both initiate, escape detection by ultrasonic (UT) examination, SRS and commercial reactor experience. and grow to instability prior to extending through-wall and being detected by the sensitive leak detection system. The CRACK NON-DETECTION PROBABILITY, PCND likelihood of these events leading to a large break is ex- The crack non-detection probability characterizes pressed as the product of four factors: the conditional probability, given the existence of IGSCC 1. Pc: The probability that a given weld heat in a weldment, that the crack is not detected by UT. The affected zone contains IGSCC. process water system piping has been subject to periodic 2- PCND- The conditional probability, given the ultrasonic (UT) examination since 1984. The UT inspec- presence of IGSCC, that the cracking will tors who have performed these examinations have been escape detection during UT examination. certified for IGSCC detection by the Electric Power Re- 3- PLND' The conditional probability, given a search Institute (EPRI). crack escapes detection by UT, that it will not The reliability of detecting IGSCC has been char- grow through-wall and be detected by leak acterized [3]. Figure I, reproduced from reference [3], age. identifies that a relatively short crack, 50% through-wall, 4. Pco: The conditional probability, given a has approximately 0.1 probability of non-detection. As a crack is not detected by leakage, that it grows crack grows in length or in depth, this probability de- to instability prior to the next UT exam. creases. This value is taken from the curve labeled These four elements describe the several conditions which "good" in Figure 1, based on the qualifications of the UT would need to coexist in order for a crack to lead to a operators used at SRS. Based on these data, the crack large break of the process water piping. Each is devel- non-detection probability is taken as 0.1 for weldments oped and discussed separately below. that receive UT examination. There also exist several

1.0 "Poor" 0.5 Ctl O

"Advanced" - Judgement That Improved Procedures^ 0.01 and Existing Technology Can Give Pro = 0 0001 for Through-Wall Flaw • • I • ... I L J 1 I l I i i I 0 01 0.02 0.05 0 10 0 2 0 5- 1.0 Crack Depth/Thickness, a/t

FIGURE 1. Detection of Iniergranular Stress Corrosion Cracking in 10-Inch Stainless Steel Pipe (from reference 3)

Second DOE Natural Phenomena Hazards Mitigation Conference - 19S9

130 welds thai are not accessible for external inspection. A approaches. Work is in progress lo develop a less conser- pipe crawler is being developed to inspect these weld- vative estimate based on an evaluation of ihe leak detec- roents. Until the pipe crawler is available, the crack non- tion system reliability. When complete, this factor will be detection probability for these weldments is taken as 1.0. revised accordingly.

LEAKAGE NON-DETECTION PROBABILITY, P[.NI) The conditional probability, given a crack escapes CRACK GROWTH PROBABILITY, PCO detection by U'K that it will not grow through-wall and be The fourth factor estimates the conditional proba- detected by leakage is assessed in this section. The SRS bility, given a crack that escapes detection by UT and leak reactors have experienced a number of cracks over the detection, that the crack grows to instability prior to the past 35 years. Before the periodic UT examination of the next UT exam. The likelihood of a crack also escaping piping was begun in 1984, most of these cracks were de- detection during the subsequent examination is modeled tected by their leakage as they grew through-wall. A total by a second application of the leak non-detection factor. of 16 such cracks have been detected in the main coolant The crack growth probability is based on three consider- loop (large piping and effluent nozzles) by the various leak ations: the crack size distribution, crack growth rate, and detection systems. These systems include stack tritium the local stresses in the pipe. monitors, closed circuit television surveillance, and visual Crack Size Distribution examinations. Thus, no large breaks have occurred while The crack size distribution is based on UT mea- 16 opportunities for a large break were averted as a result surements on SRS piping. The cumulative crack probabil- of the leak detection capabilities. A statistical treatment ity as a function of crack length is shown in Figure 2, gives the likelihood of a large break that is not prevented along with an exponential fit. This fit is expressed by the by the leak detection systems: equation [4]: 1/m l-(Probo) P(L) = (l/»exp(-L/2TrR jx (eq2) 4.2 x 1()-2 (eql) where L is the circumferential crack length, m is a param- Here, the probability of having zero large breaks is 0.5, eter fit to the data (a best estimate value of 0.05 is shown representing a statistical best estimate. Due to the small in Figure 2) and R is the mean pipe radius. To develop sample size, represented by the relatively small number of the probability that a crack exists with a length between cracks in the piping, this statistical treatment produces an two specific values, this equation is integrated between estimate much lower than would be expected from other those two values to obtain:

1-B

Analytical Fit: P(L) = (Uu) exp(-LV2irR|i) with n = 0.05

16 20 28 Crack Length (% of Circumlerence)

FIGURE 2. SRS Pipe Crack Probability UT-Detected with Exponential Fit

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

131 - [exp(-Li/2uRn) - to grow (0.4 inch/year) x (5 year) = 2.0 inches. Hence, the crack growth probability over a 5 year interval is calcu- exp(-L2/2iTR(i)J/[l-cxp(-l/(i)] (cq 3) lated from equation 3, using Lj = 26.2 inches and L2 = 28.2 Crack Growth Rate inches. The corresponding average value per year is ob- A reasonable crack growth rale is obtained from tained by dividing this result by 5. This procedure gives a laboratory test data |5|. Growth rate tests indicate steady crack growth probability of 2.4 x 10"fl per year. If a weld state crack tip extension rates of 10~6 inch/hour or less at were not inspected for a period of 10 years, the corre- prototypic conditions. The introduction of transients sponding average crack growth probability over that period (temperature, load), such as might be produced by would be 3.9 x 10~6 per year. startup/shutdown cycles and other reactor evolutions, pro- For welds that are inspected every 5 years, a crack duces eflective crack tip extension rates up to 10~4 inch/ growth probability for a longer period can still be calcu- hour; however, this maximum rate is not indicative of a lated. The crack growth probability for a second 5-year long-term average growth rate in the piping. Additionally, period would be combined with the crack non-detection variations in stresses and microstructure favorable for such probability a second time. The crack growth probability rates are generally localized. Once a crack grew beyond for a second 5-year period equals the crack growth proba- such local regions, the growth rate would decrease. bility for a 10-year period minus the crack growth proba- Therefore, a long-term average crack tip extension rate of bility for the first 5 years. This gives: 10"s inch/hour is used. Since a crack can grow from both ends, the crack Pcc(2nd 5 years) = (3.9 x 10"5 tip extension rate is doubled to obtain the crack growth - 1.2 x 10"5)/5 years = 5.4 x 10~6 per year rate. Further, to account for the possibility of multiple cracks in a single weld heat-affected zone, it is assumed Therefore, the average crack growth probability over a that two cracks exist that combine just before reaching 10-year period for inspected wcldments is: instability. Hence, the total crack growth rate within the 5 heal affected zone is 4 x 10" inch/hour, or approximately PCG( 10 years) = [PccOst 5 years) 0.4 inch/year. + Pco(2nd 5 years) x PCNDJ/10 years = 1.5 x IO~6 per year Local Stresses The local stresses in the pipe determine the Seismic Contribution length at which a crack reaches instability (Lj). For pur- The instability length developed above is based on poses of calculating the instability length, it is conserva- loads present during normal operation. A separate case to tively assumed that the crack is through-wall along its en- be considered is the addition of seismic loads. During an tire length. Since the operating history shows that no earthquake there is insufficient time to depend on crack pipes have ever broken, it is certain that no existing crack identification by leak detection means. Therefore, the has yet reached instability. Also, from the crack growth crack growth probability for the seismic case will be com- rate developed above and knowledge of the time before bined with a leak non-detection probability of unity. the next UT examination, a second crack length (L2) is When the seismic loads are added to normal opera- calculated such that a crack shorter than L2 will not grow tion loads, the instability length decreases slightly. The to instability prior to the next examination. Therefore, the corresponding crack growth probability must be multiplied crack growth probability is the probability that an existing by the earthquake probability. In practice, the instability crack has a length shorter than Li but longer than L2. length for each of a range of seismic loads is used in com- The instability length for a given pipe section va- bination with the probability of that particular magnitude ries depending on pipe dimensions and local stresses. All earthquake, and the results summed for a total seismic instability lengths are greater than or equal to 58% of the contribution to the crack growth probability. A para- pipe circumference. Hence, if one considered a 16-inch- metric study (not yet published) shows that the seismic diamcter pipe section, the instability length would be 28.2 contribution equals 0.7% of the non-seismic contribution, inches (using the mean radius of 7.75 inches). Future work may survey all local stresses to take credit for pipe RESULTS sections in which the instability length is greater than 58% Two categories of wcldmcnt arc considered. Most of the circumference. of the welds are accessible for UT' examination and are The in-service inspection plan for the SRS reac- examined every five years. A few welds have limited ac- tors process water system calls for UT examination of pipe cess: these include the weld attaching a flange lap to the wcldmcnts every five years. During the interval between pipe end and welds in piping that runs through concrete inspections, therefore, a crack would have the opportunity structures. Therefore, two cases are developed; the lim-

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

132 CONCLUSIONS ited access welds, which currently receive no UT examination, and the accessible welds, which are examined every 5 The direct failure frequency for the process wa- years. The four factors developed above arc combined to ter piping of the Savannah River production reactors produce the total piping direct break frequency. These has been conservatively estimated to be 1.6x10~6 per year. factors are summarized in Table 1, Several areas of further refinements have been identified. Additionally, work is underway to develop a pipe crawler These factors arc combined as discussed above. The re- to inspect the limited access weldmcnts from the inside. sulting non-seismic contribution is: Upon completion of crawler dcvclomcnt and inspection of 2 fi the limited access welds, further reductions in the failure PBreak(non-seis.) = (0.08X4.2xl0- X0.1Xl-5xl0- ) = 10 frequency can be realized. 5.0xl0~ per weld-year, accessible wcldmcnts. This work is part of a larger effort to characterize 2 6 PBrcak(non-scis.) = (0.08X4.2xl0- Xl.0X3.9xl0- ) = the integrity of the process water system and define the 8 1.3xlQ~ per weld-year, limited access weldmenls. maximum credible LOCA for the Savannah River produc- The corresponding seismic contribution is: tion reactors. This larger effort, combining this probabilistic work with deterministic analyses, has demonstrated PBreak(seismic) = (0.08X0.1X1. lxlO"8) = 8.8X10"11 per that the hypothetical double-ended guillotine break is not weld-year, accessible weldmcnts. a credible scenario. One long-term goal of this work is to Pnreak(seismic) = (0.08)(1.0X2.7xlQ-8)-2.2xlO-9 per define a maximum credible LOCA for use in accident weld-year, limited access weldmcnts. analyses and the establishment of power limits. Combining these respective contributions, the total pipe REFERENCES direct break frequency is: [1] DPSTA-100-1, "Safety Analysis of Savannah River PHreak(total) = 5.9xl(H° per weld-year, accessible weld- Production Reactor Operation", coordinated by J. P. ments. Church, revised September 1983. Pi)rcak(toial) = 1.5xl()~K per weld-year, limited access [2] WSRC-RP-89-126, "Leak History Reactor Primary weldments. Coolant Systems', G. R. Caskey et al., April 1989. Multiplying these two frequencies by the number of welds [3] NUREG/CR-4469, Semi-Annual Report, Volumes in each category yields the total direct break frequency for each reactor of 1.6xlO~6 per year, averaged over a period 1-4, "Nondestructive Examination (NDE) Reliability of 10 years. For further extrapolations into the future, for Inservice Inspection of Light Water Reactors", S. this estimate would increase. R. Doctor et al.

Table I. Pipe Break Frequency Input Factors

Factor Value Special Notes Weld cracking, Pc 0.08 Applies to all weldmcnts Crack non-detection, PCND 0.1 Accessible weldments 1.0 Limited access wcldmcnts 2 Leak non-detection, PLND 4.2xlO- Applies to non-seismic contribution only Crack growth, P^G 1.5xlO-6 Per year avg., 10-year period, accessible weldments 3.9xlO~6 Per year avg., 10-ycar period, limited access weldnicnts l.lxlO-8 Seismic contribution, accessible wcldmcnts 2.7xl0-R Seismic contribution. limited access wcldmcnis

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

133 (4) DPST-88-468, "Reactor Materials Program Process Water Piping Direct Failure Probability", W. L. Daug- herty, April 1988. (5J GE-88-006, "Stress Corrosion and Fracture Asses- sment Program, Monthly Program Letter #35, Report- ing Period Ending March 31, 1988", P. Aldred. Gener- al Electric Company, and GE-88-020, "Stress Corrosion and Fracture Assessment Program, Month- ly Program Letter #40, Reporting Period Ending Au- gust 31, 19888", P. Aldred, General Electric Company.

ACKNOWLEDGMENT The information contained in this article was developed during the course of work under Contract No. DE-AC09-76SR0000I (now Contract No. DE- AC09-88SR18035) with the U. S. Department of Energy.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

134 NONLINEAR SEISMIC ANALYSIS OF A THICK-WALLED CONCRETE CANYON STRUCTURE

Bob V. Winkel, Principal Engineer Gary R. Wagenblast, Principal Engineer Westinghouse Hanford Company P.O. Box 1970 Richland, Washington 99352

ABSTRACT Conventional linear seismic analyses of a thick-walled lightly reinforced concrete structure were found to grossly underestimate its seismic caoacity. Reasonable estimates of the seismic capacity were obtained by Derforming apDroximate nonlinear sDectrum analyses along with static collapse evaluations. A nonlinear time history analysis is planned as the final verification of seismic adequacy.

INTRODUCTION (20.7 m) wide. The overall length of There were a number of large the "canyonlike" box structure is over reinforced concrete canyon structures 800 ft (243.8 m) with expansion .joints constructed at U. S. DeDartment of provided every 40 ft (12.2 m). The Energy (DOE) facilities during the lower part of the structure contains 1940 to 1960 time frame. The boxlike process cells separated by 7 ft (2.1 canyon structures were generally built m) thick walls which act as shear for the purpose of housing radioactive walls in resisting transverse chemical processing. Typically, horizontal accelerations. Exterior shielding requirements dominated the wall and roof thicknesses vary from a exterior wall design, resulting in minimum of 3 ft (0.9 m) to as large thick sections with minimal as 8.5 ft (2.6 m). Sections are reinforcement. As these structures typically underreinforced relative to are qualified for new missions, current current American Concrete Institute safety requirements must be met (ACI) minimums for flexural steel, including demonstrating adequacy for which limits the usefulness of a the site Design Basis Earthquake conventional seismic analysis. (DBE). The purpose of this paper is There are a number of unique to describe the procedures used to structural features which are pertinent estimate the seismic caoacity of a to the canyon structure seismic specific canyon structure at the DOE caoacity. First of all, the thick- Hanford Site near Richland, Washington. walled construction results in significant dead weight restoring CANYON BUILDING STRUCTURAL FEATURES moments which add to the structural A scaled cross-section sketch of collapse capacity. The under- the structure in question is shown in reinforced nature of most cross Figure 1. The cross-section dimensions sections results in a crack initiation are 77 ft (23.5 m) high by 68 ft moment which is much greater than

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 135 3 ft (.9 m)

Crane Level

L^ 5 ft (1.5 m)

\ i f \ Process ,' i Cells • i i (Sjeoarated bjy Ground 7 !ft (2.1 m!) Level TJiick Walls]} 66 ft (20.1 m)

Figure 1. Canyon Structure Cross Section Sketch the cracked section's ultimate capacity. structure. Therefore, the following This unusual design feature simolifies procedure, based uDon collapse the structural caDacity analyses in capacity, was developed. the sense that it limits the nonlinearities to a relatively small (1) A soil-structural interaction number of cracked sections. analysis was performed to establish There are a number of "ore-cracked" the DBE design resoonse snectrum sections in the canyon structure in the for the canyon structure at the form of unbonded construction "joints. foundation level. This resulted These construction joints tyDically in a basemat soectra which is occur in areas of high seismic bending less than the specified surface moment (wall/roof intersections, spectra since the foundation is locations of exterior wall thickness well below the ground surface. changes, etc.). The coinciding of the high bending moments and construction (2) In situ nondestructive testing joint locations results in almost all (NDT) measurements were performed of the nonlinearities occurring at the on the canyon structure to evaluate construction joints as discussed in a the condition and current later section of this paper. properties of the aging concrete, locate and estimate the degree of SEISMIC CAPACITY EVALUATION PROCEDURE bonding in the construction joints, Preliminary analyses indicated and assess the extent of cracking. that a conventional elastic analysis was not anoroDriate for estimating the (3) An initial scooing of the seismic caDacity of the canyon structural seismic capacity was Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

136 performed by comparing section SOIL-STRUCTURE INTERACTION ANALYSIS capacities with the elastically The FLUSH computer code [4] was calculated seismic demands. Areas used to perform the soil-structure exceeding capacity were identified analysis of the canyon structure. as probable regions of nonlinear Hinges were introduced into the canyon response. Ultimate capacity moment- structure model to match the quasi- rotational relationships were linear fundamental frequency obtained developed for the sections where from the iterative response spectrum local nonlinear responses were analysis. A bounding foundation anticipated. spectrum was obtained by enveloping the results obtained from varying the (4) Linear rotational springs were parameter uncertainties (e.g. soil introduced into the structural properties). model at all locations which indicated a potential for being NDT EVALUATION loaded into the inelastic range. The NDT evaluation of the structure An approximate nonlinear seismic utilized pulse echo techniques for analysis was then performed by examining the concrete strength, iterating on the soring stiffness stiffness and quality. The pulse echo using a secant modulus aporoach results indicated relatively high to quantify the pseudo-elastic concrete strengths and "good to spring stiffness. If the iterative excellent" concrete quality. It also procedure did not develop into an demonstrated the existence of several unstable solution, it indicated essentially unbonded construction that the DBE loading was below the joints as indicated in Figure 2. seismic capacity. Although not Ground penetrating radar techniques done for the canyon structure, were employed to verify rebar this iterative procedure could be locations. continued to estimate the seismic capacity by oroDortionally LINEAR ANALYSIS increasing the DBE response spectra Finite element computer models of until an instability is achieved. the canyon structure cross section were developed using the ANSYS computer code (5) Additional insight into the seismic [2]. Both beam element and monolithic capacity was obtained by evaluating (solid element) models were developed. the static collapse capacity. The initial elastic resDonse spectrum Using limit analysis techniques, analysis of the canyon structure the horizontal g load collapse indicated several areas where capacity capacity was predicted. Both limits were exceeded. No regions were flexural and shear section identified where the shear capacity was caDacities were considered. exceeded. The regions of significant ADDropriate combinations of elastic overstress are shown in Figure horizontal and vertical 2. Note that all areas exceeding accelerations were considered. An flexural capacity coincide with estimate of the DBE collaDse safety construction joint locations. margin was made by comoaring the canyon structure exoected peak APPROXIMATE NONLINEAR ANALYSIS acceleration with the oredicted Nonlinear moment-rotation capacity acceleration. relationships were develooed for all construction joints where the seismic A summary of the results obtained demand was exoected to exceed 90^ of in aDDlying the above orocedure to the the rebar yield strength. A typical canyon structure follows. moment-rotation response is shown in Figure 3. Note that the thick-walled Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

137 ©-• Overstress Area n —— Constr. Joint £

Figure 2. Elastic Overstress & Construction Joint Locations

MOMENT Moment Ultimate Capacity

N / Dead Weight Restoring Moment * (Fd/2)

ROTATION Figure 3. Typical Construction Joint Properties Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

138 construction results in a significant of the mode of collapse and to estimate contribution to the flexural capacity the safety margin against collapse. coming from the dead weight restoring Using limit analysis techniques, moment. The yield and ultimate capacity various static collanse mechanisms magnitudes were determined from were evaluated until a lower bound calculations using the BIAX2 computer mechanism was found. The rotational code [5]. The section rotations are a capacity of each "plastic hinge" was function of the rebar bond slip, rebar considered for each mechanism to strain and concrete compression. The ensure that the individual joint model used to predict the rotations rotational capacities were not was adapted from an inelastic model exceeded prior to collapse. If the proposed by Lai [3], rotational capacity of the joint was Using the beam element finite exceeded, then the hinge moment was element model, elastic rotational reduced to the dead weight restoring springs were introduced at the moment as indicated in Figure 3. construction joint locations where an Considerations were also given to the elastic overstress was indicated. A shear/friction section caoacities to large initial spring stiffness was ensure that a shear collaDse did not selected to simulate zero relative precede a flexural collapse. rotation at the construction joints. Based upon acceleration predictions Dead weight and DBS response spectrum from the response spectrum analysis, analyses were then performed. Wherever a linear horizontal acceleration the predicted moments exceeded the first distribution was assumed with the peak knee (dead weight restoring moment) occurring at the roof and one third of of the Figure 3 rotational response, the peak at the ground level. To the spring constant was reduced to an account for the vertical and horizontal estimated secant modulus value. combined accelerations, 40% of the Iterative repetitions of this procedure peak vertical acceleration was assumed were performed until a converged to coincide with the horizontal solution was obtained. Due to the accelerations at collar>se [1]. nonsymmetric nature of the building, The lower bound collapse mode for it was necessary to repeat this the canyon structure is shown in Figure procedure for both a left and right 4. The calculated loading to produce horizontal motion. collapse resulted in a roof level peak The final iteration produced a acceleration of 0.74 g. The estimated fundamental frequency of 3.4 Hz which fundamental building frequency is about a 25% reduction in the coincided with the peak foundation original linear analysis prediction response spectrum value of 0.56 g. of 4.4 Hz. Although nonlinearities Since the peak spectral acceleration occurred in a number of .joints due to is well below the predicted collapse exceeding the dead weight restoring acceleration, a significant margin moments, the rebar yield moment was against collapse was predicted. only exceeded in two .joints (crane level joints). At these joints, only NONLINEAR TIME HISTORY ANALYSIS a small amount of yielding was As a final confirmation of the predicted. Thus, the apDroximate seismic adequacy of the canyon nonlinear response results indicated structure, a nonlinear time history that the DBE loading is well below analysis of the canyon structure is the full capacity of the canyon planned. At the time of this writing, structure. the finite element model of the canyon structure was being modified to account STATIC COLLAPSE ANALYSIS for the estimated cyclic nonlinear A static collapse analysis was behavior of the critical joints. It performed to develop an understanding is anticipated that after the Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

139 Horizontal G Loading

Figure 4. Static Collapse Failure Mode

Drescribed DBE time history evaluation REFERENCES has been performed, scaled up time histories will be applied in order to [1] ASCE, 1986, Seismic Analysis of estimate the DBE magnitude that will Safety-Related Nuclear Structures reach the canyon structure capacity. and Commentary on Standard for Seismic Analysis of Safety Related CONCLUSIONS Nuclear Structures, ASCE 4-86, The somewhat unique thick-walled, American Society of Civil lightly reinforced design of the canyon Engineers, New York, New York. structure required an innovative analytical approach. Conventional [2] DeSalvo, G. J. and R. W. Gorman, elastic approaches were found to 1987, ANSYS Engineering Analysis grossly underestimate the seismic System User's Manual, Swanson capacity of the structure. Approximate Analysis Systems, Inc., Houston, nonlinear analyses were found to Pennslyvania. provide reasonable estimates of the seismic capacity. Final confirmation [3] Lai, Shing-Sham, G. T. Will and S. of the seismic adequacy is anticipated Otani, 1984, "Model for Inelastic upon completion of a nonlinear time Biaxial bending of Concrete history analysis. Members," ASCE Journal of Structural Engineering, V. 110, No. 11.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

140 f4l Lysmer, J., T. Udaka, C. F. Tsai, and H. B. Seed, 1975, FLUSH: A Computer Program for Approximate 3-D Analysis of Soil-Structure Interaction Problems, EERC 75-30, Earthquake Engineerinq Research Center, University of California, Berkeley, California. [5] Wallace, J., 1989, "BIAX2: A Computer Program for the Analysis of R-C Sections," UCB-SEMM-89/12, University of California, Berkeley, California.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

141 SEISMIC ANALYSIS PROCEDURES FOR THE PLUTONIUM PROCESSING BUILDING OF THE SPECIAL ISOTOPE SEPARATION PLANT

C. P. Chen, Bechtel, P.O. Box 3965, San Francisco, CA 94119 F. F. Tajirian, Bechtel, P.O. Box 3965, San Francisco, CA 94119 R.A.A. Todeschini, Bechtel, P.O. Box 3965, San Francisco, CA 94119 H. J. Dahlke, WINCO, P.O. Box 4000, Idaho Falls, ID 83403

ABSTRACT This paper describes the methodology for the seismic soil-structure interaction (SSI) analysis of the Plutonium Processing Building (PPB) which is part of the Special Isotope Separation (SIS) Production Plant. The PPB consists of two structures, the enclosure building and the optics/separator area. These are founded on two independent foundations which are supported on the surface of a soil medium consisting of gravel overlying basalt. The PPB is classified as a safety related structure and is required to withstand the effects of a Design Basis Earthquake (DBE).

INTRODUCTION o The three dimensional The response of a structure nature of the problem. during an earthquake depends on the characteristics of ground To account for the above as motion, the properties of applied to the PPB structure, the surrounding soil, and the dynamic CLASSI (£ontinuum Linear Analysis properties of the structure. In for Soil Structure interaction) recent years, several techniques computer program [1] was selected. for soil structure interaction analysis have been developed. An SEISMIC CRITERIA adequate analytical procedure Design Ground Spectra should account for the following: The design ground spectra for the PPB analysis are the o Variation of soil site-specific response spectra properties with depth. developed for the Idaho National Engineering Laboratory (INEL) site o Non-linear and energy [2]. The design ground spectra absorbing characteristics of normalized to an acceleration of 1. soils. (g) for damping values of 0.5, 2.0, 5.0 and 10.0 percent are shown in o Structure-soil-structure Fig. 1. interaction effects.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 142 The maximum zero period horizontal Dynamic Properties of Soil acceleration (ZPA) is 0.33 (g) and The sub-surface profile at the the vertical is taken as 2/3 of PPB site consists of 0 to 7 feet of the horizontal or 0.22 (g). These predominantly silty or sandy gravel ZPA values are based on the fill overlying a thin, generally Preliminary Study of Earthquake cohesive, layer consisting of mostly Strong Ground Motions at the INEL lean clay with varying proportions site [3]. The design spectra are of sand and silt. Beneath this scaled using the above ZPAs. surface layer is a sandy, silty, or clayey gravel extending to the top Control Motions of the basalt bedrock at the depths The free-field seismic ground ranging from 35 to 45 feet. The motion inputs used in the potentially compressible surface soil-structure interaction layer of natural cohesive soils will analyses consists of synthetic be replaced with sandy, silty or acceleration time histories which slightly clayey gravels. The were generated for three sub-surface mean dynamic properties orthogonal directions using the calculated using downhole seismic computer program BSIMOKE [4]. wave velocity measurements [5] are These time histories are summarized in Table 1. statistically independent and their response spectra, in general In order to account for the envelop the design spectra for all uncertainties in the measured soil applicable damping values and properties, the analysis will be comply with the requirements performed using mean, upper-bound outlined in U.S. NRC Standard (1.5 times mean} and lower-bound Review Plan (SRP) sections 3.7.1. (0.67 times mean) soil properties Each time history has a duration and the results will be enveloped. of 24 seconds with a time increment of 0.01 seconds and are PLUTONIUM PROCESSING BUILDING scaled to 0.33(g) and 0.22(g) for The PPB is a two-story horizontal and vertical input reinforced concrete structure except respectively. The acceleration, for the portions on the roof over velocity and displacement time the optics/separator and laboratory history plots for two horizontal areas which are non-safety related and a vertical component are shown steel truss structures. The PPB is in Fig. 2 thru Fig. 4, approximately 310 by 370 feet and 45 respectively. The comparison of feet in height. The general view of calculated response spectra versus PPB is shown in Fig. 8. The design response spectra for 0.5, enclosure building houses the 2.0, 5.0 and 10.0 percent damping optics/separator area, special values are shown in Fig. 5 thru nuclear material processes and Fig. 7, respectively. storage, chemical support processes and local control rooms, an In the SSI analysis, the control analytical chemistry laboratory and motions are assumed to be the plant operational support equipment. free-field surface motions prescribed at the grade level. Due to vibration isolation The wave input is assumed to be requirements, the optics/separator composed of vertically propagating area is located on a foundation body waves. separated from the enclosure building foundation.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

143 Dynamic Models SOIL-STRUCTURE INTERACTION The optics/separator area is a ANALYSIS symmetrical structure. It houses CLASSI is a linear analysis six separators and seven optics walls program; thus the on a 3 foot thick basemat. these are strain-dependency of soil can not represented using thirteen single be directly considered. It mass stick models. Each mass has 3 requires separate analysis to dynamic degrees of freedom (DDOF) obtain the strain-compatible representing the E-W, N-S and equivalent linear soil properties vertical directions. The sticks are for input. In general, the placed at the appropriate locations CLASSI analysis is performed in and linked rigidly at the base to two stages as follows; form a multiple stick lumped-mass dynamic model. Site Response Analysis: to obtain strain-compatible soil The building consists of concrete walls, columns, and cast-in-place properties to be used in the slabs supported on precast beams and interaction analysis . precast panels. The superstructure Interaction Analysis: to compute is placed on a 2'-6" basemat. A the impedances for the SSI system detailed 3-D model was constructed and the resulting SSI structural using the Finite Element Method (FEM) responses. and is shown in Fig. 9. All major structural elements; walls, slabs, A flow chart describing the steps beams and columns are included in the for performing the SSI analysis model with the exception of is shown in Fig. 10. The non-safety related steel trusses following summarizes the main which are represented by masses steps in the analysis. lumped at the appropriate support locations. The cast-in-place slabs Site Response Analysis on precast beams, are modeled as The computer program SHAKE quadrilateral and triangular plate [6] is used to perform the site elements with orthotropic plate response analysis. This program properties. To reduce computer cost, models the site as a the building masses will be one-dimensional layered soil distributed to a set of horizontal column and assumes that the input and vertical dynamic degrees of motion consists of vertically freedom. The number of DDOF is propagating horizontal shear selected so that all significant waves. Soil nonlinearities are modes of vibration of the structure accounted for in SHAKE using the can be reliability evaluated. The equivalent linear method through distributed masses are checked for iterations of soil parameters, center of gravity and rotational namely the shear modulus and moment of inertia to ensure the damping ratio until structural eccentricities are strain-compatible equivalent soil properly included. properties are obtained.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

144 These strain-compatible equivalent Finally, the results obtained from linear soil properties are then used the first two steps along with the to define the soil-foundation model three control motions (two in the CLASSI analysis. horizontal and vertical) are used as input to CLASSI's SSI program For the PPB site response module to compute the foundation analysis, the site soil will be responses as well as the modeled as a 45 foot layer of gravel in-structure responses. overlying an elastic halfspace with basalt properties. The mean soil BRIEF DESCRIPTION OF "CLASSI" properties from Table 1 will be used PROGRAM as the initial soil properties. The CLASSI (Continuum Linear strain-dependent shear modulus and Analysis for Soil-Structure damping curves for gravelly soils [7] Interaction) is a linear will be used. The input motions are three-dimensional seismic SSI the horizontal components of analysis program developed by Luco synthetic acceleration time and Wong [6] in 1976 at the histories. This analysis will be University of California, San repeated for three soil condition; Diego. Since then, the CLASSI lower-bound, mean and upper-bound. program has been continuously upgraded to expand its Interaction Analysis capabilities and efficiency from The interaction analysis consists those of its initial development. of the following steps. First, Thus various versions of the develop a three-dimensional model for CLASSI program exist in the the PPB foundations. The model will industry, each covering somewhat consist of two foundations different analysis capabilities representing the enclosure building and limitations. The CLASSI and optics/separator areas. These program used for this analysis is foundations are discretized into a the Bechtel version originally number of subregions with the finer developed in 1978, and recently ones concentrated around the modified by Wong and Luco in perimeters of each foundation where 1985. This version has been the stress gradient is the highest. extensively tested against The strain-compatible equivalent analytical solutions obtained from linear soil properties obtained from the technical publications or the the SHAKE analysis are used to define comparable results obtained from the properties of the layered soil validated public domain programs. system. Frequency dependent foundation impedances and seismic The program solves the SSI wave scattering matrices are problems in frequency domain using calculated using CLASSI's GLAYER and the Fast Fourier Transformation CLAF program modules. (FFT) technique. CLASSI is comprised of program modules Second, calculate the fixed-base developed to solve the SSI problem dynamic modal properties of the PPB in separate steps. The analysis and Optics/Separator area in terms of method used in CLASSI is based on frequencies, mode shapes, modal the substructuring technique which participation factors and modal separates the analysis of dampings up to the highest frequency kinematic interaction from that of of interest. inertial interaction in two successive steps. Considering a typical structure on a rigid

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

145 foundation supported on a soil medium Sites, UCRL-53582, Rev. 1 as shown in Fig. 11, the Lawrence Livermore National substructuring technique applied to Laboratory (LLNL), this soil-structure system is California, November 1984. schematically shown in Fig. 12. [3] Woodward Clyde Consultants, The analysis of kinematic interaction "Preliminary Estimates of as shown in block I of Fig. 12, is Earthquake Strong Ground handled by first deriving the Motions at the Idaho "so-called" seismic wave scattering National Engineering matrix, which is then used to Laboratory", April 1989. transform a given free-field seismic ground wave field into a set of [4] Bechtel Power Corporation, effective free-field seismic motions "Bechtel Simulated associated with the structural base Earthquake Motions motion degrees-of-freedom. The (BSIMQKE)", Report No. analysis of inertial interaction is SFPD-c/s-84-02 Dec. 1984. handled by first deriving the foundation impedance matrix using an [5] Northern Engineering and integral equation method and Green's Testing, Inc." SIS functions of a continuum halfspace Geotechnical Evaluation, [9], The foundation impedances are February 1987. then combined with the fixed-base structural impedances to form the SSI [6] Schnabel, P., et. al, "SHAKE system, as shown in block II of Fig. - A Computer Program for 12. Finally, the interaction Earthquake Response Analysis response is calculated as shown in of Horizontally Layered block III of Fig. 12 by subjecting Sites," EERC 72-12, the SSI system to the foundation University of California, input motions at the structure base Berkeley. 1972. resulting from the kinematic interaction from block I as the input [7] Seed, H.B. et. al, "Moduli seismic excitation. For the case of and Damping Factors for multiple structures interacting Dynamic Analysis of through the foundation soil, the Cohesionless Soils", Journal above procedure is extended in a of Geotechnical Engineering, generalized sense to involve the Vol. 112, No. 11, Nov. 1986. structural base motion degrees-of-freedom of all interacting [8] Wong, H. L. and Luco, J.E. structural foundations, as detailed "Dynamic Response of Rigid in Ref. 8. Foundations of Arbitrary Shape", Earthquake REFERENCES Engineering and Structural Dynamics, Vol. 4, 1976. [1] Bechtel Power Corporation, "CLASSI (Continuum Linear [9] Bechtel Design Guide C-2.44, Analysis for Soil-Structure Rev. 0, "Seismic Analyses of Interaction), Structures and Equipments CE934", July 1988. for Nuclear Power Plants, Bechtel Power Corporation, [2] Natural Phenomena Hazards August 1980. Modeling Project: Seismic Hazard Modeling for Department of Energy

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

146 Table 1 Summary of Mean Site Dynamic Properties (Reference 4}

Compress- Shear Damping Shear* Elastic Poisons Unit ional(Vp) (VS) Ratio Modulus Modulus Ratio Weight Material ft/sec ft/sec kiDs/SF kips/SF lbs/ft3 Alluvial Gravel 3300 1400 1% 8200 22,000 0.39 135 Basalt 9200 3900 1% 71000 200,000 0.39 150

(*) For Low Strain Level of 10"4

Figure 1 Design Response Spectra Scaled to l.Og

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

147 :.lT T^TTTT iiyJiiiil- _J

r—-*•——^ •• it

Figure 2 Figure 5 Horizontal (HI) Spectrum-Compatible Comparision of Horizontal (HI) Time History Response Spectra vs Design Response Spectra

Figure 6 Figure 3 Comparision of Horizontal (H2) Horizontal (H2) Spectrum-Compatible Response Spectra vs Design Time History Response Spectra | -?yvy\/^j|^^

Figure 4 Figure 7 Vertical Spectrum-Compatible Conparision of Vertical Response Time History Spectra vs Design Response Spectra

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

148 Figure 8 PPB Structure Isometric

Figure 9 PPB Structure Finite Element Model

3-0 STRUCTURAL SHAKI ANALYSIS PROGRAM STMMCOMMmt F1XED4AMM00M.

CLASSI ( OLAYEH / ClAF )

MPtDMCt MCWAVf

CONTROL MOTIONS

CLA1SI (SSI) KNKTMHfTOffV WW.YSN

Figure 10 SSI Methodology Flow Chart

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

149 Figure 11 Description of Soil Structure System

OMMunON M

Figure 12 CLASSI Substructuring Technique

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

150 NATURAL HAZARD LOSSES: A DOE PERSPECTIVE INJURY AND PROPERTY DAMAGE EXPERIENCE FROM NATURAL PHENOMENA HAZARDS DEPARTMENT OF ENERGY 1943 - 1989

James R. Hill Office of Safety Appraisals Environment, Safety and Health U. S. Department of Energy

INTRODUCTION The author wishes to acknowledge the Historical information on life and property support of Susan Whitmore, for typing, and Janet losses can provide an insight into natural hazard Macon, for researching DOE's Occupational Injury reduction opportunities [1]. Efforts to mitigate and Property Damage data base [8] and for natural hazards have been ongoing for thousands providing the graphics. of years, but the complex infrastructure of our current world requires creative and cost-effective SUMMARY mitigation strategies. The Department of Energy DOE property damage for incidents due to (DOE) has developed several standards and natural hazards reported during the past 46 years guidelines aimed at reduction of losses caused by (1943-1989) totaled nearly $20 million. Losses earthquake, wind, tornado, flood, volcanic ash, due to all phenomena are shown by state in Figure and lightning hazards. This paper presents a 1 and are tabulated in Table 1. Losses due to historical perspective on losses due to natural specific phenomena are shown in other Figures. hazard incidents (1943-1989) at DOE and The largest incident loss for various phenomena are predecessor agencies including the Atomic Energy provided in Table 2. Loss data for the DOE power Commission [2]. This historical focus can support administrations was not readily available; however, planning for 1) loss reduction strategies for DOE informal estimates indicate annual costs of $0.5 to and 2) participation in the International Decade for $1.0 million for restoration of power due to the Natural Hazard Reduction [3] and in the U.S. effects of natural phenomena for the five Decade for Natural Disaster Reduction [4]. 'administrations'. Based upon reported and estimated loss information, natural phenomena A few comments on DOE's upgrading incidents cause over $1 million annually in damage program for natural phenomena should provide a to DOE property. Much of the loss occurs in a perspective on the review of past losses. The relatively few incidents. A listing of the 29 largest Lawrence Berkeley Laboratory was one of the first losses to occur in DOE is shown in Table 3, where DOE sites to complete a site-wide earthquake some losses exceeded $1 million. mitigation program [5]. The earthquake, flood and wind upgrading programs at several DOE sites Information is also collected on injuries were reported at the 1985 DOE Natural Phenomena caused by natural phenomena, but no dollar loss is Conference [6] and seismic mitigation activities for assigned even where medical treatment and lost DOE have been reported annually to Congress work is involved. A comparison of the number of beginning in 1983 [7]. Also, a seismic seminar injuries caused by various phenomena during the was held in 1986 to exchange design and past ten years is given in Table 4. No fatalities upgrading experience among DOE contractors in were reported that could be directly attributed to the San Francisco Bay Area. The information natural causes for the period 1943-1989. presented in this paper should help determine future upgrading priorities. Hurricane damage to DOE facilities is minor compared to the overall U.S. statistics. However,

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

151 the six of the seven hurricanes that damaged DOE • Flood damage occurred where least sites (Table 5) were among the 30 largest DOE expected. natural phenomena incidents. Through this awareness, I hope you are The DOE Computer Assisted Incident encouraged to provide your ideas and your Reporting System (CAIRS) [8] maintains records professional skills for a Decade of Natural Hazard of reported incidents caused by fire and other Reduction in the Department of Energy. accidents and annual [9] and quarterly [10] statistical reports are compiled for incidents REFERENCES including natural phenomena. Special [1] J. H. Wiggens Company, Building Losses investigations are conducted for incidents involving from Natural Hazards: Yesterday. Today and fatalities and large property damage. Tomorrow. National Science Foundation, 1977. The DOE manages or operates research, production, oil reserves and power distribution [2] Division of Operational Safety-USAEC, facilities valued at nearly $100 billion. These Operational Accidents and Radiation Exposure activities are carried on in over 10,000 buildings Experience with the U.S. Atomic Energy (>1000 sq ft) in nearly all states. Total Commission. Wash 1192, Fall 1975. construction averages about $1.5 billion annually [11]. [3] Confronting Natural Disasters: An International DcCflCJC for Natural d Losses due to natural hazards account for Reduction. National Academy Press, 1987. about 5 percent of all losses reported from 1943 to 1975, For the period 1984 to 1988 natural losses [4] Reducing Disasters' Toll: The United States were one percent of the total incidents but Decade for Natural Disaster Reduction. accounted for 20 percent of the total dollar loss. [5] Eagling, D. G., Seismic Safety Guide. LBL CONCLUSIONS 9143 (NTIS No. DE 84000542), 1983. This presentation provides a perspective of DOE losses during the past 46 years even though [6] Lawrence Livermore National Laboratory, loss data was not readily available for all DOE Proceedings: DOE Natural Phenomena operations. As such this paper is considered Hazards Mitigation Conference. CONF- preliminary and more work is needed to provide an 851D118, 1985. informed view of all DOE losses. Review of the reported historical losses has provided an [7] Hill, J. R., "Department of Energy" in Natural opportunity to create an awareness of the extent Earthquake Hazards Reduction Program: and location of a wide variety of natural Fiscal Years Activities (Annual Report to the phenomena hazards that have caused damage at United States Congress, Congress of Federal most DOE sites. Agency activities compiled by FEMA and issued annually beginning in 1983). Some suggestions and observations to consider are: [8] Department of Energy. Computer Assisted Incident Reporting System (CAIRS). EG&G • Mitigation strategies may achieve greatest Data Base, 1989. reductions in wind damage. [9] Office of the Deputy Assistant Secretary for • Most damage has occurred to conventional Safety, Health and Quality Assurance-DOE, construction. Department of Energy Safety and Health Highlights - Fiscal Year 1988 (prepared by the • Lightning damage review may provide Systems Safety Development Center, EG&G insight for design standards change. Idaho, Inc., 1988).

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 152 [10] Office of the Deputy Assistant Secretary for [11] Department of Energy. Real Property Safety, Health and Quality Assurance, Inventory System (data base current 1989). Occupational In jury and Property Damage Summary - January-March 1989 (prepared by the Systems Safety Development Center, EG&G Idaho, Inc., 1989).

TABLE 1 LOSSES DUE TO TYPES OF NATURAL HAZARDS BY STATE DOE 1943-1989

DOLLAR LOSS ($000)

STATE SEISMIC FLOOD HAILSNOW, SLEE. RAINT LIGHTNINGi FREEZING SUBSIDENCE TOTAL

CA 3429 246 258 — 7 17 370 4327 CO 447 1347 87 21 161 2090 FL 237 231 . 39 - 507 ID * 120 66 12 69 54 321 IL 510 14 36 50 88 698 KY * . 23 23 LA 415 • 33 448 MD . 188 . . 188 MO 42 478 4 5 11 540 NM 384 299 38 112 74 907 NV 40 9 8 76 . 133 NY 64 - 20 295 50 429 OH 292 - 9 43 19 363 PA 12 33 3 48 PACIFIC 399 - . . 399 SC 104 110 597 135 182 1128 TN 23 45 17 85 TX 3000 2380 25 54 5459 WV 40 . . . 40 WA 137 66 8 335 106 652 WY 10 16 - 34 - 60

TOTALS 3429 6522** 3109 3190** 1323 860 412 18845

*Mt. Borah earthquake of 1985 resulted in minimal damage. **Losses with both wind and hail damage allocated between wind and hail.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

153 TABLE 2 DEPARTMENT OF ENERGY LARGEST LOSS BY TYPE OF NATURAL HAZARD ($000)

YEAR DESCRIPTION LOSS ($000)

1980 LIVERMORE EARTHQUAKES 3,416

1980 PANTEXWJNJB AND HAJL STORM 2,546

1967 PANTEX WIND AND HAJ& STORM 1,872

1986 ROCKY FLATS PLANT FLOOD 1,174

1973 SAVANNAH RIVER SITE SJJEEI 393

1973 BERKELEY LA22D§LJDJ£ 370

1983 BAYOU CHQCTAW HURRICANE 268

1976 ARGONNE TORNADO 285

1952 PACIFIC TYPHOON 250

1989 BROOKHAVEN LIGHTNING 240

1984 PINELLASEAJH 231

1984 ROCKY FLATS PLANT FREEZING 161

1957 SAVANNAH RIVER SITE SNOW 98

1980 HANFORD AND INEL VOLCANIC ASH *

•Losses not reported.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

154 TABLE 3 MAJOR DOE LOSSES DUE TO NATURAL PHENOMENON

1943-1989

$LOSS HAZARD (OOff) YEAR STATE LOCATION DESCRIPTION

EARTHQUAKE 3416 80 CA Livermore 0.25G, 2 major shocks WIND/HAIL 2546 84 TX Amarillo 106 MPH wind, baseball-size hail WIND/HAIL 1872 67 TX Amarillo Damage to buildings and equipment FLOOD 1174 86 CO Golden Runoff flooded equipment WIND 658 89 TX Amarillo 101 MPH wind, gravel broke glass FLOOD 449 61 MO Kansas City Hurricane Carla SLEET 393 73 SC Aikcn Damaged pine trees LANDSLIDE 370 73 CA Berkeley 6 weeks of heavy rain TORNADO 285 76 IL Argonne Damaged equipment and buildings WIND/RAIN 268 83 LA Bayou Choctaw Hurricane Alicia FLOODING 258 83 CA Santa Barbara Spread Contamination TYPHOON 250 52 Pacific Damage LIGHTNING 240 89 NY Long Island Ignition of gases RAIN 231 84 FL Clcarwatcr Rain caused electrical short WIND/FLOOD 224 80 NM Albuquerque Hurricane Allen, Flooding WIND 221 88 OH Fcmald Damage to trailer and equipment WIND 210 80 CA Tupman Building burned FLOOD 200 83 CO Golden Damage transmission lines FLOOD 188 72 MD Baltimore Explosion from watcr/calcium reactions HAIL 174 77 TX Amarillo Damaged buildings and vehicles WIND 166 80 CO Golden Loosened bolts on equipment FREEZING 161 84 CO Golden Pipe rupture damage (computer) LIGHTNING 156 84 WA Hanford Range fire damage equipment WIND 140 78 IL Argonne Damaged electrical equipment WIND 125 82 CO Golden 116 MPH, property damage FREEZING 112 83 SC Aiken Pipe burst SNOW 98 57 SC Aikcn Collapsed warehouse SLEET 96 58 SC Aiken Damaged pine trees WIND 95 86 LA Bayou Choctaw Hurricane damage

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

155 TABLE 4 DEPARTMENT OF ENERGY INJURY INCIDENTS DUE TO NATURAL HAZARDS 1979-1989

ICE/SNOW 631 WIND 180 LIGHTNING 5 EXTREME COLD TEMPERATURES 11 EXTREME HOT TEMPERATURES 21 RAIN 13 MIXED PRECIPITATION 0 FOG 0 EARTHQUAKES 0 FLOOD 0

TOTAL INCIDENTS 862

TABLE 5 DEPARTMENT OF ENERGY HURRICANE LOSS SUMMARY

YEAR ($000) HURRICANE LOCATION

1961 449 CARLA KANSAS CITY, MO 1983 268 ALICIA BAYOU CHOCTAW, LA 1952 250 TYPHOON PACIFIC 1980 224 ALLEN ALBUQUERQUE, NM 1972 188 AGNES BALTIMORE, MD 1986 95 BAYOU CHOCTAW, LA 1959 50 GRACIE AIKEN, SC 1943-1989 1524 7 EVENTS 6 STATES

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

156 DEPARTMENT OF ENERGY DOLLAR LOSSES FROM NATURAL CAUSES 1943 - 1989 ($000)

HAWAII T<7 < 100 ^ 100 - 500 500 - 1000 S£3 1000 - 6000

MHcoJ Figure 1

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 157 DEPARTMENT OF ENERGY FLOOD LOSSES ($ OOO) 1943 - 1989

NORTH DAKOTA \ MINNESOTAr WISCONSIN

500 - 1500

Figure 2

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 158 DEPARTMENT OF ENERGY WIND LOSSES ($000) 1943 - 1989

WISCONSIN «> /H ~\

^ ^ MICHIGAN

HAWAII

7'2 < 25 ^25-100 r "s 100 - 500 3000 MHcft> Figure 3

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 159 DEPARTMENT OF ENERGY LIGHTNING LOSSES ($ OOO) 1943 - 1989

MONTANA I NORTH DAKOTA I MINNESOTA

>J SOUTH DAKOTA

\ IOWA NEBRASKA \ Y7"7

KANSAS

HAWAII < 25 25 - 100 100 - 500 MUc7k Figure 4

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

160 DEPARTMENT OF ENERGY EARTHQUAKE LOSSES ($ OOO) 1943 - 1989

1 OKLAHOMA ARKANSAS /_-

MS j V GEORGIA ALABAMA)

HAWAII

4000 - 5000 HU.CS Figure 5

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

161 SEISMIC ANALYSIS AND TESTING OF CLAY TILE WALLS AT THE OAK RIDGE Y-12 PLANT

K. E. Fricke and W. D. Jones Martin Marietta Energy Systems, Inc. P. 0. Box 2003 Oak Ridge, TN 37831-7231

ABSTRACT

The recent DOE 6430.1A General Design Criteria has emphasized the importance of determining the adequacy and, hence, safety of both new and old facilities to natural phenomenon hazards such as earthquakes and high winds. In order to meet the criteria, an existing unreinforced clay tile wall, which is an integral part of a new facility being placed in an old building, has been evaluated for resistance to seismic events.

Part 1 of this paper consists of the analytical studies. The facility was mathematically modeled and analyzed using a finite element program. The material properties used in the analysis are based exclusively on data available in the current engineering literature for masonry blocks and walls. The results of the analysis conclude that the wall is adequate to meet the seismic requirements per the new criteria, but the results of the testing program described in Part II will eventually need to be incorporated into the analysis.

Part II documents the results of a testing program to obtain material properties of the masonry and verify the values used in the analysis of Part I. The fact that most of the available testing data is on brick and concrete block and that the condition of the walls throughout the plants is suspect led to the testing program. The following tests on clay-tile walls, units, and panels were performed: (1) in-situ mortar joint shear strength of existing 12" walls, (2) compression strength, (3) tensile strength, and (4) diagonal tension (shear) strength of panels taken from the existing walls. The test results at this time are fairly inconclusive and have high standard deviations. The testing program is ongoing and is currently being expanded.

PART I - ANALYSIS 15910 [2]. The EUCF has been designated as a moderate hazard facility. Thus, according to UCRL-15910 the DBE INTRODUCTION is an earthquake having a return period of 1000 years and a peak ground acceleration (PGA) of 0.19g.

EUCF Description The purpose of this study is to evaluate an unreinforced clay tile wall in the Enriched Uranium Conversion Facili- The EUCF is a FY-1985, $20 million capital project, known ty (EUCF), for a design basis earthquake (OBE) according as "Dry Chemistry", that will chemically convert enriched

to the requirements of the "General Design Criteria" uranium hexafluoride (UF6) to uranium tetrafluoride powder (GDC), DOE Order 6430.1A [1]. The OBE is defined in (UF^). The facility is located in the D-Wing of the 9212 "Design and Evaluation Guidelines for Department of Energy building complex, shown in Figure 1, at the Y-12 Plant in Facilities subjected to Natural Phenomena Hazards", UCRL Oak Ridge, Tennessee. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

162 Suildinet Complex Description built-up roofing. The basement (walls, slab and piers), the first floor and the foundations are of concrete con- The 9212 building complex was constructed in various struction. The footings bear directly on rock. The stages during the 1940's and early 195Q's, with initial basement, which is only 7-ft deep from the first floor to construction beginning in 1946, The original structure the basement slab, is a very stiff, rugged structure. Two was completed in 194?; it consisted of the two story head- interior rows of piers, spaced on 12-ft centers, extend up house with a basement and the four structurally inde- to provide support for the first floor. The basement pendent one story wings (A, B, C and D) with basements. walls are 10-in. thick and are cast integrally with the The four wings were constructed of load carrying steel piers supporting the main building columns. Lateral frames, with simple connections and non-load bearing Stability in the transverse direction (H-S) is provided by hollow clay tile (HCT) walls from the ground floor up. typical steel column-truss type construction. Lateral The infill wings, A-1 to D-1 and A-2, were added during stability in the longitudinal direction (E-W) is provided later expansions. The roof construction is composed of by HCT watts, which are the object of this study. trusses and purlins which support a gypsun deck with

figure 1. Isometric of 9212 Building Complex

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

163 PREVIOUS STUDIES The 1987 Revaluation Study

General After the completion of the 1981 evaluations of the 9212 building complex, the evaluation team went on to analyze The purpose of these evaluations were: (1) to provide many of the other facilities at the Y-12 Plant. As those technical support for the facility Safety Analysis Re- studies continued, as additional research data on the ports, (2) to determine the potential hazard and risk from seismic resistance problem continued in the technical earthquakes, and (3) to determine requirements and recom- community, and as data from actual earthquakes accumu- mendations to provide seismically safe facilities and lated, it became apparent that such structures, systems reduce risks. The approach to these seismic evaluations and components tended to perform much better than pre- was to conduct state-of-the-art seismic analysis of the viously expected or estimated by analysis. For example, facilities. Because these facilities were large and many facilities examined following major earthquakes have extremely complex in the way they structurally interact, performed extremely well [71. In addition, the perfor- an approach was taken to evaluate the facilities that mance of masonry infilled walls in the Mexico City earth- maintained a state-of-the-art approach while at the same quake of 1986 [8], for example, showed that once infilled time optimizing the cost-effectiveness of such studies. walls lose their shear strength they do not necessarily The approach has been developed as an evaluation criteria collapse, but can remain within the framing members and guide [3] and has been refined and discussed at various act as excellent earthquake energy dissipators. It became national meetings tA,5], The evaluation technique for apparent to the evaluation team that the SAI 9212 building the building structures consisted of performing a response complex and 9206 building earthquake evaluation studies spectrum, two-dimensional (plane-frame), linear-elastic should be reviewed and the knowledge gained on the perfor- finite element analysis. Since the 9212 building complex mance of structures during the past several years be rests on rock a soil-structure interaction analysis was integrated into this revaluation. Thus, in mid FY-1987, not required. the 1981 study of 9212 was revisited and a report published [9]. By applying more current knowledge to the previous study, and with a better understanding of the behavior of HCT walls, and modifying the mathematical finite element The 1981 9212 Complex Study models accordingly, the revaluation was performed. The reevaluation continued to support the SAI study's previous In 1979 the operating contractor subcontracted to Sci- conclusion that failure of the D-Uing non-load bearing HCT ence Application Incorporated (SAI) to conduct a tornado/ walls is not expected to take place below the 0.3g EPGA in extreme wind and seismic vulnerability study of the 9212 the transverse (N-S) direction because the large steel building complex and Building 9206 using the evaluation frame and truss system is the dominant lateral resistance criteria specified [3]. This work was completed in 1981 structure. However, in the longitudinal (E-W) direction and a report was issued [6]. In this evaluation, D Wing where the HCT walls provide the main lateral resistance, was not specifically addressed since its construction was potential damage was now found to occur between C.10g- basically identical to the A, B and C wings. In addition, 0.15g rather than the 0.05g-0.10g of the SAI study, and each wing is essentially independent from the headhouse or expected collapse was estimated to occur around 0.16g- the adjacent infilled wings. Thus, in the evaluation, two 0.18g. These new results, showing the HCT walls being typical plane-frame models, representing the longitudinal much more resistant than previously thought, were the and transverse directions, were chosen to represent all result of using new data to obtain the shear stress four of the wings. capacity of masonry walls [10,It]] and a more realistic model of the walls. Based on the assumptions used in this study on the performance of structures during earthquakes and how the generic model of A through D wings should be constructed, SEISMIC ANALYSIS the analysis results indicated that at an earthquake excitation of around 0.05g-0.10g effective peak ground General Requirements acceleration (EPGA), the HCT walls would experience damage and permanent deformation. Also, it was postulated that In February of 1988 the 0OE Oak Ridge Operations (ORO) some significant local damage may occur due to banging office directed that implementation of the GDC DOE Order against the headhouse and adjacent wings. 6A30.1A begin. Section 0111-2.7.1 of 6430.1A, entitled "Buildings and Other Structures", states that "Earthquake

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

164 loads for buildings and other structures shall be in inelastic demand-capacity ratio, which is 1.5 for masonry accordance Kith the provisions contained in the UBC and walls; however, since the wall is a non-load bearing

UCRl 15910". Furthermore, Section 0111-99.0 "Nonreactor unreinforced wall, Fu was taken as unity. Also, since all Nuclear Facilities-General" in Sub-section 0111-99.0.4 the equipment that produces inertial loads on the wall is "Earthquakes", states that "Such systems, components and in place, only dead toad was considered. structures shall be designed to withstand a DBE ..." The DBE for the Oak Ridge facilities is defined in UCRL 15910 Model Description [2] to be a 1000 year return period earthquake having a PGA of 0.19g. For moderate hazard facilities UCRL 15910 D-Uing is a one story structure with a basement. The requires that a dynamic analysis be performed unless wing is 336-ft long, 36-ft wide, and 22-ft high. The justification is otherwise given. seismic resistance in the longitudinal direction consists of two unreinforced 12-in. HCT walls; however, since the Material Strength roof was considered to act as a rigid diaphragm, only one wall (Column line T in Figure 1) was analyzed. The HCT wall resistance to lateral loads is through shear deformation (and hence, shear strength). While the The wall was modeled using 215 nodes, 168 membrane ele- shear deformation is somewhat restrained by the steel ments, and 87 beam elements. The wall was considered to frame, the major deformation resistance of the steel be constructed of nominal 12 x 12 x 12-in. blocks using frame-wall system is the wall's shear strength charac- end construction having 9 cells, each 3-in. square. A teristics until the wall fails in shear, after which the block has two 1/2-in. thick internal webs and two 3/4-in. steel frame controls the deformation behavior. Therefore, external webs as shown in Figure 2. The head joint was the shear strength capacity of the HCT walls is a major considered to be the same thickness as the external web, factor in determining the expected performance of D-Wing so the effective thickness of the membrane elements model- in the longitudinal direction. ing the blocks was 2-in. Since the wall is the primary resisting structure, a comprehensive study was undertaken Response Spectrum to determine the properties to use in the analysis. This work (Appendix B of Reference [13]) recommends that 60 psi The Newmark-Hall response spectrum mean shape [12] was used. It was computed for the case of 12X damping as recommended [2] for masonry shear walls and anchored to a 0.19g ground motion acceleration as defined for a moderate hazard facility. The end points (on a log-log plot) for this spectrum are given below:

Point Frequency Acceleration (Hz) (g's) A' 100 0.19 A 33 0.19 B 8 0.29 C 1.51 0.29 D 0.24 0.05 E 0.10 0.01

Loads

UCRL 15910 [2] states that the demand 0 for all elements shad be computed as follows:

0 = CF(DL) + F(LL) + F(EO)]/Fu where F(OL) and F(LL) are realistic estimates for dead and Figure 2. Assumed 12-in. Clay Tile Block live loads of existing facilities and Fu is the allowable Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 165 facilities in the Y-12 plant are over 40 years old, built during or shortly after the war. The philosophy in per- 2.4' forming an analysis of an existing facility has been to attempt to obtain a realistic assessment of the actual 3 SPACES • 8'-0" strength of the facility in order to estimate the correc- 9) 100 105 tive actions needed to bring the seismic resistance to an 72 9 I 76 9> 80 11 >4 84 9 ee acceptable level. Although data from the published literature [10,11,14]] has been used to estimate the shear, bending, and tensile strengths of the walls, the use of 71 75 79 83 87 this data for our analysis is suspect due to the inherent 9! >3 uncertainties in the state of the facilities. The construction specifications for most of these facilities are 70 74 78 82 86 not available, the facilities, especially those built during the war, were designed and built assuming a fairly short life-span, and, in general, titere has been some ra 69 73 77 ai 85 deterioration of these facilities over the years due to l< 6 aging and use.

Figure 3. Typical Portion of the Wall FE Model The current atmosphere of a greatly increased emphasis to consider the impact of natural phenomenon events on the corresponding on-site and off-site consequences to the and 1,500 ksi be used for the shear strength and the general public, the plant operators, and the environment modulus of elasticity, respectively. A typical portion has had the effect of increasing the degree of verifica- of the model is shown in Figure 3. tion required to certify the safety analyses. This eventually led to the implementation of the current in-house testing program. ANALYSIS RESULTS AND CONCLUSIONS The testing program reported herein consists of the Six eigenvalues and the corresponding vectors were used following: in the response spectrum analysis. The natural frequency was 15.7 Hz with the first mode having W.2% of the effec- 1. In-situ mortar joint shear strength, tive mass. The second through the sixth frequencies 2. Compression strength of clay tile bricks, ranged up to 30.0 hz. The maximum horizontal shear, T^, 3. Tensile strength of clay tile bricks, and was 35.1 psi, well below the postulated failure value of 4. Shear strength of clay tile panels, 80 psi. The maximum compressive stress was 42.1 psi which approximately 48-in. x 48-in.. is also very low for clay tile. The fact that the assumed cross-section of the 12-in. walls was radically different All the individual units and panels were removed from from the actual case (discussed later in Part 2) impacts existing facilities that are being analyzed for resistance the results in a positive manner - since the true wall has to natural hazard events. The original testing plan a mortar bed joint width of 12-in. instead of the assumed called for three in-situ tests to be performed, but two effective 2-in., the shear strength of the HCT walls in- other walls became available and an additional six in-situ creases. It was concluded that the wall will not fail tests were run. It is expected that more such tests will inplane at a DBE of 0.19g. be performed in the future, adding to the data base and improving the statistical results.

PART II - CLAY TILE TESTING IN-SITU TESTING INTRODUCTION Introduction and Test Setup As part of the support for the natural hazards (both seismic and wind) analyses ongoing for the DOE Oak Ridge The in-situ test is a partially destructive test which facilities, a program to test the strengths of existing measures the sliding shear strength of a mortar joint by masonry walls was initiated this past winter. Many of the displacing a single unit horizontally with a hydraulic Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

166 rant. Three 12-in, unreinforced clay-tile walls were se- and it was decided to determine the mortar joint shear lected to test, with three locations in each wall. The strength across the full 12-in. width of the wall. How- testing procedure is basically that recommended by Schmid ever, in order to test the entire width of the wall it was et. al. C151 for the City of Los Angeles. The first wall necessary to modify the set-up procedure to obtain • hole tested, along column line T in Building 9312, is the sub- that was square and flush at both ends. This was accom- ject wall of the analysis described in Part 1 of this plished by attaching a 4-in. long section of clay tile paper. The other two wads are located in Building 9206 (the horizontal offset along each row was 4-in.), either at the Y-12 Plant. 4-in. or 8-in. wide depending on which side of the wall was being fitted. * 10 x 10 x 1/2-in. steel plate was The typical 12-in. thick clay tile wall that exists in attached at each end of the opening. A 30 ton EnerPak these facilities, consists of a 12 x 12 x 8-in. thick tile hydraulic ram was placed in series with a 10 ton load and a 12 x 12 x 4-in, thick tile, using a running bond cell. Linear Variable Differential Transformers (LVDT) type side (not end as assumed for the analysis) construc- were attached on each side of the wall to measure the tion, with the 8-in. and 4-in. tiles staggered both verti- horizontal displacement of the test tile and surrounding cally and horizontally as shown in Figure 4. blocks during the test. The magnitude of the applied load was measured with both the load cell and a pressure trans- ducer connected to the hydraulic ram. A typical test setup is shown in Figure 5. In this figure, three LVDT's are shown attached to the near side wad. This arrangement was used for the first test only; thereafter a single LVDT was used on each side of a wall at the elevation of the test tile. The data acquisition system was comprised of an HP-9825 computer and an HP-3497A scanner unit. An X-Y plotter provided monitoring of the load-deflection data for the LVDT's connected to the test tile.

Test Results

A preliminary functional test was run, to a low load, prior to each test to verify that all the components were operating properly. During the main load application phase, load was applied continuously to the test tile at approximately 10 kips/minute, until either the applied load was no longer increasing or the head joint gap had closed. It had been expected, once the initial bond had been broken between the mortar bed joint and the adjacent upper and lower tiles, that a substantial drop in applied Figure 4. Typical 12-in. Clay Tile Wall load would take place. This behavior was observed only in two of the tests. The normal behavior was for the load to continue to increase, though the slope of the loading The normal procedure [15J for performing the in-situ curve dropped considerably, until the head joint had mortar shear test calls for drilling out the mortar closed or some other obstruction had been contacted. The around and removing one tile immediately behind the test failure load, unless it was clearly defined, was taken to tile, removing the head joint of the test tile by drill- be the load that occurred at a horizontal displacement of ing out the mortar, and attaching a 1/2-in. thick steel 0.060-in. plate behind the test tile (in order to distribute the load to the test tile more uniformly and ensure a mortar No damage was visible to the wall further than a few joint failure and not a tile failure). A hydraulic ram, inches away from each test location. Some hairline cracks inserted in the opening, applies load to the test tile, were observed along adjacent mortar joints in a couple of with the reaction load being taken by the wall behind the locations. The general failure surface was between the ram. The construction of the walls in these facilities is mortar joint and the tiles above and below the test such that the entire cross-section will behave as a unit, block, i.e., the mortar bed displaced with the test tile.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

167 There was no visible displacement of adjacent blocks are certainty important factors to consider. Visually during any of the tests. Figure 6 shows a closeup of the there was no obvious reason for the scatter of the data. crack pattern obtained for Wall No.1 Test No.1 (designated Statistical analysis of the shear strength data can be

i-V

Figure 6. Typical In-Situ Failure Mode

TABLE 1

SUMMARY OF IN-SITU HORTAR JOINT TESTS

Nominal Failure Shear Test Building Load Stress No. No. (kips) (psi) Figure 5. In-Situ Test Arrangement 1-1 S212 28.4 99 as test 1-1). A summary of the test results from the nine 1-2 9212 20.3 71 in-situ tests conducted so far are presented in Table 1. 1-3 9212 45.0 156 2-1 9206 17.0 59 The mortar joint area is nominally 12-in. x 12-in. on 2-2 9206 25.5 88 both the top and bottom surfaces (as seen in Figure 4 the 2-3 9206 31.5 109 blocks are placed with the openings running horizontally) 3-1 9206 19.5 68 so the total mortar joint area is nominally 288 sq. in. 3-2 9206 40.0 139 Measurements taken prior to and after the tests in Build- 3-3 9206 29.5 102 ing 9212 resulted in an actual mean mortar joint area of 275 sq. in. In the table, the nominal bed area used. mean = 28.5 99 standard deviation 9.4 32.5 As denoted by the rather large standard deviation, the data indicates that a large variation exists in the shear strength of the bed joints, both within a given wall and made assuming the sample was taken from an approximately also between different walls; the reasons for this wide normal population. In the case where the population mean range are not clear, though the quality of the mortar and and variance are not known, and the sample size n < 30, a the quality of the actual construction of the bed joint confidence interval can be obtained using the t-distribu-

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

168 tion using v = n-1 degrees-of-freedom, or 8 in this case. The walls in the older buildings consist primarily of Performing these calculations results in a 5X probability the older clay tile, though with facility modifications (9QX confidence limit) that the true mean of the popula- and occasional repairs, walls generally contain a small tion is less than 79 psi, which leads to the conclusion percentage of the newer tile-. It can be seen from these that the use of 80 psi as the mean shear strength of the masonry bed joints is not unreasonable. Until more test TABLE 2 data becomes available, there is little else that can be concluded from the immediate data. Further tests are SUMMARY OF COMPRESSION STRENGTH TESTS required to investigate the repeatability of the large variance. Gross Compressive Width Length Height Area Load Strength Sample (in.) (in.(in.) ) (in/) (kips) (psi)

LABORATORY TESTS 1 8.19 6.00 11.72 49.K 188.5 3826 2 8.13 6.00 11.94 48.78 122.0 2501 Coitcressive Strength Tests 3 8.25 6.04 12.25 49.83 96.5 1937 4 8.19 6.13 12.19 49.14 90.0 1832 Fifteen clay tile Dricks were tested per ASTM C-67-85 5 8.19 6.13 12.38 50.20 77.5 1544 [16] in the direction of the depth of the brick. Nominal- 6 8.25 6.13 12.38 50.27 90.0 1780 ly, two sizes of 12-in. long x 12-in. deep clay tile 7 8.13 6.00 12.25 48.78 92.5 1895 blocks were tested, the 8-in. and the 4-in. wide units. 8 8.07 5.94 12.13 47.94 106.5 2222 The tile units were cut in half per ASTM C-67 and then 9 8.00 5.75 12.13 46.00 128.0 2783 measured in the laboratory prior to being tested. Table 10 7.88 6.25 12.19 49.25 97.0 1980 2 [19J presents the results of the fifteen compression 11 4.00 5.81 12.50 23.24 85.0 3657 tests. The compressive strength in Table 2 is simply 12 3.94 6.13 12.31 24.15 57.5 2381 calculated as C = P/A, where A is the gross area. 13 3.94 6.00 12.25 23.64 79.0 3342 14 4.00 5.88 12.25 23.52 66.0 2806 15 4.13 6.25 12.00 25.81 111.5 4320 Splitting Tensile Strength ======mean = 2587 Seven clay tile bricks were tested per ASTM C-1006 C17] standard deviation = 84V in order to determine the splitting tensile strength of the individual units. Four were original clay tiles (dark red) removed from the walls and three were new clay tiles TABLE 3 (light red). The units were nominally 12-in. long x 12- in. deep and came in 8-in, and 4-in. widths. The clay SUHHARr OF SPLITTING TENSUE STRENGTH TESTS tile bricks were each measured prior to each being tested. The bricks were tested in the same spatial orienta- Gross tion that they have in the walls (the holes running hori- Tile Dimensions Net Splitting zontally), with the lirve load running perpendicular to the W L H Area Load Tensile axis of the holes and parallel to the brick width. Table No. Color (in.) (in.) (in.) (in. ) (kips) (psi) 3 [19] presents a summary of the splitting tensile tests. 1 LT. RED 8.27 12.34 12.40 41.3 162.6 251 The splitting tensile strength, T, is calculated as 2 DK. RED 8.24 12.21 12.19 40.3 241.5 382 follows: 3 I.T. RED 8.27 13.00 12.28 39.0 127.5 208 4 DK. RED 8.14 12.09 12.10 38.3 292.0 486

T = 2P/*Anct 5 DK. RED 8.00 12.24 12.13 38.4 259.0 430 where P is the maximum applied load, and the net area is 6 LT. RED 4.17 12.59 12.t-6 21.5 76.0 225 calculated as

169 tests that the newer clay tile appears to be substantial- will be performed. Care must be taken to minimize the ly weaker than the older tile, A statistical analysis of damage to the panels. this small amount of data leads to the conclusion that the difference between the two sample means is ir fact sig- The testing program is currently being expandeJ and The nificant. The age of the clay tile units tested in University of Tennessee Civil Engineering Department has compression was not noted, so that data is not available been subcontracted to assist. In addition to the tests for comparison. already performed, there are plans to run an in-situ type panel test by isolating a section of the wall. Diagonal Tension (Shear) Tests of Masonry Panels Based on the information available at this time, there The last of the laboratory tests that were performed was is no reason to modify the 80 psi value being used for the the diagonal tension test, which was run on three -nasonry shear strength of the clay tile walls for analysis. panels removed from a wall in Building 9206. The panels were roughly 4-ft square when removed from the wall, but ACKNOWLEDGEMENTS were cut to a smaller dimension. The test procedure, per ASTH E-519 [18], tests for the diagonal tensile (or shear) The in-situ testing was performed with the assistance of strength of the masonry panel by loading it in compression the Stress Analysis Group of the Technical Services Divi- along one of the diagonals. The loading produces a diag- sion of Martin Marietta Energy Systems. The unit and onal tension failure with the panel splitting apart paral- panel testing was performed by the Geologic Associates lel to the direction of the load. Table 4 [19] shows the Division of the Engineering, Design & Geosciences (EDGE) results of these tests. The masonry panels were removed Group in Knoxville (compression tests) and Nashville from an existing facility wall. Due to health and safety (tension and panel tests), Tennessee. regulations, the panels were subject to a decontamination process, which included sandblasting and cleaning with powerful chemical cleansers and brushes. The researchers REFERENCES noticed significant damage to the wall samples including fractures at the mortar-block bonds, damage which could [1] . "General Design Criteria", DOE Order 6430.1- have been caused not only during the cleansing process, A, Li. S. Department of Energy, Washington, D.C., but also during shipping to Nashville. The exact extent Draft, December 25, 1987. to which the damage affected the results is not obvious, but the shear stress results were very low in comparison [2] Kennedy, R. P., et al, "Design and Evaluation Guide- to the previous data. These results probably will not be lines for Department of Energy Facilities Subjected factored into the analysis until further testing is done to Natural Phenomena Hazards," Report No. UCRL- to verify the low values. 15910, prepared for U. S. Department of Energy, Washington, D. C, Draft, April, 1988. SUMMART [3] Beavers, J. E., "Seismic Evaluation Criteria for the The data from the testing program is still in the proc- DOE-ORO Reservations," Union Carbide Corporation • ess of being interpreted. At present the data from the Nuclear Division, Oak Ridge, Tennessee, May 1982, diagonal tension tests is suspect, and further testing Unpublished.

TABLE 4

Diagonal Tension Tests of Masonry Panels

Gage Net Max. Shear Shear Modulus Sample Width Height Thickness Length Area Load AH Strain Stress Rigidity (in.) (in.) (in.) (in.) (in.2) (lbs) ( .001") (.001") (in./in.) (psi) (psi)

A 32.8 32.8 13.0 46.4 188.5 4391 .057 .061 .002543 16.5 6476 8 36.5 36.2 12.7 51.4 194.8 6922 .069 .069 .002684 25.1 9358 C 42.1 42.1 12.8 59.0 241.6 12407 .046 .038 .001423 36.3 25501

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 170 [4] Hanrod, w. E., Hall, W. J., and Beavers, J. E., [14] Plummer, H. C, Brick and Tile Engineering. Brick "Seismic Evaluation Criteria for Existing Critical Institute of America, 1962. Facilities," Proceedings. Earthquakes and Earth- quake Engineering: The Eastern United States. [15] Schinid, B., Kariotis, J., and Schwartz, E., "Tent- Volumes, Knoxville, Tennesee, September, 1981. ative Los Angeles Ordinance and Testing Program for Unreinforced Masonry Buildings," Proceedings of the [S] Manrod, W. E. and Beavers, J. E., "Cannon Factors SEAC Annual Meeting, 1978. in Evaluation of Critical Industrial Facilities," Proceedings. Seismic Risk and Heavy Industrial [16] ASTM C 67-87, "Standard Methods of Sampling and Facilities. San Francisco, California, May, 1983. Testing Brick and Structural Clay Tile," American Society for Testing and Materials, Philadelphia, [6] Johnson, N. E. and Walls, J. C, Seismic Resistant PA, 1987. Capacity of Y-12 Plant Facilities 9212 and 9206. Report No. SAI-148-020, Revision 1, March 10, 1981. [17] ASTM C 1006-84, "Standard Test Method for Splitting Tensile Strength of Masonry Units," American Soci- [7] Kennedy, R. P., et. al., Engineering Characteriza- ety for Testing and Materials, Philadelphia, PA, tion of Ground Motion. NUREG/CR-3805, Vol. 1, 1984. Appendix A, U. S. Nuclear Regulatory Commission, 1984. [18] ASTM E 519-81, "Standard Test Method for Diagonal Tension (Shear) in Masonry Assemblages," American [81 Bertero, v. V., "Lessons Learned from Recent Earth- Society for Testing and Materials, Philadelphia, quakes and Research and Implications for Earthquake PA, 1981. Resistant Design of Building Structures in the United States," Earthquake Spectra. EERI, Vol. 2, [19] , Hartin Marietta Prism Testing. Testing No. 4, October 1986. Report submitted by the Engineering, Design I Geosciences (EDGE) Group to Martin Marietta Energy [9] Williamson, D. H., Reevatuation of Seismic Resis- Systems, Inc., August 22, 1989. tance Capacity of Y-12 Plant Facility 9212. Martin Marietta Energy Systems, Inc., Engineering Divi- sion, Y/EN-1869, August, 1987.

[10] Woodward, K. and Rankin, F., "Shear Resistance of Unreinforced Hollow Concrete Block Masonry Walls," Proceedings. The 3rd North American Masonry Con- ference. University of Texas, Arlington, Texas, June, 1985, pp. 38-1 - 38-15.

[11] Adham, S. A., "Out-of-Plane Response of Masonry Walls," Proceedings. The 3rd North American Masonry Conference. University of Texas, Arlington, Texas, June, 1985, pp. 47-1 - 47-14.

[12] Coats, D. W. and Murray, R. C, Natural Phenomena Hazards Modeling Project: Seismic Hazard Models for Department of Energy Sites. Report No. UCRL-53582, Lawrence Livermore National Laboratory, University of California, November, 1984.

[13] Beavers, J. E. et al., Design Basis Earthqua- ke Evaluation of the Enriched Uranium Conver- sion Facility. Report No. Y/ENG/SAR6-RESP1- 2430 (draft), Martin Marietta Energy Systems, Inc., December, 1988.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

171 SEISMIC, HIGH WIND, TORNADO, AND PROBABILISTIC RISK ASSESSMENTS OF THE HIGH FLUX ISOTOPE REACTOR

S. P. Harris^ R, L. Stover P, S. Hashimojo J. O. Dizon

ABSTRACT

Natural phenomena analyses were performed on the High Flux Isotope Reactor (HFIR). Deterministic and probabilistic evaluations were made to determine the risks resulting from earthquakes, high winds, and tornadoes. Analytic methods in conjunction with field evaluations and an earthquake experience data base evaluation methods were used to provide more realistic results in a shorter amount of time. Plant modifications completed in preparation for HFIR restart and potential future enhancements are discussed.

The submitted manuscript has been authored by i contractor of the VS. Govtrament under contract No. DB- AC05440R2HOO. Accordingly, (he U.S. Government rcuini a nonexclusiv*, royalty-fret license to publish or reproduce the published form of this contribution, or allow others to do so, for U.S. Government Purposes.'

INTRODUCTION systems for the PRA. The High Flux Isotope Reactor is a high power Based upon industry precedent for older, density research reactor licensed to operate at 85 MWt. commercial nuclear plants, the Phase I analyses were It was built in 1965 as a facility to produce transuranic planned for completion before reactor restart and the elements. Phase II issues after restart. The Phase I seismic In 1986, the reactor was shut down in order to analyses were used as a basis for making plant evaluate the effects of neutron irradiation on the integrity modifications before restart. The Phase II analyses will of the reactor vessel. During this shutdown, it was also be used in conjunction with a PRA to determine the considered prudent to evaluate the HFIR design against extent of further upgrades to HFIR during its remaining current codes and practices for modern reactors. To life. support this evaluation, EQE Engineering was asked to Since HFIR was built over 30 years ago, design perform deterministic seismic, high wind, and tornado criteria for earthquakes and other natural phenomena analyses, as well as develop fragility data for a hazards were considerably less stringent than today. In probabilistic risk assessment (PRA). order to restart HFIR expeditiously, practical and cost The assessment was divided into two phases. Phase effective methods for upgrading the plant were required. I consisted of seismic analyses performed on systems A strictly theoretical approach would have been costly required for safe shutdown of the reactor, e.g., reactor and could have been overly conservative, resulting in scram, decay heat removal, and primary coolant system. unnecessary and expensive plant modifications. Phase II consisted of seismic assessment of the accident Therefore, an assessment which included analysis, mitigation systems, high wind and tornado analyses, and earthquake experience data base methods, and development of fragilities for key components and probabilistic risk assessment methods was planned

* EQE Engineering ** Oak Ridge National Laboratory Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

Research sponsored by the Office of Energy Research and j JJ Development, VS. Department of Energy, under contract DE-AC0S-840R21400 with Martin Marietta Energy Systems, Inc. in outer lo achieve a more realistic determination of the seismic evaluation of the primary coolant system. The extent of hardware modifications required. first was a large multi-degree of freedom three- dimensional plate model of the reactor vessel (Figure 3). PLANT DESCRIPTION This model was developed lo provide a precise stiffness The HF1R is located at the Oak Ridge National and force distribution within the irradiated vessel. The Laboratory in Oak Ridge, Tennessee. It is a pressurized, second model was an equivalent beam model of the light water moderated and cooled reactor operating at 85 reactor vessel, primary coolant piping, and four pump MW't. The core consists of highly enriched uranium and heat exchanger cell loops (Figure 4). As with the oxide plates clad in aluminum. There are four primary Reactor Building, analyses of the models, including and secondary cooling loops with a water-to-air cooling representations of piping and supports, reactor vessel and tower as the heal sink. The pressure vessel is 8 feet in supports, control rods and support frame, neutron beam diameter and sits in a reaclor pool 17 feet below the lubes, arid heat exchanger cell components were surface of the water (Figure 1). performed on the Cray. Frequency responses from 1 Hz The local geology and site foundation arc on a stiff to 33 Hz were calculated. The primary loop model was clay shale with an average 20-foot overburden of organic also analyzed for gravity, thermal, and pressure load top soil. The seismicity of the site has been studied cases. extensively and the earthquake selected for ihc Results of the analysis indicated that the as-designed deterministic evaluation was a Newark-Hall spectral configuration contained adequate capacity to resist the shape anchored to a 0.15g horizontal peak ground evaluation seismic load. Seismic deflections in the piping acceleration. High wind and tornado criteria were system were, however, incompatible with clearances and specified as 151) mph. seals and modifications were designed by EQE for upgrade of the primary system. REACTOR STRUCTURES The Reaclor Building, which houses the reactor, REACTOR INTERNALS coolant system, equipment, and experiment rooms, The HFIR reactor core has a high power density and consists of two major structural systems: a massive high peak thermal neutron flux. The design of the reinforced concrete substructure and a lightly reinforced control rod and scram system is unusually rapid-acting concrete-frame superstructure. EQE utilized a large and reactor shutdown is achieved in a fraction of a multi-degree of freedom, three-dimensional finite second. The design involves several complex mechanisms element model of the building (Figure 2). The reactor with close tolerances. Therefore, a detailed evaluation of bay superstructure was modeled in considerable detail lo the seismic performance of the core internals and control capture significant features of the irregular containment mechanism was performed. boundary while the substructure analysis utilized The HFIR reactor internal support structure consists simplified and conservative shear beam and rigid of two concentric cylinders boiled to the central diaphragm models. Response spectrum analyses were cylindrical fuel and reflector support sleeve assembly. performed on a Cray computer system for the two The internal structures were represented as lumped horizontal and vertical input directions, assuming that a massed models and equivalent beam elements. Seismic combination of dead and seismic loads would affect the displacements were determined to be well below shear walls, roof beams, roof slab, and columns. The allowable clearances, precluding contact between analyses indicated sufficient seismic margins for all components. Seismic load margins for support assembly structural elements to preclude collapse and predicted no connections were determined to be acceptable. A shell foundation problems. finite element model of the scram control plates was also The tornado-wind evaluation assumed the developed. These analyses also verified allowable simultaneous occurrence of three events: high-velocity clearances for the control plates under seismic load. (150 mph) wind loading, pressure differential, and tornado missile impact. High wind hazard frequency WALKDOWN OF PRIMARY SYSTEM AND characterization of the HFIR site was derived from site- APPENDAGES specific tornado history and topography. Static analyses The EQE walkdown consisted of a detailed review of were performed to simulate the direct wind and the primary coolant pressure boundary, active differential pressure load on the reactor building model. components, and the emergency coolant pump and DC power supply system. Also included was an assessment REACTOR VESSEL AND PRIMARY COOLANT of the reactor pool inventory isolation and control SYSTEM capability. The walkdown utilized the EQE earthquake Two finite element models were constructed for the experience data base methodology to evaluate most Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

173 i^J?

CIXTBCW. n.»ri oMiwas — ('.'

Figure 1: HFIR Reactor Building East-West Section

Figure 2: HFIR Reactor Building Model

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

174 Figure 3: HFIR Reactor Vessel Model

Primary Coolant Pump (typ. 4) Reactotor Vessee l

Figure 4: HFIR Reactor and Primary System Model

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

175 active components such as valves (Figure 5). FRAGILITIES AND PRA Included in the wulkdown was a review for potential The objectives of probabilistic risk studies are lo seismic interaction hazards. Seismic interaction is estimate the frequencies of occurrence of earthquake or typically caused by fallini», overturning, or deflection of high wind induced accidents, and to identify important non-seismically designeu components, resulting in impact risk contributors in the facility design. The elements of with essential components such us instrumentation, risk analysis include 1) hazard analysis of the site, 2) electrical equipment, or pressure boundary response of plant systems and components, 3) appurtenances. Interaction hazards involving block walls, development of component fragilities, and 4) plant lead shielding, test equipment, and other inadequately systems and event response sequence development. anchored components were identified, mostly on th<; Plant systems analyses and event response sequences ground floor. In most instances, modifications were were developed by Pickard Lowe & Garrick. The site recommended and upgrades designed by EQE. hazard characterizations for high wind and tornado were developed by EQE. Response of plant systems, STACK ANALYSIS components, and structures described above formed the EQE assessed the capacity of the 250-foot exhaust basis for much of the component fragilities along with stack to resist the criteria earthquake and wind loads. additional component and structure-specific analyses. The main concern was that the stack would collapse onto The objective of fragility evaluation is to estimate the the primary containment system. EQE analyzed this peak ground acceleration or peak wind velocity for which reinforced concrete tapered structure with a two- response of a given component or structure exceeds the dimensional model consisting of equivalent beam component capacity resulting in failure. Estimations of elements representing the physical properties of the outer peak response parameter are described as a family of shell. The analysis considered bending stress capacity of curves with a probability value assigned to each curve lo the stack shell, shear capacities, stack overturning reflect the uncertainty and randomness in the fragility resistance, soil bearing capacity, and foundation footing estimation. capacity. Findings indicated that under the evaluation Fourteen structural failure modes and over 35 loads, collapse of the stack is probable 120 feet above the mechanical components were evaluated. Evaluations base. Such a collapse would not reach the Reactor were based on analyses previously described, plant Building. walkdowns, application of earthquake experience data, and component or structure-specific analyses. SEISMIC/TORNADO ASSESSMENT OF Components and structures found to have low capacities ELECTRICAL BUILDING, CONTROL BUILDING, included the electrical building, reactor pool system AND WATER WING tanks, and filters and internal and external masonry walls. The HF1R facility also houses accident mitigation Seismically initiated events were found to be more and key support equipment and systems in three low-rise frequent than high wind or tornado. The dominant concrete and unreinforced masonry buildings. Seismic, seismic sequence was found to involve failure of the high wind, and tornado missile evaluations were electrical building which houses the emergency diesel preformed for these buildings. Static evaluation methods electric generators and results in the loss of all on-sitc were employed to evaluate their structural capacities. AC power. Reactor integrity and pool heat sink integrity Under evaluation seismic loads, vulnerabilities were were maintained in this event sequence. Other sequences found for exterior and interior masonry walls. Most involved seismically-induccd loss of pool heat sink structures were found to be adequate for the evaluation without loss of integrity of the reactor system. earthquake. The vulnerability of masonry walls and Significant high wind and tornado events were found missile penetration from tornado winds were also to be much less frequent. The top event involves the loss established. of the electrical building and all on-site AC power. These events in conjunction with wind damage to the MECHANICAL AND ELECTRICAL SYSTEMS emergency DC power supply to the reactor coolant Field walkdown evaluations utilizing earthquake pumps can result in core damage. experience data methodologies were performed for most accident mitigation and key support systems. These CONCLUSIONS AND RECOMMENDATIONS included auxiliary dicscl generators; heating, ventilating, As a result of the Phase I seismic assessment, 16 and air conditioning; outdoor and indoor transformers; plant areas were identified that represented potential motor control centers; switchgear; batteries; air seismic hazards to the integrity of the decay heat removal compressor systems; tanks; and piping. and primary coolant systems. Design changes and subsequent hardware modifications in these areas were Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

176 • R«com«nded limits on operator height vs. pipe ammeter.

100.04——\ 1 1 1 1 1 1 1 1 1 I

90.0 1(300*)

80.0 2(20001) 1

70.0

CO UJ 60.0 O

.C. BLOCK VALVES a SO.O (FCV 140, 142. 144, 146, UJ 161. 183, 610, 951) oa 40.0 cc UJ a. O FUEL POOL CONTROL 30.0 VALVES (FCV 464. 466, 466)

20.0

10.0

- -'' ' >'s-> W Jf „ 0.0 0.0 2.0 4.0 6.0 S.O 10.0 12.0 14.0 16.0 18.0 20.0 22.0 24.0 PIPE DIAMETER (INCHES) KEY: Number of Valv«« (Operator W«IQM+) * Operator weight it omitted il data it not available Burled pipe data omitted

Figure 5: Histogram Representing the Experience Data Base for Motor-operated Valves, with Recommended Restrictions Superimposed

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

177 completed in 1988. They consisted of 1) strengthening of internal block walls, 2) installation of snubbers and struts on the primary coolant system piping, and 3) restraining coolant lines, let-down valves, radiation monitors, battery chargers, and control rod drive supports. With these upgrades, the HFIR systems required for safe shutdown of the reactor have adequate capacity to survive the 0.15g evaluation earthquake. Other Phase I results concluded that 1) the HFIR reactor building is adequately designed to preclude collapse on the primary system at 0.15g, and 2) the exhaust stack will not collapse onto the HFIR building during a O.lSg seismic event or ISO mph wind speed tornado event. Phase II assessments provided the following conclusions: 1. Ancillary sections of the reactor building, such as the control building and water wing, have adequate capacity to preclude collapse during a O.lSg earthquake. However, structural and nonstructural damage may be anticipated. 2. The electrical building, Special Building Hot Exhaust System (SBHE), reactor system tanks, and electrical switchgear are all vulnerable to a O.lSg earthquake. None of these components affect safe shutdown of the reactor, however. 3. Some of the reactor building components, the control building, and water wing are vulnerable to failure during a ISO mph tornado event. 4. A number of interior block walls have insufficient capacity against tornado induced atmospheric pressure change. Where these block walls affect decay heat removal, design modifications have been initiated to strengthen the walls. 5. The electrical building and SBHE are vulnerable to tornado wind pressures. 6. Many parts of the reactor building are susceptible to damage from missiles during a tornado. The damage is not expected to result in core damage based upon preliminary PRA results. The Phase II results, including the fragilities, have been integrated into the HFIR PRA. Completion of the PRA will determine what, if any, further modifications are desirable to reduce plant risks.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

178 VARIABILITY OP RELATIVE SEISMIC SITE RESPONSE AT LOS ALAMOS, NM

Leigh House and W. Scott Phillips (Consultant) MS D 443 Los Alamos National Laboratory, Los Alamos, NM 87545

ABSTRACT

To estimate the range of seismic response at low strain of sites within Los Alamos National Laboratory, ground motion recordings were obtained at 13 sites from nuclear tests carried out in Nevada. The sites are distributed within a 10 X 10 km area. The ground motions recorded at each site were conceptually modelled as the result of source, path, and site contributions. Because almost all of the paths are in common, the variations seen for each source can be attributed to site response. The sites were monitored in various combinations with seven nuclear tests; each site recorded only a few of the tests. Because horizontal ground motion is more important for structural engineering and was larger than the vertical, we focussed on horizontal site response. The range of relative site response seen is about a factor of 5 to 6 at 1.5 Hz. Topography has a strong effect on response, with sites in canyons being a factor of 3 to 4 lower than nearby sites on mesas. Increased depth to seismic basement beneath some stations also correlates with higher relative site response. Relative site response does not obviously correlate with variation of seismic velocities in the near surface (e.g. upper few meters).

INTRODUCTION distances were all much larger than the differences As part of a geologic and seismologic between the source locations and between the station investigation of earthquake hazards in the area of locations, we could simplify the analysis by removing Los Alamos National Laboratory, we sought to the common path effects. estimate the range of relative site response at The sources for the ground motion recordings several locations within the Laboratory. In addition, were nuclear tests conducted at the Nevada Test we wanted to test whether response spectra for a Site. The usable frequency band of these recordings, single site are representative for the entire about .3 to 3 Hz, is narrower than was hoped for. Laboratory. We used an empirical approach, in Moreover, it is narrower than would be expected to which we recorded several sources at a number of occur during any possible nearby earthquake. different sites. This approach did not require that Nevertheless, the relative site responses found by we know the details of the geologic structure beneath this study should be representative of the relative the sites, since all wave propagation effects are site responses resulting from any possible nearby contained in the recorded data. The empirical earthquake. A study by Rogers, et al, [1] found that approach was well suited to this study because the in the Los Angeles Basin the relative site response seismic velocity structure beneath the sites is not seen in recordings of nuclear tests correlated with known in detail. the relative site response seen from the nearby San We conceptually decomposed the ground Fernando earthquake of 1971. The source to station motion records for each site into source, path, and distances studied by Rogers, et al, [1] are shorter receiver effects. Because the source to station than those used in this study, the frequency band

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

179 used by Rogers, et al, is not dramatically wider than hoped to obtain more stable estimates of the relative tho frequency band used in this study. site responses. Three different wave types were analyzed: PK, LK, and LK coda. Because they consist METHOD of different propagation modes, we analysed data Three-component ground motion data were from the different types separately. The LK coda was digitally recorded at 13 different sites. Various analyzed in five separate time windows (Cl to C5 in combinations of the sites recorded ground motion Figure 1), with results from individual windows from seven different nuclear tests; no site recorded averaged to provide the results discussed here. all of the nuclear tests. A total of 35 3-component A window of the seismic noise was taken data recordings were used in the analysis. The ahead of the Pn arrival to estimate the signal to magnitudes of the nuclear tests ranged from 5.0 to noise ratio in each of the data windows. Only data 5.8. windows with signal to noise ratio greater than 2 Digital data were recorded at a sample rate were used. Data from each window were then of 100 samples/sec, with the anti-alias filter set at bandpass filtered into several octave-width 12.5 Hz. Recordings were first corrected for frequency bands, centered at frequencies of 0.375, instrumental gains, then were windowed according 0.75, 1.5, and 3.0 Hz. to the wavetype of the arrival. The direct arrivals, Figure 1 also shows the spectrum of the Lg Pg and Lg, comprised two data windows, with the arrival, and the noise spectrum for comparison. The remaining five taken from successively later times in spectral plots make the band-limitation of the data the Lg coda. Figure 1 shows an example event clear; usable signal extends from about 0.3 to 3.0 Hz. recording, with the start of the data windows noted. The relatively narrow data bandwidth largely results Data windows were 20.48 sec long (comprising 2048 from the effects of distance from the sources, but the samples). The earliest arrival, Pn, was too small to relatively high cultural noise level in the area and use in the analysis, and is marked on the trace only the highly attenuating near surface volcanics also for reference. contributed. By using multiple data windows that provide Following Phillips and Aki [1], the observed possibly redundant but independent information, we ground motion recordings were conceptually represented by:

r ! 1 kl (1)

•:i| ': - ; •:!-..: where the indices i, j, k, 1 refer to source, site, 100 200 300 400 1 : wavetype (window), and frequenr- band, respectively; A is the observed ground motion; S is the source effect; P is the effect of the path; and R is the receiver (site) effect. The path effect has been written as depending only on the wavetype and s frequency, since the seismic wave paths were nearly < the same for all combinations of sources and sites. I We neglect any possible azimuthal effects in u the source and site terms. The direction of the seismic rays from the sources to all stations was nearly identical. Similarly, the directions of the rays arriving at the stations were nearly identical for all sources. By removing the mean amplitude for a given Figure 1. Vertical component seismogram source and wavetype (data window) from the ground recorded at site PHP (top), and spectra from Lg motion recordings, we can rewrite equation (1) as: arrival and a window of noise from before the Pn arrival (bottom). Vertical bars on the seismogram vijk "avc rijk rave (2) mark the start of 20.48 sec time windows used in the analysis of site response. Usable signal bandwidth is where r = logjo R, and the individual indices for the between about 0.3 and 3 Hz. frequency have been dropped since different bands

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

180 pg will be analyzed individually. The (ik) superscript stands for the mean value holding source and 06 wavetype indices fixed, The source and path terms have been removed from consideration in equation (2). We have assumed the sources to be isotropic, and take the effects of the different sources to be simply differences in recorded amplitudes. Removing the path term is justified because the path effect is in common for all the ground motion recordings. From the formulation of equation (2), we used standard inverse methods to solve for the individual site responses relative to the average for all sites. The variance reduction is 90% for the 1.5 Hz data band, supporting the use of our conceptual 0 375 15 30 model. Frequency (Hz) RESULTS The relative responses for the three wavetypes analyzed have similar shapes when combined in array averages (Figure 2). Relative response values plotted in Figure 2 are array averages for each wavetype and frequency band. The most important features in the figure are the relative response of the different components. The correlation coefficients of individual site measurements of Pg and Lg are 0.83, 0.91, and 0.90 for the Z, N, and E components, respectively. Because of the higher relative excitation of horizontal component ground motion the we report here the resvlts of analysis of the Lg and Lg coda arrivals. The most fundamental result of this study is •0 6 the relatively large variation in response at different 0 375 075 15 30 sites. From the 13 sites occupied, which span a distance of about 10 km, the relative horizontal Frequency (Hz) response varies by about a factor of 5 to 6 at 1.5 Hz (Figure 3). The values plotted in Figure 3 are Log10 of the response in the 1.5 Hz band at each site 06 Lg Coda relative to the average for all sites. The highest relative response at 1.5 Hz (+0.45) is at Site 7, the 04 - lowest (-0.35) is at site PSS; a difference of 0.8 in Log response which corresponds to about a factor = 10 02 of 6 difference in the relative response between the Ampli l 0 iv e Figure 2. (right) Plots of relative response £ averaged from all sites studied. Traces labelled 1 0 2 and 2 are north and east components, respectively; od traces labelled Z are vertical components. Top: •0 4 results from Pg window; middle: for Le window; bottom: for the five Lg coda windows. The vertical 06 component is lower tKan the horizontals for the Lg 0 375 0 75 15 30 and Lg coda windows. Frequency (Hz)

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

18! sites. A similar study of site response at with that at site PSS. Figure 4 shows the response microearthquake monitoring stations in central at the two sites compared to the average of all sites California [2] found site responses that differed by at each frequency band. We see the two extremes of factors of 5 or more at 1.5 Hz, Thus, the range of behavior - increasing response with increasing relative response seen in our study is reasonable. frequency (site 7) and decreasing response with Some sites that are only a few km apart, and increasing frequency (PSS). whose overall surface and near surface geological character are very similar, show a factor of 2 or more DISCUSSION difference in response. Note the differences in All of the sites studied are situated on Figure 3 between response at sites 7 and 8 (+0.45 Bandelier Tuff, which is a 1.1 MY old variably and +0.40 relative response) and the nearby sites 4 (+0.05), 6 (+0.15), and TA-55 (+0.05). Sites 7 and 8 SIT7 0 6 "1 are on mesas separated by intervening canyons. I Sites 4 and 6 are on the same mesa; TA-55 and Site 8 are both on an adjacent mesa, and yet show a large 0 4 > difference in relative response. With standard errors

(1 sigma, in units of Log10) of the relative responses 0 ?.- of less than 0.05, the differences discussed here are significant. A strong topographic effect can be seen in the quite different relative responses of nearby canyon (sites 5 and LAC) and mesa sites (sites 4, TA-55, 6, 7, S -0.2 y and 8). Responses at the canyon sites are as much as a factor of 2 to 3 lower than at nearby mesa sites. In addition to investigating the site -O.i responses at a single frequency, we can also compare -o.f I . _ 1 .. 1 the response as a function of frequency for individual 0.5 1 2 sites. In Figure 4 we compare the response at site 7 FREQUENCE (HZI

Los Alamos 0 6r-- • —

0.4-

§ I

I -0.4 f-

-O.i J J 0.5 1 FREQUENCY [HZ) Figure 3. Map view showing the locations of Figure 4. Relative site response with the sites studied and the relative horizontal response frequency for sites 7 (top) and PSS (bottom). Traces at 1.5 Hz. Relative response is shown as Logio of the labelled 1, 2, and 3 are the vertical, north, and east response at the site compared to the average for all components, respectively. Note the different sites. Note the large range of relative response, from response as a function of frequency, with Site 7 -0.3 at PSS to +0.45 at site 7 (A Log10 range of 0.75 showing increased response, and PSS showing is a factor of 5 to 6). decreased response.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

182 welded ignimbrite deposit. Seismic velocities at or favorable site response seen there is proximity to a near the sites studied range from nbout 3,000 (site possible source zone for local earthquakes, the PHP) to 15,000 (near site PSS) feet per second [31 Pajarito Fault Zone, which passes within a kilometer The thickness of the tuff beneath the mesa top sites of site PSS. The recurrence interval for earthquakes ranges from about 200 feet at sites in the SK (PUP, along the Pajarito Fault Zone is not well known, but AMS) to about 800 feet at sites in the central and may be several thousands of years. The magnitudes western portion of the area (PSS) [41. of the largest earthquakes attributable to the We do not fully understand the causes of the Pajarito Fault Zone are also not well known, but may large differences seen in the relative response of be as large as 6 to 7 131. sites. Several effects may be influencing the site The work reported here is a part of a geologic response. Resonance of the mesas may partly and seismologic study of earthquake hazards in the explain the relatively larger response seen at mesa area of Los Alamos National Laboratory. We plan to sites compared to those in canyons. Differences in calculate response spectra for selected sites from the near surface (upper several meters) velocity recordings of local earthquakes. Because of the alone may not be a major factor, since the near widely variable geologic and seismic structure surface velocities probably do not vary substantially beneath the Laboratory, those response spectra may between nearby sites. Another factor that may be more reliable for use in structural design than influence relative response of different sites may be response spectra calculated from modt

CONCLUSIONS [4] B.J. Dransfield and J.N. Gardner, The factor of 5 to 6 difference in the seismic "Subsurface Geology of the Pajarito Plateau, responses at sites situated within 10 km of each Espanola Basin, New Mexico", Los Alamos other implies that it would be unreliable to use the National Laboratory Report LA-10455-MS, response of a single site to characterize the response 1985. cf all sites throughout the Laboratory. From the basis of site response alone, sites in the western 15| M.L. Lewis, "A Gravity Study of North- portion of the Laboratory would appear to be Central New Mexico", M.S. Thesis, preferable for structures. Counteracting the University of Texas, El Paso, 1980.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~183 Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~184 Session 6 DOE Orders Codes Standards~~

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

~~185 APPLICATION OF PROJECT DESIGN PEER REVIEW TO IMPROVE QUALITY ASSURANCE~~

~~Frank E. McClure Senior Structural Engineer Building 90 G Lawrence Berkeley Laboratory One Cyclotron Road Berkeley, California 94720~~

~~ABSTRACT~~

DOE ORDER 5481.IB "Safety Analysis and Review Systems" and DOE ORDER 6430.1A "General Design Criteria" require that the design of facilities shall incorporate the necessary Quality Assurance review requirements to assure that the established program quality assurance objectives are met in the design criteria and the construction documents. The use of Project Design Peer Review to satisfy these requirements is presented. The University of California manages the Lawrence Berke- ley Laboratory, the Lawrence Livcrmore National Laboratory, and the Los Alamos National Scientific Laboratory. The 1988 University Seismic Safety Policy requires the use of independent Project Design Peer Review in its capital improvement and seismic reconstruction program.

INTRODUCTION the structures and in their immediate The principle of independent review surroundings. [1] of the seismic safety of critical facil- Independent review is the autonomous ities is a well-established element of and objective review of a proposed pro- good practice and a necessary measure for ject by qualified individuals who hold no public health and safety. A critical personal interest or claim in the pro- facility is any structure housing or ject, and who are in no way beholden to serving large numbers of people or other- those proposing or opposing the project. wise posing unusually high hazard to pub- In the State of California, in the field lic health and safety in the event of of seismic safety, the classic cases of damage or malfunction due to an earth- independent reviews in the United States quake. These critical facilities may in- are those conducted by public-sector clude (l) facilities that could pose agencies, i.e., the Office of the Cali- hazard to life and property well beyond fornia State Architect, Structural their immediate surroundings, (2) facil- Safety Section, in reviewing public ities whose continued functioning is school and hospital design and construc- necessary to maintain public health and tion, and the Division Safety of Dans in safety during and following a destruc- reviewing design and construction and tive earthquake, and (3) public or pri- operation of dar.is and reservoirs built by vate structures for housing or assembly local governments, individuals and corpor- of large populations, where failure could ations. [1] pose hazards to life and property within The following sections present the Second DOE Natural Phenomena Hazirds Mitigation Conference - 1989

186 background, vequiromentr., review of Uie .iGlr.mic drrign be made for and recommend at: ions for establishing facilities and buildings where ;\ seir.rnic and implementing a project design peer event can have a potential risk to oper- review process for use by the Department ator lives, to public .'safety, or of lar^e of Energy and its Contractors. economic loss. The review shall be made in two stages - the fi/st at the end of DEPARTMENT OF ENERGY preliminary design and the second before, DOE ORDER 5481.IB, "Safety Analy- final design is complete. For addition- sis and Review System" has broad real guidance on independent reviews, see quirements and DOE ORDER 6430.1A, "Gen- LBL-9143 and UCRL 15910. eral Design Criteria" has more specific DOE ORDER 6430.1A has Quality Assur- requirements for the review of project ance requirements related to the review design and construction documents. of design and construction documents DOE ORDER 5481.IB's purpose is to (drawings and specifications) and struc- establish uniform requirements for the tural design calculations. Specifically, preparation and review of safety analy- Section 0140, QUALITY ASSURANCE, requires ses of DOE operations, including identi- that an adequate QA program provides the fication of hazards, their elimination following assurances: (1) the design or control, assessment of the risk, and will satisfy program and project require- docimented management authorization of ments, (2) the prepared drawings and the operations. This Order requires a construction specifications adequately safety analysis which, in part, requires incorporate QA requirements, (3) construc- the identification and demonstrated con- tion can be performed in accordance with formance with applicable guides, codes, design, and (4) tests confirm the ade- and standards. It also requires evalu- quacy of design and quality of construc- ation and documentation in the facility tion and manufactured components, where safety analysis report of deviations appropriate. This Order also requires from current DOE design criteria. DOE that provisions shall be made for review ORDER 5481.IB requires the documentation and checking design calculations, draw- of all pertinent details of the analysis, ings, and construction specifications by review, and authorization relative to qualified personnel, other than those re- any DOE operation to be traceable from sponsible for the original design. To the initial identification of a hazard the extent practicable, and particularly to its elimination or the application in the case of innovative design, the de- of controls necessary to appropriately sign shall be reviewed by competent con- reduce the risk. sultants in construction or manufacturing techniques to confirm the practicability DOE ORDER 6430.lA's purpose is to of construction or manufacture. provide general design criteria (GDC) for use in the design of DOE facilities. LBL REPORT-1943, "SEISMIC SAFETY GUIDE" It requires that facility design shall DOE 643C.1A recommends application incorporate the necessary Quality Assur- of the recommendations in LBL Report 9143, ance requirements to assure that the es- "Seismic Safety Guide," September 1983,for tablished program and project quality practical guidelines for earthquake engi- assurance objectives are satisfied. Im- neering safety and management planning portant to satisfying these objectives and technical procedures for the design is the assurance that the project con- of new facilities and evaluations of ex- struction documents (drawings and spec- isting facilities. LBL Report 9143 out- ifications) conform to the project de- lines procedures for cost-effective plan- sign codes, standards and other project checks or "third-party" reviews of struc- requirements. For buildings and other tural/seismic designs and evaluations. structures designed to resist earth- Independent structural and seismic quake forces, DOE 6430.1A, Section design reviews should be made by an inde- 0111-2.7.1 requires an independent pendent consultant. These reviews should Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

187 be m tie at two stages for major facil- engineering drawings and calculations ities, tacilities having a potential and other quality assurance measures are risk to lift1 safety due to the process prescribed. For Important or Hazardous contained therein, or facilities with a facilities, all aspects of the design or potentially large economic loss. The evaluation must be included in an inde- first r;view should be made at the end pendent peer review. of the preliminary design or Title I Specifically, Section 3.4 of UCRL services, and the second, separate re- 15910, "Quality Assurance and Peer Re- view when the final design is completed, view" states "To achieve well designed but before bids are taken. and constructed facilities resistant to The structural/seismic design re- natural phenomena hazards or to assess views should include design, philosophy, whether existing facilities are well de- criteria used, franing system, construc- signed and constructed for natural phe- tion materials, and other factors per- nomena hazard effects, it is recommended tinent to the seismic capability of the that important hazardous (Categories II, proposed facility, Particularly impor- III, and IV) or unusual facilities be de- tant in the review is the check for a signed or evaluated utilizing an engi- continuous load path, or paths, and for neering quality assurance plan. Specific the adequacy of their strength and details about the engineering quality stiffness to transfer seismic forces assurance plans depend on the natural frotn point of application to final phenomena hazard considered. As a re- point of resistance. In sun, peer result, such plans are described in some view by an independent consultant or detail in each of the remaining chapters peer group need not provide a detailed of this document." UCRL 15910 also check of the spacing of reinforcing states, "In general, an engineering bars, but rather an overview to help quality assurance plan should include identify oversights, errors, conceptual the following requirements. On the de- deficiencies and other elements likely sign drawings or evaluation calculations, to cause problems during and after con- the engineer of record must describe the struction. Peer reviews can catch hazard design basis including: (1) des- costly design mistakes in judgment, cal- cription of the system resisting hazard culations, or philosophy. For a major effects, and (2) definition of the hazard facility, an independent peer review loading used for the design or evaluation. could more than pay for itself by un- Design or evaluation calculations should covering design deficiencies before they be checked for numerical accuracy and for are cast in concrete or constructed in theory and assumptions. (The author has steel. recommended the following wording be The procedures outlined in the added in text of Section 3.4, "Construc- "Seismic Safety Guide" have been imple- tion drawings and specifications should mented at the Lawrence Berkeley Labora- be peer reviewed to verify that these de- tory since the beginning of its seismic sign assumptions are implemented in the safety program in 1972. These proce- construction documents.") For new con- dures are recommended for consideration struction, the engineer of record should and use by DOE and its Contractors. specify a material testing and construction inspection program. In addition, UCRL DRAFT REPORT 15910 the engineer of record should review all UCRL Draft Report 15910, April testing and inspection reports as well as 1988, "Design and Evaluation Guidelines periodically make site visits to observe for the Department of Energy Facilities compliance with plans and specifications. Subject to Natural Phenomena Hazards," For important or hazardous facilities, requires more specific Quality Assurance all aspects of the design or evaluation procedures than DOE Order 6430.1A. For must include independent peer review. General Use, Low, Moderate and High For various reasons, a designer may not Hazard facilities, a peer review of be able to devote as much attention to

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

188 natural phenomena hazard design as he Director shall provide for the selection night like. Therefore, it is required of the reviewer and for the organization, to have the design reviewed by a quali- plan, and type of review, subject to the fied, independent consultant or group ... following: (1) a licensed structural for major hazardous facilities, it nay be engineer with demonstrated experience in prudent to have concurrent independent seismic design shall perform the review evaluations performed or to have the and prepare a written report, (2) the evaluation independently reviewed." reviewer shall be contracted for and paid by the University and not by the UNIVERSITY OF CALIFORNIA architect or engineer of record for the The University of California nanag.es design, and (3) the reviewer shall not be and operates the Lawrence Berkeley Labora- an employee of the University. Similar tory, the Lawrence Livermore National Lab- requirements pertain for an independent oratory, and the Los Alamos National Sci- review of the structural and seismic entific Laboratory. Following the occur- design of facilities being considered for rence of earthquake damage at the Uni- lease or purchase for University purposes. versity of California at Los Angeles, The approximate cost of the above caused by the 1971 San Fernando earth- independent seismic review is the same as quake, the University adopted a Univer- the Plan Review Fee which is 65 per cent sity Seismic Safety Policy in early 1975, of the Building Permit Fee in Table No. [2] & [3]. This Policy was reaffirmed 3-A, 1988 Uniform Building Code [A], At in a letter from President David Pierpont the Lawrence Berkeley Laboratory, the use Gardner, dated May 17, 1988, to the of the independent structural and seismic Chancellors and the Directors of the review process has proven to be cost- aforementioned laboratories. This effective with improved structural and Policy combined with a letter from Pres- seismic designs as well as improved con- ident Gardner, dated September 30, 1986, struction documents (drawings and spec- to the same persons addressed the "Policy ifications) with fewer design change for Independent Seismic Review of Struc- orders during construction. tures." The "Policy for Independent Seismic AMERICAN SOCIETY OF CIVIL ENGINEERS Review of Structures" requires that an The American Society of Civil Engi- independent review shall be conducted of neers has undertaken a very ambitious and the structural seismic design of all significant program to improve the qual- capital projects, whether new construc- ity in constructed projects. A major tion or remodeling, which involve struc- milestone has been reached with the pub- tural design and are intended for human lication of the Preliminary Edition for occupancy or which affect human safety. Trial Use and Comment of the Manual, The review shall be initiated early in "Quality in Constructed Project, a Guide- the project life and preferably during line for Owners, Designers, and Construc- the preparation of schematic designs so tors" [5]. According to Chapter 13, that it can be performed in conjunction "Peer Review" of the Manual, the follow- with the independent design and cost re- ing issues are very important to under- view and value engineering processes standing and implementing a "Peer Review" where applicable. Also, the review shall process. "Peer review is a technique be continued at appropriate times during that promotes quality in design organi- the design process. In all cases, work- zation and their services. It is the ing drawings and calculations shall be highest level of action to improve qual- reviev/ed for conformance of the new work ity in design of constructed projects. to the most current applicable seismic A project design peer review is a com- design code requirements prior to let- prehensive examination of the technical ting bids for such work or authorizing aspects of the project design as they structural change orders. The Chancellor relate to concept, progress or final re- or equivalent responsible Officer or sults. A peer review is conducted by

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

189 peers of the original manager, authors, owner, and the design professional. All or design professionals. A peer review of these groups who are familiar with is a special effort, not a routine pro- peer review have encouraged their use by cedure performed daily on typical pro- large or small organizations and on large jects or design processes. A peer re- or small projects. A fresh, unbiased and view has a specific purpose, scope, diplomatic review by an independent, format, and duration. A peer review is high-level professional can be a highly paid for by the commissioning authority, cost-effective measure." [5] who benefits from this valuable service. The aforementioned Manual has exten- The original design professional retains sive guidelines for project design peer all authority and responsibility for the review and is recommended to those who design and is the undisputed design have a need for or can appreciate the engineer or record. Attenpts to assign value of the project design peer review undue responsibility to a peer review process. team ultimately will result in less use of the peer review process. INDEPENDENCE, COMPETENCE AND INTEGRITY "A peer review is not simply any Report SSC-81-01, "Independent review of a document by anyone other than Review of Critical Facilities: Uith its author, even if the reviewer is at or Special Emphasis on State-Federal Re- above the author's "peer" level. A peer lationships and Dam Safety, January 1981, review is not a review by a building code [1] discusses in a very frank and official or by any other government straight-forward manner the "human prob- agency as it carries out its regularly lems" with the selection of independent mandated responsibilities covering some peer reviewers and how the peer review area of the design and construction. process can work if there is a "common Peer review is not a value engineering basis of understanding" among all those study, despite certain similarities in involved in the process. "A degree of the two processes. As a basic difference, independence is possible, even working value engineering of a design focuses on within the same organization, if a potential cost savings, while peer re- special office or staff is established view of a design examines the quality of whose principal or sole purpose is safety the design, including, in many cases, review, and safety criteria are used in the procedures used and the management of guiding their judgment (internal indepen- the process. In other words, a peer re- dent review). But full independence viewer is likely to ash first, "Is it (later called external independent review good enough?," whereas the usual value or "third-party" review) is insured only engineering question is, "Does it cost when those who are responsible for a re- too much?" view are not organizationally connected "A peer review may be voluntarily with or otherwise beholden to (1) those requested or authorized, or it may be who did the original design and computa- mandated by some authority other than the tions, or (2) the entrepreneurs of the persons or agency to be reviewed. Any project for which the design has been peer review must have adequate budgets made " [1] of time, effort, and noney. The steps of John P. finaedinger refers to the a project peer review normally consist of independent review process as "peer re- acquiring information, examining the in- view": "The principal process .... con- formation and evaluating its relevance, templated here is one whereby project thoroughness and accuracy, drawing con- plans and specifications are reviewed by clusions about the status or technical one or more independent peers with quality (or both) of the project from recognized competence in the technical these evaluations, and presenting and areas involved ...." [6] !!e empha- discussing the report. sized the torn "peer" review rather than "Peer reviews are requested, as just independent review, because "it is added safeguards for the public, the crucial that the competence and

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~190 experience be present." |~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~191 Equivalent Static Seismic Analysis Approach for Process Equipment in Moderate ind High Hazard Facilities~~

~~C. R. Hammond Martin Marietta Energy Systems, Inc. Oak Ridge National Laboratory Oak Ridge, Tennessee~~

Introduction keeping the equipment in place with secure anchorage and preventing interference, that is This paper is intended to help designers hammering pieces together. (or other seismic non-specialists) decide on the optimum course through the various paths The strategy in this report is to use the offered by our latest General Design Criteria simplest static methods for the cases with the and to provide justification for simplified least conservatism. Then, in cases with methods for design of moderate or high increasing conservatism, consideration of hazard equipment. Information is also dynamic properties also must increase. provided on writing seismic load specifications for equipment. The static equivalent seismic analysis method The General Design Criteria1 requires verification of seismic adequacy for special The general formula for static equivalent facilities"...by a dynamic analysis except seismic loads is: where it can be demonstrated that the use of a simplified approach, such as a static load method... provides assurance of adequate seismic design." Here, dynamic analysis where: means considering the dynamic properties of 2 each piece of equipment . Static load Fe is the horizontal seismic load methods, which are used to qualify C is a coefficient of static equipment in general use and low hazard equivalence facilities, are less costly mainly because they Apeak is the ratio of maximum seismic can be applied uniformly to an entire floor of acceleration to the acceleration of equipment, say. Only the weight and shape gravity of each item is needed. The requirements on is the weight of the equipment. equivalent static methods for moderate and w high hazard facilities are more conservative. This seismic force is applied at the center of mass of the equipment. A general Fortunately, equipment is usually robust. consensus among published static Experience with earthquake damage indicates coefficients is shown in Table 1 3 4 5 6. The that much equipment; including piping, peak acceleration ratio is obtained from a HVAC, cable trays, etc.; even when not response spectrum as shown on Fig. 1. The designed to resist seismic loads, survive spectrum is a smoothed plot of the maximum severe earthquakes. Although earthquakes response of single degree-of-freedom can produce damage by subtle effects, such oscillators of varying natural frequency to a as accidental torsion, most codes emphasize collection of earthquake time histories.

* The submitted manuscript has been authored by a contractor of the U.S. Government under contract No. DE-AC05- 84OR21400. Accordingly, the U.S. Government retains a nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for U.S. Government purposes.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

192 Figure 1 shows the median of peak responses coefficient of one. A complex piping system for three sets of oscillators, each with a may contain interacting modes but for a different damping value, subjected to straight pipe supported at two points, the earthquakes characteristic of the Oak Ridge NRC Standard Review Plan specifically area. The top curve is for 2% of critical allows a coefficient of one 4. damping, the middle is 5% and the bottom is 10%. UCRL-159107 recommends using 5% Torsional vibration of critical damping to represent equipment. This response spectrum is normalized to lg Torsional modes of vibration are acceleration for an infinite frequency produced when the center of mass at any oscillator and the acceleration values must be level of a structure is offset from the center of multiplied by the ground acceleration stiffness at the same level. Considerable depending on location and hazard level. For torsional damage has been observed in example in Oak Ridge the design basis buildings with a non-symmetric floor plan. earthquake magnitude for moderate hazard Current seismic building codes do not allow facilities is 0.19g and for high hazard simplified static methods for structures with facilities it is O.32g. certain plan irregularities in higher seismic zones and even for symmetric structures they Although the spectrum gives the require consideration of an "accidental" maximum response of an oscillator with only torsional moment9. A designer should be one mode of vibration, a complex structure wary of torsion for equipment with eccentric such as a building or a pipe system can mass or a base with an irregular plan. A achieve higher accelerations when two or computer analysis may be the easiest way to more modes of vibration combine. The deal with obvious cases of torsional modes. coefficient 1.5 on the first line in Table 1 is However, applying an accidental torsion an upper bound on the effect of combining moment is not necessary with a typical modes. anchorage design margin of four.

Most equipment that stands as a single Equipment mounted in flexible unit avoids the most severe effect of mode structures combination and qualifies for a static equivalence coefficient of one. In a multiple The static coefficient of one applies when degree-of-freedom system, only a fraction of the seismic response spectrum at the the total mass is effective in each modal anchorage point of the equipment is available. response. The commentary to the SEAOC When the equipment itself is flexible and is "Blue Book" shows that if the modes are mounted up in a building or other structure, independent, that is the frequencies are the simplified methods require too much widely separated, the summation of any strength to be economic. Fig. 1 is the structural response quantity is the root-sum- spectrum for single degree-of-freedom of-the-square (rss) but if the frequencies are systems mounted on ground. It would apply close together, the sum is greater than rss8. to a piece of equipment mounted on the (The interaction between frequencies depends ground, on or below the ground floor of a on damping but interaction is trivial for building, or on a rigid supporting structure fn/fn+i < 0.67.) Appendix D of Ref. 4 which is on the ground (such as a concrete considers four typical dynamic models of a shielding cell). Fig. 3 shows a response single unit of equipment. The models are spectrum at an equipment attachment point in shown in Fig. 2. Only the second model has a flexible structure. Ground acceleration is multiple frequencies, but they are all widely 0.05g and the peak amplification of ground spaced, which is typical of a simply system acceleration is more than 14 times. The with harmonics. Rigorous solutions for base bound on all in-structure spectra such as this shear and overturning moment of each of the is about 40 times ground acceleration structural models supports a static equivalent according to Duff 5 as shown in Table 1, For

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

193 a moderate hazard facility in Oak Ridge that and are not accepted by the NRC or by DOE would mean an equivalent static acceleration Nuclear Standard F9-2T. of nearly eight times the acceleration of gravity. The static coefficient of 40 is a stiff Rigorous in-structure spectra are obtained penalty to pay for simplified analysis. using time histories of earthquakes as input to However, with a change in design finite element analyses of the structure philosophy and a little more analysis, the supporting the equipment. The equipment penalty can be avoided. designer should make sure that any specification for analysis of a building Notice on Fig. 1 that oscillators with include the needs for generating in-structure fundamental vibration periods less than spectra. 0.03 s. do not amplify the earthquake motion. Structures with vibration periods less Summary of equipment seismic than 0.03 s. are considered rigid. Rigid analysis equipment mounted on the ground obviously responds only with the ground acceleration, If the equipment does not have obviously un-amplified. This is also true of rigid eccentric mass and has a regular base plan equipment mounted in structures as shown then calculate the seismic force using the on Fig. 3. If we design our equipment to be weight of the equipment and: rigid, th *n the anchorage requirements are no worse than for flexible equipment mounted 1) for equipment mounted on the on ground. Some equipment, such as electric ground, on or below the ground floor motors or small pumps, are rigid by of a structure, or on a rigid structure inspection and qualify for the simplified on the ground or if the equipment is method of designing anchorage. The rigid by inspection; use a static fundamental period of vibration of equipment equivalent seismic coefficient of one that is not obviously rigid must be determined and the peak acceleration from the and shown to be less than 0.03 s by analysis ground response spectrum for the site or test. and hazard level; If the equipment cannot be made 2) for equipment mounted in a flexible completely rigid, it is necessary to obtain an structure; then design the structure to in-structure response spectrum for the point be rigid and use the coefficient and of attachment of the equipment to avoid an acceleration as above; acceleration coefficient of 40. Generally, equipment should be designed so that the 3) obtain an in-structure response fundamental period of the equipment is well spectrum at the point of attachment below the highly amplified part of the and design the equipment to have a spectrum. In this way, the higher frequency fundamental period less than the harmonics will not be amplified and need not highly amplified region of the be checked. spectrum; then use a coefficient of one and the acceleration on the in- If the spectrum is not available The structure spectrum corresponding to Departments of the Army, Navy and Air the period of the equipment. Force technical manual Seismic Design Guidelines for Essential Buildings 2 provides Specifications for equipment a simplified way to obtain an in-structure response spectrum from the dynamic Currently many manufacturers properties of the supporting structure. Only understand requirements for dynamic manual and graphic methods are used. qualification of equipment. They are familiar Although these methods are accepted by our with general specifications such as IEEE Std General Design Criteria they are not rigorous 344, "IEEE Recommended Practices for

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~194 y it V)~~

~~0 -J >~~

~~.04 .06 08 . 4 6 8 10 20~~

~~PERIOD (sec) Figure 1. Design Response Spectrum scaled to 1.0 g (2%, 5%, and 10% of Critical Damping) Oak Ridge National Laboratory, X-10, K-25, and Y-12 Sites, Tennessee~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~195 Rigid Model~~

~~/ Second Mode / First Mode~~

~~Cantilever- Model~~

~~Flexible Base Support~~

~~Parabolic Model Figure 2. Dynamic models of typical equipment~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~1% TABU- I. STATIC EQUIVALENT SEISMIC FORCE COEFFICIENTS~~

~~CQF.FFICIRNT~~

DOE1 NRC- AECL' EPR1*

PEAK OF FLOOR RESPONSE 1.5 1.5 1.5 1.0*

~~PEAK OF GROUND SPECTRUM 12.. 10. AT FLOOR~~

~~GROUND ACCELERATION 40.~~

~~'DOE Nuclear Standard NE F9-2T.~~

~~:NRC Standard Review Plan, NUREG-75/087.~~

~~5C. G. Duff. "Simplified Seismic Analysis Methods Used by AECL for the Seismic Qualification of CANDU Nuclear Power PL.ints."~~

~~4"Seismic Verification of Nuclear Plant Equipment Anchorage," EPRI NP-5228, Vol. 1.~~

*For widely spaced natural frequencies.

~~0 0 .01 .1 1 Period (s) Figure 3. In-structure Response Spectrum~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

197 Seismic Qualification of Class IE Equipment 2) Seismic Design Guidelines for Essential for Nuc'^ar Power Generating Stations,", Buildings (A Supplement to "Seismic The designer must define the design basis Design for Buildings"), Army TM5-8O9- earthquake in dynamics terms. If we 10.1, Navy NAVFAC P-355.1, Air consider only complete, assembled Force AFM 88-3, Chapter 13.1, equipment then the seismic input is bounded Departments of the Army, Navy and Air by two response spec ja which we already Force, Washington, D. C., February have. First, if the equipment is to be 1986. mounted on the ground, on or below the ground floor of a structure or on a rigid 3) Seismic Requirements for Design of structure on ground then the input to the Nuclear Power Plants and Nuclear Test equipment is a uniform spectrum of the Facilities. Nuclear Standard NE F9-2T, ground acceleration at all frequencies. U. S. Department of Energy, Oak Ridge, Second, if the equipment is to be mounted in TN, February 1985. a flexible structure which has no closely spaced modes of vibration, then an upper 4) Standard Review Plan. NUREG-75/087. bound on the input to the equipment is the U. S. Nuclear Regulatory Commission, ground response spectrum for the correct site Washington, D. C, 1975, Section 3.7.2. with the magnitude of the spectrum at zero period equal to the ground acceleration for the 5) Duff, C. G., "Simplified Seismic correct hazard level. If the supporting Analysis Methods used by AECL for the structure contains closely spaced modes then Seismic Qualification of CANDU Nuclear all spectrum points should be multiplied by Power Plants," Second Symposium on 1.5 to bound the input motion. These ground Current Issues Relating to Nuclear Power response spectra are upper bounds since they Plant Structures. Equipment and Piping. do not consider the dynamic characteristics of Electric Power Research Institute, the building or of a reduced response of parts Seismicity Owners' Group and North of the building. The particular structure or Carolina State University, Orlando, FL, the particular mounting location in the December 1988. structure can be considered as an additional refinement. 6) Seismic Verification of Nuclear Plant Equipment Anchorage. Volume 1: The proper damping ratio should also be Development of Anchorage Guidelines, considered in the equipment specification. EPRI NP-5228, Volume 1, Electric The damping ratios recommended for use in Power Research Institute, Palo Alto, CA, structural analysis reflect the high damping May 1987. that occurs near the elastic limit in the structures being checked. If a structure 7) R. P. Kennedy, et al, "Design and containing a piece of equipment has a Evaluation Guidelines for Department of significant margin of capacity over demand, Energy Facilities Subjected to Natural the damping will be less than the design Phenomena Hazards," Draft UCRL- value. In an earthquake the equipment will 15910, Lawrence Livermore National be subjected to larger earthquake Laboratory, Livermore, CA, May 1989. accelerations than designed for. For this reason, a response spectrum reflecting low 8) Recommended Lateral Force damping (2% or less) should be specified. Requirements and Tentative Commentary. Seismology Committee, References Structural Engineers Association of California, San Francisco, CA, 1988. 1) General Design Criteria. DOE Order no. 6530.1 A, United States Department of 9) Uniform Building Code. 1988 Edition. Energy, Washington, DC, 6 April 1989. International Conference of Building Officials, Whittier, CA, 1988.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~198 SEISMIC DESIGN CRITERIA AT THE IDAHO NATIONAL ENGINEERING LABORATORY (INEL)~~

~~B.G. Harris Idaho National Engineering Laboratory EG&G Idaho, Inc. P.O. Box 1625, MS 4144 Idaho Falls, ID 83440~~

~~ABSTRACT~~

During the 40-year existence of the INEL site, the criteria used for the. seismic design of facilities have evolved based on the increased understanding of earthquakes, their effect on structures, and the specific seismic characteristics of the INEL. Several studies, including one scheduled for completion this year, have been initiated since 1970 for the purpose of utilizing evolving technology to better define reasonable, yet conservative, seismic design criteria.

INEL DESCRIPTION • research in materials science, chemistry, biotechnology, physical sciences, and The National Reactor Testing Station (NRTS) was environmental sciences established by the Atomic Energy Commission (AEC) in 1949 as a site for conducting reactor and nuclear energy Currently, the various INEL facilities are operated by five studies. In 1974, the NRTS was redesignated the Idaho major contractors - Argonne National Laboratory-West, National Engineering Laboratory (INEL) to better reflect the EG&G Idaho, Inc., Rockwell-INEL, Westinghouse Electric wide scope of activities, both nuclear and non-nuclear, being Company, and Westinghouse Idaho Nuclear Company conducted there. (WINCO). The INEL is located in the southeastern portion of Idaho, roughly equidistant from Yellowstone National Park, Salt GEOLOGICAL SETTING Lake City, Utah, and Boise, Idaho. It is a government-owned reservation encompassing approximately The INEL is located on the Eastern Snake River Plain 890 square miles and is managed by the U.S. Department of (ESRP). This plain is formed largely of volcanic rocks Energy (DOE). Support facilities are also located in Idaho interbedded with alluvial, lacustrine, and eolian Falls, approximately 29 miles to the east. See Figure 1. sediments.[l] The volcanic rocks consist of extrusive The INEL is a multipurpose site that provides applied basalts, some of which emanated from the numerous extinct research and development in support of DOE programs. volcanic craters and cones which can be found throughout Major activities at the INEL include: the plain. The sediments in the area of the INEL were laid down primarily by the Big and Little Lost Rivers and Birch • reactor technology Creek.[2] The INEL is bordered on the northwest by the Lost River, Lemhi, and Bitterroot mountain ranges. • materials testing SEISMIC ACTIVITY • recovery of uranium and other elements from spent fuels The ESRP is notably aseismic. Only five earthquakes have been centered within the ESRP since 1971, and none have exceeded a Richter magnitude of 1.0.[3] The • radioactive waste management Intermountain Seismic Belt on the cast and the Idaho Seismic Zone on the north are the two major areas of seismic activity • training of nuclear navy personnel near the ESRP. [4] Based on their proximity to (he INEL and

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~199 Bitterroot Range~~

~~INEL Boundary~~

~~Big Southern Butte~~

~~Figure 1. Idaho National Engineering Laboratory 90767~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

200 the likelihood of generating sizable earthquakes, the faults progress. In addition, the New Waste Calcining Facility considered lo be of most significance are the range front (NWCF), which is a nonreactor nuclear facility located at the faults located along the western flanks of the Lost River, Idaho Chemical Processing Plant (ICPP), was in the design Lemhi, and Beaverhead (Biuerroot) ranges.[3] phase. Consequently, there was a need to evaluate the The largest historical earthquake event in the vicinity of seismic design criteria being used for each of these facilities. the INEL occurred on October 28,1983, having a Richter A special study was conducted in 1971 for the purpose cf magnitude of 7.2 and a surface wave magnitude of 7.3. The developing conservative upper limits on seismic forces that epicenter for this earthquake was located along the western could be expected at the NRTS, and to outline a course of flank of Borah Peak in the Lost River Range, approximately action that would ultimately lead to selection of the most 50 miles northwest of the INEL. Although it was felt at the realistic design values.! 10] This study resulted in the INEL, no significant structural or safety-related damage following preliminary estimates of horizontal bedrock occurred to INEL facilities.[5] acceleration ranges at NRTS facilities: Another major earthquake occurred in 1959 near Hebgen Lake, Montana, about 100 miles northeast of the INEL. Facility Acceleration Range Initially, it was calculated to have a Richter magnitude of 7.1. However, recent re-analysis of the data from that earthquake LOFT and TAN .28-.45g has resulted in a determination that its Richter magnituue ICPP, TRA*& PBF .15 - .33g was 7.7[6] and its surface wave magnitude was 7.5[7], Test Reactor Area HISTORY OF SEISMIC DESIGN CRITERIA AT THE INEL Based on this information, and other available information, the seismic design criteria for the LOFT, PBF and NWCF projects were finalized. Horizontal bedrock Prior to 1970 accelerations of 0.34g and 0.22g were used for design of Prior to the 1960's, it was common practice to design LOFT and PBF, respectively.! 11,12] NWCF was designed structures to meet the seismic design requirements contained using a horizontal bedrock acceleration of 0.33g.[13] The in the Uniform Building Code (UBC).[8j This applied not vertical accelerations were taken to be two-thirds of the only to ordinary buildings but also to structures containing horizontal values. Housner's response spectra[14] was used nuclear material. Consequently, the first facilities for LOFT and PBF. NWCF was designed using the response constructed at the NRTS were designed using the UBC spectra contained in Regulatory Guide 1.6O[15] issued by the requirements. It should be noted that any of the reactor U. S. Nuclear Regulatory Commission in late 1973. facilities constructed during this period, which are still Two additional studies of the seismicity of the INEL were operating, have been upgraded as required. performed in 1975 and 1977.[16,4] Some of the Prior to the 1970 edition, the UBC placed the NRTS in recommendations which came from these studies were: seismic Zone 2. The zone boundaries wen revised in the 1970 UBC and the NRTS was placed in seismic Zone 3. • The NRC Regulatory Guide 1.60 response This essentially doubled the magnitude of the static spectra should be used. horizontal force used in seismic design. For most low-rise structures, the UBC seismic design method uses a static lateral force procedure which only • A probabilistic approach to determining design considers forces in the horizontal direction. accelerations is preferred.

1970 to 1979 • More geologic data was needed to determine less In the 1960's and early 1970's, extensive work was done conservative, but more realistic acceleration on a national basis to develop specific seismic design criteria values. for commercial nuclear power plants. This resulted in the issue of government regulations applicable to seismic design of such facilities. In June 1978, in order to maintain consistency with In late 1970, the Atomic Energy Commission directed the neighboring facilities (i.e., TRA), the Idaho Operations Idaho field office to cease using the UBC static analysis Office of the Department of Energy (DOE-ID) directed that method for design of the Loss of Fluid Test Facility (LOFT), future designs at ICPP were lo use a horizontal bedrock which was a test reactor located at the Test Area North acceleration of .24g. [ 11] (TAN), and to establish site-specific criteria for doing a In December 1978, the original issue of the INEL dynamic seismic analysis.[9] Architectural Engineering (A-E) Standards[18] was issued During this time period, design of another reactor facility and contained the following information on seismic design at the NRTS, the Power Burst Facility (PBF), was also in criteria:

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

201 "Unless otherwise directed, buildings and other structures Area SSE» QBE** shall be designed for a Zone 3 seismic intensity in accordance wit. the latest edition of the Uniform Building LOFT 0.35 g 0.175 g Code. Earthquake loads shall not be applied in TRA 0.24 g 0.12 g combination with wind lr^ds. More severe earthquake ICPP 0.24 g 0.12 g requirements may be imposed in criteria fo; specific PBF 0.22 g 0.11 g buildings and structures. The ID Natural Phenomena Committee is prepa\ng a seismic design criteria manual 41SSE - Safe Shutdown Earthquake for the INEL which will establish acceleration, frequency ** OBE - Operating Basis Earthquake response, etc. to be used in the design of special nuclear facilities. When approved, the report will be incorporated Vertical seismic acceleration shall be taken as two-thirds into the INEL A-E Standards in an appropriate manner." of the horizontal acceleration. Frequency spectra shall be taken from the USNRC Regulatory Guide 1.60, "Design Response Spectra for Seismic Design of Nuclear Power I98Q to Present Plants." In 1981, as a result of further data gathered by the U.S. Geological Survey (USGS), DOE-ID submitted a request to The UBC method for design of non- nuclear facilities is still the International Conference of Building Officials (ICBO) specified. The 1988 edition of the UBC placed the INEL in that the next edition of the UBC be revised such that the seismic Zone 2B. INEL would be in seismic Zone 2 instead of Zone 3. This request was reviewed and approved at the ICBO annual Impacts of DOE Order 6430.1 A business meeting in 1981.(19] This change only effected the The recent issuance of DOE Order 6430.1 A[22] has design of non-nuclear facilities. The next revision of the resulted in yet another review of the seismic design criteria INEL A-E Standards[20] reflected this change. used at the INEL. This order requires that the basic seismic A significant event relating to seismic design at the INEL parameters be derived from UCRL-53582[23] and that the occurred on October 28,1983. At 8:06 a.m., an earthquake design be done in accordance with the procedures contained of Richter magnitude 7.2 occurred about 50 miles northwest in the UBC and UCRL-15910[24]. UCRL-53582 contains of the INEL. This was the largest earthquake to occur in the site-specific seismic hazard curves and response spectra for contiguous United States since the earthquake at Hebgen twenty-six different DOE sites, including the INEL. The Lake, Montana in 1959. Two deaths and property damage in information for the INEL is based on a seismic study and excess of S2.5 million were attributed to the Borah Peak probabilistic risk assessment performed by the TERA Earthquake. [5] Most of this damage was done to older, Corporation in 1984.[25] The basic areas where these non-INEL structures which did not meet UBC standards. criteria differ from those currently in the DOE-ID A-E The structures designed to UBC standards survived with Standards are: minor or no damage. Based on detailed inspections of facilities, and • UCRL-53582 gives only one acceleration curve comparisons of measured accelerations with design values, it for the entire INEL site. For high hazard was concluded that the Borah Peak Earthquake created no facilities, the horizontal ground acceleration safety problems for INEL reactors or other facilities. Only value is 0.21g. The DOE-ID A-E Standards minor non-structural damage to some buildings in the form specify facility-s]>ecific horizontal bedrock, of hairline cracks or small settlements were found. Observed accelerations which range from 0.22 to 0.35g, earthquake motion was consistent with the earthquake zoning depending on the location of the facility with and design criteria in effect at the time.[5] The data from respect to geologic faults. this earthquake has been valuable in confirming attenuation curves, checking model and response accuracy, and • UCRL-53582 recommends an INEL response providing more data for design of future facilities. spectra based on the median Newmark-Hall The Borah Peak Earthquake has not, to date, resulted in response spectra. The DOE-ID A-E Standards any change to the seismic design criteria used at the INEL. require that the Regulatory Guide 1.60 spectra be The most recent revision of the INEL A-E Standards, now used. known as the DOE-ID Architectural Engineering Standards[21], does, however, give specific acceleration Initial review of the DOE M 30.1 A criteria indicated that values for design of special nuclear facilities. It states: they may be less representative of local conditions and "The dynamic analysis method involves use of seismic potentially less conservative than the criteria in the current parameters which have been developed for various edition of the DOE-ID A-E Siandards. locations at the INEL. Maximum horizontal bedrock Recently DOE-ID made ihe decision to utilize a acceleration shall be as follows: consultant to perform another seismic study of the INEL and Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

202 to develop site-specific seismic design parameters fur ACKNOWLEDGMENTS several INEL locations, including the proposed sites for the Special Isotope Separation and New Production Reactor This work was performed under the auspices of the U.S. Projects. Woodward-Clyde Consultants was retained to Department of Energy, DOE Contract No. DE-AC07- perform this work. The specified purposcs(26] of this study 761DO1570. The assistance of V. W. Gorman of EG&G are: Idaho, Inc. is also gratefully acknowledged.

Provide site-specific estimates of peak REFERENCES horizontal ground acceleration, response spectral ordinaies and time histories for selected sites [I] EG&G Idaho, Inc., INEL Environmental located on soil or bedrock at the INEL. Characterization Report. EGG-NPR-6698, Idaho Falls, Idaho, 1984, Revised January 1985, pp. 5-12. Develop a peak acceleration- attenuation [2] Northern Engineering and Testing, Inc., Report of relationship for ground motions from Geotechnical Investigation. SIS Geotechnical earthquakes in the magnitude range of 6 to 8 that Evaluation, April 1987, p. 5. is specific to the INEL. [3] U.S. Department of Energy. Final Environmental Provide technical support and portable digital Impact Statement: Special Isotope Separation Project. seismographs for a site response and crustal Washington, D.C., 1988, pp. 3-15 to 3-19. attenuation survey of selected sites in and around the INEL. [4] Agbabian Associates, Evaluation of Seismic Criteria Used in Design of INEL Facilities. El Sugundo, • Process and analyze the data recorded by the California, September 1977. survey to evaluate local site response, seismic attenuation, and source parameters of selected [5] V.W. Gorman and R.C. Gucnzler, The 1983 Borah earthquakes. Peak Earthquake and INEL Structural Performance. EGG-EA-6501, Idaho Falls, Idaho, December 1983. This study was scheduled for completion in August of this [6] B.A. Bolt, 'The Magnitudes of the Hebgen Lake 1959 year. It is anticipated that the results will provide additional and Idaho 1983 Earthquakes", Abstract. Earthquake valuable information for refining the seismic design criteria Notes, vol. 55, no. 1, p. 13,1984. used at the INEL. It is being performed using current, accepted technology and strict quality assurance procedures to ensure the best possible results. [7] D.I. Doser, "Source Parameters and Faulting Processes of the 1959 Hebgen Lake, Montana, Earthquake SUMMARY Sequence", Journal of Geophysical Research, vol. 90, no. B6, pp. 4537^555, May 10, 1985. The geology and seismology of the INEL have been extensively studied and the criteria used for seismic design of [8] International Conference of Building Officials, INEL facilities have evolved based on the results of these Uniform Building Code. studies, and also on the development and revisions of national standards, codes and regulations. [9a] W.H. Layman (USAEC) letter to Manager Idaho Existing facilities at the INEL have been constructed Operations Office, "LOFT Seismic Analyses", using the seismic requirements in effect at the time of their November 27,1970. design. In addition, many facilities have been upgraded to meet current requirements. [9b] R.E. Swanson (AEC-IDO) letter lo H.L. Coplen Based on currently available information, ihe present (INC), "LOFT Seismic Analyses", December 10,1970. INEL criteria are felt to be conservative. Should the current study or future studies indicate that any of these existing [ 10] Woodward-Lundgren & Associates, Preliminary criteria are not representative of actual conditions, the Seismic Risk Evaluation for liie National Reactor criteria will be revised as necessary. It is important that Testing Station. Oakland, California, June 1971. reasonable, defcndable criteria be available for both the design of new facilities such as the Special Isotope [II] V.W. Gorman and R.C. Guenzler, "LOFT SSE Separation Project and the New Production Reactor, and also Definition and Seismic Analysis Methods", for continued safe operation of existing facilities. LTR-10-19, October 8,1974. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

203 112] R.C. Guenzler and V.W. Gorman, PBF Seismic [20] U.S. Department of Energy, UNEL Architectural Analysis Method and Design Basis Input. Aerojet Engineering Standards. Idaho Falls, Idaho, Revision Nuclear Company Report TR-256, March 23,1972. No. 3, June 1982.

[13] Westinghouse Idaho Nuclear Company, Plant Safety [21] U.S. Department of Energy. DOE-ID Architectural Document, IPM-I1, Section 8.2, January 1982, Engineering SlflTUlarrti'ti Idaho Falls, Idaho, Revision p. 3.6-32. No. 6, October 1986.

[14] G.W. Housner, "Behavior of Structures During [22] U.S. Department of Energy, DOE Order 6430.1 A, Earthquakes", Journal, Engineering Mechanics General Design Criteria Manual. April 6,1989. Division, ASCE, vol. 85, no. 4, October 1959. [23] D.W. Coats and R.C. Murray. Natural Phenomena [15] U.S. Nuclear Regulatory Commission, Regulatory Ha7^rris Modeling Project: Seismic Hazard Models Guide 1.60, "Design Response Spectra for Seismic for Department of Energy Sites Lawrence Livermore Design of Nuclear Power Plants", Revision 1, National Laboratory Report UCRL-53582, Revision 1, December 1973. November 1984.

[16] Woodward-Clyde Consultants. A Seismic Hazard [24] R.P. Kennedy, et al, Design and Evaluation Guidelines Study for the LOFT Reactor Facility at the INEL. for Department of Energy Facilities Subjected to Idaho, Oakland, California, June 1975. Natural Phenomena Hazards. Lawrence Livermore National Laboratory Report UCRL-15910, May 1989. [17] C.E. Williams (DOE-ID) letter to F.H. Anderson (ACC), "Design Basis Earthquake Criteria Used at [25] TERA Corporation, Seismic Hazard Analysis for the ICPF'.June 1,1978. Idaho National Engineering Laboratory. Berkley, California, October 1984. [18] U.S. Department of Energy, INEL Architectural Engineering Standards. Idaho Falls, Idaho, Original [26] 1. Wong, et al, "Earthquake Strong Ground Motion Issue, December 1978. Studies at the Idaho National Engineering Laboratory", Paper presented at the Second DOE Natural [19] R.C. Guenzler letter to L.E. Little, RCG-38-81, "UBC Phenomena Hazards Mitigation Conference, Seismic Risk Zone Change", October 26,1981. Knoxville, Tennessee, October 3-5,1989.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~204 Session 7 DOE Orders Codes Standards~~

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

~~205 COMPARISON OF EVALUATION GUIDELINES FOR LIFE-SAFETY SEISMIC HAZARDS~~

~~Loring A. Wyllie Richard Jay Love H. J. Degenkolb Associates 350 Sansome Street, Suite 900 San Francisco, California 94104~~

~~ABSTRACT~~

The guidelines presented in "Design Evaluation guidelines for Department of Energy Facilities Subjected to natural Phenomena Hazards" (UCRL 15910 Draft; May 1989) include evaluation criteria for existing Department of Energy buildings subjected to earthquakes. These criteria were developed at the Lawrence Livermore National Laboratory for use in both the seismic design of new structures and the evaluation of existing structures. "ATC-14: Evaluating The Seismic Resistance of Existing Buildings" developed by the Applied Technology Council, consists of guidelines and criteria for "identifying the buildings or building components that present unacceptable risk to human lives." This paper compares and contrasts the two evaluation guidelines for existing buildings using a prototype building as an example. The prototype building is a seven story, concrete shear wall building assuming a General Use Occupancy.

INTRODUCTION design and evaluation criteria for The Department of Energy (DOE) has protection against hazardous natural numerous facilities throughout the United phenomena. These guidelines, entitled States. These sites have many General "Design Evaluation Guidelines for Use or Low Hazard buildings designed and Department of Energy of Energy Facilities built according to earlier Codes that do Subjected to Natural Phenomena Hazards," not address the some of the seismic (UCRL 15S10 Draft; May 1989) have been hazards that affect the strength and referenced by the General Design Manual, ductility of the structure. There is a DOE Order 6430.1A, as an acceptable current interest in the evaluation of approach to the evaluation of facilities. these existing buildings considering the This UCRL document was prepared for the probable hazards and the consequences of Lawrence Livermore National Laboratory those hazards on the both the occupants (LLNL) under the direction and review of and the functions of these facilities. the DOE natural Phenomena Hazards Panel. The DOE has published draft In a parallel effort by the private guidelines in order to provide uniform sector, the Applied Technology Council,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

206 with funds provided by a grant from the PRIMARY COMPONENTS OF A SEISMIC National Science Foundation, has EVALUATION published guidelines and criteria for A seismic evaluation of a building evaluation of seismic hazards in existing should include a description of the buildings. The document, published in desired performance goals of the 1987, is entitled "ATC-14; Evaluating the building, the ground motion criteria, the Seismic Resistance of Existing basic lateral strength requirements, the Buildings," This document was prepared by acceptance criteria, system H. J. Degenkolb Associates and reviewed configuration, and ductile detailing of a Project Engineering Panel composed requirements. The following sections of practicing engineers and academicians. discuss relevant issues for each of these (There is currently an effort, sponsored elements. by FEMA, to develop a handbook on seismic evaluation of potentially hazardous PERFORMANCE GOALS buildings. This document, ATC-22, is UCRL 15910 establishes performance based upon the ATC-14 methodology. The goals for each of four facility-use document is currently under review the categories. These facility categories Building Seismic Safety Council.) include 1.) General Use facilities, 2.) Both documents are intended to Important or Low Hazard facilities, 3.) provide guidance to the Engineer in the Moderate Hazard facilities, and 4.) High evaluation of existing structures against Hazard facilities. The performance goals seismic hazards. The UCRL document is are stated in terms of the annual broader in that it also provides guidance probability of exceedance of an event for other natural phenomena as well as that causes a prescribed level of damage. the design of new structures. However, The performance goal for a General the two documents differ in several ways U3e occupancy is a 10"5 annual probability that relate to similar types of buildings of exceedance for the onset of major and may lead to different conclusions structural damage that might endanger the regarding the most appropriate methods life-safety of the occupants. This is for analyzing, evaluating, and eventually equivalent to a five percent chance of strengthening existing structures. The exceedance over a fifty year lifetime of similarities and differences are explored the building, or ten percent chance in in order to comment on the final results 100 years. This performance goal is of evaluations for seismic hazards in essentially a life-safety goal with existing buildings. little concern for the post-earthquake building functions. PROTOTYPE STRUCTURE ATC-14 also establishes a For the purpose of comparison, a performance goal related to the life- prototype structure has been chosen to safety of the occupants of a building. compare the two guidelines. The building Future use of the building is not a is a General Use office structure located consideration in the performance goals. at the Lawrence Livermore National The criteria are intended to identify Laboratory in California. The structure potentially hazardous buildings. ATC-14 is a reinforced concrete building with defines a hazardous building as one where 3even stories. The lateral force the entire building collapses, a portion resisting system is comprised of concrete of it collapses, components fail and shear walls. Particular attention will fall, and/or exit routes are blocked be paid to the evaluation of the shear preventing evacuation or rescue of the capacity, boundary elements and coupling occupants. The criteria are based on a beams of a shear wall element. ten percent chance of exceedance in 50 years. Because both ATC-14 and UCRL-15910 General Use criteria are primarily intended for life-safety considerations,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

207 it is appropriate to compare their ATC-14 effective peak acceleration.In respective provisions for ground motion, order to compare EPA values with the same lateral strength requirements, system chance of exceedance, for example, five completeness, and ductile detailing percent in 50 years, the ATC coefficient provisions, would have to be increased from 0.4g to 0.5 g. Therefore, using a five percent GROUND MOTION CRITERIA exceedance EPA for both criteria, the UCRL specifies three alternative UCRL value is 0.37g while the ATC-14 methods for determining the ground motion value is 0.50g. This anomaly may be criteria for a particular site. These explained by the site specific nature of sources include Newmark-Hall spectral the studies that established the UCRL construction, site-specific spectra, and 15910 PGA as opposed to the more general the 1988 Uniform Building Code (UBC). studies available for ATC-14 criteria. For use in any of the three methods, UCRL 15910 specifies the expected maximum LATERAL STRENGTH REQUIREMENTS peak horizontal ground surface The lateral strength requirement for acceleration (PGA) as a function of the a structure is most often defined by a performance goal and the specific DOE required design base shear force. The site. Because the expected PGA design base shear of a structure is a represents a maximum spike that may only function of the effective peak ground occur once during a seismic event, this acceleration, the structural maximum peak ground acceleration may be amplification of the ground acceleration, reduced ten percent to establish an the period of the structure, the soil "effective" peak acceleration (EPA). characteristics, and the structure's UCRL 15910 defines the effective peak inertial mass. The design base shear, as ground acceleration as a repeatable level a function of building period, is best of acceleration in the frequency range shown in a response spectrum as the that adversely affects structures. The product of the effective peak ground EPA has been shown to be a better acceleration, Z, and an amplification indicator of the damage potential of an factor, C. The ZC values as a function earthquake. The maximum peak ground of the building period are shown in acceleration for the LLNL site is 0.41g Figure 1. The amplification factor for the General Use performance goal. includes the effect of the period of the Therefore, the effective peak structure, the assumed damping, as well acceleration is 0.37g. This EPA factor as the soil characteristics. may be used to create a Newmark-Hall For General Use facilities, UCRL- spectrum, anchor a normalized site 15910 specifically references the specific spectrum, or to substitute in criteria of the 1988 UBC for determining place of the Zone factor, Z, in the UBC the design base shear requirements for base shear equation. existing buildings. The base shear ATC-14 describes ground motion equation of the 1988 UBC, V^ZICW/R,, is criteria with an effective peak velocity- used to establish seismic strength selated acceleration (EPA) coefficient, requirements for static analyses of A,. A contour map of the United States regular buildings and/or dynamic analyses indicates the A, values for use in the of irregular buildings. However, UCRL seismic evaluation. This coefficient permits some modifications to the represents the effective peak definitions of the equation terms. The acceleration for a site with a ten first modification relates to the Zone percent chance of exceedance in 50 years. factor, Z, earlier referred to as the There is an anomaly between the UCRL effective peak acceleration. The and ATC-14 effective peak acceleration effective peak ground acceleration, 2 = values in that the UCRL 15910 criteria 0.37g, may be substituted for the UBC establishes a lower EPA correlated to a Zone factor, Z = 0.4g for the LLNL site. lower probability of exceedance than the

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~208 5=1.2~~

~~NEWMARK-MALL MEDIAN 596 DAMPED EPA=O.37g UBC 8S Z-0.4. S-1.2~~

~~i.5 2.a a. A .0~~

~~ZC vs PERIOD~~

~~Figure 1 ZC versus PERIOD~~

The second modification to the 1988 spectrum is 40 percent greater than the UBC is related to the amplification median Newmark-Hall spectrum. factor, C, in the base shear equation. However, the practical difference Instead of using the UBC equation, between the two spectra may be only C - 1.25xS/ T2/\ the amplification academic because, unless a special site coefficient may be determined from either specific study is done to justify a lower a median (50th percentile) Newmark-Hall spectrum, UCRL 15910 requires that the spectrum or from a median site specific product of the Zone factor and the spectrum. Note that the UBC equation amplification factor, ZC, be compared defining C is closely based on the median with that obtained from the standard 1988 plus one standard deviation (sigma) UBC coefficients. The larger of the two spectrum, also referred to as a one sigma values must be used. As seen in Figure spectrum. The difference between the 1, the 1988 UBC spectrum, or ZC, exceeds median and the one sigma spectra is the UCRL modified spectrum at all periods evident in Figure 1 by comparing the of vibration. Thus, the engineer must Newmark-Hall median spectrum with the evaluate the structure using full UBC 1988 UBC spectrum. In th-2 short period level forces unless special site specific ranqe, up to 0.5 seconds, the 1988 UBC justification is provided.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

209 If a site specific spectrum is used, predicated on ductile details required of a lower ZC product is acceptable as long the particular system under as any significant discrspancies between consideration. Without these special the site-specific spectrum and the UBC details, the UBC R, factors do not a.<: justified and accepted by the DOB. correspond to the expected seismic Criteria for this justification is not performance of the building. The provided in UCRL 15910. consequences of this problem and possible The ATC-14 base shear determination approaches are discussed in later is developed from a median spectrum with sections. assumed five percent damping of the system. The effective peak-velocity- EVALUATION OF SEISMIC CAPACITY related acceleration, A,, is similar to UCRL-15910 provides a general the UBC zone factor, Z. Both represent comments for the evaluation of existing an effective ground acceleration for General Use facilities and a specific analysis purposes. The -\, values ate reference to "all 1988 UBC provisions,... presented in a national map showing regardless of whether they are discussed contour lines of equivalent peak herein." Realistic estimates of the dead accelerations. The spectral and live loads for all elements are amplification factor, C, is given by the combined with the seismic forces using equation, C- 0.8xS/ (T)"\ Note that appropriate load factors to establish this equation is about 65 percent of the demand forces for all members. Member UBC 1988 amplification factor in the demand forces are then compared with short period range. This is indicative member capacities. If the member demand of the practical difference between the force is less than the member capacity, one sigma and the median response the element is deemed acceptable. spectra. Therefore, the ATC-14 design In addition to the capacity, the base shear of the structure represents a members must be reviewed to assure that median spectrum while the UCRL base shear the special detailing requirements of the represents a one sigma spectrum value. 1988 UBC are satisfied for Zone 3 and Using the prototype building as an Zone 4 sites. The special detailing example for comparison, and assuming a review requirement is appropriate to structural period of 0.5 seconds, the assure that the structural system has the ATC-14 base shear coefficient is 0.076 W. ductility that is assumed in the The UCRL 15910 base shear coefficient, development of the R» reduction factors. using the modifications to the EPA and However, the reduction factors assume a amplification factors, is 0.098 W, an level of ductile detailing frequently not increase of about 30 percent over ATC- found in existing buildings. For 14. Using the 1988 UBC, the base shear example, the boundary elements of shear coefficient would be 0.119 W, an increase walls may require closely spaced ties in of about 55 percent over ATC-14. As order to confine the concrete core and stated above, without special site prevent buckling of the compression studies, the Engineer must use the reinforcement. This requirement for greater of the two UBC coefficients, shear wall buildings was not a part of ultimately evaluating the building for a the UBC until 1985. Prior to 1985, significantly higher force than ATC-14. ductile boundary elements for shear walls The last factor necessary for were only required for dual system determination of the base shear is the R» buildings. factor. This factor reflects the If the existing building does not expected performance of the structural have the ductile details envisioned by system in the post-elastic range. A the R« reduction factor, the Engineer system with greater ductility may be must decide how to evaluate the capacity designed for lower seismic forces than a of the existing element. No guidance is less ductile structural system. However, provided in UCRL 15910 for this the R.factors given in the UBC are evaluation. One option, albeit a

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

210 potentially very expensive one, is to design force level must be increased to discount the capacity of the element and account for the lack of ductility in the recommend either upgrading the existing element. element to meet existing 1988 UBC detailing requirements, or replacing the SYSTEM CONFIGURATION existing element with a new element. UCRL 15910 provides a general This option ignores the potential of the discussion of the beneficial aspects of existing element to resist the demand structural system configuration. It forces in an elastic manner, a recommends greater care by the Engineer possibility if the capacity of the in the evaluation of irregular buildings. element is sufficiently greater than the Re-entrant corners, soft stories, and normal seismic design force levels would large changes in vertical distribution of normally require. mass are all mentioned as possible ATC-14 provides guidance to the sources of increased local damage to Engineer for evaluating the capacity of elements of the structure. Particular existing elements that do not conform to criteria for the evaluation of these current ductile detailing requirements. problems must be located in the 1988 UBC. This guidance is in the form of Capacity/Demand ratios that reflect the increased elastic capacity required of ATC-14 provides specific guidance specific elements if the ductility of for identifying potential weak links these elements is judged insufficient. caused by irregular structural The ATC-14 requirements depend on the configurations. For example, significant required ductility of the member. For torsion is defined as a condition where ductile elements, the C/D ratio need only the centers of mass and rigidity are be 1.0. For elements with semi-brittle located more than 20 percent of the plan and brittle failure mechanisms, the dimension of the structure. If this required element capacities are 0.2R, to condition occurs, the building must be 0.4R. times the code level demand forces evaluated for story drifts that include on the element. the torsional response. Vertical load Using the prototype concrete shear carrying elements must maintain their wall example, both the boundary elements capacity when subjected to deflections and coupling beams must either possess caused by O^R^ times the design forces. closely spaced ties to confine the core Soft stories, defined in ATC-14 as a 20 or they must have sufficient strength to percent decrease in yield capacity from resist 0.2 R., or 1.6, times the code one story to the story below, are treated level forces for an R, equal to eight for in a similar manner. Seismic forces shear walls. If the element can meet this must be increased by 0. 4R. times the criteria, the element is judged adequate equivalent lateral forces to account for without upgrading to 1988 UBC detailing this irregularity in configuration. requirements. This represents a significant cost savings if the element DUCTILE DETAILING REQUIREMENTS meets this criterion and therefore, does As mentioned earlier, UCRL 15910 not require seismic upgrading. requires the Engineer to review the While ATC-14 provides guidelines for details of an existing building for the acceptance of elements that do not compliance with the 1988 UBC requirements meet the Code detailing requirements, at for ductile detailing. Many existing the same time it provides guidelines for buildings will not meet these the design of the seismic upgrade of requirements; some of the ductile existing elements if they do not meet the detailing requirements are relatively ATC-14 guidelines. The design of the recent additions to the Code. upgraded element must meet the same ATC- Recommendations for the evaluation of the 14 C/D ratio used for the evaluation of useful capacity of these non-conforming the original element. Therefore, the members is not provided in the UBC.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

211 ATC-14 identifies the areas of the transverse ties. No guidance is provided structure where ductile detailing ia for evaluation of the shear wall where beneficial to the seismic performance of the boundary element details do not meet the building. If the ductile detailing the Code requirements. is not present, then the capacity of the A second example of the ductile element in question must be evaluated detailing criteria differences between against an increased demand force. If the UBC and ATC is found in the the element has sufficient capacity to provisions for deep beams between coupled resist the increased demand force, then shear walls. The UBC requires special the ductile detailing requirement may be symmetrical diagonal shear reinforcing waived. The element is judged capable to extending the full length of the coupling resist the seismic forces elastically and beams when the factored shear stress in 0 5 does not need the ductility normally the beams exceeds 4 (f'o) ' . Very few associated with design level seismic existing structures have this special forces. type of detailing. The boundary element requirements of The ATC-14 criterion applies for concrete shear walls are an example of coupling beams located over means of ductile detailing requirements found in egress. In these instances, the coupling the both the 1988 UBC and ATC-14. In beam is deemed adequate if the beam has ATC-14, the boundary element criterion is transverse stirrups spaced at 8 db or less automatically satisfied if the spacing of and are anchored into the core with 135 transverse tie3 is less than eight times degree hooks. If the beam does not meet the diameter of the longitudinal this requirement, then the capacity must reinforcement. This criterion is be compared to 0.2 R,, times the demand intended to prevent buckling of the forces. If the beam capacity is greater longitudinal bars and is based on the than this value, the beam is judged critical buckling length of the adequate for life-safety. If the beam longitudinal bars. does not meet this requirement, then Comparing with the UBC criterion, the either its ductile detailing or its ductile boundary elements, where capacity must be upgraded to meet the required, must have transverse ties ATC-14 criterion. meeting the requirements for concrete In summary, the 1988 UBC ductile columns in a special moment resisting detailing criteria are significantly more space frame (SMRSF). The transverse ties stringent than the ATC-14 detailing may be spaced at no greater distance than criteria. This should be expected from a four inches apart, irrespective of Code intended for the design of new longitudinal bar diameter. This SMRSF structures. The Code ductile detailing spacing criterion was developed primarily requirements improve the expected seismic to prevent shear failures in columns due performance of the structure to a higher to seismic bending moments and to degree than that required for life- increase the ultimate strain capacity of safety. The Code criteria are intended the concrete through confinement of the to limit structural damage as well as column core concrete. A secondary protect lives. This is a higher benefit of the close spacing is the performance goal than just life-safety prevention of buckling of the and should be recognized as such. If longitudinal reinforcement. Because life-safety is the only performance goal, shear failures in the boundary elements then the ATC-14 detailing requirements of tall shear walls is not a primary are intended to offer a level of concern, such close spacing is not protection commensurate with this goal. warranted to prevent longitudinal bar buckling. Therefore, using the UBC Code to review the detailing of the boundary member results in a much more restrictive criterion for acceptable spacing of

~~Second DOE Natural Phenomena Hazards Mitigation Conference -- 1989~~

212 CONCLUSIONS In addition to the required base The stated intention of the UCRL- shear issue, the evaluation guidelines 15910 guidelines for evaluation of should address a method for establishing existing DOE structures is to provide the reliable capacity of critical life-safety level performance of the elements with non-conforming details in structure. In referencing the 1988 UBC an existing building. This can be done requirements for the determination of the by 1.) either using increased C/D ratios basic lateral strength, configuration, for certain elements of the structure member capacity determination, and that are critical to the seismic ductile detailing requirements, UCRL performance of the building, or 2.) 15910 requires the existing building to decreasing the R,, for the entire meet the same requirements as a new structure to account for the lack of structure. As many existing buildings do ductility in some of the structural not have the basic design strength nor details in the existing building. Both the ductile details now required of new methods effectively require greater structures, it is expected that they will strength to account for lack of not meet these criteria. ductility; however, the former method ATC-14 represents a guideline for the defines only key elements of the building evaluation of existing buildings that while the latter method affects the recognizes that existing buildings should evaluation of the entire building. It is not necessarily have to meet the the former method of evaluating only the requirements of the new code. Using a critical elements of a structure for lower basic design force and providing a life-safety performance goals where cost method to evaluate non-conforming savings may be realized. Whatever structural details, ATC-14 represents a approach is taken, the evaluation of life-safety level approach to the existing structures for seismic forces evaluation and upgrading of existing requires judgment and care by the buildings. Engineer in selecting and utilizing the DOE may wish to consider alternatives most appropriate criteria for the task at to the 1988 UBC requirements for the hand. evaluation of life-safety of existing buildings. The use of a lower lateral strength requirement, reflected by a median spectrum, appears proper for evaluation of existing buildings. The requirement that the larger of the two base shear coefficients, ZC from a median response spectrum and ZC from the UBC, results in a significant increase in the required base shear.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~213 PROBABILITY-BASED JUSTIFICATION FOR REDUCTION OF SEISMIC DEMAND FOR SHORT DURATIONS~~

~~Masoud Moghtaderi-Zadeh Sohrab Esfandiari IMPELL Corporation P.O. Box 5013 San Ramon, CA 94583~~

~~ABSTRACT~~

Currently, in seismic design of nuclear structures and facilities two levels of site ground motion are specified. One is termed the safe shutdown earthquake (SSE) and the other is termed the operating basis earthquake (OBE). Recent probabilistic seismic hazard evaluations for nuclear sites have generally demonstrated that the seismic requirements may be too conservative.

The current requirements are to declare a system as operable only when it meets all the requirements of the original design criteria, including the seismic requirements. This means that for any given short duration a system must have seismic capacity comparable to what is required during its design life. The probability of occurrence of the design SSE and OBE during any given short period of time is lower compared to the probability of occurrence of the design SSE and OBE during the plant life. Consequently, the current practice is an excessively conservative one. Using a probabilistic approach, revised ground motion levels for short durations may be obtained such that comparable risk is maintained.

This philosophy has been applied, in concept, in the nuclear industry. Gener- ally, deviations from design basis requirements (acceptance criteria) are granted for shorter finite durations. Such deviations allow for less conservative requirements for shorter durations in order to allow the applicant more time to fully comply with the design basis requirements, meanwhile allowing the plant to be operational for the shorter durations.

~~This paper provides a probability-based justification for such deviations for short durations. A probabilistic method is developed to obtain revised ground motion levels for short duration.~~

INTRODUCTION quake (SSE) and the other is termed the operating Currently, in seismic design of nuclear structures basis earthquake (OBE). Generally, the design SSE and facilities two levels of site ground motion are for a nuclear power plant represents the expected specified. One is termed the safe shutdown earth- ground motion at the site from the largest historic

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

214 earthquake within the tectonic province within levels at all frequencies. which the site is located. Alternatively, this may be determined from an assessment of the maximum Recent probabilistic hazard studies [e.g., 9] have earthquake potential of the appropriate tectonic generally demonstrated that the design basis spec- structure or capable fault placed at the closest tra, obtained using the deterministic approach, for epicentral distance to the site. In any case, a design some plants may be too conservative. It is this layer earthquake is defined in terms of earthquake charac- of conservatism that is utilized to develop a lower teristics, such as, its Modified Mercalli Intensity, spectrum for shorter durations. Furthermore the magnitude and distance, or peak ground accelera- probability of occurrence of a design earthquake tion, velocity and displacement at the site. during any given short period of time is much lower than the probability of occurrence of the design From the design earthquake, the design SSE re- earthquake during the plant life. Consequently the sponse spectrum is then obtained using existing current accepted practice of requiring systems to earthquake records for the site (or for sites with meet original seismic design criteria while undergo- similar characteristics), or predefined spectral ing short-term repairs, is a conservative one. By shapes, such as Housner [1], Reg. Guide 1.60 [2] and using a probabilistic approach, it is possible to NUREG/CR-0098, [3]. The design OBE spectrum, calculate the probability of experiencing a seismic for a site, is generally taken as one half the SSE event during any given period of time. From this, spectrum, [2J. revised ground motion intensities and response spectra for short durations may be obtained such The process described above is essentially a deter- that comparable risk over the life of the plant is ministic definition of the design SSE and OBE maintained. It is the use of the short-term spectra spectra. At every step in this process, various layers that alleviate the need for interim hardware repairs, of conservatism are introduced. thereby permitting the plant to remain operable during the short-term interim period. There is, In the past two decades, modern probabilistic however, the need to complete all required long-term seismic hazard evaluation methodologies have been hardware repairs prior to the end of the interim pe- developed [4,5,6] and evolved into state-of-the-art riod. analytical techniques [7, 8] to develop site-specific response spectra for design of structures, specifically Generally, deviations from design basis conditions nuclear power plants. These techniques incorporate are granted (whether in terms of seismic demand or uncertainties in modeling techniques and incomplete acceptance criteria) on a case-by-case basis for data, as well as inherent randomness, in the occur- shorter periods of time, such as one refueling cycle rence time, location and characteristics of earth- for nuclear power plants. Such deviations allow for quakes, in a rational manner using probabilistic less conservative acceptance criteria for shorter methods. In a probabilistic site-specific seismic risk durations in order to allow the utility more time to assessment, the hazard is usually quantified in fully comply with the design basis conditions and, terms of annual probability of exceedance of a meanwhile, allowing the plant to remain operational ground motion intensity level at the site. The for one refueling cycle. ground motion intensity at the site may be, peak ground acceleration (PGA), peak ground velocity This paper provides a probability-based justifica- (PGV), peak ground displacement (PGD), or spectral tion for such deviations for short durations. A proba- acceleration (SA), for a specific damping ratio and bilistic method is developed to obtain revised ground frequency of an oscillator. The hazard curves, which motion levels for sh-Tt durations. are plots of annual exceedance probability (or return period) versus ground motion intensity levels, are then constructed. From the hazard curves for PGAs, PROBABILISTIC SEISMIC HAZARD and SAs, for a given damping ratio at various fre- EVALUATION quencies, uniform probability response spectra may The methodology to calculate the seismic hazard be obtained. A uniform probability response spec- at a site is well established in the literature [4,5, 6, trum for a given damping ratio has the same annual 7, 8]. The fundamental method requires specifica- probability of exceedance of spectral acceleration tion ofthree types ofinputs:

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~215 1. Geometry of the seismic sources.~~

~~2. Seismicity within each source (distribution of earthquake magnitude and their occurrence in time and space).~~

3. An attenuation law allowing estimation of ground motion at the site as a function of earthquake magnitude, distance and other characteristics of earthquake, properties of the medium in which earthquake waves travel, and site conditions.

The purpose of a probabilistic seismic hazard analy- Peak Ground Acceleration Level (g) sis is to incorporate uncertainties and randomness of Figure 1 the above input, and compute a site-specific seismic hazard. In its simplest form, which is given here for the purpose of illustration, the probability that a Damping Ratio * 5% ground motion parameter at the site, Y, exceeds a Frequency - 5 Hz level, y, in any given year is: » jj P[Y>y] = £ v, P[Y>y I i.m.r] 1 (HI*.) fMli(m, i)drdm a) where P[Y>y I i,m,r] is the conditional probability that ground motion parameter, Y, exceeds the level, y, when an earthquake, with magnitude, m, and distance from the site, r, occurs in source, i. v i is the annual rate of occurrence of earthquakes in source, i- fRli,m(r> i»m) is the conditional probability Spectral Acceleration Level (g) density function of distance to the site given an Figure 2 earthquake with magnitude, m, occurs in source, i, fMli(m,i) is the probability density function for magnitude of an earthquake occurring in source, i, and Nt is total number of seismic sources in the Design Basis region of the site. Response Spectrum Various methods may be used to compute annual probability of exceedance of ground motion parameter P[Y>y], from Eq. (1) or similar equations, e.g., using analytical integration, numerical integration, Monte-Carlo simulation, or first-order and asymp- Uniform Probability Response Spectrum totic second-order reliability methods. (iff* Probability of Exceedance) By changing the intensity level, y, and computing P[ Y>y], the hazard curve for ground motion parameter, Y, is obtained. As mentioned earlier, a hazard Frequency (Hz) curve is usually the plot of annual probability of exceedance of a ground motion parameter versus Figure 3

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 216 ground motion intensity level. Figure 1 shows a for optimal design, systems should be designed for typical hazard curve for PGA for a site, Figure 2 pe = p0, However, as it was discussed previously, shows a typical hazard curve for SA for an oscillator most plants have been designed using some sort of with 5% damping and frequency of 5 Hz, for the site. deterministic process with built-in layers of conser- Similar SA hazard curves can be obtained for other vatism in the process to deal with uncertainties, in frequencies and damping values. One should note an ad-hoc manner. As a result of this kind of deter- that due to uncertainty and lack of perfect knowl- ministic approach, the plants may have been de- edge, the hazard curves, in general, have uncer- signed for earthquake levels which may indicate an tainty associated with them. The curves in Figures 1 actual risk level of p0 greater or smaller than accept- and 2 represent the mean value of the hazard. able level. Recent probabilistic methods have generally demonstrated that most plants have been From the hazard curves for PGA and SA at various designed for very conservative earthquake levels. frequencies, uniform probability response spectra at a site are then obtained. A uniform probability It is not the purpose of this paper to establish the response spectrum for 10"4 annual probability of acceptable levels of risk for design purposes. But exceedance is shown in Figure 3. In this figure let's assume for a design earthquake, e.g. SSE, that spectral acceleration levels at all frequencies have the acceptable risk level, p0, is given. Furthermore, the same annual probability of exceedance of 10'4. the probabilistic site-specific studies have been performed, and actual risk levels, pe, have been SEISMIC RISK FOR FINITE DURATIONS It is often of interest in the design of a system to Risk define the probability that an event, e.g., the design SSE level, will be exceeded during a specific period,

e.g., life of the system. If pc is the actual annual p probability of exceedance of a design level, then the o complementary probability, 1 - pe, is the annual probability that the design level will not be exceeded. The value of pe can be obtained from Eq. (1), i.e., pe = P|Y>yd], where yd is the design level, or from hazard curves similar to Figures 1 and 2. If the exceedance (a) R Time events in any given year are assumed to be independent events, then the probability that the design Risk level will not be exceeded in a duration of D years, e.g., design life, is (1 - peT- Thus the exceedance probability in the duration D, PD is:

PD=1-(1-P/ (2) where D is expressed in years. | s | R Time For a system with an acceptable annual probability of exceedance of design level of p0, then the acceptable total risk, over the life of the system, L years, is Figure 4 from Eq. (2): obtained. Two cases may occur. The first case is when po

The second case is when po>pe, i.e. the actual risk is Suppose the actual probability of exceedance of a less than the acceptable level (Figure 4a). This is the design level for an existing system is pc. In general, case where the procedure presented in this paper

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

217 may be used to justify the increased risk for a short new facility, Eq, (7) may be used by replacing R with finite duration (Figure 4b),, i.e. reduction of seismic L, i.e. the remaining life is the same as design life of demand. the facility.

If the system is used foi a short period, with a SUMMARY reduced strength, such that the risk in the short period increases from actual annual risk of p0, for The procedure described in this paper provides a normal strength, to ps, for reduced strength, then the probability-based justification for reduction of total risk over the intended remaining life, R years, seismic demand for short durations. The same is: methodology may be applied to hazard curves for spectral accelerations,, to derive applicable response PR=1-(1-PC)R"S(1-PS)S (5) spectrum for short periods. This is feasible if at all frequencies of interest, the spectral accelerations corresponding to design basis spectrum exceed those If pa p0 the total risk PR over the remaining life must be limited to the acceptable total risk, (from Eq. 4), then the corresponding acceptable annual risk for the short period, p£, is obtained by equating PK(from Eq. 5) to Ff (from Eq. 4), thus:

~~p (1-Po) (6) Design Basis L(i-Pc)RSJ Response Spectrum Note that the above equation gives acceptable risk for a short period for when pe~~

Let's examine the case where the seismic resistance Figure 5 of a facility for an unknown period S, is reduced from corresponding to allowable uniform probability (e.g. its normal strength (which would cause an actual 104) response spectrum. Figure 5 shows such typical risk of pep0). Then, in order to obtain response spectrum is conservative over frequency the maximum durations, which the facility may range of interest relative to an acceptable uniform operate with the lower strength, the total risk from probability spectrum. Thus for a short period, e.g. Eq. (5) is equated to the acceptable risk, i.e.: six months, a reduced response spectrum may be abtained using the procedure described earlier. - P (7) Applications of this procedure to existing older In1 -Ps "1 -Pe plants, e.g. typical of DOE plants, or older nuclear where Smm is the maximum length of duration within plants, is self-evident. Justification for lower seismic which the facility could be operated with reduced demand over a finite short duration (e.g. one year, or strength such that total risk over the remaining life one operational cycle) may be demonstrated provided is equal to the total acceptable risk. Note that for a the overall risk is not impaired. Justifications for continued operation of some systems with temporary

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

218 lower strength may be demonstrated in order to of America, Vol. 58,1968, Pp. 1583 -1606. allow time for long term fixes, without having to shut down the facility. In summary, if a facility 5. McGuire, R. K., "FORTRAN Computer maintains an acceptable risk level throughout its Program for Seismic Risk Analysis:", U. S. life, using the procedure stated in this paper, it may Geological Survey Open-File Report 76-67, be demonstrated that for a short finite period, the 1976, P. 99. facility may be operated with increased level of risk (reduced seismic demand), or reduced strength. 6. Der Kiureghian, A. and A. Ang, "Fault Rupture Model for Seismic Risk Analysis", REFERENCES Bulletin of Seismological Society of America, Vol. 67,1977, Pp. 1173-1194. 1. Housner, G. W. and P. C. Jennings, "Earth- quake Design Criteria", Earthquake Engi- 7. "Seismic Hazard Methodology for the Central neering Research Institute, Berkeley, CA, and Eastern United States", Vol. 1-10, 1982. Electric Power Research Institute, EPRINP- 4726, Final Report, July 1986. 2. USNRC Regulatory Guide 1.60, "Design Response Spectra for Seismic Design of 8. NUREG/CR-5250, "Seismic Hazard Charac- Nuclear Power Plants", U. S. Atomic Energy terization of 69 Nuclear Plant Sites East of Commission, Revision 1,1973, the Rocky Mountains", UCID-21517, Law- rence Livermore National Laboratory, 3. NUREG/CR-0098, "Development of Criteria January 1989. for Seismic Review of Selected Nuclear Plants", May 1978. 9. Coats, D. W. and R. C. Murray, "Natural Phenomena Hazards Modeling Project: 4. Cornell, C. A., "Engineering Seismic Risk Seismic Hazard Models for Department of Analysis", Bulletin of Seismological Society Energy Sites", UCRL-53582, Rev. 1, Law- rence Livermore National Laboratory, 1984.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~219 SEISMIC DESIGN CRITERIA FOR SPECIAL ISOTOPE SEPARATION PLANT STRUCTURES~~

M. W. Wrona, Bechtel, P.O. Box 3965, San Francisco, CA 94119 S. J. Wuthrich, Bechtel, P.O. Box 3965, San Francisco, CA 94119 D. L. Rose, DOE-ID, 785 DOE Place, Idaho Falls, ID 93402 J. Starkey, WINCO, LLNL, P.O. Box 5508, Bldg. 3725, Livermore, CA 94550

ABSTRACT This paper describes the seismic criteria for the design of the Special Isotope Separation (SIS) production plant. These criteria are derived from the applicable Department of Energy (DOE) orders, references and proposed standards. The SIS processing plant consists of Load Center Building (LCB), Dye Pump Building (DPB), Laser Support Building (LSB) and Plutonium Processing Building (PPB). The facility-use category for each of the SIS building structures is identified and the applicable seismic design criteria and parameters are selected.

INTRODUCTION CLASSIFICATION OF SIS STRUCTURES The Special Isotope Separation The selection of the appropriate (SIS) production plant will be facility-use category for the SIS located at the Idaho National structures is based on the Engineering Laboratory (INEL). definitions included in UCRL-15910 [2]. This document identifies four The plant will be used for facility-use categories: separation of plutonium isotopes using the Atomic Vapor Laser o General Use Isotope Separation (AVLIS) process o Important or Low Hazard developed by Lawrence Livermore o Moderate Hazard National Laboratory (LLNL). The o High Hazard nuclear materials will be processed in the reinforced Guidelines for the facility-use concrete Plutonium Processing categories are defined in Table 1, Building (PPB). The support which is taken from UCRL-15910. facilities will consist of three DOE Order 6430.1A also identifies steel frame structures: the Laser facilities as critical or Support Building (LSB), Dye Pump non-critical based upon their Building (DPB) and Load Center national importance to DOE Programs. Building (LCB). In accordance with the DOE Order and The design of the plant will its associated references the comply with DOE Order 6430.1A [1] following classifications have been and the related referenced codes determined. and standards.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~220 TABLE 1~~

~~FACILITY USE CATEGORY GUIDELINES~~

FACILITY-USE CATEGORY DESCRIPTION General Use Facilities Facilities which have a non-mission dependent purpose, such as administration buildings, cafeterias, storage, maintenance and repair facilities which are plant or grounds oriented.

Important or Low Hazard Facilities which have mission dependent Facilities use (e.g., laboratories, production facilities, and computer centers) and emergency handling or hazard recovery facilities (e.g., hospitals, fire stations).

Moderate Hazard Facilities Facilities where confinement of contents is necessary for public or employee protection. Examples would be uranium enrichment plants, or other facilities involving the handling or storage of significant quantities of radioactive or toxic materials.

High Hazard Facilities Facilities where confinement of contents and public and environment protection are of paramount importance (e.g., facilities handling substantial quantities of in-process plutonium or fuel reprocessing facilities). Facilities in this category represent hazards with potential long term and widespread effects.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

221 o The LSB, DPB, LCB and the 5 percent damping median response non-safety class portions spectra for the facility. Using of the PPB, (i.e., this definition, the parameters for structural steel portion the SIS site are Z=0.14 and the of the building and the maximum C is 2.3. It is also PPB change room and load specified that if the value of ZC center area) have no from the 1988 UBC provisions is nuclear safety-class higher, it should be used in the functions or systems. design. Therefore, the Z factor for These structures are Zone 2B of 0.2 and maximum C value categorized as Important of 2.75 are the values to be used in or Low Hazard, rather design of the LSB, DPB and LCB. than General Use, due to their overall importance The 1988 UBC provides a simplified to DOE programs. static approach in place of a dynamic analysis. The use of a o The safety class portion, higher I value, 1.25 for "Important or confinement portion, Facilities" versus 1.0 for "General of the PPB is classified Use Facilities", provides for as a "Moderate Hazard greater protection of the Facility". This structures, systems and components. structure should be This higher importance factor means capable of withstanding a the "Important or Low Hazard Design Basis Earthquake Facilities" will be designed for a (DBE) such that the 25% greater lateral seismic force hazardous materials may than a "General Use Facility" be controlled and located in the same seismic zone. contained. The increased lateral design force provides greater elastic capacity in SEISMIC DESIGN CRITERIA FOR SIS the structure's lateral force STRUCTURES resisting system. This, in turn, provides a higher level of LSB, DPB, AND LCB investment protection,than that provided for a "General Use UCRL-1591O, Section 4.2.2 gives Facility", by reducing the amount of guidance for the seismic analysis potential damage for a given level and design of Important or Low of earthquake. Hazard DOE facilities. In accordance with this section, PPB Important or Low Hazard facilities shall be designed to the seismic Initial DOE Order 6430.1A required provisions of the 1988 UBC [3] for that a safety class structure be Zone 2B with an Importance Factor designed to withstand a DBE and to (I) of 1.25. continue to operate after the occurrence of an OBE. Interim Final UCRL-15910 defines Z as the version of DOE Order 6430.1A, dated seismic zone factor equivalent to 1/9/89, removed the OBE the peak ground acceleration for requirement. Also, DOE Order an annual exceedance probability 6430.1A and UCRL-15910 require that of 1x10"^ and C as the spectral the structural loads for the DBE be amplification at the fundamental determined by a dynamic analysis. frequency of the facility from the

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

222 Therefore, the safety class PPB anticipating licensing concerns, the structure and safety class higher return period of 10,000 years equipment will be designed to a is roughly the upper bound of the DBE utilizing structural dynamic return period that most pressurized analysis in accordance with or boiling water reactors are UCRL-15910 and DOE Order 6430.1A. designed for. The following discussion describes Due to the recent occurrence of the the technical parameters used for 1933 Borah Peak earthquake and the the PPB Design Basis Earthquake potential for similar earthquakes to (DBE) and seismic criteria derived occur closer to the INEL, an from the DOE Order. additional survey of seismic attentuation and site response is Determination of DBE being conducted at INEL. Based on DOE Order 6430.1A, Section the preliminary results of that 0111-99.0.4 states, "To determine survey [5], it was decided to use a the DBE, site specific earthquake peak horizontal ground surface hazard models and response spectra acceleration of 0.33g for the DBE given in UCRL-53582, Rev. 1, shall analysis of the PPB structure. The be used to select the appropriate vertical component is taken as 2/3 seismic ground acceleration." of horizontal (i.e. 0.22g). UCRL-15910 defines the maximum The preliminary survey [5] shows the surface, accelerations at DOE peak ground acceleration (PGA) facilities for various ranging from 0.12 to 0.33 g with the probabilities of annual standard model having PGA of 0.20g. exceedance. It recommends that Consequently the 0.33g PGA selected for Moderate Hazard Facilities a for use in the PPB seismic probability of annual exceedance structural analysis is a of 1x10"^ be used, which gives a conservative number. Should a lower maximum acceleration of 0.14g. It value be determined at a later date also recommends that a vertical by the ongoing survey, the seismic component, equal to 2/3 the analysis will be re-run for the horizontal, be used. purpose of generating the equipment design response spectra within the For conservatism, the recommended building. approach is that the maximum acceleration value for the safety The ground acceleration response class PPB structure will be spectra developed in the preliminary determined using a probability of survey [5] are in general agreement annual exceedance of 1x10 . with the site specific response This corresponds to a return spectra for INEL given in period of 10,000 years. From UCRL-53582. Consequently the Figure 2, "Earthquake Hazard at UCRL-53582 spectra are selected for INEL" (Return Period vs. Peak use in seismic analyses of the PPB. Acceleration), in UCRL-53582, a return period of 10,000 years yields a peak acceleration of 0.24g at the ground surface. From a changing regulatory perspective,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

223 Dynamic Analysis D4428, will be used in this A detailed three dimensional additional geotechnical soil-structure interaction investigation. analysis of the PPB will be performed. Accelerations and Detailed information on the seismic forces in the PPB building will be analysis procedures for the SIS compmed as well as acceleration Plutonium Processing Building is response spectra at selected contained in a paper on the subject locations where safety-related [7] presented at this conference. equipment will be located. These spectra will be used in the design CONCLUSIONS of equipment and piping. The seismic design criteria for the The sub-surface profile at the PPB SIS facilities comply with the DOE site consists of 0 to 7 feet of Order 6430.1A and the related codes predominantly silty or sandy and standards. These criteria are gravel fill overlaying a thin, also in general compliance with the generally cohesive, layer related NRC regulations and consisting of mostly lean clay guidelines. with varying proportions of sand and silt. Beneath this surface REFERENCES layer is sandy, silty or clayey gravel extending to the top of the [1] DOE Order 6430.1A, General basalt bedrock at a depth ranging Design Criteria Manual, from 35 to 45 feet. The April 6, 1989. potentially compressible surface layer of natural cohesive soils [2] UCRL-15910, Design and will be replaced with sandy, silty Evaluation Guidelines for or slightly clayey gravels. Department of Energy Facilities Subjected to Natural Phenomena The site will be modelled as a 45 Hazards, by R. P. Kennedy, ft. layer of gravel overlying a S. A. Short, J. R. McDonald, M. basalt halfspace. The soil W. McCann and R. C. Murray, properties are defined in the SIS April 1988. Geotechnical Report [6]. To account for uncertainties in the [3] Uniform Building Code, 1988 soil properties, the analysis will Edition. be performed using mean, upper-bound (1.5 times the mean) [4] UCRL-53582, Rev. 1, Natural and lower-bound (0.67 times the Phenomena Hazards Modeling mean) soil properties. The Project: Seismic Hazard Models responses for the three subsurface for Department of Energy Sites, conditions will be enveloped. The by D. W. Coats and R. C. response spectra will be widened Murray, November 1984. by ±10% to account for other modelling uncertainties. [5] Preliminary Estimates of Earthquake Strong Ground The shear wave velocities shown in Motions at the Idaho National reference [6] will be confirmed by Engineering Laboratory, by additional geotechnical Woodward-Clyde Consultants, investigation to be conducted at dated April 5, 1989. the SIS site. The crosshole method, in accordance with ASTM

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

224 [6] Report of Geotechnical Evaluation by Northern Engineering and Testing, Inc., Boise, Idaho, February 1987. [7] Seismic Analysis Procedures for the Plutonium Processing Building of The Special Isotope Separation Plant, by C. Chen, et. al. September 1989.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~225 Simplified Seismic Analysis Applied to Structures Systems and Components With Limited Radioactive Inventories~~

~~By O.D. Stevenson Stevenson and Associates 9217 Midwest Avenup Cleveland, OH 44125~~

ABSTRACT This paper presents a review of the current status of simplified methods of seismic design and analysis applicable to nuclear facility structures, systems and components Important to public health and safety. In particular, the International Atomic Energy Agency, IAEA TEC DOC 348 procedure for structures and the Bounding Spectra Concept for equipment as being developed by Seismic Qualification Utility Group and the Electric Power Research Institute will be discussed 1n some detail.

INTRODUCTION o new fuel storage facilities. This paper presents a review of the o spent fuel storage or processing current status of simplified methods of facilities with at least three seismic design and analysis applicable months of decay time. to nuclear facility structures, systems o radwaste processing or storage and components Important to public facilities. health and safety. In particular, the Uranium mines, mills, and mill International Atomic Energy Agency, IAEA tailings deposits are considered outside TEC DOC 348 procedure for structures and the scope of this paper. However, 1t 1s the Bounding Spectra Concept for recognized that mill tailings deposits equipment being developed by Seismic may need some kind of seismic analysis Qualification Utility Group and the similar to those suggested herein. Electric Power Research Institute will The basic philosophy 1n presenting be discussed 1n some detail. the simplified methods contained 1n this The simplified methods and procedures paper Is to minimize sophisticated described 1n this paper are considered calculations and emphasize the Important to be applicable to the following construction and design detailing facilities with limited radioactive procedures which generally govern Inventories or risks. overall seismic safety. Rational o research reactors of up to a few seismic design should also consider the HW(th) continuous power. relative seismic ruggedness of o hot cells with a small Inventory Industrial process systems and of radioactive materials. components as compared to the structures o uranium or plutonium fuel which house or support them. Based on fabricating and purification strong motion earthquake damage plants. experience, process equipment In general

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

226 1s more seismically rugged than the seismic Induced failure building structure which houses or of a nominal 600MWe support them assuming neither has been nuclear power plant. designed to be earthquake resistant. Class II Items that can deform However, for specific structural 1nelast1cally to a arrangements, the simplified approach moderate extent without may not be valid. In cases where a loss of function and large safety margin 1s required for also Includes seismic nuclear public health and safety reasons category I piping and a more complete and refined approach, duct systems exclusive comparable to that presented by current of active valves. applicable USNRC Regulatory Guides and Structures and Standard Review Plans for power reactors structural elements critical structures may be within major safety r equ1 red .C»2»3,4,5,6,7) related structures which do not form part of the Seismic Classification primary load carrying The nuclear facilities with limited path and structures radioactive Inventory are classified as housing Hems of Class I A, B, and C and their structures and or I-S that must not be components are further categorized to permitted to cause take Into account the consequences of damage to such Items by failure. Nuclear power piantsP] and excessive deformation of facilities at nuclear plants with the structure. limited radiological consequences of Ductility factor, v = failure are classified as II and III. 2.0 for primary load Class definitions are as follows: path structures Class III Facilities which may subjected to significant and Class C be designed in P - A effects, shear and non compact section accordance with the bending modes of failure normal building or other and >i = 3.0 for conventional Industrial primary load paths codes. compact section bending Class B Facilities for which the and membrane tension following should be modes of failure and all ensured. C1v1l non-primary load path structures will not structures. collapse, the pool and containment or The alphabetic classification are confinement structures associated with structures in facilities will preserve normal with limited radiological Inventories leak tightness and no and consequences. The Roman numeral disruption of nuclear classification 1s for structures at fuel or core resulting large nuclear power plants as shown in from falling debris (1f Table 1. In general simplified seismic this can result in design and analysis methods would not be critically or melting of used for Class I and I-S structures at the fuel). nuclear power plants. Class A Facilities for which the consequences of seismic induced failure will be [1] See Table 1 for the definition of two orders of magnitude more critical Class I and I-S or less than would be facilities located at Nuclear Power determined from the Plants. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

227 For facilities of Class A or 8, a IAEA-TECDOC-348 for the International further categorization of Hems 1s made Atomic Energy Agency prepared by Watabe a follows: et. al.<12) The ASCE Committee 1s currently reviewing both publications, Category (a) Items whose safety and 1t 1s assumed the ASCE committee functions must be will make use of concepts and procedures maintained 1n the 1n both documents 1n developing Us event of the Design guidelines. Bails Earthquake Since most attendees of this conference are more familiar with the Category (b) items whose loss of U.S. Department of Energy Guidelines, function may be this paper will concentrate on permitted but shall presenting the IAEA TEC DOC 348 be designed against simplified procedure for building collapse 1n the design. The more rigorous dynamic time event of the Design history and response spectrum modal Basis Earthquake analysis methods of seismic design and analysis applicable to nuclear Non-Categor1zed: All other Items facilities are well documented 1n the literature*1* 13) and win not be Category (a) only 1s assumed discussed further 1n this paper. applicable to Class II facilities as Simplified methods of seismic defined herein. building analysis have been directed If the behavior of a Category b or primarily toward nuclear facilities with non-Categorized Items results 1n the limited radioactive Inventories. loss of function of Category (a) or Category (b) Items respectively, these Seismic Design Items should either be categorized 1n The earthquake resistant design of category a and category b respectively building structure should be performed or the potential of damage should be according to the previous defined appropriately evaluated and reduced by classification as follows: change of design or location. Class C and III Uniform Building SEISMIC DESIGN OF BUILDING STRUCTURES Code with 1=1.25 With regard to building structures, the Dynamic Analysis Subcommittee of the Class B: Simplified Nuclear Structures and Materials Equivalent Static Committee of the Structural Division, Approach may be used. American Society of Civil Engineers, Is developing guidelines to simplified Class A and II Simplified Dynamic seismic design of nuclear facilities Approach should be with limited radioactive Inventories used. (research reactors, spent fuel storage, radwaste processing and storage The story lateral load, 1s given by: facilities etc.). In performing this task, the Committee essentially has Fi = a x C, x (1) selected two existing guidelines to g choose from, l)Des1gn and Evaluation where: Guidelines for Department of Energy Facilities Subject to Natural Phenomena = design acceleration Hazards. UCRL-15910 prepared by Kennedy g applicable at the ground et. al.( ) and 2)Earthquake Resistant surface level. The value Design of Nuclear Facilities with of a may be modified as limited Radioactive Inventory, g

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

228 a function of foundation K-): Standard Response Factor to be soil at categorization, determined from F1g. 2 as shown In Table 2 where Kg: Coefficient related to damping ag * J* 2 and Z 1s to be determined from Tables 5 defined 1n Table 3 and and 6. Figure 1. The load distribution throughout the Cis or ^01: coefficient height of the building 1s given by S^ related to the building, as defined 1n Table 4. characteristics and defined In order to use Figure 4 to determine by Eqs (3) and (4) for the the Standard Response Factor, the simplified dynamic and fundamental period of the structure, equivalent static Tt, may be calculated using, 1/2 approaches respectively TD = d HB (0B)- (5) the total weight of the where: building at 1-th floor to Hg = height of the building, (m) be calculated by, DB = dimension of the building 1n a direction parallel to the 0.25 (2) applied lateral loads, (m) d * 0.10 for rigid Reinforced where, D^ and l\ are the design dead Concrete structures, and and live loads for the 1-th floor, equal 0.12 for ordinary respectively Reinforced Concrete Simplified Equivalent Static Approach structures. The seismic coefficient C^s (the If the fundamental period of the additional subscript s 1s Introduced to structure, TD, 1s shorter than 0.3 denote "static" approach) of Equation 1 seconds, the product K-j K2 should is defined as follows: not be less than 1.5. SIMPLIFIED SEISMIC DESIGN OF INDUSTRIAL (3) PROCESS SYSTEMS AND COMPONENTS It 1s Interesting to note that the current 1988 and past editions of the where: Uniform Building Code and ANSI A 58.1 the coefficient giving the StandardO*> 15) have had provisions distribution of the seismic for seismic design of attachments and coefficient throughout the some equipment located In or supported height of the building as by conventional building structures. defined 1n Table 4. The codes developed for design of conventional Industrial process systems y = the ductility coefficient and components (ASME B31-Series, ASME defined 1n Table 1 Boiler and Pressure Vessel Code and IEEE etc.) have generally been silent on the Simplified Dynamic Approach subject of seismic design leaving It up The seismic coefficient ZQ\ (the to the Engineer to decide how seismic additional subscript D 1s Introduced to phenomena will be treated or make denote "dynamic" approach) of Eq. 1 1s reference to either the Uniform Building defined as: Code or ANSI A58.1 for guidance. In the past, this has generally lead to very CD1 (4) For Uttle explicit seismic design of Industrial process systems and TD < 0.1 s; =1 0.1s < T < 0.5 s; components outside regions of recognized D = (2y high se1smU1ty such as California. 0.5s < TD where y 1s defined 1n Table 1 and Over the past 10 years various Federal

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

229 Guidelines have been developed 1n this the conservative threshold of damage to area which are available for use 1n new the plant. Such damage would consist of construction of Federal Buildings.(8» 9) unanchored narrow width high aspect However, 1t 1s not clear 1f or to what ratio cabinet overturning, some extent this Guidelines have been applied dislocation of hung celling tiles, to design of equipment within buildings cracking of ceramic Insulators, or other Industrial process systems and activation of unfiltered vibration alarm components. It 1s generally sensors and tripping of vibration acknowledged that the use of the methods sensitive relays. of seismic design of systems and components based on the Uniform Building Bounding Spectra Method Code or Inclusion of a seismic design or Seismic qualification of Industrial performance requirements 1n procurement process systems and components described specifications for Industrial process above 1s referred to a the Bounding systems or components significantly Spectrum Methods. This method 1s under Increases the cost of such components active development by the Electric Power and systems. The basic question that Research Institute for potential must be addressed 1s as follows: application to nuclear power plant piping 1n regions of low to moderate Is the additional cost of seismic seismicHy. analysis design and qualification of The purpose of this technique Is to process systems and components provide a rational and consistent basis justified by the risk associated with for the seismic evaluation of Industrial potential seismic Induced damage or process systems and components without failure of such systems? recourse to explicit seismic stress analysis or testing. The technique has Since 1982, there have been major been developed consistent with the research activities sponsored by the observed behavior of equipment Seismic Qualification Utility Group, Industrial facilities and thermal power SQUGOQ) and the Electric Power plants subjected to strong motion Research InstUuteO1) and the U.S. 6 earthquakes. Nuclear Regulatory Comm1ss1on0 ) to The technique Involves the definition answer this question. The answer to of a Bounding Spectrum and Caveats this question, except for regions of governing the construction of process recognized relatively high seismidty system and components. If the Caveats with zero period ground design are met and the Bounding Spectrum accelerations equal to or greater than envelops the seismic design basis ground about 0.33g[2] as defined by the response spectrum for the site, then the Uniform Building Code 1s a qualified equipment may be considered qualified no. In Table 7 1s a summary of failures for seismic Inertia loads without and significant damage to piping systems further evaluation. 1n large number of Industrial facilities during a large number of strong motion A threshold of significant damage earthquakes. Similar Tables are Bounding Spectrum based on the averaging available for other Industrial of response spectra data from Industrial equipment.(10) The qualification to facilities and power plants which have the "no" 1n answer to the basic question been subjected to earthquakes of Richter 1s that certain specific mitigating design and construction details need to be employed to develop this level of [2] For nuclear power plant, seismic and construction ruggedness. conservative ground response spectra Using normal design and practices, are defined which effectively without any consideration of ductile reduces this ZPGA threshold of design, a ZPGA of about 0.12 g would be damage to about 0.25g.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

230 magnitude of about 6.0 to 7.8, and site structures at nuclear facilities with peak ground acceleration equal to or limited radioactivity has been presented greater than 0.4g has been developed as as a possible alternative or basis to shown 1n Figure 3. modify recently developed DOE Observations made at these sites and guidelines. In addition, a simplified other power and Industrial plant sites non-analytical procedure has been which were subjected to strong motion suggested for seismic qualification of earthquakes Indicate there would be no equipment 1n buildings and Industrial significant damage or failure provided and process systems and components. certain Caveats as determined by It has been estimated that the experience were met 1n the construction application of current Uniform 8u1ld1ng of these process systems and components. Code requirements to seismic design of An example of a group of caveats buildings and qualification of equipment which might be applied to Industrial or attachments 1n building 1n California piping 1n order to seismically qualify adds about 8 percent to the overall cost such piping without explicit analysis or of the facility with about two thirds of test might be as follows: this Increase attributed to equipment o Lateral supports or restraints are seismic design and qualification. In placed at a maximum of 5 times region of low to moderate seismidty, deadweight span lengths, the total cost Increase should be o Analysis to determine effect of approximately 3 to 5 percent. The significant differential seismic Increase 1n engineering cost 1n both (eg > + 2.0 Inch) anchor motion 1s cases 1s about 10 to 15 percent. performed. This contrasts with a nuclear power o Equipment to which piping 1s plant where seismic design at a nominal attached Is anchored and designed moderate se1sm1dty 0.2g ZPGA site adds to carry seismic loads, about 12 percent to overall plant costs o Field walkdown of piping 1s and Increases engineering costs by about performed to Identify potential 20 percent. Approximately 10 percent of for seismic Interaction (Impact) this overall plant cost Increase 1s with other equipment or structures associated with equipment seismic design o Piping connections are limited to and qualification. The engineering cost welded connections attributable to equipment seismic design o Piping and piping support material and qualification 1s about 18 percent. limited to ductile weldable steels Application of Bounding use of Spectra o Piping connections to equipment Method technique for seismic are made as flexible as practical qualification of equipment would (should accommodate at least a ± dramatically reduce the costs of such 2.0 Inch displacement) equipment qualification 1n the range of Similar caveats can be developed a factor of 2 or 3. applicable to other classes of equipment (pumps, valves, tanks, vessels, heat REFERENCES exchanges, switch gear, relays, motor (1) Nuclear Regulatory Commission control centers etc.) Using the "Standard Review Plan" Section 3.7 Bounding Spectra Method, seismic -Seismic Design NUREG-0800 July resistant designs can be developed for 1981. low to moderate seismicity sites (2PGA < (2) Nuclear Regulatory Commission without recourse to expensive, time "Standard Review Plan Section 3.8 consuming seismic engineering analysis -Standard Design NUREG-800 July or test qualification. 1981. CONCLUSIONS AND SUMMARY (3) Nuclear Regulatory Commission A simplified seismic analysis and "Standard Review Plan Section 3.9 design procedure for safety related

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

231 -Standard Design NUn:G-800 July Inventory," IAEA- TEC00C-348, 1981, International Atomic Energy Agency, (4) Regulatory Guide 1.60 "Design 1985. Response Spectra for Seismic (13) ASCE 4-86, "Seismic Analysis of Designof Nuclear Power Plants," Safety-Related Nuclear Structures U.S. Atomic Energy Commission, Rev. and Commentary on Standard for 1 Dec. IV73. Seismic Analysis of Safety Related (5) Regulatory Guide 1.61 "Damping Nuclear Structures," American Values for Se1sr-1c Design of Society of C1v1l Engineers, Sept. Nuclear Power Plants," U.S. Atomic 1986. Energy Commission, Oct. 1973. (14) "Uniform Building Code 1988 (6) Regulatory (Vilde 1.92 "Combining Edition," International Conference Modal Responses and Spatial of Building Officials, 1988. Components 1n Seismic Response (15) ANSI A58.1-1982 "Minimum Design Analysis," U.S. Nuclear Regulatory Loads for Buildings and Other Commission, Rev. 1 Feb. 1976. Structures," National Bureau of (7) Regulatory Guide 1.122, Standards, 1982. "Development of Floor Design (16) Seismic Design Task Group, "Summary Response Spectra for Seismic Design and Evaluation of Historical of Floor-Supported Equipment or Strong-Motion Earthquake Seismic Components," U.S. Nuclear Response and Damage to Above Ground Regulatory Commission, Rev. 1 Feb. Industrial Piping," NUREG-1061 Vol. 1978. 2 Addendum, U.S. Nuclear Regulatory (8) Interagency Committee on Seismic Commission, April 1985. Safety 1n Construction, "Seismic Design Guidelines for Federal Buildings," NBSIR 87-3524-ICSSCRP-l, National Bureau of Standards, Feb. 1987. (9) Kennedy R.P. et. al. "Design and Evaluation Guidelines for Department of Energy Facilities Subjected to Natural Phenomena Hazards," UCRL-15910 Prepared for Office of Nuclear Safety USDOE by Lawrence Livermore National Laboratory, April 1988. (10) EQE Inc., "Program for the Development of an Alternative Approach to Seismic Equipment Qualification Volume I: Pilot Program Report," Prepared for Seismic Qualification Utilities Group, 1982. (11) EQE Inc., "Recommended Piping Seismic-Adequacy Criteria Based on Performance During and After Earthquake, Vol. 1: Summary Document," EPRI NP-5617, Electric Power Research Institute, Jan. 1988. (12) Advisory Group on Earthquake Resistant Design of Nuclear Facilities with Limited Radioactive

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 232 Tttlt Mcomeended Seismic Design dloOil Ouctilltv Table 4 taiMfication Coefficient ai < function or lulleine Height tor nuclear Power Plant Structures 4119 Component) cuss DESCRIPTION luildini level 5, (1) I-S Dm which Include structures, equipment. Instruments, or components performing vital function* that must remain operative Embedded fart 0.7} (potentially move or change state) during 4nd after earthquakes. Item (hit are subject to nottducttle, Including elastic Ouckling and brittle fracture modes of failure. Ductility factors, » • 1.0. Ground level 1.00 I Urn? that must remain operative after an earthquake out need not MOWI or chino,t tt4tt durtnq tht tvtnt. Major saftty rtUttd nuclur sow*r pUnt structurii. Ductility f«ctor, » • 1.1 for imlolr.9 Center of Cravity Level 1.S0 1nelnt1c Duelling, shiar inn non cwmict mtion bending mdts of fillur* 0.75. trays, conduit) and all other Items vnich are usually governed by ordinary seismic design codes. Structures requiring seismic resistance In order Co be repairable after an earthquake. Ductility factor. » • } to S, depending on material, type of construction, f - 1 effects, design of Joint details, »nd control of quality.

~~[1) Revises version of Table given In Refs. 3 and < Table 5 Percentage of Critical Damping.~~

~~Type of Structure Percentage of I (X) critical damglng~~

~~riveted steel structure i~~

~~Table 2 Oefmition of X Factor as a Function of Soil Category Mtded steel structure 3~~

~~reinforced concrete or steel concrete structure S Soil Categorization Category 1 Category Category 3~~

or steel concrete structure with shear wall Other soil F1U ground Firm Searing Than those or Alluvium Strata defined In ground which Categories Is thicker These values may be Increased by 2X and ix for Category 2 and Category 3 1 and 3 than 25 m. soils (Taele t) respectively, 1n order to take Into account the additional damping provided by loll.

~~Table i Darning coefficient as a Function of Percent Critical Damgtng~~

~~ing coefficient Table 3 Seismic 2one Factor Z 0.5 1.0 2.0 5-0 M> ».O 10.0 12.0 13.0 1S.0~~

~~lone 1 2> il 3 4 «2 1.7J 1.55 1.32 I.U 1.00 0.11 O.H 0,!| 0.M O.1J 0.7 Z 0.075 O.H 0.20 0.30 o.to~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~233 T«»lt 1 Ho\ng OiMgt *Mur« in >n>tr >l~~

~~CitMOU Oinuot ». Piping • «bov» ground StllMVc Uuttt llnchor Nevmnt IS 0 C 10 Mn.atl^M Joints (Priairtl) ThrtidH) 31 10 _1_ Totil Piping ^^~~

~~C. «nttrnil i 15~~

~~CstHutta totjl «»ount of Siloing it rijt: UO0OO0 ft. &«1tO on « totjl of 40 glints jno in ivtrt e Of 20000 f| of~~

~~EltKu'.M totil n~~

~~Figure 1 Seismic Zone Acceleration Hap of the United States~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~234 Bofl MM) !«• 1 • Jl .::~~

~~- -V •• 4. .w -v, i . . , i~~

~~1 ! i i '< ' • ; 1 i i! i i; A y^ri " -•>y r •••• 1! •» \P1\x ; i i :~..:V. hr jtiiF; r-t- ! ": ! i t;HP 1'T \: 1 ! i !;H; i i j | ! ; UgtnL d \ X a 1 P«rto4 of Scruccuc* \t*)~~

~~RC. 3 3E37CNSE OJUVSS FCt OIFRSENT INTENSTiSS ANO SOIL CATEGORIES~~

~~I I ( 3 u~~

OAOUNS ACttHU 1UT10N • 0J3 f*

~~• .* ^ 12 ii as~~

~~K 0W5 •SUOOn HOAHSNTAL CHOUNO HO1ICH~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~235 Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~236 Session 8 Seismic Hazard~~

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

237 PROBABILISTIC ASSESSMENT OF THE SEISMIC HAZARD FOR EASTERN UNITED STATES NUCLEAR POWER PLANTS*

~~Jean Savy, Don Bernreuter, and Richard Mensing Lawrence Livermore National Laboratory 7000 East Avenue Livermore, California 94550~~

ABSTRACT The purpose of the seismic hazard characterization of the Eastern United States project, for the Nuclear Regulatory Commission, was to develop a methodology and data bases to estimate the seismic hazard at all the plant sites east of the Rocky Mountains. A summary of important conclusions reached in this multi year study is presented in this paper. The magnitude and role of the uncertainty in the hazard estimates is emphasized in regard of the intended final use of the results.

INTRODUCTION 1. To develop a seismic hazard characterization methodology for the entire region of the United The impetus for this study came from two States east of the Rocky Mountains. unrelated needs of the Nuclear Regulatory Commission (NRC). One stimulus arose from the 2. To apply the methodology to selected sites to NRC funded "Seismic Safety Margins Research assist the NRC staff in their assessment of the Programs" (SSMRP). The SSMRP's task of implications in the clarification of the USGS simplified methods needed to have available data position on the Charleston earthquake, and the and analysis software necessary to compute the implications of the occurrence of the recent seismic hazard at any site located east of the Rocky earthquakes such as that which occurred in Mountains which we refer to as the Eastern United New Brunswick, Canada, in 1982. States (EUS) in a form suitable for use in probabilistic risk assessment (PRA). The second The methodology used in that 1985 study evolved stimulus was the result of the NRC's discussions from two earlier studies that the Lawrence with the U.S. Geological Survey (USGS) Livermore National Laboratory (LLNL) performed regarding the USGS's proposed clarification of for the NRC. One study, [2], was part of the their past position with respect to the 1886 NRC's Systematic Evaluation Program (SEP) and Charleston earthquake. is simply referred hereafter to as the SEP study. The other study was part of the SSMRP. The results of this study were presented in Eemreuter et al., [1], and the objectives were: At the time (1980-1985), an improved hazard analysis methodology and EUS seismicity and ground motion data set were required for several reasons:

~~•This work was performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under Contract W-7405-Eng-48.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

238 In addition, we critically reviewed our • Although the entire EUS was considered at methodology which lead to minor improvements the time of the SEP study, attention was and we also provided an extensive account of focused on the areas around the SEP sites— documentation on the ways the experts interpreted mainly in the Central United States (CUS) our questionnaires and how they developed their and New England. The zonation of other answers. Some of the improvements were areas was not performed with the same level necessitated by the recognition of the fact that the of detail. results of our study will be used, together with results from other studies such as the EPRI study • The peer review process, both by our Peer or the USGS study, to evaluate the relative hazard Review Panel and other reviewers, identified between the different plant sites in the EUS. some areas of possible improvements in the SEP methodology. METHODOLOGY • Since the SEP zonations were provided by The methodology used in this study is developed our EUS Seismicity Panel in early 1979, a around three basic elements: number of important studies had been completed and several significant EUS 1. The estimation of the seismic hazard (the earthquakes had occurred which could impact hazard model) is analytically defined by the the Panel members' understanding of the now classical Cornell model [1,3,4,5, 6]. seismotectonics of the EUS. The various elements, seismicity and ground • Our understanding of the EUS ground motion attenuation are expressed separately and motion had improved since the time the SEP integrated to provide an estimate of the study was performed. probability of exceedance of the peak ground acceleration (PGA) and of the 5% damping By the time our methodology was firmed up, the spectral velocity at five frequencies. expert opinions collected and the calculations performed (i.e. by 1985), the Electric Power 2. It is recognized that every element of the hazard Research Institute (EPRI) had embarked on a modeling is subject to uncertainties. The parallel study. random (or physical) uncertainties in the prediction of the ground motion are analytically We performed a comparative study, [3] to help in accounted for in the hazard model. Other understanding the reasons for differences in results uncertainties, random and model uncertainties, between the LLNL and the EPRI studies. The are propagated in the analysis by means of a three main differences were found to be: (1) the Monte Carlo simulation technique. minimum magnitude value of the earthquakes contributing to the hazard in the EUS, (2) the The uncertainty in the seismicity distribution is ground motion attenuation models, and (3) the fact accounted for by generating a large set of that LLNL accounted for local site characteristics zonation maps and associated seismicity and EPRI did not. Several years passed between parameters (a- and b- values, upper magnitude the 1985 study and the application of the cutoffs) and the uncertainty in the ground methodology to all the sites in the EUS. In motion modeling is accounted for by using a recognition of the fact that during that time a set of 11 alternative models for the attenuation considerable amount of research in seismotectonics of the peak ground acceleration (PGA) and for and in the field of strong ground motion prediction, the attenuation of the spectral velocity at the in particular with the development of the so called five frequencies. random vibration or stochastic approach, NRC decided to follow our recommendations and have a The local soil site conditions at each site were final round of feedback with all our experts prior to acknowledged and each time a site correction finalizing the input to the analysis. factor was used, it was described by a

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 239 probability distribution function to model the These three elements, seismic hazard model, use of uncertainty found in site amplification data. expert opinion and propagation of the uncertainty form the basis of our methodology. Other methods In the Monte Carlo technique, each hazard available today, may vary in several respects. For estimate (simulation) is done with a fixed value example, many seismic hazard analyses do not of the parameters. (A single zonation map, recognize the variability in the estimates due to the ground model, a single pair of a and b-values source of the input (i.e. they may use only one for each zone of the zonation map, a single expert to define the seismicity or the ground motion value for the ground motion uncertainty-sigma, model). Some do not account for the uncertainty in and a single value for the site correction factor). the many uncertain parameters in such an analysis The process is repeated many times (2750 (e.g., a and b-values, upper magnitude cutoff, etc). times in our analysis for PGA and 1650 for spectra), and in each simulation a new value of The method developed by the Electric Power the uncertain parameters is drawn from their Research Institute (EPRI) [7] has many common respective probability distribution. elements to ours. The basic hazard model is the same, the input is obtained through the elicitation The result of this experiment is a large set of of expert's opinion and all types of possible artificial estimates of the seismic hazard from uncertainties are recognized including the which the 50th (median), 15th and 85th variability in the expert opinions. The overall percentile hazard curves are calculated to uncertainty is propagated by the means of the represent the central tendency (median) and enumeration method where all the possible total uncertainty (random and modeling) in the combinations of parameters are considered, in seismic hazard. contrast to using a Monte Carlo method which selects alternatives at random from known 3. This study was intended to represent the probability distributions. opinion of the scientific community with respect to seismic hazard estimation. To this GROUND MOTION MODELING effect, two panels of experts were formed. The seismicity panel, composed of 11 experts (S- The seismic hazard model requires an estimate of experts) in the field of seismicity of the EUS, the probability of exceedance of a given level of and the ground motion panel, composed of five ground motion (i.e., PGA or PSRV), for an experts (G-experts), represent a cross section earthquake with known magnitude and location. of the various schools of thoughts and opinions We divided the problem into two independent currently present in the scientific community. ones. The opinion of each of these experts was elicited via questionnaires to develop the input 1. For a known earthquake (i.e., magnitude and necessary to the hazard model. distance from the earthquake source to the site considered), and assuming generic soil type at The questionnaires were such that they enabled the site (i.e., rock or deep soil), we estimate the the experts to express their opinion as to which ground motion at that site. This is usually models, or parameter values were the most referred to as the ground motion attenuation likely to represent the true state of nature. In modeling. addition, whenever necessary, the experts had to describe their opinion on the uncertainty 2. If the subsoil conditions at the actual site are about the parameter value they selected in a different from the generic conditions, we apply qualitative and quantitative fashion. a correction to the estimate of the ground motion. This part is usually referred to as the Several feed back meetings were organized to site soil correction. ensure the experts opinions were interpreted correctly and to give the experts an opportunity For smaller studies, (i.e. when studying only one to critically review their answers to the site), the above two items need not be separated. questionnaires and eventually to modify them. Actually, with the necessary resources of time and effort, a site specific type of modeling is almost

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 240 always preferable to a generic type of approach. In our experience, the improvements reside more in the amount of uncertainty than in the actual values of the estimates. For example, the median hazard estimates of the site generic case may be only a few percent away from the specific case, but the 15th and 85th percentile curves may show much wider uncertainties. In this section, we review the attenuation part of the ground motion modeling. The next section is concerned with the site l0 correction aspect r Consistent with the philosophy of our methodology, the ground motion model input was developed by elicitation of the G-cxperts opinion, with two rounds of feed back. The intent was not to obtain what some would be tempted to call "The Model", but rather to sample the experts to ensure that all the models that the experts deemed rational and possible be considered. Each expert was free to select as many models as he wished and assign whatever weight he wanted to each one of them. OISTANCC-HM The total weight of the models for a given expert Figure 1 Best estimate PGA models plotted for was unity, and the weight of each expert was the magnitudes of 5 and 7. Rock base case: normalized self weight given by the expert himself (the self weights were roughly 1/5 for each of the five G-experts).

The final set of attenuation models selected by the experts includes a range of available models including the empirical, intensity based models (Veneziano [8], Trifunac model [9], the empirical model of Nuttli [1], and the theoretical models of the Atkinson type [10] also called random vibration models (RV models). The difference between the various types of models is shown in Figs. 1 and 2 for the attenuation of the PGA for two events, of magnitude 5 and 7. Figure 1 shows the best estimate models for each of the 5 G-experts for a rock site (i.e. the model which they believed were the most appropriate to represent the median PGA for a given magnitude and distance). Figure 2 shows the additional models selected by the experts to express their uncertainty in the models.

~~DISTMICC-IU Figure 2 Remaining PGA models plotted for magnitudes of 5 and 7. Rock base case.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

241 Examination of these two figures immediately In spite of these drastic changes in the ground draws several conclusions. motion models from our previous study on 10 test sites in the EUS [6], and some changes in the 1. The dispersion betwe< a the models is large, seismicity modeling, the estimates of the seismic approximately a factor of 10, in the range of hazard in terms of exceedance of PGA did not distances of 50km to 200km. drastically change, as can be seen in Figs. 4 and 5. 2. The dispersion is much less within 50km. O o Previous Results -2 3. The RV models are much lower than the other 10 Updated Results models. As much as a factor of 10 for distances greater than 100km, at both magnitudes 5 and 7. 4. Model number 3 in Fig. 1 (Trifunac's model), is higher than the rest of the models by a factor of approximately two, up to 200km for magnitude 7. However, Fig. 3 shows that the difference is much smaller for the deep soil case.

~~Z ACCELERATION CM/SCC"2~~

BRA IDWOOD Figure 4 Comparison of the 15th, 50th and 85th percentile CPHCs aggregated over all S and G- Experts between the new input and the previous input from the S and G-Experts for the Braidwood site.

The ground motion attenuation models of the response spectral velocity, in addition to being of the three types, empirical, semi-empirical and theoretical, could be either based on original spectral shapes (the RV models, and Trifunac's model), or Newmark-Hall type models constructed DIST»NC€-KM with PGA and velocity models of the types Figure 3 Best estimate PGA models corrected to described for PGA: Fig. 6 shows the best estimate generic deep soil for magnitudes of 5 and 7. spectral models for the five G-experts for the rock case. Model 1 of Fig. 6, is a "pure" RV model, and model 3 is a Newmark-Hall model [11] This selection of models is somewhat of a constructed with RV models of PGA and velocity. departure from our previous analyses, where the Model 2 is the Trifunac model, and models 4 and 5 RV models were not used. In this analysis, the RV are Newmark-Hall models based on semi-empirical models account for approximately 50%, Nuttli's relationships of PGA and velocity attenuation model 20%, Trifunac model 20% and Veneziano's models. model 10%.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 242 0 0 Previous Results Updated Results In toto, the contribution from RV models was approximately 53% (including 44% of "pure" RV models and 9% of Newmark-Hall-RV models), 27% of Newmark-Hall-semi-empirical models, and 10 20% for the Lee and Trifunac model [12]. In other words, the pure RV models accounted for 44%, Newmark-Hall models 36% and Lee and Trifunac, 20%. Figure 6 shows the difference in behavior between the various spectral models. These

1Q median models are defined at five periods (0.04, 0.1, 0.2,0.4, and 1.0 second). Examination of Fig. 6 leads to the following immediate -« 10 conclusions: 1. The variability for 0.2,0.4, and 1.0 sec. is much greater than for 0.04 and 0.1 sec, with ACCELERATION CM/SCC2 the variability being the greatest at 1.0 sec.

MILLSTONE This is due mostly to the vastly different Figure S Comparison of the 15th, 50th and 85th behavior of the "pure" RV model (model percentile CPHCs aggregated overall S and G- number 1 in Fig. 6) by comparison to the Experts between the new input and the previous Newmark-Hall type model. input from the S and G-Expens for the Millstone site. 2. At low period (0.04 sec), the "pure" RV model is higher than the other models, especially for the higher magnitudes, and it is substantially lower at higher periods. Thus, when comparing previous results to the present study, the present estimates show a drastic change in the estimated spectral shape, with relatively much higher levels at low period and much lower levels at higher periods. Figure 7 shows this "flattening" effect for a typical site in the North Central region of the EUS. SITE CORRECTION The ground motion attenuation models selected by the G-experts were derived either for rock site conditions or deep soil. The Trifunac model was derived for rock, deep soil or some other category which the author calls intermediate. In each trial, -I 10 the Monte-Carlo simulation process selects at • • raioo (xc) 2 random one of the ground motion models, at a rate proportional to the weights assigned by the Figure 6 Best estimate 5 percent damped relative experts. Thus, each time the ground motion model velocity spectra models listed in Table 3.6 plotted selected did not match the soil site conditions, a for magnitudes of 5 and 7 at a distance of 25 km. correction was applied. The opinions of the Rock base case.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

243 The simple correction consists in applying a fixed multiplication factor to the median attenuation ooo Provtous Results curve. By contrast, the correction applied in the Updated Results categorical correction has some variability. This variability was described by a probability distribution function from which the correction factors were drawn in each of the Monte Carlo trials. In terms of interpretation, and usage, of the 0 <» final seismic hazard results, there is a fundamental difference between the two methods of correction. One, the simple correction, deterministic, does not account for the uncertainty innerent in predicting the amplification at a shallow soil site, the other does (Probabilistic case). In the deterministic case, the hazard curve derived for a soil site would be exactly the hazard curve derived for a rock site at the same location '2 '2 PERIOD (SEC) 2 2 multiplied (in the PGA axis direction) by the 8RAI0W000 correction factor from rock to soil. Figure 7 Comparison of the 1000 year return This is not true in the probabilistic case where period 15th, 50th and 85th percentile CPUHS for 5 several random variables with various symmetrical percent damping aggregated over all S and G- or unsymmetrical probability density functions. Experts between the new input and our previous With the choices of probability density functions in input from the S and G-Experts for the Millstone our analysis, the probabilistic correction leads to a site. median hazard curve slightly (up to 10% in PGA) lower than if a deterministic case with correction factor equal to the median of the probabilistic experts were elicited to define what type of distribution of the probabilistic correction factor corrections and how the corrections should be were used. In addition, the uncertainty bounds on applied. In the end, two types were selected. the total hazard are larger. The 85th percentile hazard curve can be up to two times higher (in 1. The simple correction approximates the soil PGA values) than the deterministic case. sites conditions to only two generic categories, either rock or soil, or three categories in the The conclusion is that it is appropriate to perform case of the Trifunac's model. the correction on the hazard curve if one believes that the correction factor is a fixed value, known 2. The categorical correction differentiates all of without uncertainty. It is not appropriate to the EUS sites into eight different generic perform such a correction on the hazard curve if categories. Rock, deep soil, shallow till-like the correction factor is known with uncertainty. soil (three different depths) and shallow sand- like soil (three different depths). Thus, in our analysis where one-fifth of the corrections were deterministic and four-fifth were The experts were also given the opportunity to probabilistic, it would not be appropriate to apply no correction at all, if they did not believe perform the site correction on the hazard curve. that the data available showed any definite trend depending on the site soil type. This latter One interesting consequence of having several alternative was not selected by any of the experts. types of correction factors is that it makes it Four of the G-experts opted for the categorical regionally dependent, due to the complex (eight categories) correction method and one G- interaction between zonation and seismicity-ground expert opted for the simple correction. motion and site correction. The reader is referred to Bernreuter et al., Vol. 6 [1] for more details on

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

244 the mechanics of this regional dependency. Table • For the shallow soil case, the large, distant 1 gives a summary of the site correction results for earthquakes are also dominant, and G-experts 12 sites in the EUS. In this table, the soil category 5's model falls within the cluster of other is one of the eight categories discussed earlier in models, thus, the median will be this section, columns 1,2, and 3 are the ratios of representative of all the models, and in PGA values, read on the median hazard curves, at particular again close to the median without fixed probability of exceedance values (10"3,10"4, Expert 5. and 10"5) between the soil and rock cases. Column (4) is an average of columns 1, 2, and 3. Column The result is that the final ratio of PGA between (6) is the deterministic correction factor for the shallow and rock cases for these three sites is close Trifunac model and column (7) is the median of the to the case when only the categorized correction is probabilistic correction factor used in the used (i.e., the correction recommended by all but categorical correction method. Column (5) is a G-expert 5). weighted average of columns (6) and (7). If there were no regional effects, we would expect the This discussion indicates that in general the hazard actual correction factors (i.?. column 4), to be close curve computed for shallow soil is close (here, to the weighted averages (i.e. column 5). The never more than 10% higher in the direction deviation from the value in column (5) is due to the parallel to the ground motion (PGA) axis and more complex interaction between ground motion often within 1 or 2%) to what would be obtained models, seismicity zones and seismicity by simply applying a median correction factor to parameters. Depending on all those factors the the hazard curve for rock. However, the impact will be that the correction advocated by G- complexity of the process makes it very difficult to expert 5 will have more or less weight, relative to isolate the parameters which make this ratio deviate the other 4 experts. For Oconee, the combination at some sites from the expected ratio and thus of the above mentioned interactions leads to an makes it, for the time being, impossible to predict. impact of G-expert five greater than the equal weight case. For the other sites, but Three Mile Thus, when using the results of our analysis, it is Island, the effect is reversed and the impact of incorrect, even approximately, to routinely correct G-experts 5 is diluted. a rock hazard curve to get an estimate of a soil hazard curve, when a combination of site The case of Arkansas, Callaway and Duane Arnold corrections are used, in addition the site correction requires additional scrutiny. For those three sites, is in effect region dependent. Table 1 shows that the effective amplification factors (column (4)) obtained in our simulation are USE OF THE RESULTS OF THE close to the case when G-expert 5's model is not SEISMIC HAZARD ANALYSIS used (compare columns 4 and 7). This phenomenon seems extreme and can be explained The results of a seismic hazard analysis can be as follows, (remembering that we are comparing used in a variety of ways either in a relative or median hazard curves for rock and for soil): absolute sense.

• For the rock case, the contribution to the Hazard curves used in Probabilistic risk hazard comes from distant large earthquakes. assessment (PRA) studies rely on the estimates as Figure 1 shows that in that range, G-experts true estimates of the hazard (absolute sense). So 5's ground motion model (number 3 on Fig. does any investigation of a single site without 1) is much higher than the rest of the models. comparing it to other sites. For this reason it is Thus, the resultant median value is more important to incorporate the entire specification of representative of the other four ground the hazard, including its uncertainty, rather than a motion models. point estimate or even a mean or median value. To this effect, most PRA now use a family of curves to represent the uncertainty in the seismic hazard

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 245 TABLE 1 RATIOS OF PGA VALUES BETWEEN SHALLOW AND ROCK CONDITIONS FOR FIXED VALUES OF THE HAZARD~~

~~Ratio Shallow/Rock~~

~~All Prob. of Exceed. Equal Only W/O Site SoU 10-3 10-4 io-5 Avg. Weight G5* G5** Category (1) (2) (3) (4) (5) (6) (7)~~

1 Nine Mile Point Sand-1 1.57 1.58 1.59 1.58 1.47 0.73 1.65 2 Susquehanna Till-2 1.30 1.30 1.30 1.30 1.25 0.73 1.38 3 Three Mile Island Sand-1 1.50 1.47 1.44 1.47 1.47 0.73 1.65 4 Browns Ferry Sand-1 1.56 1.66 1.68 1.63 1.47 0.73 1.65 5 Catawba Sand-1 1.59 1.58 1.55 1.57 1.47 0.73 1.65 6 Farley Sand-1 N/A 1.56 1.49 1.53 1.47 0.73 1.65 7 North Anna Sand-1 1.51 1.50 1.51 1.51 1.47 0.73 1.65 8 Oconee Sand-1 1.37 1.44 1.47 1.43 1.47 0.73 1.65 9 Summer Sand-1 1.47 1.62 1.61 1.57 1.47 0.73 1.65 10 Arkansas Till-1 1.51 1.50 1.50 1.50 1.39 0.73 1.55 11 Callaway Sand-1 1.65 1.70 1.72 1.69 1.47 0.73 1.65 12 Duane Arnold Till-1 N/A 1.50 1.50 1.50 1.39 0.73 1.55

•Ratio of PGA shallow/rock given by G-expert 5 only. •*Ratio of PGA shallow/rock given by G-experts 1,2,3 and 4 only. estimates. Comparison between plant sites, Another use of the results is in comparing the regions or groups of sites rely mostly on the spectral shapes of the uniform hazard spectra relative level of hazard between the sites. CUHS) at different sites. Figure 8 shows the median hazard curves for 19 The spectral level is sensitive to both the rate of sites in the north eastern pan of the EUS and Fig. 9 occurrence and earthquake magnitude. The longer for 16 sites in the north central region. It is clear period part of the CPUHS is very strongly based on these median values that on the average, influenced by magnitude. Thus, sites which are the seismic hazard in the north central region is influenced by very large earthquakes, e.g., around lower than that of the north east. the New Madrid region, will have more longer period energy than sites in New England where the

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~246 ACCELERATION CM/SEC"2 Figure 8 Comparison of the median CPHCs for the 19 sites in the North East.~~

~~ACCELERATION CM/SEC»«2 Figure 9 Comparison of the median CPHCs for the 16 sites in the North Central EUS.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

247 local activity from smaller earthquakes is could be estimated for all sites, More simple important. There is some influence of attenuation methods yet can be thought of which in some on the short period end of the spectrum, but it is manner consider some aspect of risk. relatively small. In a follow up project, our charter was to develop This is illustrated in Fig. 10 where we compare, grouping techniques and identification of outliers for two rock sites, the spectral shapes between a purely on the bases of the seismic hazard at the 69 site (Arkansas) where very large earthquakes plant sites. Without specifically involving risk, we dominate the hazard as contrasted to a site considered the probability of exceedance of the (Limerick) at which the seismic hazard is governed SSE values and multiples of the SSE, 0.3g and primarily by smaller nearby earthquakes. The main 0.5g. In addition we defined a new hazard difference in spectral shape is at the longer periods. parameter equal to a linear combination of the There is some difference at the short period end but hazard at the five periods available. This new it is relatively small. measure of the hazard places the emphasis on different periods at will, emphasizing the periods Another important use of the hazard estimates which are more important for a given plant. For consists in sorting the various sites according to example the 0.4 sec. to 0.1 sec. period window is criteria based on the probability of exceedance of in general more important than the rest of the some pre-chosen ground motion value. As an spectrum. In other cases one might want to example, Fig. 11 shows an ordering of the 69 sites in the EUS according to the median value of the hazard at 0.2g. Figure 11 shows that, depending on the type of criteria one would choose, several kinds of groupings could be obtained. The first two sites in the ordering, could be considered as forming a group by themselves, then the next five sites could form a second group, etc On the other hand, if the sites were ordered according to the arithmetic mean of the hazard (shown by the symbol "A" on Fig. 11), the order would be quite different. The same would be true of the 85th percentile (represented by the "*" symbol in Fig. 11). Furthermore, using the hazard at 0.6g instead of 0.2g would also lead to different results. Thus, it is quite obvious that ordering the sites on the basis of seismic hazard alone could be misleading at best and always tainted with some arbitrariness. •2 '2 ttKioo S 2 Risk based criteria could help in ordering the sites Figure 10 Comparison of the 10,000 year but could also be misleading if one is not careful in return period CPUHS between the Arkansas and selecting the criteria for ordering. One alternative Limerick sites (both rock sites). would be to order the sites on the basis of probability of core melt, or even on the total emphasize the low period range, smaller than 0.02 consequences of release, but clearly this would sec. require enormous efforts to include all sites. This methodology was reported in Bernreuter [13]. More simply however, generic plant fragility functions could be developed from the 20 or more existing PRA's, and the probability of core-melt

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~248 -2~~

~~A A « . A~~

1; A A" U'* * A -J ;A 1 • A 'i <','• A> A^ A A i ^ 1 B* S B> ; •1 t '? °e 'A • « » A t A 8 , i M M| ? I' 1' t| I'l » • i i ' ' J. 1 B?'I ii i 1 9 ' T ' i ' A-& .',9 ip. •. -4 •' j ... B 1 J 1 !' • ; 1 j 1 1 ?' ri A! ' ! ' h 3. ' B 8' il.. <1 1 1 if i I ' ** • !ii ii -5 ' n ;•! N • « . i '. | '' 1' o * o «• ! ' 1 ,11. 1; i i -6 * * i

-7

SITE NUMBER Figure 11 The 69 sites have been ordered by median probability of exceeding 0.2g. The symbols A, B, M and *, respectively represent the arithmetic mean, the best estimate, the median and the 85-15th percentile hazards. Note that if the site had been ordered according to other than the median hazard, a different order would have been obtained.

CONCLUSION factor of 4 in the ground motion for the range of values in the PSRV for a given return The detailed conclusions reached in the course of period. this study are given in [1]. The following is a summary of the most important ones: The range between the 15th and the 85th percentile hazard curves represents the total (1) There is substantial uncertainty in the uncertainty in estimating the seismic hazard at estimated hazard. The typical range in the a site due to two sources of uncertainty: value of the probability of exceedance between the 15th and 85th percentile curves • The uncertainty of each expert in the for the PGA is on the order of 40 times, for zonation, models and values of the low PGA; it is more than 100 at high PGA parameters of the analyses. values. This translates into an approximate factor of 4 in ground motion for the 15th- • The variation in the hazard estimates due to 85th range of values in the PGA given a fixed the diversity of opinirns between experts. return period, and similarly an approximate

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

249 The latter, or inter-expert variation is an introduced by zonation and recurrence important contributor to the total uncertainty models is also significant and of the same in the estimated hazard. Specifically, the order. magnitude of uncertainty introduced by the diversity of opinions between experts is of (5) We found that the correction for the site's soil the same order, jn the average, as the category had an important effect on the uncertainty in the hazard due to the estimated hazard. uncertainty of an individual expert in the value of the parameters. However, at rimes (6) We found that with our methodology, in the uncertainty between experts can be very general the site soil correction is not a linear large. operation on the hazard curve. Thus it is, in general, incorrect to modify a hazard curve (2) The 50th percentile CPHC appears to be a calculated for a rock site by multiplying by a stable estimator of the seismic hazard at the constant number (i.e., mean or median site. That is, it is the least sensitive to correction factor) to obtain the hazard curve at changes in the parameters, when compared to the same site for a different soil condition. other estimators considered in this study. Performing this incorrect operation could lead to errors in th;? estimate of the PGA, for a (3) The process of estimating the seismic hazard fixed return period, by as much as in the EUS is reasonably stable. Comparison 10 percent. However, we found that for with our previous results indicated that there some sites, multiplying the median hazard has not been a major shift in results over the curve for rock by the median correction factor past few years, although there have been would have given approximately the same some significant perturbations in the form of median hazard curve we obtained by recent occurrences of EUS earthquakes and performing the full analysis with our the completion of several major studies of the probabilistic correction factors. seismotectonics of the EUS. In the feedback Unfortunately, at the present time, we have performed in this study, there were some not been able to develop criteria to identify changes introduced by members of both the when performing such operation is Seismicity and GM Panels. However, the acceptable. computed hazard when aggregated over all experts did not significantly change. (7) Although the soil site correction is not region However, the introduction of the "new" dependent, we found that other complex random vibration models introduced a interactions, with zonation seismicity and significant change in the spectral shape by ground motion models, made the site raising the spectral values in the high correction actually region dependent in our frequency range and lowering it in the low methodology. frequency range. ACKNOWLEDGMENTS (4) It is difficult to rank the uncertainties, because zonation and the parameters of the The NRC project manager was Gus Giese-Koch. recurrence models are hard to separate. Nevertheless, our results indicate that the REFERENCES uncertainty in zonation, and ground motion models are more significant than the [1] D. L. Bernreuter, J. B. Savy, R. W. uncertainty associated with the seismicity Mensing, and J. C. Chen (1938), "Seismic parameters. The largest contribution to Hazard Characterization of 69 Plant Sites modeling uncertainty comes from the East of the Rockv Mountains." UCID-21517 uncertainty of the ground motion. The and NUREG/CR5250. correction for local site effects is a significant contribution to the overall uncertainty [2] D. L. Bernreuter and Minichino (1983), introduced by the ground motion models. Seismic Hazard Analysis, Overview and However, as already noted, the uncertainty

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 250 Executive Summary, NUREG/CR-1582, [ 121 V, W, Lee and M. D. Trifunac (1985), Vol. 1 (UCRL-53O3O, Vol. 1). Attenuation of Modified Mercalli Intensity for Small Fipicentral Distances in California. |3] D. L. Bernreuter, J. B. Savy, and R. W. University of Southern California Report Mensing, Seismic Hazard of the Eastern CEtfi-01. United States: Comparative Evaluation of the LLNL and EPRI Studies. USNRC Report [ 13] D. L. Bernreuter, R. W. Mensing, and J. B. NUREG/CR-4885 (1987). Savy, "Methodology on Plant Grouping for Decision Making," NUREG report to be [4] C. A. Cornell (1968), Engineering Seismic published for the NRC, November 1988. Risk Analysis, BSSA, Vol. 58, pp. 1583- 1606. [5] R. K. McGuire, (1976), Fortran Computer Program for Seismic Risk Analysis. U.S. Geological Survey, Open-file Report 76-67, 90 pages. [6] D. L. Bernreuter, J. B. Savy, R. W. Mensing, J. C. Chen, and B. C. Davis, Seismic Hazard Characterization of the Eastern United States. Vol. 1 and Vol. 2. LLNL UCID-20421, Vol. 1 and Vol. 2 (April 1985).

[7] Electric Power Research Institute (1986), Seismic Hazard Methodology for the Central and Eastern United States, EPRI-NP-4726, Volumes 1 through 10. [8] D. Veneziano and M. Heidari, "Statistical Analysis of Attenuation in the Eastern U.S.," Section 4 of Methods of Ground Motion Estimation for the Eastern U.S., EPRI RP 2556-16 (1986). [9] M. D. Trifunac, f!986). A Note of the Range of Peak Amplitudes of Recorded Accelerations. Velocities and Displacements with Respect to the Modified Mercalli Intensity. Earthquake Notes 47, pp. 9-24. [10] G. N. Atkinson, (1984), Attenuation of Strong Ground Motion in Canada from a Random Vibration Approach. BSSA. Vol. 74, pp. 2629-2653. [11] N. M. Newmark and W. J. Hall, Development of Criteria for Seismic Review of Selected Nuclear Power Plants, Nuclear Regulatory Commission Report NUREG/CR-0098, May 1978, 49 p.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~251 KEEPING PACE WITH THE SCIENCE: SEISMIC HAZARD ANALYSIS IN THE CENTRAL AND EASTERN UNITED STATES~~

~~Kevin J, Coppersmith and Robert R. Youngs Geomatrix Consultants, Inc. One Market Plaza Spear Street Tower, Suite 717 San Francisco, California 94105 United States of America~~

~~ABSTRACT~~

Our evolving tectonic understanding of the causes and locations of earthquakes in the central and eastern U.S. (CEUS) has been a challenge to probabilistic seismic hazard analyses (PSHA) methodologies. We summarize some of the more significant advances being made in characterizing the location, maximum earthquake size, recurrence, and ground motions associated with CEUS earthquakes. Seismic sources for PSHA in the CEUS have typically been based on the pattern of historical seismicity, but tectonic data including stress data are increasingly being incorporated into source identification. Paleoseismic and historical seismicity studies are evaluating the spatial stationarity of source zones in the CEUS. Techniques have been developed to use physical characteristics to quantitatively diagnose source activity and to assess alternative tectonic structures that might explain seismicity in a region. Assess- ments of maximum magnitudes are difficult to make in the CEUS and ongoing studies have extended the seismicity data base by compiling information on large earthquakes that have occurred within regions worldwide that are tectonically analogous to the CEUS. The frequency of earthquake occurrence is typically constrained by the historical seismicity record and extrapolations thereof. Recent paleoseismic investigations suggest pronounced spatial and temporal clustering that must be accounted for in hazard analyses. In terms of ground motion prediction, wave propagation studies have shown a pronounced effect of crustal structure on ground motion at intermediate distances. The wide range of interpretations of ground motion attenuation based on empirical and numerical analyses have underscored the necessity of considering alternative ground motion models in PSHA.

INTRODUCTION States (CEUS) is especially difficult because the area lies within a "stable The assessment of earthquake hazards continental region" (SCR) away from tec- within the central and eastern United tonic plate boundaries. In these areas,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

252 ; uiuimiii-nt al aspects of the earthquake parameters (e.g., peak ground acceler- probK-ii! are uncertain such as: stress ation, spectral acceleration at various stale and raechani sins for crustal stress, frequencies;, etc. ) , h) Calculated KSHA cau.sit.ivo faults and geologic struc- results ("hazard curves"), which express tures, validity of the historical record the probability of exceeding various relative to the future location and size levels of ground motion at a site, of earthquakes, and attenuation of seis- usually in terms of annual probability. mic wave energy from seismic sources to The hazard calculation has a reasonable sites of interest. Despite these physical basis in that it aggregates the issues, the historical occurrence of occurrence of earthquakes on various large and damaging earthquakes in the sources, at various distances, having CEUS, such as that in 1811-1812 New various magnitudes and rates of recur- Madrid. Missouri, and 1886 Charleston, rence. Efficient methods have been South Carolina, as well as the large developed for incorporating the uncer- concentration of population and engi- tainties in the PSHA elements into the neered structures demands that seismic calculated hazard results [3]. hazards be evaluated. Recent years have seen the completion of very large proba- The issues related to earthquake source bilistic seismic hazard analyses (PSHA) characterization are very different in tor commercial nuclear power plants in the eastern U.S. than the western U.S. the CEUS '1, 2). In conjunction with In the west (roughly that region lying these studies, as well as through west of the Rocky Mountains), the issues ongoing scientific research, our under- related to future location, maximum size, standing, of the causes and locations of and recurrence of earthquakes deal CEUS earthquakes is increasing. We here largely with fault behavior (see Youngs summarize some of these recent advances and Coppersmith, this volume). Eastern and their implementation in PSHA. North America is a "stable continental region" (SCR) region away from plate The basic elements of a PSHA are: 1) boundaries (see definition of SCR given Definition of seismic sources that are in [4]). Here the issues related to assessed to be the locations of future earthquake source characteristics are in earthquakes. Seismic sources may be some ways more fundamental than in the re-presented by areal source zones or by West. What are the causes of crustal faults depending on the level of stress? What are the mechanisms for knowledge available, 2) Earthquake eastern earthquakes? (reactivation recurrence relationships for each seis- processes, stress amplification, strain ::. ic source that describe the frequency localization?) Hov; much can we rely on of occurrence of various magnitude seismicity data to estimate future earthquakes up to the maximum. Recur- earthquake locations, maximum size, and rence relationships may be developed recurrence rate? It is these issues from one or a combination of seismicity that we wish to address in this paper. ar.d geologic data. The maximum magnitude that a source is capable of EARTHOUAKF SOURCE IDENTIFICATION generating is usually a difficult parameter to assess because of its Historically, the guiding philosophy be- raritv relative to the period of hind estimating where earthquakes will historical observation, 'i) Ground motion occur in the future in the CEL'S ('i.e. . attenuation relationships that express defining seismic sources; has been, in t':.•• rate- of decay of seismic wave energy essence, to let the spatial pattern of fro:?. th>- seismic source to the site as historical seismicity represent the •: '.\\~.'t : o;: of magnitude a: id distance. pattern of future occurrences (i.e.. '..< : ••i.'ilr.y, on the engineering purposeassum for e general spatial stationaritv;. t:.-- ?SHA, attenuation relationships are- A principal reason for this is that the •iv-ve loped for a variety of ground motion general observed spatial distribution or Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 seismicity in the East, which is remark- sediment loading may be locally signifi- ably non-uniform (Figure 1), has re- cant but are secondary to the CEUS as a mained relatively stable when the older whole;. historical epicenters are compared with the more recent instrumental locations. This is particularly true for the regions where large historical earthquakes have occurred (see Figure 2). However, ILLINOIS most seismologists are fully aware that MISSOIHI the historical record in the CEUS may 38" \ r*Ji not be sufficiently long to tell us where we might expect all future earthquakes of engineering significance to occur.

* TKNNF2BEK

~~CXPIANA1I0N 3 0 £ mj, £ 3.9 -~~

~~9l«30 W30' 89*30 39W • - • •~~

~~38W~~

~~Figure 1. Epicenters of earthquakes mb > 3 during the period 1500-1984 [1].~~

A promising approach to evaluating the future locations of earthquakes is to consider the geologic structures and 36.""% tectonic history of the region, in much the same way that faults might be iden- MiK tified as potential seismic sources in the West. Recent analyses of the stres- 25 X ses in the upper crust in the CEUS (Fig- ure 3) show that the stresses are com- Figure 2. Comparison of historical pressive and are remarkably uniform in seismicity with recent instrumental orientation from northeast-southwest to seismicity in the New Madrid seismic east-west (e.g., [6, 7]). Adams [8] has zone. The epicenters in 'A' are mb > 3 arrived at a similar result in eastern for period 1811 - June, 1974. In 'B', Canada. This uniformity across various the time period is July 1974 - June 1978. geologic terranes and based on a variety Note that the pattern of more recent of stress measurement methods strongly instrumental seismicity data (through suggests that the primary stress mechan- 1988) is very similar to that shown in isms are far-field plate tectonic 'B'. The comparison shows a clear mechanisms (e.g., ridge-push). Other spatial concentration over the 167-year mechanisms such as glacial unloading or period (from [b}).

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

254 candidate tectonic features exist as wall as amplified stresses. However, the consistency in stress orientation as well as the spatial associations made between seisraicity and profound crustal features at New Madrid, the St. Lawrence rift, and elsewhere around the world [9] have lent more credibility to the "reactivation" hypothesis, which appears to be favored by the seismological community at the present time. For example, in a major study of the seismic hazards at nuclear sites in the CEUS [1], a large group of experts developed seismic source

~~MAXIMUM HORIZONTAL interpretations by specifically consider- COMPRESSIVE STRESS ing the earthquake potential of tectonic~~

MECHANISM features. Criteria were identified for IN SITU STRESS assessing the probability that any par- GEOL ticular feature was seismogenic and then MIXED methodology allowed for a quantitative analysis of this probability. The criteria for activity included characteris- Fit:, 11 re 3. Orientation of maximum tics such as the spatial association of horizontal stress in the eastern U.S. as the feature with seismicity orientation determined from a variety of crustal of the feature relative to the maximum stress indicators (from [6]). compressive stress direction, evidence for Cenozoic brittle reactivation, and the presence of intersecting features. Basically, two alternative mechanisms have been proposed for the generation of In conclusion, the identification of earthquakes in the relatively uniform seismic sources in the CEUS is based conipressional stress environment in the largely on the emerging belief that the CEUS: 1) the reactivation of geologic maximum horizontal compressive stress zones of "weakness" that are favorably orientation is relatively consistent and oriented relative to the contemporary that earthquakes are the result of reac- regional stress field, and 2) local tivation of existing zones of "weakness" amplification of stresses leading to or tectonic structures. Because ongoing earthquake stress release. These models or historical seismicity is perhaps the imply different manifestations and, in best indication of reactivation in the turn, different methods to search for present tectonic regime, seismicity data future seismic sources. The reactiva- strongly control the configuration of tion hypothesis would suggest that we CEUS seismic sources. However, consider- identify tectonic features that are ation of the tectonic features, their oriented properly relative to regional orientation relative to the contemporary stresses. The stress amplification stress field, and geologic evidence for hypothesis suggests that we identify reactivation can also supplement the high stress regions or attempt to historical seismicity data. Typically, identify geologic conditions where the resulting seismic source zones then stresses might be expected to be have configurations that reflect both the amplified (e.g., areas of high rigidity locations of tectonic features as well as contract as along the margins of a the locations of seismicity (see, for pluton). Of course, these models are example, [1, 2] ) . not necessarily mutually exclusive — one might expect earthquakes where

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 to bo statistically significant in MAXIMUM EARTHQUAKE MAGNITUDE assessing the maximum earthquake magnitude. The estimation of the maximum magnitude that a seismic source can generate is Methods Based On Global Data Base very often difficult but is vitally The most common methods for esti- important to both probabilistic and mating maximum earthquakes in the central deterministic seismic hazards assess- and eastern United States have been based ments and probably best exemplifies the on seismioity data, as discussed above. differences in approaches to seismic The chief weakness of these approaches is hazard assessment in the CEUS versus the the generally short time period of his- West. Because of the need for site- torical observation. One way to overcome specific hazard predictions, we attempt this problem is to substitute space for to establish the expected largest event time to make the historical record more that may be generated by each seismic meaningful. To do this, one can expand source of significance to the site the region of data collection beyond the (i.e., the maximum magnitude is the central and eastern United States to point of truncation of the recurrence other parts of the world that are analo- curve). The kinds of seismic sources gous in terms of geologic and tectonic that are identified typically in the characteristics important to the earth- central and eastern United States for quake process. The logic is simple: hazards assessments include tectonic Although the likelihood of any given provinces, seismicity zones, and tec- seismic source having experienced its tonic features. Unlike the western maximum earthquake is low, the chances United States or other interplate are very good that a tectonically-similar regions, rarely are the causative geo- source somewhere in the world will have logic structures recognized active had its maximum event. The largest ob- faults. Therefore, approaches to esti- served earthquake, then, for that "type" mating maximum earthquakes that assume of seismic source would provide a basis a knowledge of the seismogenic fault or for estimating a maximum magnitude for details of its behavior (e.g., see our source of interest. Youngs and Coppersmith, this volume) will not be generally applicable in the Drawing "tectonic analogies" in an infor- CEUS. Appropriate methods for the CEUS mal way has been done for some time in must be based on the types of data that estimating a maximum magnitude. An on- are generally available for the eastern going major study for the Electric Power seismic sources such as historical seis- Research Institute [4, 9, 11] builds on ir.icity, regional tectonics, etc. this space-for-time substitution in a more formal manner to allow utilization Maximum earthquakes have been estimated of a global earthquake data base. The in the CEUS primarily from the seis- focus is on the identification of factors micity record because seismicity data that control or limit the maximum size of are commonly available for identified earthquakes within stable continental source zones—-not because of a general regions (SCR) such as the CEUS. The acknowledgement that seismicity-based approach is first to delimit explicitly methods are scientifically supportable. the study regions and then to systemati- These methods have been evaluated by cally examine the larger earthquakes that Coppersmith et al. [A, 10] and will not have occurred within them. be discussed further here. It is, however, important to note that the most Some of the preliminary conclusions significant limitation to all of the coming out of the EPRI study that are methods is that the historical germane to the maximum magnitude issue seismicity record for an individual include: seismic source usually cannot be shown

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

25b • The rate of occurrence of large earth- earthquakes associated with known rifts, quakes within stable continental crust margins, and extended curst. All, eight is very low relative to plate boundary M £ 7 events are associated with (incerplate) regions, although the extended continental crust. area of stable crust is about three time larger. Fewer than sixty earth- EARTHQUAKE RECURRENCE quakes have occurred in the past 200 years having magnitude equal to or Probabilistic seismic hazard analyses greater than 6 and only eight events require the specification of the recur- of magnitude 7 or greater. rence or frequency of occurrence of earthquakes of various magnitudes. Stan- The seismic moment release rate in SCR dard practice in the CEUS has called for is less than 1% of the global total heavy reliance on historical seismicity rate. Expressed in terms of moment data in assessing recurrence, mainly in magnitude per year, plate- boundary the absence of other types of data. Of regions generate one moment magnitude course, to do this, the assumption must 8.6 per year and SCR regions generate be made that both the location of zones one 6.8 per year. About 64 years have of seismicity in the historical record passed since the last M > 7 event and the observed rates of earthquake occurred in stable continental regions occurrence will remain the same in the worldwide and over a century and a future (i.e., spatial and temporal sta- half since the last M a 8 event. tionarity) . An obvious potential problem is that the historical record may not be • Most (68%) locations of large events sufficiently long to conclusively define appear to have occurred within the earthquake locations (discussed earlier) location of prior seismicity. Far or rates of recurrence. Few seismicity fewer cases of no prior seismicity zones in the CEUS have experienced more (18%) are known and for many cases than a single large-magnitude earthquake (14%) no data are available. This — hardly enough to characterize the suggests that the occurrence of large recurrence of these events. events away from recognized seismicity is uncommon. As discussed in an accompanying paper (Youngs and Coppersmith, this volume), • Intra-continental rifts and passive the recurrence issue is addressed in the margins (formerly extended continental western United States by relying on geo- crust) are important loci of SCR logic data — either recurrence intervals earthquakes. Virtually all of these derived from dating paleoseismic events tectonic features are under compres- or fault slip rate. The problem of sion in the present tectonic stress applying these approaches to the CEUS is, regime. About 75% of all SCR earth- of course, our inability to recognize quakes (M > 5) are associated with causative faults. extended continental crust; 35% with passive margins, 40% with imbedded So what are the problems with using seis- rifts. Rifts whose most recent exten- micity data (or extrapolations thereof) sional activity occurred in Mesozoic to estimate recurrence in the CEUS? time appear to be the most seismically Arguments have been made that during the active. 100 to 300 year (depending on location) historical period, the locations of zones • From the standpoint of the occurrence of seismicity have remained relatively of large earthquakes, which is most constant. It is concluded, therefore, important to the maximum earthquake that one can expect these zones to remain problem, the presence of extended where they are in the short-term future crust appears to be highly (e.g., 50 years) of engineering significant. 70% of the M > 6 significance. However, even if one Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

257 argues for such spatial stationarity, it showing that temporal "clustering" last is not clear that the race of earthquake for several seismic cycles over several occurrence has been stable over the thousand years, and are not just a single historical period. For example, Arm- "surprise" event [21]. Likewise, the bruster and Seeber [12] argue that in spatial extent of the active areas is the 19th century before the Charleston being assessed from paleoseismic work on earthquake, the seismicity in the meizo- the distribution of liquefaction seismal area was relatively low, unlike associated with prehistoric earthquakes the distinctly high seismicity rate that (e.g., [14]). Tentative conclusions for has occurred subsequently in this area. the Charleston and New Madrid regions Other issues are coming to light suggest that the active clusters exist regarding the stationarity issue. within the regions that would be defined by the present small-magnitude Temporal And Spatial Clustering seismicity. This is comforting because Geologic (paleoseismic) studies at seismic sources for hazard analysis are Charleston [13, 14], New Madrid [15], usually assessed this way. Exceptions are Charlevoix [16] and the Meers fault [17, locations such as the Meers fault, which, 13] all show evidence for repeated large has not been associated with seismicity magnitude earthquakes having recurrence and would require geologic recognition in intervals of several hundred to a few order to be identified as a potential thousand years. Extrapolation of his- seismic source. torical seismicity recurrence curves also suggest similar intervals (e.g., In conclusion, earthquake recurrence is [19, 20]). Coppersmith [21] argues that typically estimated for CEUS seismic these high rates of recurrence cannot be sources using the historical seismicity easily reconciled with the observed lack record, which usually contains small- of geologic evidence for Quaternary magnitude events but very few or no deformation without calling on fairly large-magnitude events. When using these pronounced temporal and spatial cluster- data, consideration should be given to ing of earthquake activity. Such behav- whether the seismic source is within an ior is being increasinly identified active "cluster". Ongoing and future within intraplate as well as plate- paleoseismic studies will be developing boundary environments (e.g., [22]) and data to better define the long-term appears to be especially significant in recurrence behavior and stationarity of SCR such as the CEUS. This suggests seismic sources. Paleoseismic data, that, at least for these zones and which are few in the CEUS, are critical possibly for others, the time period for identifying potential seismic sources over which the active cluster persists such as the Meers fault in Oklahoma, is at least a few thousand years. which displays abundant evidence of geologically recent activity but has been The implication of temporal clustering aseismic during the historical period. to recurrence analysis is that one must assess whether or not a given seismic GROUND MOTIONS source is within an active "cluster" period or between clusters. If it is The lack of a large data base of CEUS within a cluster (such as New Madrid or strong motion recordings has in the past Charleston), the appropriate recurrence necessitated the use of macroseismic data rates are probably those developed from from historical earthquakes and analyses historical seismicity and the recent of crustal wave propagation characteris- geologic record. If the source is tics to development of ground motion between clusters, a "background" recur- attenuation relationships. These rela- rence rate from the regional seismicity tionships (e.g., 23, 24) These relation- data is probably more appropriate. For- ships show significant difference from tunately, preliminary studies are those developed from WUS strong motion Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

258 data, particularly a much lower rate of attenuation with distance. However, 20 •i i "i•• within the past several years analyses of newly gathered ground motion data together with theoretical models of strong ground motion attenuation have o 10 suggested that scaling of CEUS ground motions with magnitude and distance is more nearly like WUS motions than had & 5 been implied by the older attenuation relationships [25, 26], These results are complicated by wave propagation O studies that indicate crustal structure o can have a pronounced effect on ground .to motions at intermediate distances [27], Eastern US, mb 6 at 30km Somervilie et al. (28] have suggested ? 1 Boore and Atkinson [25] that these wave propagation effects may o • • • • Nuttli [30] account for the relatively large ampli- p, • McGuire, Toro, and Silva [31] tudes of recorded motion at distances of ^ ,5 Toro and Me Guir« [26] about 100 km from the 1988 Saguenay, Quebec earthquake. The observed motions are somewhat larger than would be pre- .3 dicted using the recently developed 10 attenuation relationships for the CEUS [29],

Thus there is currently a much larger level of uncertainty in predicting ground motions in the CEUS than in the VUS. Figure 4 contrasts predictions of response spectra for the CEUS and WUS made using recently developed attenuation relationships. The top plot in L'.' Western US, Mw 6 at 30km Figure 4 shows the response spectra for Idriss [32] a magnitude 6 earthquake predicted by • • • • Joyner and Fumal [33] four attenuation relationships used in Sadigh et al. [34] studies of seismic hazards at CEUS nuclear power plants [1, 2]. The bottom plot shows similar estimates for an I ] I 1 | I l I I 1 I 1 I earthquake in the WUS made using pub- .05 .1 .2 .5 12 lished response spectral relationships. The level of uncertainty in estimating Period (sec) CEUS ground motions underscores the need to include alterative ground motion Figure 4. Predicted response 5%-damped models in PSHA. spectra for magnitude -6 earthquakes using attenuation relationships developed for CEUS (bottom) and WUS (top) earthquakes. Note greater variation in predicted spectra for CEUS.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

259 CONCLUSION [8) J. Adams, "Crustal stresses In eastern Canada," in S. Gteiierson and P,H, Basham («ds.), Earthquakes at Horth-Atlantlc Passive Margins; Heotectonlcs and While our understanding of the Postglacial Rebound, Kluwer Academic Publishers, earthquake generation process in the pp. 289-297, 1989. CEUS is far from complete, significant [9] K.J. Coppersmith, R.R. Youngs, A.C. Johnston, and L.R. Kanter, Methods for assessing earthquakes in advances have bee*, made in m.-mv areas the central and eastern United States. EPRI Research that affect the assessment o£ seismic Project 2556-12, Electric Power Research Institute, hazards. We have described many of Palo Alto, California, 1989. these advances and have indicated how (10] K.J. Coppersmith, A.C. Johnston and W. Arabasz, "Methods for assessing maximum earthquakes In the they are being Jncorporated into PSHA. central and eastern United States: A progress It is also important to recognize that report," Proceedings of the Hatlnal Center for Earthquake Engineering Research Symposium on Seismic there are uncertainties in applying Hazards. Ground Motions. Soil Liquefaction, and these new models. Modern PSHA should Engineering Practice In eastern Horth America. incorporate the alternative hypotheses October 20-21, NCEER-87-0025, 1987b. and data interpretations that may affect [11) A.C. Johnston, "The seismlclty of "Stable Continental Interiors," lr. S. Gregersen and P.W. the computed results. In this manner, Basham, eds., Earthquakes at Worth Atlantic Passive a realistic assessment of the Margins; Neotectonics and Postglacial Rebound. uncertainty in the computed hazard can Kluwer Academic Publishers, pp. 299-327, 1989. be made to guide in the use of PSHA [12j J.G. Armbruster and L. Seeber, "Seismlclty and seismic zonation along the Appalachians and the results in design and safety evaluation. Atlantic seaboard from Intensity data," Proceedings of NCEER Symposium Seismic Hazards, Ground Motions, Soll-Llqiiefactlon and Engineering Practice in REFERENCES Eastern Horth America. October 20-22, 1987, Technical Report, J1CEER-87-0025, pp. 163-177, 1987. !1] Electric Power Research Institute (EPRI), Seismic hazard methodology for the central and eastern [13] S.F. Obermeier, G.S. Gohn, R.E. Weems, R.L. Gellnas, United States. EPRI NP-4726, Project P101-21. 11 and M. Rubin, "Geologic evidence for recurrent volumes, July 1986. moderate to large earthquakes near Charleston, South Carolina," Science, v. 227, pp. 408-411, 1985. (2) D.L. Bernreuter, J.3. Savy. R.W. Menslng, and J.C. Chen, Seismic hazard characterization of 69 [14] S.F. Obermeier, R.B. Jacobson, D.S. Powars, R.W. nuclear plant sites east of the Rocky Mountains. Weems, O.C. Hallbick, G.S. Gohn, and H.U. Markewich, NUREG/CR-5250, UCID-21517, 8 volumes. 1989. "Holocene and late Pleistocene(?) earthquake-induced sandtlows in coastal South Carolina," Proceedings of [3] K.J, Coppersmith and R.R. Youngs, "Issues the Third U.S. national Conference on Earthcuake regarding earthquake source characterization and Engineering. Earthquake Engineering Research seismic hazard analysis within passive margins and Institute, v. 1, pp. 197-208, 1986. stable continental interiors," E. Gregersen and P.W. Basham, eds., Earthquakes at North-Atlantic (15) D.P. Russ, Style and significance of surface Passive Margins: Neotectonics ar.d Postglacial deformation in the vicinity of New Madrid. Missouri. Rebound. Kluwer Academic Publishers, pp. 601-631, U.S. Geological Survey Professional Paper 1236-H, 1989. pp. 94-114, 1982.

; - j K.J. Coppersmith, A.C. Johnston, A.G. Metzger, and [16] R. Doig, "A method for determining the frequency of U.J. Arabasz, Methods for assess ir.g maximum large magnitude earthquakes using lake sediments," earthquakes in the central and eastern United Canadian Journal of Earth Sciences, v. 23, p. 93C- States. Electric Power Research Institute, 937, 1986. Research Project 2556-12, January 1987. [17] A. Crone, "The Meers fault, Southwest Oklahoma; ,'5] W. Stauder, "Present-day seismiclty and evidence of multiple episodes of Quaternary surface identification of active faults in the !iev Madrid faulting," (abstract) Geological Society of America seismic zone," in Investigations of the New Abstracts with Programs. Annual Meeting, p. 630, Madrid. Missouri. Earthquake Region. FA. McKeovr. 1937. and L.C. Pakiser, eds., Geological Survey Professional Paper 1236, pp. 21-30, 2 982. [18) F.H. Swan, Ongoing studies of the Meers fault. Oklahoma. Contract Ho. NRC-04-077 to Geomatrix |6] M.D. Zoback, and M.L. Zoback, "Tectonic stress Consultants, San Francisco, California. field of the continental U.S. ," in L, pakiser and W. Mooney, eds. , Geophysical Framework of the [19] O.W. liuttli, "On the problem of the maximum Continental United States, Geological Society of magnitude of earthquakes," ^in proceedings of America Memoir, 1988 (in press). Conference XIII, Evaluation of Regional Seismic Hazards and Risk. 'J.S. Geological Survey Open-File I?; K Zoback, S.P. Nishenko, R.M. Richardson, H S. Report 81-437, p. 876-885, 1931. Kascgawa. and M.D. Zoback, "Mid-plate stress, deformation, and seismic ity , " l£ Vogt , P.P.. , and [20] A.C. Johnston, and S.J. Nava, 1985, "Recurrence Tucr.olke. 3 E. , eds . The Geology of N'orrh raves and probability estimates for the New Madrid Arr.or lea. v. M, The Western North Atlantic Region. seismic, zone," Journal of Geophysical Research, Geological Society of America, bou!t:er, Coiorado, v 90, ,-o B3, pp. 6737-6753. 1985. 'Jr.;.ted States of America, p. 297-112, !'j8t

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 260 [31] K.J, Coppersmith, "Temporal and spatial clustering of earthquake activity in tht central and eastern United States," Se}smoloiloal Research Letters, v. 59, no. *, pp. 299-30*. 1989,

[22] R.E. Wallace, "Grouping and nitration of surface faulting and variation* in ilip r»t«s on faults in the Great lasIn province," Bulletin of the Seisrooloaical Society of America, v. 77, pp. 868- 876, 1987.

~~(23] O.W. Nuttli, "Seismic wave attenuation and magnitude relations for eastern North America", Journal of Geophysical Research, v. 78, p. 876- 885, 1973,~~

12*] O.W. Nuttli, The relation of sustained maximum around acceleration and velocity to earthquake Intensity and magnitude. St«ie-of-the-Art for Assessing Earthquake Hazard in the United States, Report 16, Waterways Experiment Station, Corps of Engineers, Vicksburg, 100 p., 1979.

[25] D.H. Boor* and G.M. Atkinson, "Stochastic prediction of ground motion and spectral response parameters at hard-rock sites in eastern North America, " Bulletin of the Seismoloalcal Society of America, v. 77. pp. 440-467, 1987,

[26] G.R. Toro, and R.K. McGulre, "An Investigation Into earthquake ground motion characteristics In eastern North America," Bulletin of the Seismoloalcal Society of America, v. 77, n. 2, pp. 468-489, 1987,

[27] R.U. Burger, P.G. SomervlUc, J.S. Barker, R.B. Herrmann, and D.V, Helmberger, "The effect of crustal structure on strong ground motion attenuation relations In eastern North America," Bulletin of the Seismoloalcal Society of America, v, 77 pp. 420-439, 1987.

[28] P.G. SomerviUe, J.P. McLaren, and N.F. Smith, "Wave propagation modeling of ground motion attenuation of the Saguenay earthquake sequence of November 25, 1988," (abs) Selsmoioglcal Research Letters, v. 60, n. 1, p. 19, 1989.

[29] G.M. Atkinson, and D.M. Boore, "Preliminary analysis of ground motion data from the 25 November 1988 Saguenay, Quebec earthquake," (abstract) Selsmoioglcal Research Letters, v. 60, n- 1, p. 18, 1989.

~~[30] O.W. Nuttli, Letter dated September 19, 1986 to J.B. Savy. Reproduced In: [2], v. 7.~~

~~[31] R.K. McGuire, G.R. Toro, and W. Silva, Engineering model of earthquake ground motions for eastern North America. Technical Report NP-6074, Electric Power Research Institute, 1988.~~

~~[32] I.M. Idriss, "Evaluating seismic risk in engineering practice," Proceedings Eleventh International Conference on Soil Mechanics and Foundation Engineering, August 12-16, v. 1, pp. 225-320. 1985.~~

[33] W.B. Joyner and T.E. Fumal, "Predictive mapping of earthquake ground motion". In Evaluat ing earthquake hazards in the Los Angeles region: U.S. Geological Survey Professional Paper 1360, pp. 203-220, 1985.

134) K Sadlgh, C.-Y. Chang, F. Makdisi, and J.A. Egan, "Attenuation relationships for horizontal peak ground acceleration and response spectral acceleration for rock sites," (abstract), Selsmologlcal Research Letters, v. 60, p. 19, 1989.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~261 KEEPING PACE WITH THE SCIENCE: SEISMIC HAZARD ANALYSIS IN THE WESTERN UNITED STATES~~

~~Robert R, Youngs Kevin J, Coppersmith Geomatrix Consultants One Market Plaza, 717 Spear Street Tower San Francisco, California 94105~~

~~ABSTRACT~~

Recent years have witnessed rapid advances in the understanding of the earthqiiake generation process in the western U.S. , with particular emphasis on geologic studies of fault behavior and seismologic studies of the rupture process. We discuss how probabilistic seismic hazard analysis (PSHA) methodologies have been refined to keep pace with scientific understanding. Identified active faults are modeled as three-dimensional surfaces with the rupture shape and distribution of nucleation points estimated from physical constraints and seismioity, Active blind thrust ramps at depth and sources associated with subduction zones such as the Cascadia zone off Oregon and Washington can also be modeled. Maximum magnitudes are typically estimated from evaluations of possible rupture dimensions and empirical relations between these dimensions and earthquake magnitude. A rapidly evolving technique for estimating the length of future ruptures on a fault is termed "segmentation", and incorporates behavioral and geometric fault characteristics. To extend the short historical record, fault slip rate is now commonly used to constrain earthquake recurrence. Paleoseismic studies of fault behavior have led to the "characteristic" earthquake recurrence model specifying the relative frequency of earthquakes of various sizes. Recent studies of have indicated the importance of faulting style and crustal structure on earthquake ground motions. For site-specific applications, empirical estimation techniques are being supplemented with numerical modeling approaches.

INTRODUCTION that each source is capable of generating, 3) the recurrence rate of earth- Conceptually, the basic elements of quakes of all magnitudes up to the maxi- probabilistic seismic hazard analysis mum on each source, and 4) the attenua- (PSHA) have not changed significantly tion of strong ground motions as a func- since proposed by Cornell [1]. That is, tion of source-to-site distance and to calculate the probability of exceed- magnitude. What has changed in the past ing various levels of ground motions at two decades has been our scientific abil- a site of interest, we must assess: 1) ity to estimate these quantities with the possible sources of earthquakes that more confidence and to quantify more might affect ground motions at the site, appropriately the uncertainties asso- 2) the maximum magnitude earthquakes ciated with the input parameters.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

262 Recent years have witnessed rapid models faults as planes that extend from advances in the understanding of the their surface trace down to the bottom earthquake generation process in the the seismog.enic crust, The base of the western U.S., particularly with regard seismogenlc crust (typically about 10-15 to geologic studies of fault behavior km in the western U,S.) can be identified and soisraological studies of the rupture by the maximum depths of well-located process. In seismic hazard assessments hypocenters in the region or by consider- for engineering purposes, we are parti- ation of physical properties of the crusi cularly interested in incorporating (e.g., [4]), these scientific advances into our methodologies for calculating hazard. Of The recent Coalinga (1983) and Whittier course, this process of updating leads Narrows (1987) earthquakes were generated to additional levels of complexity in along thrust faults whose rupture did not PSHA modelling, but, as we will discuss reach the surface. These moderate-magni- in this paper, this complexity can be tude events, which caused considerable readily accommodated. localized damage, are reminders that significant earthquakes can occur along Our objective in this paper is to pro- faults that are not mapped at the sur- vide a brief summary of those scientific face. In both cases, however, as well as advances in our understanding of earth- in other well-documented cases of subsur- quakes in the western U.S. that are face thrust-faulting, the earthquake was being incorporated into PSHA methodolo- accompanied by readily-measurable coseis- gies. The issues associated with esti- mic uplift reflecting permanent deforma- mating earthquake hazards in the eastern tion associated with fault slip at depth. and western United States are suffi- Further, it has now been shown that the ciently different that we treat them coseismic uplift occurred coincident with separately (see Coppersmith and Youngs, the axes of Quaternary folds that extend this volume). The theme in both papers to the surface and show good evidence of is that PSHA methodologies are now flex- geologically-recent activity [5]. The ible and sophisticated enough to incor- point here is that subsurface thrust porate our most recent thinking about faults associated with active folds can the earthquake process and the asso- be identified before the earthquake as ciated ground motions. Recent years potential seismic sources using essen- have also witnessed the development of tially the same geologic techniques that efficient methods for incorporating are used to identify active faults. In uncertainty into PSHA, such as the use addition, quantitative structural geology of logic trees [2, 3], techniques (e.g., [6]) may help provide estimates of the geometry of buried Note that we discuss here the seismic thrust faults for purposes of seismic hazards associated with earthquake hazard analysis. shaking, but the wider range of seismic hazards including surface faulting, Another category of seismic sources, liquefaction, and the like are also which are unique in the U.S. to the implied. Alaska, Washington, and Oregon areas, are sources related to plate subduction EARTHQUAKE SOURCE IDENTIFICATION zones. Numerous historical examples around the world confirm that most of the Studies of aftershocks of large earth- largest observed earthquakes have quakes, as well as inversion of geodetic occurred in subduction zone environments. and seismic data, show that earthquake Seismologic and geophysical studies of fault ruptures can be reasonably approx- subduction zones show that the three- imated by -^ree-dimensional fault dimensional geometry of earthquakes is planes. As a result, PSHA typically complex and vitally important to

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

263 estimates of seismic shaking at various locations at the surface. For example, the C\seadia subduction zone along the Oregon and Washington coast is probably best, modeled as consisting of three distinct types of seismic sources (Fig- ure 1): 1) a plate "interface" between the subducting Juan de Fuca plate and the North American plate, 2) "intraslab" sources that occur within the Juan de Fuca plate (subducting oceanic crust is usually termed a 'slab'), and 3) "crustal" sources that occur within the upper part of the North American plate. The geometries of these sources can be specified for a PSHA. However, an interesting problem arises in assessing the seismogenic potential of these various Cascadia subduction zone sources inas- much as the observed earthquake behavior Fipure 1. Diagrammatic sketch of of each has been quite different histor- Cascadia subduction zone showing the ically. The "intraslab" source has gen- potential earthquake sources: the plate erated historical earthquakes larger interface, intraslab, and shallow crustal than magnitude 7 in historical times and sources (modified from [7]) is known to be the source of abundant ongoing small-magnitude seismicity. The 1949 and 1965 earthquakes that caused rarity of maximum earthquakes on seismic damage in Seattle were intraslab events. sources (hundreds to thousands of years), In contrast, the plate "interface" has the historical seismicity record is not been conspicuously seismically quiescent typically used to estimate maximum magni- during the 150-200 year-long historical tudes. In the western U.S., maximum mag- period, despite the fact that we typi- nitudes are usually estimated from evalu- cally observe the largest subduction ations of the maximum expected dimensions zone earthquakes (M > 8) along plate of a future fault rupture (rupture interfaces. The National Earthquake length, rupture area, maximum displace- Hazard Reduction Program is currently ment, etc.). These dimensions are empir- focussing its attention on the seismic ically related to earthquake magnitude hazard in the Pacific Northwest to help through observations of a large number of resolve this problem. In the meantime, historical earthquakes worldwide. The it is possible to incorporate the uncer- past ten years have seen rapid advances tainties associated with the potential in our ability to estimate future rupture seismic sources in this region for seis- dimensions, based primarily on geologic mic hazard analysis using available studies of prehistoric earthquakes as approaches and expert scientific opinion well as detailed evaluation of historical [8]. ruptures. An example will serve to illustrate the progress being made. MAXIMUM EARTHQUAKE MAGNITUDE For many years it has been recognizeo Estimates of maximum earthquake magni- that the length of fault rupture at the tudes for each seismic source are neces- surface is related to the magnitude of sary elements of both probabilistic and the associated earthquake--longer rup- deterministic approaches to evaluating tures are associated with larger earth- seismic hazard. Because of the usual quakes. With the compilation of

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

264 measurements of rupture lengths and • Changes in cumulative slip magnitudes of the associated earthquakes, linear regressions were devel- As an example of the proposed segmenta- oped (e,g, (9)) that can provide the tion o£ a fault zone, the segments along magnitude of a postulated earthquake the Wasatch fault in Utah are given in along a fault of interest, given that Figure 2, In this case, changes in the one is able to estimate the maximum recency of faulting along the fault zone rupture length expected along the fault. as well as structural and geometric Estimating this maximum rupture length changes along the fault zone were the is not trivial. basis for defining those segment boundaries shown. In recent years, the study of historical ruptures as well as detailed geologic studies of the displacement and extent of prehistoric ruptures, has led to the development of a aew technique for estimating rupture Jci.gth termed fault "segmentation" (e.g., [10, 11]). Studies of the segmentation of ruptures have grown out of earthquake research and will undoubtedly have predictive implications that go well beyond merely predicting rupture length, but at the present time rupture length estimation is one immediate application. Geologic and seismic studies are suggesting that the locations of ruptures do not occur randomly along a fault zone. Rather, through time, particular segments of faults repeatedly rupture through several seismic cycles, suggesting the persistence of barriers to rupture propagation [12, 13]. The locations of fault segments and the boundaries between segments are probably physically controlled. Research is now underway to quantitatively evaluate the worldwide database Figure 2. Segmentation model for the of historical surface ruptures to deter- Wasatch fault zone. Stippled bands mine those geologic characteristics that define segment boundaries identified by are most diagnostic of segmentation for [15]; dashed bands are additional boun- use in assessing the future rupture daries interpreted by [16], behavior of a fault of interest [14], Some of the more important features to look for along a fault zone in assessing Once the segmentation of a fault zone can its segmentation appear to be: be estimated, one is able to estimate the length of rupture and, in turn, the maxi- Changes in the recency of fault slip mum magnitude of the segment(s) of the Changes in slip rate fault of most significance to the site. ChangeJ in the sense of slip Major bends or discontinuities in Work is continuing on the updating of the Large stepovers or double-bends empirical relationships between earth- Cross-faults or folds quake magnitude and fault dimensions. Changes in trace complexity The most common relationships used in

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

265 magnitude assessment are chose EARTHQUAKE RECURRENCE relating magnitude to: 1) rupture length, 2) rupture area, 3) maximum Because the seismic sources in a typical displacement per event, 4) total fault western U.S. PSHA are faults, earthquake length, and 5) seismic moment. With the recurrence must be estimated for indi- documentation of recent historical vidual faults. In nearly all cases, the earthquake ruptures as well as further historical seismicity record, while pos- study of older events, new empirical sibly adequate to define the recurrence relationships are being published ([17] rates for smaller magnitude earthquakes, see Figure 3). Of course, in order to is inadequate to define fault-specific use any of these relationships to pre- recurrence for larger magnitude events. dict the maximum magnitude for a fault In fact, in very few cases around the of interest, data must be developed on world have we observed the re-rupture of the expected fault parameter (rupture the same segment of a fault. This recog- length, etc.). Assuming that such data nition has led to the emerging field of can be developed, typically we make paleoseismology whose focus is the iden- estimates of magnitude using multiple tification of prehistoric earthquakes, published relationships and a variety of their sizes, and their timing. By using fault parameters. Techniques have been the geologic record, particularly the proposed for using the statistics of the fault-specific record, to extend the his- relationships to arrive at a consensus torical record, we have a much stronger estimate of maximum magnitude [18]. basis for assessing the recurrence characteristics required for PSHA.

For certain well-studied faults, such as the San Andreas and Wasatch faults, geologic studies have successfully identi- fiiiiii t i i i 11 in i i i i I MM fied paleo-earthquakes and their timing. For example, exploratory trenches across Mw Regression Line normal faults like the Wasatch fault 8 89 Data Points allow identification of individual scarp- forming earthquakes and associated collu- vial wedges. The displacement associated with each earthquake can also be assessed 0)c 7 •o as well as the timing, if geologic condi- 3 tions so allow. Detailed geologic studies and dating of prehistoric earthquakes have been done along the part of the San Andreas fault that ruptured in 1857 (e.g., [19]), as shown in Figure 4. Recurrence intervals derived from data of Mw « 4.14 + 0.99 log (Rupture Area) the type shown in Figure 4 can be quite useful to a seismic hazard analysis, i i i iiml i i i i 11 ill i i i i i ml except that the magnitude of the prehis- 10 100 103 10+ toric earthquakes must be estimated. Progress has been made in this area as Rupture Area (km2) well, as discussed below.

Statistical studies of the historical seismicity of large regions have shown Figure 3. Empirical relationship bet- that the number of earthquakes is ween fault rupture area and moment mag- exponentially distributed with earthquake nitude [17] magnitude. The general form of this

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

266 Oott(AO) the same size or "characteristic" earth- ZOOOr quakes having a relatively narrow range of magnitudes at or near the maximum [15, 20], The characteristic earthquake hypothesis was developed by observations of the displacement-per-event associated with individual faulting events at points along the Wasatch and south-central San Andreas faults. These data showed that at a point the amount of displacement during successive surface-faulting earthquakes remained essentially constant. A major implication of this conclusion is that earthquake recurrence on an individual fault does not conform to an exponential recurrence model, but rather one that has a non-constant b-value, such as that shown in Figure 5.

B C 0 F I N R T V X Z - older EARTHQUAKE younger — 10° Figure 4. Estimates of the dates of paleo-earthquakes (95% confidence intervals) at Pallet Creek along the south- ,o-. central San Andreas fault (from [19])

~~2 10-2 recurrence model is the familiar Gutenberg-Richter frequency-magnitude relationship: Z 10-3~~

log N(ra)= a - bm (1) 10-4 where N(m) is the cumulative number of earthquakes greater than or equal to m, and a and b are constants. In the 10-5 absence of fault-specific recurrence 4.0 5.0 6.0 7.0 B.O data, it has commonly been assumed that Magnitude, m the exponential recurrence model is appropriate to individual faults as it is to regions. However, recent geologic studies of late Quaternary faults Figure 5. Comparison of recurrence strongly suggest that the exponential curves based on exponential magnitude recurrence model is not appropriate for distribution (solid) and a characteristic expressing earthquake recurrence on earthquake distribution (dashed). Both individual faults or fault segments. curves assume the same fault slip rate, Instead, these studies are suggesting maximum magnitude, and b-value (from that many individual faults and fault [21]) segments tend to generate essentially

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

267 At the present time, the type of geo- available (Figure 5). The model was logic recurrence interval data developed tested against the historical seismicity, for the Wasatch and San Andreas faults geologic recurrence intervals, and slip is not generally available for most rate data of the Wasatch and San Andreas faults in the western U.S. Thus, PSMA faults. Note that the basic differences methodology must incorporate another between the exponential and characteris- type of geologic recurrence data that is tic models are that, for a given slip more generally available. Fault slip rate, the characteristic model predicts rates, derived from the amount of slip significantly fewer moderate and small- that has occurred over a geologically- magnitude earthquakes, and slightly more defined interval, offer the advantage large-magnitude events. Depending on the over historical seismicity data of span- distance from a site of interest to the ning several seismic cycles of large fault, this difference in recurrence earthquakes on a fault and they can be models can have a significant effect on used to estimate average earthquake the seismic hazard results. recurrence rates. Slip rates are determined simply by assessing the amount of Earthquake recurrence is usually fault displacement of a geologic unit expressed as the average number of events having a known age. Typically, the best per year and hazard is usually expressed geologic units (either stratigraphic as the annual probability of exceeding units or geomorphic surfaces) for some level of ground motion. Increasing- assessing .slip rate for recurrence pur- ly, however, efforts are being made to poses are late-Quaternary or Holocene develop "real-time" or "time-dependent" units. Assessing slip rates over rela- hazard estimates that express the proba- tively young units will avoid averaging bility of ground motions in, say, the out long-term changes in the slip rate next five years as opposed to any five due to, perhaps, regional changes in year period. To do this, simple models tectonic stresses. The assumption is that consider the earthquake process to that the fault slip rate reflects the be one of strain accumulation and release average rate at which strain energy along faults can be used. For example, (seismic moment) accumulates along the if the fault slip rate is assumed to fault and is available for release. reflect the average rate of strain accum- Slip rates are readily translated into ulation between large earthquakes, and seismic moment rates (e.g., [22]), given the size of the earthquake is related to the three-dimensional geometry of the the slip, then the time to the next fault. earthquake is directly related to the time since the most recent earthquake Once a seismic moment rate has been cal- [26]. A "renewal model" can account for culated for a fault, it must be parti- these dependencies (e.g., [10]). A tioned into earthquakes of various magn- recent example of the use of time-depen- itude according to a recurrence model. dent recurrence models was a study of the Several authors (e.g., [22, 24, 25]) probabilities of large earthquakes on the have developed relationships between San Andreas fault system for use in fault slip rate and earthquake recur- regional planning [27]. rence rates assuming an exponential recurrence model. However, as discussed GROUND MOTIONS previously, the characteristic earthquake model is probably more appropriate The assessment of ground shaking hazard for individual faults than the exponen- in the western U.S. (WUS) has benefited tial model. Youngs and Coppersmith [21] from an extensive data base of strong developed a generalized characteristic motion recordings. These have been used earthquake recurrence model that can be to develop ground motion attenuation used when fault slip rate data are relationships that allow reliable

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

268 estimation of the expected levels of strong ground shaking that can occur ac distances of about 10 km or more from earthquake ruptures in as large as Ms 7.5 incorporating the effects of site classification and type of faulting. Joyner and Boore [28] present a detailed summary of the available attenuation relationships appropriate for predicting ground motion in the WUS.

There are situations, however, when the La Villila NS - Mo«ico Valparaiso Nf'SM N7OE - Cttilo Keccfded available empirical attenuation rela- Simulalmn ^ j tionships may not be applicable or require significant extrapolation beyond the available data base. Examples 10 I include evaluations for a site located a few kilometers of the San Andreas >\ 7 fault or a site directly above the postulated seismogenic subduction zone ll.l.l along the coast of Washington and .05 .1 .2 .5 12 5 .05 .1 .2 .5 12 5 l3cnntl (we) I'trwil (see) Oregon. Hazard assessment in these situations has benefited from the advances made in understanding of the dynamics of earthquake rupture and Figure 6. Comparison of 5%-damped numerical modeling of earthquake ground response spectra for recorded and simu- motions. Investigators are now able to lated motions from 1985 My 8 Valparaiso, simulate strong ground motion in the Chile and Michoacan, Mexico earthquakes frequency range of interest for engi- (from [29]) neering design. For example, Figure 6 compares response spectra for observed and simulated ground motions at sites are being implemented in PSHA. It is directly above Mu 8 subduction zone important to recognize that uncertainty earthquakes. Numerical simulations such exists in applying these new methods as these can be used to extend the empi- arising from alternative hypotheses rical attenuation relationships to and/or data interpretations. Modern PSHA closer distances and/or larger magni- can readily and should accommodate these tudes. As one example, Youngs et al. uncertainties formally into the analysis [29] have used numerical simulations of to provide an assessment of the uncer- large subduction zone earthquakes to tainty in the computed hazard. An develop attenuation relationships for assessment of uncertainty is an important predicting response spectra for rocks aspect of the use of PSHA results in sites directly above subduction zone design and safety evaluation. earthquakes as large as M,, 9. These relationships have been used in a PSHA REFERENCES for a site in western Washington [8]. [1] C.A. Cornell, "Engineering seismic risk analysis," Bulletin of the Selsmolotlcal Society of America. CONCLUSION v. SB, pp. 1583-1606, 1968. [2] R.R. Youngs, K.J. Coppersmith, M.S. Power, and F.H. We have described many of the recent Swan, III, "Seismic hazard assessment of the Hanford region, eastern Washington state," advances in our understanding of the Proceedln«s DOE Natural Phenomena Hazards Mitiga- earthquake generation process in the tion Conference, p. 169-176, 1985. western U.S. and have indicated how they

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~269 K.J, Coppersmith and R.R, Yountis, "issues regard- 116] M,N. Machncm, S.F. Personlus, and A.R. Nelson, ing earthquake source characterisation and seismic 1~~

D.P. Schwartz, "Geology and seismic hazards: [24] J.G. Anderson, "Estimating the seismicity from moving Into the 1990's," Earthquake Eng. Soil geological structure for seismic-risk studies," Dynamics II - Recent Advances In Ground Motion Bulletin of the Selsmologlcal Society of America, Evaluation. ASCE Geotechnical Special Publication v. 69, pp. 163-168, 1979. 20, pp. 1-42, 1988. [25] J.G. Anderson and J.E. Luco, "Consequences of slip [12] K. Aki, "Characterization of barriers o . an earth- rate constraints on earthquake occurrence rela- quake fault", Journal of Geophysical Research, v. tions ", Bulletin of the Seisroologlcal Society of 84, no. Bll, pp. 6140-6148, 1979. America, v. 73, no. 2, pp. 471-496, 1983.

[13] K. Aki, "Asperities, barriers, and characteristic earthquakes," Journal of Geophysical Research, [26] K. Shimazaki and T. Nakata, "Time-predictable v. 89, pp. 5867-5872, 1984. recurrence model for large earthquakes," Geophys- ical Research Letters, v. 7, pp. 279-282, 1980. [14] P. Knuepfer, "Implications of the characteristics of end-points of historical surface fault ruptures [27] Working Group on California Earthquake Probabili- for the nature of fault segmentation,1* U.S. ties, Probabilities of Large Earthquakes Occurring Geological Survey Conference on Fault Segmentation in California on the San Andreas Fault. U.S. and Controls on Rupture Initiation and Geological Survey Open-File Report 88-398, 1988. Propagation, U.S. Geological Survey Open-File Report, pp. 193-228, 1989. [28] W.B. Joyner and D.H. Boore, "Measurement, characterization, *nd prediction of strong ground [15] D.P. Schwartz and K.J. Coppersmith, "Fault motion," Proceedings of Earthquake Engineering i behavior and characteristic earthquakes: Examples Soil Dynamics II. ASCE, Park City, Utah, June 27- from the Wasatch and San Andreas faults," Journal 30, 1988. of Geophysical Research, v. 89, pp. 5681-5698, 1984. [29] R.R. Youngs, S.M. Day, and J.L. Stevens, "Near field ground motions on rock for large subduction earthquakes," Proceedings of Earthquake Engineering & Soil Dynamics II. ASCE, Park City, Utah, June 27- 30, 1988.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 270 Session 9 Seismic Hazard~~

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

~~271 EARTHQUAKE RECURRENCE RATE ESTIMATES FOR EASTERN WASHINGTON AND THE HANFORD SITE~~

~~A. C. Rohay Pacific Northwest Laboratory Richland, Washington~~

ABSTRACT The historical and instrumental records of earthquakes were used to estimate earthquake recurrence rates for input to a new seismic hazard analysis at the Hanford Site in eastern Washington. Two areas were evaluated, the eastern Washington region and the smaller Yakima Fold Belt, in which the Hanford Site is located. The eastern Washing- ton historical earthquake catalog was incomplete for Modified Mercalli Intensity (MMI) IV, and was judged to be complete since 1905 for MMI V, and since 1390 for MMI VI. Only one MMI VII earthquake was reported in the last 100 years. A recurrence curve for the historical earthquakes was calculated using the maximum likelihood method. The Gutenberg-Richter intensity-magnitude relationship was found to fit the subset of earthquakes for which both intensity and magnitude were reported. The slope of the recurrence curve was found to be -1.15 after correcting for the width of the magnitude groups that result from the intensity-to-magnitude conversion. Instrumentally detected earthquakes from 1969 to the present were used to supplement the historical earthquake data. For earthquakes that had magnitudes between 3 and 5, the b-value ranged from -1.07 to -1.12. From this analysis, we conclude that t" e recurrence relationship,

log(N) = log(0.5) - 1.15 (ML - 4) v/here N is the cumulative number of earthquakes in 100,000 km2 v/ith magnitudes greater than or equal to ML, is appropriate input for estimating seismic hazards at the Hanford Site. No significant differences were found between the area-normalized rates of earthquakes calculated from the historical and the instrumental data, nor between the smaller Yakima Fold Belt and the eastern Washington region.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

272 DESCRIPTION OF THE HISTORICAL 2) to estimate the magnitude EARTHQUAKE CATALOG recurrence from the intensities. The catalog of earthquakes produced by Woodward-Clyde Consul- EVALUATION OF THE HISTORICAL tants (Washington Public Power EARTHQUAKE CATALOG Supply system 1981, Table 2.5-5; Counts of the number of earth- DOE" 1988*, Table 1.4-2) has been quakes for each intensity are used to develop recurrence rate shown in Figure 4 in 5-year incre- estimates for eastern Washington. ments and also grouped into over- An earlier version of this paper lapping 15-year periods. Figure 5 (Rohay, 1989) describes the cata- plots the number of earthquakes log and the methods used in more per year in each of these 15-year detail. periods. It appears that there Figure 1 (from DOE 1988, Fig- are too few intensity IV earth- ure 1.4-5) shows the locations of quakes to be considered repre- every earthquake with either a sentative of the true number. Modified Mercalli Intensity (MMI) The total numbers and annual of IV or greater or a magnitude rates of intensity V and VI earth- (on any of several scales) greater quakes per year are tabulated in than 3 that were reported prior to Table 1, averaged over 50- to 1969 in the area of eastern Wash- 80-year periods that end in 1970 ington. The smaller box shows (no intensity V earthquakes the area of the Yakima Fold Belt. occurred in 1969 or 1970). The The Hanford Site is located in the rates of intensity V earthquakes eastern half of the Yakima Fold decrease significantly for periods Belt. These regions represent longer than 65 years, the pre- possible choices for characteriz- viously estimated period of coming the seismo-tectonic environ- plete reporting (Rohay, 1989). ment of the Hanford Site. Annual rates of intensity V and Figure 2 (from DOE 1988, Fig- VI earthquakes were calculated ure 1.4-6) shows the locations of for the 60- or 65-year period every earthquake reported since prior to 1970 (either period gives 1969 with a magnitude greater than approximately the same rates). 3.0 for the same regions. Nearly This assumes that both records are ail of these earthquakes were lo- complete back to 1905 or 1910. cated by the eastern Washington Although these estimates are seismic network that was installed slightly lower than they would be in 1959. Because network coverage if estimated for a 50- or 55-year has not extended to the eastern period, the record of intensity V and southern boundaries of the earthquakes does appear to be region in Figure 2, the area indi- complete for 65 years. It does cated by the larger box represents not seem reasonable to assume a the area of complete detection by shorter period for the intensity the seismic network at the magni- VI data? in fact, a longer period tude 3 level since 19 69. (80 years) gives n rate that is Figure 3 plots intensity versus only slightly different. Because magnitude for the earthquakes only one intensity VII earthquake shown in Figure 1. The Gutenberg- has occurred in the region since Richter relationship appears to 1885 (when intensity IV, V, and VI fit rhe data best, and was used earthquakes were first reported), 1) to estimate intensities from the rate of intensity VII earth- magnitudes for all earthquakes quakes is estimated to be 1 in that occurred before 1969 and then 100 years, or 0.01.

~~Second DOE \atural Phenomena Hazards Mitigation Conference - 1989~~

~~273 I1«~~

~~' r i>~~

~~na- 117=~~

MAGNITUDE INTENSITY EXTENT OF THE COLUMBIA RIVER a 3 0 34 a 45 49 IV BASALT GROUP 0 35 39 SO 5 4 A V a 100 KILOMETERS a 4 0 4 4 a >5 5 A VI A IX PS8609267A FIGURE l. Historical Seismicity of the Columbia Plateau and Surrounding Areas (from DOE 1988). All earthquakes between 1850 and 1969 with a Modified Mercalli Intensity (MMI) of IV or larger or with a magnitude of 3 or greater are shown. The Yakima Fold Belt is enclosed by the smaller box (118.75°-121.0° W, 45.75°-47.25° N).

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 274 117 lit~~

~~EXTENT OF THE COLUMBIA RIVE BASALT GROUP~~

~~121-~~

~~MAGNITUDE 0 SO 100 KILOMETERS EXTENT OF THE O 30 34 Q 45 4 9 | COLUMBIA RIVER 1 1 BASALT GROUP D 3S 3» Q] i50 1 1 0 SO MILES • 4 0 44~~

FIGURE 2. Instrumental Seismicity of the Columbia Plateau (from DOE 1988). All earthquakes between 1969 and 1986 with magnitudes greater than 3 are shown. The larger box (117.5°- 121.5s W, 45.5°-48.5° N) represents the region covered by the seismic network since 1969. The Yakima Fold Belt is C enclosed by the smaller box (118.75 -121.0° Wf 45.75°- 47.25° N).

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~275 j 1~~

~~6 0~~

~~SS - A /~~

~~5" 50 / A~~

~~45 A~~

~~A 40 /~~

A VI VII MMI FIGURE 3. Relationship Between the Maximum Observed Modified Mercalli Intensity (MMI) and Richter Magnitude (ML) for Eastern Washington. The relationship of Gutenberg and Richter (195P) , FIGURE 5. Fifteen-Year-Average ML = 1 + 2/3 MMI, appears to fit Annual Rates (n/yr) of Earthquake these data. Occurrence in Eastern Washington (206,660 km?). Modified Mercalli Intensity IV (circles), V (squares), and VI (triangles) are shown. Points are plotted at the center of each 15-year interval. Open squares represent the result 8 5 4 2 of removing three MMI V aftershocks of the 1936 MMI VII Milton-

s 11 Freewater earthquake. Figure 6 plots the cumulative 1 i • 3 1 3 recurrence rates per 100,000 km2 • 3 3 for the data described above. 1 i 3 1 This plot uses the Gutenberg- •< Richter intensity-magnitude rela- FIGURE 4. Number of Earthquakes tionship (Figure 3). The data in 5- and 15-Year Periods in points at intensities V and VI Eastern Washington. imply a slope (the b-value) of - 1.07, although this value must be RECURRENCE CURVES corrected for the width of the Table 2 gives incremental and magnitude groups used in plotting cumulative recurrence rate esti- the data. mates (n/yr and N/yr, respec- Sources of uncertainty in the b- tively) for the eastern Washington value include 1) the intensity- area (Figure 1) and normalizes the magnitude relationship, 2) the cumulative rates to a 100,000-km2 length of the complete catalog, area. and 3) whether aftershocks should Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

276 Years before 197Ci MMI = V MMI = VI MMI = VII (T) n n/vr s . ei _LJ~*../._ j*i n/yr s. i n n/yr s. e.-t-S 50 39 0.73 0.12 10 0.20 0.06 55 43 0.78 0.12 10 0.18 0.06 60 4 5 0.75 0.11 10 0. 17 0.05 65 49 0.75 0.11 11 0.17 0.05 70 49 0.70 0.10 11 0.16 0.05 75 50 0.67 0.09 11 0.15 0.05 80 50 0.62 0.09 13 0.16 0.04 100 1 0.01

(a) Standard error is estimated from the n-l/2 relationship, approximately valid for n > 10. (b) Standard error is based on Weichert (1980, Figure 1) for n = 1; the 68% confidence bounds are factors of 3 .3 and 0.2 of the average rate. TABLE 2. Cumulative Recurrence Rates for Eastern Washington Intensity Incremental Cumulative Area-normalized (MMI) Rate (n/vr) Rate fN/yr) N/vr/100.000 km* V 0.75 + 0 .11 0.93 + 0.12 0.45 ± 0.06 VI 0.17 + 0 .05 0.18 + 0.05 0.087 + 0.03 VII 0.01 0.01 0.005 or should not be removed from the eastern half of the Yakima Fold catalog. The largest uncertainty Belt. The rate (per unit area) of is caused by scatter in the rela- intensity V earthquakes is tionship between intensity to slightly lower in the eastern magnitude. The recurrence rates half, but the rate of intensity VI for intensities V and VI decrease earthquakes is unchanged relative slightly for periods from 55 and to the entire Yakima Fold Belt. 75 years. Removal of aftershocks On the basis of these analyses, (if appropriate) would slightly it was concluded that the earth- reduce the rate of intensity V quake recurrence rate chosen to earthquakes and would tend to represent the Hanford Site (in increase the b-value. the eastern half of the Yakima Recurrence rate data for the Fold Belt) cannot be distinguished Yakima Fold Belt are also plotted from the rate estimated from the in Figure 6. For this smaller entire eastern Washington area. region, it was assumed that the The two assumptions, 1) that the catalog was complete from 1888 to recurrence rate is uniform 1989, a 100-year period that in- throughout the Yakima Fold Belt cludes two intensity VI earth- and 2) that the recurrence rates quakes and thirteen intensity V in the east and west halves of the earthquakes (Rohay, 1989) . Fig- Yakima Fold Belt are different, ure 6 also shows recurrence data give recurrence rates that are for earthquakes in just the higher and lower, respectively,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~277 1.0 Utsu's formula for determining the b-value is~~

(1) X tanh M ave~Mmin 0.1 where d is the width of the magnitude groups (equal to 0.67 as \| calculated from the Gutenberg- N \.b-1.07 Richter relationship), Mave is the average magnitude of the earthquakes, and Mmin is the minimum \. magnitude of the lowest group 0.01 \ (magnitude 4.0 in this case). Using this formula results in a I—1 1 1 corrected b-value of -1.15. A method of replotting the MMI recurrence data was developed that V VI VII reproduces this correction exact- 1 ! 1 i L ly, properly adjusts the fitted 4.0 5.0 6.0 ML line downward, and does not depend on equal-width magnitude ranges. FIGURE 6. Historical Earthquake The first step involves assuming Recurrence Rates Normalized to that half of the earthquakes are 100,000 km2. Cumulative recur- larger than the center of each rence rates for eastern Washington magnitude range. In a cumulative (squares) are shown with their recurrence curve, then, half of standard errors. The slope the earthquakes for a given range (b-value) of the line, fitted to should be added to the total num- the points at MMI V and VI, is ber of earthquakes in all larger- -1.07. The one MMI VII earthquake magnitude ranges and replotted has a known magnitude of 5.75 + below the center point (Figure 7). 0.2. Cumulative recurrence rates This approach would be correct if for the entire Yakima Fold Belt this point represented the median (solid triangles) and for the magnitude of each intensity group. eastern half of the Yakima Fold For the few observations availa- Belt (open triangles) are not sig- ble here that can be used to nificantly different from the rate estimate an intensity-magnitude for eastern Washington. relationship, the Gutenberg- Richter relationship appears to than the recurrence rate for predict the median magnitude for eastern Washington. Therefore, if intensity V and VI (see Figure 3) these two assumptions are equally fairly well. Using this assump- weighted, the eastern Washington tion, the b-value is estimated recurrence rate provides a rea- from points replotted below the sonable average. center of the range. The b-value determined, -1.15, agrees exactly CORRECTION FOR WIDTH OF MAGNITUDE with the slope predicted by the GROUPS Utsu relationship. Utsu (1966), weichert (1980), Correcting for the magnitude and Bender (1983) have given grouping depends on whether the formulas for data grouped into values predicted by the intensity- equal-width magnitude ranges. magnitude relationship represent

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

278 set where the median magnitude value is located within each intensity group, assuming that the magnitude distribution within each group is exponential. The b-value determined in the first step is used to estimate how far the point should be offset to the left to account for the more frequent occurrence of earthquakes in the left half of the magnitude range. The position within the range x (above the lower limit) is calcu- o.ot - lated from

1 - = (0.5)(1 - 10"bd) (2) This point represents the median of a magnitude group whose dis- 0.001 tribution is governed by the 4.0 b-value. The value determined for d « FIGURE 7. Correction of Histori- 0.67 and b = -1.15 is x = 0.202 cal Earthquake Recurrence for (compared to 0.333 for the center Magnitude Grouping. All symbols of the range), and this correction are the same as in Figure 7, is shown in Figure 7. This step except for correction to the moves the points an equal amount lower limit of each magnitude to the left and does not affect range (open squares). The cor- the b-value because, in this case, rections reflect the assumption the magnitude distributions have that centers of intervals repre- been assumed to be the same for sent the median value of the each intensity group. If each groups (open circles) and the intensity group had a different assumption that distribution magnitude distribution, the median within the groups is exponential value would be offset a different with a b-value of -1.15 (solid amount from the center for each circles). group, and a small adjustment to the b-value would result. Such the median, the mean, or some adjustments lead to an iterative other estimate of the distribution process of updating the exponen- of magnitudes for that intensity. tial magnitude distribution within This distribution was estimated groups according to the distribu- from a small subset of the data tion of the whole data set. because the distribution of the Because of the uncertainty in whole data set is unknown. If the estimating the median magnitude data within each group were dis- within each of the intensity tributed exponentially with the groups, two alternative recurrence same b-value as the whole data relationships are shown in Fig- set, there would be more earth- ure 7. The upper line corresponds quakes in the lower half than in to the assumption that the the higher half of the range. Gutenberg-Richter relationship The second step in replotting is predicts the median magnitude in to estimate for the whole data each group. The lower line

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

279 corresponds to the assumption that MC-ML comparison was made only the earthquakes within each group for earthquakes smaller than mag- are distributed exponentially and nitude 4 and so may not be appli- that the median magnitude is cable above magnitude 4. Both the smaller than the central value of Mc~ML anc* the MMI-ML relationships the range. These two alternative have such large uncertainties (on relationships predict that the the order of 0.3 units) that the average recurrence rate (in apparently precise agreement be- 100,000 km2) for magnitudes >4 is tween the instrumental and his- 0.45/yr to 0.65/yr. torical recurrence rates in eastern Washington may merely be RECURRENCE RATE FOR INSTRUMENTAL coincidence. MAGNITUDE DATA Since 1969, the eastern Wash- 10.0 ington seismic network has provided a catalog of earthquakes with 5 magnitude precision of 0.1 units. However, magnitude accuracy is probably only about 0.2 to 0.3 units. The post-1969 earthquakes larger than coda-length magnitude 3,0 have been listed by DOE (1988, Table 1.4-2). The numbers of earthquakes in each magnitude group for the regions shown in Figure 2 are plotted in Figure 8. The catalog of earthquakes with magnitudes below 4.0 may be incomplete in the Idaho, Oregon, and northeastern Washing- ton portions of Figure 2. The instrumental recurrence rates for eastern Washington and 0.05 - the Yakima Fold Belt are similar to the rates determined from the historical data. High b-values 0.02 are consistently indicated, and the rates of earthquakes greater than magnitude 4 are comparable. FIGURE 8. Earthquake Recurrence The apparent agreement between the Curves, 1969-1986. Cumulative historical and instrumental earth- annual recurrence rates are based quake recurrence rates depends on on coda-length magnitudes (Mc) for the relationship between local three regions, normalized to a 2 magnitude (ML) and coda-length 100,000-km area: Yakima Fold Belt magnitude (Mc). The NRC (1982) (triangles), eastern Washington has suggested that Mc is larger (filled circles), and eastern than M^ by 0.3 units. However, Washington and Oregon and northern the original investigators of the Idaho (open circles). The slopes MC-ML relationship (UWGP 1979) (b-values) determined using the indicated that no correction to Mc maximum likelihood method are was substantiated because of the shown at lower right. large scatter in the data. The

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

280 CONCLUSION University of Washington Geophys- In summary, this analysis has ics Program (UWGP). 3 979. Annual determined the following cumula- Technical Report 1979 on Earth- tive annual recurrence rates, nor- quake Monitoring of the Hanford malized to an area of 100,000 km2, Region. Eastern Washington. Geo- where N is the number of earth- physics Program, University of quakes greater than magnitude 4: Washington, Seattle, Washington.

Data N b-value U.S. Department of Energy (DOE). Instrumental Data 1988. Consultation Draft Site Eastern Washington 0.45 -1.07 Characterization Plan. Reference Yakima Fold Belt 0.60 -1.12 Repository Location. Hanford Site. Historical Data Washington. DOE/RW-0164, U.S. Eastern Washington 0.45 -1.15 Department of Energy, Washing- Yakima Fold Belt 0.52 ton, D.C. East half Yakima Fold Belt 0.3 5 U.S. Nuclear Regulatory Commission (NRC). 1982. Safety Evaluation From this analysis, we conclude Report Related to the Operation of that the recurrence relationship WPPSS Nuclear Project No. 2. NUREG-0892, Supp. No. 1, U.S. log(N) = log(0.5) Nuclear Regulatory Commission, - 1.15 (ML - 4) (3) Washington, D.C. is appropriate input for estimat- Utsu, T. 1966. "A Statistical ing seismic hazards at the Hanford Significance Test of the Differ- Site. ence in b-Value between Two Pacific Northwest Laboratory is Earthquake Groups." J. Phys. operated for the U.S. Department Earth 14:37-40. of Energy by Battelle Memorial Institute under Contract DE-AC06- Washington Public Power Supply 76RLO 1830. System. 1981. Final Safety Analysis Report for Washington REFERENCES Nuclear Plant 2. Vol. 2, Sect. Bender, B. 1983. "Maximum Like- 2.5.2 Amendment 18, pp. 2.5-220 to lihood Estimation of b-Values for 2.5-385. Washington Public Power Magnitude- Grouped Data." Bull. Supply System, Richland, Washing- Seisin. Soc. Am. 73:831-851. ton. Gutenberg, B., and C. F. Richter. Weichert, D. H. 1980. "Estima- 1956. "Earthquake Magnitude, tion of the Earthquake Recurrence Intensity, Energy, and Accelera- Parameters for Unequal Observation tion." Bull. Seism. Soc. Am. Periods for Different Magnitudes." 46:105-145. Bull. Seism. Soc. Am. 70:1337- 1346. Rohay, A. C. 1989. Earthquake Recurrence Rate Estimates for Eastern Washington and the Hanford Site. PNL-6956, Pacific Northwest Laboratory, Richland, Washington.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~281 GEOLOGIC ASPECTS OF SEISMIC HAZARDS ASSESSMENT AT THE IDAHO NATIONAL ENGINEERING LABORATORY, SOUTHEASTERN IDAHO~~

Richard P. Smith, Idaho National Engineering Laboratory, EG&G Idaho, Inc., P.O. Box 1625, Idaho Falls, ID 83415 William R. Hackett and David W. Rodgers, Department of Geology, Idaho State University, Pocatello, ID 83209

~~ABSTRACT~~

The Idaho National Engineering Laboratory (INEL), located on the northwestern side of the Eastern Snake River Plain (ESRP), lies in an area influenced by two distinct geologic provinces. The ESRP province is a northeast-trending zone of iate Tertiary and Quaternary volcanism which transects the northwest-trending, block-fault mountain ranges of the Basin and Range province. An understanding of the interaction of these two provinces is important for realistic geologic hazards assessment. Of particular importance for seismic hazards analysis is the relationship of volcanic rift zones on the ESRP to basin-and-range faults north of the plain.

The Arco Rift Zone, a 20-km-long belt of deformation and volcanism on the plain just west of the INEL, is colinear with the basin-and-range Lost River fault. Recent field studies have demonstrated that Arco Rift Zone deformation is typical of that induced by dike injection in other volcanic rift zones. The deformation is characterized by a predominance of dilational fissuring with less extensive development of faults and grabens. Elongate positive magnetic anomalies, closely associated aligned volcanic vents and eruptive fissures, and local perturbations in groundwater flow suggest that basalt dikes are present beneath the deformational features. Although the Lava Ridge-Hells Half Acre Rift Zone, which crosses the INEL south of the Lemhi fault, is not as well preserved or as well studied, its exposed features are similar to those of the Arco Rift Zone.

Available K-Ar ages of lavas in the Arco Rift Zone suggest that volcanism and associated deformation are Pleistocene in age. Cumulative vertical displacements over the past 0.6 Ma are an order of magnitude lower than those associated with the Arco Segment of the Lost River fault to the northwest.

The evidence suggests that the northeast-directed extension that produces the block fault mountains of the Basin and Range is expressed by dike injection and volcanic rift zone development in the ESRP. Seismicity associated with dike injection during rift zone development is typically of low magnitude and would represent only minor hazard compared to that associated with the block faulting. Since the ESRP responds to extension in a manner distinct from basin-and-range faulting, it is not appropriate to consider the volcanic rift zones as extensions of basin-and-range faults for seismic hazard analysis.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

282 INTRODUCTION INUKX MAP The interaction of basin-and-range (B&R) with East- . IDAHO ern Snake River Plain (ESRP) tectonism has received f I , INEL Boundary the recent attention of several researchers. Such ! J vUmhl Fa(ih '.s studies [1,2,3] show that passage of the Yellowstone "j hotspot to produce the ESRP has also influenced the movement histories of B&R faults located to the north and south of the plain. The nature of that influence is Lost Rlv«r Fault/ HOVVeQ j"' generation or activation of B&R faulting in front of and adjacent to the hotspot, and gradual migration of fault activity away from the plain in the wake of the hotspot. This has left a gradually widening zone of waning fault activity with distance southwestward from Yellowstone, and a zone of latest Quaternary fault activity that fans East Butte 1 M away from the ESRP margins. Although it must be Middle Butte " related somehow to crustal uplift and profound %, t Big Southern Butt* changes in the thermal structure of the lithosphere with passage of the hotspot, the precise nature of fault control is a subject of debate. Figure 1. Map of volcanic rift zones near INEL.

Our study of seismic hazards associated with ESRP and B&R tectonism focuses on one aspect of the inter- Ground Deformation action of the two provinces: the relationship of ESRP Ground deformation in the Arco rift consists of volcanic rift zones to the basin-and-range activity. fissures and faults that cut basalt lava flows at the Since the Arco rift zone and the Lava Ridge-Hells Half surface. The fissures are open vertical cracks whose Acre rift zone are colinear with the B&R Lost River and walls have experienced only dilational opening, with no Lemhi faults respectively, there has been suspicion apparent dip slip or strike slip displacement (Figure 2). that those faults, with their associated seismic hazards, Individual fissures are up to 3 km long, with dilational extend onto the ESRP. The Arco rift zone was selected offsets of up to 1 meter, and observable depths of up for detailed study because it is the best exposed and to 6 meters. Fissures make up 80% of the total length most well developed of the rift zones in the vicinity of of ground deformation features in the Arco rift zone the INEL [4,5]. DESCRIPTION OF THE ARCO RIFT ZONE The Arco rift zone can be traced for a total distance of more than 20 km and extends from a point 6 km south of Arco to the Big Southern Butte area (Figure 1). The northwesternmost manifestation of the rift zone occurs within 3 to 4 km of the northwest edge of the ESRP. The Arco rift is characterized both by ground deformation and by constructional volcanic features. This suggests a close genetic association between deformation and volcanism as has been described in other volcanic rift zones.

~~Figure 2. Aerial view of open fissure.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~283 GEOLOGIC MAP OF THE ARCO RIFT~~

~~Volcanic Vent K/Ar Ag« Fissure R«f. [7,8]~~

~~Flexure Fault Scarp (Dltplactmtnt In mtttrt) Qs Quaternary Sediments 0 km 5~~

~~Quaking Aspen Big Southern Butte ButU~~

~~Figure 3. Geologic map of the Arco Rift Zone. (References [4,5,6]).~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

284 Faults are expressed on the surface as monoclinal flexures, in which slabs of basalt are draped over underlying faults, and as vertical scarps. The maximum observed vertical displacement on faults in the rift zone is about 10 meters, and the maximum fault length is 5 to 6 km (Figure 3). Most offsets are down-to-the- southwest, but down-to-the-northeast offsets are common (Figure 3). Both flexures and scarps cause vertical displacement of surface lava flows and the expression of faults commonly changes over short distances from flexures to scarps (Figure 4).

Fissures and faults occur in a branching, en echelon, overlapping pattern in a northwest-trending zone approximately 20 km long by 6 km wide. From the northwest end of the rift zone, the fissures and faults branch to the southeast into two subparallel zones, Figure 5. Aerial view of Box Canyon Graben. each about 0.6 km wide and separated by about 4 km of relatively undeformed surface rocks (Figure 3). Constructional Volcanic Features Constructional volcanic features are an important part of the Arco rift. They consist of aligned volcanic vents, elongate volcanic vents, and eruptive fissures. Several northwest-trending linear alignments of volcanoes occur in the rift. The one that extends southeast, from Pond Butte and the one that includes Teakettle Butte and Coyote Butte are the most notable examples (Figure 3).

A more common and more striking feature of the Arco Rift is the northwest elongation of many of the volcanic vents. The most obvious of these elongated vents are Tincup Butte, Teakettle Butte, the unnamed butte northeast of Teakettle Butte, Pond Butte, and Crater Peak (Figures 3 and 6). Figure 4. Aerial view of flexure and scarp.

Each of the branches contains a northwest-trending graben. The most obvious graben (Figure 5) occurs at the head of the Box Canyon of the Big Lost River in the southwesternmost branch of the rift. The floor of the Box Canyon graben has been downdropped up to 10 meters from the block to the northeast but only about 6 to 8 meters from the block to the southwest. The graben in the northeastern branch of the rift zone is even more asymmetric, with a vertical displacement along the northeast side of about 8 meters and along the southwest side of about 1 to 2 meters.

~~Figure 6. Aerial view of Teakettle Butte.~~

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 285 Some of the volcanic vents in the Arco Rift are MECHANISM OF FORMATION OF THE eruptive fissures. Coyote Butte consists of a conical RIFT ZONE main vent with a string of spatter cones and spatter In Iceland it has been demonstrated that fissuring ramparts extending for almost a kilometer to the north- and faulting is induced by emplacement of shallow west, along a rift-zone fissure. The same fissure dikes from a magma chamber beneath the Krana cal- extends for about 1,5 km to the southeast of the main dera into rift zones north and south of the caldsra [9]. cone and hosts minor eruptive spatter along much of Each dike-injection episode is accompanied by that length (Figures 3 and 7). Another example of an deflation of the caldera, inflation of the rift zones, seis- eruptive fissure is the vent that lies just to the southeast mic activity migrating down the rift zones, widening of of Pond Butte. fissures and creation of new fissures in the rift zone, and renewed volcanic and geothermal activity in the rift zone.

Recent theoretical derivations and experimental work [10,11] have furnished an understanding of the mechanism of formation of fissures and faults above shallow dikes in volcanic rift zones. This work shows that the injection of magma into dike systems at shallow depths below the surface generates stress fields that produce zones of fissuring and graben development on the surface (Figure 8). As the dike thickens and as additional dikes are injected, two zones of fissuring and faulting fan upward from the top of the dike and a zone of surface subsidence that commonly is expressed as a graben forms directly over the top of the dike. Figure 7. Aerial view of Coyote Butte. Comparison of the theoretically and experimentally predicted surface deformation to that actually observed AGE OF THE ARCO RIFT ZONE in volcanic rift zones [10,11] shows remarkable agreement. For instance, the magnitude of extension due to Fissures in the northwest part of the Arco rift near fissuring and vertical displacement due to faulting in the Box Canyon cut basalt lavas that have been radio- the Inyo Craters rift zone [11] closely matches that metrically dated at about 609 ±92 Ka and a iava flow which is predicted by the theoretical and experimental from Crater Peak (Figure 3) has been dated at 516±52 model. Drilling there has intersected a rhyolite dike of Ka [7]. These two ages furnish an older age limit for appropriate thickness and depth below the surface to the fissuring and show that volcanism from northwest- produce the observed ground deformation. elongated vents related to rift-zone development was in progress between 500,000 and 600,000 years ago. The mapped pattern of ground deformation and volcanic features in the Arco rift is also consistent with Radiometric age determinations have been made on dike-induced deformation. The magnitude of vertical lava flows from drill hole 77-1 (Figure 3, DH 77-1) [8], displacement in the graben faults is within the range of The uppermost of these lava flows has been inter- displacement observed at Inyo Craters and the pattern preted [6] to have erupted from Quaking Aspen Butte of fissures (Figure 3) is typical of that observed in other and to have flowed across the Arco rift zone to reach volcanic rift zones. The observed deformation is con- the drill hole site. Our work has shown that the lava sistent with that to be expected from two dikes that flow covers rift zone fissures in older lavas and partially diverge from each other near the northwest end of the buries the west end of the fissure eruption at Coyote rift and converge again near Teakettle Butte. A third Butte. Its determined age of 95±50 Ka, therefore, dike may be present beneath the Pond Butte area. imposes a younger age limit on the fissuring and volca- Based upon the geometry of the theoretical and exper- nism of the Arco rift. imental models (Figure 8), the depth to the top of these dikes should be <500 meters.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 286 The second indication that basaltic dikes are present beneath the Arco rift is the northwest-elongated positive aeromagnetic anomaly that exists over the rift [12], It is very similar in magnitude to, but smaller in size than, the one that exists over the north end of the Great Rift at Craters of the Moon (Figure 9).

Figure 8. Mastin and Pollard's (1988) idealized depiction of fault and fissure growth above a growing shallow dike (Modified from reference [11]). Figure 9. Aeromagnetic anomalies over the Great Rift and the Arco Rift Zone (Modified from Reference [12]). The evidence that such dikes exist at depth beneath the Arco Rift is threefold. The strongest evidence is the The third indication is related to perturbations in the existence of fissure eruptions and elongate, aligned southwestward flow of groundwater in the Snake River volcanic vents along the rift. The analogy with fissure Plain aquifer. At the Great Rift the water table elevation eruptions along the Great Rift and the presence of the drops steeply from appoximately 4400 feet above sea Craters of the Moon lava field at the north end of the level northeast of the rift to about 4100 feet southwest Great Rift is strongly compelling. These eruptions must of the rift [13], This 'damming' of the aquifer flow be supplied by flow of magma through dikes which (aquiclude) is interpreted to result from the presence of reach the surface in some areas and remain at depth in nearly vertical, relatively unfractured basalt dikes others. The areas of graben formation and faulting beneath the rift [14]. A similar but smaller magnitude correspond to areas where the dikes are deepest and perturbation of water table elevations occurs at the the areas of fissure eruptions correspond to areas Arco Rift and can be explained by a dike system similar where the dikes reach the surface. to that at the Great Rift. IMPLICATIONS FOR SEISMIC HAZARDS The crustal extension that affects southeastern Idaho is accomodated by block faulting in the B&R to the north and south of the ESRP. That block faulting, on

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 287 the north side of the plain, has produced north- with documented maximum earthquake magnitudes of northwest-trending mountain ranges with elevations as 4 to 5 associated with magmatism in Hawaiian rift high as 3600 meters and structural offsets of as much zones [19]. as 2,7 km within the past 10 Ma. Estimated rates of fault movement in the Arco and Thousand Springs CONCLUSIONS segments of the Lost River fault are 0.1 m/1000 years Although some ESRP volcanic rift zones are colinear and 0.3 m/1000 years respectively [15]. with normal faults in the adjacent B&R province, they have a distinctly different style of deformation and The fact that no large-displacement normal faults mechanism of formation. They cannot be considered offset lavas as old as 0.5 Ma on the ESRP indicates as extensions of B&R faults for seismic hazard analysis. that the mechanism of accomodation of extension is They form in a dilational manner, are associated with different on the plain. Fault slip rates as high as those dike injection and volcanic activity, and are not capable associated with the Lost River fault would have proof generating large B&R-style earthquakes. duced structural relief on surface lava flows as great as 50 to 150 meters if the mechanism were the same. The ACKNOWLEDGEMENTS vertical displacement of some surface lava flows is a This work was supported by the United States near-surface phenomenon and probably dies out at Department of Energy under DOE Contract No. DE- very shallow depths (Figure 8). AC07-76ID01570. We are grateful for thoughtful reviews by Ivan Wong, W. E. Harrison, and J. C. The style of deformation observed in the Arco rift Walton. H.T. Ore assisted with aerial photography. zone indicates that the mechanism of accomodation of extension in the ESRP is dike injection and volcanic rift REFERENCES zone formation, probably with associated seismicity [1] K.L. Pierce, W.E. Scott, and L. Morgan, 'Eastern ypical of that occurring in Icelandic rift zones. The Snake River Plain neotectonics: Faulting in the north-northwest trend of the rift zones is controlled by last 15 Ma migrates along and outward from Yel- the northeast-southwest direction of extension, and the lowstone hotspot track," Geological Society of location of some of the rift zones seems to be con- America. Abstracts with Programs, vol. 20, no. 6, trolled by the positions of adjacent B&R faults. Exten- p. 463, March 1988. sion is almost entirely accomplished, however, by dike injection in the rift zones. [2] M.H. Anders, J.W. Geissman, LA. Piety, and J.T. Sullivan, "Parabolic distribution of circumeastern At least three seismogenic mechanisms could have Snake River Plain seismicity and latest Quater- been associated with dike injection in the Arco rift nary faulting: Migratory pattern and association zone. (1) Shear failure on nearly vertical faults pro- with the Yellowstone hotspot," Journal of duced the grabens above the dike tops (Figure 8). (2) Geophysical Research, vol. 94, no. B2, pp. Conjugate fault planes connecting en echelon dikes 1589-1621, February 1989 and fissures [16] may have allowed for shear failure in short sections between dikes. (3) Tensile failure of [3] D.W. Rodgers, W.R. Hackett, and H.T. Ore, "Ex- rocks during fissure and crack propagation [17] ahead tension of the Yellowstone Plateau, Snake River of and above advancing dike tips preceded the dila- Plain, and Owyhee Plateau," Geology, in press. tional displacement necessary for dike emplacement. All of these mechanisms are associated with relatively [4] R.P. Smith, W.R. Hackett, and D.W. Rodgers, small failure surfaces. For instance, the faults respon- "Surface deformation along the Arco Rift Zone, sible for mechanism (1) are limited in vertical extent to Eastern Snake River Plain, Idaho," Geological about 1 km or less (Figure 8) and their length does not Society of America Abstracts with Programs, vol. exceed about 5 km (Figure 3). In comparison, the 21, no. 5, p. 146,1989. failure surfaces associated with magnitude 7 or greater earthquakes in the B&R to the north of the ESRP are [5] M. A. Kuntz, "Geology of the Arco-Big Southern typically in the range of 20 km in vertical extent by >20 Butte area, eastern Snake River Plain, and poten- km in length [18]. Seismicity associated with Arco Rift tial volcanic hazards to the radioactive waste Zone dike injection would, therefore, be expected to be management complex, and other waste storage of much smaller magnitude than that associated with B&R tectonism. These observations are consistent

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 288 and reactor facilities at the Idaho National Engi- [15] W.E. Scott, K,L Pierce, and M.H. Hart, 'Quater- neering Laboratory, Idaho," U.S. Geological Sur- nary tectonic setting of the 1983 Borah Peak vey. Open-File Report 78-691.1978. earthquake, central Idaho," U.S. Geological Survey. Open-File Report 85-290-A. 1985. (6] MA Kuntz, B, Skipp, W.E, Scott, and W.R. Page, 'Preliminary geologic map of the Idaho National [16] D.P. Mill, 'A model for earthquake swarms,' Jour- Engineering Laboratory and adjoining areas, nal of Geophysical Research, vol. 82, no. 8, pp. Idaho,* U.S. Geological Survey. Open-File Report 1347-1352, 1977. 84-281.1984. [17] G.R. Foulger, "Hengill triple junction, SW Iceland: [7] MA Lanphere and D.E. Champion, unpublished 2. Anomalous earthquake focal mechanisms and letter to Department of Energy, Idaho Operations implications for processes within the geothermal Office, January 1985 reservoir and at accretionary plate boundaries,' Journal of Geophysical Research, vol. 93, no. [8] D.E. Champion and MA. Lanphere, "Evidence for B11, pp. 13,507-13523, 1988. a new geomagnetic reversal from lava flows in Idaho: Discussion of short polarity reversals in the [18] D.I. Doser, "The 1983 Borah Peak, Idaho and Brunhes and late Matuyama polarity chrons," 1959 Hebgen Lake, Montana Earthquakes: Mod- Journal of Geophysical Research, vol. 93, no. els for normal fault earthquakes in the Intermoun- B10, pp. 11,667-11,680, October 1988. tain Seismic Belt," U.S. Geological Survey. Open-File Report 85-290-A. 1985. [9] A. Bjornsson, G.Johnsen, S. Sigurdsson, G, Thor- bergsson, and E. Tryggvason, 'Rifting of the plate [19] F.W. Klein, R.Y. Koyanagi, J.S. Nakata, and W.R. boundary in north Iceland 1975-1978," Journal of Tanigawa, "The seismicity of Kilauea's Magma Geophysical Research, vol.84, no. B6, pp. System", U.S. Geological Survey Professional 3029-3038, June 1979. Paper 1350. pp. 1019-1186,1987.

~~[10] A.M. Rubin and D.D. Pollard, 'Dike induced faulting in rift zones in Iceland and Afar," Geology, vol. 16, pp. 413-417, May 1988.~~

[11] LG. Mastin and D.D. Pollard, "Surface deformation and shallow dike intrusion processes at Inyo Craters, Long Valley, California," Journal of Geophysical Research, vol. 93, no. B11, pp. 13221-13235,1988.

~~[12] I. Zietz, F.P. Gilbert, and J.R. Kirby, Jr., "Aeroma- gnetic Map of Idaho," U.S. Geological Survey- Map GP-92Q. 1978.~~

[13] G.F. Lindholm, S.P. Garabedian, G.D. Newton, and R.L. Whitehead, "Configuration of the water table, March 1980, in the Snake River Plain regional aquifer system, Idaho and eastern Ore- oon.' U.S. Geological Survey. Open-File Report 82-1022. 1983.

~~[14] J.T. Barraclough, pers. comm., 1989.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~289 VARIATIONS OF EARTHQUAKE GROUND MOTIONS WITH DEPTH AND ITS EFFECT ON SOIL-STRUCTURE INTERACTION~~

~~C.-Y. Chang1, Wen S. Tseng11, Y.K. Tang111, and Maurice S. Power1~~

I-Geomatrix Consultants II-Bechtel Group, Inc. 111 Electric Power Research One Market Plaza 50 Beale Street Institute Spear Street Tower, San Francisco, CA 94119 3412 Hillview Avenue Suite 717 Palo Alto, CA 94303 San Francisco, CA 9410S

~~ABSTRACT~~

Data from a free-field downhole ground notion array at Lotung, Taiwan indicated that both peak acceleration and response spectra of ground notions varied significantly with depth below the ground surface. Data trends were found to be reasonably consistent with predictions fron deconvolution analysis assuming vertically propagating body waves. Soil- structure interaction analyses of a reactor containment aodel indicated that analyses excluding ground motion variations with depth led to significant overestimation of structural responses. It is concluded that appropriate variations of ground motion with depth should be included in carrying out soil-structure interaction analyses and characterizing foundation input motions for embedded structures.

INTRODUCTION included and excluded, and structural Spatial variations of earthquake responses from these analyses were ground motion can have important compared with recorded responses. effects on the seismic soil-structure interaction (SSI) of embedded VARIATIONS OF EARTHQUAKE GROUND MOTION structures. In this study, variations WITH DEPTH of earthquake ground motion with depth Two downhole arrays DHA and DHB were examined using data from a free- were installed at the Lotung LSST site field downhole array installed by the to record earthquake ground motions at Electric Power Research Institute depth (depths of 6 m, 11 m, 17 m, and (EPRI), in cooperation with the Taiwan 47 m) as well as at the ground surface. Power Company (TPC) at a site near The locations of these two arrays are Lotung, Taiwan as part of the Lotung shown in Figure 1. Array DHA is Large-Scale Seismic Test Program (LSST) located approximately 3 m from the edge HI- of the containment model and may be affected by soil-structure interaction The significance of ground motion of t..e model, whereas array DHB is variations with depth on structural located approximately 48 m from the response was examined by soil-structure model and may be considered interaction (SSI) analyses of an representative of free-field instrumented 1/4-scale reactor conditions. Thus, the ground motion containment model at the LSST site. data front the free-field array DHB are Analyses were made in which ground examined in this study. motion variations with depth were

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~290 FA1-5~~

~~30.48m~~

~~DHB6- DHB11 — 1.5m DHB17-~~

~~Arm 3 30m~~

~~Triaxial accelerometers~~

~~(A) Surface Instrument Arrays (B) Downhole Instrument Arrays (After Tang. 1987)~~

~~Fig. 1 Location of Surface and Downhole Instrumentation, Lotung Experiment Site~~

Geological materials at the Lotung Lotung site since the completion of the downhole array site generally consist installation of the downhole instru- of Holocene alluvium and Pleistocene ments in October 1985. Ground motion materials to a depth of approximately data from a selected number of earth- 400 m over a Miocene rock. The upper quakes (ten events) analyzed in the layer of about 30 to 35 m thickness study are summarized in Table 1. These consists predominantly of silty sand data include those from earthquakes and sandy silt with some gravel. The having magnitudes ranging from ML 4.5 soils beneath this layer consist to ML 7.0 and spicentral distances predominantly of clayey silt and silty ranging from 4.7 km to 77.9 km. Peak clay. The measured shear-wave velocity ground surface accelerations range from profile increases gradually from 0.03 g to 0.21 g for horizontal motions approximately 110 m/s at the ground and from 0.01 g to 0.20 g for vertical surface to approximately 200 m/s to motions. 220 m/s at a depth of approximately 18 m. Below this depth, the shear-wave In this study, ground motion and velocity increases gradually to structural response data recorded approximately 250 m/s to 280 m/s at a during the May 20, 1986 earthquake depth of 60 m. The shear-wave (event LSST07, magnitude ML 6.5) were velocities are approximately 320 m/s at used in the ground response analyses depths of 60 m to 80 m and 480 m/s at and the soil-structure interaction depths of 80 m to 150 m. analyses. Acceleration time histories of horizontal motions recorded in the A number of moderate to strong downhole array and at the ground earthquakes have been recorded at the surface (station FAl-5) during

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

291 event LSST07 are shown in Figure 2. depth, and acceleration amplitudes as Response spectra of the horizontal well as their response spectral values motions in the downhole array and at at depths of importance to soil- the ground surface are shown in Figure structure interaction analysis of 3, Data in Figures 2 and 3 show that embedded structures (i.e., generally ground motions vary significantly with shallower than 17 m depth) are

~~1 ' \ 1 1 1 1 ' 1 ' 1 1 1 ' - - 6m •*»>t+to**fl 1 n r\/\ l\ * r\ o O^v^- \j \J \^ — V V o<~~

~~..I... i I . 1 I 1 L , 1 1 1 I 1 1 1 1~~

~~.1 "en 0 AC C (~~

~~.2 1 1 ' 1 1 1 i 1 1 ' -—v .1 — -1 7m "en s lift /I |"»L t A A /> \ A f\ V J~\ 0 V\JV \J 1 \J AC C (~~

~~1 , 1 i I i~~

~~i 1 ' 1 ' 1 ' - -47m s 1 u y|M < -.1 -~~

~~1 1 ! 1 , | 1 0 5 10 15 20 25 30 0 5 10 15 20 Time (sec) Time (sec) 2 Recorded Accelerograms in Lotung Downhoie Array DHB, Event LSST07~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

292 generally lower than those of the surface accelerations from this event motions at the ground surface. are small, i.e., equal to 0.03 g.) The spectral ratios in Figure 4 clearly To examine magnitudes of reduction show that for the small-magnitude in response spectral values from the events there are substantial reductions ground surface motion, ratios of the of the spectral values at depth from response spectra of the motions at the spectral values at the ground depth to the response spectra at the surface. The largest reduction, which ground surface were computed at the occurs at approximately the fundamental corresponding frequencies for each frequency of the soil column between earthquake summarized in Table 1. Sta- the ground surface and the depth tistical results of the spectral ratios considered, amounts to about 70 per- are shown in Figure 4. Figure 4 shows cent. It is noted that the data the median, 16th and 84th percentile scatter is relatively small at frequen- ratios of 6m/surface, llm/surface, cies between 2 Hz to 10 Hz as shown by 17m/surface and 47m/surface for the the narrow range of values between 16th three larger magnitude events (LSSTO7, and 84th percentile values. For the LSST12 and LSST16) generating peak large-magnitude events generating ground surface accelerations ranging strong shaking, the amount of reduction from 0.13 g to 0.21 g and for the six is smaller than lor the small-magnitude smaller magnitude events (LSST06, events. At a depth of 17 m, the lar- LSST08, LSST09, LSST10, and LSST14 and gest reduction is about 60 percent. LSST17) generating peak ground surface Smaller reductions in spectral values accelerations smaller than 0.07 g. at depth for the large-magnitude events (Data from event LSST08 were included (i.e., smaller amplifications of the in this group because peak ground motion at depth to the ground surface)

~~Table 1 Summary of Earthquake Ground Motion Data Analyzed~~

~~Epicentral Focal Peak Distance Depth AZIM Ground Surface Ace*. g Earthauake Date Magnitudes fkmi (km) ide&l EW NS Vert.~~

LSST06 4/8/86 5.4 31.4 10.9 174 0.04 0.03 0.01 LSST07 5/20/86 6.5 66.2 15.8 195 0.16 0.21 0.04 LSST08 5/20/86 6.2 69.2 21.8 192 0.03 0.03 0.01 LSST09 7/11/86 4.5 5.0 1.1 146 0.07 0.05 0.01 LSST10 7/16/86 4.5 6.1 0.9 162 0.03 0.04 0.02 LSST11 7/17/86 5.0 6.0 2.0 90 0.07 0.10 0.04 LSST12 7/30/86 6.2 5.2 1.6 131 0.16 0.19 0.20 LSST14 7/30/86 4.9 4.7 2.3 119 0.04 0.05 0.02 LSST16 11/14/86 7.0 77.9 6.9 174 0.13 0.17 0.10 LSST17 11/14/86 „ _ 0.04 0.04 0.02

* Peak ground surface acceleration at Station FAl-5

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

293 .: i 1 2 I l)!0 iO«l ! i I ! 5 10 50 50 100 frequency (H;) frequency (Hr) .1 .2 .5 12 5 10 20 S0100.1 ? .5 12 5 10 20 Ml00 Response Soectra{5% Damping) of Recorded Honz. Motions frequency (Hj) frequency (M:) m Lotung Downhole Array DHB. Event L5ST07 Fig. 4 Ratio of Spectral Acceleration for Horizontal Components Lotung Downhole Array Ground Motion Dato are probably due to an increase in soil of soils due to strong shaking during damping during strong shaking. The the large-magnitude events. data scatter for the larger-magnitude events is larger at frequencies higher Deconvolution analyses assuming than about 6 Hz due primarily to the vertically propagating shear waves were high frequency motions at depths for conducted for the May 20, 1986 event LSST12. Because of limited earthquake (event LSST07), In these number of data sets available for the analyses, the ground motions recorded large-magnitude events, the 84th and at the ground surface were used as 16th percentile values correspond input motions. Motions at other depths closely to the upper-bound and lower- were computed and compared with the bound values. recorded motions. The analyses were conducted using the computer code SHAKE Comparisons of frequencies at which [2], Non-linear soil response is the largest reduction in spectral approximated by the equivalent linear values occurs at corresponding depths method implemented in SHAKE. Compari- for the small- and large-magnitude sons of the response spectra of the events indicate that the frequencies computed and recorded motions at depths are lower for the large-magnitude are shown in Figure 5. Excellent events than for the small-magnitude agreement was obtained between the events, indicating a reduction in response spectra of the computed and fundamental frequencies of the soil recorded motions, even to depths as column or a reduction in shear modulus great as 47 meters. Comparisons of

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

294 acceleration time histories of the computed and recorded motions at depths 3' 5 f}f'"""T''fl\ T MHT r "' T "id I \ ITTT SurfocefNSj of 6 m and 17 m are shown in Figure 6 I " Input and again indicate excellent agreement.

EFFECTS ON STRUCTURAL RESPONSE OF $0 NEGLECTING GROUND MOTION VARIATIONS S 5 WITH DEPTH 4 In practice, soil-structure i -6m(NS) interaction analyses of embedded I: structures have sometimes been conducted using an approach that 2 .5 neglects embedment effects on the 1 4 foundation input motion (i.e., that -nm(EW) " -llm(NS) " neglects the variations of ground motion with depth and excludes

kinematic interaction effects). In 3 .5 -T rrmui 1.4 such analyses, the translational a 3 components of the foundation input I -17m(EW) " 17m(NS) " motion have been taken directly as 3 2 those of the control motion (as fta prescribed at the ground surface), and 3 5 the rocking components of the 1.4 O foundation input motion have been -47rn(NS) " neglected. To assess the effects of this practice of excluding variations of ground motion with depth on .1 .2 .5 1? S 10 20 50100.1 .2 .5 12 5 10 10 50 100 structural response, soil-structure Frequency (He) Frequency tHl) interaction analyses of Fig. 5 Comparison of Response Spectra (5% Damping) of Computed and Recorded Motions. Deconv.. Event LSST07, Lotung Site

~~A-Z^-AAA ^r^ Mwov^M^VVl~~

~~I T I ' I ' I -17 m(N-S)~~

~~i[Ay» At. A -- /\.~~

~~_L 10 15 20 25 30 0 10 15 20 25 30 Time (sec) Time (sec) g. 6 Comparison of Computed and Recorded Acceleration Time Histories, Event LSST07, Lotung Site~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

295 a 1/4 scale reactor containment model cated that earthquake ground motions at the Lotung site were conducted in can vary significantly with depth below this study. The analyses were con- the ground surface in the depth range ducted for two cases -- one incorpo- of typical embedment depths of nuclear rating variations of earthquake ground power plant structures and other motion with depth and kinematic inter- buildings. The results of the action and the other excluding these deconvolution analyses conducted in effects. Computed structural responses this study indicated that deconvolution are compared with recorded responses. procedures assuming vertically propagating body waves provided Figure 1 shows that the containment reasonable estimates of the variations model has a diameter of 10.52 m and is of ground motion with depth. embedded at a depth of 4.57 m [1]. The model also includes a steam generator The results of the soil-structure model. In this study, the SSI analyses interaction analyses of the reactor were performed for th? May 20, 1986 containment model installed at the earthquake (event LSST07). The free- Lotung, Taiwan site indicated that the field ground surface motions recorded practice, which is sometimes followed, at station FAl-5 were used as the con- of excluding ground motion variations trol motion prescribed at the ground with depth and kinematic interaction surface. Computed structural responses led to overestimation of structural are compared with recorded responses at response. the top of the basemat of the containment model and at the top of the steam On the basis of the results of this generator model. study, it is concluded that appropriate variations of ground motion with depth Structural responses computed from should be incorporated in carrying out the analysis that excluded effects of soil-structure interaction analyses and ground motion variations with depth and characterizing foundation input motions kinematic interaction are compared with for embedded structures. those from the analysis in which these effects were included in Figures 7 and ACKNOWLEDGMENTS 8. Also shown in Figures 7 and 8 are This paper is based on research the recorded structural responses. The studies sponsored by the Electric Power comparisons in Figures 7 and 8 t.how Research Institute, Inc. that the analysis incorporating variations of earthquake ground motion with REFERENCES depth and kinematic interaction results [1] H.T. Tang, "Large-Scale Soil- in computed structural responses that Structure Interaction," Report are in reasonably good agreement with EPRI-NP-5513-SR, Electric Power the recorded responses. However, the Research Institute, Palo Alto, analysis excluding kinematic inter- California, November 1987. action results in substantially higher structural responses than the recorded [2] P.B. Schnabel, J. Lysmer, and responses. The results indicate that H.B. Seed, "SHAKE - A Computer the practice of excluding kinematic Program for Earhquake Response interaction can lead to significant Analysis of Horizontally Layered overestimation of structural responses. Sires," Report No. UCB/EERC 72- 12, Earthquake Engineering SUMMARY AND CONCLUSIONS Research Center, University of Analysis of the ground motion data California, Berkeley, 1972. recorded in the free-field downhole array at the Lotung, Taiwan site indi-

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~296 1.0 — Recorded response 5% Damping — Without kinematic interaction at With kinematic interaction NS CO c g I 2 o~~

~~to I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I 1 5% Damping EW .8 CO ion i « .6 ;ele i o .4~~

~~VV' ( tral , .2 / Spe c~~

~~0 0.1 1 10 100 Frequency, Hz~~

~~Fig. 7. Comparisons of Response Spectra of Computed and Recorded Motions at top of Basemat of the Containment Model, Event LSST07~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~297 2.0 I 1 1 1 1 1 1 1 — Recorded response 5% Damping -- Without kinematic interaction O) With kinematic interaction NS~~

~~o i * .8 2 ' i ' A i ,'•< ,''M' co .4~~

~~2.0 i i iiitii 5% Damping EW p1.6 1 ll CO 11 I 1.2 1 1 I 1 1 1 1 1 1 1 1~~

~~.8 1 \\ 1~~

~~I -*• 1~~

1 / ': *

~~Q. CO /~~

~~»——"T 1 t t 1 1 1 i i (tit*, J i j i I ! i : 0.1 1 10 100 Frequency, Hz~~

~~Fig. 8. Comparisons of Response Spectra of Computed and Recorded Motions at top of the steam Generator Model, Event LSST07~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~298 NATURAL PHENOMENA ANALYSES, HANFORD SITE, WASHINGTON A. M. Tallman Westinghouse Hanford Company Richland, Washington ABSTRACT~~

The Hanford Site has been the subject of numerous natural hazard studies. These studies have supported the siting and construction of nuclear reactors, the siting and characterization of a potential high-level nuclear waste repository, the routine operation of process facilities, and waste management. The natural hazard models resulting from these studies are compared to the models promulgated by the Lawrence Liveraore National Laboratory for use in the design and evaluation guidelines for U.S. Department of Energy facilities. Probabalistic seismic hazard studies completed for the Washington Public Power Supply System's Nuclear Plant 2 and for the U.S. Department of Energy's N Reactor sites, both on the Hanford Site, suggested that the Lawrence Livermore National Laboratory seismic exposure estimates were lower than appropriate, especially for sites near potential seismic sources. A probabilistic seismic hazard assessment was completed for those areas that contain process and/or waste management facilities. The lower bound magnitude of 5.0 is used in the hazard analysis and the characteristics of small- magnitude earthquakes relatively common to the Hanford Site are addressed. The recommended ground motion for high-hazard facilities is somewhat higher than the Lawrence Livermore National Laboratory model and the ground motion from small-magnitude earthquakes is addressed separately from the moderate- to large-magnitude earthquake ground motion. The severe wind and tornado hazards determined for the Hanford Site are in agreement with work completed independently using 43 years of site data. The low- probability, high-hazard, design-basis flood at Che Hanford Site is dominated by dam failure on the Columbia River. Further evaluation of the mechanisms and probabilities of such flooding is in progress. The Hanford Site is downwind from several active Cascade volcanoes. Geologic and historical data are used to estimate the ashfall hazard.

1.0 HAZARD/SAFETY CLASSIFICATION AND are intended (1) to control the level NATURAL PHENOMENA of conservatism introduced in the design/evaluation process so that Design and evaluation guidelines for natural phenomena hazards are treated U.S. Department of Energy (DOE) facili- in a consistent and uniform fashion at ties subjected to natural phenomena haz- the various DOE sites, and (2) to ensure ards have been promulgated by the that the level of conservatism is Lawrence Livermore National Laboratory appropriate for the use/mission of the (LLNL) for the DOE. These guidelines facility being assessed. Four facility

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

299 uae/hasard classification levels are data, multiple tectonic models to aid presented and an annual probability of in the interpretation of faulting in exceedance (hazard level) or annual the area, deterministic assessments of number of events are given for each the seismic hazard at the Supply System phenomena at the four classifications. site, and a detailed probabilistic seismic hazard evaluation for the Probabilistic assessments of nat- Supply System site. ural phenomena hazards occurrence at the DOE sites were completed by LLNL The primary earthquake-generating and their contractors. A probabilistic structures considered to be capable by seismic haxard study of the Hanford Site the U.S. Nuclear Regulatory Commission was completed by TERA Corp., Inc. in (NRC) [9,10] are the Rattlesnake-Wallula 1978 with the results reported by LLNL Alignment (RAW) and the Gable Mountain/ In UCRL-53582 [1], The hazard models Gable Butte structure (Figure 1). The for tornadoes and extreme winds for facilities on the Hanford Site are at eight DOE sites including the Hanford various distances from these and other Site are reported in UCRL-53526 [2]. potential earthquake sources, there- A probabilistic flood hazard study was fore, the total seismic exposure is completed for the N Reactor [3] and is not the same for all locations. It reported, along with the other DOE was decided that a probabilistic seismic sites, in UCRL-53851 [4], hazard analysis should be done to address the various facility locations These natural hazard models of the and ensure incorporation of geologic Hanford Site were reviewed and compared and seismic data not available in 1978 to Site-specific studies completed on when the TERA study was completed. the Ranford Site. This paper presents the data and interpretations that support A probabilistic seismic hazard as- Vestinghouse Hanfords assessment of the sessment was completed by Woodward- natural phenomena design criteria and Clyde Consultants (WCC) for the 100-N, provides technical support for any 200 East, 200 West, 300, and 400 Areas revisions to the design criteria. on the Hanford Site [11j. The methodology and data base used are generally 2.0 EARTHQUAKE STRONG GROUND MOTION the same as those used by WCC for the Supply System WNP-2 FSAR (see Appendix 2.1 BACKGROUND 2.5K [5]), in the seismic hazard assessment of the Hanford Site [12], and for The Hanford Site has been the subject the seismic exposure study of the N of many seismotectonic studies, espe- Reactor [13]. The earthquake recur- cially during the last 10 yr. These rence model used In the original study studies are summarized in the final [5] does not represent the seismiclty safety analysis reports (FSAR) for of the Site as it is understood today. Washington Public Power Supply System A recent detailed study of the histori- (Supply System) nuclear power plants cal and instrumentally recorded seismic IS,6] the preliminary safety analysis record was completed by Pacific North- report (PSAR) for the Skagit/Hanford west Laboratory (PNL) [14], The results Site nuclear power plants [7], and the of this seismlcity study show that the Hanford Site Characterization Plan- earthquakes of magnitudes 4.0 to 5.7 Consultant Draft [8] that summarizes were underestimated in the Supply seismic data and tectonic interpretations System study and that earthquakes of completed for the Basalt Waste Isolation magnitudes greater than 5.7 were over- Project. These data include, but are not estimated (Figure 2). The revised limited to, the identification of recurrence model [14] was used in the specific potential earthquake-generating probabilistic analyses supporting this faults, detailed earthquake monitoring study [11]. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

~~300 119° m~~

~~.100N~~

G«bk Mountain- Gabl* Butt*

~~Potential source considered in seismic analysis; 10 tics indicate segment boundaries described in text 76905163.11 Anticlinal ridge beyond 50-km radius Figure 1. Location Hap.~~

1,1 LOWBR BOUND MAGNITUDE ground-motion characteristics were recorded [17]. Two conclusions can be The range of earthquake magnitudes drawn from these and other related that should be included in a seismic studies: hazard analysis must be defined. The upper bound magnitude is defined by 1. Earthquakes with ML <5.0 determining the maximum size of the appear to be generally in- earthquakes that can be generated by the capable of damaging even sources (faults) in the area. The lower stiff (high-frequency), well- bound or smallest magnitude earthquake engineered structures and to be considered is determined by the equipment [18]. engineering characteristics of the 2. Small-magnitude earthquakes earthquake ground motions and their (ML <5.0) generally have a potential for causing damage to facili- narrow frequency band, short ties. Recent studies have been made of duration and often lack low- the engineering characteristics of the frequency ground motion [16], earthquake ground motions 115,16] and As such they are not repre- eapirical data have been gathered from sented by a broad-frequency- engineered facilities that have been band response spectra like the subjected to ground motion where the Newaark-Hall spectrum. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

301 lower probability of moderate (5 (H( 7.0) earthquakes. Because the sels- nicity la dominated by micro- and •mall earthquakes, their Inclusion in the seismic hazard model would Increase the seismic exposure of the Hanford Site compared to only considering magnitudes greater than 5.0. This is especially true in locations proximal to the Gable Mountain/Gable Butte structure where the majority of the expected earthquakes are less than magnitude 5.0. The faults on this structure are considered to be capable generating magnitude 5 ^ 0.5 earthquakes [9,10]. ~~

As previously discussed, current studies Indicate that the ground motion from small earthquakes (H < 5.0) is not a significant factor in damage to laterally braced structures and that the response spectra for these earthquakes Figure 2. Earthquake Recurrence are not represented by a broad-band Comparison. response spectra such as Newmark-Hall. Therefore, the ground motions considered Probabilistic seismic hazard studies in the WCC study [11] include contribu- have often Included small-magnitude tions from magnitude 5.0 to 7.5 earth- earthquakes,. A comparison of probabi- quakes. The upper-magnitude earthquake listic seismic hazard studies of the is based on the fault characteristics eastern United States (EUS) nuclear as discussed in the VCC study. However, power plants (NPP) [19] demonstrated because the Hanford Site and surrounding that Including small-magnitude earth- area has a relatively high rate of quakes considerably Increased the low-magnitude earthquakes, a response seismic exposure at those sites with spectra characteristic of small- relatively high occurrence of small magnitude earthquakes in the area earthquakes. However, there Is a should be considered in the design and general lack of data that Indicate qualification of components and equip- damage to engineered structures, espe- ment that must remain operable during cially NPPs, from small-magnitude earth- and/or immediately following a seismic quakes [20]. The most recent study of event. EUS NPPs completed by LLNL for the NRC Is, therefore, based on a lower bound magnitude of 5.0 ML [21]. The LLNL 2.3 SEISMIC HAZARD MODEL cautioned that the seismic safety of brittle components of NPP Systems The seismic hazard curves for the (e.g., relays) may need to consider the 100-N, 200 East, 200 Vest, 300, and effect of the small-magnitude earth- 400 Areas are shown in Figure 3. These quakes that are not Included In the curves represent the randomly oriented hasard studies. peak instrumental horizontal ground motion from earthquakes of magnitude <5.0 from known potential selsmogenlc The Hanford Site Is within an area sources. The hazard curves can be of primarily micro-earthquakes divided into two families. The 300 and (1.0 < M < 3.0) and small-magnitude 400 Area hazard curves are lower (appro- earthquakes (3.0 < M <5.0) with a much ximately 0.12 to 0.14 g at 2 x 10~* Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

~~302 2.4 RESPONSE SPECTRA~~

A Ntwmark-Hall median response spectrum shape for a sediment sit* la recommended for the Hanford Site In th« LLNL guidance. This broad-frequency- Gtntraiuw 500 yi band shape most closely represents the ground motion for earthquakes of magni- Mod«attandLowHaurd 1,000 yt tudes greater than 6.0 and strong motion durations longer than 3 s. Earthquakes of magnitudes less than 6.0 generally contain a narrower frequency band of High Hazard 5,000 yr strong ground motion and are of short duration. In general, earthquakes of less than 6.0 magnitude do not have sufficient energy content to be capable of producing high acceleration of long duration and broad-frequency content, all a part of the Newmark-Hall spectrum 300 -. > 200 Eail(North Boundary) \ [16]. In the 100-N and 200 Areas less 100N ,V\ than 15% of the ground motion at 0.20 g 0.0 01 0,2 0.3 0.4 05 is from earthquakes greater than 6.0 Ptak ground accttoiatlon Ig) M. In the 300 and 400 Areas less than Figure 3. Seismic Hazard Curves. 30% of the ground motion is from greater than 6.0 M. The use of the broad- probability of exceedance or events/year) frequency-band Newmark-Hall shape pro- than the remaining areas (approximately vides considerable conservatism as it 0.16 to 0.18 g at 2 x 10~* events/year). tends to overestimate the structural The lower values at the 300 and 400 response to earthquakes less than Areas primarily result from a lower con- magnitude 6.0 [22]. tribution from the Gable Mountain/Gable Butte structure (see Figure 1). The RAW Site-specific response spectra for is the major contributor to the 300 and a magnitude 4.0 +_ 0.2 earthquake was 400 Areas seismic hazard. The exposure developed for the Supply System's WNP-2 from 4.0 £ M < 5.0 earthquakes, not in- site to address the impact of a small- cluded in the, seismic hazard curves, is magnitude earthquake very near the Site considerably less in the 300 and 400 [5] (Amendment No. 37). The amplifica- Areas because of the greater distance tion factors and spectral shape are an from the Gable Mountain/Gable Butte. At appropriate representation of small 10-3 events/year, all sites are less than (less than 5.0 magnitude) earthquakes 0.08 g and at 2 x 10"3 events/year, all on the Hanford Site. sites are less than 0.05 g(see Figure 3). 2.5 RECOMMENDATIONS The 100-N and 200 Areas have quite similar seismic hazard curves; however, The Newmark-Hall median response the individual contributors are differ- spectra anchored to a 0.15 g peak, rent. For example, the 100-N Area has horizontal free-field, ground accele- considerably more exposure from the ration are appropriately conservative Saddle Mountains and Frenchman Hills for the design and analysis of high- (see Figure 1), but less from the other hazard structures in the 300 and 400 sources. A slightly higher exposure at Areas (Figure 3). In the 100-N and 200 200 West than at 200 East results pri- Areas, the Newmark-Hall spectra anchored marily from the proximity of Yakima to 0.20 g peak, horizontal free-field, Ridge to the 200 West Area (WCC 1989). ground acceleration are recommended for Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

~~303 1.0~~

~~0.5 IS 0.3 / / >. N 0.20 g \ 0.15 g .2 a / £> / o u / j 7/ 0.05 A JJr 0.03 // 0.01 0.1 0.3 0.5 1.0 3.0 5.0 10 30 50 100~~

~~Frequency (Hz) 78905163.1~~

~~Figure 4. Newmark-Hall Median Response Spectrum, Solid Line; Small Magnitude Median Response Spectra, Dashed Line.~~

~~Table 1. Seismic Hazard Comparison.~~

~~Facility use~~

~~Low hazard/ Moderate High General use important hazard hazard~~

~~Hanford Site safety class (SC) SC 4 SC 3 SC 2 SC I~~

~~LLNL Study 0.09 g 0.12 g 0.12 g 0.17 g This Study 0.05 g 0.08 g 0.08 g 0.15 ga 0.20 gb~~

~~a 300 and 400 Areas, b 100-N and 200 Areas.~~

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 304 • high-haxard structure. The sp«ctr« category. Except for high-hazard facil- anchored «t 0.20 g are recommended for ities in the 100-N and 200 Areas, the high-haiard faciliti«s throughout the acceleration valuet determined in this Hanford Sit* as it is conv«nitnt to have study are all less than the UCRL-15910 the same sp«ctra for all ar«as(?igur« 4), recommendation. This value, 0.20 g, may b* usad as the maximum horizontal instrumental peak It is recommended that the Hanford acceleration for design analyses because Sits seismic hazard value for high- it represents a 12% or greater increase hazard facilities be changed from 0.17 in the values shown in Figure 3. However, g to 0.20 g and the small magnitude if an existing structure within the 300 earthquake be addressed for Safety or 400 Area does not qualify using the Class 1 systems, components and equip- 0.20 g spectra, the structure qualifi- ment. The more conservative LLNL accele- cation should be to the 0.15 g spectra. rations for the rest of the classes provide additional margin in the lateral The seismic hazard at the various load capacities of these lower hazard areas on the Hanford Site is quite simi- facilities. lar at the 1 x i0~3 events/year or a return period of 1,000 yr, varying from 3.0 SEVERE WIND AND TORNADO 0.06 to 0.08 g(Figure 3). Based on these values, a peak horizontal acceleration Coats and Hurray [2] compiled the of 0.08 g is appropriate for moderate- extreme wind and tornado hazard models and low-hazard structures in all the for the DOE sites in UCRL-53526, Rev.l. areas. Two Hanford Site wind and tornado models were developed for the LLNL. One used Safety Class 1 systems, components, 26 yr of Hanford Site data and the other and equipment that require operability analyzed data from Valla Walla and during and/or immediately after a seis- Yakima, WA. The Hanford Site model was mic event should also consider a small- judged most appropriate and is the magnitude, narrow-frequency-band earth- basis of the values in Coats and Hurray quake with a duration less than 3 s. [2]. The straight wind speed exceeds The exposure to small earthquakes is the tornado at both the moderate- and less in the 300 and 400 Areas than in high-hazard probabilities; therefore, the 100-N and 200 Areas because of the straight winds govern the design relative distance to the Gable criteria for nonreactor facilities. Mountain/Gable Butte (see Figure 1). The results of a study using 43 yr of Hanford Site data compiled for the NRC A near-field earthquake with peak by PNL [23] verify the values recom- horizontal acceleration of 0.15 g is mended in UCRL-53526, Rev. 1. recommended for the 300 and 400 Areas and a peak horizontal acceleration of 4.0 FLOODING 0.20 g for the 100-N and 200 Areas (Figure 4). These median accelerations A probabilistic flood hazard study approximate a magnitude of 4.0 and 4.5 of N Reactor site was done for Office earthquakes respectively at a distance of Nuclear Safety as part of the N of about 3 km. The magnitudes are Reactor Probabilistic Risk Assessment representative of the most likely (PRA) [3 & 4]. However, there are some magnitude contributor at the location. aspects of the report that have not been resolved. The revision of the The results of this study are com- flood hazards is being deferred until pared to the LLNL recommended ground a complete evaluation of the motion in Table 1. The Westinghouse probabilistic assessment has been Hanford Company safety classification made. is also related to the facility use Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

305 5.0 VOLCANIC ASH for maximum ashfall at the Sita would ba conaidarably grafter than 500 to There arc several major volcanoes In 1,000 yr. Based on these long return the Cascade Rang* vast of tha Hanford periods, the design basis ashfall is Sit*. Tha only potential volcanic haz- only recommended for high-hazard or ard to tha Slta Is ashfall from a larga Safety Class 1 facilities. aruptlon. Tha closest volcano Is Mount Adams at a distance of about 160 km. Mount St. Helens, currently the most active, is approximately 200 km from the Hanford Site. 6.0 REFERENCES

A volcanic hazard {ashfall) probabi- [1] D. V. Coats and R. C. Murray, 1984, listic assessment has not been done. Natural Phenomena Hazards Modeling However, based on the geologic record of Project; Seismic Hazard Model3 for the Cascade Range, a recommendation for Department of Energy Sites, ashfall design criteria can be made. UCRL-53582, Rev. 1, Lawrence The Supply System's VNP-2 design basis Livermore National Laboratory, ashfall [5,6] (Table 2) is equivalent to Livermore, California. an eruption that is expected to occur every 2,000 to 3,000 yr [24]. The pro- [2] D. V. Coats and R. C. Murray, 1985, bability of the Hanford Site being in Natural Phenomena Hazards Modeling the maximum ashfall plume for any single Project; Extreme Wind/Tornado eruption is considerably less than 1. Hazard Models for Department of Therefore, the return period for the Energy Sites. UCRL-53526, Rev. 1, design basis ashfall would be greater Lawrence Livermore National than 2,000 to 3,000 yr. It is recom- Laboratory, Livermore, California. mended that the design basis ashfall for the Supply System's WNP-2 is appro- [3] M. W. McCann, Jr. and A. C. priately conservative for a high-hazard Boissonnade, 1988, Probabilistic facility at the Hanford Site. Flood Hazard Assessment for the N Reactor, Hanford Washington. UCRL- A small volume eruption, similar to 21069, prepared for the U.S. the Mount St. Helens May 1980 eruption, Department of Energy under contract is likely to occur every 500 to 1,000 yr to Lawrence Livermore National [24], Again, because the axis of the Laboratory, Livermore, California. maximum ashfall will not always be on the Hanford Site, the return period for

~~Table 2. Characteristics of Design Basis Ashfall.~~

Duration of ashfall 20 h Potential compacted thickness 3 in. Estimated X compaction 20 to 40X Average ashfall rate 0.15 in./h Average density Dry, loose 72 Ib/ft3 Dry, compacted 96 Ib/ft3 Wet, compacted 101 Ib/ft3

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

306 141 J. b, Savy and R. C. Murray, 1988, [12] R. R. Youngs, K. J. Coppersmith, Natural Phenomena Hazarda Modeling M. S. Power, and F. H. Swan III, Project: Flood Hazard* Models for 1985, "Seismic Hazard Assessment Department of Energy Sites, UCRL- of the Hanford Region, Eastern 53851," prepared for "the U.S. Washington State," in Proceedings Department of Energy by Lawrence DOE Natural Phenomena Hazards Livermore National Laboratory, Mitigation Conference*. CONF 85 Livermore, California. lOH8,""o"ffice "of "Nuclear Safety, U.S. Department of Energy, t5] WPPSS, 1981, Final Safety Analysis Washington, D.C. Report, WPPSS Nuclear Project No. 2, Amendments 18, 21, and 36, Washing- [13] WCC, 1987, Evaluation of Seismic ton Public Power Supply System, Exposure for the N Reactor Plant, Richland, Washington. Hanford, Washington, UNI-4426, prepared by Woodward-Clyde Con- [6] WPPSS, 1986, Nuclear Project Nos. sultants for UNC Nuclear 1 and 4, Final Safety Analysis Industries, Richland, Washington. Report, Washington Public Power Supply System, Richiand, Washington. [14] A. C. Rohay, 1989, Earthquake Recurrence Rate Estimates For [7] PSPL, 1981, Skagit/Hanford Nuclear Eastern Washington and the Hanford Project, Preliminary Safety Analysis Site, TNL-6956, prepared for Report, Amendments 23, 724, 25, and Westinghouse Hanford Company by 26, Puget Sound Power and Light Co., Pacific Northwest Laboratory, Bellevue, Washington. Richland, Washington.

[8] DOE, 1988, Consultation Draft Site [15] R. P. Kennedy, S. A. Short, Charac teriza tIon Plan Reference K. L. Merz, F. J. Tokarz, I. M. Repository Location, Hanford Site, Idrlss, M. S. Power, and K. Washington, DOE/RW-0164, U.'s. Sadigh, 1984, Engineering Department of Energy, Washington, Characterization of Ground Motion D.C. Task I: Effects of Characteris- tics of Free-Field Motion on [9] NRC, 1982a, Safety Evaluation Report Structural Response, NUREG/CR- Related to the Construction of 3805, Vol. 2, prepared for the Skagit/Hanford Nuclear Project, U.S. Nuclear Regulatory Com- Units 1 and T, NUREG-0309, Supple- mission by Structural Mechanics ment No. 3, U.S. Nuclear Regula- Associates, Inc., Newport Beach, tory Commission, Washington, D.C. California and Woodward-Clyde Consultants, Walnut Creek, [10] NRC, 1982b, Safety Evaluation California. Report Related to the Operation of WPPSS Nuclear Project No. 2, [16] R. P. Kennedy, R. H. Kincaid, and NUREG-0892, Supplement No. 1, U.S. S. A. Short, 1985, Engineering Nuclear Regulatory Commission, Characterization of Ground Motion Washington, D.C. Task II: Effects of Ground Motion Characteristics on Structural [11] WCC, 1989, Evaluation of Seismic Response Considering Localized Hazard for Non-Reactor FacilitresL Structural Nonllnearitles and Hanford Reservation,_ Hanford, Soil-Structure Interaction Washington, WHC-^MR-dc-23, prepared Effects.,~~NUREG/CR-3805, Vol. I, for Westlnghouse Hanford Company prepared for U.S. Nuclear Regula- by Woodward-Clyde Consultants, tory Commission by Structural Consultants, Oakland, California. Mechanics Associates, Inc., Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

307 Newport Beach, California and [22] G. E. Cummings, D. L. Bernreuter, Woodward-Clyde Consultants, Walnut R, C. Murray, and J. B. Savy, Creek, California. 1987, Sujutarjr JUggoc;t jp£ the §MR»L(BAuJli ~2IL-jlskVLK* HLd- JSS?A [17] S. W. Swan and N. G. Horstman, Siting^Criteria for Nuclear 1989, "Seismic Experience In Power Power .Plants'," "NURBG/CP-0087, and Industrial Facilities as It Lawrence Livermore National Relates to Small Magnitude Laboratory, Liveraore, Earthquakes," In Proceedings: California. Engineering Characterization of Small-Magnitude 'Earthquakes*,* [23] J. V. Ramsdell, D. L. Elliot, Electric Powef" Research Trfst 11ute, C. G. Holladay, and J. M. Hubbe, Palo Alto, California. 1986, Methodology for Estimating Extreme Winds for Probabilities. [18] R. P. Kennedy, 1989, "Engineering Rls'k "Assessments, NUREG/CR-449*2, Characteristics of Small Magnitude prepared by Pacific Northwest Earthquakes," in Proceedings: Laboratory for the U.S. Nuclear Engineering Characterization of Regulatory Commission, Small-Magnitude Earthquakes, Washington, D.C. Electric Power Research Institute, Palo Alto, California. [24] D. R. Crandell and D. R. Mullineaux, 1978, Potential [19] D. L. Bernreuter, J. B. Savy, and Hazards from Future Eruptions of R. W. Menslng, 1987, Seismic Hazard Mount St. Helens VolcanoK Characterization of the Eastern WashTnJttjon, Bulletin 1383-C, United States: Comparative U. S. Geological Survey. Evaluation of the LLNL and EPRI Studies, NUREG/CR-4885, prepared by Lawrence Livermore National Laboratory for U.S. Nuclear Regulatory Commission, Washington, D.C.

[20] M. W. McCann, Jr., and J. W. Reed, 1989, "Engineering Characterization of Small-Magnitude Earthquakes: An Overview," in Proceedings: Engineering Characterization of Small-Magnitude Earthquakes, Electric Power Research Institute, Palo Alto, California. [21] D. L. Bernreuter, J. B. Savy, R. W. Menslng, J. C. Chen, and B. C. Davis, 1985, Seismic Hazard Characterization of the United States: Vol. 1 and Vol. 2. UCID- 20421, Vol. 1 and Vol. 2, Lawrence Livermore National Laboratory, Livermore, California.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~308 SessidrilO Poster Session~~

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

~~309 A PORTABLE BACKUP POWER SUPPLY TO ASSURE EXTENDED DECAY HEAT REMOVAL DURING NATURAL PHENOMENA-INDUCED STATION BLACKOUT~~

~~L. D. Proctor, L. D. Merryman, W. E. Sallcc Oak Ridge National Laboratory P. O. Box 2008 Oak Ridge, Tennessee 37831~~

~~ABSTRACT~~

The High Flux Isotope Reactor (HFIR) is a light water cooled and moderated flux-trap type research reactor located at Oak Ridge National Laboratory (ORNL). Coolant circulation following reactor shutdown is provided by the primary coolant pumps. DC-powered pony motors drive these pumps at a reduced flow rate following shutdown of the normal ac-powered motors. Forced circulation decay heat removal is required for several hours to preclude core damage following shutdown. Recent analyses identified a potential vulnerability due to a natural phenomena-induced station blackout. Neither the offsitc power supply nor the onsitc emergency diescl generators arc designed to withstand the effects of seismic events or tornadoes. It could not be assured that the capacity of the dedicated batteries provided as a backup power supply for the primary coolant pump pony motors is adequate to provide forced circulation cooling for the required time following such events. A portable backup power supply added to the plant to address this potential vulnerability is described.

PLANT DESCRIPTION the atmosphere by way of the induced draft The High Flux Isotope Reactor is a cooling tower located near the reactor light water cooled and moderated flux-trap confinement. type research reactor located at Oak Ridge Reactor coolant enters the reactor National Laboratory on the DOE Oak vessel through two pipes near the top of the Ridge Reservation. The reactor, rated at 85 vessel and after passing through the core megawatts (MW), was originally designed as exits via a pipe near the bottom of the part of the overall program to produce vessel. Nominal primary coolant flow rate is trans-uranic isotopes for use in heavy- 16,000 gallons per minute (gpm) with a core element research. inlet temperature of 120 degrees Fahrenheit The reactor is housed in a (F) and an outlet temperature of 155 F. reinforced concrete confinement building. The water-solid system is maintained at an Within the building, the reactor vessel is operating pressure of 468 pounds per square located in a pool containing approximately inch (psig) by one of two centrifugal 85,000 gallons of water. The reactor pressurizer pumps and a system of letdown cooling system, shown schematically in valves. Figure 1, includes four pump-heat exchanger Decay heat removal following loops (three in operation, one in standby) reactor shutdown is normally accomplished which circulate coolant through the core by driving the primary cooling pumps with and reject heat to the secondary cooling large AC motors. If AC power is not system during reactor operation. The available, the primary pumps can be secondary cooling system transfers heat to operated at a reduced flow until the level of

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~310 ORNL DWG 89-14804~~

~~FE,00 TE100-lA 2A, 3A~~

~~, PUMP SEALS MAGNETIC 468 PSlG COUPLER 15 KGPM~~

~~PRESSURE CONTROL LETDOWN VALVES PCV-127-7~~

~~PRIMARY COOLANT TE100-1B, 2B. 3B SYSTEV PUMP SEAL H2O 5 KGPM/PUMP 600 HP - 2 3 KV AC + 3 HP - 120V DC lEMERGENCY) CONTROLLED 3Y VESSEL INLET TEMPERATURE SIGNAL~~

~~SECONDARY 377 (MANUAL TRIM) COOLANT OTHER HEAT SYSTEM LOADS. k POOL HEAT EXCHANGER AIR CONOITlCNIN TRU. ..~~

~~2 AT 30 HP TEMPERATURE CONTROLLED~~

~~COOLING TOWERS~~

~~E'/==JENCY)~~

~~FIGURE 1. SIMPLIFIED HFIR PROCESS FLOW DIAGRAM~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

311 decay heat is such that natural circulation in supply, with backup from the onsite the reactor and convective heat transfer to emergency diesel generators. Dedicated the pool can remove the heat generated. batteries are also provided in the event both An auxiliary DC motor called a "pony offsite and onsite ac power are lost. The motor" is coupled to the main motor shaft dedicated batteries have been shown to be of each primary coolant pump to accomplish able to supply adequate power to run the this reduced flow decay heat removal. One pony motors (and thus provide forced pony motor driven pump is sufficient to circulation) for a period of 6 to 7 hours. remove the decay heat; although three are With proper operator actions to stage the normally in operation during shutdown number of operating pumps, this can be conditions. The secondary cooling system is extended to a minimum of 12 hours. not required during this time. For any accident in which offsite power is available or in which the onsite THE NEED FOR EXTENDED DECAY emergency diesel generators are available, it HEAT REMOVAL can be assured that forced primary coolant During the recent shutdown of the circulation would be provided for as long as HFIR, it was considered prudent to re- is necessary. However, for events in which examine various analyses upon which the these sources of power are not available, plant Safety Analysis is based. The early power for the pony motors must be analyses indicated that transition from provided from the dedicated batteries. The forced circulation to natural circulation design of the offsite power feeders and the cooling could be accomplished when decay onsite emergency diesel generators is not heat reached 1 MW. However a review of adequate to assure that they would survive these analyses resulted in a concern that the extreme natural phenomena such as time necessary to reach the transition tornadoes or earthquakes. Thus, it is likely point may be longer than that originally that it would be necessary to rely on the estimated in the Accident Analysis. dedicated batteries to power the pony Therefore, an extensive analytical effort was motors for such events. This type of event initiated to determine the time period in is analogous to an extended "station which forced circulation would be required. blackout" condition recently addressed by The analytical approach used in the effort the Nuclear Regulatory Commission. was application of a power reactor thermal Due to the uncertainty in the decay analysis code, RELAP5, modified to heat calculations and the uncertainty represent the HFIR conditions. Best- associated with restoration of offsite power estimate calculations using this approach or repair of the onsite emergency power indicated that forced circulation would be supply, it could not be established that the necessary for 2 to 4 hours following reactor capacity of the dedicated batteries would be shutdown from the rated 85 MW power sufficient to provide forced circulation for level. However, because of uncertainties in the required time. Failure to provide the the analysis models, input parameters, and necessary forced circulation could lead to correlations, the calculated time could not core damage in the reactor. Although this be conservatively bounded without would not be expected based on the best- experimental validation. Development of estimate calculations, it was determined that such experimental data would be time it would be appropriate to undertake consuming and difficult to accomplish in the additional measures to assure that adequate near term. decay heat removal would be provided for The pony motors on the primary the unlikely event of an extended station coolant pumps are normally powered blackout. through inverters from the offsite ac power Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

312 OPTIONS CONSIDERED extended time. The siting criteria for the Various options were considered in units was to include a requirement that the developing a plan to address this potential units would have a high probability of vulnerability. The first option investigated surviving severe natural phenomena, such as was the possibility of procuring larger earthquakes or tornadoes. The units would capacity batteries. It was determined that be moved to the site and connected to the the space requirements for replacement pony motor circuits. The system was batteries were larger than the space designated the Auxiliary Emergency Power available. In addition, due to the Generator System (AEPGS). uncertainty associated with the decay heat calculations, it was not certain what battery DESCRIPTION OF AUXILIARY capacity would be adequate. This option EMERGENCY POWER GENERATOR was not pursued further. SYSTEM The feasability of removing the top The AEPGS consists of two mobile, hatch from the reactor after an appropriate redundant diesel generator units each time was investigated. This would allow capable of meeting the power requirements establishment of a natural convective flow of the four pony motor battery chargers and from the reactor to the pool. However, it associated equipment during station blackout was determined that the weight of the hatch conditions. Each portable generator has a is greater than the capacity of the manual 50-kW continuous duty rating to assure that bridge crane. Of course, power for the adequate power is provided. Each unit is overhead crane would not be available in capable of being positioned and manually such situations. Therefore this option was connected to the existing pony motor discarded. electrical supply system in less than six A variation of this option was also hours. considered. The possibility of removing the During a station blackout condition, hatch over the target area was considered. one of the two portable generators will be It was determined that it could not be positioned near the pony motor battery assured that adequate flow area would be room. Five 300-foot long trail cables supply available unless the target tower was also power from the AEPGS to the four pony removed. Removal of the target tower motor battery chargers and the battery room requires the overhead crane; thus, this exhaust fan as required. Each generator is option was also dismissed. equipped with a 50-gallon fuel tank which The cost of upgrading the emergency provides approximately 15 hours of diesel generators and associated equipment operation. Portable, flexible, 100-foot long was also investigated. This option would fuel lines are also included and can be con- have essentially required complete nected to the existing, seismically qualified replacement of the current equipment and 4,000-gallon underground diesel fuel tank to structures. Therefore this course of action provide an extended fuel supply for up to was considered to be too expensive. 49 days of operation. The only change to The final option considered and the the existing facility was the addition of quick option chosen was to procure an auxiliary connect/disconnect fittings and a check valve power source to provide forced primary to the existing 4,000-gallon diesel fuel tank coolant circulation via the pony motors. supply and return lines and modification of This was to be accomplished by a set of the electrical supply for the battery room portable dicscl generators. These would be exhaust fan to allow operation from the stored near the site such that they would be AEPGS. accessible in the unlikely event that both One of the two AEPGS generators offsite and onsite ac power were lost for an is stored outside at the Molten Salt Reactor Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

313 (MSR) site approximately 1,700 feet north experimental data are developed to of the HFIR. The second generator is demonstrate that the transition from forced stored at the ORNL main complex, circulation to natural circulation cooling can approximately 5,000 feet east-northeast of be accomplished within the 6-hour capacity the MSR and 6,500 feet northeast of the of the dedicated batteries, use of the HFIR. Both generators are secured to the portable, AEPGS can be discontinued ground with quick release cables to keep without significant economic penalty. them from turning over during a severe The consideration of industry wind storm. Power has been provided at precedent took the form of a review of the storage locations for battery chargers guidance from the Nuclear Regulatory and water jacket heaters on the AEPGS Commission regarding design requirements units to maintain operational readiness. for station blackout and onsite power Analyses were completed by EQE supplies. Documents reviewed included [1], under contract to ORNL, to evaluate guidance in the Standard Review Plan for the vulnerability of the AEPGS to seismic Safety Analysis Reports [2] and Regulatory events and tornadoes. The approach taken Guide 1.155 [3]. In particular, it was noted was to combine deterministic analyses, field that guidance in Regulatory Guide 1.155 evaluations, probabilistic analyses, and allows an alternate AC power supply in the application of EQE's earthquake experience form of commercial grade, portable diesel data base. The evaluation concluded that generator units. The AEPGS conceptual the system is capable of surviving the 0.15 g design was compared to the guidance in the HFIR evaluation earthquake and that the Regulatory Guide and the Standard Review probability of concurrent damage to all Plan and found to meet or exceed all emergency power sources due to the 150 requirements [4]. mph evaluation tornado is acceptably small, approximately 2 E-6 per year. BENEFIT RESULTING FROM THE AEPGS BASES FOR CHOOSING THIS OPTION A Probabilistic Risk Assessment of The selection of a portable auxiliary the HFTR had been completed for ORNL power supply was based on considerations by Pickard, Lowe, and Garrick (PLG) [5]. of practicality, cost, and industry precedent. However, the study was issued prior to the All major components in the system are determination that the time required for commercial grade items and are thus readily forced circulation may have been available. A minimum amount of underestimated in the Safety Analysis. modification to the portable diesel generator Therefore, the effect of the AEPGS was units was required to provide the not included in that study. To estimate the connections necessary for mating the system reduction in core damage frequency with the existing electrical supply for the attributable to the AEPGS, the system was pony motors. Similarly, only minor evaluated by PLG for three event severities modifications were required in existing [6]. Effects of component failure, human equipment and structures to accommodate error, and inability to move the units to the the addition of the AEPGS. From a cost- site were considered in the evaluation. benefit standpoint, this made the option an The least severe event considered attractive alternative. Use of commercial was a station blackout which affects the grade components avoided the extra cost of HFIR area only. This type of event could procuring "nuclear qualified" components. be caused by a fire in the electrical The cost of additional major modifications equipment building which damages the to existing equipment and structures was offsite power feeders and the onsite also avoided. Finally, if subsequent emergency dicsel generators. The second Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

314 level of event severity was a station would thus be reduced by a factor of about blackout which affects the overall ORNL fifty-nine (i.e. AEPGS conditional area and results in minor damage to access unavailability for this severity level, 1.7 E-2, routes leading to HFIR. Such an event times the earthquake frequency). could be characterized by the HFIR evaluation earthquake. The final and most CONCLUSION severe event considered was a station A portable auxiliary power supply to blackout event affecting the overall ORNL backup the offsite and onsite power supplies area in combination with significant damage provides a high level of assurance that to the HFIR area and other areas at forced circulation will be available for the ORNL, including partial blocking of access required time period following reactor routes. shutdown. This approach is consistent with The probability that at least one of guidance from the commercial nuclear the two AEPGS units could be connected power industry. It is a cost-effective means to the pony motor battery system within six of achieving the desired capability. If hours and would operate for an additional experimental data are developed in the 18 hours was calculated for the three event future such that the capability is no longer severity levels. Results of the analysis needed, the plant can easily be restored to indicated that the conditional unavailability the original configuration if desired. of the AEPGS ranged from about 1.6 E-2 for the least severe event category, to about 1.7 E-2 for the second severity level, to about 6.0 E-2 for the most severe event category. These are believed to be somewhat conservative; since results of emergency drills have indicated that the units could be transported to the site and connected in about a third of the time assumed in the probabilistic analysis. Blockage of access routes was not simulated in the emergency drills, however. The effect that the AEPGS has on reducing core damage frequency can be seen by considering the following. Under the assumptions of the PRA, the HFIR evaluation earthquake would lead to a station blackout because neither the offsite power supply nor the onsite emergency diesels are qualified to withstand the event. Since it can not be assured that the capacity of the dedicated batteries is sufficient to provide forced circulation for the required time, a core damage event is assumed to result. Therefore the frequency of the event is also the core damage frequency contribution. For HFIR, the evaluation earthquake has a frequency of about 1.4 E- 3 per year. By adding the AEPGS, this contribution to core damage frequency Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

~~315 REFERENCES~~

~~[1] "Seismic Evaluation of the Auxiliary Electric Power System at the High Flux Isotope Reactor," EQE Report No. 87257.0l-R-002, March 9, 1989.~~

~~[2] NUREG-0800, U.S. NRC Standard Review Plan Section 8.3.1, "AC Power Systems (onsite)," July 1981.~~

~~[3] U.S. NRC Regulatory Guide 1.155 Revision 1, "Station Blackout," August 1988.~~

~~[4] "Applicability of Regulatory Guide 1.155 (Station Blackout) and Standard Review Plan 8.3.1 to the High Flux Isotope Reactor Auxiliary Emergency Power Generator," ORNL/RRD/INT-18, September 1988.~~

~~[5] "The High Flux Isotope Reactor Probabilistic Risk Assessment," ORNL/RRD/INT-36, January 1988.~~

~~[6] "High Flux Isotope Reactor Auxiliary Emergency Power Generator Study," Pickard, Lowe, and Garrick, Inc. March 3, 1989.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

316 EARTHQUAKE STRONG GROUND MOTION STUDIES AT THE IDAHO NATIONAL ENGINEERING LABORATORY Ivan Wong, Walter Silva, Robert Darragh, Cathy Stark and Douglas Wright Woodward-Clyde Consultants 500 12th Street, Suite 100 Oakland, CA 94607 Suzette Jackson, Glen Carpenter, Richard Smith, Dennis Anderson, Hollie Gilbert and Don Scott Idaho National Engineering Laboratory EG&G Idaho, Inc. P.O. Box 1625 Idaho Falls, ID 83415

ABSTRACT Site-specific strong earthquake ground motions have been estimated for the Idaho National Engineering Laboratory assuming that an event similar to the 1983 Mg 7.3 Borah Peak earthquake occurs at epicentral distances of 10 to 28 km. The strong ground motion parameters have been estimated based on a methodology incorporating the Band-Limited-White-Noise ground motion model coupled with Random Vibration Theory. A 16-station seismic attenuation and site response survey utilizing three-component portable digital seismographs was also performed for a five-month period in 1989. Based on the recordings of regional earthquakes, the effects of seismic attenuation in the shallow crust and along the propagation path and local site response were evaluated. This data combined with a detailed geologic profile developed for each site based principally on borehole data, was used in the estimation of the strong ground motion parameters. The preliminary peak horizontal ground accelerations for individual sites range from approximately 0.15 to 0.35 g. Based on our analysis, the thick sedimentary interbeds (greater than 20 m) in the basalt section attenuate ground •otions as speculated upon in a number of previous studies.

INTRODUCTION along the southern segments of the On 28 October 1983, a surface wave northwest-trending Lost River and Lemhi magnitude M 7.3 (moment magnitude faults respectively (both Basin and M 6.9) earthquake occurred in the Range normal faults), have uncovered vicinity of Borah Peak along the Lost evidence for multiple earthquakes. The River Range approximately 90 km most recent events are characterized by northwest of the Idaho National offsets of more than 3 m that possibly Engineering Laboratory (INEL) which is occurred approximately 15,000 to 30,000 located within the eastern Snake River years ago [1]. These data suggest that Plain (Figure 1). Previous geologic there exists a potential for future studies of the Arco and Howe scarps earthquakes similar to the 1983 Borah

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~317 111" ___~~

~~^ ClAKK CANHOK WYOMING T mnexvom tAKf. .~~

~~J-«"v, \ \ West o \ si YELLOWSTONESTONE%\ \ LAKE *W~~

~~, JACKSON + ! / S I LAKE~~

~~Howev ,v Rexourg I~~

~~» j* /~~

~~/ L.™"Z_.^T. ..J NATIONAL / A V. ENGINEERING / Jackson I.ABORATORV ,'~~

~~HESERVOm BlacKfoot~~

~~AMERICAN FALLS RESEKVOm, Sprinns . Alton~~

~~Twin Fails Burley / 5 / z v~ Cl Montpelier * • !o \y Malad City Z BEAft °;i 2. LAKE 5:° *L ID*HO J-~~

~~Magnitude ML A WCC/EG & G Seismographic Station~~

~~O =1.0 A INEL Seismographic Station~~

Figure 1. Regional sttting of the INEL and locations of earthquakes recorded by the seismic survey. Also shown are the permanent stations of the INEL seismographic network. Peak event which may occur at close motion have been performed for the INEL distances to facilities at the INEL. In based on data from earthquakes occurring the past, empirical studies of ground in other regions worldwide. This is due

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

318 to the lack of recorded strong ground An additional important aspect of motion data for distances less than 90 the BLWN model employing a simple source km at the INEL, or for that matter, the model is that the site- and region- intermountain U.S. For this reason and dependent parameters can be evaluated by because of the uncertainties regarding observations of local or regional propagation path and site response earthquakes. Thus, results from the effects on ground motions, site-specific seismic survey will provide data for strong ground motion parameters for the input into the strong motion predictions INEL are required for input into seismic in addition to providing site-specific safety analyses. This paper describes ground-motion recordings to calibrate (1) the studies being jointly performed the BLWN-RVT model. by Woodward-Clyde Consultants (WCC) and EG&G Idaho, Inc. to assess potential Previous Studies earthquake strong ground motions at the In previous evaluations of potential INEL and (2) the preliminary estimates strong-ground motions at the INEL dating of strong ground motion that might be back to the early 1970's, the sources of experienced during a hypothetical the design earthquakes have been the earthquake, similar to the 1983 Borah southern segments of the Lost River Peak event, occurring along the southern fault (near Arco) and the Lehmi fault segment of the Lemhi fault. (near Howe) (Figures 1 and 2). The magnitudes of the events have generally Scope of Work ranged from 6 3/4 to 7 3/4 with Specific objectives of this study resulting peak horizontal ground are: 1) provide site-specific estimates accelerations of 0.15 to 0.45 g at the of peak horizontal ground acceleration ground surface [2]. (For other and response spectra for selected sites pertinent discussions, see Harris [2] located on soil or bedrock; 2) develop a and Dahlke and Secondo [3] in this peak acceleration-attenuation • volume.) relationship for earthquakes in the In 1977, Agbabian Associates [4] range M 6 to 8 that is specific to the reviewed the evidence for two factors INEL; 3) perform a seismic survey of that had been speculated upon as selected sites using portable digital possibly diminishing the levels of seismographs; and 4) process and analyze earthquake ground motions within the the data recorded by the survey to Snake River Plain: (1) "decoupling" of evaluate local site response and seismic the Plain by perimeter faulting and attenuation along the propagation path (2) attenuation due to the interbedded and in the very shallow crust. alluvial layers within the basalts. It In this study, the strong ground was concluded that there was no evidence motion parameters will be estimated at that time to indicate the existence based on a methodology incorporating the of either process. The latter was Band-Limited-White-Noise (BLWN) ground considered unlikely because the seismic motion model coupled with Random waves would not be affected by the Vibration Theory (RVT). The BLWN model interbeds due to the differences between incorporates the general characteristics their much larger wavelengths and the of the source and wave propagation as thinness of the interbeds. well as propagation path and site In 1983, the Borah Peak earthquake effects. The model is appropriate for occurred providing the first strong an engineering characterization of motion records at the INEL although at ground motion since it captures the distances exceeding 90 km. Thirteen general features of strong ground motion accelerographs which were the closest in terms of peak acceleration and instruments to the event were triggered, spectral composition with a minimum of recording peak horizontal accelerations free parameters. ranging from 0.022 to 0.078 g at

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~319 "UINEL Boundary~~

~~MIDDLE BUTTE~~

~~| '"<- SOUTHERN "W~ BUTTE~~

Figure 2. Locations of seismic stations and major facilities at the INEL. Also shown is the surface projection of tht hypothetical "Howe" earthquake rupture plane on the southern segment of the Lemhi fault. ' basement or free-field sites at the earthquake, Jackson and Boatwright [6] facilities ANL, ATR, CPP and PBF [5] calculated values of 0.21 to 0.5A g for (Figure 2). In an attempt to estimate distances of 18 to 11 km, respectively, near-field accelerations for the 1983 based on the observed attenuation of the Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

320 largest aftershock. These estimates, Amplification by near-surface velocity however, may only be appropriate for a gradients is accounted for in the site located within the Basin and Range detailed site-specific velocity models. province and probably not for the Snake The P(f) filter is an attempt to River Plain province. model the observation that acceleration spectral density falls off rapidly APPROACH beyond some region-dependent maximum frequency. This observed phenomenon Band-Limited-White-Noise Point Source truncates the high frequency portion of Model the spectrum and is responsible for the The BLWN ground motion model first band-limited nature of the stochastic developed by Hanks and McGuire [7J model. Following the Anderson and Hough (sometimes referred to as the stochastic [14] attenuation model, the form of the model) in which the energy is P(f) filter is taken as distributed randomly over the duration of the source has proven remarkably P(f) = (2) effective in correlating with a wide range of ground motion observations where r is epicentral distance and <(r) [7,8,9]. The ground motion model uses is a distance-dependent damping factor an u-square Brune source model with a that represents the loss in energy of single corner frequency and a constant- the seismic waves as they propagate. At stress parameter [8], RVT is used to zero epicentral distance, < represents relate rms (root-mean-square) values to energy attenuation in the shallow crust peak values of acceleration [8], and beneath the site [15]. oscillator response [10,11] computed As r increases, the rays penetrate from the power spectra to expected peak deeper into the crust where the time domain values. Details of the mechanism of attenuation contributes a methodology are described in Silva et frequency dependence to the damping. al. [12] and Silva and Darragh [13]. This part of the path attenuation is The shape of the acceleration generally modeled with a frequency spectral density a(f) is given by dependent Q of the form irfR (3) M Q(f) - Qo (f) a(f)=C P(f)A(f)e 2 R o l+(f/fc) (1) where Q is the reference Q at frequency o* where H is the seismic moment, R the To summarize, the parameters that hypocentral distance, 6 the shear wave are required for the ground motion velocity at the source, Q(f) the estimates are the source parameters of frequency dependent quality factor, A(f) the earthquake (seismic moment or moment the near-surface amplification factors, magnitude, stress parameter, and source P(f) the high-frequency truncation depth), shortest distance to the rupture filter, f the source corner frequency, plane, propagation path parameters and C a constant. (Q(f), shear wave velocity [V ], and Source scaling is provided by density [p]) and the detailed geology specifying two independent parameters, beneath each site as characterized by the seismic moment (M ) and the high- Vs, p and Qg for each layer. Modulus frequency stress parameter and damping curves modified from Seed Ao relates f to M . The spectral shape and Idriss [16] are used to characterize of the source model is then described by the response of the soils in the the two free parameters M and Aa. f Q c geologic profiles having Vg less than increases with the shear-wave velocity 0.75 km/sec. and with increasing stress parameter, both of which are region dependent. Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

321 Seismic Attenuation and Site Response Survey In early February 1989, the first of 16 stations in the seismic survey were installed either at or near facilities of interest at the INEL or as part of an array aligned approximately with the trend of active earthquake sources within the region (Figure 2). A desire to have stations located on a variety of subsurface geology was also considered in the site selection. Finally, the need to avoid high levels of ground 0. 8. 16. 24. 32. 40. 48. 56. noise due to activities associated with Time (seconds) the operations at the INEL and to be near a borehole in which detailed information on the subsurface geology was available, governed the exact station locations. Each site was equipped with a Sprengnether DR-100 digital event recorder and an orthogonal three- component set of Mark Products L-4C or Teledyne-Geotech S-13 1.0 Hz seismometers. Data were recorded at 100 sps per channel and bandpass-filtered between 0.2 and 30 Hz. The seismometers were generally buried to a depth of 1 m 0. 8. 16. 24. 32. 40. 48. 56, to minimize wind noise. Power was Time (seconds) provided by external batteries recharged by solar panels. Digital cassette tapes Figure 3. Typical regional earthquake as recorded by the were generally changed every two days seismic survey. This event (M[_ 3.5) occurred and the internal clocks of the DR-100's near Hebgen Lake (Figure 1) on 6 June 1989. calibrated with a portable reference operate the Utah regional network, the clock. Station locations were Montana network and the Teton network, determined using a portable Satellite respectively. A few events, principally Navigation System. Calibrations of the in the Borah Peak area, were located by digital seismographs were performed at the INEL eastern Snake River Plain the beginning, middle and end of the network (Figure 1). The best recorded survey in mid-July. earthquakes were two events of My > 4 Approximately 40 regional which occurred at the Utah-Idaho border earthquakes of sufficient magnitude in early July and two ML > 3.5 events (local magnitude ML > 2.5) were well- near Jackson, Wyoming in late June recorded on 200 seismograms during the (Figure 1). seismic survey (Figures 1 and 3). All events were processed from the data INPUT PARAMETERS cassettes and the largest are being used in the analysis of K and Q. Hypocentral Earthquake Source Parameters locations and magnitudes for most of the An earthquake similar to the 1983 earthquakes were provided by the Borah Peak earthquake but occurring University of Utah Seismograph Stations along the southern segment of the Lemhi (UUSS), the Montana Bureau of Mines and fault near the town of Howe was assumed Geology (MBMG), and the U.S. Bureau of as the maximum earthquake pertinent to Reclamation (USBR). These agencies seismic safety (Figure 2). The Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

322 following source parameters were assumed most of the propagation path was within based on the 1983 earthquake [17]: the Basin and Range province. For the Howe earthquake along the southern Lemhi moment magnitude M 6.9 fault at the edge of the Snake River 26 seismic moment MQ 2.1 x 10 Plain, a QQ of 200 and n of 0.55 may be dyne-cm more appropriate (Q was assumed to be stress parameter 75 bars equal to Q ). A technique which fits source depth 16 km the non-linear model in Equation 1 to observed spectra from selected regional The stress parameter is probably an earthquakes using the Levenberg- upper bound value based on the estimates Marquardt method is presently being of the static stress drop for the 1983 performed to further refine the values event [17]. Turko [18] has estimated a of Q and r\. maximum earthquake of M 6.9 for the southern segment of the Lemhi fault Site Parameters based on its estimated rupture length. Three site parameters need to be A conservative assumption of a maximum specified as a function of depth for the magnitude of M 6.9 (M 7.3) incorporates BLWN-RVT model: Vg, Qs and p. Thus the estimated uncertainty in the geologic profiles were developed for the definition of this segment and the individual sites. Stratigraphic data assumption that only this segment will for the upper 200-300 m at each site rupture in a maximum event. were extracted from the nearest A hypothetical rupture plane for the available well or borehole. The "Howe" earthquake is assumed to be a 45° remainder of the 3-km geologic profiles southwest-dipping normal fault with an were characterized based upon data from initial point of rupture at 16 km depth a 3159 m deep exploratory well (INEL-1). at the southwestern corner of the Much of what is known about the rupture plane identical to the 1983 subsurface geology beneath the I NET. is Borah Peak event [17] (Figure 2). based on INEL-1. The lithologic log of Closest distances to the plane of INEL-1 shows that at least 52 rupture are conservative and range from distinctive layers were encountered 10 to 28 km for the various facilities. [22]. The upper section above a depth of 745 m consists of basaltic Lava flows Propagation Path Parameters with interbedded sediments of alluvial, For the propagation path between the Lacustrine and volcanic origin. The closest point of the rupture plane of Lower section consists principally of the Howe earthquake and the sites, a rhyolitic welded ash-flow tuffs half-space model characterized by a V interbedded with devitrified of 3.55 km/sec and a p of 2.7 g/cm was rhyolites. A thick (87 m) tuffaceous assumed based on Sparlin et al. [19]. Layer separates the basalt and welded Based on an analysis of L waves, Singh tuff sections. At a depth below and Herrmann [20] determined a regional approximately 2460 m, a rhyodacite crustal coda Q of 450 and a n of 0.2 porphyry was encountered [22],

for Q(f) (see equation 3). In contrast, Values of Vg and p for each layer in Braile et al. [21] observed high the geologic profiles were estimated attenuation in a seismic refraction from (1) sonic (V ) and density logs experiment within the eastern Snake performed in INEL-1 and INEL 2-2A, a River Plain and they attribute it to low 915-m deep hole and (2) laboratory Q values in the volcanic rocks (Q 20 to measurements of core samples from the 200) and throughout the crust (Q 160 to two holes under in-situ conditions of 300) where Q is the P-wave quality temperature and pressure. Uncertainties factor. Thus for preliminary modeling in V values are greatest for the of the 1983 Borah Peak earthquake, a QQ sedimentary interbeds and devitrified of 45C and n of 0.2 were assumed because rhyolite compared to the volcanic rocks.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

323 For the soil sites, we have assumed o a V of 0,41 km/sec and a p of 2.00 o g/cm3 for the Lost River flood-plain sands, silts, and gravels overlying the basalt. From approximately 3 km (the bottom of INEL-1) to a depth of 5 km. a 3 Vs of 3.05 km/sec and p of 2.54 g/cm were assumed based on Spar1 in et al. [19]. Preliminary values of Q were estimated based on V for each layer. 0) The resulting Qs profiles contained average values over the 5 km depth in en the range 100 to 150, which is conservative when compared with the Avar«gaof ANL, ATR average Q structure beneath the Snake and PBF Racords Extrama of ANL, ATR River Plain to this depth as proposed by and PBF Racords Braile et al. [21]. K and its standard error at each site are being estimated by least' squares fits to the distance-corrected 10'2 10 -1 10 ° 10 spectra of selected regional events with Period (seconds) ML > 3.0. These events were chosen because they have the largest signal-to- Figure 4. Absolute acceleration response spectra (5% damp- noise ratios at high frequencies. In ing) comparing the average of 10 INEL horizon- addition, several events were large tal strong motion records on rock of the 1983 enough to be recorded on many stations Borah Peak earthquake and the BLWN-RVT in the seismic survey. These recordings _ model prediction at INEL-1. are presently being used in spectral- ratio analyses to define the differences in K between the various stations. These two procedures will help constrain the site-specific attenuation in our final models.

RESULTS AND DISCUSSION Model Predictions of the 1983 Borah Peak —O) Mainshock Ground Motions Of the eight lowest structural level en or free-field strong motion sites at the — Av«r»8» of ATR INEL, five sites were founded on basalt Racords or in the basements of structures with Extramaof ATR Racords piers embedded in basalt and three were Racords soil sites. Employing the BLWN-RVT • LWN-RVT Pradiction methodology, absolute acceleration o response spectra (5Z damping) for the 10-2 10 "] 10 ° 1983 earthquake for several of the 10 strong motion sites were estimated to Period (seconds) compare with averaged response spectra Figure 5. Absolute acceleration response spectra (5% damp- from the actual recordings. The ing) comparing the average of four horizontal predicted spectra for the INEL-1 rock records of the 1983 Borah Peak earthquake as site and a rock and soil site at ATR are recorded at ATR (see Figure 2) and BLWN-RVT shown in Figures 4 and 5. model predictions for a soil and rock site at ATR.

~~Second DOE Natural Phenomena Hazards Mitigation Coiiference - 1989~~

324 In general, the spectra predicted from the ground motion model lie within the extretna of the data (Figures 4 and 5) and are a good representation of the averages of the data. The observed peak accelerations, however, tend to be slightly higher than those predicted by the preliminary modeling. For the ATR site, the model predictions for rock closely match the observed spectra over the displayed range of periods compared to the predictions for soil (Figure 5). This may be due to the fact that the strong motion sites are located in the basement 5%, Standard gaologic profila of a building supported by piers or -5%, Gaologic profila with no intarbadt columns that are embedded in the basalt in lowtr basalt layer. The inference is that the soil layer does not have a large effect on the response of this structure due to -2 1 its support in the basalt. 10 10 - 10 10 Period (seconds) Site-Specific Ground Motion Estimates Figure 6. Predicted acceleration response spectra of the For the site-specific estimates of Howe earthquake at the INEL-1 rock site with strong ground motion from the Howe and without sedimentary interbeds in the lower earthquake, values of peak horizontal basalt section. ground acceleration and response spectra from 207 to 660 m replaced with a (5% damped) have been computed. homogeneous basalt layer is also shown Preliminary peak accelerations for the in Figure 6. As expected, the response various sites range from approximately spectrum of the altered INEL-1 profile 0.15 to 0.35 g. shows fewer resonant peaks, some The preliminary peak horizontal shifting in the frequency of resonant acceleration predicted for the INEL-1 peaks, and generally a higher level of site (rock) from the Howe earthquake is ground motions. The peak horizontal 0.36 g at a distance of 16.9 km. The acceleration based on the detailed model acceleration response spectrum for the is 0.36 g compared to 0.43 g for the INEL-1 site exhibits substantial detail, profile with the homogeneous lower i.e. resonant peaks, as might be basalt section. This 19% difference in expected for a site-specific spectrum peak values reflects the attenuation of (Figure 6). These peaks are the result the seismic waves by the thicker of large velocity contrasts at geologic sedimentary interbeds (> 20 m?) in the contacts which are numerous in the lower basalt section as the waves complex INEL-1 geologic profile. The propagate up through the near-surface preliminary estimation of ground motions geology. This effect of the thicker for a shallow soil site (approximately interbeds, which has been speculated 12 m of alluvial soil overlying basalt) upon in a number of previous studies shows substantial amplification [4], appears to be a significant factor resulting in an approximate 1.5 increase in lowering potential strong ground in peak horizontal ground acceleration motions at the INEL. compared to a corresponding rock site. For the purposes of comparison, peak A comparison of 5% damped absolute horizontal ground accelerations for the acceleration response spectra based on INEL-1 site were computed based on the the INEL-1 geologic profile and the empirical acceleration-attenuation profile with the detailed basalt section relationships of Seed and Idriss [23] Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

325 (probably the most widely used SUMMARY relationship in the earthquake Site-specific strong ground motion engineering community), Idriss [24], estimates in terms of peak horizontal Joyner and Boore [25] and Campbell [26] ground acceleration and absolute (unconstrained) and are shown in Table 1 acceleration response spectra are with the preliminary estimate obtained presently being determined for various in this study. The relationship by Seed facilities at the INEL based on the and Idriss is based on rock site BLWN-RVT methodology. A normal faulting recordings. The Idriss, Joyner and earthquake of M 6.9 (Mg 7.3) similar to Boore, and Campbell relationships the 1983 Borah Peak earthquake but incorporate data from selected soil occurring along the southern segment of sites. However, according to Joyner and the Lemhi fault is considered as the Boore [25], a statistical evaluation of maximum event for seismic safety their data suggests that their analyses. Such an earthquake would relationship is also directly applicable occur at closest rupture distances of 10 to rock sites. All the relationships to 28 km from various facilities. provide estimates of free-field surface Detailed geologic profiles based ground motion. The preliminary peak principally on borehole data have been acceleration for the INEL-1 site developed for each site and used in the estimated in this study is in general ground motion estimates. Estimates agreement with the values computed from of K and Q(f) are being obtained from an the empirical relationships (Table 1). analysis of the regional earthquakes Differences are not unexpected, however, recorded during the seismic survey. given the site-specific nature of our The near-surface geology, preliminary estimate compared to the specifically the shear wave velocity empirically-based estimates which are contrasts between the basalt layers and generally dominated by California strong the sedimentary interbeds and the Qs ground motion data. assigned to each layer, have a major effect on the ground motion estimates. The stress parameter also significantly Table 1 influences the level of ground Comparison of Median Peak motions. The preliminary peak Horizontal Accelerations at the horizontal ground accelerations for INEL-1 Rock Site for the various sites at the INEL range from Howe Earthquake approximately 0.15 to 0.35 g depending principally upon the distance to the Distance PGA rupture plane of the hypothetical Howe Relationship Magnitude (km) (g) earthquake, and the details of the

Seed and Ms 7.3 10.6* 0.45 geologic profiles. The thicker Idriss sedimentary interbeds in the basalt section and the tuffaceous interbed Joyner and M 6.9 10• 6* 0.30 separating the basalt and welded tuff Boore sections attenuate ground motions

Campbell Ms 7.3 16. 9** 0.23 according to the BLWN-RVT methodology as speculated upon in a number of previous Idriss Ms 7.3 16.9** 0.29 investigations. Currently, the effects This study M 6.9 16.9** 0.36 of a finite earthquake source on the ground motion estimates, e.g., radiation * Shortest horizontal distance to pattern and rupture directivity, are surface projection of rupture being evaluated for sites at the INEL. plane. ** Shortest distance to rupture plane.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

326 ACKNOWLEDGMENTS Earthquake, U.S. Geological Survey Our appreciation to the DOE, EG&G Open-Pile Report 85-290, pp. 385- Idaho, Inc., Westinghouse Idaho Nuclear 400, 1985. Co., Westinghouse Electric Corp., and [6] S. M. Jackson and J. Boatwright, Argonne National Laboratory for support "Strong ground motion in the 1983 of these studies and in particular to Borah Peak, Idaho earthquake and Bob Secondo, Al Bowman, Bob Guenzler, its aftershocks," Bulletin of the Vince Gorman, Hans Dahlke and Earl Seismological Society of America, Krenz. We would like to acknowledge the vol. 77, pp. 724-738, 1987. contributions of Russell Sell, Jim [7] T. C. Hanks and R. K. McGuire, Humphrey, Sam Spencer, Dewitt Cheng, "The character of high-frequency Pete Knuepfer, Diane Doser, Don Armour, strong ground motion," Bulletin of Holly Adams and Susan Martin. We are the Seismological Society of grateful to Mike Stickney (MBMG), Chris America, vol. 71, pp. 2071-2095, Wood (USBR) and Linda Sell (UUSS) for 1981. providing earthquake data. Our thanks [8] D. M. Boore, "Stochastic to Tom McEvilly and Ernie Majer of simulation of high-frequency Lawrence Berkeley Laboratory for the use ground motions based on of their instruments. seismological models of the radiated spectra," Bulletin of the REFERENCES Seismological Society of America, [1] H. E. Malde, "Quaternary faulting vol. 73, pp. 1865-1884, 1983. near Arco and Howe, Idaho," [9] D. M. Boore and G. M. Atkimon, Bulletin of the Seismological "Prediction of ground motion and Society of America, vol. 77, spectral response parameters at pp. 847-867, 1987. hard-rock sites in eastern North [2] B. G. Harris, "Seismic design America," Bulletin of the criteria at the Idaho National Seismological Society of America, Engineering Laboratory," in vol. 77, pp. 440-467, 1987. Proceedings of the Second DOE [10] D. M. Boore and W. B. J^yner, "A Natural Phenomena Hazard note on the use of random Mitigation Conference, 1989 (this vibration theory to predict peak volume). amplitudes of transient signals," [3] H. J. Dahlke and R. J. Secondo, Bulletin of the Seismological "Comparison of proposed seismic Society of America, vol. 74, pp. design criteria for the INEL," in 2035-2039, 1984. Proceedings of the Second DOE [11] W. J. Silva and K. Lee, "WES Natural Phenomena Hazard RASCAL code for synthesizing Mitigation Conference, 1989 (this earthquake ground motions, State- volume). of-the-Art for Assessing [4] Agbabian Associates, "Evaluation Earthquake Hazards in the United of seismic criteria used in the States," Report 24, U.S. Corps of design of INEL facilities," Engineers Waterways Experiment unpublished report prepared for Station, Miscellaneous Paper S-73- the Energy Research and 1, 120 pp., 1987. Development Administration, INEL, [12] W. J. Silva, R. B. Darragh, R. K. 151 pp., 1977. Green, and F. T. Turcotte, [5] S. M. Jackson, "Acceleration data "Spectral characteristics of small from the 1983 Borah Peak, Idaho magnitude earthquakes with earthquake recorded at the Idaho application to western and eastern National Engineering Laboratory," North American tectonic in Proceedings of Workshop XXVIII environments," Report to U. S. On the Borah Peak, Idaho, Army Corps of Engineers Waterways

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

327 Experiment Station, Vicksburg, Ansorge, M. R. Baker, M. A. Mississippi, 84 pp., 1968. Sparlin, C. Prodehl, N. M. [13] W. J. Silva and R, B. Darragh, Schilly, J. H. Healy, St. Mueller, "Engineering charac'cerization of and K. H. 01 sen, "The Yellowstone- strong ground motion recorded at Snake River Plain seismic rock sites," unpublished report profiling experiment: Crustal prepared for Electric Power structure of the eastern Snake Research Institute, 1989. River Plain," Journal of [14] J. G. Anderson and S. E. Hough, "A Geophysical Research, vol. 87, model for the shape of the Fourier pp. 2597-2609, 1982. amplitude spectrum of acceleration [22] D. J. Doherty, L, A. McBroome, and at high frequencies," Bulletin of M. A. Kuntz, "Preliminary the Seismological Society of geological interpretation and America, vol. 74, pp. 1969-1993, lithologic log of the exploratory 1984. geothermal test well (INEL-1), [15] S. E. Hough and J. G. Anderson, Idaho National Engineering "High-frequency spectra observed Laboratory, Eastern Snake River at Anza, California: Implications Plain, Idaho," U.S. Geological for Q structure," Bulletin of the Survey Open-File Report No. Seismological Society of America, 79-1248, 10 pp., 1979, vol. 78, pp. 692-707, 1988. [23] H. B. Seed and I. M. Idriss, Ground motions and soil [16] H. B. Seed and I. M. Idriss, "Analysis of ground motions at liquefaction during earthquakes, Union Bay, Seattle during Earthquake Engineering Research earthquakes and distant nuclear Institute Monograph Series, blasts," Bulletin of the vol. 4, 134 pp., 1982. Seismological Society of America, [24] I. M. Idriss, "Evaluating seismic vol. 60, pp. 125-136, 1970. risk in engineering practice," in [17] D. I. Doser and R. B. Smith, Proceedings, Eleventh "Source parameters of the 28 International Conference on Soil October 1983 Borah Peak, Idaho Mechanics and Foundation earthquake from body wave Engineering, San Francisco, vol. analysis," Bulletin of the 4, pp. 255-320, 1985. Seismological Society of America, [25] W. B. Joyner and D. M. Boore, vol. 75, -pp. 1041-1066, 1985. "Measurement, characterization and [18] J. M. Turko, "Quaternary : prediction of strong ground segmentation history of the Lemhi motion," in Proceedings of the fault Idaho," M.A. Thesis, State Conference on Earthquake University of New York of Engineering and Soil Dynamics: Binghampton, 91 pp., 1988. Recent Advances in Ground Motion [19] M. A. Sparlin, L. W. Braile, and Evaluation, American Society of R. Bo Smith, "Crustal structure of Civil Engineers, pp. 43-103, 1988. the eastern Snake River Plain [26] W. W. Campbell, "Predicting strong determined from ray trace modeling ground motion in Utah," in of seismic refraction data," Evaluation of Regional and Urban Journal of Geophysical Research, Earthquake Hazards and Risk in vol. 87, pp. 2619-2633, 1982. Utah, U.S. Geological Survey [20] S. Singh, and R. B. Herrmann, Professional Paper (in press), "Regionalization of crustal coda Q 1987. in the continental U.S.," Journal of Geophysical Research, vol. 88, pp. 527-538, 1983. [21] L. W. Braile, R. B. Smith, J.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~328 PROBABILISTIC SEISMIC HAZARDS: GUIDELINES AND CONSTRAINTS IN EVALUATING RESULTS~~

~~R. K. Sadigh and M. S. Power Principal Engineers, Geomatrix Consultants, Inc. One Market Plaza Spear Street Tower, Suite 717 San Francisco, California 941 OS~~

~~ABSTRACT~~

In conducting probabilistic seismic hazard analyses, consideration of the dispersion as well as the upper bounds on ground motion is of great significance. In particular, the truncation of ground motion levels at some upper limit would have a major influence on the computed hazard at the low-to-very-low probability levels. Additionally, other deterministic guidelines and constraints should be considered in evaluating the probabilistic seismic hazard results.

In contrast to probabilistic seismic hazard evaluations, mean plus one standard deviation ground motions are typically used for deterministic estimates of ground motions from maximum events that may affect a structure. To be consistent with standard deterministic "maximum" estimates of ground motions values should be the highest level considered for the site. These "maximum" values should be associated with the largest possible event occurring at the site. Furthermore, the relationships between the ground motion level and probability of exceedance should reflect a transition from purely probabilistic assessments of ground motion at high probability levels where there are multiple chances for events to a deterministic upper bound ground motion at very low probability levels where there is very limited opportunity for "maximum" events to occur.

In Interplate Regions, where the seismic sources may be characterized by a high-to-very- high rate of activity, the deterministic bounds will be approached or exceeded by the computer probabilistic hazard values at annual probability of exceedance levels typically as high as 10r2 to 10"\ Thus, at these or lower values probability levels, probabilistically computed hazard values could be readily interpreted in the light of the deterministic constraints. In contrast, in Intraplate Regions, where the seismic sources may be characterized by a low-to-intermediate rate of activity, the deterministic constraints may not be approached until very low probability levels, e.g., l(r*or lower annually. Thus, deterministic constraints may not be sufficient in guiding interpretations for probability levels 10~J to 10"*. Therefore, it is essential that uncertainties in all the major factors that influence the computer exposure values be incorporated into the analysis.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~329 SITE-SPECIFIC RESPONSE SPECTRAL ATTENUATION RELATIONSHIPS: SIGNIFICANCE ON GROUND MOTION PREDICTIONS~~

~~Maurice S. Power and Khosrow Sadigh Geomatrix Consultants One Market Plaza Spear Street Tower, Suite 717 San Francisco, California 94105~~

~~ABSTRACT~~

This paper compares response spectral shapes obtained from analyses using appropriate response spectral attenuation relationships with generalized spectral shapes in common usage. The generalized spectral shapes incorporate the effect of subsurface conditions but do not explicitly recognize the effects of seismic-environment related variables. It is shown that the relative frequency content (i.e., spectral shapes) of both de- terministically and probabilistically obtained response spectra are strongly dependent on these variables, particularly on earthquake magnitude. Thus, in estimating site-specific response spectra, the use of ground motion relationships that are appropriate for the seismic environment and that incorporate the influence of parameters such as earthquake magnitude and distance, as well as the influence of subsurface conditions, is important.

INTRODUCTION of Newmark and Hall [3, 4] also incorporate site-dependent considerations The objective of this paper is to review through the use of different values of the desirability of using response spec- relationships between the peak ground tral attenuation relationships directly motion parameters acceleration, a, to specify design ground motions and velocity, v, and displacement, d (i.e., illustrate the limitations of using gen- relationships v/a and ad/v2) for dif- eralized spectral shapes anchored to ferent subsurface categories. This ap- peak ground acceleration. Commonly used proach is similar to that of Mohraz [2] generalized spectral shapes include but Mohraz defined more categories of those that are site-independent and subsurface conditions than the dual site-dependent. In this context, "site- categories of "soil" and "rock" of dependent" means that the spectral Newmark and Hall. shapes are dependent on the subsurface conditions. The best known example of The mean spectral shapes of Seed and a site-independent spectrum is the others [1] are illustrated in Figure 1. Reg. Guide 1.60 spectrum of the U.S. These spectral shapes were a primary Nuclear Regulatory Commission. Examples basis for the site-dependent spectral of spectra that are site-dependent shapes developed in the ATC 3-06 project include those developed by Seed and [5] and subsequently adopted in the others [1] and Mohraz [2]. The spectra building code recommendations of Che

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

330 Structural Engineers Association of predicted response spectrum normalized by California [6] and the Uniform Building the corresponding predicted value of peak Code [7]. The spectral shapes developed ground acceleration) are strongly by Seed and others [1], like those de- dependent on earthquake magnitude as well veloped by the other investigators cited as on subsurface conditions. The depen- above, are based on statistical analyses dence of the spectral shape on earthquake of groups of strong motion records. Ex- magnitude is illustrated in Figure 2, in amination of the data base used by Seed which spectral shapes for a "soil site" and others indicates that most of the condition are shown for earthquake moment records are from earthquakes of magni- magnitudes in the range of 5 to 8 using tude « 6Jj and intermediate source-to- the attenuation relationships of Sadigh site distances. et al. [11].

~~-I 1 r 1 1— -1 1 1 , 1— SPECTRA FOR S% DAMPING Daap>»f ratio - 0.05~~

~~Deep, stiff soils uj /-SOFT TO MEDIUM CLAY ft SAND 2.5 //-KIP COHESIONLESS SOIL ' ' -STIFF SITE CONDITIONS NOCK Iv v \\N 1.5 \ \\ 15 20 PERIOD-SECONDS N \~~

~~\ Figure 1. Site-dependent response \ \ spectral shapes (from [1]). N. .5 DETERMINISTIC ESTIMATES OF SITE-SPECIFIC SPECTRA~~

.5 1 1.5 ? In recent years, several attenuation re- PERIOD (seel lationships for response spectral values have been developed through regression analyses of the strong ground motion data base primarily from the western Figure 2. Effect of earthquake magnitude U.S. [8, 9, 10, 11, 12, 13, 14]. With on response spectral shape-based on these relationships, response spectral attenuation relationships by [11] predictions for given structural periods and damping ratios are made as a function of earthquake magnitude, distance, Attenuation relationships for response category of subsurface conditions, and spectra have also been developed for (with some relationships) other vari- tectonic and geologic environments other ables such as type of faulting, type of than the plate margin western U.S. envi- structure housing the accelerograph, ronment. Recently, attenuation relation- etc. Examination of these relationships ships have been developed for the subduc- indicates that the response spectral tion zone environment (such as pertains shapes for a given deterministic to the Puget Sound, Washington area) [15, response spectrum prediction (i.e., the 16]. Attenuation relationships have

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

331 also been recently developed for the motion analysis (i.e., hazard analysis), eastern U.S. (e.g., [17, 18]), Figure an equal hazard spectrum (i.e., a spec- 3 compares the spectral shapes for a trum having the same probability of rock site for a moment magnitude 6,5 exceedance at every period) can be earthquake occurring at a distance of 30 obtained. By normalizing the resulting km using the relationship of Sadigh and spectrum to the corresponding value of others [11] for the western U.S. and peak ground acceleration, the corre- those of Boore and Atkinson [17] and sponding spectral shape is obtained. McGuire and others [18] for the eastern U.S. Clearly, there are dramatic dif- Spectral shapes from a hazard analyses ferences in these spectral shapes, for soil sites in Salt Lake City, Utah reflecting differences in both earth- [19], are shown in Figure 4. For these quake source characteristics and the analyses, the attenuation relationships nature of the rocks (harder in the of Joyner and Boore [8] and Sadigh and eastern U.S.) between the eastern and others [11] for soil were used. The western parts of the country. figure shows the spectral shapes obtained using each attenuation relationship for

~1 1—I—TTTTT -| 1—I I I I I I probability of exceedance levels of 10% in 10 years (annual exceedance probability « 0.01), 10% in 50 years (annual 2.5 exceedance probability « 0.002), and 10% in 250 years (annual exceedance probability * 0.0004). For reference purposes, the ATC 3-06 spectrum for deep stiff soil 9 2 (Soil Type S2) is superimposed on the results. These results indicate a strong dependence of the spectral shapes on the probability of exceedance level, with the ,a shapes broadening and becoming relatively richer in long-period spectral content with decreasing annual exceedance probability. This spectral broadening reflects the increasing contribution of large magnitude events as the probability .5 - - Western US (11) V, • Eastern US (17) V level decreases. (Sites in Salt Lake Eastern US (18) City are located close to the Wasatch fault, which is capable of causing earthquakes as large as magnitude = lh-) .02 .05 .1 .2 .5 1 Period {sec) Figure 5 presents spectral shapes obtained from hazard analyses for two sites in the San Francisco Bay Area. These results were obtained using Sadigh and FiEure 3. Comparison of spectral shapes others [11] attenuation relationships for for a magnitude 6.5 earthquake using soil and are shown for a probability of western U.S. [11] and eastern U.S. [17, exceedance level of 10% in 50 years. 18] relationships Again, the ATC 3-06 spectral shape for soil is shown for comparative purposes. One site is located close to the Hayward PROBABILISTIC ESTIMATES OF fault, which has the potential for SITE-SPECIFIC SPECTRA earthquakes as lar^e as magnitude = 7. The other site is closer to the San Using response spectral attenuation Andreas fault, which can produce relationships in a probabilistic ground earthquakes up to magnitude * 8.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~332 10% in 10 Years 10% in 50 Years 10% in 250 Years (a) (c) J.b~~

~~2.S~~

~~'•5 / \ \~~

~~1 Joyiw and \ ^ * JoyiMr ond V Jayrar ond \\ ~~~

~~Hoar. (1982) \ \ BOOM (1982) \v • Boor. (1962) \\ .5 »TC \~~

~~0 3~~

~~• (e) • (') 2.S~~

1.3 \\ \ . \ \ '•/ if \\ \ 2 / \ \ V / ( \ \ '/ \\ 7 \ - Sadoh ond Sadiah and ' Sodioh and \ \ " othon V " othw* ()9«6) (1986) \ • t«in (I9K) MC V - ZZT. ' \ " \ 1 , . , . . ...t .1 i , , , ml . , .04 .1 3 £ 1 2 * .04 .1 5 .5 I 2 4 .04 .1 ,2 .5124 Ptriod (»c) Period [IK] Period (w)

~~Figure 4. Response spectral shapes resulting from probabilistic hazard analysis for sites underlain by deep stiff soils in Salt Lake City, Utah~~

Comparisons of these two results indi- than in the western U.S., due both to the cates that the spectra for the second dominance of lower-magnitude events to site are broader, reflecting the larger the hazard in many parts of the Eastern magnitude contributions from the San U.S., as well as differences in Andreas fault. applicable attenuation relationships for the two parts of the country as Figures 4 and 5 illustrate the impor- illustrated earlier. tance of the seismic environment and probability level on the relatively SUMMARY frequency content of equal hazard spectra. Obviously, for locations where the Site-dependent response spectral shapes hazard is dominated by lower-magnitude in common use account for the effect of events, resulting spectral shapes will subsurface conditions on spectral shape be narrower and less rich in longer- but they do not explicitly recognize the period motions. Hazard analysis results seismic-environment-related effects, for the eastern U.S. with which the which can be as important as the subsur- authors are familiar suggest that, in face conditions. Earthquake magnitude general, spectral shapes may be narrower has a very important effect on

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 333 REFERENCES Damping ratio > 0.05 10% probability ol txcttdanct in SO ytars [1] H.B. Seed, C. Ugai, and J. Lysmer,~~

~~—— Sit* b«twe*n San Andrias and "Site Dependent Spectra for Earth- Hayward fauns quake-Resistant Design," Bulletin of~~

- — - Sit* n*ar Hayward 1auH the Seismological Society of America, vol. 66, no. 1, pp. 221- ATC (Soil Type S2) 224, February 1976. [2] B. Mohraz, "A Study of Earthquake Response Spectra for Different Geo- logical Conditions," Bulletin of the Seismological Society of America, vol. 66, no. 3, pp. 915- 935, June 1976. [3] N.M. Newmark and W.J. Hall, "Development of Criteria for Seismic Review of Selected Nuclear Power Plants," Nuclear Regulatory Sadigh and others (1986) Attenuation Relationships Commission Report NUREG/CR-0098, May Used tor Probabilistic Analyses 1978. i , i . i . i 1 1.S 2 2.5 PERIOD (s*c) [4] N.M. Newmark and W.J. Hall, Earth- quake Spectra and Design. Berkeley, California: Earthquake Engineering Research Institute, Berkeley, Figure 5. Response spectral shapes California, 1982, 103 pp. resulting from probabilistic analyses for sites underlain by deep stiff soil [5] Applied Technology Council, Tenta- in San Francisco, California, Bay Area tive Provisions for the Development of Seismic Regulations for Buildings. ATC 3-06, 1978, 599 pp. shapes, and thus the appropriate spectral shape for a deterministic ground [6] Recommended Lateral Force Require- motion prediction depends on the magni- ments and Tentative Commentary. tude as well as the site conditions. Structural Engineers Association of Spectral shapes may differ significantly California, 1988, pp. . for different seismic environments, e.g., Eastern U.S. versus Western U.S. [7] Uniform Building Code. International In probabilistic ground motion predic- Conference of Building Officials, tions, because of magnitude and other 1988 Edition, 926 pp. effects, spectral shapes can be significantly different for different [8] W.B. Joyner and D.M. Boore, "Pre- probability levels. Thus, the use of diction of earthquake response response spectra attenuation relation- spectra," Proceedings of the 51st ships that are appropriate for the seis- Annual Convention of Structural mic environment and include the effect Engineers Association of California, of the seismic-environment-related also U.S. Geological Survey Open- parameters as well as the subsurface File Report 82-977, 1982, 16 pp. conditions is important in developing site-specific response spectra.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 334 [9] W.B. Joyner and T.E. Fumal, "Pre- [14] K. Campbell, "Empirical prediction diction Mapping of Earthquake of near-source ground motion for Ground Motion," in Evaluating the Diablo Canyon Power Plant site, Earthquake Hazards in the Los San Luis Obispo County, Califor- Angeles Region - An Earth Science nia," Deparment of the Interior, Perspective. U.S. Geological Survey U.S. Geological Survey Open File Professional Paper 1360, pp. 203- Report 89-484, Denver, Colorado, 220, 1985. 1989.

[10] K, Sadigh, "Considerations in the [15] R.R. Youngs, S.M. Day and J.L. development of site-specific Stevens , "Near-field Ground Motions spectra," in Proceedings of for Large Subduction Zone Earth- Conference XXII. Site-Specific quakes ," in Earthquake Engineering Effects of Soil and Rock on Ground and Soil Dynamics II - Recent Ad- Motion and the Implications for vances in Ground Motion Evaluation. Earthquake Resistant Design: U.S. Proceedings of the ASCE Specialty Geological Survey Open File Report Conference in Park City, Utah, 83-845, 1983. June 27-30, 1988, pp. 445-462.

[11] K. Sadigh, J. Egan and R, Youngs, [16] C.B. Crouse, Y.K. Vegas and B.A. "Specification of Ground Motion for Schell, "Ground Motions from Seismic Design of Long-Period Subduction Zone Earthquakes," Structure Labs," in Earthquake Bulletin of the Seismological Notes vol. 57, no. 13, 1986. Also Society of America, vol. 78, in paper by W.B. Joyner and D.M. pp. 1-25, 1988. Boore, "Measurement, Characteriza- tion, and Prediction of Strong [17] D.M. Boore and G.M. Atkinson, Ground Motion, in Earthquake "Stochastic Prediction of Ground Engineering and Soil Dynamics II - Motion and Spectral Response Recent Advances in Ground Motion Parameters at Hard-rock Sites in Evaluation. Proceedings of the ASCE Eastern North America, Bulletin of Specialty Conference in Park City, the Seismological Society of Utah, June 27-30, 1988, pp. 43-102. America, vol. 77, pp. 440-467, 1987. [12] K. Sadigh, C.-Y. Chang, F. Makdisi, and J.A. Egan, "Attenuation rela- [18] R.K. McGuire, G.R. Roro and W.J. tionships for horizontal peak Silva, "Engineering Model of ground acceleration and response Earthquake Ground Motion for spectral accelerations for rock Eastern North America," Technical sites," abstract, Seismological Report NP-6074. Electric Power Research Letters, vol. 60, no. 1 Research Institute, 1988. (January-March, 1989), p. 19; Seismological Society of America, [19] R.R. Youngs, F.H. Swan, M.S. Power, 1989 Annual Meeting, April, D.P. Schwartz and R.K. Green, Victoria, British Columbia, 1989. "Probabilistic Analysis of Earth- quake Ground Shaking Along the [13] I.M. Idriss, "Evaluating seismic Wasatch Fault, Utah," in Earthquake risk in engineering practice, Hazards and Risk in Utah. U.S. Proceedings of the Eleventh Geological Survey Open File Report International Conference on Soil 87- , vol. II, pp. , 1987; Mechanics and Founda t ion also U.S. Geological Survey Engineering. San Francisco, Professional Paper, in preparation. California, August 12-16, 1985, pp. 255-320. A.A. Balkema, Rotterdam.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~335 STRUCTURAL EVALUATION OF SAFETY CLASS COMPONENTS TO NATURAL PHENOMENA LOADINGS~~

~~T. J. Conrads Wcstinghouse Hanford Company P.O. Box 1970 Richland, Washington 99352~~

~~ABSTRACT~~

This paper addresses the efforts completed at the U.S. Department of Energy Hanford Site near Richland, Washingotn, to qualify structurally a number of existing safety class components in the Plutonium Finishing Plant complex. Design, fabrication, and installation of the facility occurred in the 1950s and 1960s and were based on the Uniform Building Code criteria for wind and earthquake loads. Recently the buildings were qualified to site- specific wind and seismic hazards. The methodology employed to qualify scismically the safety class components is discussed.

All gloveboxes, filterboxes, and connecting ductwork as well as other items which could pose an unacceptable seismic interaction were evaluated. The timely and cost-effective completion of the work presented a number of challenges peculiar to the detailed structural evaluation of existing equipment in older-operating-secured facilities. Some of the obstacles which had to be overcome were the following:

• Number of structural analysts having the proper security clearance • Availability of as-built component fabrication drawings • Availability of general arrangement drawings • Limited access to facility and specific area of interest • Consistent structural evaluation methods and criteria • Short schedule.

Some methods for surmounting these hurdles may be of benefit to others responsible for similar tasks. For example, since access could not be guaranteed, the walkdown teams complemented sketches and marked-up drawings with photographs and video tapes so that the need to return for additional information was minimized. The use of a desk-top procedure depicting a uniform analysis methodology and acceptance criteria precluded excessive revisions of analysis packages.

~~The seismic qualification of existing equipment in older facilities is a challenge, but some preliminary planning can make it manageable.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~336 TRANSMISSION OF LOW-MAGNITUDE SEISMIC EXCITATION INTO HANFORD SITE STRUCTURES~~

~~E.O. Weiner Westinghouse Hanford Company P.O. Box 1970 Richland WA 99352~~

~~ABSTRACT~~

Several Hanford Site buildings were analyzed using simplified models to gain insight as to what extent the free field motion of a small- magnitude earthquake is transmitted into building structures as a result of soil-structure interaction effects. Building selection included the Plutonium Processing Plant, B-Plant and the Fast Flux Test Facility containment which represented a variety of stiffnesses, weights, and embedments. An artificial time history for the free field has a peak response at 13 Hz. This motion represents a median for magnitude 4 and 4.5 earthquakes, respectively. Floor response spectra were compared with results from analyses to design basis ground motions using the same structural models. Considerable attenuation of the small-magnitude free-field motion was found in the case of stiff, deeply embedded structures. This attenuation is attributed to kinematic interaction in addition to attenuation with depth in the free field. Even with such attenuation, there are exceptions where small magnitude responses exceed design basis responses. They are generally associated with 10 to 20 Hz modes with vertical motion.

INTRODUCTION Plutonium Processing Plant, B-Plant and the Fast Generally, magnitude 4 to 5 earthquakes Flux Test Facility (FFTF) containment. These are believed to cause little, if any, damage to buildings represent a variety of stiffnesses, competent structures. In regions near the weights, and embedments. epicenter, such small-magnitude events typically have substantial maximum accelerations, high Earthquake Motions frequency content, durations less than 2.5 s, and A small-magnitude earthquake ground few cycles of excitation. The effect on active motion is defined in terms of response spectra equipment having high frequency response for horizontal motion. Based on a median components such as electrical relays may be of spectral shape for M-4 motions [1], the peak concern. Responses to the free field motion of a amplification is 3.3 occurring at 12 to 15 Hz. small-magnitude earthquake may be higher at The maximum acceleration is 0.20g in the 200 relatively high frequencies than a design basis, Area and 0.15g for the 400 Area [2]. Artificial free field spectrum. Thus, relay chatter or time histories were generated with 2-s durations. equipment operability is not immediately Figure 1 shows the 200 Area response spectra. precluded by analysis performed to the design Vertical motion is 60% of the horizontal motion. basis requirements. The purpose of this effort is The design basis earthquake (DBE) spectrum, to analyze a few typical buildings to gain insight also shown in Figure 1, consists of a Newmark- as to what extent the free field motion is Hall median spectrum with a 0.20g ZPA. The transmitted into building structures. Soil- 400 Area FFTF design basis motion shown in structure interaction effects are the basis of the Figure 2 has a 0,25g ZPA, and the vertical analysis. Building selection includes the component is two-thirds of the horizontal. All

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

337 analyses used surface control motions, i.e., the 234-5Z spectra of Figures 1 and 2 are to be found at the The 234-5Z or Plutonium Finishing Plant surface in the free field solutions, (PFP) is a lightly framed structure on grade. The flexible NS character of the building is Buildings Descriptions captured with a simple stick model. Baseplate Emphasis is placed on buildings in the and clip flexibilities are modeled with short 200 Area, Only the FFTF containment is in the flexible beam elements fixed to the column at 400 Area. Building models are approximate, one end and restrained against rotation at the emphasizing known fixed-base fundamental other. Similar considerations were used for the frequencies and mass distributions. The intent EW direction, but a shear wall provides a stiff was not to duplicate any existing analyses or use connection from the ground to the second floor. the results for qualifying equipment. Instead, Finally, the vertical model consists of a column small-magnitude and design basis solutions were with attached oscillators representing floor panel sought for similar models in order to compare frequencies of 4, 5 and 6 Hz chosen as typical floor responses at all frequencies. Structural of those found in various buildings. The column damping values were generally at 7% for design is supported on a footing with a soil spring basis motions and at 4% for the small-magnitude providing a 15 Hz column mode. A 7% earthquake motions. Exceptions are discussed damping value was used for the baseplates and below. The FLUSH computer program [3] was clips in the small-magnitude case. used in all analyses. AZ Tank FFTF Containment The AZ tank is a reinforced, concrete The FFTF containment is a stiff structure with a double-steel liner buried from structure with an 80-ft embedment which is -7 to -54 ft. Excitation is expected to be expected to attenuate the free field motion by governed by the free field motion. The tank is kinematic interaction with the nearly rigid assumed full, but the low sloshing mode foundation. The basement, steel containment warrants inclusion of only the impulsive mass structure, crane support and reactor are modeled which is about half the total in the horizontal with quad elements. direction. Vertical excitation uses all the fluid Iteration for strain compatible soil mass. Provision was made for vertical motion of properties with the small-magnitude motion the overburden which is resisted by arch action produces very large bedrock accelerations which of the roof and circumferential expansion of the were associated with strains at about the 150 ft haunch area. A vertical load path was provided depth. Thus, bedrock was set at the 140 ft so that the vertical overburden load could be depth leaving 60 ft of flexible soil between the transmitted down the walls to the basemat at the basemat and bedrock. The design basis solutions outside perimeter. used a 200-ft flexible soil column. Any Light Structure B-Plant Simple oscillators set to the free field B-Plant is a narrow, heavy building with motions in the 200 or 400 Areas demonstrate moderate embedment of 22 ft. Some attenuation response possibilities to the different motions. with depth is expected. Basement excitation should be less than the free field, and upper Soil Properties ievels should appear somewhat isolated. The area Soil stiffnesses for the 200 Area were below the deck level was modeled with quads taken as the upper end of the data for recent using shear area rules. Concrete beams were used alluvial sands down to 100 ft. Iteration for in the roof structure of the North-South (NS) strain compatible soil properties was carried out and vertical models. Symmetry was assumed in for the horizontal free field motion, and the all cases. The East-West (EW) direction is so resulting profile was used for both horizontal long that only a one-eighth endpiece of the and vertical building models without further building was modeled using constraints to iteration. In all cases, a surface control motion prevent rotation. was prescribed.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

338 SUMMARY OF RESULTS Maximum Accelerations Table I summarizes the maximum or Embedment Effects zero period accelerations (ZPA) of resulting The general character of the small motions at the basemat and surface levels in the magnitude and design basis solutions for free field and buildings. With horizontal embedded structures is illustrated by the FFTF excitation, there is more attenuation with depth response spectra shown in Figures 3a and 3b. In in the small-magnitude motion. In the FFTF the horizontal direction, the basemat responses to case, there is a considerable drop from the free the small-magnitude excitation are much more field to the building at the basemat level. These attenuated than the design basis solutions. This effects are less pronounced with vertical is attributed to kinematic interaction of the rigid excitation. foundation and a small excitation wavelength. Vertical excitation has a similar attenuation Exceptions of Greater Small-MagnitudeResponse effect, but not nearly as strong. By comparing small-magnitude and Of particular note is the comparison of design basis floor spectra, a list of exceptions, responses of a component with its base in both where the small-magnitude response exceeds the the small-magnitude and design basis motions. design basis response at some frequency, was For example, the FFTF crane support structure generated. There are 21 exceptions out of the shows a 0.12g response at its base (the operating 48 locations and directions for the four buildings deck) in the small-magnitude case and 0.20g in studied. All occur at frequencies greater than the design basis case. Yet, the relationship is 5 Hz, and the significant ones involve vertical reversed at the top of the support structure - motion at frequencies greater than 10 Hz. 0.84g versus 0.7g - as seen in Figures 4a and 4b. From the appearance of these exceptions, It appears that the operating deck has more high it is concluded that analysis to a design basis frequency content in the small-magnitude case motion does not ensure equipment qualification in spite of showing a lower response, and the in the presence of a small-magnitude motion. 9 Hz crane support structure resting on the This does not necessarily mean that there are operating deck responds more to the small- equipment qualification problems in a building magnitude motion. Thus, conclusions about the with equipment analyzed or tested to a design response of a component cannot be drawn basis motion. In many cases, Category I directly from responses at its base. equipment requiring operability during a seismic

~~Table 1. Summary of Maximum Accelerations (g).~~

~~Small-Magnitude Design Basis Free Field Building Free Field Building Surf Bsmat Surf Bsmat Surf Bsmat Surf Bsm Building Embed Dir ZPA ZPA ZPA ZPA ZPA ZPA ZPA ZVA~~

~~FFTF 80 ft HRZL .153 .066 .051 .033 .252 .150 .201 .155 VERT .089 .037 .047 .039 .167 .153 .164 .161~~

~~234 0 ft NS .201 .201 .202 .202 .197 .197 .199 .199 EW .201 .201 .204 .204 .197 .197 .212 .212 VERT .115 .115 .104 .104 .120 .120 .112 .112~~

~~B 22 ft NS .201 .087 .071 .068 .194 .142 .151 .143 EW .201 .087 .076 .072 .194 .142 .156 .155 VERT .116 .084 .072 .069 .118 .112 .109 .105~~

~~AZ 56 ft HRZL .197 .083 .112 .092 .194 .148 .165 .139 VERT .115 .094 .079 .070 .118 .114 .117 .089~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~339 Response Damping • .05 Reeponte Damping * .05~~

Smal MaanHud* 1 JSiSMlf. OJ Time Hlttefy, DBE r _D«»lanBofj«__ 3 0,4 1 \ o.t \\\ .8 ft •\ 0.3 \ ** 1 \ 0.4 j \ N f / smd IMognitude • 1 f 0.2 s o.i f1 .Igra?!?. y Design Bosis . 0.0 — 0.3 1 0.3 wo Frequency (Hz) Frequency (Hz)

~~Figure 1. 200 Area Horizontd Free Field Response Spectra Figure 2. 400 Area Horizontal Free Field Response Spectra~~

~~Response Damping = .05, Structural = .04 Response Damping * .05, Structural«.04~~

~~Free field raM.syj>fisrt. OJ- I •RX -ri. sz! IPS'- LT - i Bottom or rwoctor < |ft it >\~~

~~0.4 JVfA V'•*> r OJ- 7 o.v s F" -T-- .—*• \ 0.0 -* 10 100 Frequency (Hz) Frequency (Hz)~~

~~Figure 3a FFTF Horizontal Small Magnitude Responses. Figure 3b. FFTF Horizontal Small Magnitude Responses.~~

~~Response Damping = .05, Structural = .03/-O5 Response Damping = .05, Structural = .03/.05~~

~~Free field Basemat, Q 468 ft ..QlHM.^iHEfiX! Si^jf550J™ JS& S/L Cwjjuj'j;L>gj Deck. Vertical Bottom of Reactor~~

~~o.i Frequency (Hz) Frequency (Hz)~~

~~Hgure 4a FFTF Horizontal Design Basis Responses. Figure 4b. FFTF Horizontal Design Bosis Responses.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

340 event has been qualified to rather high response Typical fundamental building frequencies accelerations. Required response spectra for at the Hanford Site are 3 to 6 Hz involving equipment testing sometimes are found up to 10 horizontal motion. Compared to a design basis to 20g in the 2 to 8 Hz range, especially if motion, it has been shown that a smaller fraction motion at the equipment base is determined by of the small-magnitude motion studied herein cascaded analyses. The corresponding test will get into an embedded structure like the response spectra, which must envelope the FFTF containment. Therefore, it is considered required response spectra peaks, usually do not unlikely that the 12 Hz small-magnitude motion drop off much at higher frequencies. Finally, will have a greater damaging effect in the response spectra associated with the above horizontal direction than the DBE motion having exceptions are typically narrow- banded and peak responses in the 2 to 8 Hz range. unidirectional. It is suggested [4] that such Turning to the vertical direction, spectra be reduced by a 0.5 factor before frequencies of modes with vertical motion are comparing with equipment ruggedness spectra generally much higher than those involving based on broadband and multiaxial testing. horizontal motion. The engineering characterisation study of [6] is based on Peak Response Envelopes ductility, and damage tolerance is expressed in Figures 5 and 6 show the power spectral terms of demand-capacity ratios. The highest densities (PSD) based on the free field motions natural frequency appearing in [6] is 8.S Hz, and used in this study. Strong motion durations (Td) the demand-capacity ratios for this frequency do are typically three-quarters of the total not appear to decrease with decreasing durations. There is more small-magnitude energy magnitude as they do for lower natural in the 10 to 20 Hz range compared to the design frequencies. One might then argue that small- basis motions. Response of simple oscillators of magnitude earthquakes are important in the various natural frequencies to the small vertical direction. With the possible exception magnitude motion at Ig ZPA is obtained, and of floor panels having lower frequencies, the envelope of the response peaks is shown in building structures are generally quite strong in Figure 7 along with the corresponding 200 Area this direction, thanks to dead and live load design basis results at the appropriate maximum design criteria together with the associated load accelerations. Similar results for the 400 Area factors. Thus, it is considered unlikely that are in Figure 8. These figures look more like small-magnitude motions will result in damage the power spectral density comparisons than the to typical site structures. response spectra comparisons. The peak response envelopes could be used to estimate peak floor REFERENCES responses given modal parameters of a building with insignificant soil-structure interaction. [1] "WPPSS Nuclear Project #2 Final Safety Based on the general nature of the results for Analysis Report," Washington Public Power embedded structures, it is expected that the Supply System, 1986. comparison of small-magnitude and design basis peak response envelopes will give the most [2] A.M. Tallman, "Natural Phenomena pessimistic exceptions of small-magnitude Hazards: Hanford Site, Washington," WHC- response exceeding design basis response. SA-0606A, Westinghouse Hanford Company, Richland, Washington, May Structural Integrity 1989, presented at the Second DOE No specific analytical effort was made Natural Phenomena Hazards Mitigation herein to investigate structural integrity with Conference, Knoxville, Tennessee, October small-magnitude earthquake motions. Instead, 3, 1989. the intent was to provide an idea of how small- magnitude earthquake motions can be [3] J. Lysmer, T. Udaka, C.F. Tsai, and H.B. transmitted into typical Site buildings. Seed, "FLUSH - A Computer Program for However, discussions are provided, e.g., [5], Approximate 3-D Analysis of Soil- which support a rational engineering basis for Structure Interaction Problems," Report neglecting earthquakes with magnitudes below EERI 75-30, Earthquake Engineering S.O for nuclear power plants. Research Center, University of California,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~341 0.00V 0.001' SmaM Maanltuda, Id * 1.51 \ \ Design Basis, Td = 12.6 • 1 1 •A i 111 n I i PJ • 1 t k "iBIllll 1 1 w ; 1 i A I inHBIir M l Q : 1 f J > a. ; 0.000001: I Hi~~

~~/ 0.000000V illl 0.3 1 » too Frequency (Hz) Frequency (Hz) Figure S. 200 Area Horizontal Power Spectral Densities. Figure 6. 400 Area Horizontal Power Spectral Densities.~~

~~Response Damping = .05 Response Damping — .05~~

~~A DBS 020a ZPA. Damolna = .07 A DBE 0.25a ZPA, Damping > .07 a Small Mqgnitud* 0.20a ZPA, Damping. = .04 a SM 0.15a. ZR*, Darnplng = .04~~

~~/ \\ \ s / \ \ y \ \ \ \ \ I <* s~~

~~100 100 Frequency (Hz) Frequency (Hz)~~

~~Figure 7. 200 Area Oscillator Horizontal Response Envelopes. Figure 8. 400 Area OsciNaior Horizontal Response Envelopes.~~

Richmond, California, November 1975. [6] R.P. Kennedy, "Engineering Characterization of Small-Magnitude [4] K.L. Merz and P. Ibanez, "Guidelines for Earthquakes," in Preliminary Workshop Estimation of Cabinet Dynamic Proceedings - Engineering Amplification," presented at the Second Characterization of Small-Magnitude Symposium on Currrent Issues Related to Earthquakes, sponsored by EPRI Nuclear Nuclear Power Plant Structures, Equipment Power Division, p. 2-70, January 1987. and Piping sponsored by EPRI, Orlando, Florida, December 1988.

[5] M.W. McCann, Jr., and J.W. Reed, "Lower Bound Earthquake Magnitude for Probabilistic Seismic Hazard Evaluation," EPRI-NP-6437D, presented at the Second Symposium on Current Issues Related to Nuclear Power Plant Structures, Equipment and Piping, sponsored by EPRI Seismicity Owners Group, Orlando, Florida, December 1988.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~342 ADVANCES IN SEISMIC CRITERIA TO QUALIFY STRUCTURES, SYSTEMS AND COMPONENTS IN OPERATING REACTORS~~

~~Miguel A. Manrique Walter R.Bak Impell Corporation San Ramon, CA 94583~~

ABSTRACT This paper describes improved seismic evaluation criteria and analysis methodologies used as part of the seismic reevaluation of San Onofre Nuclear Generating Station, Unit 1. The plant had originally been designed for 0.25g ground acceleration and was required to be upgraded to a 0.67g ground acceleration as part of the plant's Long Term Service Seismic Reevaluation Program. The application of the criteria and methods described in this paper to demonstrate the seismic capability of the plant resulted in efficient plant modifications with considerable cost savings to the plant owner. The NRC accepted these criteria and methods based on favorable results of reviews, audits and independent verification of the theories, bases and implementation procedures of the proposed criteria and analysis methods.

INTRODUCTION From the outset, the plant owner recognized that the indiscriminate and uniform application of "stan- This paper presents key aspects of the design dard" existing seismic criteria and analysis methods criteria and analysis methods used to demonstrate resulted in an unacceptable number of plant up- the seismic capability of an older vintage nuclear grades and associated expenditures. Instead, power plant. Specifically, the criteria and analysis supplementing existing criteria and methods with methods described in this paper were developed for more advanced techniques that would provide a the requalification of structures, systems and more realistic assessment of the inherent seismic components at the San Onofre Nuclear Generating resistance of the existing structures, systems and Station, Unit 1 (SONGS-1). components was necessary. A description of these criteria and analysis methods is the subject of this SONGS-1 was originally designed to a 0.25g paper. ground acceleration using equivalent static analyses and the codes and standards in effect at the time of The overall philosophy used for seismic requalifi- its construction (1964 to 1967). In the late 1970's cation of SONGS-1 was that proposed by Dr. N.M. SONGS-1 was selected for seismic reevaluation as Newmark in NUREG/CR-0098 (Reference 1), which part of the Nuclear Regulatory Commission's (NRC) in page 2 states: Systematic Evaluation Program (SEP). The SEP reevaluation included reverification of the plant's "It is well known that upgrading and retrofitting systems and components necessary for hot safe constitute expensive operations when they can be shutdown under a 0.67g peak ground acceleration. accomplished at all. In many cases it is economi- Such a design criterion represents a significant cally, if not physically, impossible to carry out increase over the original design basis, is one the significant seismic upgrading improvements. In highest in the USA, and is especially severe for a those cases where it is possible economically it is plant that has been operating for over 20 years. desirable to take advantage of the latest concepts pertaining to development of seismic resistance.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 343 Thus in the evaluation of existing facility, and in the New floor response spectra were generated using subsequent detailed design studies for physical current, state-of-the-art methods of analysis for soil- upgrading of structural and mechanical systems, the structure interaction analysis (SSI). The new floor authors believe it is possible (and desirable) to take response spectra were generated using the computer into account the modest amount of nonlinear behav- codes SASSI and CLASSI. ior that can be permitted in many portions of such systems without significant decrease in the margin A desirable feature of the procedure for SSI of safety against safe shutdown or containment." evaluation was to be able to directly use existing structural model information. Models had already It was further observed that "the inherent seismic been generated and reviewed as part of the initial resistance of well designed and constructed systems SEP evaluations. Dynamic model properties, in is usually much greater than that commonly as- terms of fixed-base mode shapes, frequencies, mass sumed, largely because nonlinear behavior is mobi- participation factors and modal damping ratios had lized to limit the imposed forces and accompanying already being calculated, and it was highly desirable deformations. For such systems where the resistance to use this information already available. Addition- is nondegrading for reasonable deformations, the ally, because the models were relatively large (of the requirements for retrofitting may be nonexistent or at order of thousands of degrees of freedom) direct use most minimal.'' of the existing model information resulted in significant cost savings associated with model recoding, Based on the philosophy described above, current reanalysis rechecking, verification, review and methods of analysis coupled with more realistic licensing. seismic criteria were developed and implemented to demonstrate functionality of the plant's systems and Thus, an approach was selected in which the components. The implementation of the above SASSI code was used to generate the frequency- approaches to this older vintage reactor led to a dependent, complex valued impedance functions significant decrease in hardware modifications, as representing the site/foundation system; and the compared with those resulting from the application CLASSI code was used to perform the response of "standard" criteria, without a significant decrease calculations using the SASSI-genera ted impedance in the plant's safety margins. functions and the already available fixed-base dynamic properties of the superstructure. Key features of the SONGS-1 criteria and analysis methodologies consist of the following: Use of The existing Reactor Building model consisted of: state-of-the-art soil-structure interaction (SSI) (1) a 3-dimensional representation of the Reactor analysis methods to evaluate seismic response; use Containment Building which includes the primary of direct generation of response spectra technique shield wall, secondary shield wall, refueling canal for computation of floor response spectra; use of a walls, operating deck and steam generator compart- strain-based criterion to qualify piping systems; use ments; (2) the Containment Sphere which is a ball- of ductility limits to qualify structural steel; use of shaped, partially embedded steel structure; and (3) ASME Code Case N-411 damping values with both, the Enclosure Building, which is a surface-founded the Envelope Response Spectrum analysis method reinforced concrete structure surrounding and and the Multiple Level Response Spectrum (MLRS) enclosing the Steel Sphere structure and Reactor analysis method; and use of improved methods for Containment Building. modal combination of modal responses, among others. Each of these criteria and methods of The foundation of the Reactor Containment analysis are described in detail in References 2 and Building is deeply embedded and is of a hemispheri- 3 and are summarized in Section 3.0 of this paper. cal shape. In addition, it is in close proximity to the Enclosure Building foundation which is a surface- ANALYSIS METHODS AND EVALUATION founded ring-shaped footing surrounding the CRITERIA Reactor Containment Building foundation. Founda- tion-to- foundation interaction would influence the Generation of Floor Response Spectra Using in-structure response of the structures and, there- the SASSI and CLASSI Computer Programs fore, impedances were developed for the coupled two-foundation system. The SASSI program is

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

344 capable of generating impedance functions for significant reduction of unnecessary plant upgrades multiple foundations, each of arbitrary geometry with a concomitant cost savings and increased plant and embedment configuration, The frequency- reliability. dependent foundation impedances were explicitly calculated using a 3-dimensional finite element Qualification of Piping Systems Using Strain- model of the Reactor Building and Enclosure Build- Based Criteria ing foundations. The qualification criteria for piping systems were The procedure used to generate the impedance based upon the requirements set forth in Section III, functions is to first generate the compliance (flexi- Subsection NC of the 1980 ASME B&PV Code bility) coefficients by applying harmonic loads of (Reference [4]) with stress allowables taken from unit amplitude at frequencies covering the fre- the SEP criteria, as previously applied to SONGS-1. quency range 0.5 to 20.0 Hz. and tracking the corresponding displacements. The unit harmonic As an alternative to the above criteria, piping loads are applied at the centroid of each foundation systems were qualified by inelastic strain criteria (hemispherical Reactor Containment Building and using a stress-strain correlation. This qualification ring-shaped Enclosure Building foundations) and required that the piping systems strain associated correspond to each global degree of freedom (i.e. x, y, with an elastic-calculated primary stress be deter- z, xx, yy, and zz directions). To reduce model size, a mined and limited, as defined below: one-quarter model of the coupled foundation system is used. A separate analysis is performed for each £| £ 1 percent for carbon steel type of load application using the appropriate £l £ 2 percent for stainless steel boundary conditions. Displacements resulting from the application of the unit loads are the flexibility where &t = maximum piping membrane-plus- coefficients necessary to form each term of the bending strain (amplitude) due to Code Level D compliance matrix. The impedance matrix is then loadings. obtained by direct inversion of the 12x12 compliance matrix. The impedance functions are frequency- The basis for these strain limits are, in part, dependent and complex-valued. The real part of the Code Case N-47 (Reference [5]) and component impedance term represents the stiffness and the testing programs (References [6] and [7]). A de- imaginary part represents the damping of the site/ tailed discussion of these basis can be found in foundation system. References [2] and [9].

The SASSI-generated impedance functions are To calculate the piping strains up to the specifed then used by CLASSI together with the dynamic limits, a stress-strain correlation methodology is properties of the superstructures to obtain struc- used. This methodology is based on the fatigue tural time history responses from which in-structure evaluation procedure of the ASME Code (Reference response spectra was generated. [4]) and is further verified by comparison with Greenstreet elbow test results (Reference [8]). The The applied methodology including the theory, selected approach allows the use of standard pro- implementation aspects and limitations of the duction type linear elastic analysis techniques to approach were extensively reviewed by the NRC convert the elastically calculated stresses to strains. staff and consultants. The review took the form of audits of the program's documentation, formulation The conversion is as follows: of test problems to compare SASSI/CLASSI solutions with other solutions, and independent verifica- For Carbon Steel Et = Ks ^ tion of plant specific results. Based on the favorable 2 Q CTc results from this extensive review, the NRC ac- For Stainless Steel£t = Ks ' cepted the approach and the results obtained. where The applied approach resulted in significant E| = pipe strain as previously defined reductions in floor response spectra, as compared CTe = elastic-calculated stress for with the soil-spriri approach and resulted in pressure, gravity and DBE

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

345 loadings, as defined in the Code (1) In calculating the intensified primary stress Class 2 and 3 Equation (9) - Level ao, at least 50% of oo is due to earthquake D, using stress intensification loading. factor approach, psi (2) In calculating moments due to earthquake loading, a response spectrum method is used, with damping not exceeding that specified in Code Case N-411 [22]. (All terms are defined in the Code [4] and the product of 0.75 i (3) Diameter/wall thickness ratio (DQ/t) does not shall never be taken as less than exceed 50. 1.0) E = Young's modulus, at operating (4) Weldments as well as piping base materials temperature, psi are ductile. (No quenched and tempered Ks = Strain correlation factor. ferritic steels or cold worked austenitic stainless steels.) The strain correlation factor Kg is defined as follows: (!?) Joints are butt welded or girth fillet welded. =1.0 when 3.4 ge < 1.0 (Bolted-flanged joints are qualified per the requirement of NC-3658 of the ASME Code [4].

= 1.0 (6) The cumulative usage factor due to a Modified Housner Design Spectrum event does not when 1.0 < 3.41°-y (7) A clearance check for pipe displacement was performed for large bore piping qualified by whenm<3.4°£ the strain criteria. S where (8) A boundary load capacity check (pipe supports, mechanical equipment, penetrations, S = Piping material yield strength at maximum valves, etc) was performed for large bore operating temperature, psi stainless steel piping qualified by the strain n = Strain hardening exponent criteria and with strain exceeding 1 percent. m = Code-defined parameter to produce correct correlation. For stainless steel piping, two additional checks were performed if the calculated strain was in the The material parameters n and m used on range of 1 to 2 percent(both checks 1 and 2 needed SONGS-1 piping are defined in Table NB-3228.3(b)- to be satisfied): 1 of the ASME Code [4] and are summarized below: (1) Compressive Wrinkling Check: To avoid com- Material m 11 pressive wrinkling failure, the calculated strain for stainless steel piping was limited to Stainless Steel 1.7 0.3 2% strain and 0.2t/R. This latter check is Carbon Steel 3.0 0.2 recommended in NUREG 1061, Volume 2 Referencel 101) as a simple and conservative means of preventing compressive wrinkling Application Limits failure in straight pipe. Thus, Piping was qualified by the strain criteria above, X subject to the following limitations: where

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

346 input time history's response spectrum must enve- t = nominal wall thickness of pipe, in lope the design ground response spectrum over an R = mean radius of pipe, in appropriate range of damping ratios. This over- conservatism is typically of the order of 20% to 30% (2) Low Cycle Fatigue Check: To avoid low-cycle and is eliminated by use of direct response spectra fatigue, the elastically-calculated stress generation technique. The FLORA (Reference |4|) O.75SM/Z due to SSE loading was limited by a computer program methodology was applied for fatigue check based on Markl's correlations on certain SONGS-1 structures to remove this excess moment-loading fatigue tests (Reference I 111). conservatism. FLORA uses the "smooth" design The allowable usage factor (Ua) due to seismic ground response spectrum and directly computes loading was limited to 0.25. amplified floor response spectra. Existing dynamic properties (mode shapes, frequencies, mass matrix, d mass participation factors and modal damping N ratios) calculated as part of previous analyses were where used as input for the direct response spectra genera- n = Number of significant cycles tion computations. The FLORA methodology was N = Number of allowable cycles accepted by the NRC. The acceptance of this Ua = Allowable usage factor method was based on a detailed review of the theory, and the application procedures. FLORA was N was calculated as follows: benchmarked against time history methods as part 91.875 of the acceptance process. N = 0.75i M Mode Combination by Complete Quadratic where ZJ Combination (CQC) Rule M = Resultant elastically calculated moment due to the SSE. The Complete Quadratic Combination (CQC) rule for mode combination was used in the analysis of The adoption of strain criteria is a significant certain SONGS-1 systems. The CQC method is an departure from current practice. It is, however, accurate method for modal response combination, consistent with observed behavior of piping under particularly for systems such as piping which seismic loading. Recent studies and tests (Refer- typically contain closely spaced, coupled modes. The ences [12] and [13]) indicate that the failure mode of CQC method is implemented in computer program piping is strain-controlled and the loading to failure, SUPERPIPE (Reference [16 j) for piping analysis defined as excessive flow area reduction, is many applications and EDSGAP (Reference [18]) for times greater than the Code allowable values. The general linear elastic structural analysis applica- allowable stresses defined in current criteria are tions. The method has been validated by time based on limit load theory which limits piping history methods, and is found to give more accurate stresses to levels that preserve the elastic load- responses than the Regulatory Guide 1.92 (Refer- deflection characteristics of the piping components. ence [19]) methods, particularly when combining Instead, the strain criteria allow a limited excursion closely spaced, laterally-coupled modes. The CQC beyond the limit load levels. The strain limits of 1 method reduces to the SRSS method for well spaced percent for carbon steel and 2 percent for stainless modes. steel are set such as to prevent: (1) the onset cf plastic tensile instability, (2) low cycle fatigue or The CQC method requires that all modal re- plastic ratcheting, (3) local buckling, (4) excessive sponse terms be combined as: deformation, and (5) function failure of pipe mounted equipment. R =

Direct Generation of Floor Response Spectra where R = Combined Response Py = Correlation coefficient between Floor response spectra generated using time modes i and j (based on Ran- history approach typically contains excess and dom Vibration Principles) unaccounted conservatism due to the fact that the

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

347 Rj, Rj = Modal responses for modes i accepted by the NRC for qualification of structural andj (including directional steel, sign for each mode) CONCLUSIONS The cross-modal correlation coefficient (Pij) is a function of the duration and frequency content of In this paper, we have presented advanced the loading, modal frequencies, and damping ratios seismic criteria and methods of analysis used for the of the piping system. The CQC formulation can be seismic reevaluation of SONGS-1 - an older operat- found in Reference [15], ing nuclear power plant whose seismic design basis was significantly increased as part of seismic The coefficients P^ vary between zero (for uncor- reevaluation programs. By addressing both the related modes- modes with very different frequen- seismic capacity and seismic demand aspects, the cies) and one (for modes which respond in-phase). evaluation criteria and methods presented permitted improved quantification of plant margins and Multiple Level Response Spectra provided for consistent and more realistic levels of conservatism. The NRC approved the criteria and The multiple level response spectra (MLRS) methods based on confirmatory analysis and exten- method allows the appropriate floor response sive detailed reviews performed to assure that spectra to be input at the piping system's support adequate margins of safety exist in the proposed attachment locations. Because each support loca- approaches. The application of the criteria and tion may have different input, the MLRS method method resulted in effective and efficient seismic allows for a better representation of the actual input improvements to the plant, and in considerable to the piping support system. In the MLRS method savings to the plant owner. the overall response of the piping system is calculated considering the dynamic (inertial) and the REFERENCES pseudostatic (seismic anchor movement) components of the motion. 1. Newmark, N.M., Hall, W.J., "Development of Criteria for Seismic Review of Selected Nuclear When applying the MLRS method, the SRSS Power Plants", NUREG/CR-0098. Prepared for method of response combination was used for the U.S. Nuclear Regulatory Commission, May combination of different level responses provided 1978. that the correlation coefficient of the input is between +0.16 i.e., when the inputs are uncorrelated. 2. Safety Evaluation Report by the Office of Directional responses were combined by SRSS. Nuclear Reactor Regulation. San Onofre Nu- Modal responses were combined in accordance with clear Generating Station, Unit 1, Long Term Regulatory Guide 1.92 rules. Seismic Reevaluation Program. July 1986.

As an alternative to the above procedures, mode 3. Safety Evaluation Report by the Office of and level combination could be combined using cor- Nuclear Reactor Regulation. San Onofre Nu- relation coefficients derived using random vibration clear Generating Station, Unit 1. Long Term methods. This approach has been implemented in Seismic Reevaluation Program, September RV-SUPERPIPE. The RV-SUPERPIPE methodol- 1985. ogy (References [17] and [20]) gives results which are consistent with time history analysis results 4. ASME Boiler & Pressure Vessel Code, Section III, 1980 Edition with Addenda through Winter Use of Ductility Limits for Structural Steel 1980. Qualification 5. ASME Boiler & Pressure Vessel Code, Code In cases where the elastically calculated tension Case N-47-21, "Class 1 Components in Elevated and bending stresses of structural steel members Temperature Service, Section III, Division I", (exclusing columns) exceed the stress allowables approved December 11, 1981. under the Level D loading, ductility calculations were performed. Ductility ratios of up to 3 were 6. Imazu, et al., "Plastic Instability Test of Elbows

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

348 Under In-Plane and Qut-of-Plane Bending", Paper E6/5, Sixth SMIRT Conference, Paris, 14. FLORA: A Computer Program for the Direct France, August, 1981, Generation of Amplified Response Spectra", Impell Corporation u 7. Teidoguchi, HM Experimental Study on Limit Design for Nuclear Power Plant Facilities 15. Wilson, E.L., Der Kiureghian, A., Bayo, E.P.," A During Earthquakes", Japanese Report 50-1705 Replacement for the SRSS Method in Seismic issued to US NRC, February, 1975. Analysis", Earthquake Engineering and Struc- tural Dybamics, Volume 9,1981. 8. Greenstreet, W.L., "Experimental Study of Plastic Response of Pipe Elbows", ORNL/ 16. SUPERPIPE: A Computer Porgram for Analysis NUREG24, February, 1978. of Piping Systems, Impell Corporation.

9. Stawniczy, G., Bak, W.R., J*au, G., "Piping 17. RV-SUPERPIPE: A Random Vibration Based Stress-Strain Correlation for Seismic Loading". Computer Program for the Analysis of Piping Transactions of the ASME Journal of Pressure Systems. Impell Corporation. Vessel Technology, Vol. 110, November, 1988. 18. EDSGAP: A General Structural Analysis 10. NUREG 1061, Volume 2, "Report of the U.S. Program, Impell Corporation. Nuclear Regulatory Commission Piping Review Committee, Evaluation of Seismic Designs—A 19. Regulatory Guide 1.92, "Combining Modal Review of Seismic Design Requirements for Responses and Spatial Components in Seismic Nuclear Power Piping", April, 1985, pp. 444-450. Response Analysis", U.S. Nuclear Regulatory Commission, February 1976. 11. Markl, A.R.C., "Fatigue Tests of Piping Compo- nents", Transactions ASME, Vol. 74,1952, pp. 20. Asfura, A.," A New Combination Rule for 287-303. Seismic Analysis of Piping Systems", Proceed- ings of the 1985 Pressure Vessel and Piping 12. NUREG/CR-3893, "Laboratory Studies: Dy- Conference", New Orleans, Vol. 98-3, June 1985. namic Response of Prototypical Piping Sys- tems", Prepared by ANCO Engineers, Inc. for 21. Russel, M.J., Shieh, L.C., Tsai, N.C., Cheng, the U.S. Nuclear Regulatory Commission and T.M., "Lessons Learned from the Seismic the Electric Power Research Institute, August, Reevaluation of San Onofre Nuclear Generating 1984. Station, Unit 1. Paper Presented at the SMIRT Conference, 1987. 13. EPRI Report No. NP-3746, "Dynamic Response of Pressurized Z-Bend Piping Systems Tested 22. ASME Boiler & Pressure Vessel Code, Code Beyond Elastic Limits and with Support Fail- Case N-411, "Alternate Damping Values for ures", Prepared by ANCO Engineers, Inc. for Seismic Analysis of Class 1, 2 and 3 Piping", EPRI, December, 1984. Section [1], Division 1, Approved February 1986

~~Second DOE Naiural Phenomena Hazards Mitigation Conference - 1989 Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 350 Session 11 Probabilistic Risk Assessment~~

~~.••• •~~

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

~~351 OVERVIEW OF SEISMIC PROBABILISTIC RISK ASSESSMENT FOR STRUCTURAL ANALYSIS IN NUCLEAR FACILITIES~~

~~John W. Reed Jack R. Benjamin and Associates, Inc. 444 Castro Street, Suite 501 Mountain View, CA 94041~~

ABSTRACT Probabilistic Risk Assessment (PRA) for seismic events is currently being performed for nuclear and DOE facilities. The background on seismic PRA is presented along with a basic description of the method. The seismic PRA technique is applicable to other critical facilities besides nuclear plants. The different approaches for obtaining structure fragility curves are discussed and their applications to structures and equipment, in general, are addressed. It is concluded that seismic PRA is a useful technique for conducting probability analysis for a wide range of classes of structures and equipment.

INTRODUCTION Mile Island, PRA has become increasingly popular Probabilistic Risk Assessment (PRA) for as a tool to evaluate in probabilistic terms the seismic events is routinely performed for nuclear potential threat from nuclear power for a variety of power plants and DOE facilities. Although the initiating events. A number of PRAs have been practice of PRA has developed procedures and recently published which include consideration of conventions which are particular to the nuclear seismic events (3, 10, 15). Based on these power plant industry, the seismic PRA studies, the results of the analyses indicate that methodology is applicable to other structure and earthquakes may, after all, be a significant equipment environments. To the casual observer contributor, particularly for older plants. The state- PRA has the appearance of a new probabilistic- of-the-art has been advancing, and the analysis based procedure different from the other structural procedures for seismic PRA for nuclear power reliability approaches which have been evolving plants are becoming accepted in the nuclear power over the recent years. In reality, PRA does not plant industry. Other industries and critical contain any new mathematical concepts, but rather facilities may benefit as well from the formalism represents a "repackaging" of probabilistic provided by the seismic PRA method. techniques designed to solve a specific engineering problem (i.e., what is the risk of operating a This paper summarizes the background of the nuclear power plant in the presence of potential seismic PRA method as currently being used in the seismic events?). nuclear power plant industry. A description of the seismic PRA method for developing structure and The Reactor Safety Study (RSS) was one of equipment fragility curves is presented. The the first attempts using PRA for nuclear power background and description provides a basis for plants and included a small effort at evaluating the extending the PRA procedures to other critical effects of seismic events (20). Based on the RSS it facilities. The potential for applying seismic PRA was believed that earthquakes were not significant to other environments and opportunities for contributors to risk when compared to other expanding this technique are discussed. internal event initiators. Since the accident at Three

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

352 BACKGROUND ON SEISMIC PRA In addition to seismic hazard curves, The basic pans which are needed for a information on the capacity of structures and seismic PRA represent the same information which equipment is needed in a seismic PRA. The must be used for any complete seismic probabilistic deterministic perspective that the nuclear power study. First, the ground motion hazard in the plant complies with the criterion safe shutdown seismic PRA takes the form of hazard curves earthquake (SSE) peak ground acceleration value which portray the annual frequency of exceedance does not lead immediately to a realistic capacity for as a function of the ground shaking parameter use in a seismic PRA. For example, for the Zion (e.g., peak ground acceleration or spectral plant, which has an SSE of 0.17g, the frequency acceleration). Figure 1 shows an example set of of occurrence from Figure 1 is in the range of hazard curves. Note that more than one curve is approximately 10"4 to 103 events per year. This given since the underlying mechanisms which frequency, if it were the frequency at which failure cause earthquakes are uncertain. Subjective will occur, would not be an acceptable safe level of weights (i.e., probabilities) are assigned to the earthquake hazard. curves which by convention must sum to unity. Although each single curve is developed using The objective in the seismic PRA for nuclear standard seismic hazard calculational procedures power plants is to systematically account for the which involve earthquake sources, frequency of conservatisms which exist due to the design occurrence of different size events, maximum process by scaling the SSE capacity to a more magnitudes, and ground motion attenuation, the realistic failure value. The final result of this individual underlying parameters are not known process is a family of fragility curves such as with certainty. This leads to postulating sets of indicated in Figure 2 for three components. Again, hazard parameters which are assigned a single curve is not shown but rather a family of corresponding probability weights. Because of a curves is indicated by the dashed lines which lack of knowledge and limited amount of data (i.e., reflect the uncertainty in quantifying the various the effects of earthquakes have been recorded for factors of safety built into the design. The word only a few hundred years in the U.S.) there does "fragility" has been popularized in seismic PRA not exist the intellectual luxury of basing and is simply the inverse of capacity. Thus, the uncertainty on only the statistical variability due to fragility curve(s) gives the frequency (or more limited data. The Bayesian viewpoint must be correctly, the fraction) of failure between 0 and 1 adopted to allow the assignment of probabilities as a function of the ground motion parameter. based on subjective judgement. However, in contrast to some classical statisticians who prefer to It is important to recognize that the results of allow onl; recorded data, this procedure is the many structural reliability approaches can be perfectly acceptable to most engineers. expressed in the same final fragility curve format as shown in Figure 2. The approach used in seismic It is important to note in Figure 1, the PRA for nuclear power plants is different from increasing spread of the hazard curves with higher traditional reliability approaches which start from accelerations. In other words, the frequency of scratch and perform the analysis without regard to occurrences of high accelerations is very uncertain. the original design process. However, the results This is significant to the structure and equipment from the different reliability approaches can be capacity analysis since it is the range of higher expressed in the fragility curve format. This is acceleration which generally dominate the risk in discussed in more detail below. seismic PRAs for nuclear power plants. Because of the layers of conservatisms built into nuclear In addition to hazard and structure/equipment power plant design, the capacities of most fragility curves, the third component in the seismic components are relatively high. Hence, due to the PRA process is the logic model which relates the large uncertainties in the ground shaking hazard it failures of the different structures and equipment to is unproductive to refine the structure and a radiological consequence (e.g., early fatalities, equipment capacity calculations to accuracies which injuries, or latent cancers). In an analogy to are inconsistent with the hazard uncertainty. traditional reliability methods, the higher order

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~353 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 Peak Acceleration (g) Figure 1 Example seismic hazard curves.~~

~~0 OfFSITETOWCR CERAMIC INSULATORS~~

~~1 1 1 1 1 3.0 3.0 4.0 10 •« 10 tf~~

~~O SfRVKEWAMRniMfS~~

~~Figure 2 Example seismic fragility (3). Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

354 event in nuclear power plant analysis is like a stales. If the events are independent, the structure failure in a structural reliability analysis. frequencies of each event are simply multiplied together (note that since failure frequencies are The logic model takes the form of event trees small, success frequencies are essentially equal to and fault trees. Event trees display the success or unity). However, particularly for seismic initiating failure of various safety systems leading to the events, the event frequencies are not generally higher order event. On the other hand, fault trees independent due to the common cause effects of can answer the question of how can a particular earthquakes. system fail. Thus for each of the systems in an event tree there is a corresponding fault tree which Figure 4 shows an example of a partial fault relates the various structure and equipment failures tree for loss (i.e., failure) of electric power, which in a logical manner. is the top event. An example of an "or" and "and" gate is shown. An "or" gate defines the upper An example event tree is shown in Figure 3 event failure as failure of one or more of the lower where a large pipe break is the initiating event and events immediately below the gate. In contrast, an radioactive release is the failure event. The events "and" gate defines the upper event failure as failure across the top are ordered in time and the upper of all of the events immediately below the gate. branches represent success and the lower branches The branches of a fault tree proceed downward in represent failure. Thus, the top-most branch is increasing refinement until the most basic failure success (i.e., everything works in the event of a events are reached. pipe break) while the lower branches are all failure

~~C Pioe Electric Fisjion Containment ECCS Break Power Product Integrity Removal~~

~~PEi~~

~~?E2~~

~~init.'at.Tsg Event ?A PQ1~~

~~P02~~

~~Figure 3 Example event tree (20).~~

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 355 Through the logic of the event and fault trees, i the fragilities of the basic events (usually individual ItSfll structures and equipment) are related by a Boolean equation for failure event under consideration (e.g., core melt or radioactive material release). Thus, the frequencies of failure of the individual structures and equipment are combined to produce a higher order event frequency of failure such as Lon o< AC al DC *ovm to ESFi to esfi core melt, for example. This is directly analogous to failure of a building which is logically related to T the failures of the individual parts, which may be either failure modes in series (i.e., determinate case) or parallel (i.e., indeterminate case). Figure 5 shows the final core melt fragility curves for the

lw of LOB 01 Zion PRA which relates the frequencies of failure n ESFi to €5Fi of the following components through the Boolean expression which is given as Equation 1. Figure 4 Example fault tree (20).

~~0.40 DSO OH Ground Motion Parameter Figure 5 Example core melt fragility curve family (3).~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~356 list of components:~~

Service water pumps Auxiliary building—failure of concrete shear wall Refueling water storage tank Interconnecting piping/soil failure beneath reactor building Condensate storage tank Crib house collapse of pump enclosure roof 125 VDC batteries and racks Service water system buried © pipe 48" (§) CST piping 20" Collapse of pressurizer © enclosure roof

Core melt = where: C\ = "and" Boolean algebra symbol (intersection joint occurrence) KJ = "and" Boolean algebra symbol (union-occurrence of either or both) Once the hazard curves and fragility curves for a failure event are obtained such as core melt, dH then the two sets of curves are combined two at a ~fa = derivative of the hazard curve with time (i.e., one hazard curve and one fragility curve) respect to the ground motion to obtain the probability distribution on the unconditional frequency of failure, Pf, where Pf is Using as examples the curves from Figures 1 and obtained as follows: 5, each of the five fragility curves is integrated with the nine hazard curves (total of 45 integrations). The probability of each of the 45 resulting frequency values (obtained using Equation 2) is f~J n* da found by multiplying the probabilities associated with the two curves used in each integration (note that the sum of the 45 new probability values are where: equal to unity). If the 45 frequency values are plotted as a histogram and converted to a Pfla conditional core melt frequency (i.e., continuous representation, the result is a density fragility curve — see Figure 5, for function such as shown in Figure 6. example)

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

357 The entire process of seismic risk analysis for could be used to obtain various values of frequency nuclear power plants is shown in Figure 7, which of failure (for different Ag values) which plot as a relates the seismic hazard analysis, component- fragility curve as shown in Figure 8a. fragility evaluation, plant system and accident- sequence analysis (i.e., the event and fault trees), The problem is that A is not known with and finally the consequence analysis. The certainty. (It is assumed that the lognormal model interrelationship of the various parts are shown in and pr value are known in a relatively certain Figure 7. sense). Thus, a second lognormal distribution for A is used to quantify the uncertainty for this DESCRIPTION OF THE FRAGILITY parameter. This second distribution is determined METHOD by two parameters: the median value, A, and the The approach which has been used in all of logarithmic standard deviation for uncertainty in the the commercial nuclear power plant seismic PRAs median value, p The probability density function performed to date for developing fragility curves is u called the Zion method which is characterized by for A is shown in Figure 8b. two principal features. First, the methodology is based on a double lognormal distribution model. Now, depending on what value of A is Both the distribution on the median and the random picked from the distribution on A as shown in variation of frequency of failure are assumed to be Figure 8b, a corresponding fragility curve can be lognormal. Second, the probabilistic analyses use calculated. For example, if the 95 percent the results from the original design analysis as the probability fragility curve is derived, the A would initial basis for the seismic fragility. The median be selected such that there is a 0.95 probability that fragility values are obtained using the dynamic a larger median value would occur. responses and capacities from the design analysis which are scaled to eliminate conservatisms. For example if A is 0.77g and pu equals Variabilities, such as randomness and uncertainty, 0.39, then for the 0.95 probability level, A equals are estimated based on some data, but mostly on 0.4g. This value comes from the following engineering judgement. equation, which is the mathematical representation of the solution shown in Figure 8b. When using the Zion method it is assumed that the capacity of a structure, in terms of ground = Aexp[puO-J(l-p)] (3) acceleration, is lognormally distributed. Thus, the frequency of failure is a function of these where: parameters: = Standard cumulative normal distribution • the median capacity value, A and *•• is the inverse function • the logarithmic standard deviation for p = Probability value (e.g., 0.95) capacity, pr, and

« the ground motion input acceleration value, Now, if the fragility frequency of failure Ag- value, assuming p, is 0.36, is desired corresponding to a ground acceleration Ag equal to Note that any randomness in the ground 0.2g, the answer can be found from the lognormal motion or building response is included in the pr distribution with median value to 0.4g (see Figure value. Figure 8 shows the capacity frequency 8a; pr equals 0.36). The answer is 0.027 and is density function, which is determined by A and pr. found from the following equation: If the ground motion value is Ag, then failure occurs for all values of A less than Ag. Thus, the frequency of failure is just the area under the In density function between A equal to 0 and Ag. The f(Ag) = analysis could stop at this point and this procedure (4)

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~358 MfANfBtOUtNCY* 0«~~

~~0.4~~

~~0-2~~

~~to* ANNUAL C01E MELT FDCQUCNCY Figure 6 Example probability functions for core melt.~~

~~Haiard analysis exceeoenc e Frequenc y o f~~

~~Hazard intensity~~

i Weather data Release r 1 category Atmospheric I? dispersion M . 1 1 5 —•>• ex c 2 Evacuation Fre o a. 1 Event trees Health edicts Fault trees Frequency Properly damage Damage Containment analysis

~~Plant iyslttm and fU'lease Consequence Risk analysis~~

~~failur e / t* O 3 u> E / // O a o US Senmic rnotiun parameter~~

~~Coiuporibr.1 fraQilMy/ vulnerability evaluation Figure 7 Nuclear power plant seismic PRA flowchart.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~359 F(A)~~

~~1.0~~

~~(corrasoonds to frequency of faiiurs)~~

~~Density Function Fraaiiitv Curve~~

~~(a) Capacity Frequency Density Function and Fragility Curve~~

~~arsa is probability p~~

~~A (corrssponds to probability p)~~

~~(b) Probability Density Function for X~~

~~Figure 8. Probability and frequency functions for fragility analysis.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

360 Thus the Zion method requires three fatality risk, etc.). In this context values are parameters to characterize the fragility of each estimated for the significant range of contributing structure: vthe median ground level acceleration earthquake accelerations. capacity, A, and the logarithmic standard deviations for randomness and uncertainty, p and The assumption that the various response and r capacity factors are independent is made primarily pu respectively. The procedure for determining these parameters for components can be found in for analytical convenience. In reality, many of the parameters are related. For example, soil structure Reference 2 and in past safety studies such as interaction, structure damping, and inelastic energy performed for Zion and Indian Point (3,10). absorption are not in general independent. For critical structures bench-marking non-linear time In summary, the fragility analysis starts with history analysis may be required where the effects the design capacity usually corresponding the SSE of the various response and capacity parameters are design (e.g., 0.20g). Then the factors of safety are varied simultaneously. An approach for including systematically factored-out for the following dependency is given in Reference 12. response and capacity factors in a fragility analysis for structures (the fragility analysis for equipment APPLICATION OF PRA TO OTHER is similar but additional factors are required): FACILITIES The concepts embodied in seismic PRA for • Spectral shape nuclear power plants can be applied to other types • Soil-structure interaction of engineered facilities. The basic parts of a seismic PRA are hazard curves, component • Damping fragility curves, and a logic model which relates • Frequency component failure to an overall failure state. These parts also are present in probabilistic analyses of • Mode shape non-nuclear facilities. • Modal combination The development of hazard curves is a well • Combination of earthquake established discipline and can be generally component performed independent of the risk analysis in • Strength which they are used. This is not completely correct since the hazard parameter (e.g., peak ground • Inelastic Energy Absorption acceleration) must be compatible with the parameter used in the fragility analysis. However, the Because of the assumed underlying lognormal hazard/fragility interrelationship can be just as model, median factors of safety are simply easily considered to be part of the fragility analysis multiplied together and the result used to scale the as the hazard analysis. design ground motion capacity. Logarithmic standard deviations are obtained for each factor The structure and equipment fragilities can based on a first-order Taylor series expansion of come from a variety of sources including the the relationships between response or capacity and following which are discussed below: the underlying physical variables. It is generally assumed that the factors are independent and the • Factor of safety analysis starting with design corresponding logarithmic standard deviations are capacity combined by the square root of the sum of the squares (SRSS) procedure. • Reliability analysis starting with the structural information (i.e., drawings, material The assumptions of the lognormal model and properties, past mathematical models, etc.) independence between factors may not be • Design code requirements appropriate in all cases. Only portions of the fragility curves contribute significantly to the • Test data (both component specific and results of the safety analysis (e.g., core melt, early generic) • Engineering judgement (expert opinion)

~~Second DOE .Natural Phenomena Hazards Mitigation Conference - 1989~~

361 piping runs in a nuclear power plant make it • Data from past seismic events virtually impossible to develop fragility curves for • Combinations of sources of information every single segment of pipe that exists. Thence, generic pipe fragility curves are developed which are tailored to the seismic input and design code The factor of safety analysis approach, which requirements used for the plant. By understanding starts with the design capacity, is one of the the governing code requirements used to design a methods used in seismic PRA for nuclear power component, an estimate of its median failure plants. This approach is equally applicable to capacity and the corresponding variability can be conventional facilities where a clear understanding made. In addition to code rules, information of the of the design steps are known. By learning what failure capacities in relation to the elastic limit also was assumed at each design step, realistic median must be used. This information is usually obtained analysis procedures and values can be substituted, from either tests or nonlinear failure analyses (or compared to the values assumed in the original both). Fragility curves developed in this manner analysis, and safety factors developed accordingly. are most useful when the capacities are found to be In addition, the variabilities (i.e., randomness and uncertainty) can be calculated based on an assumed relatively high and are not significant contributors underlying model such as lognormal, using a to the calculated risk. If a component fragility Taylor series expansion of the underlying based on design code requirements is a dominant equations where only the first order terms are contributor, than a more component-specific retained. fragility analysis may be required. Test data has been a valuable source of The reliability analysis approach is limited information used in the development of fragility only by the constraints of the engineers curves. Both generic tests and tests performed for imagination. Many of the reliability procedures specific components have been used. The former which have been developed can be used to develop type of tests lead to generic fragility curves, while fragility curves. Multiple failure modes which are the latter type are used for facility specific partially dependent along with various equipment fragility curves. In many cases, combinations of loads which may occur particularly in the nuclear industry, qualification simultaneously can be included. An example of rather thin fragility tests are performed. For the this approach is the detailed seismic PRA former type, failure of the equipment normally methodology developed in the Seismic Safety does not occur, hence, the results represent at best Margins Research Program where multiple linear a lower bound on the fragility level. In some time history analyses are performed in a Latin cases, using qualification data even with some Hypercube simulation experiment (16). A second upward adjustment leads to a conservative example is a probabilistic procedure which representation of capacity. This is appropriate, if combines random vibration input with finite the results do not significantly contribute to the element models to obtain fragility curves for a calculated risk. If it does contribute, then the nuclear power plant containment (7). For this resulting probability distribution on frequency of example a limit state which is defined as either failure will be conservatively shifted. In PRA, the concrete crushing or reinforcing steel yielding was analysis should be performed in an unbiased used. Many other reliability-based procedures also manner. In contrast, decisions made using the could be modified to express their results in the results of PRA studies should be conservative. form of fragility curves. Similar to the factor of More times than not, the decision maker is not the safety analysis approach, reliability analysis is same person who performs the PRA. In directly suited for developing fragility curves for transferring information to the decision maker, individual components. conservatism should not be built into the results. Otherwise, the decision maker is put in an Using design code requirements to obtain awkward position of making a conservative fragility curves is primarily applicable for classes decision using information with unknown of components rather than for individual structures conservatisms already built in. or equipment. An example in the nuclear industry is piping fragility. The many piping systems and

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 362 Reference 4 contains generic fragility values analysis performed using the seismic PRA for 37 types of equipment. The values given are approach. Other industrial facilities have been based on generic fragility tests and judgments from analyzed in this manner with the use of fragility experts which have been combined. Judgments of curve input (8). Important industrial plants with experts based on their engineering experience is an critical functions that depend on complex equally valid basis for developing fragility curves. interrelationships of structural, mechanical, and Systematic procedures have been suggested for electrical components can be modeled with event soliciting expert opinion and combining this trees/fault tree logic. At the other extreme, information with other sources (4). This approach individual structures consisting of dependent also has been used in developing fragility data for failure states also can be fit within the seismic PRA the effects of tornado winds (6). format. The use of seismic PRA has promoted the engineering viewpoint to keep the fragility analyses Data from past earthquakes is a valuable as simple as possible within the limited constraints source of fragility information particularly for of the available data. Rather than expending large various classes of structures in the Southern resources on detailed analyses which outstrip the California area after the San Fernando earthquake quality of the data, simple PRA analyses are in 1971 (21). Other sources of data of this type performed and the results are tested using include References 1, 5,13, 17, 18, and 19. sensitivity analyses which consider reasonable During the time of the underground nuclear ranges of values. This approach is done in nuclear explosion program conducted in Colorado to power plant seismic PRA analyses and also is unlock natural gas sources, data from damage to useful to other critical facilities. conventional structures was collected and used to calibrate theoretically-derived fragility models (11 REFERENCES and 14). One difficulty in working with damage [11 Algcrmisscn, S. T., K. V, Stcinbruggc, and H. J. data is often only information on the damaged Lagorio, "Estimation of Earthquake Losses for buildings is available. The number and Buildings (Except Single Family Dwellings)," USGS characteristics of the undamaged buildings is Open-File Report 78-1441,1978. equally important to the development of fragility curves. [2] American Nuclear Society, "PRA Procedures Guide, A Guide to the Performance of Probabilistic Risk Assessments of Nuclear Power Plants," NUREG/ Damage data from past earthquakes and the CR-2300, Vols. 1 and 2, prepared for the U.S. Nuclear corresponding fragility curves for buildings, Regulatory Commission, January 1983. bridges, dams, electrical facilities, underground utilities and other lifeline systems, etc., are [3] Commonwealth Edison Company, Zion Nuclear Plant particularly useful in assessing probability of Units 1 and 2 Probabilistic Safety Study. Dockets different levels of damage in a region which could 50-295 (Unit 1) and 50-304 (Unit 2), September 8, occur after a major earthquake. Results of these 1981, available at NRC Public Document Room. types of studies are useful for effective and efficient emergency planning. For example, [4] Cover, L. E., et al., "Handbook of Nuclear Power Plant fragility curves have been developed for 16 types Seismic Fragilities," Prepared for U.S. Nuclear of structures common to cities of the Mississippi Regulatory Commission, NUREG/CR-3558, December Valley Region (9). 1983 (draft). [5] Earthquake Engineering Research Institute, Finally, combinations of the various sources "Reconnaissance Report, Imperial County, California, of information discussed above can be made to Earthquake, October 15,1979," D. J. Leeds, Editor, obtain fragility curves. Several examples have 1980. already been given above. Either ad hoc procedures based on engineering judgement or [6] Hart, G. C, "Estimation of Structural Damage Due more formal procedures using Bayesian techniques to Tornados," Proceedings of the Symposium on can be used. Tornados — Assessment of Knowledge and Implications for Man. Texas Tech University, Nuclear power plants are not the only June 1976. facilities that can benefit from a probabilistic

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

363 17] Hwang, H., et al,, "Reliability Assessment of 117] Steinbrugge, K. V, and D. F. Moran, "An Engineering Reinforced Concrete Containment Structures," Study of the Southern California Earthquake of July 21, Prepared for U.S. Nuclear Regulatory Commission, 1952 and Its Aflcrshocks," Bull. Seism. Soc. Am.. 44, NUREG/CR-3227, February 1983, 1954.

[8] Jack R. Benjamin and Associates, Inc., "PUREX Plant [18] URS/Blumc and Associates, "Seismic Damage Equipment Fragility Analysis," JBA 117-0505-HF-O1, Assessment for High-Rise Buildings," URS/J AB 8020 Prepared for URS/John A. Blume & Associates, report prepared for U.S. Geological Survey, 1982. Engineers, San Francisco, California, August 1981. [19] U.S. Department of Commerce, "San Fernando, [9] Kircher, C. A., and M. W. McCann, Jr., "Development California, Earthquake of February 9,1971," L. M. of Seismic Fragility Curves for Sixteen Types of Murphy, Scientific Coordinator, Vol. MI, 1973. Structures Common to Cities of the Mississippi Valley Region," Jack R. Benjamin and Associates, Inc., JBA [20] U.S. Nuclear Regulatory Commission, "Reactor Safety 186-020, Prepared for Allen and Goshall, Inc., Study: An Assessment of Accident Risks in U.S. Memphis, Tennessee, December 1984. Commercial Nuclear Power Plants," NUREG-75/014, October 1975. [ 10] Power Authority of the Slate of New York and Consolidated Edison Company of New York, Inc., [21] Whitman, R. S., S. T. Hong, and J. W. Reed, Indian Point Probabilistic Safety Study. Units 2 and 3, "Damage Statistics for High-Rise Buildings in the Dockets 50-247 (Unit 2) and 50-286 (Unit 3), March 5, Vicinity of the San Fernando Earthquake," Technical 1982, available at NRC Public Document Room. Report R73-24, Department of Civil Engineering, MIT, 1973. [11] Reed, J. W, and J. A. Blume, "Predicting Damage Probabilistically for Buildings Subjected to Ground Motion," American Society of Civil Engineers Specialty Conference on Probabilistic Methods in Engineering, Stanford, California, 1974.

[12] Reed, J. W., et al., "Analytical Techniques for Performing Probabilistic Seismic Risk Assessment of Nuclear Power Plants," Presented at the 4th International Conference on Structural Safety and Reliability, Kobe, Japan, May 1985.

~~[13] Ross, G. A., H. G. Seed and R. R. Migliaccio, "Bridge Foundation Behavior in Alaska Earthquake," ASCE. Journal of the Soil Mechanics and Foundations Division. 95, No. SM4, 1969.~~

[14] Scholl, R. E., Editor, "Effects Prediction Guidelines for Structures Subjected to Ground Motion," URS/John A. Blume & Associates, Engineers, JAB-99/115, Prepared for U.S. Energy Research and Development Administration, July 1975.

~~[15] Science Applications, Inc., "Probabilistic Risk Assessment, Limerick Generating Station," Prepared for Philadelphia Electric Company, 1981.~~

~~[16] Smith, P. D., "Seismic Safety Margins Research Program — Phase I Final Report," NUREG/CR-2015, 10 Volumes, September, 1981.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~364 PROBABILISTIC EVALUATION OF MAIN COOLANT PIPE BREAK INDIRECTLY INDUCED BY EARTHQUAKES SAVANNAH RIVER PROJECT L & P REACTORS~~

~~Stephen A. Short and Donald A. Wesley Impell Corporation Mission Viejo, California~~

~~Nabil G. Awadalla Westinghouse Savannah River Company Aiken, South Carolina~~

~~Robert P. Kennedy RPK Structural Mechanics Consulting Yorba Linda, California~~

~~ABSTRACT~~

A probabilistic evaluation of seismically-induced indirect pipe break for the Savannah River Project (SRP) L- and P-Reactor main coolant (process water) piping has been conducted. Seismically- induced indirect pipe break can result primarily from: 1) failure of the anchorage of one or more of the components to which the pipe is anchored; or 2) failure of the pipe due to collapse of the structure. The potential for both types of seismically-induced indirect failures was identified during a seismic walkdown of the main coolant piping. This work involved: 1) identifying components or structures whose failure could result in pipe failure; 2) developing seismic capacities or "fragilities" of these components; 3) combining component fragilities to develop plant damage state fragilities; and 4) convolving the plant seismic fragilities with a probabilistic seismic hazard estimate for the site in order to obtain estimates of seismic risk in terms of annual probability of seismic-induced indirect pipe break.

INTRODUCTION FRAGILITY METHODOLOGY A probabilistic evaluation of pipe break for the The seismic risk of indirect reactor main Savannah River Project (SRP) L- and P-Reactor coolant (RCL) pipe break involves consideration main coolant (process water) piping has been of the seismic capacities (fragilities) of the conducted. This paper covers the portion of this important components and structures. These effort including the evaluation of seismically- components were identified based on discus- induced indirect pipe break. L- and P-Reactors sions with the SRP staff, review of L- and are very similar structures with nearly identical P-Reactor drawings, and plant walkdowns equipment which might affect main coolant loop conducted by Drs. R. P. Kennedy and D. A. piping. The P-Reactor building is much weaker Wesley. The main tank, main coolant (Bingham) than the L-Reactor building in the local region pumps, and the heat exchangers are all of suffi- which governs building fragility such that overall cient mass that failure was conservatively con- seismic risk is somewhat higher for P-Reactor. sidered to result in failure of the pipe. The location The L-Reactor evaluation is primarily described of these components as well as the general plant herein with results for both L- and P-Reactors configuration is shown in Figure 1. presented at the end of the paper.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~365 • An-}~~

Figure 1 - SRP Reactor Building Configuration Failure of the building structure above Elevation Because of this uncertainty, great precision in 0 as well as failure of the forest support and attempting to define the shape of these curves is control rod system was investigated due to the unwarranted. Thus, a procedure which requires potential for impact-induced pipe failure of the a minimum amount of information, incorporates exposed nozzles located above the operating uncertainty into the fragility curves, and easily floor slab at Elevation 0. Failure of the building enables the use of engineering judgment was structure below Elevation 0 was not considered used. credible at the seismic levels which govern the The entire fragility curve for any mode of failure indirect pipe break plant risk. an J its uncertainty can be expressed in terms of The seismic fragilities and resulting plant risk the median ground acceleration capacity, A, and developed in this investigation are based on logarithmic standard deviations of the inherent breach of the fluid boundary. The approach to randomness (failure fraction) about the median developing the conditional probability of RCL and the uncertainty (probability) in the median pipe break is based on combining the failure value, fijp and fty, respectively. Inherent ran- probabilities of individual structures and com- domness is associated primarily with the earth- ponents. Component fragilities are treated as independent which gives a conservative bound quake characteristics themselves, and on the combined probability of seismic indirect uncertainty is associated with other lack of pipe break. Fragility curves must be developed knowledge. In general, it is not considered primarily from analysis combined heavily with possible to significantly reduce randomness by engineering judgment supported by very limited additional analysis or test based on current test data. Suchfragility curves will contain a great state-of-the-art techniques. Uncertainty, on the deal of uncertainty, which must be recognized. other hand, is considered to result primarily from analytical modeling assumptions and other lack

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

366 of knowledge concerning variables such as reduction of seismic input with depth of embed- material strength, damping, etc., which could in ment of the structure, and a factor to account for many cases be reduced by additional study or the effect of soil-structure interaction. test. Similarly, the equipment response factor, When developing the median ground accel- FREE, includes a spectral shape factor (including eration, it is computationally convenient to work the effects of peak floor response spectra with a median factor of safety, F, such that: broadening and smoothing, and artificial time history generation), an equipment damping fac- 0) tor, a modeling factor, a factor to account for is tne eak where ASSE P ground acceleration conservatism in combining modal responses, associated with the Safe Shutdown Earthquake and a factor to account for conservatism in (SSE) or Design Basis Earthquake (DBE). The combining earthquake components. overall median factor of safety, F, may in turn be The approach to developing the civil structure broken down (separation of variables) into indi- fragility is essentially the same as for equipment vidual factors of safety representing the important fragilities except that no equipment response variables contributing to the seismic capacity and factor, FRE, is included. response variables associated with each structure or component. COMPONENT FRAGILITIES For equipment, the factor of safety can be Structure Response modeled as the product of the three random The principal variables contributing to the structure response factor of safety, FRS, include variables. the considerations of the shape of the response RS RE (2) spectra, structure and soil damping, modeling considerations, embedment effects, soil- The capacity factor, Fc. for the equipment is a structure interaction (SSI) effects, and earth- product of a strength factor, Fs, and an inelastic quake directional component combinations. The energy absorption factor, F „. The strength factor, basis for establishing the structural response Fs. represents the ratio of ultimate strength to the ln factor of safety and variabilities as well as the stress calculated for ASSE- calculating the structure seismic loads is the Quadrex soil- value of Fs, the non-seismic portion of the total structure interaction model and analysis results load acting on the support is subtracted from the described in Ref. [5]. strength. The inelastic energy absorption factor, Ground Response Spectra f u, is a measure of the strength of an element The soil-structure interaction analysis was failing in a ductile mode beyond the yield strength based on recommended ground response or plastic hinge stress. spectra developed by J. A. Blume (Ref. [6]) for The structural response factor, FRS, recog- the SRP site. Although not specifically stated, nizes that in the design analyses, the structural these spectra appear to be approximately response was computed using specific (often median plus one standard deviation spectra (i.e., conservative) deterministic response parame- 84% nonexceedance). As a result of the radiation ters for the structure. The structural response of energy from the base slab into the soil (radi- factor, FRS, is expressed as a product of the ation or geometric damping), very high equiva- factors influencing the variability on building lent modal damping ratios (20 to 40 percent of response. critical) are computed from the Quadrex soil-structure interaction analysis (Ref. [5]). An FRS~ FSA' F6- FM' FSD- FSS (3) average factor of safety of about 11.08 results for including a spectral shape factor representing the equivalent 20% to 40% damping over the 2 the ratio of the median site-specific ground to 10 Hz frequency range of interest. In addition, response spectra to the ground spectra used for the average ratio of the spectral acceleration design, a damping factor representing the vari- developed by the artificial earthquake time his- ability in response due to difference in actual tory to the specified ground response spectral damping and design damping, a modeling factor acceleration is about 1.08 over the frequency accounting for the uncertainty in response due range of interest (1 to 5 Hz). The overall factor of to modeling assumptions, a factor to reflect the

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

367 safety, FSA< to the ground response spectra Soil-Structure Infraction used in the analysis compared to the expected The soil-structure interaction analysis per- median value is about 1.08 x 1.08 or about 1.17. formed by Quadrex was conducted using current Damping state-of-the-art techniques and is considered to A comparison of the damping values used in be essentially median-centered. Variations due the Quadrex analysis and those expected in the to possible changes in the soil shear modulus at range of severe building distress is: shear strains above the SSE were accounted for in the factor of safety for modeling, FM, and the Damping (% Critical) effect of embedment was accounted for sepa- QUADREX FRAGILITY rately in FSD as previously discussed. Structure 4 5 Combined Structure Respon— Effect* The combined factor of safety, FRS. anc* Soil Hysteretic 9 13 associated variabilities may then be developed Geometric 7 to 27 7 to 27 for the above parameters. The combined factor of safety for the horizontal direction is 1.55. Ppj The average damping factor of safety, F\ is 0.18 and pu is 0.31. developed from the ratio of the spectral accel- Structure Fragility eration from the Quadrex analysis compared to The primary lateral load-carrying system of the spectral accelerations at the damping expected structure is of reinforced concrete construction. at structure or equipment failure of 1.02 was used. For reinforced concrete structures, the strength Modeling factor is a function of material strengths asso- The factor of safety due to modeling is ciated with the concrete and the reinforcing steel expected to be essentially unity with no ran- and relationships expressing strength of domness. A lognormal standard deviation for concrete walls in flexure and shear. uncertainty, Py, of 0,1 was estimated in order to Strength of Concrete and Reinforcing St—I account for mode shape effects. Concrete for the L- and P-Reactor structures Embedment was specified to have a minimum compressive When a structure is embedded such as the L- strength of 2500 psi at 28 days. The average and P-Reactors, the earthquake input at the base 28-day strength for 2500 psi concrete was con- slab can be expected to be significantly reduced servatively estimated to be approximately 3000 from the surface motion. Since the hazard curves psi with a logarithmic standard deviation of 0.12. are developed in terms of the free-field surface A factor of 1.2 was applied to the 28-day strength was acceleration levels, a factor of safety, FSD> to develop the strength of the aged concrete, developed in order to reflect any conservatism resulting in a median value of 3600 psi. A expected in the response computed by Quadrex logarithmic standard deviation associated with due to embedment as wel! as the variabilities aging was estimated to be 0.10. Combining the associated with this factor. effects of average 28 day strength with subse- In Ref. [5], a deconvolution analysis was quent aging gives an overall Py of 0.16. Based performed to determine the expected seismic on a survey of test results from other nuclear input at the base slab of the structure compared plants, the madian yield strength, /y, and loga- to the free-field surface input. These results are rithmic standard deviation, Pu, for the Grade 40 considered to be median-centered. However, reinforcing steel are 47 ksi and 0.09, respectively. because of licensing restrictions, the reduction in Strength of Shear Walls in Flexure spectral acceleration was not allowed to Equations to predict the overturning (in-plane) decrease below 60 percent of the free-field moment capacity of rectangular shear walls acceleration in the Quadrex analysis. To prevent containing uniformly distributed vertical rein- greater reduction, the response of the structure forcement were derived from the basic ultimate was multiplied by 1.3 in the Quadrex analysis. strength design provisions for reinforced Consequently, a factor of safety of 1.3 is used in concrete members subjected to flexure and axial the fragility analysis based on the assumption loads contained in the ACI code. These provi- that the Quadrex deconvolution analysis is sions are based upon the satisfaction of force essentially median-centered,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

368 equilibrium and strain compatibility and have The median capacity/demand ratio or been verified by testing, Hence, they are judged strength factor for the L-Reactor building is 2,83 to be median centered. Uncertainty in the median as described above. Uncertainty in this median flexural strength is estimated to be \\\j of 0.10 factor results from the following sources: 1) based on test data. material strength; 2) the strength relationship for Shear Strength of Concrete Walls shear; and 3) uncertainty introduced because a Recent studies have shown that the shear simple stick model was used to represent the strength of low-rise concrete shear walls with complex load distribution. Combining all sources boundary elements are conservatively predicted of uncertainty results in a total fi\j of 0,22, by the ACI code provisions. Median shear Structure Inelastic Energy Absorption strength which is based on test data of low rise The Riddell-Newmark ductility modified walls is given by an equation from Ref. [3]. Based response spectra approach (Ref. [1]) has been on an evaluation of the same experimental data, used to predict the inelastic energy absorption the logarithmic standard deviation of the median factor, F u, corresponding to the system ductility, shear strength equations was estimated to be [i. An effective ductility corresponding to shear 0.15. wall structures and earthquake magnitudes on Structure Strength the order of 6 (i.e. 5.3 < M < 6.3) has been used. A strength factor for this building is evaluated The majority of seismic risk for the SRP plant is as the structure capacity as given in median flexure or shear provisions with median material estimated to result from earthquakes in this properties divided by the corresponding shear magnitude range. and moment from the Quadrex soil-structure In order to estimate the system ductility, a story interaction analysis (Ref. [5]) for east-west drift approach has been utilized, where system earthquake ground shaking. The structure is ductility is given by: much weaker against east-west ground motion. Above El. 48, the building is a relatively strong box-type structure. Below El. 48 feet, the east- west running walls in the tower region are discontinued. There is a 5 foot thick roof slab at El. 48 feet, which extends to the north of the tower where Wj is the story weight, A M t is the total story region about 110 feet to an expansion joint which drift at failure, and A e , is the elastic story drift at structurally separates this portion of the building from additional structures to the north. In addi- yield level. The story drifts at failure are estimated tion, this 5 foot thick slab extends to the south of by developing a structure deflected shape in thetowerregiontoa5footthickshearwall. Shear which the story with the greatest amount of loads from the tower structure above and from inelastic behavior (i.e. the smallest capacity/de- the massive slab at El. 48 feet must be resisted mand ratio) is set to a drift judged to correspond by the shear walls to the south of the tower region to failure with stories below deforming at yield and by out-of-plane bending of the underlying level and stories above deforming in a manner north-south running walls. The capacity of the which is compatible with the inelasticity lower in building walls between Elevations 34 to 48 feet is the structure. It is estimated that the median story a factor of 2.83 times the shear load at this drift corresponding to structural failure due to elevation of the building. shear is 0.7 percent of the story height. Uncer- tainty is evaluated by taking 0.4 and 1 percent to Below El. 34 feet, there are additional shear be 10 and 90 percent probability levels. The walls at the southern portion of this building and deflected shape at yield is taken to be the to the northwest of the tower region to resist deflected shape from the Quadrex analysis lateral forces. Consequently, the capacity/de- scaled by the minimum strength factor of 2.83. mand ratios for lateral load resisting elements By this approach, the median inelastic energy from El. 0 to El. 34 feet are larger than for the absorption factor is evaluated to be 1.62. resisting elements between El. 34 and 48 feet. In addition, capacity/demand ratios for elements above El. 48 feet are larger than 2.83.

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 Structure Capacity tension and shear but with very limited bending The median capacity for the L-Reactor build- as shown in Figure 2a. For the case of thin shims, ing can be approximated by the median response the bolt is free to displace laterally a relatively factor times the median strength factor times the large amount to the point at which plastic hinges median inelastic energy absorption factor times are formed at the top and bottom of the bolt. the design free-field ground acceleration (0.2g). Plastic hinges are formed at a very low load level In this manner, the fragility of the structure as (about a factor of 0.63 times the loads developed from Quadrex floor spectra). However, the for- represented by the median capacity, the loga- mation of plastic hinges does not constitute rithmic standard deviation associated with ran- failure under transient earthquake loading. The domness, |3R, the logarithmic standard deviation joint will continue to slip until the bolt is stretched associated with uncertainty, f\j. and the 95 sufficiently to provide a clamping force between percent confidence of 5 percent probability of the shims which then carries the shear loads failure (HCLPF) level are given by: while the bolt is loaded in axial tension with A = 1.42g bending at the base and sole plates. A sketch pR = 0.18 illustrating this behavior is shown in Figure 2b. PU = 0.40 For the case of anchor bolts with solid shims, HCLPF = 0.55g median capacity was evaluated in accordance Equipment Fragilities with an interaction formula for combined shear Heat Exchanger* and tension (bending is assumed to be negli- The heat exchangers are horizontal, saddle- gible). The factor of safety for anchor bolts with mounted cylindrical tanks; mounted on four solid shims is calculated to be 3.02. Combining wheel rail trucks; and located at elevation -20 feet. the heat exchanger factor of safety and uncer- Seismic bracing has been added to the heat tainty with the response factor of safety, ran- exchangers to provide resistance to overturning domness, and uncertainty results in the following (transverse response) or rolling (longitudinal heat exchanger fragility (solid shim case): response). Bracing is bolted to the floor and to A = 0.94g steel members connecting it to the heat (3R = 0.18 exchanger. Provisions are made for either thin 3U = 0.33 laminated steel shims or solid block shims at base plates and the top of the bents where the bracing HCLPF = 0.41 g is bolted to the floor and to the heat exchanger, For the case of the anchor bolts with thin respectively. The shims are significant because shims, the ultimate shear capacity is given by the bending of the bolts at the base of the bracing horizontal component of the bolt tension as the which is permitted by ihin laminated shims gov- bolt has been displaced laterally, Pboits\n 6plus erns the seismic capacity of the heat exchanger. the clamping force on the shims times a coeffi- The design basis loads were not available so cient of friction, n(^bo»cos9- pS9ismicy The seismic loads were recalculated using the factor of safety on strength was therefore Quadrex floor response spectra at Elev. -20. With determined from the relationship: the exception of potential failure of the anchor P ,,(sin 0 + ucosG) (5) bolts protruding from the concrete floor, the b0 factor of safety for all other failure modes was 4 V + uP v seismic V-' seismic or more. The controlling capacity for the heat where 0 is the angle formed by the offset and [iis exchangers was found to be the base plate anchor bolts. The capacity of these bolts is the coefficient of friction between the shims and dependent on whether laminated shims or solid plates block shims are used since the joint behavior The angle 0 is related to the average bolt strain, under lateral loads is different. For the case of a £bolt by the following relation: solid shim, the joint will slip a small amount until the clearances in the shim and plate bolt holes are reached at which time the bolt bears against 1 boll the shim. For this case, the boil is loaded in

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

370 Main Tank The main tank consists of top tuba sheet, a bottom tube sheet, and a cylindrical tank. The top tube sheet is connected to the cylindrical tank wall by a flexible expansion joint and is supported -1 from the concrete by bearing supports. The bottom tube sheet and cylindrical tank are supported on a separate set of bearing pads. Thus, the bottom tube sheet and tank can respond essentially independently from the top tube sheet during a seismic event. Seismic loads and stresses in the top tube sheet and nozzle system are much lower than in the main tank and bottom tube sheet assembly. Six outlet nozzles are located just above the top plate of the bottom tube sheet. These nozzles are embedded in concrete. This provides a very a. Solid Shims stiff connection between the tank wall and the concrete which results in correspondingly high nozzle stresses occurring due to increased tank seismic displacements after failure of the bottom anchorage has occurred. The bottom tube sheet and tank assembly were modeled as part of the overall structure seismic mode) by Quadrex (Ref. [5]). Loads from this model were then applied to a detailed finite element model of a sector of the tank and nozzle. Seismic loads and stresses developed from these two analyses formed the basis of the fragility evaluation. Buckling of the tank wall was checked by Quadrex and found to have a very high factor of safety so that this mode of failure is not expected to control. Fragility of the main tank was based on failure of the nozzles in shear (i.e., trunnion loading). A factor of safety b. Thin, Laminated Shims on strength of 3.3 with Pu of 0.14 were calculated Figure 2 - Behavior of Shims for the main tank based on the above assump- Based on a median coefficient of friction between tions. Combining the factors of safety and the laminated shims of about 0.5 (steel on steel variabilities for strength with those for structure sliding friction) and an average strain over the and equipment response, the fragility for the main bolt length of 1% at failure, a median factor of tank was evaluated as: A = 1.04g safety of the bolts of about 2.27 was calculated fiR = 0.20 for the thin shim condition. Including the building PU = 0.44 response factor of safety and variabilities, the HCLPF =0.36g fragility for the thin shim condition is: A = 0.70g Pumpsp . Control Rods. & For—t PR = 0.18 It was determined that these items are not PU = 0.33 contributors to the seismic risk of seismic indirect HCLPF = 0.30g RCL pipe break when their capacities are compared to that of the main tank, heat exchangers, and building.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~371 1.0~~

~~0.1 51 Confidence~~

~~l.O«g~~

~~0.4~~

~~0.2~~

~~'0.28g O.47g Q.77g Q.OS~~

~~0,2 0.4 0.6 O.S 1.0 1 .2 1.4~~

~~PEAK GROUND ACCELERATION (g)~~

~~Figure 3 - Plant Fragility (Laminated Shims)~~

PLANT DAMAGE STATE FRAGILITIES curves of the second component. The resulting Plant damage state fragilities for seismic- nxn curves are then condensed back to n curves, induced indirect RCL pipe are developed from which are then combined with the n curves of the "he individual component fragilities described in third component. This process is continued for the previous section. Following the rules of all components in the Boolean expression, Boolean algebra, the individual component fra- resulting in n plant damage state fragility curves. gilities were combined, two at a time, using the Plant level fragility including heat exchangers Discrete Probability Distribution (DPD) approach with thin laminated shims, the tank, and the (Ref. [2]) to form plant damage state fragility building with independent randomness and curves. For plant damage states corresponding uncertainty for all components is a median to seismic-induced indirect RCL pipe break, the acceleration capacity of about 0.64 g with a high Boolean expression is: confidence, low probability of seismic failure of Pipe Break = Tank u Heat Exch u Bldg about 0.28 g. A plant damage state fragility curve indicating that the plant damage state consists of for the case of laminated shims is shown in Figure the union of the tank with the heat exchangers 3. If solid block shims are assumed, the median with the building. capacity increases to about 0.76 g with a HCLPF With the DPD approach, the individual com- of about 0.33 g. If the heat exchangers are strengthened to the extent they do not influence ponent fragility curves are first discretized into a the plant fragility, the median capacity increases family of fragility curves, each with a probabilistic to about 0.93 g with a HCLPF of 0.36 g. weighting, representing the uncertainty (characterized by the (3y value) in the fragility evaluation. Each one of the, say n, curves of one component is then combined, according to the rules of Boolean algebra, with each one of the n

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

372 SEISMIC RISK In order to obtain the annual occurrence The seismic risk of RCL pipe failure frequencies of the plant damage states, the family indirectly-induced by earthquakes is developed of plant damage state fragility curves discussed by a convolution of seismic hazard curves with above are convolved with the family of seismic fragility curves representing the plant damage hazard curves. The convolution between the states. For this study, the set of seismic hazard seismic hazard and the plant damage state fra- curves given in Figure 4 (from Ref. [4]) was used. gility is carried out by selecting a hazard curve j The uncertainty in the earthquake hazard is and a fragility curve i; the probability assigned to accounted for by developing a family of curves the plant damage frequency resulting from the and assigning a subjective weighting factor (in convolution is the product of the probabilities pj this case, 0.1) to each curve. Due to the high and qi assigned to these two curves. The capacity of the plant fragilities, the hazard curves convolution operation consists of multiplying the had to be extrapolated to lower annual frequency frequency of occurrence of an earthquake peak of exceedances than shown in Figure 4 in order ground acceleration between a and a+da with to compute the low seismic risks. These curves the conditional frequency of the plant damage were linearly extrapolated on semi-log paper state, and integrating such products over the down to 10*9 for this rjSk study. In addition, these entire range of peak ground accelerations 0 to extrapolated curves were truncated at 1.5 g peak 1.5 g. Comparisons of the seismic risk of RCL ground acceleration, since hazard curves in Pipe Failure for several cases are shown in Table excess of 1.5 g are not considered credible for 1. the SRP site. Table 1 L-Reactor Seismic Risk of RCL Pipe Failure Mean Median 95% to 5% Confidence Laminated 6x10"6 7.8x10-7 3.3x10"5-< 10"8 Shims Solid 2.9x10-6 2.9X10-7 1.6x10-5-<10-3 Shims Upgraded 2.0x10-6 7.4x10"8 9.6x10-^IO-8 HE HE & Tank 2.9x10-7 <10"S 1.4x10-6-<10-8 Upgrade

Contribution to Seismic Risk from DWfwnt Acceleration Range* Ranges of acceleration which contribute most significantly to the overall frequency of occurrence of the damage state can be evaluated by integrating over small acceleration ranges and comparing the occurrence frequency obtained with that obtained by integrating over the entire range of accelerations. Figure 5 gives the percent contributions for mean risk from various acceleration ranges assuming a maximum credible peak ground acceleration of 1.5 g. Ac.r.r-1 (r»~cz : n

~~Figure 4 • Seismic Hazard Curves~~

Second DOE Natural Phenomena Hazards Mitigation Conierence - 1989 373 The contribution from the acceleration ranges 40 below 0.25 g is very small for both the mean and median seismic risk. The majority of the mean

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~374 LARGE BREAK FREQUENCY FOR THE SRS PRODUCTION REACTOR PROCESS WATER SYSTEM~~

~~W. L. Daugherty, N. G. Awadalla, R. L. Sindelar S. H. Bush Westinghouse Savannah River Company Review & Synthesis Associates Savannah River Site, Aiken, SC Richland, WA~~

ABSTRACT The objective of this paper is to present the results and conclusions of an evaluation of the large break frequency for the process water system (primary coolant system), including the piping, reactor tank, heat exchangers, expansion joints and other process water system components. This evaluation was performed to support the ongoing PRA effort and to complement deterministic analyses addressing the credibility of a double- ended guillotine break. This evaluation encompasses three specific areas: the failure probability of large process water piping directly from imposed loads, the indirect failure probability of piping caused by the seismic-induced failure of surrounding structures, and the failure of all other process water components. The first two of these areas are discussed in detail in other papers. This paper primarily addresses the failure frequency of components other than piping, and includes the other two areas as contributions to the overall process water system break frequency. The most vulnerable components are the expansion joints. The large break frequency for the expansion joints is estimated to be 9.7 x 10-6 per reactor-year. This break scenario is equivalent in severity to a double-ended guillotine break of the adjoining pipe. A limited break with restricted flow area is a much more likely failure scenario for these components, with an estimated frequency of 5.6 x 10"3 per reactor-year. The combined large break frequency for the entire process water system under directly imposed loads is about 1.5 x 10'5 per reactor-year. Added to this is the indirect failure probability. Although the indirect failure probability was calculated specifically for the piping, it serves as a rough estimate of the indirect threat to the entire process water system. This source contributes 7.8 x 10'7 per reactor-year (median estimate) for L or K reactor. Differences in the reactor building for P reactor lead to slightly different results.

INTRODUCTION initiated in 1985 to characterize the integrity of the process water system (primary coolant system) and The Savannah River Site (SRS) production estimate the remaining useful lifetime of the reactors operate at low temperature and pressure. reactors. One subtask of this program was to The material of construction for the primary estimate the failure frequency for each component pressure boundary is Type 304 stainless steel. of the process water system. This paper reviews These reactors were built in the 1950's, and have and summarizes the failure frequency estimates. undergone various modifications and upgrades The failure frequency for the piping (by both direct since that time. The Reactor Materials Program was and indirect means) is discussed in detail in

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

375 companion papers and is summarized herein for Each head is held in place by 84 staybolts. completeness. Seventy-two C-clamps around the periphery provide additional restraint, although restraint is DISCUSSION not their primary purpose. Figure 1 illustrates the heat exchanger and details of the head The process water system loop is comprised of configuration. the reactor tank (including outlet nozzles), main During 720 heat exchanger-years of experience circulation pump, two heat exchangers, inlet at SRS with the current heat exchanger design there nozzles to the tank plenum and the interconnecting have been 77 cases of cracking in the heads, piping. In addition, the process water system primarily in the inlet heads. Nine of these cracks contains valves, expansion joints and flanges. The leaked. The ductile behavior of the austenitic failure frequency of each of these components has stainless steel in the heat exchanger heads, and the been evaluated and is discussed separately below. sensitive leak detection system provide high The main concern of this paper is the frequency confidence that cracking will not lead to large of a sudden large break of the primary pressure failures. On the basis of engineering judgment, the failure frequency for the heat exchangers is boundary. Loss of coolant through pump shaft 7 seals or valve stem leakage are of no concern to estimated to be 1 x 10' per heat exchanger-year. this evaluation. Likewise, a through-wall crack with its ensuing leakage is not of concern here, REACTOR TANK unless it could lead to a rupture before being The reactor tank wall is constructed of 0.5 inch detected and corrected. thick stainless steel plate. Because of the similarity The SRS production reactors have been between the tank and the process water piping, the operating successfully for approximately 35 years. probability of cracking and leakage is expected to Due to their age, history and various unique be similar to the corresponding value for the features they do not lend themselves readily to piping. However, the likelihood of a catastrophic standard probabilistic analyses. In particular, while failure is extremely low, based on mechanistic failure analyses of piping has been developed in analyses showing the tank would leak before great detail for commercial reactors and could be breaking. On this basis, the tank failure frequency adapted to SRS reactor piping, no comparable is judged to be similar to or less than values analytical tools are readily applicable to SRS typically cited for power reactor vessels. In reactor components such as the reactor tank or general, power reactor PRA's use the WASH-1400 expansion joints. For this reason, the failure value for pressure vessel failure of 2.7 x 10'7. frequency estimates discussed in this paper have Therefore, a value of 3 x 10"7 is appropriate for the been developed in part from engineering judgment. SRS reactor tanks. Several consultants, with vast experience in the industry, have assisted with these evaluations. PLENUM INLET NOZZLES Comparison to industry experience or statistical The plenum inlet nozzles are constructed of treatment of operating data is included where wrought plate and stainless steel castings which are relevant information is available; however, in many welded together with internal flow vanes. The flow cases the operating conditions and unique design vanes act to reinforce the nozzle against pressure features of the SRS reactors limit the applicability loads. The inherent toughness of the material and of such data from other sources. Further the sensitive leak detection system provide high discussion of the basis for the component failure confidence that a crack would not lead to sudden frequencies is given in reference [1], rupture. Therefore, the failure scenario of concern is the failure of the flow vane attachment (the vanes HEAT EXCHANGERS are attached by either staybolts or fillet welds) The SRS reactors each have 12 horizontal combined with a severe overpressurization once-through heat exchangers. Portions of the heat accident. exchanger pressure boundary that contact the Among the 5 SRS reactors (three of which primary coolant are the inlet and outlet heads and remain operational) there are approximately 750 the tubes. A failure of one or several tubes docs not nozzle-years of successful operation without a constitute a severe accident in terms of a threat to failure. Statistical treatment of this data produces a the core. The primary source of a large LOCA from median estimate of the failure frequency of the heat exchangers is the inlet or outlet heads. 3.1 x 1Q-4 per year. This estimate is controlled by

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~376 33' 6"~~

~~Moderator Inlet Outlet~~

~~River Water Inlet~~

~~Figure la. Heat Exchanger Configuration~~

~~Shell~~

Heat Exchanger Tubesheet Seal Membrane Head Figure lb. Detail of Heat Exchanger Staybolt Figure lc. Detail of Heat Exchanger C-Clamp the relatively low number of operating years. A the primary system, the SRS reactors contain review of additional data supports a much lower expansion joints as part of the primary pressure estimate. boundary. The expansion joints contain stainless Several original casting defects exist in the steel convolutes. The expansion joint convolutes nozzle castings. These are inspected periodically are the only part of the primary pressure boundary and have shown no significant change. No other for which analyses have not demonstrated a leak- degradation has been observed in the nozzles, before-break (LBB) capability. As such, the although accessibility to the flow vanes is difficult possibility of a sudden rupture of the convolutes is and non-destructive examination in the cast assumed to present a very real threat. sections is of limited reliability. Nevertheless, Approximately 2250 expansion joint-years of efforts are in progress to verify the integrity of the operating experience in the primary system has entire nozzle, including flow vanes. Service been developed at SRS. There have been 19 leaks conditions are generally mild, with low operating in SRS expansion joints, but no breaks. Most of stresses. On this basis, the failure frequency is these leaks were induced by fatigue during the estimated to be less than 1 x 10"8 per nozzle-year. early years of operation. Subsequent modifications Verification of the flow vane attachment integrity greatly reduced the incidence of leakage. With no will further enhance nozzle integrity and provide breaks in the operating history, a statistical support for a lower estimate. treatment gives an estimate of the true failure rate. Assuming failures are randomly distributed in time, EXPANSION JOINTS we have the expression: Unlike commercial power reactors where expansion joints are limited to application outside

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

377 Additional discussion of expansion joints may be found in references [1] and [2], PUMPS AND VALVES Due to the functional requirements on pumps and valves for rigidity, these components are typically much thicker than required by the ASME Code from a pressure boundary standpoint. Leakage through seals and valve packing is common, but such leaks in most cases are small and will not be considered here. The valve and Flange pump bodies are made of cast stainless steel. This material contains sufficient ferrite to render them not susceptible to intergranular stress corrosion cracking. Because of the structural overdesign of these components in order to meet functional requirements, a pump or valve would most likely fail as a result of bolting failure; either the flange Figure 2. Expansion Joint Design bolting, valve bonnet bolting or pump suction cover bolting. In general, as one or more bolts X = 1 - (Pr)l/m = 1 - (0.5)1/2250 = failed, the joint would tend to open and allow 4 leakage, thus leading to early detection of the 3.1 x 10" per expansion joint-year. condition. The failure frequency estimate for the valves is Taking Pr equal to 0.5 represents a 50% probability of obtaining zero breaks in m expansion based on failure of the bolted joints. Counting the joint-years. bonnet bolting and one-half of each end flange as Failure of the convolutes alone will not produce part of the valve gives two bolted joints per valve. a large break, due to the presence of an internal Using the failure frequency for flanged joints developed below, the valve failure frequency flow sleeve (see Figure 2). A large break would 9 require either of two additional events: (1) the fldw becomes 2 x (5 x 10" ) = 1 x 10"** per valve-year. sleeve is not present (due to prior failure or a An O-ring seal in the pump suction cover might previous omission of the sleeve), or (2) the prevent detection of such leakage, however. Therefore, the pump failure frequency is higher, or external restraints (tie rods) fail and allow the joint 7 to stretch out. Stretchout of the joint is assumed to 1 x 10" per pump-year. lead to convolute failure at the same time. Inspection of the flow sleeves and their attachment FLANGED JOINTS welds provides confidence that the probability of a The process water system contains flanged missing sleeve is no greater than 10"3 per joints, both B16.5 150 and 300 Class, connecting expansion joint-year, based on engineering various components to the piping and connecting judgment. The evaluation of external restraints pipe segments. There have been, or are now, an conservatively considers only the tie rods and estimated 30 million B 16.5 150 or 300 Class ignores any contribution from nearby supports and joints, 6 NPS and larger, in service throughout the components. Based on a statistical treatment of tie world over the past 60 years. A review of available rod experience (three tie rod failures, but none information has failed to identify any breaks in 6 leading to joint failure) and inspections to verify NPS or larger B 16.5 joints [3]. Assuming that any current tic rod integrity, the probability of joint breaks would be distributed randomly in time and stretchout is estimated to be 2.3 x 10"7 per applying the statistical treatment presented above expansion joint-year at a 50% confidence level. for expansion joints, the failure frequency for Combining these numbers provides an overall flanged joints is 1 x 10*7 per joint-service life. This estimate of a expansion joint large break break: value is based on a 95% upper bound confidence level. Assuming an average useful service life of 9 4 3 7 20 years gives a failure frequency of 5 x 10" per Pr(EJ) = (3.1 x 10- ) x (1 x 10- ) + 2.3 x 10* = joint-year. 5.4 x 10'7 per expansion joint-year. Second DOE Natural Phenomena Hazards Mitigation Conference - 1983

378 PROCESS WATER PIPING large break frequency for the process water system The piping is mentioned here for completeness. is obtained by summing each individual Details are provided in two companion papers contribution. As seen from Table 1, the expansion (4,5). Piping failure is evaluated for two cases; as a joints contribute approximately two-thirds of the result of loads acting directly on the pipe, and as a total direct failure frequency. The acceptability of result of seismic loads acting indirectly, such that this contribution is assessed in terms of the risk it the failure of a nearby component or building presents. Preliminary results from a level 1 PRA structure leading to the failure of the piping. The for the SRS reactors show that a large break in the piping direct failure frequency is estimated to be expansion joint contributes no more than 28% of 6 the total core damage frequency [6J. The failure 1.6 x 10" per reactor-year. frequency for each of the remaining components is The indirect failure frequency was evaluated for 6 seismic loads ranging up to 1.5g peak ground extremely low, on the order of 10" or less. acceleration. For this extreme case, the median The indirect failure frequency was developed failure frequency is 7,8 x 10"7 per year for L and K specifically for the piping. However, those indirect reactors, and 1.3 x 10'6 for P reactor. The different failure scenarios which threaten the piping also value for P reactor is due to a difference in the threaten other components. Similarly, there would reactor building, which affects the building be few failure scenarios that threaten other fragility. components that do not also threaten the piping. Therefore, the indirect piping failure frequency approximates the indirect failure frequency for the RESULTS entire process water system. A summary of the failure frequency for each component is given in Table 1. The total direct Table 1. Summary of Component Failure Frequencies No. Components Failure Frequency (per year) Component Per Reactor Per Component Per Reactor Process Water Piping (Direct) 1.6 xlO"6

~~Heat Exchangers 12 1 x 10-7 1.2 xlO-6~~

~~Main Tank 1 3 x 10-7 3 x 10-7~~

~~Plenum Inlet Nozzles 6 1 x 10-8 6 x 10-8~~

~~Expansion Joints 18 5.4 x 10-7 9.7 x 10-6~~

~~Pumps 6 1 x 10-7 6x10-7~~

~~Valves 28 1 x 10-8 2.8 x 10-7 Ranged Joints (4 NPS and Larger) 144 5 x 10-9 7.2 x 10-7~~

~~Total (Direct) 1.5 x 10-5~~

~~Indirect (1.5g, median) 7.8 x 10-7, L and K reactors 1.3 x 10"6, P reactor Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

379 CONCLUSIONS 151 "Probabilistic Evaluation of Main Coolant Pipe Break Indirectly Induced by Earthquakes The failure frequencies for each component in Savannah River Site L & P Reactors", the process water system of the SRS production S. A. Short, D. A. Wesley, N. G. Awadalla reactors has been evaluated. This work has been and R. P. Kennedy, Proceedings of the performed in support of the SRS PRA effort and as Second DOE Natural Phenomena Hazards an adjunct to deterministic analyses of the integrity Mitigation Conference, October 3-5,1989, of the process water system. The probabilistic Knoxville, Tennessee. analyses, combined with the deterministic analyses, strongly support the conclusion that a [6] DPST-88-826, "Preliminary Analysis of the sudden large break in the process water system is Frequency of Core Melt from Primary Cooling incredible. Water Leaks", W. H. Baker, March 1989.

ACKNOWLEDGEMENT REFERENCES The information contained in this article was [ 1 ] "Failure Frequencies and Probabilities developed during the course of work under Potentially Applicable to Savannah River Contract No. DE-AC09-76SR00001 (now Reactors", S. H. Bush, attachment 1 to Contract No. DE-ACO9-88SR18035) with the DPST-88-469, "Reactor Materials Program U. S. Department of Energy. Process Water Component Failure Probability", W. L. Daugherty, April 1988. [2] "Evaluation of Portions of SRP Process Piping Pressure Boundary Consisting of Expansion Joints", E. C. Rodabaugh, attachment 1 to DPST-88-434, "Evaluation of Expansion Joints in the Reactor Process Water Piping", E. R. Hartman, G. A. Abramczyk and W. L. Daugherty, July 1988 [3] "Evaluation of Portions of SRP Process Piping Pressure Boundary Consisting of Bolted Flanged Joints, Valves and Pumps", E. C. Rodabaugh, attachment 1 to DPST-88- 404, "Reactor Materials Program LBB Assessment - Flanges, Valves and Pumps", E. R. Hartman, G. A. Abramczyk and R. L. Sindelar, March 1988. [4] "Failure Probability Estimate of Type 304 N. G. Awadalla, R. L. Sindelar, H. S. Mehta and S. Ranganath, Proceedings of the Second DOE Natural Phenomena Hazards Mitigation Conference, October 3-5, 1989, Knoxville, Tennessee.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~380 Session 12 Probabilistic Risk Assessment~~

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

~~381 EXTERNAL EVENT PROBABILISTIC RISK ASSESSMENT FOR THE HIGH FLUX ISOTOPE REACTOR (HFIR)~~

~~G. F. Flanagan Oak Ridge National Laboratory P. O. Box 2008 Oak Ridge, Tennessee 37831~~

~~D. H. Johnson, D. Buttemer, H. F. Perla, S. H. Chien Pickard, Lowe, and Garrick, Inc. 2260 University Drive Newport Beach, CA 92660~~

~~ABSTRACT~~

The High Flux Isotope Reactor (HFIR) is a high performance isotope production and research reactor which has been in operation at Oak Ridge National Laboratory (ORNL) since 1965. In late 1986 the reactor was shut down as a result of discovery of unexpected neutron embrittlement of the reactor vessel. In January of 1988 a level 1 Probabilistic Risk Assessment (PRA) (excluding external events) was published as part of the response to the many reviews that followed the shutdown and for use by ORNL to prioritize action items intended to upgrade the safety of the reactor. A conservative estimate of the core damage frequency initiated by internal events for HFIR was 3.11 x 10"\ In June 1989 a draft external events initiated PRA was published. The dominant contributions from external events came from seismic, wind, and fires. The overall external event contribution to core damage frequency is about 50% of the internal event initiated contribution and is dominated by seismic events.

INTRODUCTION The High Flux Isotope Reactor (HFIR) is a Partly as a result of this review process, a high performance isotope production and Probabilistic Risk Assessment (PRA),[1] of research reactor which has been in operation at HFIR was completed for internal initiated the Oak Ridge National Laboratory (ORNL) events in January 1988. This was the first PRA since 1965. Its main missions are the on a large research reactor in the United States. production of transuranic and cobalt isotopes, The PRA initiated by external events was materials irradiation research, and neutron completed in draft form in June 1989. The scattering research. approach (used for the external events In late 1986 a special internal post-Chernobyl initiators) and results of the external events review of HFIR discovered unexpected neutron assessments will be presented in this paper. embrittlement of the reactor vessel. As a result of the discovery the reactor was shutdown in HIGH FLUX ISOTOPE REACTOR DESIGN November 1986. The Department of Energy The HFIR is an 85 MW flux trap reactor. A (DOE) and ORNL began an extensive review of schematic of the reactor is contained in Figure the reactor design, safety, operation, 1. It is water cooled and beryllium moderated. maintenance, and management. Over twenty It operates at 468 psi pressure with an inlet reviews of various depths have been conducted temperature of 120-F and outlet temperature of to date by DOE, ORNL and independent 158°F. The peak thermal flux in the flux trap oversite groups such as the National Academy is 5 x 1015 n/cm2-scc which makes the HFIR the of Science/National Research Council and the highest thermal flux reactor in the world. The Advisory Committee on Nuclear Facility Safety. core of the reactor is small (17 1/2 inches

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~382 Figure 1. Vertical Section of HFIR Reactor Vessel and Core~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

383 diameter, 24 inches in height) with a 5 inch earlier published internal event initiated diameter target hole through its center. The assessment, a brief summary of the results of core contains about 9.6 kg of highly enriched the internal PRA are presented. (93%) Uy\ arranged in two concentric cylindrical elements. The inner clement HFIR PROBABILISTIC RISK ASSESSMENT contains 171 involuted plates and the outer 369 (INTERNAL EVENTS) involuted plates. The core is made up of a The HFIR PRA[1J was developed with UjOg/Al mixture clad in aluminum. The core is several uses in mind. Foremost, it was required replaced every 24 days. The Be moderator by the DOE design review team; within ORNL surrounds the core and is about 1 ft thick. it is used for safety improvement and to help Control is achieved by 4 safely plates arranged prioritize the many design and administrative in a cylinder around a solid control cylinder. changes required by the numerous review The outer cylinder is raised and the inner committees. In addition it is also used for lowered to increase reactivity and keep a operator and engineer training, emergency symmetric flux profile. These control cylinders planning, technical specification modification, are sandwiched between the core and the Be maintenance improvements, and to help define reflector and are composed of Eu2O,, and Ta. and document the safety design basis of the The reactor core is contained in an 8 foot plant. diameter pressure vessel that is about 19 ft high. The project was subcontracted to Pickard, The pressure vessel is located near the bottom Lowe and Garrick Inc. (PL&G), Newport of a large pool (36 ft deep and about 18 ft Beach, CA and work began in July 1987 with across) containing 85,000 gallons of water. the final report (excluding external events) The pressure of 468 psi is maintained by issued in January 1988. Several basic compressing the primary system water using a assumptions were set forth to guide the work. prcssurizcr pump in combination with a system These are shown in Table 1. of letdown valves. The flow (16,000 gal/min) is The system was modeled using the so-called achieved by 3 out of 4 AC motor driven PL&G methodology commonly used on several primary pumps and it is downward through the commercial nuclear power plants. It consists of core and target regions. Decay heat is removed large event trees with only sparse use of fault using a small DC motor to drive the primary trees. This is different than the approach used pumps. The power to the DC motor is in the classic Reactor Safety Study (WASH- supplied using a dedicated battery power supply 1400) [2] and used by the Nuclear Regulatory or by using off-site power, on-site diesel Commission in several of their internal generators, or portable diesei generators assessments. (AEPG's) connected to inverters. The internal initiating events were selected by A schematic of the HFIR process flow system applying the following six steps: (1) examine is included in Figure 2 and a schematic of the the 20 years of operating history and the electrical power distribution system is included quarterly technical reports, (2) review the HFIR in Figure 3. Accident Analysis Rcport,[3] (3) review the The reactor is contained in a large reactor HFIR design, drawings, and operational building 128 x 160 x 110 ft which is maintained procedures, (4) hold discussions with the at a slight vacuum. Exhaust fans continuously original HFIR design team, (5) extensively pull air from the building through a scries of review the incidents at other research reactors filters and exhaust up a 250 ft stack. The and applicable commercial nuclear power building, filters, fans and stack act as a dynamic reactor experience, and (6) create a master logic confinement in the event of an accident. diagram (MLD) which generally examines how The reactor was built in 1965 to Uniform the HFIR core could be damaged. Building Code Seismic Standards resulting in a Because of the simplicity of the HFIR design, seismic design acceleration of about O.OSg's. the plant models were rather simple to develop The primary coolant system was upgraded in and were expressed using event sequence 1987 to enable it to withstand O.15g's, which is diagrams (ESD) in addition to using event trees. the safe shutdown earthquake for the HFIR. The ESD was found to be helpful in explaining Since the external events Probabilistic Risk accident scenarios to review teams, management, Assessment (PRA) made extensive use of the and operators. For the most part, the HFIR

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~384 MAGNETIC COUPLER~~

~~PRESSURE CONTROL LETDOWN VALVES PCV-127-7~~

~~PRIMARY COOLANT SYSTEM~~

~~S KGPM/P'JMP 600 HP - 2.3 KV AC f 3 HP - 120V OC (EMERGENCY) CONTBOLLED BY VESSEL INLET TEMPERATURE SIGNAL~~

~~SECONDARY 377 (MANUAL TRIM) COOLANT OTHER HEAT SYSTEM LOADS fr POOL HEAT EXCHANGER. AIR CONDITIONING TRU. ..~~

~~2 *T ;C HP Ti'/PE3AT C0NT3CLL£3~~

~~CCCL.NG rc-.VEPS~~

~~=.- ?J V.~~

~~Figure 2. HFIR Process Flow Schematic (primary and secondary systems)~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~385 4MVN0MMAI~~

~~400V NC~~

~~I MCCIM Mcc/r I I I cx>l I v J™ V J~~

~~oM SWITCHOtANI IjSSnSg1*" j~~

~~O I \ | roun rout MOTOW* 7~~

~~TPWNSFIK SA TKANSf [X f I nwmir \ \ OtijU U!JLJ MNO.T UQHTING OlSTKiaUTON *U*\. ICHCSCL-TIST LOAOl E I M 2MMI0V~~

~~«I»CTOK I 4C~~

~~I-I'l't-UVDC~~

~~I »»»1L '.S' • 31 -0 • - UV 3C > •owtCOT«O* I »NO sr~~

~~Figure 3. HFIR Electrical System Schematic Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

386 PRA used established assessment methods. opinion coupled with data from fuel There were two initialing events, however, manufacturing, inspection and quality control thai were unique 10 HFIR and required a new procedures. innovative approach. These were flow blockage events and fuel damage resulting from Results of ihe HFIR PRA (Internal Events) manufacturing defects and assembly errors. In order to facilitate source term determination and subsequent off-silc Flow Blockage consequence analysis, the results of the HFIR Because of the very high power density and PRA were expressed in terms of plant damage narrow fuel channels in the HFIR and states. The plant damage state matrix is shown experience at other research reactors it was in Figure 4. The matrix categorizes the end acknowledged by the original designers that core state of an event tree as to (1) the extent of the damage could occur as a result of small flow damage, (2) whether the primary system is intact blockages. In order to reduce this vulnerability, following the accident, (3) in case of a loss of a strainer containing small orifices was inserted coolant accident (LOCA), whether the break is in the inlet primary coolant pipe, and the plant inside or outside the reactor pool (which confinement, shielding, and water clean-up provides fission product scrubbing), and (4) .systems were designed to accommodate a core whether power is available to one, two or all melt which might be expected to occur during three exhaust fans. the lifetime of the plant. The results of the PRA are also expressed in There arc no system models or explicit data terms of frequency of core damage. Table 2 to use for modeling and quantifying flow indicates the overall frequency of core damage blockage scenarios, since such blockages arc not as a result of internal initiated events. expected in the commercial nuclear power plants. EXTERNAL EVENTS PRA APPROACH A very structured expert opinion approach[4] Following the internal events assessment the was developed by S. Kaplan of PL&G which same subcontractor began to examine the risks used the HFIR designers, operators, and associated with external event initiators. The engineers as the knowledge base. initiators considered consisted of the 9 major categories below: Fuel Element Defects Because of the high power density and Seismic narrow flow channels, fuel defects associated Wind/Tornado with the manufacturing and assembly of the fuel Fire/Smoke plates and elements could cause narrowing of Floods (External and Internal) the flow channels, in turn causing surface hot Spray (Steam and Water) spots or flow starvation. These may be of Explosions sufficient severity so as to cause fuel damage. Missiles Interaction with fuel experts at ORNL and Caustic Attack Argonne National Laboratory identified a Falling Objects comprehensive list of 10 potential defects. Based on further interaction with fuel experts Except for the first two initiators, the olher and detailed examination of the manufacturing contributors were all assessed using the same process including the inspection and quality general approach. These initiators will be control procedures, the 10 defects were reduced referred to as Internal Hazard Initiators (IHI). to five categories which consist of 3 fuel inhomogcneilics and 2 assembly errors. Specific Internal Hazard Initiators (Approach and details on the manufacture and inspection Rcsulls') processes relating to the defects on (he short The assessment of the internal hazard list were obtained from the fuel manufacturer initiators begins with an identification of and the ORNL fuel experts. Based on the initiators and an assessment of potential approach taken for the analysis of flow interactions between the hazard and the plant blockages, small event trees were developed and equipmenl, referred to as spatial interaciions. branch points were quantified using expert This is accomplished by an extensive

~~Second DOE Natural Phenomena Hazards Miligation Conference - 1989~~

~~387 Table 1. Basic Assumptions Used in the HFIR PR A~~

~~1. Core damage will be defined as occurring at (he onset of incipient boiling.~~

~~2. The reactor configuration assessed would be that at restart (includes power reduction and all prc-rcstart design modifications) with the addition of the portable dicscl generators.~~

~~3. The probability of vessel failure would come from the "Evaluation of HFIR Pressure-vessel Integrity Considering Radiation Embritllcment" - ORNL/TM-10444, edited by R. D. Cheverton.~~

~~4. The plant specific HFIR data is to be used wherever possible and to the extent possible.~~

~~5. Consideration should be given to accidents which have occurred at other research reactors when exploring initiating events.~~

~~6. Results should be expressed in such a way as to facilitate ease in calculation of off-site consequences.~~

~~7. Models should be fluid (easily modified as the design changes) in order to make the assessment a "living PRA".~~

~~EXTENT OF CORE DAMAGE~~

~~PARTIAL CORE DAMAGE TOTAL CORE DAMAGE~~

~~PRIMARY SYSTEM INTACT? PRIMARY SYSTEM INTACT?~~

~~YES NO YES NO~~

~~BREAK IN POOL? BREAK IN POOL?~~

~~YES NO YES NO~~

~~SBHE AVAILABLE?~~

~~YES NO YES NO YES NO YES NO YES NO YES NO~~

~~PLANT DAMAGE 1 2 3 4 5 6 7 8 9 10 11 12 STATE:~~

NOTE: AN "E" SUFFIX IS ATTACHED FOR THOSE SEQUENCES IN WHICH CORE DAMAGE OCCURS EARLY IN CORE LIFE. A "B" SUFFIX INDICATES SCENARIOS WITH CONFINEMENT BUILDING FAILURE. A "D" SUFFIX INDICATES A SCENARIO WITH A DRY POOL

~~Figure 4. HFIR Plant Damage State Matrix~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 388 Table 2. Initiating Event Categories, Mean Frequencies, and Contribution to Core Damage~~

Mim Contribution to Cora Mean Cor* Cattgory Frtqutncy Damage Frtquancy Damage (yr<1> (percent) Frequency 1A Manual Scram 21,4 7,1 2,2-5 1B Inadvertent Control Plate Drop 3.16 1.2 3.7-6 1C Inadvertent Scram 1.01-1 0.1 3.6-7 2A Complete Loss of Offsite Power 4.43-1 18.0 5.6-5 2B Loss of Preferred Feeder 4.10-1 2.4 7.4-6 2C Loss of Switchgear DC 3.79-3 0.1 3.9-7 3A Runaway Pressuri2er Pump 1.34-1 1.5 4.8-6 3B Loss of Running Pressurizer Pump 9.65-1 0.5 1.6-6 4A L1 Scenarios 1.5-5 4.8 1.5-5 4B L2 Scenarios

L2E 2.02-5 6.8 2.1-5 L2MCP 3.42-5 10.9 3.4-5 4C L3 Scenarios 2.00-5 6.4 2.0-5 40 L4 Scenarios (Total) 2.12-5 6.8 2.1-5 5 Small Break LOCA 4.56-3 5.1 1.6-5 6 Large Break LOCA 3.30-5 10.6 3.3-5 7 Beam Tube Failure 5.94-4 0.7 2.1-6 8A Reactivity Insertion 1.60-1 4.2 1.3-5 88 Degraded Secondary Cooling 2.32-1 0.6 2.0-6 8C Loss of Instrument Air 8.13-2 5,5 1.7-5 80 Degraded Primary Flow 2.68-1 7,1 2.2-5

~~Note Exponential notation is indicated in abbreviated form; ,e.. 2.9-5 - 2.9 x lO"5. Total (Internal Events) - 3.11-4.~~

~~Table 3. Contribution of Fire, Flood, and Other Environmental Hazards lo Core Damage Frequency~~

~~Contribution Core Percent of Internal Initiating Description Damage Frequency Event Core Damage Frequency~~

Fire Scenarios 1.83xiO5/year 5.88 Flood and Other Environmental Hazard Scenarios 1.81 xiO6/year 0.58 Fire, Flood, and Other Environmental Hazard Scenarios 2.01 xiO5/year 6.46 Internal Initiating Event 3.11 xiO*/year

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

389 examination of plant drawings, plant layout, and cxposurcd to a fire. a detailed plant walk-down, In the case of the The annual frequency of a fire at HFIR HF1R, 207 possible accident scenarios involving outside of the main and auxiliary control room IHTs were identified for further analysis. was estimated at 0.0117 fircs/yr. This was based The 207 scenarios were grouped into one of on the fire occurrence data from nuclear power 5 categories: (1) the scenario docs not affect plants modified by zero reported fires at HFIR any safety system and docs not cause any since it went critical in 1965. initiating events; (2) the scenario directly causes The main and auxiliary control room fire an initiating event; (3) the scenario affects one occurrence frequency was estimated to be 6.55 x or more systems relative to plant safety; (4) the 10'Vyr and 5.49 x 10"7yr, respectively. These arc scenario has the potential to directly cause plant generic frequencies modified by the HFIR or core damage; and (5) the scenario has the information. potential to propagate to vital areas of the The dominant sequence initiated by a fire plant, especially the main and/or auxiliary resulted in a total loss of all AC power to the control room. HFIR (fire in the electrical building) and a A screening process to reduce the number of failure to maintain decay heal removal capability scenarios to a more manageable number for using the Auxiliary Electric Power Generators detailed analysis was implemented based on the (AEPG's). The mean frequency of this scenario following rules. was 1.08 x 10"5/yr with the total fire initiated Scenarios in category 1 were eliminated from mean core damage frequency being 1.83 x 10' s further analysis. Because of the importance of /yr. Thus 60% of the risk was a result of this the main and auxiliary control rooms to the single scenario. scram function of the reactor, all scenarios in All other internal hazard initiators were category 5 were retained for detailed analysis. analyzed in the same manner as the fire hazard. A conservative estimate of the frequency of Table 3 summarizes the internal hazard occurrence and an assessment of the potential initiators contribution to the mean frequency of plant damage state was assigned to scenarios in core damage for the HFIR. the remaining categories 2-4. For each plant damage state, the scenarios were ranked by Seismic Risk frequency. The scenario was eliminated from The seismic risk analysis consisted of 5 steps: further analysis if the scenario contribution to a (1) determine the seismic hazard for the HFIR damage state was less than 2% of the frequency site (frequency of ground motion acceleration of contribution of the internal events to that various sizes), (2) perform a fragility analysis damage state. (response of structures and/or components to After screening, 62 scenarios remained in various magnitudes of ground acceleration), (3) categories 2-4 and 16 scenarios were retained in analyze the plant response to the seismic category 5. Fires appeared to dominate the failures resulting from steps 1 and 2, (4) obtain internal hazard scenarios and are discussed in a mean (point estimate) of the core damage detail in the following paragraph. frequency and assign core damage states After the screening, the fire scenarios were resulting from a combination of steps 1-3, (5) analyzed in detail as to (1) the occurrence rate finally, perform an uncertainty analysis for those of the fire, (2) the physical effect of the fire and scenarios found to be dominant contributors to (3) the response of the plant to the fire. the seismic risk. The frequency of the plant damage state due The seismic hazard curves were obtained by to a fire is a multiple of 5 factors: (1) the combining the site specific curves generated by annual frequency of a fire in room (Z), (2) the the Electric Power Research Institute (EPR1) fraction of fires in a specific area (J) of room and those generated by Lawrence Livermore (Z) (Geometry Factor), (3) fraction of fires in National Laboratory. The combination process room (Z), area (J) with severity sufficient to did not include the results of Livcrmore ground cause damage (severity factor), (4) fraction of motion expert #5, because this data set was an fires that are not suppressed before damage outlier by those performing the analysis. The occurs to equipment (non-suppression factor) sensitivity of the HFIR core damage frequency and (5) the conditional frequency of reaching to this assumption will be addressed later in this damage state (X) due to failure of equipment paper. The statistical combination of the two

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

390 current ha/ard curves was performed by Risk The wind hazard curves were generated by Engineering Inc., the results arc portrayed in EQE Engineering, Inc. Because of the large Figure 5. distances to the seacoast, hurricanes were The seismic fragilities for the HFIR were excluded from the data sources. The Rcinhold- generated by EQE Engineering Inc. The Ellingwood model was used for tornado hazards components and structures for which fragilities with variability. The variability included (1) the were generated were based on the results from uncertainty in plant site (2) uncertainly in the the internal events PRA. Earthquake tornado data (frequency of occurrence, area, and characteristics, system damping, load path length), and (3) uncertainty in tornado combinations, model combination, combination damage area and length. Extratropical winds of responses to earthquake directional were derived from data from the Knoxville components, and structural modeling Weather Services and combined with the considerations were all included in the tornado frequency curves. The results for development of the fragilities. The failure tornados and extratropical wind hazard are criteria for the majority of components was shown in Figure 7. defined as a failure of the component to The wind fragilities were also generated by perform its design function and for structures, EQE Engineering Inc. The structures that were the failure criteria was defined as inelastic considered in the analysis included: (1) major deformation sufficient to interfere with the elements of the reactor building and the control opcrability of equipment. Table 4 show the water wing; (2) the electrical building, and (3) fragilities for the components and structures the exhaust stack. The fragilities arc shown in which were used in the PRA. Table 5. The plant logic and assembly of seismic Potential tornado missile damage was hazard and fragility information used current assessed for the reactor building high bay and accepted practices and is shown schematically in the control/water wing. The reactor high bay Figure 6. was protected by the thick concrete walls from The dominant seismic initialcd-scqucnce penetration. The first floor frequency of involves a total loss of AC power to the HFIR penetration was less than 5.7 x 10'5/yr, the and a failure to obtain and connect at least one control room level less than 7.8 x 10"5/yr and the of the two AEPG's within the 6 hours of reactor building roof less than 5.8 x lO'5/yr. battery lifetime. With the loss of the pumps to Because of the location of internal barriers, circulate the coolant, the decay heat cannot be presentation area, and restricted angles, the core removed resulting in core damage. The mean damage frequency from missiles was less than frequency of this scenario is 4.92 x 10s/yr. The lO^/yr. With such a low missile damage total seismic initiated core damage frequency is frequency, the plant logic models are initiated 1.55 x lOtyr. For the majority of the by failures associated with pressure changes due sequences, the reactor primary system is intact to the wind. Loss of all on-site AC power is and/or the pool is available for scrubbing fission assumed for all wind scenarios due to the low products. However, one dominant sequence wind velocity damage threshold of the electrical docs involve a seismically induced loss of building. coolans accident with loss of pool integrity with The dominant wind initiated sequence is a mean frequency of 9.65 x 10 */yr. damage to the reactor building roof, pieces of which fall into the reactor pool and cause High Winds and Tornados primary boundary damage and leading to core The analysis of the effects of high winds and damage (large LOCA). The second dominant tornados on the HFIR follows the same wind initiated sequence is a loss of all AC approach as for seismic analysis. The steps arc: power and cither a failure of the batteries (due (1) create tornado/wind hazard curves (frequency to collapse of one of the interior walls) or of wind events at various velocities), (2) perform failure to transport one of the two AEPG's to a fragility analysis, (3) perform a tornado the site and connect to the sysiem before the missile analysis, (4) combine steps 1-3 with the battery lifetime of 6 hours is exceeded. This plant logic and obtain an estimate of core scenario leads to a loss of decay heat removal damage frequency and plant damage states, and capability. The total wind mean core damage (5) perform an uncertainty analysis. frequency is 3.] x lOVyr. For the wind

~~Second DOK Natural Phenomena Hazards Mitigation Conference - 1989~~

~~391 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 Peak Acceleration (g) Figure 5. Aggregate Seismic Hazard Curves [EPRI and LLNL (w/o Expert #5)] for the HFIR Site~~

~~1.0~~

~~1.0 f MQUENCY ros pnoiAD ILITY .1 ,J .3. . . .10 AOOTO.Ol fAIOUCNClCS JF • HOMOTM* «O COHTHI1UTORI~~

~~LOGIC/FnACIUTY FIANTICVCI FACILITY MATRIX FnACILITY FOA CACHFOI ANALYSIS Of AGGREGATION FOA EACH rot KEY STRUCTURES, COMPONENTS~~

~~Figure 6. HFIR Seismic and Tornado/Wind PRA Analysis Approach~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~392 Table 4. Sicsmic Fragility Summary~~

Item Number Component a(g) ft Pu Pc HCLPF (g) 1 Pool Caution Exchanger 0.11 0.29 0.20 0.35 0.05 2 Pool Cleanup Filter .15 .29 .20 .35 .07 3 Electrical Building .16 .38 .45 .59 .04 4 Vent Stack Inner Liner .20 0.25 0.30 .39 .08 5 Pool Demineralizer .23 .29 .20 .35 .10 6 Control Room Ceiling .25 .25 .30 .39 .10 7 CHOG/SBHE MCC Block Wall .25 .25 .30 .39 .10 8 Offsite Power .31 .25 .43 .50 .10 9 Pool Anion Exchanger .35 .29 .20 .35 .16 10 Reactor Building High Bay Leakage .38 .26 .22 .34 .17 11 Reactor Building Control Bay Masonry Walls .40 .26 .30 .40 .16 12 Water Wing Masonry Walls .40 .26 .30 .40 .16 13 Instrument Air .40 .26 .30 .4 .16 14 Reactor Building Control Bay Roof .48 .27 .40 .48 .16 15 Reactor Building Observation Window .48 .27 .40 .48 .16 16 Pony Motor Batteries .54 .29 .19 .35 .24 17 Pool Deaerator .77 .26 .20 .33 .36 18 Vent Stack Outer Stall .83 .45 .44 .63 .19 19 Reactor Vessel Supports .88 .32 .35 .47 .29 20 Pony Motor Battery Room Masonry Walls 1.0 0.26 0.30 0.40 0.40 21 Primary Pump Support 1.1 0.33 0.33 0.45 0.36 22 PrecondenserC6-13 1.1 0.31 0.57 0.65 0.25 23 After Condenser C6-C 1.1 0.31 0.47 0.56 0.30 24 Retrofit Block Walls (primary shielding) 1.1 0.27 0.46 0.53 0.33 25 Reactor Building High Bay Frame 1.2 0.30 0.35 0.46 0.41 16 Control Rod Drives 1.2 0.20 0.35 0.40 0.48 27 Pool Demineralizer after Filter 1.2 0.29 0.20 0.35 0.54 28 Pressurizer Pumps 1.3 0.20 0.40 0.45 0.48 29 Reactor Building Substructures 1.5 0.30 0.35 0.46 0.51 30 Emergency Depressurization System Valves 1.5 0.20 0.50 0.54 0.47 31 CHOG/SBHE Steel Shed 1.7 0.35 0.40 0.53 0.50 32 Piping 1.8 0.30 0.50 0.58 0.48

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~393 10 "S~~

~~10 "S J1 s 10 -'4 I 10-J «g 10 g xi a * 10-s C8 3 1 o.i~~

~~-8 _ 10 ~i i I I I i I I i [ I I rn i i i r l j T I I F I I r ~ir[TirriiTi i 0.00 100.00 200.00 300.00 400.00~~

~~Figure 7. Combined Tornado and Extratopical Wind Hazard for the HFIR~~

~~Table 5. Summary of Structural Wind Pressure Fragilities~~

~~Structure Failure Mode (mph) PR Pu~~

1. Control Bay/Water Wing Failure of Control Room Block Walls 99 0.06 0.16 2. Control Bay/Water Wing Failure of Other Block Walls 99 0.06 0.16 3. Reactor Building High Bay Observation Window Failure 99 0.06 0.16 4. Electrical Building Failure of East Exterior Block Wall 116 0.08 0.18 5. Vent Stack Failure of Outer Shell* 175 0.06 0.17 6. Reactor Building High Bay Roof Beam Failure 214 0.07 0.19 7. Reactor Building Failure of 4-Inch Pony Motor 226 0.08 0.19 Battery Room Block Wall" 8. Reactor Building, Control Bay/Water Wing Substrate Shear Wall Failure >300 9. Interior Shielding Wall Collapse >300

~~'Only limited probability of impact onto reactor building. "Fragility is based on proposed conceptual retrofit scheme.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

394 initiators there is always a loss of building The effect on the HFIR core damage confinement and fans, but the pool is intact and frequency by including the ground motion model therefore (here is adequate fission product of expert #5 in the seismic hazard curves is an scrubbing, but with ground level releases. increase in the seismic contribution to core damage from 1.55 x RESULTS 10"Vyr to 4.1 x 10'Vyr, thus increasing the total Table 6 summarizes the core damage core damage frequency to 7.71 x frequencies resulting from the major external 10'4/yr. An effort is under way to technically initiated events and compares this to the justify the elimination of expert #5 ground internal event initiated core damage frequency. motion model from the data base by using The total core damage (external and internal) detailed analysis of the data from the recent frequency for the HF1R is 5.16 x 10 V- Sagucnay, Canada earthquake.

CONCLUSION The HFIR PRA results for external events will be addressed as part of the overall HFIR Table 6. Summary of External Events Results Risk Management Program to ascertain if design changes arc needed. Mean Core Currently, there appears to be only one area Initiator Damage Frequency where some significant risk reduction can be achieved at a reasonable cost. The external Fire 1.83 x 10"5/year event initiated core damage can significantly be Wind 3.01 x 10'5/year reduced if one cither improves the reliability of -4 the decay heat removal power supplies or Seismic 1.55 x 10 /year 6 proves by experiment and analysis that the Other 1.81 x 10" /year HFIR core can be cooled by natural circulation after a period of forced circulation which is Subtotal - External 2.05 x 10*/year shorter than the pony motor battery lifetime, Subtotal- Internal 3.11 x 104/year thus eliminating the need for the AEPG's. Total HFIR Mean Core Damage 5.16 x 104/year The experiment/analysis program is currently underway.

~~REFERENCES~~

~~[1] "The High Flux Isotope Reactor Probabilistic Risk Assessment," ORNL/RRD/INT-36, Jan. 1988.~~

~~[2] "Reactor Safety Study, An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants," Report No. WASH-1400, U.S. Nuclear Regulatory Commission, Oct. 1975.~~

~~[3] "The High Flux Isotope Reactor Accident Analysis," ORNL 3573, 1967.~~

(4] S. Kaplan, "Expert Information" vs. "Expert Opinion" Another Approach to the Problem of Eliciting/Combining/Using Expert Opinion in PRA in proceedings of the PSA '89 International Topical Meeting Probability, Reliability, and Safety Assessment conference, pp. 593-602, 1989.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~395 N REACTOR EXTERNAL EVENTS PROBABILISTIC RISK ASSESSMENT~~

~~J. T. BAXTER, P. E. Westinghouse Hanford Company P. 0. Box 1970, HO-31 Richland, WA~~

ABSTRACT An external events probabilistic risk assessment of the N Reactor has been completed. The methods used are those currently being proposed for external events analysis in NUREG-1150. Results are presented for the external hazards that survived preliminary screening. They are earthquake, fire, and external flood. Core damage frequencies for these hazards are shown to be comparable to those for commercial pressurized water reactors. Dominant fire sequences are described and related to 10 CFR 50, Appendix R design requirements. Potential remedial measures that reduce fire core damage risk are described including modifications to fire protection systems, procedure changes, and addition of new administrative controls. Dominant seismic sequences are described. The effect of non-safety support system dependencies on seismic risk is presented.

INTRODUCTION conduct the N Reactor external The 1382 Triennial Review events risk assessment under [1] recommended that a subcontract to Westinghouse probabilistic risk assessment Hanford Company in late 1987. (PRA) should be performed on the They were chosen because of N Reactor, the only large water- their recognized expertise in cooled, graphite-moderated reactor this area. Objectives of the in operation in the United States. contract were to (1) evaluate In May 1986, a Level 1 PRA was the contribution of external initiated. After the Chernobyl events to plant core damage incident, the scope of the study frequency, (2) identify plant was expanded to include Level 2/3 modifications to reduce expected studies with treatment of external risk of operation if needed, (3) events. Results of the N Reactor furnish input for the Level 2/3 Level 1 study were published in PRA, and (4) transfer technology August 1988 [2]. Core damage of external events PRA methodology frequency was estimated as to Westinghouse Hanford Company. 6.4 E-O5/yr. Sandia National The N Reactor external events Laboratories was selected to analyses were started in January Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

396 1988 and completed in September plants as part of NUREG-115O 1989. Four types of external (draft 1989) [4] and to the hazards survived screening based N Reactor as reported here. A on a mean rejection frequency of full description of these 1 E-O6/yr. They were earthquake, procedures is given in Bohn and fire, extreme winds and tornadoes, Lambright (draft 1988) [5]. and external flooding. Detailed risk assessments were conducted FACILITY DESCRIPTION for earthquake and fire hazards, A plot plan of the N Reactor while bounding analyses were used is shown in Figure 1. The to evaluate risks from extreme buildings of principal importance winds and external flooding. are 105-N, 109-N, 181-N, 182-N This paper presents initial results and 184-N. and a brief discussion of The reactor and its cooling engineering insights gained from systems are housed in two the study. adjoining buildings, 105-N and 109-N. The reactor, including METHODOLOGY its control and trip systems, The PRA procedures used for the control room, and cable the N Reactor analyses are based spreading room are located in on the following general concepts: 105-N. Primary coolant pumps, steam generators, the 1. External events analyses are pressurizer, graphite shield based on the internal event cooling components, and other risk assessment plant system auxiliaries are located in 109-N. models and fault trees. Located within 105-N, the reactor itself consists of a 2. Systematic screening is used horizontal array of 1,003 Zircaloy to evaluate all external process tubes penetrating a 1,800 events to which the plant ton graphite cuboid approximately might be exposed and eliminate 33 ft by 33 ft at the face and 39 unimportant events. ft long. Interlocking graphite bars make up the cuboid. A 3. Evaluation of similar events composite steel and high-density ! is coordinated to avoid concrete thermal shield box duplication of effort and structure surrounds the core. minimize data-gathering Both the core and thermal shield efforts. are supported by the reactor pedestal, which is a separate 4. Computer-aided screening structure from the 105-N building. techniques and generic failure The river pump house (181-N) data are used before detailed is located at a lower elevation component failure analyses adjacent to the Columbia River. to minimize effort on failure Electrical pumps supply the analyses. circulating raw water system (CRW) during normal operation. Procedures based on these concepts Three diesel-powered, emergency have been applied (in whole or in core cooling (ECCS) low-lift part) to six power plants as part pumps are also housed in this of the U.S. NRC-sponsored structure. Unresolved Safety Issue A-45 The 182-N Building houses resolution program [3], to the five diesel-driven pumps servicing Peach Bottom and Surrey power the ECCS and confinement fog

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~397 116-N STACK '~~

Figure 1. N Reactor Plot Plan. sprays. High-pressure injection that the probability of a maximum pumps and both low- and high- wind speed of 175 mi/h is around pressure auxiliary water systems 1 E-07/yr. At higher wind speeds are also located in the basement tornado winds govern design over of 182-N. straight wind for this site. Wind design loads for the WIND BOUNDING ANALYSIS original plant design were based N Reactor was evaluated for on a maximum straight wind load risk resulting from tornadoes and of 78 mi/h at a height of 50 ft straight winds. Hurricanes were above ground level in accordance not considered because of the with the 1961 Uniform Building plant's inland location separated Code. the ocean by a high mountain Plant standards were revised in range. 1974 to include a design basis The reactor site is located tornado with a maximum speed of in region III of the U.S. NRC 175 mi/h. Critical reactor tornado risk regionalization scheme facilities have been evaluated [6]. An alternative against the design basis tornado regionalization scheme by Twisdale and were either accepted or and Dunn [7] places the plant in upgraded in the interval since region D of a four-region scheme. 1974. Thus, the site has the lowest During a design basis tornado tornado occurrence rate in the both onsite and offsite power will continental United States. Site- be lost resulting in a demand specific studies, when extrapolated for the ECCS. Critical to the N Reactor site, indicate facilities required to provide a Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

398 success path for ECCS are shown response to these concerns, plant in Figure 1. These include modifications were made. These 105-N (the building housing the included erection of new fire main control room), 182-N (high- walls between adjacent diesel lift pump house), 181-N (river pump pumps and re-routing of power and house), and 109-N (heat exchanger control cabling to provide building). The ECCS system was physical separation between ECCS upgraded and qualified for both diesel trains in the 181-N and design basis earthquake and 182-N buildings. As with previous tornado as part of a recent upgrades, the decisions were seismic safety enhancement based on qualitative "engineering program. Provision of the success judgement" rather than a path is dependent on completion quantitative decision making of several minor tornado process. resistance upgrades to buildings 181-N and 182-N before facility Plant Vulnerabilities and restart. Engineering Fixes Assuming completion of the The total fire-induced core upgrades, the tornado risk at N damage frequency computed for the Reactor is considered acceptable. N Reactor is 1.8 E-O4/yr. Results of the tornado risk Transients dominate the accident assessment support the original sequences as shown in Table 1. decision to qualify the ECCS system. Table 1: Fire accident FIRE RISK ASSESSMENT sequences with annual The N Reactor was authorized core damage frequencies in 1958 and achieved its design thermal power rating in 1964; thus, its design and construction predate many contemporary Mean Core requirements. Damage After the Browns Ferry fire Sequence Fire Area Freg./vr additional fire protection requirements were imposed on T8 109-N Access 1.3 E-04 existing commercial nuclear power Corridor and facilities. These requirements 109-NT Bsmnt. are specified in 10 CFR 50, Appendix R. In particular, Main Control 3.2 E-05 separation requirements are Room specified for independent trains of safety-related equipment. 105-N Access 4.6 E-05 Modifications have been made Corridor over the years at the N Reactor to bring portions of the safety 105-N Cable 6.6 E-07 related systems into compliance Spreading Rm. with Appendix R. Post-Chernobyl safety reviews T4 Bldg 184 Cable 1.3 E-06 identified physical separation of Runs (Plant redundant ECCS diesel pump trains Air) in buildings 181-N and 182-N as new Appendix R concerns. These concerns stemmed from walkdown Sequence T8 results from an notes by various reviewers. In early failure of the

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

399 primary/secondary cooling systems compliance with a level of (PCS/SCS) caused by a transient protection called "improved risk" initiating event and early failure as used by the insurance industry. of ECCS. Sequence T4 results from an early failure of the Cable Travs in Building 109 PCS/SCS caused by an initiating The 12 cable trays event and subsequent failure of identified in Figure 2 contain ECCS flow in one of 16 risers control and power wiring for (partial core involvement). many plant sytems. These include Over 98% of this frequency is the primary cooling system pumps attributable to a "pinch point" and high-pressure injection, in control and power wiring where actuation, and power circuits for it exits the cable spreading the ECCS system, controls for room. The general arrangement of the confinement fog spray system, this area is shown in Figure 2. the plant raw cooling water

~~CONFINEMENT BOUNDARY~~

~~DUCT BANK TO AUXILIARY BUILDINGS~~

~~109-N TURBINE 109-N STEAM GENERATOR BUILDING 105-N REACTOR ANNEX BUILDING BUILDING~~

~~Figure 2. Power and Control Wiring "Pinch Point."~~

The cable trays and conduit supply, portions of the secondary in this area are the original cooling system, and communications installation with limited circuits from the main control additions and modifications which room to control rooms in the have taken place over the years. pump houses and the local power There is no separation between plant in 184-N. control and power wiring for Both the access corridor separate trains of cooling and the Turbine Annex basement systems. area are large, open areas. All areas in Figure 2 have Cabling in this area has a existing fire protection systems. "Flamastic" coating so self- Fire protection systems at N ignited fires from "hot-shorts" Reactor are designed to provide were not considered. Transient Second DOE Natural Phenomena Hazards Mitigation Conlerence - 1989

400 combustibles are the main capability exists at the N initiators for this area. Small Reactor, the control room fires do not damage the trays in procedures are not yet approved both areas. Large fires can and in place. Assuming that the damage the trays anywhere in the procedures are implemented as corridor in 2 to 6 min, based on part of a restart effort, recovery COMPBRN simulation runs. Similar will reduce this frequency by a results were obtained in the factor of two. basement area immediately surrounding the trays. Risk Reduction Measures The cable trays are protected For the 109-N access corridor by automatic sprinklers (wet and 109-NT basement area standpipe system) below and above modifications are proposed that the trays. Lower sprinkler heads will provide for earlier detection are spaced approximately 10 ft and suppression because of the apart, centered below the middle very short time to damage. Early tray. The upper run is centered warning smoke detectors installed over the middle tray run with in accordance with National Fire heads on 20-ft centers. All Protection Association standards sprinklers are equipped with and tied into the existing control fusible link heads. Suppression panels and alarm system would credit was given for the provide adequate detection. sprinklers below the trays if Alternate automatic actuation any point of the fire pool was circuitry for the existing wet within 1 meter radially of a standpipe sprinkler system may sprinkler head. This represents be considered to improve 50% of the floor area immediately suppression times. Additional below the trays. Fires at other fire protection blankets for locations damage the trays and passive protection may be an wiring before the sprinklers alternative to sprinkler respond. modifications. Similar arguments apply to The access corridor in those cable tray sections passing 105-N is a dead-end, low-traffic through the building 105-N access area. Administrative controls corridor and in the immediate area may be implemented to make this surrounding the cable tray area a flammable-free area. penetrations of the cable This would imply a full-time spreading room. Short times to fire watch for future plant damage indicate a need for better maintenance in this area. fire detection and suppression in Assuming that these these areas. modifications are properly implemented, the mean frequencies Main Control Room for the two most dominant Control rooms at the N Reactor sequences would be reduced by a are continually manned and factor of from 10 to 20. Coupled equipped with adequately designed with development of additional and maintained halon systems. control room procedures, the Scenarios are based on suppression overall fire risk core damage of 99 out of 100 fires before frequency can be easily reduced control wiring is damaged. The to values in the neighborhood of remaining fire scenario assumes 5 E-05/yr; comparable to control room abandonment because commercial nuclear power plants. of smoke from a cabinet fire. Although remote shut-down

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

401 SEISMIC RISK ASSESSMENT to least least severe. Events in the hierarchy above an Background initiating event are precluded A seismic upgrade program from the accident sequence for has been conducted at the N that event. The sum of initiating Reactor in preparation for restart event probabilities must be one at the same time as the external for each increment of the seismic events PRA. Elements of the hazard curve. upgrade program were developed to respond to post-Chernobyl review Dominant Accident Sequences comments on the seismic Initial seismic screening qualification of the N Reactor reduced the number of sequences ECCS and graphite shield cooling to 13. Subsequent quantification systems (GSCS). Commitments were with best estimate random failures made to qualify the ECCS cooling means and best estimate seismic system and the necessary supporting fragilities and responses systems for seismic and wind identified five dominant environmental conditions (i.e., sequences: safe shutdown earthquake and design basis tornado). This o Building 182 (45%) resulted in a $24 million upgrade o LOSP T8 (21%) program with 212 category I fixes, o PTR-1 (19%) analysis of 1480 seismic III/I o SLOCA T8 (10%) problems, and 530 III/I fixes. o TRANS T8 (2%) Walkdowns and plant modifications stemming from the Percentage contributions are based seismic upgrades program have on a Monte Carlo uncertainty influenced the PRA results. There analysis. The total mean core are no instances of localized damage frequency was determined failure in non-safety class to be 6.7 E-05/yr. The range equipment or building partitions factor (defined as the ratio of (seismic III/I failures) that would the 95th percentile to the 50th lead to dominant core damage percentile of the distribution) sequences. on total core damage frequency was found to be 35. Initiating Events Seismic risk assessment of Description of Accident Sequences the N Reactor is based on the same N Reactor has three cooling event trees developed for the systems available to mitigate internal events analysis of the accidents. System G (PCS/SCS), plant. Initiating events System A (ECCS), and System C considered include the following: (GSCS). The dominant accident o Process tube rupture (ECCS sequence results from the failure ineffective) of building 182. Building failure o Large LOCA leads to accident sequence GA-C. o Small LOCA No other failures are required to o Building 182 failure cause this transient accident o Transient Type 1 (LOSP) sequence. Building collapse is o General transient (PCS assumed to fail all PCS/SCS initially available) support pumps and all of the ECCS high-lift diesel pumps. The initiating events are listed GSCS does not fail because its in a hierarchy from most severe pumps are located elsewhere.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

402 The next most important o 13.8-kV Bus 5.7% sequence is process tube rupture. o 4-kV Bus 11.9% This arises at high ground-motion o Primary shield input levels (0.75 g) when the wall 1.4% primary shield wall of the reactor 0 Ceramic starts to fail. This failure insulators 1.3% takes the form of uplift and o Building 105 0.4% rocking of the primary shield wall, which leads to shear failure Building 182 and the ECCS of the process tubes at the inlet silo structure were both and outlet face of the primary seismically upgraded to qualify shield. No other failures are for a 0.25-g SSE earthquake in required to cause this accident recent restart programs. This sequence. This sequence is similar has influenced the results of to reactor vessel rupture sequences the current external events reported in NUREG-1150. analysis. The next three sequences in order of importance are TRANS T8, REFERENCES SLOCA T8, and LOSP T8. All result from the same logical combination [1] H.J. Kouts and M.C. Leverett, of cutsets and component failures. "Report of Three-Year Review In each case, the sequences of UNC Nuclear Safety Review involve GA-C system failures. System," Brookhaven National The dominant cutsets consist of Laboratory, Upton, New York, failure of the ECCS silo 1982. structure, which fails the ECCS system in combination with [2] M.D. Zentner et al., "N electrical bus failures (both 13.8 kV and 4 kV) which fail Reactor Level 1 Probabilistic PCS/SCS. As before, GSCS is not Risk Assessment," WHC-SP- failed. The PCS/SCS failures are 0087, Westinghouse Hanford caused by dependency of instrument Company, Richland, air on 4-kV busses. Washington, August 1988. [3] NRC Task Action Plan (Rev. Basic Event Importance 2), "Shutdown Decay Heat The mean seismic core damage Removal Requirements (Task frequency for the N Reactor is A-45)r" draft program comparable to that for two description issued by Generic commercial plants, Surry and Peach Issues Branch, U.S. Nuclear Bottom, as reported in NUREG-1150 Regulatory Commission. (draft 1989) [4]. Results of the analysis have not been available [4] U.S. NRC, "Severe Accident for sufficient time to evaluate Risks: An Assessment for potential risk reduction measures. Five U.S. Nuclear Power Basic event importance to the Plants," draft, Office of mean values was determined by Nuclear Regulatory Research, setting the seismic failure U.S. Nuclear Ragulatory probability to zero for each Commission, Washington, DC, component and recalculating the NUREG-1150, vol. 1, June mean point estimate. Risk 1989. reduction potentials are:

~~o ECCS silo 37.6% o Building 182 19.4% Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

403 [5] M.P. Bohn and J.A. Lambright, "Recommended Procedures for Simplified External Event Risk Analyses," (draft) Sandia National Laboratories, Albuquerque, NM, NUREG/ CR-4840, February 1988. [6] E.J. Markee et al., "Technical Basis for Interim Regional Tornado Criteria," WASH-1300, U.S. Government Printing Office, Washington, May 1974. [7] L.A. Twisdale and W.L. Dunn, "Probabilistic Analysis of Tornado Wind Risks," Journal of the Structural Divisionf ASCE, February 1983.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 404 USE OF A PHASED APPROACH TO EVALUATE THE SEISMIC PROBABILISTIC RISK ASSESSMENT FOR PRODUCTION REACTORS AT THE SAVANNAH RIVER SITE~~

~~H E.Wingo Westinghouse Savannah River Company Savannah River Laboratory Building 773-41A Aiken, S. C.29808~~

The phased approach used to evaluate seismic risks for the production reactors at the Savannah River Site was designed to use all of the classical components of a full scale probabilistic risk assessment, but made extensive use of existing analyses and engineering judgment as a first approach to determine the magnitude of the risk, and if a more detailed analysis was needed. The assessment produced a quantification of the risk, an understanding of reactor systems contributing to the risk, and indicated a more detailed evaluation was needed, which is underway.

INTRODUCTION ground shaking of Intensity VII (Modified Mercalli Intensity) at the site. A probabilistic seismic hazard Seismic studies and upgrades for reactors at the analysis was chosen for application to the site. The Savannah River Site have been underway since the primary hazard analysis selected was that based on late 1960*s, but an overall quantification of the risk of the work done by the Lawrence Livermore National operation was not available, and the major Laboratory as reported in reference 1 because of the contributors to that risk were unknown. Therefore, a availability of this evaluation for the nearby seismic probabilistic risk evaluation was undertaken VOGTLE nuclear Power plant. to understand the reactor systems contributing most to the risk and to quantify the risk. The assessment The impetus for this particular study of seismic consisted of the following components: hazard was to develop a methodology for application to all nuclear power plant sites East of the Rocky Mountains to assist in assessing implications of the • Seismic hazard evaluation USGS position on the Charleston earthquake. An • Plant familiarity additional impetus was to provide a simplified • Equipment fragility analysis method and data for computing the seismic hazard at • Plant performance modeling any site located east of the Rocky Mountains in a form • Seismic hazard/equipment fragility interface suitable for use in probabilistic risk assessments modeling (PRA). Expert judgment was used extensively in this • Risk assessment hazard analysis because of the sparsity of seismicity data and ground motion attenuation in the Eastern SEISMIC HAZARD EVALUATION U. S. This hazard assessment was selected for use although no other seismic PRA was discovered that The seismic hazard evaluation provided an used this specific hazard evaluation. estimate of the frequency of damaging ground It was assumed that little or no difference existed shaking that could be experienced at the site The between the seismic hazard for the Savannah River Savannah River Site is located 160 km from Site and the VOGTLE site, which are located within Charleston, and the largest known earthquake to 30 miles, 50 km of each other, and in the same occur in the vicinity was the Charleston earthquake Atlantic Coastal Plain. of 1886 which has been estimated to have caused

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

405 The seismic haiard results for the Vogtle power plant percentile curves followed a lognormal distribution. are shown in Figure 1, with results being expressed From the lognormal distribution at each acceleration, in terms of the 15th, 50*h, and 85lh percentile of 20 Latin Hypercube random samples were taken to probability of exceedance per year Uncertainty in calculate the mean frequency The lognormal the seismic hazard assessment is indicated by the distribution was truncated at ± 3 standard deviations spread in these percentile curves. in the calculations.

~~10" 100 200 300 400 500 600 700 BOO 900~~

~~Peak Ground Acceleration (cm lite /ire) Peak Ground A«tl«f»tK>n,~~

~~FIGURE 1 FIGURE 3 SEISMIC HAZARD FOR VOGTLE MEAN SEISMIC HAZARD \PPUED TO SAVANNAH RIVER REACTORS FOR SAVANNAH RIVER REACTORS FROM VOGTLE RESULTS~~

An historic seismic hazard analysis also was made, as shown in reference 2, to provide a direct The mean frequency of occurrence at each comparison of actual ground snaking calculated to acceleration level was developed as shown in Figure 2 have occurred at the site versus that as predicted from from these 15th, 50th, and 85th percentile reference 1. distributions, based on the assumption that the

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~406 Steps used to develop this historic seismic hazard are shown in Figure 3.~~

~~LogV Attenuation Modal A Ground Motion O.DIsparsion Intensity~~

~~Historic Catalog Of Log Distance Earthquakes~~

~~Data, Location, l0 or~~

~~Data, Location, lo or~~

~~Log P~~

~~,-1 Dlvldt by 10 Number of Yaara i-2 Function Proportional to 10 Historical Exceedanca Rat* 10"3 10"4 Nonparametrte Function~~

.01g .tg Log Y •^-Distribution of Log Historical Sita lnttn*iti«s Hazard Function Historic Cxcaadanca Rata FIGURE 3 HISTORIC SEISMICITY ANALYSIS STEPS Two estimates were made of the historic seismicity. The first used the input provided in reference 1 The second included a correction for the Each earthquake in the historical catalog of period of completeness of the earthquake catalog, earthquakes was attenuated to the site and the selection of attenuation models developed specifically frequency of occurrence of each event was assessed for South Carolina, and consideration of alternative By incorporating alternative earthquake catalogs Modified Mercalli Intensity-magnitude conversions and attenuation models, the uncertainty in the Use of two approaches eliminated the need to know or historic seismidty analysis was estimated, and which assume the shape of the frequency distribution of resulted in a probability distribution of the frequency actual earthquakes. of peak ground accelerations.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

407 The second set of results gave the preferred estimate PLANT FAMILIARIZATION of the historic seismicity at the Savannah River Site A familiarization study was conducted for and is shown in Figure 4 along with that from consultants used in evaluation of the seismic reference 1. probabilistic risk assessment. This familiarization

U-00 r was necessary because of the unique design of the Savannah River production reactors, which made insights and assumptions from other reactors inappropriate. A schematic of the reactors and their cooling system is shown in Figure 5. This IE-OU familiarization study determined the insitu condition of the plant and gave firsthand knowledge of the plant

>0 equipment and layout, and included on site inspections, interviews, reviews of plant drawings, and reviews of seismic studies previously done. The it-ora list of items important to safety that were evaluated was supplied to the consultants based on judgment of personnel familiar with the reactor systems. The list i of items was generated in an all encompassing mode in an attempt to support an expanded and inclusive 3 IE-904 seismic modeling effort. Thus some items were evaluated that later were shown not to be needed in c the assessment. 1 EQUIPMENT FRAGILITY ANALYSIS it-oes The equipment fragility analysis used the DELPHI method to determine the probability of failure of items important to reactor safety as the result of seismic events. Seismic capacity of individual components was estimated by a team of lE-DOi. seismic experts acting independently and their .9 I individual judgments were combined to define a Peak Ground Acceleration, g seismic status and a seismic capacity for each item evaluated, as shown in reference 4. FIGURE 4 HISTORIC SEISMICITY COMPARED TO PREDICTION FOR VOGTLE (REF 1)

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

408 mammm*

~~KOTONMEll~~

~~KHOKM~~

~~FIGURE 5 SAVANNAH RIVER REACTOR SCHEMATIC~~

~~PERCENT Of CATEGORY DEFINITION ITEMS INSPECTED~~

A STRENGTH ADEQUATE I NO FURTHER 44 The analysis provided a categorization of the MODIFICATIONS NEEDED seismic strength of equipment, structures, and e STRENGTH PROBABLY ADEQUATE BUT FURTHER 37 support facilities as shown in Table 1. The ANALYSIS NEEDED c STRENGTH PROBABLY NOT ADEQUATE BUT 4 categorization indicated the overall assessment of the FURTHER ANALYSIS NEEDCD condition of the plant and provided a listing of D HARDWARE MODIFICATION NEEDED 14 E HARDWARE MODIFICATION NEEDED TO ADIACENT 1 equipment that was obviously weak and in need of EQUIPMENT upgrading against seismic challenges even before the TAIII.K I seismic probabilistic risk assessment was completed SEISMIC CATEGORY OF EQUIPMENT KHOM DELPHI K VALUATION

~~Second DOIi Natural Phenomena Hazards Mitigation Conference - 1989~~

409 Additionally, the evaluation produced directly for as indicated by the few examples in Table 2. Nearby use in a probabilistic risk assessment, the seismic non-safety related equipment that could impact capacity for each item evaluated, at the 10th, 50th, safety functions was also identified and evaluated and 90th percent probability of failure for seismically when needed to determine the capacity of safety induced accelerations applied horizontally and related equipment. vertically to the center of gravity of the equipment or structure. Values were defined for functional failure and for failure to meet requirements of design codes,

~~INDIVIDUAL TEAM MEMBER ESTIMATES, G'S MEAN STANDARD DEVIATION~~

~~FUND Surv Based on Code Limits Based on Functional Failure SYS FREQ. Prob REMARKS ID HZ S Expert i Expert 2 Expert 3 Expert) Expert 2 Experts Code Function Code Function~~

~~H V H V H V H V H V H V H V H V H V H V~~

~~1 10-15 90 0.4 OS 0.4 * 1.0 2.0 1.2 0.4 0.4 1.2 1.5 0.2 o.s SO - 0.6 1.0 1.0 • 1.5 3.0 3.0 1.0 0.8 30 2.2 - 02 .75 10 0.8 1.2 2.6 2.0 4.0 8.0 2.6 1.0 8.0 3.0 0.2 1.0~~

~~ISd 5-10 90 0.4 0.2 04 0.8 0.4 0.9 il l 0.4 0.2 0.8 0.4 .02 .04 SO 0.5 0.4 08 1.0 0.8 • 1.6 0.6 0.4 1.3 0.8 .15 • .30 • 10 0.6 06 1.4 m 1.2 1.0 2.9 1.0 .0.6 2.0 1.0 .42 .80~~

~~19b 25 90 06 0.4 0.4 1.0 0.6 IS 0.9 0.7 0.4 1.1 0.6 0.2 0.3 SO 0.8 0.5 1.0 1.5 0.8 2.5 - 2.0 • 1.1 0.5 2.0 0.8 0.3 0.4 • ': ; 10 1.0 07 i n b 2.2 2.0 1.0 35 4.4 1.9 0.7 3.3 1.0 0.7 1.0~~

18c 5 90 .25 .20 .10 .40 .75 .50 .10 .80 .25 .20 55 .50 .10 .32 SO 35 .40 25 - .70 1.5 .70 .25 1.8 .43 40 1.2 .70 .18 .65 10 SO .50 .75 1.4 ! 2.0 1.0 .75 3.8 .88 .50 2.2 1.0 .40 1.25 ; 1* 5-10 90 2.5 2.0 5.0 136 5.0 3.0 7.0 2.71 2.95 20 4.90 3.0 1.53 1.76 AKumet 50 30 3.0 60 277 - 7.0 4.0 9.0 - 5.54 392 3.0 7.18 4.0 1.47 1.42 - duct over 10 3.5 4.0 8.0 - 5.66 9.0 5.0 11.0 11.3 - 5.72 4.0 10.4 50 1.84 - 1.03 com- preiior doesn't govern

~~TABLE 2 SUMMARY OF SELECTED LIMITING ACCELERATION VALUES, g APPLIED AT CENTER OF GRAVITY OF COMPONENT~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

410 PLANT PERFORMANCE MODELING These floor response spectra were generated for A logic model was developed to describe plant use in design and thus reflected conservative responses and event sequences that could result in assumptions. In a seismic probabilistic risk core damage when initiated by earthquakes. assessment, the median capacity of a component, Documents were reviewed to gain an understand of using the lognormal model and the assumed the interplay among various systems used to operate multiplicative property of the process, is to be used the reactors and to protect them from upsets. The based on the median values of the underlying modeling development used results of the fragility parameters. Hence factoring out conservatisms in the analysis in an iterative fashion to concentrate on floor response model was necessary. items of equipment and structures that were subject Fragility values for some components that were to failure from seismic challenges and contributed to hot evaluated by the DELPHI method were the risk of core damage. Models took the form of determined using an approximate analysis. In all event trees and fault trees. Models were generated on cases the fragility values were compared to published the assumption that seismic upgrades were values for a reasonableness check. Values used in the completed for items previously identified as obviously analysis are shown in Table 3. in need of repair. The seismic fragility values generated by the DELPHI method, which were in terms of COMPONENT FRAGILITY, accelerations for failure applied at the center of HORIZONTAL GROUND gravity of the components, were converted to values EQUIPMENT CAPACITY of acceleration at ground level by use of floor response spectra generated for the reactor building as Offsite Power 0.20 0 35 indicated in reference 5. The local capacity of Control Room Catling 1.M 0*3 equipment was combined with the calculated Safety Rod System 0.73 0.73 structure response to estimate mean seismic fragility safety Rod System Support 0.7* 0.73 curves in terms of shaking at ground level. In this Supplementary shutdown system »M 0*3 fashion accelerations capable of failing equipment Cooling water piping 1.H 0.53 throughout the plant were expressed on a common ECS Piping in Heat Exchanger lay 0.71 O.K basis. ECS Piping in Emergency Pump Room 1.70 0.S7 ECS block Valve 4.7S 0.70 Relay Chatter, * IS' Elevation 0.79 1.S7 Relay Chatter, 0' Elevation 094 1.S7 Pump Seal Pressure Tank 1.21 0*7 Reactor Top Shield 1.4t 0*4 Reactor Bottom Shield 1.4« 0*5 Reactor Effluent lines 1.40 O.tt Main Heat Exchangers 2*5 0.49 Emergency power Diesel 931 0 52 Safety Computer 0.73 0*3

~~TABLE3 COMPONENT FRAGILITY VALUES USED IN RISK ASSESSMENT~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

411 SEISMIC HAZARD/EQUIPMENT FRAGILITY response frequency, ductility, size of damaging INTERFACE MODELING earthquakes that can occur, the choice of a site- An interface analysis was developed to provide response spectrum, and the duration of ground compatibility between the seismic hazard and the shaking. Table 4 shows the interface approach used fragility parts of the probabilistic risk assessment. to make use of data available for the analysis and to Factors involved included the choice of appropriate yield appropriate conservatisms. parameters to characterize the seismic hazard, the

ATTRIBUTE USF. COMPONENTS Response frequency Frequency range from DELPHI analysis Damping characteristics Five percent assumed for equipment at failure Energy absorbtion capability Included in judgment of failure acceleration Failure Mode Given from DELPHI analysis SEISMIC HAZARD Earthquake Magnitude Not included Strong motion duration Not included Site-Response characteristics Broad band ground response spectrum shape Ground motion parameter Peak ground acceleration

~~TABLE4 HAZARD/FRAGILITY INTERFACE PARAMETERS~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

412 RISK ASSESSMENT Therefore a more robust evaluation of the seismic risk To quantify the seismic risk, the seismic hazard, has been undertaken as shown below: systems analysis, and fragility evaluations were combined. The quantification process was performed • Obtain improved estimates for the seiimic in two steps. First, a sequence level fragility was hazard at the Savannah River Site. (The seismic determined by incorporating the component fragility hazard developed by the EPR1 has been selected). data into probability expressions as dictated by the • Evaluate behavior of the site soils during large system models. This was done for all sequences and earthquakes. (Additional soil borings and the results added to determine the total or plant level analysis are being made. ) core damage fragility. This plant level core damage • Evaluate in detail, spectral response of the fragility provided a composite measure of the reactor buildings over the range of earthquake capacity of the plant when exposed to seismic events. magnitudes needed. (Building model response is Plots of the plant level core damage fragility curves being evaluated using actual earthquake input.} are shown in Figure 6. In the second step, the • Evaluate underground piping behavior during sequence fragility curves were integrated with a earthquakes. (Probabilistic assessments have mean seismic hazard curve to estimate the mean been made, including soil liquefaction and nil frequency of core damage. The total risk of core ductility considerations.) damage was calculated in reference 5 to be 1.85x10-5 • Generate improved system models as required per reactor year when using the historic seismicity, to develop a more detailed logic model for plant and l,56x 10-* per reactor year when using the seismic response. (Heavy reliance is being placed seismicity of reference 1. on the completed internal events study to The analysis indicated the risk of reactor generate expanded models,) operation when exposed to seismic events was not insignificant. References: 1) Bernreuter, D. L., J. B. Savy, R. W. Mensing, J. C. Chen, B. C. Davis, Seismic Hazard Characterization 100% of the Eastern United States. Volume 1: Methodology and Results for Ten Sites. Lawrence Livermore National Laboratory, UCID-20421 Vol. 1, April, 1985. 2) McCann, Jr., M. W., Historical Seismicitv Analysis ao% SAVANNAH RIVER for the Savannah River Plant. Jack R. Benjamin and REACTORS Associates, Inc. JRB153-010-02, Prepared for E. I. DuPont de Nemours and Company, December, 1985. 3) Stevenson J. D., W. Djordjevic, C. A. Kircher, DELPHI Evaluation of Seismic Resistance of L Reactor Mechanical and Electrical Components and Distribution Systems. Stevenson & Associates Inc, 1 40H 84C1257-1, February, 1985. 4) Tang, C. C. Seismic Analysis of "L" Area Buildings at the Savannah River Plant Site. QCAD-1-85-016, Prepared for E. I. DuPont de Nemours and Company, November, 1985. 5) McCann, Jr. M. W. and J. W. Reed, Preliminary Probabilistic Seismic Risk Assessment of L-Reactor Savannah River Plant. Jack R. Benjamin and Associates, Inc. JRB 153-010-01, Prepared for E. I.

~~PEAK GROUND ACCELERATION * DESIGN BASIS EARTHQUAKE DuPont de Nemours and Company, December, 1985. FIGURE 6 PLANT LEVEL CORE DAMAGE FRAGILITY, SAVANNAH RIVER REACTORS. ZION AND OCONEE POWER REACTORS~~

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~413 PROBABILISTIC RISK ASSESSMENT OF EARTHQUAKES AT THE ROCKY FLATS PLANT AND SUBSEQUENT UPGRADE TO REDUCE RISK~~

~~S. A. Day Rockwell International, Rocky Flats Plant P. O. Box 464, T886A Golden, CO 80402-0464~~

~~ABSTRACT~~

An analysis to determine the risk associated with earthquakes at the Rocky Flats Plant was performed. Seismic analyses and structural evaluations were used to postulate building and equipment damage and radiological releases to the environment from various magnitudes of earthquakes. Dispersion modeling and dose assessment to the public were then calculated. The frequency of occurrence of various magnitudes of earthquakes were determined from the Department of Energy Natural Phenomena Hazards Modeling Project. Risk to the public was probabilistically assessed for each magnitude of earthquake and for overall seismic risk. Based on the results of this Probabilistic Risk Assessment and a cost/benefit analysis, seismic upgrades are being implemented for several plutonium-handling facilities for the purpose of risk reduction.

INTRODUCTION The Golden fault was determined not to be a Probabilistic Risk Assessments (PRAs) were capable structure according to the U.S. Nuclear performed for natural phenomena at Rocky Flats Regulatory Commission criteria. No evidence was Plant (RFP). Earthquakes, tornados, high winds, found to indicate that the graben is technically or lightning, meteorites, snow loadings, and floods structurally related to the Golden fault. No other were analyzed. In conducting PRAs only credible faults were found and none of the features hazards (i.e., frequency of occurrence >10"6/year) identified present a seismic hazard. The 1882 are considered for consequence development. At earthquake is thought to have originated in RFP, only earthquakes, high winds, and tornados northwestern Colorado, at a location north of the are considered credible hazards that could cause a town of Rifle, Colorado, with an estimated release of hazardous materials. The objective of magnitude of 6.5. this paper is to address the seismic analysis. The TERA Corporation researched the regional seismic history and used that data to SEISMIC ANALYSIS develop a seismic hazard curve in a study Regional, local and historical seismic analyses performed for the Department of Energy (DOE) of the RFP were conducted by Dames and Moore called the DOE Natural Phenomena Hazards (I], TERA Corporation [2], and URS/Blume and Modeling Project. The seismic hazard curve is Associates [3]. presented in Figure 1. (Note that the ordinate Dames and Moore investigated the capability subdecade markings apply only to the return of the Golden fault and the tectonic significance of period.) This curve is used to determine the a graben located north of Clear Creek, the seismic frequency of occurrence of varying magnitude significance of features near Rocky Flats Plant earthquakes. suspected to be faults, and the 1882 Colorado These studies were then used to determine the earthquake. design criteria for RFP.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~414 10~~

~~/ / / / / t/ t' A f-f-i 1 f / / f f I~~

~~1O1 \!\L -I 10~~

~~10' 10 100 200 300 PEAK ACCELERATION~~

* OMlgn Baal* Earthquake

~~Figure 1. Earthquake Hazard at Rocky Flats Plant~~

John A. Blume and Associates, Engineers, to determine the Operating Basis Earthquake evaluated site and local geology, dynamic soil (OBE) and the Design Bases Earthquake (DBE). properties at the site, and regional seismic history The design criteria is shown in Table I.

~~Epicentral Peak Horizontal Earthquake Magnitude Distance. Miles Acceleration, g's~~

~~OBE 5.6 16 0.07 DBE 6.0 16 0.14~~

~~Table 1. Earthquake Design Criteria~~

STRUCTURAL EVALUATION their associated support facilities at RFP. Several Rocky Flats Plant structures were evaluated by of the buildings were determined by conservative Agbabian and Associates [4], URS/BIume and finite element computer modeling to have incipient Associates [5] [61, and by Los Alamos Technical failures occurring at peak horizontal accelerations Associates (LATA) [7J. ranging from 0.04 g to 0.10 g. Agbabian and Associates studied the structural URS/BIume and Associates evaluated vital resistance of the plutonium-handling buildings and equipment at the plant for performance during

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

415 natural phenomena events. Some of the equipment building during the seismic event. This value is tested was expected to malfunction during the calculated as a function of the inventory, the DBE. Some of the findings for electrical residence time of the inventory (the amount of time equipment were inconclusive, so shaking-table inventory is available), the physical and chemical testing was performed. This study concluded that form of the material, and the process being nearly all of the electrical equipment at RFP is performed. capable of performing properly when subjected to A damage ratio is applied to the material-at- the postulated earthquake if its devices are risk and is based on the percentage of gloveboxes anchored properly inside cabinets. that are expected to be breached during the LATA performed a realistic evaluation of earthquake. Release fractions as a function of the consequences from natural phenomena events for form of material are then applied to determine the the plutonium-handling buildings at RFP. initial source term. All the initial source terms for Threshold and total damage seismic levels were the area are summed. Resuspension factors are determined for each plutonium-handling building applied to the initial source terms to account for and its associated auxiliary buildings. The the resuspension of liquids and powders over time threshold seismic value gives the magnitude of after the earthquake. These factors are l0"10/sec earthquake at which hazardous materials will begin for powders and lO*11/sec for liquids. The to be released from the building. The total seismic duration of a release can be hours or days, based on value gives the magnitude of earthquake at which the amount of damage; rescue efforts can prolong the building will be rendered inhabitable and must and contribute to the release. be demolished. LATA defined a realistic At RFP, three analyses were performed; (1) assessment for the progressive failure of composite instantaneous release, (2) the first 24 hours, and (3) systems consisting of the building structure, the next 24 hours. mechanical and electrical systems, and other vital A building source term is an estimate of the equipment elements for the design basis earthquake release of hazardous or radioactive material from a through total failure, including intermediate values. building. Leakpath analyses and an exponential air If total damage did not occur during the DBE (or exchange model are applied to the initial source slightly higher, to address uncertainty), total terms to determine the building source term. damage was considered incredible. These values range from .08 g to >.14 g for threshold damage to Dispersion Modeling .09 g to > .14 g for total damage. The values given The material released from the building is for equipment are those at which critical damage dispersed throughout the atmosphere. This process will occur. These values range from .01 g to >.14 g is modeled with a bivariate Gaussian plume model for the equipment evaluated. that has been modified by representing the crosswind horizontal dispersion by a rectangular CREDIBLE ACCIDENTS function for horizontal dispersion and an expansion Risk is the product of frequency of occurrence factor to account for plume meander. (This and consequences [8]. Population risk was representation provides an average concentration calculated as explained in subsequent sections. over a lateral width of 3a , which is within 20% of Worker and facility risk were also calculated. Gaussian peak value.) Meteorological dispersion Seismic hazards for each building were analyzed for values (x/Q) are estimated at the plant boundary risk due to each level damage, from the threshold and at various distances out to SO miles. x/Q values to intermediate levels, to the design basis for each scenario are based on a star deck earthquake or total damage. established from measured onsite winds. Population affected by the release of hazardous CONSEQUENCE MODELING material was estimated using the sector-averaged Consequences due to earthquakes are estimated Gaussian plume equation combined with 1980 by assessing the amount of hazardous or radioactive demographic data broken out by sector and distance material, applying a dispersion model to the release, to 50 miles. and calculating an expected dose to the public [9]. Dose Assessment Source Terms Organ and effective committed dose The material-at-risk is the amount of equivalent were calculated for the maximum offsite hazardous or radioactive material present in a individual (maximally exposed off-site individual) Second DOE Natural Phenomena Hazards Mitigation Conference - 1989

416 and for the population for various exposure pathways. Consequences were expressed in units of rems and latent cancer fatalities (LCF). LCF are the result of exposure to toxic material usually due either to a single large overexposure or continuing low-level exposure. Latent cancers do not appear immediately, but develop after a delayed period of time, which may be many years after the exposure. Rem (Roentgen equivalent man) is a unit of radiation dose equivalent which is numerically equal to the absorbed dose in rads multiplied by a quality factor (a distribution factor and any other modifying factors) to express the biological Upgrad* Opt (on effectiveness of the radiation relative to X or gamma radiation. Figure 2. Cost and Risk Impacts of Upgrading RISK ASSESSMENT Risk is determined by multiplying the frequency of occurrence (as determined by the CONCLUSION TERA seismic hazard curve) by the consequences. Cost/benefit analysis based on PR A is an Risk is expressed as rem/year and LCF/year. effective method for making recommendations and decisons that pertain to risk-reducing structural RISK MANAGEMENT upgrades. Risk assessment is used by contractor and DOE The PRA performed for earthquakes at RFP management primarily to identify vulnerabilities was used to recommend improvements to the for subsequent risk management decisions. This facilities. The risk to the population, worker and process includes risk comparisons, risk acceptance, facility was determined using geological, and risk reduction. The expected risk is compared engineering, meteorological, and health physics to the WASH 1400 study [9], and for all natural methods. phenomena events determined to be less than that A cost/benefit analysis was performed using from the data presented in that study. The Long- the risk assessment and the results of the Range Rocky Flats Utilization Study [10] concluded engineering studies for damage to the structures that natural phenomena events contribute 97% of and equipment by LATA. the composite RFP risk and that hardening the Structural and equipment upgrades at RFP are buildings would reduce the risk by 90%, thereby currently in construction which will reduce the risk reducing the risk for natural phenomena to further to the population by an order of magnitude. negligible levels. Cost/benefit analysis was performed to identify the most cost effective risk REFERENCES reduction options. [1] Geologic and Seismologic Investigations for Rocky Flats Plant. Dames and Moore, Denver, Cost/Benefit Analysis Colorado, July 1981. The consequences due to an earthquake were calculated for various structural upgrades. A [2] Seismic Risk Analysis for Rocky Flats. generic graphical depiction of the cost of upgrades Colorado. TERA Corporation, Berkeley, California, versus the reduction of risk is shown in Figure 2. April 1978, revised September 1982. Notice that cumulative costs increase as risk levels off. At a point it is not economical to increase [3] Seismic and Geologic Investigations and Design resources to reduce risk. This information has been Criteria for Rocky Flats Plutonium Recovery and used to make decisions on upgrading facilities. Waste Treatment Facility. URS/Blume and Associates, San Francisco, California, September Structural Upgrade 1972, revised June 1974. Upgrades are currently in construction and reduce the risk to the public by approximately one [4] Structural Evaluation of Existing Plutonium order of magnitude. Buildings and Auxiliary Structures at Rockv Flats

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~417 Plfln{. Agbabinn and Associates, El Segundo, California, February 1980.~~

~~[5] Study of Vital Equipment at Rockv Flats Plant for Forces Produced bv Natural Phenomena. URS/Blume and Associates, Danvers, Massachusetts, May 1981.~~

~~[6] Shaking-Table Testing for Seismic Evaluation of Vital Equipment at the Rockv Flats Plant. URS/Blume and Associates, Danvers, Massachusetts, April 1982.~~

[7] Realistic Evaluation of Consequences from Natural Phenomena Events for Building 707 Complex - Buildings and Vital Equipment. RFO06360(01), Los Alamos Technical Associates, Los Alamos, New Mexico, April 1985.

~~[8] Rockv Flats Risk Assessment Guide. Rockwell International, Golden, Colorado~~

~~[9] T. L. Foppe, "Natural Phenomena Risk Assessment at Rocky Flats Plant," in Proceedings of the DOE Natural Phenomena Hazards Mitigation Conference. 1985, pp. 70-75.~~

~~[10] Reactor Safety Study. WASH-1400 (NUREG- 75/014), U.S. Nuclear Regulatory Commission, Washington, D.C., October 1975.~~

~~[11] Long-Ranee Rockv Flats Utilization Study. ALO-1983, U.S. Department of Energy, Albuquerque, New Mexico, February 1983.~~

~~Second DOE Natural Phenomena Hazards Mitigation Conlerence - 1989~~

~~418 Invited Speakers~~

~~Walter Hays, U.S: Geological Survey, presented an after dinner talk on the International Decade for Natural Hazards Disaster Reduction Mitigation Conference~~

Peter Yanev, EQE Engineering, Jim McDonald, (left) Texas Tech University, Lunch Speaker, discussed the gave a special lecture on the effects of Hurricane perfomance of power plants and Hugo. Jim Beavers, (right) Martin Marietta industrial facilities during the Energy Systems, brought local charm to the 1988 Armenia Earthquake conference by acting as Master of Ceremony at the Dinner.

~~Second DOE Natural Phenomena Hazards Mitigations Conference - 1989~~

~~419 THE PERFORMANCE OF THE ARMENIA NUCLEAR POWER PLANT AND POWER FACILITIES IN THE 1988 ARMENIA EARTHQUAKE~~

~~Peter I. Yanev EQE Engineering, Inc. San Francisco, California~~

INTRODUCTION traced to 40.94l'N and 44.275*E, approximately The region affected by tire December 7,1988 75 kilometers northwest of the Armenian capital of Armenia (Spitak) earthquake included the cities of Yerevan. The focus was estimated at 15 Leninakan, Spitak, Kirovakan, and Stepanavan. kilometers. Surface wave magnitude of the main More than 700,000 people were affected. The shock was 6.9, moment magnitude 6.8, and body region is highly industrialized and contains wave magnitude 6.2. Strong motion lasted about numerous light and heavy industrial facilities and 30 seconds and was followed four minutes later by various power facilities including the 2-unit an aftershock with a body wave magnitude of 5.8. Armenia nuclear power plant. Armenian engineers estimate that peak The industry of the area was devastated by ground accelerations in Leninakan may have the earthquake. The Soviet authorities reported that reached 0.4g. The much stronger ground motion the earthquake caused major work stoppages in at generated in the Spitak area is comparable to the least 130 factories in the strongly shaken region 0.5g to l.Og that was experienced in the Sylmar and that about 80% of the engineered structures in area during the 1971 San Femando, California Leninakan, a city with a population of about earthquake. 290,000 people, were destroyed. EFFECTS ON CONVENTIONAL POWER The effects to the power facilities of the FACILITIES region were much less severe. Several substations The generalized electrical power grid of were damaged extensively. Two thermal power Armenia is shown in the accompanying figure. plants were affected: one in Kirovakan and one in The map dates from about 1975. At least three 220 Razdan. The damage was light. kV substations were located in the high intensity area. One of these, in Leninakan, was examined. The earthquake struck about 75 km north of Several 110 kV substations experienced extensive the two-unit Armenia nuclear power plant. Only damage; only one of these, in Nalband (the only minor shaking and no damage occurred at the one shown in the figure), was visited, and another plant. was briefly observed in Spitak. Both suffered severe damage. It was observed that control In terms of human and economic loss the full electrical equipment that was anchored appeared to impact of the Armenia earthquake will not be be structurally undamaged, with no obviously known for many months, although on February failed components such as relays. That was 20,1989, the Soviet press reported that property particularly noticeable at the Nalband 110 kV losses amounted to over $16 billion. The financial substation. losses from business interruption and closing of the Armenian nuclear power plant may double this The Nalband Control Building, a one- and figure. two-story load-bearing masonry wall structure with a concrete-frame and precast roof planks, collapsed OVERVIEW OF SEISMOLOGY on the control equipment. Even then, much of the On December 6,1988, a Richter magnitude equipment cabinets retained their shape and did not 3.0 earthquake began the seismic sequence that appear to be substantially damaged. U.S. nuclear culminated the following day with the main industry has spent much time and energy testing earthquake. The epicenter of the main event was similar cabinets on shake tables. Their inherent

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 420 strength under inertial seismic loads is well illustrated by the earthquake. Other anchored The Unit 2 plant building itself is excavated control equipment in the Nalband station that was into hard basalt. A sand or soil layer that is about not impacted by collapsing debris appeared to be 10 feet thick is sandwiched between two basalt structurally undamaged. layers in the foundation. Beneath the reactor shaft this soil layer has been reinoved and replaced with France, under the sponsorship of the French concrete. This is common engineering procedure earthquake engineering society and UNESCO, also for construction of power plants of all kinds in sent a team to Armenia. The team visited the volcanic terrain. Kirovakan thermal power plant. The plant has four natural-gas-fired units. Two units are rated at 12 Armenia Plant Response to the Earthquake. The MW, one is 25 MW, and one is 50 MW. The plant is equipped with three motion detectors in the plant experienced minor damage and tripped due to office building, chimney stack, and electric relay actuation. substation - that are designed to activate at a peak ground acceleration of 0.05g, which plant Piping throughout the visited facilities engineers have correlated to shaking intensity generally performed well. Severe damage was MSK-64 VI (MSK intensities are similar to caused by collapsing buildings and other debris. Modified Mercelli intensities used in the United During our relatively brief visit no failures of States). The reactors are programmed to scram and welded pipes were observed that were caused by shut down if two of the three detectors are inertial loads only. A leak was found at the plant triggered. Intensity at the site was estimated in the which was caused by relative displacement of mid-V range and no triggering occurred. supports. A flexible 200 mm main header moved However, vibration-reduction dampers connected far enough to damage a stiff 60 mm branch line at to equipment activated after 2.4 millimeters the connection. The damage was due to the displacement caused by the earthquake. The inertially induced displacement of the larger line horizontal peak ground acceleration at the facility against the smaller branch line. The damage was reported at about 0.03g with amplified reported by the French team at Kirovakan Chemical building response at about 0.05g. Following the Plant is typical. earthquake, the plant was shut down for 48 hours for a safety inspection. No significant damage was THE ARMENIA NUCLEAR POWER found, and upon restart all systems functioned PLANT normally. Armenia Units 1 and 2 comprise the nuclear facility at Oktembryan, about 90 kilometers south The original seismic design of the plant was of the epicenter. The units are in the Soviet V VER- based on MSK-64 intensity VII, which, during 440 class; Unit 1 began operation in 1977, Unit 2 design was upgraded to VIII as required for in 1980. Two planned 1,000 MW reactors for the important facilities in the Soviet Union. Following same site were cancelled recently. Officials the 1977 magnitude 7.2 Vrancea, Romania reported that the December 7, earthquake caused earthquake that damaged the Kozloduy nuclear neither scram nor damage at the plant, which plant on the Danube River in Bulgaria, the Soviets continued to operate. began implementation of more rigorous seismic design criteria for nuclear facilities in high Twenty-five percent of the power generated earthquake risk areas that include the Armenian by the Armenian nuclear plant was transmitted to site. Criteria are based on two design basis Georgia, which sold a portion of this to Turkey. earthquakes, a 100-year event and a 10,000-year Operation of the plant had caused growth of a event. working and residential community of approximately 10,000 people around the site. The criteria specify that equipment and structures be segregated into three categories: I for Armenia Plant Site Geology. Based on the inventory essential to safe shutdown; II for regional geologic map of Armenia, the plant site inventory generating power not directly critical to appears to be located on a large field of flat-lying the integrity of Category I equipment; and III for Quaternary volcanic rocks. inventory comprising all equipment not contained

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~421 -To Akstatu Alavordi 1 » 2 , yt* EPICENTER * '/•• ^, A Dznragrf~~

~~Klrovakan rov.|kiJ2~~

~~• Kirovakan 1 / lonlnakan-2 •, /Sevan Razdan™ / .~~

~~A Ciumush • Lake Armenia Nuclear. jf7.rinm Kamo Sevan i Powor Plant ' A**™*>>~~

~~Shaumlan-2 ^Yerovan Llgh MYerevan - .J~~

~~^Yehegnadzor O 50 Arara^-2. — KILOMETERS "~ ,.-— Shlnuair '~~

V m Thermal-electric power stations A y * * A Hydroelectric power stations Armenia Nuclear Plant: From left to right, office building « 11Q-kV substations Kafan , s • Kadzaran (prccasl-concrete-framc), Unit 1 reactor building (concrete- # 220-kV substations — 330-kV power transmission lines shear wall), Unit 1 turbine building (steel-frame with 220-kV power transmission lines precast-concrete facia).

~~Simplified map of ihc electrical power grid of Armenia.~~

~~. COMPRESSED GROUND SEAMS'.'"~~

~~BASALTS"1 CONCRETE PLATE~~

Armenia Nuclear Plant: Cross section of Main Building, Unit 2. Numbers identify seismic upgrades made following the 1976 Vrancea earthquake. 1 = steel tics; 2 = rcinforccd-concrcte wall; 3 = concrete base mai; 4 = sand fill; 5 = longitudinal concrete rib; 6 = rcinforccd-concretc rig. (Drawing courtesy of the U.S. Department of Energy.)

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

422 Armenia Nuclear Plant: Unit 2 control room. The panel Armenia Nuclear Plant: Overhead seismic bracing of indicates that Unit 2 was operating at 392 MW, approxi- electrical cabinetry and relay panels in control room area mately 9(W( of original, new plant capacity, during U.S. of Unit 2. Original start-up of Unit 2 was delayed three investigation two weeks after earthquake. years while seismic modifications such as this were made.

~~Nalband Electric Substation: Equipment that apparently remained in tact or had limited damage within collapsed control building. Electrical control cabinet*; supporting roof fragment.~~

lxninakan Electric Substation: Undamaged control equipment and relay-board cabinets are welded to steel baseplates. Note cracked rear wall of heavily damaged control house. The equipment was npcralional when power was restored within a lew flours to days following (lie earthquake.

~~Second DOK Natural Phenomena Hazards Mitigation Conference i<)~~

Spitak Sugar Refinery: Toppled electrical cabinets and detail of base where two poorly executed plug welds (arrows) failed to stabilize unit. Another cabinet (right) remained welded to its steel base-plate and was undamaged.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

424 in the first two categories. A retrofitting program occur, the Soviet Council of Ministers announced for critical equipment was implemented at the that Unit 1 would be shut down as early as Armenian site shortly after imposition of the February 25, 1989. Shutdown of Unit 2 would enhanced criteria, which specified that reactors, follow on March 18,1989. In a related pumps, steam generators, and valves be designed announcement made shortly after the Armenia to resist shaking intensity IX. Soviet scientists earthquake, officials stated that construction at six now believe that the plant is located in an area of nuclear plants, three of these in the Caucasus expected maximum magnitude of about 7.0 instead Mountains regions, would be halted or suspended of the M6.5 expected before. for seismic and other safety-related reasons. The following additional seismic During the first half of 1989 Soviet officials strengthening features were observed by the writer will study the feasibility of converting the during a visit to the plant after the December 7, Armenian plant to a natural gas facility. Even if 1988 earthquake: such a conversion is successful, the process will be lengthy and will only partly compensate for power • All electrical cabinets were bolted down and lost by the elimination of the nuclear units. extensively braced to each other in the Additional power will eventually be provided by control room Unit 2. two other nuclear facilities: in the Republic of Russia the Rostov plant, which has not yet been • The suspended ceiling of the Unit 2 control commissioned to operate, and in the Republic of room was seismically braced. Kazakh the Razdan plant, which is currently undergoing physical upgrade. In the meantime, • The plant manager was planning to strict power rationing and sharing will be strengthen the steel-frame roof trusses of implemented in the Armenia/Georgia/Azerbaijan the turbine building of Units 1 and 2. region. • The office building for the plant, which is CONCLUSIONS outside the power plant, is a precast The Armenia earthquake prompts concrete-frame building of the type that observations and lessons that are applicable in both performed poorly in the December 7 the Soviet Union and the United States. earthquake. The earthquake was a major disaster. While • The steel-frame turbine building is a Soviet seismologists had expected such an event, massive, well designed building that Soviet society had not prepared for it. The primary appears to have a large seismic capacity. cause of death, injury, and destruction was the total collapse of modem residential, commercial and Despite the upgrading program the Armenian notably industrial engineered structures that were units still lack a complete emergency core cooling not designed for the expected earthquake loads. system and a containment, both of which are mandatory in the United States. U.S. investigators In terms of damage and death directly caused also noted that the control room for Unit 2 had by seismic shaking, the United States has not windows, a feature that has been eliminated from experienced anything comparable to the Armenian critical structures at U.S. nuclear plants to reduce disaster. However, the combination of the risk from tornado and other potential missiles. seismologic and engineering conditions that Soviet officials acknowledge that to remain resulted in this disaster in the Trans-Caucasus exist operationally safe, the plant would have to be in such areas as Puget Sound in Washington; the retrofitted to resist intensity X, a program they say region centered on New Madrid between St. Louis, would be prohibitively costly. Missouri and Memphis, Tennessee; the region around Charleston, South Carolina; and the Salt Because of the plant's proximity to the Araks Lake City region in Utah. Given the regional valley, the major agricultural area in Armenia, and similarities in earthquake potential, building types, to quell the population's fears that a nuclear and the absence of earthquake preparedness accident similar to the Chernobyl disaster may programs, the possibility for an earthquake with

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 425 consequences of comparable proportions is ACKNOWLEDGEMENTS realistic. While such disasters are feared by the The U.S. investigation of the 1988 Armenia U.S. earthquake engineering community, U.S. earthquake was jointly sponsored by the U.S. society has not exhibited consistent concern about Academy of Sciences and the Soviet Academy of the risk. Sciences. The effort was a cooperative one in which information was collectively gathered and Steel-frame buildings performed well even freely exchanged. EQE's direct participation in the though they were not apparently designed for investigation and hence this report would not have higher seismic loads than the nearby precast been possible without the timely aid and financial structures. That is consistent with observations support of the Electric Power Research Institute in from numerous other earthquakes. The Soviet Palo Alto, California. Union, as well as building owners in the United States would do well to learn from that experience, REFERENCES particularly for important industrial structures. [1] EQE Engineering. Investigation of the San Salvador Earthquake of October 10,1986: Properly anchored power and industrial Effects on Power and Industrial Facilities. equipment like that in nuclear power plants were EPRINP-5616. Palo Alto, CA: Electric undamaged in areas of the highest intensity Power Research Institute, 1988. shaking. However, much equipment that appeared undamaged was not yet tested for function. In the [2] The 1986 North Palm Springs Earthquake: United States, industrial equipment, particularly Effects on Power Facilities. EPRINP-5607. electrical control equipment, is typically Palo Alto, CA: Electric Power Research unanchored. Substantial reduction of equipment Institute, 1988. losses through proper anchorage is a lesson that has been repeated without fail in all major [3] U.S. Department of Energy. Assistant earthquakes affecting industrial facilities. Secretary of Nuclear Energy. Overall Plant Design Descriptions: VVER, Water-cooled, The structural behavior of electrical cabinets Water-moderated Energy Reactor. DOE /NE- like those in nuclear power plants was very good 0084, revision 1. Washington, DC, October and demonstrates that shake table testing is usually 1987. unnecessary except for operability of components in critical facilities. Such components, however, can be tested individually at much lower costs. Welded piping without seismic design performed well. This again demonstrates that major changes to the ASME code for the seismic design of piping are necessary. Inertial loads should be made secondary loads and loads caused by relative support displacement should be upgraded to primary loads.

Financial loss from long-term business interruption will probably equal the direct-damage cost of the earthquake. It is apparent that events such as the Armenia earthquake affect all sectors of a society, and businesses that are not properly protected face complete ruin or major disruption. Corporations with California facilities are reducing their risks through risk analyses and selective strengthening. Corporations with facilities in the other earthquake-prone regions of the United States need to follow the example from California.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 426 INTERNATIONAL DECADE FOR NATURAL DISASTER REDUCTION~~

~~Walter W. Hays U.S. Geological Survey Reston, VA 22092~~

The concept of a Decade for Natural Disaster Reduction emergency management, and public administration. The has evolved considerably since it was proposed by Dr. committee is supplemented by working members from Frank Press, President of the U.S. National Academy and of the Federal Agencies having natural hazard of Sciences, in July 1984 at the Eighth World programs (for example, U.S. Geological Survey, Conference on Earthquakes Engineering. Now, the National Science Foundation, National Oceanic and United States and at least 28 other nations and Atmospheric Administration, Federal Emergency organizations have taken steps to organize and plan for Management Agency, National Institute of Standards concerted national and international actions during the and Technology, the U.S. Forest Service, National 1990's to reduce loss of life and economic losses from Aeronautics and Space Agency, Office of U.S. Foreign disasters triggered by natural hazards. Approximately Disaster Assistance, Corps of Engineers, and the State 100 nations arc expected to accept this goal and to join Department). with the United States and others following the 43rd General Assembly of the United Nations in the fall of The Federal Agencies Subcommittee on Natural 1989. They arc expected to forge unilateral, bilateral, Disaster Reduction and the U.S. National Committee and multilateral partnerships to make their country and must deal with three critical problems in the the world safer from floods, windstorms (typhoons, development of a U.S. Decade program. They arc: cyclones, hurricanes, and tornadoes), landslides, earthquakes, volcanic eruptions, wildfires, tsunamis, drought, and insect infestation. These programs arc • Leadership, expected to be multihazard, multifunctional, and multiorganizaUonal in scope. • Motivation, and • Funding. The United States, which faces annual losses of approximately $10 billion from the natural hazards Each of these complex problems is being addressed listed above, is developing this program the Decade cooperatively. The goal of the cooperative efforts is to: through a partnership involving: • Develop a vision of where we go as a Nation during • The Federal Agencies, which are organized through the decade. the Committee on Earth Sciences as the Subcommittee on Natural Disaster Reduction. • Identi fy a rallying point that all participants in the Decade throughout the Nation can associate with • The National Research Council of the National (for example: a) zeal for protecting our planet from Academy of Sciences which has organized a U.S. the disastrous consequences of natural hazards, b) National Committee on the Decade for Natural personal pride in protecting our homes, families, Disaster Reduction to advise the Federal Agencies. and workplaces, c) national pride that comes from gaining a position of preeminence in the world in • Institutions, organizations, and individuals having natural hazards research or in disaster prevention, abroad range of expertise throughout the nation and d) the challenge of working together to make who have responded to an "Invitation to Participate the world safer and more productive). in the Decade" extended by the U.S. National Committee in May 1989. • Create partnerships at all levels throughout the Nation to carry out programs to accomplish the The U.S. National Committee, chaired by Dr. Richard vision (for example: a) Federal-Federal, b) Federal- Hallgrcn, American Meteorological Society, consists of State, c) State-State, arid d) Federal-regional 15 members having backgrounds and broad experience partnerships). in the earth sciences, hydrology, wind engineering, earthquake engineering, fire safety, weather, political science, communication, insurance, the environment,

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

427 • Attack complex programmatic issues one step at a • Coordination and integration of the natural hazard time (for example: a) the linkage between programs of the Federal Agencies, State and local researchers and practitioners, and b) the interface governments, academia, and the private sector. between disciplines). • Development of hazard warning and prediction • Work smaller, not just harder (for example: a) take systems. advantage of the exiting body of fundamental knowledge on natural hazards developed through • Creation and sharing of multihazard databases and research, and b) utilize modem technology such as mitigation techniques. geographic information systems, satellites, and computer networks). • Implementation of post disaster data acquisition, data analysis, and data sharing programs. • Communicate (for example: a) use a nationwide speakers bureau to communicate the vision of the • Execution of research to close critical gaps in Decade to everyone, b) use a national news letter, c) fundamental knowledge on topics such as extreme improve the capability of credible sources of events and the implications of regionally and hazards and risk information to use all of the temporally varying natural hazards occurring available channels to reach decisionmakers and singularly or in combinations. policymakers and their constituencies with a meaningful message). • Provision of education and training throughout the Nation to increase awareness of natural hazards and • Simplify (for example, some loss reduction to enhance the capability and skills of professionals techniques for each natural hazard can be applied to to deal with their adverse societal impacts. another natural hazard). • Improvement of existing systems to communicate • Evaluate (for example: a) use the anniversary dates natural hazards and risk information, especially to of past notable disasters as a time to take stock of public officials, policymakers, and professionals progress and to examine gaps in knowledge or who can provide leadership for hazard mitigation. capability and b) use each new disaster as a window of opportunity to exiting capability). The U.S. National Committee on the Decade for Natural Disaster Reduction will join with the These seven actions are expected to provide solutions to committees and entities of other nations and the United the problems associated with leadership, motivation, Nations in carrying out the overall Decade program. and funding. The United Nations, which will have a major role in facilitating the Decade program, started their planning in The U.S. Committee, which met for the first time on March 1988 by forming a 25-member International Ad June 21-22,1989, will produce a comprehensive report Hoc Group of Experts on the Decade. Chaired by in 1990 containing model programs and Frank Press, this group delivered a report to the recommendations on how to implement them. These Secretary General of the United Nations on June 1, programs wiU call for 1989, containing model programs and recommendations for an organization to implement • Pilot projects to build local, State, regional, and them. The proposed organization for the United national partnerships. Nations consists of: • National projects to accelerate the application of loss • A Board of Trustees to marshall political support reduction measures, and and to seek funds. • Tnlcrnflt'nna^ projects to share the technology for • A program committee to solicit, develop, evaluate, hazard mitigation with other nations, especially and recommend programs to individual nations for developing countries. the Decade. The overall goal is to save lives and to reduce economic • A secretariat drawn from existing UN organizations losses in the United States. The particular thrusts of the to carry out operational requirements. U.S. Decade programs will be on achieving:

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

428 The report also recommended that a trust fund be • Wildfires, such as those that broke out in the Great established to provide resources to assist program Khingan Range in northern China on May 5,1987, development, especially for developing nations. The and the great Yellowstone wildfires of 1988 in the trust fund and the funds available to each national Western United States. committee or national entity would constitute the resources for the Decade program. • Drought, like the Dust Bowl drought in the 1930's that persisted in the Great Plains States of the The challenge of the International Decade for Natural United States for 10 years, and the long-term Disaster Reduction is unprecedented. If the past is an drought beginning in 1968 in the Sahcl countries of indication of what will happen in the 1990s and West Africa. afterward, the United States and the world will once again face potential disasters from: The goal of the Decade is to keep recurrences of these natural hazards from becoming disasters. The • Earthquakes, such as those that occurred in Alaska concerted actions of all nations working together in the in 1964, Algeria and Italy in 1980, Chile and 1990's can make this goal a reality. Mexico in 1985, Armenia, SSR, in 1988, and California in 1989. • Volcanic eruptions, such as those that occurred in Mount St. Helens, Washington in 1980, Ncvado del Ruiz, Colombia in 1985, and Izu-Oshima, Japan in 1986. • Roods, such as those that occurred in Florence, Italy in 1966, Nagasaki City, Japan in 1982, and Bangladesh in 1988. • Typhoons, cyclones, and hurricanes, such as those that occurred in Japan from typhoon Iscwan in 1959, in Pakistan from a cyclone in 1970, on the eastern seaboard of the United States from hurricane Agnes in 1972, in Jamaica and other Caribbean countries from hurricane Gilbert in 1988, in the Caribbean countries and South Carolina in the United States from hurricane Hugo in 1989.

• Tornadoes, such as the Palm Sunday outbreak that struck Iowa, Illinois, Indiana, Michigan, and Wisconsin in 1965; and the super outbreak of tornadoes that struck 11 Midwestern States and Canada on April 3,1974. • Landslides, such as those that occurred in Alaska in 1964 in conjunction with the Prince William Sound earthquake, in Peru on the west bank of the Manatro River in 1974, in Puerto Rico in 1983, in Ecuador in 1987, and in Tajckistan, SSR in 1989. • Tsunamis, such as the Showa Sanriku earthquake- tsunami that struck Japan in 1933, the Chilean earthquake-Tsunami which struck Hawaii and affected the cost of almost all of the countries of the Pacific rim on May 22,1960, and the Mindanao earthquake-tsunami that struck the Philippines on August 7,1975.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~429 Natural Phenomena Hazards Bibliography~~

Andrac, R. W., P. K. Tang, and W. S. Gregory Lawrence Livermorc National Laboratory (1985;, TVENT1: A Computer Code for Analyzing Tornado- Proceedings DOE Natural Phenomena Hazards Mitigation Induced Row in Ventilation System, LA-9836-MS. Conference, Lawrence Livcrmorc National Laboratory report CONF-8510118. Brynda, W. J., C. H. Scarlett, G, E. Tanquay, and P. R. Lobncr (1986), Nonrcactor Nuclear Facilities: McCann. M. W. and A. C. Boissonnade (1988), Preliminary Standards and Criteria Guide, DOE/TIC-11603-Rcv. 1, Flood Hazard Estimates for Screening Department of BNL-51444, Rev. 1. Energy Sites, Lawrence Livcrmorc National Laboratory report UCRL-21045. Coats, D. W. and R. C. Murray (1984;, Natural Phenomena Hazards Modeling Project: Seismic Hazard Models for McCann, M. W. and A. C. Boissonnadc (1988), Department of Energy Sites, Lawrence Livermorc National Probabilistic Flood Hazard Assessment for the N Reactor, Laboratory report UCRL-53582, Rev. 1. Hanford, Washington, Lawrence Livcrmorc National Laboratory report UCRL-21069. Coats, D. W. and R. C. Murray (1985), Natural Phenomena Hazards Modeling Project: Extreme Wind/Tornado Hazard McCann, M, W. and H. J. Owen (1985), Overview of Flood Models for Department of Energy Sites, Lawrence Considerations, Lawrence Livcrmorc National Laboratory Livcrmorc National Laboratory report UCRL-53526, report UCRL-15745. Rev. 1. McDonald, J. R. (1985a), Extreme Winds and Tornadoes: Eagling, D. G., Seismic Safety Guide (1983), LBL-9143 An Overview, Lawrence Livermore National Laboratory (NTIS No. DE84000542). report UCRL-15745.

ED2 (1985), Suspended Ceiling System Survey and Seismic McDonald, J. R. (1985b), Extreme Winds and Bracing Recommendations, Lawrence Livcrmorc National Tornadoes: Design and Evaluation of Buildings and Laboratory report UCRL-15714. Structures, Lawrence Livcrmorc National Laboratory report UCRL-15747. EQE Incorporated (1986), Practical Equipment Seismic Upgrade and Strengthening Guidelines, Lawrence McDonald, J. R. (1988), Structural Details for Wind Livermore National Laboratory report UCRL-15815. Design, Lawrence Livermorc National Laboratory report UCRL-21131. Gregory, W. S., B. D. Nichols, J. A. Moore, P. R. Smith, R. G. Steinke, and R. D. Idzorek (1987), Computer Nicolctli, J. P., L. E. Malik and P. G. Griffin (1985), Seismic Code Simulations of Explosions in Flow Networks and Design, Lawrence Livermore National Laboratory report Comparisons with Experiments, LA-11089-MS. UCRL-15744.

Hill, J. R. (1983), "Department of Energy" in Natural Savy, J. B. and R. C. Murray (1988), Natural Phenomena Earthquake Hazards Reduction Program: Fiscal Year Hazards Modeling Project: Flood Hazard Models for Activities (Annual Report to the United Slates Congress of Department of Energy Sites, Lawrence Livermorc National Federal Agency activities complied by FEMA beginning Laboratory report UCRL-53851. in 1983). Stcinke, R. G. (1987), NF85: A Three Dimensional, Air - Kennedy et al. (1989), Design and Evaluation Guidelines Dynamics Computer Code for Analyzing Explosions in for Department of Energy Facilities Subjected to Natural Structures, LA-11158-M. Phenomena Hazards, Lawrence Livermorc National Laboratory report UCRL-15910. Tang, P. K., W. S. Gregory, and C. R. Rickctts (1982), A Numerical and Experimental Investigation of Simulated Kennedy et al. (1989), Workshop Notes on Design and Explosions Inside a Flow Network, LA-9340-MS. Evaluation Guidelines for Department of Energy Facilities Subjected to Natural Phenomena Hazards. Yanev, P. I., S. Horn, C. A. Kirchcr and N. D. Bailey (1985), Emergency Preparedness, Lawrence Livcrmorc Kennedy, R. P. and S. A. Short (1985), Seismic National Laboratory report UCRL-15746. Analysis, Lawrence Livcrmore National Laboratory report UCRL-15742.

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

~~430 Attendees - NPH Conference~~

Vcmon L, Anderson James A. Carlson K. J. Coppersmith U,S, Department of Energy/SAN Lawrence Livermorc National Laboratory Gcomalrix Consultants 1333 Broadway P.O. Box 808, L-431 Spear Street Tower, #717 Oakland, CA 94612 Livcrmore, CA 94551 One Market Plaza (415) 273-6439 (415)422-1691 San Francisco, CA 94105 (415)957-9557 T. A. Angelelli Glen Carpenter Martin Marietta Energy Systems, Inc. Idaho Nationnl Engineering Laboratory Hans J. Dalilke P.O. Box 2003 P.O. Box 1625 Westingluiusc Idaho Nuclear Co. Oak Ridge, TN 37831 Idaho Falls, ID 83415 P.O. Box 4000. MS 332 (615) 576-1995 (208)5264166 Idaho Falls, ID 83403 (208)526-9814 Gregory R. Ashley C. Y. Chang Impcll Corporation Geomalrix Consultants William L. Daugherty 333 Research Cl. Tech. Park Spear Street Tower, #717 Wcstinghouse Savannh River Co. Norcross, GA 30092 One Market Plaza P.O. Box 616 (404)441-5195 San Francisco, CA 94105 Aikcn, SC 29802 (415)957-9557 (803)725-4180 Donald S. Asquith Martin Marietta Energy Systems, Inc. C. V. Char Kimbcrly A. Davis Oak Ridge, TN 37830 C. T. Main, Inc. U.S. Department of Energy/ (615)574-5583 P.O. Box 240236 Savannah River Charlotte, NC 28224 P.O. Box A N. G. Awadalla (704)554-1100 Aikcn.SC 29802 Wcstinghouse Savannah River Co. (803) 725-4407 P.O. Box 616 Carl P. Chen Aiken, SC 29802 Bcchtcl National, Inc. Sandra A. Day 50BealcSt.,P.O.Box3965 Rockwell International JolinT. Baxter San Francisco, CA 94119 Rocky Flats Plant Westinghouse Hanford Co. (415) 768-8823 P.O. Box 464 P.O. Box 1970 Golden, CO 80402-0464 Richland, WA 99365 Jor-Shan Choi (303) 966-7440 (509)376-5791 Lawrence Livcrmorc National Laboratory P.O. Box 808, L-390 Richard M. Drake James E. Beavers Livcrmore, CA 94551 Fluor Daniel, Inc. Martin Marietta Energy Systems, Inc. (415)423-8038 3333 Michclson Drive Oak Ridge, TN 37830 Irvine, CA 92730 (615)574-3117 Tony Chung (714)975-5791 Science Applications Int'l. Corp. Daniel W. Benjamin 800 Oak Ridge Turnpike John Eidingcr Lawrence Livcrmore National Laboratory Oak Ridge, TN 37831 Impcll Corporation P.O. Box 808, L-545 (615)482-9031 6500 West Freeway, Suite 400 Livcrmore, CA 94551 Fort Worth, TX 76116 (415)423-1339 Alan J. Colburn (817)737-1050 U.S. Department of Energy/Richland Michael A. Boyd 335SnydcrRd. John C. Elder U.S. Department of Energy Richland, WA 99352 Los Alamos National Laboratory Oak Ridge Operations (509) 375-3248 P.O.Box 1663 P.O. Box 2001 Los Alamos. NM 8754-4 Oak Ridge, TN 37831-8733 (FTS) 843-3366 (615)576-0821

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

431 Norman Eng Richard II. Oilier William A, lleineken NUTECH Wcstingliou.se Hanforcl Co. Martin Marietta Energy Systems, Jnc, 145 Maninvalo Ln. P.O. Box 1970 P.O. Box Mil) San Jose, CA 95119 Rieliland, WA 99352 Puducaii, KY 4200] (408) 281-6007 (509)376-4710 (512)441-6101

H. M. Engle, Jr. Bob Graham H. Tom Hicks Englc & Englc Structural Engineers Science Applications Inl'l. Corp. U.S. Department of Energy/ 32 Ross Common, Suite 230 8115 Beaver Ridge Rd. Savannah River Ross.CA 94957 Knoxvillc.TN 37931 P.O. Box 4125 (415)461-0630 (615)693-2188 Martinez, GA 30907 (803) 725-2027 Steven F. Fenncr Richard C. Gwattncy Lockwood Greene Engineers Oak Ridge National Laboratory James R. Hill P.O. Box 3561 P.O. Box 2009 U.S. Department of Energy, EH-33 Oak Ridge, TN 37831 Oak Ridge, TN 37831-8051 19901 Germ antownRd. (615)483-8080 (615)574-0726 Gcrmantown, MD 20545 (301)353-4508 George F. Flanagan C. Richard Hammond Martin Marietta Energy Systems, Inc. Oak Ridge National Laboratory John A. Hoffmeistcr P.O. Box 2008,7964A, MS-6390 P.O. Box 2009 Martin Marietta Energy Systems, Inc. Oak Ridge, TN 37830 Oak Ridge, TN 37831-8051 P.O. Box 2009 (615)574-8541 (615)574-6499 Oak Ridge, TN 37831-8101 (615)574-0261 Gary E. Frccland Mian A. Huq Lawrence Livcrmorc National Laboratory Kaiser Engineers Hanford R. M. Holmes P.O. Box 808, L-654 P.O. Box 888 Martin Marietta Energy Systems, Inc. Livcrmore, CA 94551 Richland, WA 99352 Bldg. 1000, MS-6333 (415)422-9411 (509) 376-6526 P.O. Box 2008 Oak Ridge, TN 37831 Ken Frickc Brent G. Harris (615)574-6375 Martin Marietta Energy Systems, Inc. EG&G Idaho Oak Ridge, TN 37830 1903 W. 190th St. John Hortel (615) 576-0465 Rexburg, ID 83440 Martin Marietta Energy Systems, Inc. (208) 356-6387 P.O. Box 628, MS-1203 Roy M. Gale Pikcton.OH 45661 Baltelle Pacific Northwest Laboratory Raymond E. Harris (614) 897-4002 P.O. Box 999, MS-P7-63 U.S. Department of Energy Richland, WA 99352 P.O. Box 2001 Leigh House (509) 376-0550 Oak Ridge, TN 37831-8733 Los Alamos National Laboratory (615)576-0841 MS-D443 Ron Gallagher Los Alamos, NM 87545 R. P. Gallagher Associates, Inc. Steven P. Harris (505)667-1912 116 New Montgomery St., #206 EQE Engineering, Inc. San Francisco, CA 94105 595 Market St., 18th Floor Bruce Howard (415) 882-9190 San Francisco, CA 94105 Gilbert Commonwealth, Inc. (415)495-5500 320 Cedar Bluff Rd. WalmerT. Gay, JD Knoxvillc,TN 37923 Gilbert Commonwealth, Inc. Walter Hays (615)690-4389 320 Cedar Bluff Rd. U.S. Geological Survey Knoxvillc.TN 37923 Department of the Interior R. Joe Hunt (615)690-4389 905 National Center Martin Marietta Energy Systems, Inc. Reston. VA 22092 P.O. Box 2009 (703) 648-6712 Oak Ridge, TN 37831-8101 (615)574-1979

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 432 Randy Inglebarger Raymond H. Kincaid Masoud M-'/addi Lockvi'ood Greene Engineers EQE Engineering, Inc. Impel) Corporation P.O.Box3S61 3150 Bristol St.. Suite 350 5000 Executive J'kwy Oak Ridge, TN 37831 Cosia Mesa, CA 92626 SanRumon.CA 94583 (615)483-8080 (714)850-9299 (415)943-4500

Suzctte Jackson Charles Kircher Lincoln E. Malik Idaho National Engineering Laboratory Jack R. Benjamin Associates URS/Blumc P.O. Box 1625 444 Castro St. 1504th St. Idaho Falls, ID 83415 Mountain View, CA 94087 San Francisco, CA 94103 (208)526-4293 (415)969-8212 (415)957-5300

Kenneth Jaquay James P. Knight Miguel A. Manriquc Rockwell International U.S. Department of Energy Impcll Corporation 6633 Canoga Ave. Washington, D.C. 20545 350 Lennon Ln. Canoga Park. CA 91303 (301)353-4435 Walnut Creek, CA 94598 (818)700-4035 (415)943^734 Gregg Lagcrbcrg Clifford Jarman Science Applications Int'l. Corp. Matthew E. Maryak NUS Corporation 800 Oak Ridge Turnpike Westinghousc Savannah River Co. 900 Trail Ridge Rd. Oak Ridge, TN 37830 P.O. Box 616 (706-13C) Aiken, SC 29801 (615)482-9031 Aikcn, SC 29802 (803) 649-7963 (803) 557-9069 Kenneth Lanham Thomas M. Jclinck EBASCO Frank E. McClure U.S. Department of Energy 9050 Executive Park Lawrence Berkeley Laboratory P.O. Box 2001 Knoxville.TN 37931 Building 90G Oak Ridge, TN 37831-8733 (615)691-6955 One Cyclotron Road (615)576-0838 Berkeley, CA 94720 Manrico C. Lara (415)486-5715 James J. Johnson Lawrence Livcrmorc National Laboratory EQE Engineering, Inc. P.O. Box 808, L-362 Lorcn Mead 595 Market St., 18th Floor Livermore, CA 94551 Martin Marietta Energy Systems, Inc. San Francisco, CA 94105 (415)423-1745 P.O. Box 628. MS-1203 (415)495-5500 Piketon.OH 45661 Richard C. Lee (614)897-2750 Robert A. Just SAIC Oak Ridge National Laboratory 101 Convention Center Dr. Thomas M. Monahon Oak Ridge, TN 37831 Las Vegas, NV 89109 Wcsiinghouse Savannah River Co. (FTS) 624-6497 (702) 658-0134 Savannah River Site Aikcn, SC 29801 F. Edward Kendall Jay Love (803) 725-9532 U.S. Department of Energy H. J. Degcnkolb Associates 2325 Black Bear Rd. 350 Sansomc St. Thomas J, Moran Knoxville.TN 37923 San Francisco, CA 94104 Argonnc National Laboratory (615)576-9439 (415)392-6952 9700 S. Cass Avc. Argonne, IL 60439 Darrcl R. Ketcham Donna E. Lucas (312)972-5901 Westinghousc Savannah River Co. Battcllc, Pacific Northwest Laboratory Savannah River Site P.O.Box999,MS-P7-78 Steven G. Mouty Aikcn.SC 29802 Richland, WA 99352 Gilbert Commonwealth, Inc. (803) 557-9773 (509)376-1331 320 Cedar Bluff Rd. Knoxville, TN 37923 Jeffrey K. Kimball Donald T. Lynch (615)690-4389 U.S. Department of Energy Allied Signal 1000 Independence Avc. 2000 East 95th St. Washington, D.C. 20585 Kansas City, MO 64131-3095 (202)586-1063 (817)997-5844

Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 433 Joan Muccko M.K, Kavinriru A. R. Schado Manin Marietta Engineering Systems, Inc. EQE Engineering, Inc. Westinghouse Hanford Co, IH24 El Ptado Dr. 3150 Bristol St., Suite 350 P.O. Box 1970, MSIN R3-(W Knoxvillc, TN 37922 Costa Mesa, CA 9262ft Ridilaml, WA 99352 (615)574-8733 (714)850-9299 (509) 373-3075

Kenneth G. Murphy John W. Reed R. B. Schappcl U.S. Department of Energy Jack R. Benjamin & Associates, Inc, Martin Marietta Energy Systems, Inc. Germantown, MD Mountain Bay Plaza, Suite 501 P.O. Box 2003 (301)353-6514 444 Castro St. Oak Ridge, TN 37831 Mountain View, CA 94041 (615)574-8124 Robert C. Murray (415)969-8212 Lawrence Livcrmorc National Laboratory Don Schlick P.O. Box 808, L-197 Peter Rieck U.S. Department of Energy Livermorc. CA 94551 Gilbert Commonwealth, Inc. Nevada Operations Office, MS-523 (415)422-0308 Rd. 1 Box 1498 Las Vegas, NV 89193 Reading. PA 19603 (702) 794-7275 John M. Musacchio (215)775-2600 Paul C. Rizzo Associates, Inc. David A. Sharp 300 Oxford Dr. Alan C. Roliay, K6-84 Wcsiinghousc Savannah River Co. Monroeville, PA 15146 Battcllc Pacific Northwest Laboratory P.O. Box 616 (412)856-9700 P.O. Box 999 Aikcn.SC 29808 Richland, WA 99352 (803)725-5406 Bruce My»u (509) 943-2578 Impcll Corporation David W.Sheffcy 5000 Executive Pkwy Ronald W. Rucker Martin Marietta Energy Systems, Inc. San Ramon, CA 94583 U.S. Department of Energy P.O. Box 203 (415) 275-4866 P.O. Box 2001 Oak Ridge, TN 37831-7155 Oak Ridge. TN 37831-8711 (FTS) 626-8499 Jim Myers (615)576-1835 Science Applications Int'l. Corp. Stephen A. Short 7524 Gyncvere Dr. R. K. Sadigh Impcll Corporation Knoxvillc, TN 37931 Gcomatrix Consultants 27401 Los Altos, Suite 480 (615)947-2131 Spear Street Tower, #717 Mission Vicjo, CA 92691 One Market Plaza (714)582-9178 Emie C. Ocoma San Francisco, CA 94105 Westinghouse Hartford Co. (415)957-9557 Robert L. Sindelar P.O. Box 1970 Westinghouse Savannah River Co. Richland, WA 99352 William E. Sallec P.O. Box 616 (509) 376-5015 Martin Marietta Energy Systems, Inc. Aikcn, SC 29802 P.O. Box 2008 Kermit R. Peters Oak Ridge, TN 37831-6837 Elwood A. Smictana U.S. DOE/Idaho (615)574-6491 EQE Engineering, Inc. 785 DOE Place 3150 Bristol St., Suite 350 Idaho Falls. ID 83402 R. Brian Sanchez Costa Mesa, CA 92626 (208)526-0170 Los Alamos National Laboratory (714) 850-9299 P.O. Box 1663, MS-M984 Larry Proctor Los Alamos, NM 87545 Richard P. Smith Martin Marietta Energy Systems, Inc. (505)667-7351 EG&G Idaho, Inc. P.O. Box 2008 P.O. Box 1625. MS 2107 Oak Ridge, TN 37831-6390 Jean B. Savy Idaho Falls. ID 83415 (615) 574-8555 Lawrence Livermore National Laboratory (208) 526-9896 P.O. Box 808, L-196 Livermorc, CA 94551 (415)423-0196

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

434 Jedd Starkey Michael A. Tenbus Donald L. Williams WINCO Martin Marietta Energy Systems, Inc. Martin Marietta Energy Systems, Inc. Lawrence Livormore National Laboratory 332 Wardley Rd. P.O. Box 2002 P.O. Box 808, L-440 Knoxville. TN 37922 Oak Ridge, TN 37831 Livermorc.CA 94551 (615)574-9818 (615)574-8710 (415)423-3365 George C. Thomas H. Elwyn Wingo Date E. Stephenson Stevenson & Associates Westinghousc Savannah River Co. Weslinghou.se Savannah River Co. 9217 Midwest Ave. Building 773^1A Savannah River Site Cleveland. OH 44125 Aiken. SC 29808 Aiken, SC 29802 (216)587-3805 (803) 725-2930 (803)725-5217 Nishikant R. Vaidya Ivan G. Wong Roger L. Stover Paul C. Rizzo Associates, Inc. Woodward-Clyde Consultants Martin Marietta Energy Systems, Inc. 300 Oxford Dr. 500 12lh St., Suite 100 P.O. Box 2008 Monrocville, PA 15146 Oakland, CA 94607 Oak Ridge. TN 37831-6390 (412)856-9700 (415)874-3014 (615)574-8544 Raman M. Venkata Matthew W. Wrona C. V. Subramanian Bechtcl Savannah River, Inc. Bechiel National, Inc. Sandia National Laboratories 1700 Market St., Suite 1816 50BcaleSt.,P.O. Box 3965 P.O. Box 969 Philadelphia, PA 19103 San Francisco, CA 94119 LivermorcCA 94551 (215)587-6849 (415)768-6930 (415)294-2311 Gary R. Wagcnblast Steve Wuthrich Delano F. Surdahl Wcstinghousc Hanford Co. Bcchicl National, Inc. U.S. DOE/Albuqucrquc P.O. Box 1970 421 LcJcan Way Pennsylvania & H St. Richland, WA 99352 Walnut Creek, G. 94596 Albuquerque, NM 87115 (509)376-5015 (415)930-9403 (515)844-2070 Larry A. Walker Loring A. Wyllic, Jr. Mark Swatta Holmes & Narvcr H. J. Dcgcnkolb Associates Impell Corporation 1050 E. Flamingo Rd., Suite 330 350 Sansome St. 6500 Executive Pkwy Las Vegas, NV 89109 San Francisco. CA 94104 San Ramon, CA 94583 (707) 295-0800 (415)392-6952 (415)275-4760 David L. Wall Peter I. Yanev Greg A. Swoyer U.S. Department of Energy EQE Engineering, Inc. Allied-Signal Aerospace P.O. Box 2001 959 Market St. P.O. Box 419159 Oak Ridge, TN 37831-8733 San Francisco, CA 94105 Kansas City, MO 64141-6159 (615)576-2550 (415)495-5500 (816)997-5440 Evan Wciner Robert Youngblood Mclvin N. Taie Wcsiinghouse Hanford Co. Brookhavcn National Laboratory Lawrence Livcrmore National Laboratory P.O. Box 1970 Building 130 P.O. Box 808, L-384 Richland, WA 99352 Upton, NY 11973 Livcrmorc. CA 94551 (509)376-5152 (516)282-2363 (415)423-6018 Mark E. Whiting Ann M. Tallman Impell Corporation Wcstinghouse Hanford Co. 333 Research Ct., Tech. Park P.O. Box 1970 Norcross, GA 30092 Richland. WA 99352 (404)441.5195 (509)376-4038

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989 435 Index of Authors~~

Anderson, Dennis M., 317 Hays, Walter W., 427 Sallec, William E., 310 Awadalla, Nabil G., 129,365,375 Hill, James R,, 151 Savy, Jean B., 18,238 Bak, Walter R., 343 House, Leigh S., 179 Scott, Don L., 317 Baxter, John T., 396 Jackson, SuzctlcM,, 317 Scott, Mark A., 121 Bcrnrcutcr, Don, 238 Johnson, David H., 382 Sccondo, Robert J., 120 Bingham, Gail E., 66 Johnson, James J., 28,35 Scidcnslickcr, R., 46 Bush, Spencer H., 375 Jones, W. Dale, 162 Short, Stephen A., 2,365 Bultcmer, David R., 382 Kelly, J.M., 46 Silva, Walter J., 317 Carpenter, Glen S., 317 Kennedy, Robert P., 2, 365 Sindelar, Robert L., 129, 375 Chang, Ching Y., 290 Kctcham, Darrcl, 35 Smictana, Elwood A. 109 Chen, Carl P., 142 Kincaid, Raymond H., 109 Smith, Richard, 317 Chicn, Shan H., 382 Kirchcr, Charles A., 46 Smith, Richard P., 282 Conrads, Thomas J., 336 Love, Richard J., 206 Stark, Cathy L., 317 Constantinou, M., 46 Manrique, Miguel A., 343 Starkcy,JcddG.,220 Coppersmith, Kevin J., 252,262 McCann, Martin W,, 18 Stevenson, John D., 226 Dahlkc,Hans,J., 120,142 McClurc, Frank E., 186 Stover, Roger L., 28, 172 Darragh, Robert B., 317 McDonald, James R., 12 Subramanian Chittcr V., 58 Daughcrty, William L., 129,375 Mchta, Hardayal S., 129 Sumodobila, Basilio N., 28 Day, Sandra A., 414 Mcnsing, Richard, 238 Tajirian, Fredrick F., 46,142 Dizon, John 0., 172 Merry man, Larry D., 310 Tallman, Ann M., 299 Drake, Richard M., 122 Moghtaderi-Zadeh, Masoud, 214 Tang, Y. K., 290 Eder, Stephen J., 35 Monahon, Tnomas, 35 Todcschini, Ricardo A.A., 142 Eidingcr, John M., 46 Murray, Robert C, 2 Tseng, Wen S., 290 Esfandiari, Sohrab, 214 Ocoma, Ernie C, 73 Vaidya,N.,46 Flanagan, George F., 382 Ovadia, D., 46 Wagcnbast, Gary R., 135 Fricke, Kenneth E., 162 Perla, Harold R., 382 Weincr, Evan O., 337 Gallagher, Ronald P., 89 Phillips, W. Scott, 179 Wesley, Donald A., 365 Gilbert, HollieK., 317 Power, Maurice S., 290,329,330 Wingo, H. Elwyn, 405 Giller, Richard A., 80 Proctor, Larry D., 310 Winkel.BobV., 135 Griffin, Michael J., 66 Quinn Robert D., 99 Wong, Ivan G., 317 Hackett, William R., 282 Ranganath, Saniath, 129 Wright, Douglas H., 317 Hammond, C. Richard, 192 Rccd, John W., 352 Wrona, Matthew W., 220 Hardy, Gregory S., 35,66 Rodgcrs, David W., 282 Wuthrich, Steven J., 220 Harris, Brent G., 199 Rohay, Alan C, 272 Wyllic, Loring A., 206 Harris, Steven P., 28,172 Rose, David L., 220 Yancv, Peter I., 420 Hashimoto, Philip S., 172 Sadigh, Ross K., 329,330 Youngs, Robert R., 252,262

~~Second DOE Natural Phenomena Hazards Mitigation Conference - 1989~~

436 Disclaimer This document was prepared as an account of work sponsored by an agency of the United States Government, Neither the United Slates (invernment nor the University of California nor any of their employees makes any warranty, express or implied, or assumes any legal liability or responsibility lor the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial products, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or the University of California. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or the University of California, and shall not be used for advertising or product endorsement purposes.

~~Work performed under the auspices of (he U.S. Department of Finergy by I awrence l.iverrrmre National 1 aboratory under Contract VV-7405-l-:ng-48.~~