— IOO
Reproduced by IAEA XA9952643 - s i
SMiRT 13 Post Conference Seminar 16 (Advanced Technology Seminar)
SEISMIC EVALUA TION OF EXISTING NUCLEAR FACILITIES
PROCEEDINGS OF THE SMiRT 13 POST CONFERENCE SEMINAR NO. 16
organized by the INTERNATIONAL ATOMIC ENERGY AGENCY
and the NATIONAL UNIVERSITY OF CORDOBA (ARGENTINA)
Iguazu, Argentina August 21 - 23, 1995
NOTE The material in this document has been supplied by the authors and has not been edited by the IAEA. The views expressed remain the responsibility of the named authors and do not necessarily reflect those of the government(s) of the designated Member State(s). In particular, neither the IAEA nor any other organization or body sponsoring the meeting can be held responsible for any material reproduced in this document. / 30-46 DISCLAIMER
Portions of this document may be illegible in electronic image products, Images are produced from the best available original document. Reproduced by IAEA Limited Distribution
SMiRT 13 Post Conference Seminar 16 (Advanced Technology Seminar)
SEISMIC EVALUA TION OF EXISTING NUCLEAR FA CILITIES
PROCEEDINGS OF THE SMiRT 13 POST CONFERENCE SEMINAR NO. 16
organized by the INTERNATIONAL ATOMIC ENERGY AGENCY
and the NATIONAL UNIVERSITY OF CORDOBA (ARGENTINA)
Iguazu, Argentina August 21 -23, 1995
NOTE The material in this document has been supplied by the authors and has not been edited by the IAEA. The views expressed remain the responsibility of the named authors and do not necessarily reflect those of the government(s) of the designated Member State(s). In particular, neither the IAEA nor any other organization or body sponsoring the meeting can be held responsible for any material reproduced in this document. SMiRT-13, POST CONFERENCE SEMINAR No. 16 on
"Seismic Evaluation of Existing Nuclear Facilities "
ORGANIZERS
INSTITUTIONS:
* INTERNATIONAL ATOMIC ENERGY AGENCY
* NATIONAL UNIVERSITY OF CORDOBA (Argentina)
and the sponsorship of:
* SCIENCE AND TECHNOLOGY RESEARCH COUNCIL of Cordoba Province (Argentina) ORGANIZING COMMITTEE:
Antonio R. GODOY and Aybars GURPINAR International Atomic Energy Agency Division of Nuclear Installation Safety Engineering Safety Section
Carlos A. PRATO National University of Cordoba (Argentina)
Heki SHIBATA Yokohama National University (Japan)
Norbert KRUTZIK Siemens-KWU (Germany)
John D. STEVENSON Stevenson and Associates (USA) LOCAL ORGANIZING COMMITTEE:
Carlos A. PRATO Department of Structures National University of Cordoba Cordoba, Argentina
Ricardo ROCCA National University of Cordoba Cordoba, Argentina
Emilio REDOLFI National University of Cordoba Cordoba, Argentina
Alejandro GIULIANO National Institute for Seismic Prevention (INPRES) San Juan, Argentina
Luis M. ALVAREZ ENACE S.A. Buenos Aires, Argentina PREFACE
At the International Atomic Energy Agency the past twenty years has seen the development of an internationally recognized set of nuclear power plant safety standards (NUSS) consisting of 5 codes and 55 safety guides providing basic requirements and recommendations on governmental organization, siting, design, operation, and quality assurance. Since the beginning of the process of standard development, the safety of nuclear power plants in relation to natural external events, mainly earthquakes, had a significant role and a number of the safety guides refer to this topic.
Concurrent with the development of NUSS, the past six years has witnessed the increase in the safety review services as an element of Agency's national, regional and interregional technical assistance and co-operation programmes. These review services have the main purpose to assist Member States for the implementation of requirements and recommendations of the IAEA codes and safety guides as well as standards of international practice to ensure consistent and uniform assessment and enhancement of safety. Issues mainly related to the seismic safety of operating nuclear facilities resulted in the organization and performance of more than 120 engineering safety review services between 1989-1996. Thus, after 1991 an increasing activity was carried out regarding the evaluation and upgrading of the seismic safety of former Soviet design nuclear power plants. The vulnerability of WWER type reactors to earthquakes received special attention and consequently a large number of services were provided by the Agency in coordination with other international and national organizations (regulatory bodies, utilities/operators and engineering and consultants firms). Some of the common problems encountered during these revisions led to the initiation of the Coordinated Research Program on "Benchmark study for the seismic analysis and testing of WWER type NPPs" in 1993, with 25 participating institutions from 15 countries.
The need to provide transfer of information and experience in the evaluation of seismic safety was recognized by the Member States in a Technical Committee Meeting held in Tokyo, Japan, in August 1991, following the SMiRT-11 International Conference. In that meeting it was emphasized that effective exchange of information is essential for further development of international co-operation in this field. It should also be mentioned that in the programme of SMiRT-11 a special session on IAEA activities was included as part of the national and international research and standard programmes associated with SMiRT.
To respond to the request of Member States and to reflect the work done and the experience gained, the Agency organized this post conference seminar on the subject as part of the activities of SMiRT. This is the second time that experts involved in seismic re-evaluation and upgrading of operating nuclear facilities convened to discuss the issues of mutual interest and the experience that they had, first in Vienna in 1993 at the time of SMiRT-12 and then in Iguazu in 1995.
We are happy to have had the opportunity to contribute to this effort.
Antonio GODOY and Aybars GURPINAR SMiRT-13, POST CONFERENCE SEMINAR No. 16 on
"Seismic Evaluation of Existing Nuclear Facilities "
CONTENTS
OBJECTIVES OF THE SEMINAR 1
SUMMARY 2
KEYNOTE PAPER: "An International Nuclear Safety Regime" 3 Mr. Morris ROSEN Assistant Director General International Atomic Energy Agency
SESSION I: "EARTHQUAKE EXPERIENCE AND SEISMIC RE-EVALUATION"
(1.1) "Seismic re-evaluation of nuclear facilities worldwide: overview and status" 13 Messrs. J. J. JOHNSON, R. D. CAMPBELL, G. S. HARDY, M. K. RAVINDRA and A. J. HOY (EQE International, USA)
(1.2) "On Southern Hyogo Prefecture Earthquake and some related activities in Japan" .. 37 Mr. H. SHIBATA (Yokohama University, Japan)
(1.3) "Latur earthquake and its impact on the aseismic design of structures in India" .... 55 Mr. P.C. BASU (Atomic Energy Regulatory Board, India)
SESSION II: "COUNTRY EXPERIENCE IN SEISMIC RE-EVALUATION PROGRAMME"
(II. 1) "U.S. experience in seismic re-evaluation and verification programs" 77 Mr. J. D. STEVENSON (Stevenson and Associates, USA)
(II.2) "A regulatory view of the seismic re-evaluation of existing nuclear power plants in the United Kingdom" 95 Messrs. J. E. INKESTER and P. M. BRADFORD (HSE, HM Nuclear Installations Inspectorate,U.K.) (11.3) "Seismic re-evaluation of French Nuclear Power Plants" 105 Mr. R. ANDRIEU (EdF, France)
(11.4) "Nuclear Power Plants - Seismic Review Programme in Spain" Ill Messrs. J. G. SANCHEZ CABANERO and A. JIMENEZ JUAN (Consejo de Seguridad Nuclear, Spain)
SESSION III: "GENERIC WWER STUDIES"
(III. 1) "Seismic safety of nuclear power plants in Eastern Europe" 129 Messrs. A. GURPINAR and A. GODOY (Division of Nuclear Installation Safety, IAEA)
(111.2) "Comparison of ex-USSR norms and current international practice in design of seismicaily resistant nuclear power plants" 155 Mr. M. DAVID (Consulting, Engineering, Czech Republic) and Mr. B. HAUPTENBUCHNER (Technical University Dresden, Germany)
(111.3) "Seismic PRA, Approach and Results" 167 Mr. R. D. CAMPBELL (EQE-International, USA)
SESSION IV: "ANALYTICAL METHODS FOR SEISMIC CAPACITY RE- EVALUATION"
(IV. 1) "Seismic Design of Nuclear Power Plants: Where are we now?" 189 Mr. J. M. ROESSET (The University of Texas at Austin, USA)
(IV.2) "Dynamic analysis of WWER type NPPs using different procedures for consideration of soil-structure interaction affects" 207 Messrs. L. HALBRITTER and N. J. KRUTZIK (Siemens-KWU, Germany)
(IV.3) "Dynamic analysis of WWER-1000 Nuclear Power Plants" 225 Messrs. A. ASFURA and M. J. JORDANOV (EQE-International, USA)
(IV.4) "In-structure spectra generation for Kozloduy NPP, Bulgaria" 243 Mr. M. KOSTOV (CLSMEE, Bulgaria)
(IV.5) "Applications of seismic damage hazard analysis for the qualification of existing nuclear and offshore facilities" 251 Messrs. P. BAZZURRO, G.M. MANFREDINI and I. DIAZ MOLINA (D'Appolonia S.p.a., Italy and Argentina)
11 SESSION V: "EXPERIMENTAL METHODS FOR SEISMIC CAPACITY RE- EVALUATION"
(V.I) "Full scale dynamic structural testing of Paks NPP" 281 Messrs. E. M. DA RIN and F. P. MUZZI (ISMES Spa, Italy)
(V.2) "Full scale dynamic tests of Atucha II NPP" 289 Messrs. T. KONNO and S. UCHIYAMA (KAJIMA, Japan), L. M. ALVAREZ (ENACE SA, Argentina), A. R. GODOY (IAEA), M.A. CEBALLOS and C. A. PRATO (University of Cordoba, Argentina)
(V.3) "Experimental and numerical determination of the dynamic properties of the Reactor Building of Atucha II NPP" 311 Messrs. M. A. CEBALLOS, E. J. CAR, T.A. PRATO, C. A. PRATO (University of Cordoba, Argentina), L. M. ALVAREZ (ENACE SA, Argentina), and A.R. GODOY (IAEA)
(V.4) "Shaking table testing of mechanical components" 329 Messrs. D. JURUKOVSKI, L. TASKOV, D. MAMUCEVSKI and D. PETROVSKI (Institute of Earthquake Engineering and Engineering Seismology, Republic of Macedonia)
(V.5) "Experimental and computer analyses of control rods drive systems seismic capacity" 347 Messrs. V. KOSTAREV, V. ABRAMOV, A. BERKOVSKI, P. S. VASILIEV and A.J. SCHUKIN (CKTI-Vibroseism, Russia)
(V.6) "Shaking table testing of electrical equipment in Argentina" 367 Messrs. J. S. CARMONA, F. ZABALA, J. SANTALUCIA, C. SISTERNA, M. MAGRINI and L. OLDECOP (Universidad Nacional de San Juan, Argentina)
SESSION VI: "CASE STUDIES"
(VI. 1) "Design and Implementation Experience of Seismic upgrades at Kozloduy and Paks NPPs" 377 Messrs. V. G. BOROV, V. TRICHKOV, A. ALEXANDROV and M. JORDANOV (EQE-Bulgaria, Bulgaria)
(VI.2) "Seismic upgrading of WWER 440-230 structures, Units J4, Kozloduy NPP" ... 393 Messrs. D. STEFANOV, M. KOSTOV, H. BONCHEVA and G. VARZANOV (Academy of Sciences, Bulgaria)
(VI.3) "Seismic upgrading of piping supports in WWER 1000 MWe" 401 Mr. M. F. SCHMIDT (Stussi and Partner, Switzerland)
(VI.4) "Methodology and results of the seismic probabilistic safety assessment of Krsko Nuclear Power Plant" 407 Messrs. M. K. VERMAUT and Ph. MONETTE (Westinghouse Energy Systems Europe,
iii Belgium) and R.D. CAMPBELL (EQE International, USA)
TIMETABLE 435
LIST OF PARTICIPANTS 439
IV OBJECTIVES OF THE SEMINAR
Programmes for re-evaluation and upgrading of safety of existing nuclear facilities are presently under way in a number of countries around the world. An important component of these programmes is the re-evaluation of the seismic safety through definition of new seismic parameters at the site and evaluation of seismic capacity of structures, equipment and distribution systems following updated information and criteria.
The Seminar is intended to provide a forum for the exchange of information and discussion of the state-of-the-art on seismic safety of nuclear facilities in operation or under construction.
Both analytical and experimental techniques for the evaluation of seismic capacity of structures, equipment and distribution systems are discussed.
Full scale and field tests of structures and components using shaking tables, mechanical exciters, explosive and shock tests, and ambient vibrations are included in the seminar programme with emphasis on recent case histories.
Presentations at the Seminar also include analytical techniques for the determination of dynamic properties of soil-structure systems from experiments as well as calibration of numerical models.
Methods and criteria for seismic margin assessment based on experience data obtained from the behaviour of structures and components in real earthquakes are discussed.
Guidelines for defining technical requirements for capacity re-evaluation (i.e. acceptable behaviour limits), and design and implementation of structure and components upgrades are also presented and discussed.
SESSIONS PROGRAMME
Session I: Earthquake experience and seismic re-evaluation
Session II: Country experience in seismic re-evaluation programme
Session III: Generic WWER studies
Session IV: Analytical methods for seismic capacity re-evaluation
Session V: Experimental methods for seismic capacity re-evaluation
Session VI: Case studies
Session VII: Panel discussions. SUMMARY
A summary in figures of the SMiRT-13 - Post Conference Seminar 16 on "Seismic Evaluation of Existing Facilities" is as follows:
26 papers presented;
a key note lecture by the IAEA Assistance General Director on Nuclear Safety;
52 participants;
from 17 countries:
* Argentina * Armenia * Bulgaria * Belgium * Czech Republic * Finland * France * Germany * India * Japan * Korea * Macedonia * Russian Federation * Slovenia * Switzerland * UK * USA;
financial support to 9 participants. XA9952644
Structural Mechanics in Reactor Technology
An International Nuclear Safety Regime
Morn's Rosen Assistant Director General for Nuclear Safety International A tomic Energy Agency
For the many of us closely involved with the safe use of nuclear power, the opening for signature of the Convention on Nuclear Safety and the ongoing work to prepare a Convention on Radioactive Waste Safety are particularly important milestones. The Nuclear Safety Convention has been signed by almost 60 countries and will likely come into force early next year. It is the first legal instrument that directly addresses the safety of nuclear power plants worldwide. The Convention on Radioactive Waste Safety is presently in an early stage of preparation, but a political will exists for its early adoption and it may possibly be ready late next year.
The two safety conventions are only one facet of international collaboration to enhance safety. A review of some cooperative efforts of the past decades and some key provisions of the new safety conventions, will show how international cooperation is increasing nuclear safety worldwide.
International Nuclear Safety Regime
International collaboration has been continuously evident during nuclear power's evolution. There have been a multitude of bilateral and multilateral collaborative efforts which you will hear about in the course of this meeting. In the global governmental arena, the IAEA has supported this internationalization process through a number of initiatives which have led to what is now called an international nuclear safety regime. This regime encourages adherence to high standards of safety through internationally agreed safety recommendations, through binding agreements, through an array of safety services and through a wide range of international cooperative and assistance efforts.
At the IAEA, the past twenty years has seen the development of an internationally recognized set of nuclear power plant safety standards consisting of 5 codes of practice and 55 supporting safety guides covering regulatory organizations, siting, design, operations, and quality assurance. These standards were adopted in its entirety by China as a basis for its growing nuclear power activities and other countries have used them in part or as a reference for their own standards. There are additional international recommendations such as those contained in the Basic Safety Standards for Radiation Protection, and the Regulations for the Safe Transport of Radioactive Material. Important binding international agreements cover the physical protection of nuclear materials in international transport, civil liability for damage following nuclear incidents, and -2- the early notification and assistance in the event of a serious nuclear accident or radiological emergency. There also exists formalized incident reporting procedures and an international nuclear event scale for communicating the severity of operational events to the public.
A concurrent development at the IAEA of its widely used safety review services has promoted a worldwide exchange of information and experience, particularly in the key operations area. International teams of experts organized by the Agency have visited and provided advise to all countries with nuclear power plants. There are peer review services also in siting, design and operations and a new service to promote an adequate safety culture at all organizations involved with nuclear power activities. The IAEA efforts have given incentive and support to the current international nuclear safety regime.
The Convention on Nuclear Safety
Turning to the Nuclear Safety Convention itself, the substantial efforts to prepare the document date back four years. Although, many believed an international safety agreement was premature and others considered it unnecessary or even undesirable, a major consensus supporting such an instrument developed at an international safety conference held at the IAEA headquarters in Vienna in September 1991. After almost three years of intense negotiations and hard work at seven meetings of a Group of legal and technical experts from 53 countries, a Diplomatic Conference was held in June of last year to adopt the Nuclear Safety Convention.
Structure and Contents
The Convention itself is relatively simple in structure. It consists of an introductory preamble and four chapters consisting of 35 articles. The first chapter delineates the principal objectives. The second and most substantive contains the various obligations. The third chapter deals with the required periodic meetings to review national reports on the measures taken to implement each of the obligations, while the last contains the final clauses and other judicial provisions common to international agreements.
The fundamental principle of the Convention is that overall "...responsibility for safety rests with the State where a nuclear installation is located". Each nation within its own legislative and regulatory framework must govern safety. Bilateral and multilateral mechanisms are available to provide assistance and support.
For the purposes of the convention, a nuclear installation is defined as only,
"...any land based civil nuclear power plant...including such storage, handling and treatment facilities for radioactive materials as are on the same site and are directly related to the operation of the plant. " -3- Nevertheless, the preamble supports the broad agreement which now exists for a future binding convention to deal with the safety of radioactive waste management also suggests the future development of instruments for other parts of the nuclear fuel cycle.
The Convention seeks to achieve its objectives through adherence to general as opposed to detailed binding technical standards. The fundamental safety principles contained in the Convention are based largely on the IAEA document "The Safety of Nuclear Installations" (Safety Series No. 110) which present general principles rather than detailed and prescriptive ones. The more specific detailed requirements, which are continuously evolving and in a practical sense cannot be placed into a treaty, would when necessary be dealt with on a plant by plant basis through the Convention's review process.
Individual obligations to meet these general objectives are contained in 16 articles which are grouped under three headings. The first group covers the essential prerequisite for safety, that is the requirements for Legislation and Regulation. It calls on each country,
• to establish and maintain a legislative framework and an independent regulatory body, separate from other bodies concerned with promotion and utilization of nuclear energy, and
•to govern safety through a system of licensing, inspection, and enforcement.
The second group of obligations concern a number of the more General Safety Considerations and contains requirements for each country,
• to ensure an effective safety management system maintained throughout the lifetime of the installation.
This is to be realized by steps to assure that all organizations involved with safety give priority to safety through a number of measures such as,
•to provide adequate financial and human resources, trained staff, quality assurance programmes, safety assessments and verification activities, radiation protection of workers and the public, and emergency plans which are tested.
The final group of articles dealing with obligations addresses some specific aspects of the Safety of Installations. They require appropriate steps,
•to ensure that technical aspects of siting, design and construction, and operation are considered and continuously assessed throughout the lifetime of the installation.
These are to be achieved through a number of measures such as, -4- • comprehensive and systematic safety assessments, safety analyses to define safe operating conditions, commissioning programmes, and the reporting and analyses of safety events.
In the Convention's obligations special attention is given not only to the responsibilities for emergency planning within a State with nuclear power plants, but also to the responsibilities towards neighboring States. With regard to emergency planning,
"...(countries) must provide competent authorities of the States in the vicinity of their nuclear installations with appropriate information for emergency planning and response. "
There is also a complementary requirement,
"...(that countries) which do not have a nuclear installation on their territory, but are likely to be affected in the event of a radiological emergency in a neighboring State, take the appropriate steps to ensure that emergency plans have been prepared and tested that cover the activities to be carried out in the event of an emergency. "
In recognition of the general concerns of neighboring countries, there is also a specific obligation calling,
"...(for consultations with countries) in the vicinity of a proposed nuclear installation, insofar as they are likely to be affected by that installation and, upon request providing the necessary information ... , in order to enable them to evaluate and form their own assessment of the likely safety impact of the installation. "
In concluding with the specific obligations spelled out in the Convention, there is one that deals with the current problem of existing reactors, such as those of Soviet design, with recognized safety deficiencies. It states,
"When necessary ... the Contracting Party shall ensure that all reasonably practicable improvements are made as a matter of urgency to upgrade the safety of the installation. If such upgrading cannot be achieved, plans should be implemented to shut down the installations as soon as practically possible. The timing of the shut-down may take into account the whole energy context and possible alternatives as well as the social and economic impact. "
There is no obligation for a State to shut down a plant immediately as this step may not be advisable when considering the overall social and economic impact. But, the Convention's review mechanism will bring the difficult plant into international and public discussion. The difficulties of dealing with deficient nuclear power plants is currently demonstrated by the situation in countries operating Soviet designed reactors. No country has yet decided to permanently -5- close any plant. Improvements are being made through national and international efforts.
Implementation and Peer Review Process
Turning now to the most vital issue for the Convention's success. Its success is dependent on compliance with the agreed obligations. Thus, a principle feature of the Convention is the implementation mechanism which requires each country to demonstrate this compliance through written reports submitted for review. There is a key requirement that,
"each Contracting Party (is to) submit for review ... a report on the measures it has taken to implement each of the obligations of the convention."
These reports will be dealt with through periodic review meetings. The first of these meetings would not be later than two and one-half years after the Convention enters into force. Future meeting intervals would not exceed three years. Extraordinary meetings could be held if a majority of countries request it.
Within 6 months of entry into force, a preparatory meeting is to be convened to lay out the structure of the required national reports and the review process. The review process will have to be efficient, involve reasonable costs, and not place an undue burden on national reporting. Reports cannot be a detailed item by item review of national programmes, but it nevertheless will have to sufficiently demonstrate compliance and how this was accomplished. The review process will have to identify problems, concerns, uncertainties and omissions in national reports.
In order to begin an early exchange of opinion so as to be ready for the formal preparatory meeting, the Agency's Secretariat convened an informal meeting of representatives of signatories in early March of this year. Forty eight countries took part. The meeting concentrated on the review process and the contents of national reports. A further meeting is to take place in November.
One review possibility is to create three sub-groups; one for governmental and organizational matters, a second for siting design and construction, and a third for operations. Another approach would create groups of countries which would review only the reports of those in the group. Each group would consist of countries having a diverse number of nuclear power plants ranging from a high number to none. Among the advantages of this approach would be a more overall look at safety, avoiding the difficulty of only examining a limited number of safety areas. It could also bring increased quality of review through smaller groups. A disadvantages would be the possible inconsistency of reviews between country groups. It would also conflict with the Convention's requirement that all parties can comment on the national reports of all others. -6- There is some support for the right of each country to submit a report in the form and length it deems necessary. Others see the desirability of harmonizing reports to the extent possible, for the purpose of an efficient and effective review. Support exists for a report having an extended summary or introduction to identify main themes followed by an article-by-article or more topical discussion. A recently received proposal prepared by Japan and Germany calls for a six part document as follows:
A. Introduction providing an overview of national nuclear policy and nuclear programmes.
B. Legislative and regulatory framework covering nuclear installations.
C. Financial and human resources, human factors, quality assurance, radiation protection as well as safety principle covering siting design and construction.
D. Operational safety.
E. Planned activities to improve safety.
F. Annexes containing data on the installations, reference such as for laws and standards, published reports including those of international review missions.
The question of language is still open. One proposal calls for national reports to be prepared in the national language or in a single designated language. If it is not submitted in the designated language, a translation of the report would be provided.
The review meeting itself could begin with a short plenary session with general statements followed by work in sub-groups, composed by topic or by groups of countries. The meeting would likely be limited to two or three weeks. A document addressing the safety issues discussed and the general conclusions reached would be adopted by consensus and made available to the public.
The IAEA would be the Secretariat for the Convention. It would convene, prepare and service all meetings and transmit information received or prepared in accordance with the provisions of the Convention. The IAEA could provide other services if requested.
Entry into Force
As for the date of entry into force, the Convention states that,
"...(it will) enter into force on the ninetieth day after the date of deposit with the Depositary of the twenty second instrument of ratification, acceptance
8 -7- or approval, including the instruments of seventeen States, each having at least one nuclear installation which has achieved criticality in a reactor core.
When the Convention was opened for signing on 22 September, Canada which chaired the meetings of legal and technical experts was first to sign. This was followed by 37 countries including 23 of the 30 countries with nuclear power plants. At this point there are almost 60 signatures in place and eight countries have ratified the Convention, these being the Czech Rep., Japan, Norway, Poland Romania, Slovakia, Spain and Turkey. Five of these have nuclear power plants. The political will which enabled the preparation of the Convention in a relatively short time period may also help bring it into force early next year.
Concluding Remarks
The safety philosophy and practices involved with a formal legal framework for the safe use of nuclear power will foster a collective international involvement and commitment. It will be a positive step towards increasing public confidence in nuclear power. It will also be a valuable example for other potentially hazardous industries of our industrial world including those involving other energy sources.
This brief scan of the nuclear safety convention will undoubtedly not have made you an expert in the subject, but I hope it has given you an opportunity to more fully appreciate the significance of the document. Further details of the Convention on Nuclear Safety can be found in the Legal Series No. 16 publication of the IAEA. CONTENTS OF THE NUCLEAR SAFETY CONVENTION
PREAMBLE
CHAPTER 1. OBJECTIVES, DEFINITIONS AND SCOPE
ARTICLE 1. OBJECTIVES ARTICLE 2. DEFINITIONS ARTICLE 3. SCOPE OF APPLICA TION
CHAPTER 2. OBLIGATIONS
(a) General Provisions
ARTICLE 4. IMPLEMENTING MEASURES ARTICLE 5 REPORTING
ARTICLE 6 EXISTING NUCLEAR INSTALLATIONS
(b) Legislation and Regulation
ARTICLE 7. LEGISLATIVE AND REGULATORY FRAMEWORK ARTICLE 8. REGULATORY BODY
ARTICLE 9. RESPONSIBILITY OF THE LICENSE HOLDER
(c) General Safety Considerations
ARTICLE 10. PRIORITY TO SAFETY ARTICLE 11. FINANCIAL AND HUMAN RESOURCES ARTICLE 12. HUMAN FACTORS ARTICLE 13. QUALITY ASSURANCE ARTICLE 14. ASSESSMENT AND VERIFICATION OF SAFETY ARTICLE 15. RADIATION PROTECTION ARTICLE 16. EMERGENCY PREPAREDNESS (d) Safety of Installations ARTICLE 17. SITING ARTICLE 18. DESIGN AND CONSTRUCTION ARTICLE 19. OPERATION
CHAPTER 3. MEETINGS OF THE CONTRACTING PARTIES
ARTICLE 20. REVIEW MEETINGS ARTICLE 21. TIMETABLE ARTICLE 22. PROCEDURAL ARRANGEMENTS ARTICLE 23. EXTRAORDINARY MEETINGS ARTICLE 24. ATTENDANCE ARTICLE 25. SUMMARY REPORTS ARTICLE 26. OBSERVERS AT MEETINGS OF CONTRACTING PARTIES ARTICLE 27. CONFIDENTIALITY ARTICLE 28. SECRETARIAT
CHAPTER 4. FINAL CLAUSES AND OTHER PROVISIONS
ARTICLE 29. RESOL UTION OF A GREEMENTS ARTICLE 30. SIGNATURE, RATIFICATION. ACCEPTANCE. APPROVAL. ACCESSION ARTICLE 31. ENTRY INTO FORCE ARTICLE 32. AMENDMENTS TO THE CONVENTION ARTICLE 33. DENUNCIATION ARTICLE 34. DEPOSITARY ARTICLE 35. AUTHENTIC TEXTS
10 SESSION I
"EARTHQUAKE EXPERIENCE AND SEISMIC RE-EVALUATION"
NEXT PAGE(S) left BLANK 11 III! XA9952645
SEISMIC REEVALUA TION Ob NUCLEAR FACILITIES WORLDWIDE: OVERVIEW AND STATUS
by
Robert D. Campbell Dr. James J. Johnson Alan J. Hoy Greg S. Hardy EQE International . EQE International Ltd. Dr. Mayasandra K. Ravindra 44 Montgomery, #3200 500 Longbarn Blvd. EQE International San Francisco, CA 94104 Birehwood, Warrington 18101 Von Kantian, #400 (415)989-2000 Cheshire WA2 OXI" Irvine, CA 92715 (415) 433-5107 (Fax) United Kingdom (714)833-3303 (01925)838372 (714)833-3392(Fax) (01925) 838396(Fax)
ABSTRACT
Existing nuclear facilities throughout the world arc being subjected to severe scrutiny of their safety in the event of an earthquake. In the United States, there have been several licensing and safety review issues for which industry and regulatory agencies have cooperated to develop rational and economically feasible criteria for resolving the issues. Currently, all operating nuclear power plants in the United States are conducting an Individual Plant Examination of External Events, including earthquakes beyond the design basis. About two- thirds of the operating plants are conducting parallel programs for verifying the seismic adequacy of equipment for the design basis earthquake. The U.S. Department of Energy is also beginning to perform detailed evaluations of their facilities, many of which had little or no seismic design. Western European countries also have been reevaluating their older nuclear power plants for seismic events often adapting the criteria developed in the United States. Wilh the change in the political systems in Eastern Europe, there is a strong emphasis ftom their Western European neighbors to evaluate and upgrade the safety of their operating nuclear power plants. Finally, nuclear facilities in Asia are, also, being evaluated for seismic vulnerabilities. This paper focuses on the methodologies that have been developed for reevaluation of existing nuclear power plants and presents examples of the application of these methodologies to nuclear facilities worldwide.
INTRODUCTION
Nuclear facilities designed, constructed, and operated over the past 40 years have been subjected to substantially varying levels of seismic analysis, design, qualification, and operating procedures. These variations are due to many reasons; principal among them being significant changes in the state-of-the-art, -science, and -engineering of earthquakes and their effects on natural surroundings and man-made facilities. Broadly speaking, the technical disciplines for which significant advances have occurred include seismic hazard prediction; geotechnical engineering; structural, mechanical, and electrical engineering; systems analysis; mathematics and software development, especially as they relate to statistical evaluations of data and probabilistic treatment of uncertainties in all disciplines; general computer hardware and mkt223/jjj/papr696a
13 sol I \\ .nc. :nui cousin it 111 MI pi act ices. In all ol I hese a leas, obsci v;ilion;il dala. i|iialitat ivc ami (|II;)I)1)I;IIIVL-, has served lo substantially advance the piolcssion s slalc ot knowledge, l! is these advances in know ledge liasc along u ilh aging ol llie laeiliU and a licighlcncd concct n lor saletv (hat have motivated the need tor periodic evaluations ol the stale oi nuclear facilities with respect lo seismic risk.
Seismic hazard prediction is ihc single most influential factor in motivating the evaluation ol the seismic performance of nuclear facilities. It is, likewise, the single most important factor in determining a facility's seismic design. Although there have been enormous gains in the understanding of seismic hazard over the past 40 years, there remains considerable uncertainty in seismic hazard prediction. Advances in all areas of seismic hazard prediction have occurred. Physical measurements ot' fault characteristics and plate movements aid in estimating activity tales, maximum earthquake potential on fnuIt systems, existence of previously unknown fault systems, etc. Recorded slronu ground motions ol hundreds of earthquakes have expanded the data base from which ground motion models are created. In fact, recorded ground motions over the past 25 years have altered the profession's belief as to the maximum accelerations that can be produced by an earthquake—recordings at and above l.Og clearly demonstrate the broad range ot motions possible dependent on earthquake magnitude, fault characteristics, site conditions, frequency content, etc. The important effect of site soil conditions has been repeatedly reinforced over the last few decades—qualitatively, in terms of built facility performance, and quantitatively, from comparative ground motion measurements. In spite of these and many other advances in the profession's understanding, there remains considerable uncertainty in seismic hazard predict ion. This uncertainty and the often accompanying ehaiiiiinu perception of the seismic hazard at specilie sites has been (he prime motivator m the evaluation and recvaluation of nuclear facilities. In addition, it has forced seismic hazard to be characterized probabilistically, as it should, including an explicit treatment of uncertainty. The uncertainty associated with flic seismic hazard, also strongly motivated the development of the seismic margin methodology, as described in the ensuing text. Seismic margin methodologies focus on determining earthquake ground motion levels at which one has high-confidence-of-low- prob;ibility-ol-fai!urcs (! IC1.PF) of the facility. One can then interpret the 11CLPF with respect to current and future estimates of the site seismic hazard for decision-making purposes.
Cieotechnical engineering advances have focused on ground motion transmission, soil material behavior, soil failure modes, buried structure and component performance, and foundation performance. Substantial advances in understanding of the variation in ground motion in rock/soil media have been made over the last 15 years (Johnson and Asfura, 1992). The increase in knowledge is due principally to extensive recordings of ground motion within the soil and on partially and fully buried structures. Whereas seismic hazard prediction is characterized by rapidly changing perceptions over the last several decades, the increased understanding of ground motion at the sile has permitted more realistic evaluations o! '.he impact of revised site seismic hazard on facilities of interest.
Structural, mechanical, and electrical engineering have gained substantial experience and understanding of the performance of structures, equipment, components, and commodities when subjected to strong earthquake ground motion. This experience is derived from laboratory testing experience and facility performance during earthquakes. Capitalizing on this experience has led to the development of design-by-rule guidelines for equipment, components, and commodities in a majority of typical applications. The ensuing text describes elements of this evolution.
mkl223/jjj/p:ipr6%a
14 Systems analysis has evolved into an essential element in the design and evaluation of nuclear facilities. The behavior of primary and support systems necessary for safe operation and shutdown of the facility arc modeled with respect to earthquake risk. Systems analysis extends beyond the facility boundaries when considering accident mitigation systems behavior for the surrounding communities. Systems analysis has evolved into the decision tool for prioritizing the various elements of the facility for evaluation.
The evolution of computer hardware and software has permitted one to examine individual and combined phenomenon in increasing detail. In all cases, when used properly, results from these evaluations enhance the decision-making process.
Given this background, three situations have arisen over the last 15 years which are addressed by methods described herein.
(i) Seismic design basis has been established during the licensing process and remains in tact. Questions concerning the ability of structures, equipment, components, and/or commodities to meet the design basis have been postulated and verification of seismic adequacy is required. For commercial nuclear power plants within the United States, several unresolved safety issues (USI), are in this category. Approaches to the resolution of these USIs are discussed in the ensuing text.
(ii) Beyond design basis seismic events are considered as part of the Independent Plant Examination of External Events (IPEEE) for U.S. commercial nuclear power plants. Of the multiple purposes of the 1PE program, evaluation of plant risk to beyond design basis events is principal. The several methods appropriate for seismic IPEEE are discussed here.
(iii) Revised or newly implemented seismic design criteria. For facilities not originally designed for seismic events or for which the design ground motion has changed substantially, requirements have often been instituted for verification of the seismic adequacy of these plants. Hybrid methods, combinations of seismic design, margin, and verification programs, have been effectively used to address facility concerns in this category. U.S. Department of Energy (DOE) facilities and many Eastern European nuclear power plants are in this category. SEISMIC EVAL UA TIONMETHODOLOGIES IN THE UNITED STA TES
The evaluation of seismic vulnerabilities in earlier operating plants in the U.S. began in the late 1970s during the Systematic Evaluation Program (SEP). Initial activities were conducted by a Senior Review Pane! funded by the USNRC, and consisted of the analytical evaluation of selected structures, walkdowns and sample calculations for equipment. Subsequent activities by the utilities and their contractors were in response to the findings of the USNRC consultants and focused on piping analysis and evaluation of selected structures and equipment. Operability of equipment was not verified during the SEP program. Unfortunately, these utility programs did not always focus on priority issues, often..because of non risk based perceptions of governing vulnerabilities and excess conservatism contained in regulatory requirements at that time.
mkt223/jxj/papr696a
15 In general, the Sl-.P program allowed lor more liberal acceptance criteria tor MiucturcN by allowing the response to go beyond the elastic limit. Newmark and I lall (I 979) developed criteria lor evaluation ot structures and equipment which included the use of inelastic response spectra. For structural systems which undergo inelastic deformation, the effective dynamic response could be defined by a linear elastic rc^nonsc analysis using a reduced response spectrum to define the input motion. The reduction in spectral acceleration was based upon the allowable inelastic deformation (ductility) and the frequency of the structural system and was based upon exhaustive analytical studies of structures subjected to real earthquake records and on observed behavior of structures during strong motion earthquakes.
The evaluation of structures in many cases utilized the inelastic spectra concept but this was not carried over to piping systems. Piping evaluations were very conservatively conducted using classic linear clastic response spectrum analysis, low damping and conservatively defined input motion. Subsequently, some of these conservatisms have been reduced in efforts to develop
more rational criteria for resolution of other seismic issues and for new design'b1 .
There were a number of unresolved seismic safety issues in the U.S. during the 1980s and early 1990s. Many of these issues were consolidated into two major programs. The first of these programs is the demonstration of the operability of safety related equipment during and after the design basis earthquake. This activity is limited in scope to only address those issues which have not previously been resolved for the design basis earthquake. The second major issue is the evaluation of the plant response to seismic events beyond the design basis.
The first issue is Unresolved Safety Issue (USI) A-46, dealing with operability of safe shutdown equipment in 72 of the earlier U.S. NPPs. The scope, however, was expanded to include long term decay heat removal equipment (USI A-45), selected seismic design basis issues (US! A-40) and seismic spatial systems interactions (USI A-17). Some additional passive items have also been included (cable raceways, tanks and heat exchangers. Structures and piping are not included in this program since they have been addressed in other programs. The second issue is the Individual Plant Examination of External Events (IPEEE). In this program, all structures, piping and equipment essential for safe shutdown must be evaluated for seismic events greater than the design basis. This program also included seismic spatial systems interactions, scismic- fire and seismic-flood interactions, and to some extent Generic Issue (Gl)-57 which includes the consequences of inadvertent activation of fire suppression systems during a seismic event.
The U.S. Department of Energy has numerous test and production reactors and process facilities located on government reservations which have not been subject to the U.S. nuclear regulatory process for power reactors. DOE Order 5840.28 (DOE, 1992) requires that these facilities be reevaluated for natural phenomena hazards and brought up to safety standards commensurate with the public risk involved.
Resolution for USI A-46
A Generic Implementation Procedure (GIP), (SQUG, 1991), has been prepared over a several year period to provide criteria and methods to resolve most of the outstanding seismic issues related to the design basis earthquake. The GIP is based heavily upon the use of earthquake and testing experience in lieu of analysis and testing of components. A large database of earthquake and testing experience has been reviewed by a Senior Seismic Review and Advisory Panel (SSRAP, 1991), and the USNRC and rules have been formulated to demonstrate survivability
mki223/jjj/papr6%a
16 and opcrability of a large generic class of equipment. A testing database lias also been collected and reviewed to establish operability limits for equipment and relays (Mcrz, 1991a and 1991 b).
A final Safety Evaluation Report (SER) on the GIP has been issued and the affected U.S. utilities completed or are nearing completion of programs to apply this procedure to their NPP's. Newer plants whose equipment has been seismically qualified to IEEE 344-1975 or later are exempt from this issue.
The steps involved in the applications of the GIP to resolution of USI A-46 are:
• Development of safe shutdown equipment list • Development of seismic demand (in-structure response) • Equipment walkdown and screening • Relay evaluation • Outlier resolution • Reporting
The safe shutdown equipment list defines that equipment which must function to safely shutdown the reactor after a design basis earthquake. A single shutdown path is defined, but redundancy must be maintained for decay heat removal functions. Accident mitigation equipment is not required.
The seismic demand is that specified for the design basis (safe shutdown) earthquake. Many plants are choosing to develop new spectra using more modern and less conservative methods than originally used. In some cases, the NPPs have elected to change their licensing basis by using a USNRC Regulatory Guide 1.60 spectral shape to define the ground motion rather than the spectrum originally used. By using the Regulatory Guide Spectral Shape, more recent and more liberal regulatory criteria may be used for analysis of structure and equipment response. Numerous studies have been performed to quantify the calculational conservatism in a wide variety of seismic analysis procedures used to define in-structure response spectra for the evaluation of the seismic adequacy of equipment, components, and commodities. Substantial conservatism can exist in the seismic demand for design and qualification of structures, equipment, components, and commodities. These excess conservatisms can exist for older plants analyzed using approximate conservative approaches appropriate at the time as well as newer plants. Plant configurations for which substantial conservatisms in seismic demand often exist are those located on soil sites with plant structures having embedded foundations and partially embedded structural elements. The lessons learned from field observation (over the past 15 years) related to the spatial variation of motion with depth in the soil (generally a reduction in amplitude) have permitted removal of these conservatisms and more realistic prediction of seismic demand.
The equipment walkdown and screening and relay evaluation procedures are based upon seismic experience and testing experience. For the case of anchorage evaluation, exhaustive studies have been conducted to develop inspection and strength criteria for concrete expansion anchors.
mkt223/jjj/papr696a
17 I <_>L I .\;>ci K-IICC D;il;i Base
SI re HI g-mot IDII earthquakes frequently occur in high seismic areas, such as Caliloi ma and Latin American countries, where power planl.s or industrial tacilitics arc included in the affected areas. By studying the performance of these earthquake-affected (or data base) facilities, a large inventory ol various types of equipment installations can be compiled that have experienced substantial seismic motion.
The primary purposes of the seismic experience data base are summarized as follows:
• To determine the most common sources of seismic damage, or adverse effects, on equipment installations typical of industrial facilities • To determine the thresholds ol seismic motion correspondinu to various types ol seismic damage • To determine the performance of equipment during earthquakes, regardless ot the levels of seismic motion • To determine minimum standards in equipment construction and installation, based on past experience, to assure their ability to withstand anticipated seismic loads
To summarize, the primary assumption in compiling an experience data base is that the actual seismic hazard to industrial installations is best demonstrated by the performance of similar installations in past earthquakes.
FACILITIES SURVEYED IN COMPILING THE DATA BASE
The seismic experience data base is founded on studies of over 100 facilities located in the strong motion areas of more than 60 earthquakes that have occurred worldwide since 1971
The data base was compiled through surveys of the following types of" facilities:
• Fossil-fueled power plants • Hydroelectric power plants • Electrical distribution substations • Oil processing and refining facilities • Water treatment and pumping stations • Natural gas processing and pumping stations • Manufacturing facilities • Large commercial facilities (focusing on their} 1 VAC plants).
In general, data collection efforts focused on facilities located in the areas of strongest ground motion for each earthquake investigated. Facilities were sought that contained substantial inventories of mechanical or electrical equipment, or control and instrumentation systems. Because of the number of earthquake-affected areas and types of facilities investigated, there is a wide diversity in the types of installations included in the data base. For the types of equipment
mki223/hj/papr6%a
18 of focus, this includes a wide diversity in age, si/.e, configuration, application, operating conditions, manufacturer, type of building, location within building, local soil conditions, quality of maintenance, and quality of construction.
The data base includes more than 60 earthquakes, usually with several different sites investigated in each earthquake-affected area. The earthquakes investigated range ii, Richter magnitude from 5.7 to 8.1. Measured or estimated ground accelerations for data base sites range from 0.1 Og to 0.85g. The bracketed duration of strong motion (on the order of 0. lOg or greater) ranges from 5 seconds to over 40 seconds. Local soil conditions range from shallow soil over rock to deep alluvium to rock. The buildings housing the equipment of interest have a wide range in size, and type of construction. As a result, the data base covers a wide diversity of seismic input to equipment, in terms of seismic motion amplitude, duration, and frequency content.
TYPE OF DATA COLLECTED
Information on each data base facility, its performance during the earthquake, and any damage or adverse effects caused by the earthquake were collected through the following sources:
• Interviews with the facility management and operating personnel usually provide the most reliable and detailed information on earthquake effects. At most facilities several individuals were consulted to confirm or enhance details. In most cases interviews are recorded on audio tape. • Facility operating logs are a written record of the conditions of the operating systems before and after the earthquake. Operating logs list problems in system operation associated with the earthquake and usually tabulate earthquake damage to the facility. Operating logs are useful in determining the amount of time the facility may have been out of operation following the earthquake and any problems encountered in restarting the facility. • The facility management often produces a report summarizing the effects of the earthquake following detailed inspections. These reports normally describe causes of any system malfunctions or damage, and typically include any incipient or long term effects of the earthquake. • If the facility can be surveyed immediately following the earthquake, as has been the case in many earthquakes included in the data base, earthquake damage can often be inspected prior to repairs.
Standard procedures used in surveying data base faciiities focus on collecting all information on damage or adverse effects of any kind caused by the earthquake. For a large majority of the facilities surveyed in the data base, this is not a lengthy task. Except for sites that experienced very high seismic motion, seismic damage to well-engineered facilities is normally limited to only a few items.
mk!223/jjj/papr696a
19 liarilu|iiake experience data procedures have been developed lor luenly classes ol equipment:
1. Motor Control Centers M. Low-voltage Swiiehgears 2. Medium-voltage Switchgcars 12. Transformers 1. Horizontal Pumps 13. Vertical Pumps 4. Fluid-operated Valves 14. Motor-operated and Solenoid-operated Valves 5. Fans 15. Air Handlers 6. Chillers 16. Air Compressors 7. Motor Generators 17. Distribution Panels 8. Batterv on Racks 18. Battery Chargers and Inverters 9. Engine Generators 19. Instruments on Racks 10. Temperature Sensors 20. Instrument and Control Panels and Cabinets
Beyond the twenty classes are others which were added to aid in the evaluations: cable trays, conduit, and raceway systems; tanks and heat exchangers.
IPEEE
Criteria for IPEEE have likewise been developed in parallel to the G1P but are applicable to seismic levels beyond the plant design basis. The USNRC has recently issued the Generic Letter, (USNRC, 1991a), and NUREG 1407 (USNRC, 1991b) for IPEEE. There are three methodologies which may be used.
• Seismic Probabilistic Safety Assessment (USNRC, 1983) • NRC Seismic Margins Method, (Budnit/. 1985 and Prassinos. 1986) EPR1 Seismic Margins Method (EPR1, 1988)
For all of the methods, the goal is to determine the seismic shaking level at which there is a high- confidence-of-low-probability-of-failure (HCLPF). This HCLPF is mathematically defined as 95% confidence of less than 5% probability of failure.
Seismic PSA
In the PSA method, fragility curves for essential equipment, piping and structures are defined as a conditional probability of failure versus a seismic input parameter (either peak ground acceleration or spectral acceleration within a defined frequency range). A seismic hazard is defined as a frequency of occurrence versus seismic input parameters (peak ground acceleration or spectral acceleration). The plant systems are modeled as event trees and fault trees from which Boolean equations are derived. Using the Boolean equations, the seismic hazard and the component fragility curves, the frequencies of core damage and release from containment can be derived. Figure 1 shows the seismic PSA process from the modeling and input parameters up through the analysis of the consequences of an accident. As a by product of the risk modeling, the plant level HCLPF can be computed from the Boolean equations and the fragility curves.
mki223/jjj/papr696a
20 I Ins computation delines llic dominant accident sequences lhat lead (o core damage and release and the IICI.PF for each.
It a PSA is elected to satisfy the J 1*11111-1, ii is only required that core damage frequency (Level 1 PSA) plus an evaluation of containment performance be performed. The computation of release frequency (Level 2 PSA) is not required but many utilities have elected to go to this extent. It ;., further stipulated that only a point estimate of core damage frequency is required. This involves the use of a mean seismic hazard prediction and a single mean fragility curve (Figure 2). The use of the full uncertainty spread of the seismic hazard and fragility curves was not required but. many utilities have elected to carry out this uncertainty analysis.
NRC Seismic Margins Method
The NRC seismic margins method was developed by USNRC contractors and is a truncation of PSA. The plant systems are modeled and seismic fragility curves are developed, just as in a PSA, and the plant level UCI.iM-' is computed. However, only the most important safety functions are considered. The frequency of core melt and release are not determined. In applying the NRC margins method, seismic capacity screening is conducted to eliminate many components from fragility computations. This capacity screening is based primarily on results of past seismic PSAs and on the successful performance of certain classes of equipment in past strong motion earthquakes.
The NRC seismic margins method involves the following steps (Prassinos, 1986):
• Selection of the Review Level Earthquake • Development of Systenis Models • Initial Component Ruggcdness Screening Plant Walkdown • Development o! Component and Structural/Fragilities • System Analysis • Determination of Plant Level 11CLPF
The procedure is virtually identical to the PSA procedure except that the systems analysis step does not involve the use of a seismic hazard for computation for core damage frequency. The systems models and fragility curves are used to determine the dominant accident sequences and the plant level 1ICLPF.
F-PRi Seismic Margins Method
A deterministic seismic margins method was developed by Electric Power Research Institute contractors and is very similar to the methodology contained in the GIP for resolution of USI A- 46. This similarity was deliberate to minimize the required activity to resolve both USI A-46 and IPEEE. In this method, safe shutdown paths are defined and components and structures in the safe shutdown paths arc dctcrministically evaluated to calculate component HCLPFs. The weakest component in a shutdown path then defines the plant level HCLPF for that path.
inkt223/|ij/p;ipr696a
21 1 he steps in the lil'RI seismic margins evaluation methodolouv are:
• Selection of the Review Level Earthquake • Selection of the Assessment Team • Preparatory Work Prior to the Walkdown • Success Path Selection • Seismic Capability Walkdown and Screening • Seismic Evaluation of Unscreened Components • Documentations
In this case, the success path selection must include a primary success path and an alternate success path utilizing to the greatest extent possible, different equipment. One of the paths must also have the capability to mitigate a small pipe break. The process is virtually identical to the A-46 process except that the alternate success path and the small break mitigation are additional requirements. Also, since the review level earthquake is specified to be beyond the design basis, all structures and equipment including piping, that are important to the success paths must be included.
Selection of Method
One of the above three methods was applied to all U.S. operating plants. The choice of method was determined by the review level earthquake specified for the plant, the utility desire to combine USI A-46 and IPEEE resolutions and the utility preference for methodology.
The plants have been placed into three review level earthquake (RLE) bins. Most plants are to be evaluated for a 0.3g RLE and have elected to do an EPRI seismic margin methodology evaluation, although some have elected to do PSA and a few are opting for the NRC margins methodology with the goal of expanding the margins evaluation to a PSA at some future date. There are a few plants which are placed in the 0.5g RLE level and most have elected seismic PSA for their IPEEE. Two California NPPs have RLEs exceeding 0.5g and arc required to conduct a PSA.
Even though the steps to perform tiie evaluation are summarized somewhat differently in the governing documents, all of the methods require similar procedures as does the resolution of USl A-46. The NRC has emphasized the integration of the A-46 and IPEEE programs for plants which must do both. Figure 3 compares the A-46 resolution process to the EPRI seismic margins process. Figure 4 compares the A-46 resolution process to the seismic PSA process. The NRC seismic margins process follows the steps in Figure 4 to the point of seismic risk quantification. At that point the margins process involves the computation of the plant level HCLPF using the systems models and fragility curves. As can be seen, the actual steps and scope of work are very similar.
Numerous seismic PSAs have been conducted in the U.S. prior to the IPEEE requirement. These PSAs will require enhancements, principally the performance of a detailed walkdown, the addition of equipment associated with containment performance and the use of more recent estimates of seismic hazard.
mkt223/jij/papr696a
22 Pilot studies that have been conducted using the margins methodologies include: the NKC method for Maine Yankee, (Ravindra, 1987), the HPRI margins melliod lor Catawba, (Campbell, 1989), and a combined IIPR1 margins and A-46 methodology for Plant 1 latch, (Southern Company Services, 1991).
Department of Energy Criteria
The U.S. Department of Energy has developed criteria for evaluation of DOE reactors and process facilities for natural phenomena hazards (UCRL, 1990). The criteria are structured with respect to the performance goals and risk inherent in the process. There are four categories of facilities. The performance goal for each category is based upon a frequency of occurrence of the event and the probability of failure, given the event. The most critical of the facilities has a seismic hazard defined for a very low frequency of occurrence, similar to that for defining the safe shutdown earthquake (SSE) for power reactors. The use categories and performance goals are shown in Table 1.
There is ongoing effort to update and expand the DOE criteria. In particular, the DOE has undertaken a program to develop a complete evaluation criteria parallel to the Generic Implementation Procedure for resolution of USI A-46. This procedure will be principally based upon earthquake and testing experience with supplemental analysts for anchorage and strength of supports. Several DOE laboratories are currently in the process of evaluating their major structures and equipment. Some test and production reactors have completed PSAs.
APPLICA TIONS OUTSIDE OF THE UNITED STA TES
Some past and ongoing projects outside of the U.S. have utilized the methods described above as full or partial resolution of seismic issues. Several PSAs which include external events have been performed for plants in Switzerland, Taiwan, Korea and Japan.
None of these PSAs have been conducted specifically to address seismic issues; external events have been a logical extension of the PSAs initiated to study internal event vulnerabilities. Some selected applications of the above described methodologies have been applied in Switzerland, Finland, Sweden, Belgium and Bulgaria. Finland and Sweden have low seismic hazard and as a result, the emphasis on seismic events is somewhat limited. At the Tihange plant in Belgium and the Beznau plant in Switzerland, the seismic design basis is similar to that of an eastern U.S. site. At the Kozloduy site in Bulgaria, the seismicity has been recently redefined and results in ground motion input levels about two times the previously predicted level. Several earthquakes have affected the site. The largest earthquake was in 1977 resulting in approximately a 0. Ig peak ground acceleration at the site which caused some structural damage. The currently specified earthquake for seismic reevaluation is 0.2g.
United Kingdom
The United Kingdom region is a region of low to moderate seismicity and the instances of damaging earthquakes are rare. In common with the worldwide trend of increasing safety standards, seismic design of new nuclear facilities is now the practice. For older power stations, major studies to evaluate the seismic capability have been initiated in support of the periodic safety reviews or continued operation safety cases. In the United Kingdom, nuclear power stations are licensed by the Nuclear Installations Inspectorate (Nil). The licensing regime is
mkt223/iij/p?pr696a
23 i ion-prescript i vc ruui the onus is on I lie opei a lor oi I lie plant to pi eseui the safety case \\ h ich meets the licensing coiulil ions. I he approach lo the, seismic evaluation of nuclear power stations not designed lor seismic loads has evolved lor more than a decade. The approach currently adopted, and beniii implemented lor the A(iR power stations and some of the older Maimox power stations, is outlined below, typically, the desnjn ol each power station is relatively individual and this had inhibited the adoption of uencne approaches. The seismic hazard at the site is determined by a site-specific study to determine the peak ground acceleration with a probability ofexecedance of 10 per year. The site ground motion is characterized by means of a uniform risk spectra. This probability of exceedance represents an infrequent initiating event. The objective of the seismic evaluation is to demonstrate the capability of a safe shutdown and post-trip coolinu path against the 10 ~ seismic event. I he buildings, plant, and equipment associated with, path are termed the Bottom Line Plant.
In order to demonstrate delense m depth a second, diverse path is assessed against a frequent initiating event which is a 0.1 n seismic event characterized by the UK ground motion spectral shapes derived in the early I9S0S. The building plant and equipment associated with this path are termed the Second Line Plant. The 0. Ig event corresponds approximately to a 10'"' event and is a convenient benchmark to compare between different plants. It also represents the minimum earthquake level recommended by the IAEA.
Safctv related buildums are sub|ected to dynamic analysis, taking account of soil-structure interaction where site conditions dictate. These analyses arc used to determine member loads and calculate in-sirueuue response spectra for subsequent plant and equipment evaluation. Major safety-related plan! items are subjected to structural analysis. For the assessment of mechanical ami electrical equipment in the Bottom Line and Second Line systems, seismic waikdowns are performed. The methods adopted are heavily based on the GIP, with some minor modifications lo suit UK conditions and practice. All the elements of the seismic evaluation arc drawn together in a seismic safety report which forms part of the Periodic Safety Review which is submitted to the NIL
Sweden
The older nuclear power plants in Sweden were not specifically designed to withstand earthquakes since Sweden has a relatively low seismicity. In the last several years, a number of studies have been conducted lo assess (he seismic hazard at the Swedish nuclear power plant sites and estimate the seismic margins of older plants.
The first study was to estimate the seismic margin of the mitigation systems at Oskarshamn Units I and 2. The mitigation concept developed by OKG to address the unlikely event of a severe accident at Oskarshamn consists of the filter vented containment and an independent containment spray system. These systems are designed lo meet seismic standards currently used in Sweden. Some of the components of these systems interface with the existing systems in Oskarshamn Units I and 2 which were not designed to current seismic criteria. A pilot study (Landelius, et a!., 1989) was performed lo assess the seismic margins of these interfacing systems and to verify that they would perform successfully in a major earthquake.
The Swedish nuclear industry and the regulatory agency funded an investigation - PROJECT SEISMIC SAFETY - to develop a characterization of seismic ground motions for probabilistic analyses of nuclear facilities in Sweden. The study (Engclbrcklson, 1989) has prodticed uniform
nikl223/jii/papi696a
24 hazard ground inolion spectra for hard rock and soft soil silcs al annual frequencies ol exccedance of 10"^, 10'" an Finland Imalran Voima Oy has performed a probabilistic safely assessment of l.oviisa Nuclear Power Plant. This PSA is explicitly treated the seismic events. Loviisa plant was not designed lor any specific seismic criteria. Therefore, the seismic fragility evaluation had to rely on seismic walkdowns and use of earthquake experience data in the development of seismic capacities of structures and equipment. An initial scoping study (Ravindra, Hardy and Hashimoto. 1989) identified certain components as needing further fragility evaluation. Imatran Voima Oy performed the seismic hazard assessment and the probabilistic response analysis to develop realistic floor spectra (Varpasuo and Puttonen, 1991). Using these spectra and based on the plant information and walkdown findings, the seismic fragilities have been developed for selected components (Ravindra, ct al., 1991). The results of the seismic risk analysis show a very low (4 X 10" /yr.) frequency of seismic induced core damage (Varpasuo, 1993). This low core damage frequency resulted from the very low seismic hazard prediction (—0.05g PGA at 10"7yr.). Teollisuuden Voima. Inc. (TVO) is performing a seismic probabilistic safety assessment of the BWR plant at Olkiluoto—two units of 710 M\V each. The objective is to verify seismic adequacy of the plants. This project was initiated in 1996 and is ongoing. The seismic PSA procedure follows those developed in the United States. Switzerland Seismic PSAs have been conducted for Bcznau, Gosgcn and Muelburg in Switzerland. The Beznau PSA was the first in Switzerland and was used to specify design requirements for a dedicated safe shutdown facility which has been added to each of the two PWRs. AH new equipment in the safe shutdown facilities has been seismically qualified by currently specified U.S. standards (ASME, IEEE, etc.). The piping and equipment in the containment and the steam and fecdwater piping outside of containment ahead of their respective isolation valves were required to be rcqualificd. In this rcqualificatiou program a variety of methods were utilized (Sahgal, 1990). Large bore piping has been evaluated to current ASME standards using dynamic analysis. Small bore piping has been evaluated using chart type screening methods based on ASME code stress allowables, selected dynamic analyses with increased allowables and lo some extent, using seismic experience based criteria. All equipment (valves, heat exchangers, tanks, piping penetrations) and systems interaction issues have been resolved using deterministic seismic margins methods which rely heavily on seismic experience based screening and selected calculations. This program has worked weli to apply practical, yet technically justifiable methods for seismic requalification of existing piping and components. In this program, the appiicabiiity of the seismic experience based screening criteria to European equipment had to be mkt223/nj/papr696;i 25 demonstrated. Only minor modifications have been required lo (.leiIUnisir.ilc ihe ability of existing equipment to withstand ihe safe shutdown earthquake. Belgium Die Belgian Utility, EJectrabcl, is a member ,>1 the Seismic Qualification Utility Group and has applied the G1P to the Tihange 1, 2 and 3 nuclear power plants. The Belgian work began before the finalization of the GIF. As reported in Lafaiile, 1990, the issues to be resolved with the Belgian authorities could mostly be addressed by use of the GIF methodology with some minor charmes in procedures and some additional study for equipment that could not be demonstrated to be represented by the earthquake experience database which forms the basis for G1P screening rules. This program appears to have been successful and was the first application of the GIP in Europe. Bulgaria Ko/loduy units 1-4 are Soviet designed VVER 440 model 230 I'WKs. In an initial IAEA mission, (Monette, et. al., 1991), a short walkdown was conducted and HCLPFs were calculated for the most seismically vulnerable components identified in the walkdown. The HCLPF calculations were based upon the fragility method. The principal reason for selecting the fragility method was the ability lo treat uncertainties regarding incomplete information as to the seismic input, structural response and equipment construction. The deterministic method requires that these parameters be defined in accordance with stated rules. Subsequent to this initial IAEA study, two follow on programs were simultaneously initialed. IAEA defined terms of reference for a VVANC) sponsored program to design priority seismic upgrades for Kozloduy I and 2. The scope of work for the terms of reference was developed based upon the prior IAEA mission and risk priorities derived from results of a top level risk assessment of Kozloduy 1-4 (BEQE, 1992). The program was defined in four phases and the first two phases have been completed. They included the evaluation and upgrade design for equipment anchorage, the diesel generator building, and the service water pump house. In addition, the main building, which consists of the reactor confinement, the auxiliary building and turbine hall, have been analyzed and in-structurc spectra have been developed. Phases 3 and 4 would include further evaluation and design of upgrades for the main building, evaluation and upgrade of the primary circuit, and a walkdown and experience based evaluation of piping and cable raceways. These phases have not yet begun. A very similar program funded directly by the plant was carried out for Kozloduy units 3 and 4 by local Bulgarian engineers with assistance from their U.S. counterparts. To date, most equipment anchorage and many masonry wall upgrades have been completed. Structural upgrades have not been initiated lo date. Upgrading of the unit 3 and 4 pumphouse is planned for the near future. The upgrade program must be compatible with outages and electricity demand, thus the design of structural upgrades must minimize outage lime. Slovakia There are four WWER 440s at Bohinicc, two models 230s and two model 213s. The plants were not originally designed to resist earthquakes, however, more recent seismic hazard assessments reveal (hat the hazard could be similar to that in Bulgaria. Major structural backfits have been mki223/i.ij/p;ipr696a 26 conducted by ('/.celt and Slovak ciiiiineers. A Western huropenn contractor lt;is been sclcclc Hungary There are four WWER 440 model 213s at Paks in Hungary. Several ongoing programs are addressing different aspects of the seismic issues. A unified criteria has evolved for applications at Paks which was merging of criteria being used by several contractors for structural evaluations and easy fixes of equipment and masonry walls. The criteria are a combination of ihc SQUG GIF, Seismic Margins Methodology and DOE standard for structures. The easy fix project for anchorage of equipment and stabilization of masonry walls have been completed. The phase 1 evaluation of piping and mechanical equipment is completed but actual backfit is not commencing pending further refinement in the site seismicity. The main building complex at PAKS consists of the reactor building and turbine hall interconnected by gallery buildings with each building on a separate foundation mat. The structures are a combination of reinforced concrete and structural steel frames with concrete infill panels. These structures were not designed for earthquakes, hence are potentially vulnerable to current estimates of the seismic hazard. Seismic evaluations of the existing structures are being conducted based on current seismic hazard estimates and upgrades are being developed for identified deficiencies. The criteria being considered for the design of upgrades is that defined in DOT, Standard 1020-94 with suitable restrictions placed on the ductility of existing precast structures. Slovenia The KRSKO NPP in Slovenia is a Weslinghousc PWR and was designed for a 0.3g peak ground acceleration. The site is quite seismically active and close in, high acceleration, low energy, earthquakes frequently occur. These close in low energy earthquakes are, however, not damaging to engineered structures and equipment. Recent seismic hazard studies show that a 50 percentile, 10,000-ycar return period earthquake producing low frequency damaging virbratory motion is about 0.4g, which exceeds the design basis. This is a similar situation to that in many U.S. plants where beyond design basis earthquakes must be addressed in IPEEE. The utility has elected to conduct IPE and IPEEE for the plant using the U.S. methodology. This work is being carried out by Western European and U.S. contractors. A consortium of Spanish utilities have been long standing SQUG members and are beginning now to implement their A-46 program. In an NPP in Taiwan, seismic experience was utilized to resolve seismic qualification issues. In this case though, the plant was a turn key plant that utilized almost all US manufactured equipment, thus there was little issue regarding the applicability of seismic experience. mkl223/jjj/papr696a 27 In ;i Japanese PSA, where specific qualification data were not available, the database was used in a few instances to develop fragilities. In cases where there was a lot of data, especially at high acceleration sites, a technique known as survival analysis was used 'o develop probability of failure of specific classes of equipment at increasing acceleration levels. In conducting this study, it was determined in a walkdown that the Japanese manufactured equipment was at least as rugged as database equipment. Since Japan is a country that experiences frequent severe earthquakes, most products manufactured have good seismic resistance by way of proper anchorage and attention to load path. Eartiiquake experience has been used on a limited basis to qualify a new diese! generator to be installed in a WER. This was acceptable to the authorities on the basis that similar projects had been conducted in the US for the qualification of Diesel Generator Systems. Conclusions Well defined criteria for evaluation of outstanding seismic issues in the U.S. have been developed and are rapidly being applied to existing power reactors, test reactors and nuclear facilities. Some limited applications of these methodologies have been made for European plants. In Europe, there is a wide diversification of regulating authority and seismic hazard at plant sites so it is unlikely that U.S. requirements for IPEEE US1 A-46 and DOE will be applied across-the-board for all plants in all countries. There is merit in selecting practical aspects of these methodologies and methods for application to specific issues. Some limited applications have been presented to demonstrate the applicability. Screening criteria associated with these methods must be used with caution. Some equipment in European reactors is not adequately represented in the experience database to confidently apply the U.S. screening criteria. While most equipment is genericaily rugged, there are some unique constructions that have been observed for which the screening criteria are cieariy not applicable without additional justification. In a SQUG training course held in Brussels, a trial plant walkdown was conducted on a Belgian fossil power plant. In this brief training walkdown. several instances were found where the intent of the screening criteria was not satisfied. In particular, the 6.3 kv switchgear contained ceramic parts. The GIP criteria limit voltage to 4.1 kv due to the fact that ceramics are often used in higher voltage switchgear. 6.3 kv is common in the Eastern European plants and a similar design with ceramic parts has been observed in Russian supplied switchgear. The seismic evaluation procedures as characterized by the GIP and earthquake experience data, in general, are becoming standardized by the IEEE and the ASME for electrical and mechanical equipment, respectively. It is anticipated that such procedures will become industry standards worldwide. mkt223/Ijjj/papr696;i 28 Table ] PERFORMANCE GOALS FOR EACH USAGE CATEGORY Performance Goal Performance Goal Annual Usage Category Description Probability of Excccdancc Genera! Use Maintain occupant safety 10"3 of the onset of major structural damage to the extent that occupants are endangered Important or Low Hazard Occupant safety, continued 5x!0 of facility damage to operation with minimal the extent that the facility interruption cannot perform its function Moderate Hazard Occupant safety, continued 10~4 of facility damage to the function, hazard confinement extent that the facility cannot perform its function High Hazard Occupant safety, continued 10"-* of facility damage to the function, very high confidence extent that the facility cannot of hazard confinement perform its function confinement mkt223/jjj/papr696a 29 Figure !: Risk Assessment Methodology for Seismic Events 1 A • 087s ^— -0.2S ^~ •u -0.35 / 06 ' 0.7S 95% Mean 3 Confidence 1 f I y ConfkhKK* ax / ^) i 0 I PEAK GROUND ACCELERATION (g) 2: Cotnponant FragMy Curves Figure 2: Component Fragility Curves mkt223/jjj/papr696a IS 30 R«v>ew Ptant A-46 Safely Systems a/KJ MoCify fP£ Event 4 FauM Tree* Sod iqudtacb T.. Study Develop O«v«top Safe Shutdown Compooems Ust Equipment i Coolainment Systems . Y _ . . Devetop Oevefop Median Perform Prooaoihsbc Median Ffoor -< * Flow Spectra Response Analyses Spectra lor SSE for RLE jerRi Teai Y Relay Chatter Fcx A46 SSEL. For Other Components Evaluation Perform a Follow GlP Follow EPRf perGIP Plant Wai and Coded Info Resoive Outbers Modify Components as Needed ...... Y. Develop HCLPFS Report on A46 Review Components. EsUmattf Planl A46 Review Seisrmc Margtn Completed ...... Y. ... Prepare 1P6EE Report ! Seis/TK fnputto UUtty IP€EE r Comp4«r*o Management 1 Figure 3: Seismic IPEEE Integrated Seismic Margin Assessment and A-46 Evaluation mkt223/yj/papr696a 31 Rrww Pl»nt A-48 Safety Sf5(«*ns ' £W«*T| & F«ul Trees ! So- >quel»cno T Sttxly Sale Shutdown Comptxvemj. Let Contawnent I j T Develop Flow Spectra and Pertwm Probab«ic(»c Sin^aursi Rezponse , espooie Aj**fyss Spear? fw SS£ Seyood Des^n Base . ...T. PCK A<6 SSEl. FooowCiP Plant Wafcdown and Cotfed Into l_ n Ou( Compootots From Total PRA Ltst _.. i Y Devf too £ecs.(n< Ptcpafe Repon Fi»QM>t1, of on A46 j- Fwl Tree*. I O«vetolop Sequence I EququalKKt; t I NRC Margin I PSA Quantificauon 1 T Seanicl>U(g«i 1 PRAOutpmi 1 1 1 KlSfQtta Report 1 i . _1. I T If 1 V£££ L tnpu( to UUMy 1 Figure 4: Seismic IPEEE Integrated Seismic PSA and A-46 Evaluation mkt223/jjj/papr696a 32 UKFKRKNCKS 1. Asfura, A.P., ct. al., 1991, "Pilot Study of Reactor/Containment Building: Oskarshamn 2 and Barscbcck I and 2, Probabilistic Response and Capacity", Report prepared for Svdkraft and OKC Akticbolag by UQU Engineering, Inc. and Westinghouse Energy Systems International. 2. BEQE, 1992, "Top Level Risk Study For Kozloduy Units 1 to 4," Prepared for the Committee on the Use of Atomic Energy for Peaceful Purposes by BEQE Ltd., April. 3. Budnitz, R.J., et. al., 1985, "An Approach to the Quantification of Seismic Margins in Nuclear Power Plants," Lawrence Livermore National Laboratory. NUREG/CR-4334. 4. Campbell, R.D. et. al., 1991, "Seismic Fragility Methodology for Evaluation of Liquid Metal Reactors." Structural Mechanics in Reactor Technology, Paper Ml 1(11)/1. 5. Campbell, R.D., et. al., 1989, "Seismic Margin Assessment of the Catawba Nuclear Station," EPRI NP-6359. 6. Electric Power Research Institute (EPRI), 1988, "A Methodology for Assessment of Nuclear Power Plant Seismic Margin," NP-6041. 7. Engelbrektson, A., 1989, "Characterization of Seismic Ground Motions for Probabilistic Safety Analyses of Nuclear Facilities in Sweden," Vol. Kl, pp. 37-42, Transactions of the 10th International SMiRT Conference. 8. Lafaille, J.P., et. al., 1990, "Experience of Seismic Walkdowns of Belgian Plants, Proceedings of Third Symposium on Current Issues Related to Nuclear Power Plant Structures, Equipment and Piping," North Carolina State University, December. 9. Landclius, M., M.K. Ravindra, G.S. Hardy, P.S. Hashimoto, 1989, "Seismic Margin Assessment of Mitigation Systems in Oskarshamn,'' presented at the I Oth International Conference on Structural Mechanics in Reactor Technology, Anaheim, California. 10. Merz, K.L., 1991a, "Generic Seismic Ruggedness of Power Plant Equipment", Prepared by ANCO Engineers for the Electric Power Research Institute, EPRI NP-5223. 1 1. Merz, K.L., 1991b, "Seismic Ruggedness of Relays", EPRI NP-7147, Prepared by ANCO Engineers for the Electric Power Research Institute. 12. Monette, P., R. Baltus, P. Yanev, R. Campbell, 1991, "Seismic Assessment of Kozloduy VVER 440, Model 230 Nuclear Power Plant," Structural Mechanics in Reactor Technology, Paper SD 006/5. 13. Newmark, N.M. and WJ. Hall, 1978, "Development of Criteria for Seismic Review of Selected Nuclear Power Plants," NUREG/CR-0098, May. 14. Prassinos, P.G., M.K. Ravindra and J.D. Savay, 1986, "Recommendations to the Nuclear Regulatory Commission on Trial Guidelines for Seismic Margin Reviews of Nuclear Power Plants," Lawrence Livcrmore National Laboratory, NUREG/CR-4482. inkl223/jjj/papr696a I 5. Ravindra, M.K., cl. al.. )987, "Seismic Margin Review of the Main Yankee Atomic Power Station." NUREG/CR-4426, Vol. 3, Prepared by EQE Inc. for Lawrence Livcrmore National Laboratory. 16. Ravindra, M.K., et. al., 1991, "Seismic Fragilities of Selected Components in Loviisa Nuclear Power Piant", Prepared for Imatran Voima Oy by EQE Engineering, Inc. and Westinghouse Energy Systems International. ! 7. Ravindra, M.K., G.S. Hardy and P.S. Hashimoto, 1989, "Scoping Study on Seismic Fragilities for Seismic Risk Analysis of Loviisa Nuclear Power Plant", report prepared for Iinatran Voima Oy by EQE Engineering, Inc. December 1989. 1 8. Sahgal, S., M. Culot, R. Campbell, P. Monette, 1990, "Application of Experience Based Methodology to the Seismic Qualification of Beznau Nuclear Power Plant," Third Symposium on Current Issues Related to Nuclear Power Plant Structures, Equipment and Piping," North Carolina State University. 19. Seismic Qualification Utility Group (SQUG) 1991, "Generic Implementation Procedure (G!P) for Seismic Verification of Nuclear Plant Equipment", Rev. 2. 20. Senior Seismic Review and Advisory Panel (SSRAP), 1991, "Use of Seismic Experience and Test Data to Show Ruggedness of Equipment in Nuclear Power Plants", Rev. 4. 21. Southern Company Services, 1991, "Seismic Margin Assessment of the Edwin I. Hatch Nuclear Plant, Unit I," EPR1 NP-7217, Prepared by Southern Company Services for Electric Power Research Institute. 22. U.S. Department of Energy, December 1992, DOE Order 5480.28 Natural Phenomenon Hazards Mitigation. 23. U.S. Nuclear Regulatory Commission, 1983, "PRA Procedures Guide," NUREG/CR- 2300. 24. UCRL, 1990, "Design and Evaluation Guideline for Department of Energy Facilities Subjected to Natural Phenomena Hazards," UCRL-15910, University of California, Lawrence Livermore Laboratory. 25. USNRC, 1991 (a). Generic Letter 88-20, Supplement 4, "Individual Plant Examination for External Events (IPEEE) for Sever Accident Vulnerabilities - 10 CFR 50.54(0- 26. USNRC, 1991(b), "Procedural and Submitta! Guidance for the Individual Plant Examination of External Events (IPEEE) for Severe Accident Vulnerabilities," NUREG - 1407. 27. Varpasuo, P. and J. Puttonen (1991), "Development of Probabilistic Floor Spectra for Loviisa Nuclear Power Plant," 11th International Conference on Structural Mechanics in Reactor Technology, Tokyo, Japan; mkt223/jjj/papr696a 34 28. Varpasuo, P., J. I'uUoncn and M.K. Kavindra, "Seismic Probabilistic Safety Analysis of LoviisaNPP, Unit I," Proceeding of Structural Mechanics in Reactor Technology, 12, Paper MKO5/3, August 1993. 29. Asfura, A.P. and Johnson, "Soil-structure Interaction (SSI): Observations, Data, and Correlative Analysis," In Proceedings of the NATO Advanced Study Institute on Developments in Dynamic Soil-structure Interaction, Kemei, Antalya, Turkey, July 1992. ' ' ' inkt223/jju/papr696a NEXT PAGE{S) left BLANK XA9952646 SMiRTl3 IAEA-WS August 1995 Iguasu, Argentin On Southern Hyogo-prefecture Earthquake and Some Related Activities in Japan Heki SHIBATA, Professor-Dr. Mechanical Eng'g. and Materials Sci. Faculty of Engineering Yokohama National University 156 Tokiwadai, Hodogaya, Yokohama 240, JAPAN Fax 81/45-331-6593 Abstract This paper consists of three parts. At first the reporter discusses on the earthquake event on January 17, 1995, and then on the summary of the report of examining the adequecy of the guideline of seismic design of nuclear power plants in Japan by the task group, Nuclear Safety Commission. And also on the activity of "the sub-committee on the research of seismic safety" for the future research subjects during 1996 ~ 2000 F.Y. Part 1 On the Southern Hyogo-prefecture Earthquake 1.1 Introduction The event occured at 5:46 am on January 17, 1995 was one of the most serious earthquake disasters since the Kwanto earthquake in 1923 in Japan. In the Kwanto earthquake, approximately 140,000 were killed in Tokyo and Yokohama area. Most of casualties were caused by the extended fires in Cities of Tokyo and Yokohama. The direct deaths are estimated as 10,000. In the earthquake disaster of this time, officially called as "Hanshin-Awaji Great Earthquake Disaster", approximately 5,500 were killed, and those were directly caused by structural failure mostly. One of the serious natural disasters, which caused such a big number of deaths is "Isewan Typhoon and High-tide" in 1959. 37 As seismic events, this earthquake is one of the greatest ones, may be once or twice every century through Japan. In this area, we experienced several destructive seismic events since 8 Q. but they might be not so serious ones according to historical records including 16 9 one: which is famous. Some one estimated its return period might be more than 5,000 years (in Fig.8). The characteristics of its ground motion was very unique, that is, very high velocity as well as high acceleration and short duration (Table 1). This caused many casualties and serious, very uneque structural failures. The feature of the ground motion, damage of various kind of structures, and other unique events will be described in the following chapters, which were presented at SMiRT 13 in Porto Alegre last August. Of course, this area is one of the most densely populated and highly industrialized areas. The western part of Kei-han-shin area, which is the second busiest area in Japan, is centered by City of Kobe. This is the reason why this event is called as Kobe earthquake. Officialy. Southern Hyogo-prefecture earthquake 1995 and the disaster itself is called as Hanshin-Awaji Great Earthquake Disaster. 1.2 Purpose of Field Survey What are the purposes of the survey on seismic damages? How do we learn the lessons from the seismic event. Usually we start immeadately after the event from sur- veying what are happening and how they are. For this, video is a very powerful tool now. The author started its recording a half hour later of the event when he recognized that the event had been very large scale event, and had been continuing for six hours. The major items and purposes are as follows : 1) How are there any phenomenon or failure mode which has been never observed:' In this case, they will be reported as examples in the field. 2) How adequate the models, which have been used for the design, are proven or verified through their behaviors observed? 3) How the statistics of structures and other systems are, for example, damage ratio of underground pipeline etc.? 4) How to recover structures and systems from their damaged state in a short range, hours and day, and in long a range, month and year? 5) How did or does the event give the impact to individuals and the society? 1.3 Facts on the Event Magnitude : M = 7.2 Focal Depth : H = 14 km 38 Time and Date : 05:46 , 1/17/1995 , JST Casualties : 5,500 deaths + 35,000 injures Damaged Buildings & Houses : App. 160,000 . Estimated Total Loss : $ 100 Billion Those numbers on damages are still moving 1.4 Fault, Ground Motion and Intensity The slips of related fault have been studied by seismologists, and various kind of new- facts have been being found. One of the features is high velocity ground motion. Japanese Meteological Intensity Seale exceeded VI in this event. This lias been the first event since it was denned after Fukui earthquake in 1948. The definition of Intensity VII is as follows : / = VII > 400 gal and Damage rate of buildings > 30 % There were several issues, for example, how to defined the damage rate and so on this time, but JMA decided that in the area of the center of Kobe City the intensity was VII. Following issues on their feature of ground motions which were made immeadiatly after the event, now some of them are clear, for example : How strong the effective PGA? Is the recorded maximum ground velocity reliable? Which faults did cause main shocks? The short duration of ground motion, is very unique as Japanese eartuquake. And initial several peaks were significant to cause structural damages as shown in Fig.l. and Table 1. Main shock consits of three shocks, and the waves were focused into the eastern part of the City Fig.2. Distribution of peak ground accelerations is shown in Fig.3. And their attenation curves are shown in Figs.4 and 5. Through the activities of the seismologists after the event, we feel the vecessity of the establishment of the engineering seismology for estimatimation of local ground motions, that is, micro-zoning. 1.5 Damage State of Various Kind of Structures i) Structures Highway Bridge : collapsing and sliding Building ; High-rise Building : less damage 39 Reinforced Conventional Building : large damages, three modes ; lower level collapse, one particular level collapse, and PS effect failure Steel Building : large damages, especially degraded one and brittle failure Wooden house : collapsing (and burnt) Piles ; failure by shear force mainly Pier and Embankment : side slip and subsidence Tower : less resonance-type failure in the area Tank : liquefaction effect, and sloshing in outside of Kobe area Comuter : overturn and cable failure Lifelines : various mode ii) Mechanism Single shock failure Resonance failure P-<5 effect Liquefaction Brittle Failure Mechanism of brittle failure has not been known well. However, the understanding on the phenomenon is diverged at each field as shown in Table 2. 1.6 Time-history and Damage The time history, that is, the partterns of ground motion of this event are very unique, and it has been proved by seismologists that there are exact reasons to induce such ground motions. The features of this event are quite different from other destructive earthquakes which have been recorded in Japan since 1880's instrumentally. The most serious destructive earthquakes, which were recorded in the past, are inter- plate type huge earthquakes and their epi-centers were in the ocean and their epicenter distances are usually more than 100 km. Durations of such earthquakes were over one minute in general. They induced resonance type failure to structures and this S-phase was followed by the surface wave period. Main points are as follows : i) Very short duration from initiation of ground motion to main peaks ; approximately 2 sec as shown in Fig.l and Table 1. ii) Similar wave form of acceleration to displacement : 40 This means, that the waves consist of rather simple component distribution, and it makes easy to analyse them from seismological view-points. Comparison of NS motions at Kobe Ocean Observatory to the images of Video, recorded at 10 sec ahead at the NHK office in the area shows the process of events as Table 1. The shaking tests of human on the table, demonstrate the strong effect to human body as well as to structures. The large amplitude of displacemet in a short duration may cause many unique features of this event, especially, many deaths. It should be mentioned that such a type of ground motions may be expected more in low seismicity area rather than high seismicity area, even though the probability of occurence might be low. 1.7 Response Spectrum and Particle Motions. Role of Vertical Component They have been discussed, but their direct effect has not known exactly yet. Some features are as follows : i) Dominant in longer period range, 1 ~ 1.5 sec in the response spectrum (in Fig.9 and 10) ii) Corelation of horizontal motions and vertical motions is recognized iii) Video-recordings at super-markets have a certain role for seismological study iv) Behavior of box-shape articles, over-turning, sliding and jamping As far as the second item, we feel the strong necessity of more study. 1.8 Damage of Industrial Facilities There were very serious damages of industrial facilities such as ship-builders, macline factories, habor facilities, plants, lifeline and so on. Most of them came from liquefaction rather than acceleration effect. Therefore, those were found more in areas near to the coast line compared to ordinary buildings and residential houses. 1) Effect of liquefaction and land-sliding : i) Overhead piping system above-ground deformation ii) Crane failure including largesize container crane iii) Settlement of heavy articles iv) Deformation of rails for O.H. crane and other traveling machines v) Ship-builders' yard and dock vi) Settlement of cylindrical tank vii) Lifeline, underground pipings viii) Switching station 41 2) Accerelation effect i) Railway train derailing ii) Power plant, boiler and pipings iii) Switching station and transmission line iv) Towers-type crane v) Tower for micro-wave transmission and other purpose 1.9 Damage Modes. Newly Observed The following modes were observed. Some of them, especially items ii) and v) are not known on their exact reasons. i) Land-slide in flat ground and coastal area ii) Brittle failure of steel column iii) One-direction shear failure of R.C. column iv) New-type cracking of brittle material v) Mid-story collapsing of building — propagation wave model vi) Partial failure of structure, and electric equipment and distribution system caused by other failed structural member or element vii) Overturning of buildings by P-£ effect, especially R.C. buildings viii) In some places vertical acceleration are higher than horizontal one, and both are corelated ix) Low acceleration near to surface fault and very high acceleration in regions where slighty for from the fault where surface fault is not found. Those items shall be studied intensively, and also their damage survey systems must be developed. 1.10 Standardization of Design Basis Earthquake Central Commission for Disaster Prevention, Chairman : Prim-minister, issued "the new fundamental principle for disaster prevention" on July 18, 1995 based on the facts which we experienced. In this article, two levels of counter-measures, including Design Basis Earthquake are stated. 1) For any events which are expected to occure several times through its life, it must be remaining as functional without significant damage. 2) For any rare event, human lives must be kept without serious damage. This comes from the original principle of our nuclear power plant design. There are approximately 40 seismic codes in Japan. These codes must be modified and unified 42 according to their fundamental concept and philosophy. Following items will be reviewed next two years. i) Factor of Importance ii) Concept of Zoning iii) Final Protection for the Safety iv) Standard Design Response Spectrum v) Vertical Ground Motion for the Design vi) Longer-period-range Ground Motion vii) Duration of Ground Motion viii) Measure for Rare Catastophic Event and its Level ix) Active / Capable Fault Protective Practice x) Liquefaction Protection xi) Land-slide Protection in Flat Area near Coastal Line xii) Land-slide Protection for Large Scale Slope xiii) Allowable Damage and Loss in Various Cases xiv) Insurance and Financial Back up by Official Organization xv) Mitigation of Social Impact 1.11 Liquefaction and Sliding Most of damages of industrial facilities were induced by liquefacion and sliding of embankment and pier. Soil, mixed with gravel and silt, caused liquefaction phenomenon by very high response of soil layer, and sliding ground toward the sea. The phenomenon of liquefaction was said to occure with only uniform granular sand. 1.12 Conclusion and Recommendation 1) This event, Kobe earthquake, is a really rare event. 2) There are very unique and unexpected features in the view points of seismological studies. 3) It is very difficult to predict its features. 4) Safety design must be done in consideration with two levels of events as described in the previous chapter. 5) The level of the severer event must be settled based on the knowledge on rare events. However, it is very difficult to develop it only by engineers, and they must ask the positive assistance of scientists. 43 6) The design for safety systems under seismic conditions is very significant, and it must be carefully examined in the sens^ of system dynamics. 7) P-6 effect of steel frame and ductile R.C. buildings were observed, and they are some different from ordinary ones. 8) The cooperation between seismologists and engineers including safety engineers must be encolleged. a) Continuous efforts to improve models for design must be done based on facts which we obser%'ed. Part 2 The-State-of-the-Arts of Seismic Design of Nuclear Power Plants in Japan. 2.1 Introduction After the event, the Nuclear Safety Commission organized a task group, chaired by Professor Kojima. This TG consisted of nine specialists including the reporter. They examined the following points to clarify that the Guideline for (examining) the Seismic Design of Nuclear Power Plants in 1981 NSC, which has been used for the regulatory purpose, is adequate: i) If a nuclear power plant is constructed according to the Guideline in Kobe Area, the design would be reasonable? ii) How S2 ground motions would be strong? iii) How the response spectrum of S2 would be conservative? Nine meetings and one field survey were made. 2.2 Design Basis Earthquakes in Japan Two levels of "Design Basis Earthquakes" have been employed in Japan since almost beginnings, 1960's. In the Guideline, Si and S2 are defined as follow : Si, the strongest earthquakes, and S2, the limit earthquakes. The second one may be interpreted as the upper-bound earthquake. Si earthquake may be the historical maximum earthquake in the site region. The historical records on destructive earthquakes have been kept since 5C, and those since 9 ~ 10$ can be listed in the seismic catalogue. The number of them from 416 to 1882, non-instrumental era listed is 267 and from 1884 to 1993, instrumental era, is 163. Some compensations to decide Si earthquake are made based on seismological knowledge, because, their return periods in some regions may be more than 1000 years, for which we can find historical written recods. And the periodical change of frequency of occurence might be observed in some area. 44 So earthquake is the upper-bound earthquake, whose magnitude M can be estimated in region by region as shown in Fig.6. According to the practice in Japan, the annual probability of occurence of a certain earthquake doesn't follow the stochastic relation in the stronger level, and there is the limitation as explained in Fig.7, and this value can be estimated by the seismotechtonic structure of the region (in Fig.6). However, level of intensity may be more diverged. 2.3 Comparison of Design Basis Earthquakes to the Kobe Earthquake According to the requirement of the Guideline. S2 should be as follows : M = 7| A = 7 km If we assume that a nuclear power plant in the area M95 = 7.2 and A95 = 16 km based on Japan Meteorogical Agency. Those values, officially reported, are formalized as the definition as to concentrate to one focus. The survey result by the seismologist the situation is more complicated. The focal distance Rg^ can be defined not so clearly. The above value was decided under consideration of such a situation, and the value based on original definition, it might be more than 50 km. A is the epicenter distance, and the focal distance is the distance to the focus of the event. However, the definition of a focus is the point which the initial slip of a fault movement had started. In the case of this event, its depth H is estimated as 14 km. S2 earthquake in the Guideline is defined in two ways : the maximum earthquake which might occure at the point estimated by seismological survey on active fault distribution, and that ; M = 6.5 underneath of the site, that is A = 7 km, H = 7 km. In this case, the former definction is applicable. 2.4 Response Spectrum of the Event Even though there are many records of strong seismograph in the area, those of rock site are limited. The site of a nuclear power plant in Japan must be rock site. Therefore, the records observed in the tunnel at the campus of Kobe University were only met to this requirement. These the design basis spectra, which are called as Ohsaki spectra based on approxi- mately 40 ground motion records in rock site observed in the world wide, are the standard spectrum which is recommended in the siipplimental explanation of the Guideline. Ac- cording to the response spectrum, in the lower frequency region, the response spectra of them are dominated compare to the design basis response spectra wliich are used for the design (in Figs 9 and 10). The comparison of this standard spectra to the response spectra of ground motions in the area was made, and the standard spectra are more conservative than the spectra of the event at the Kobe University except in the lower fre- quency region than 06 sec. Eigen-periods of significant structures, piping systems and 45 equipment are generally shorter than this value. This means that the standard spectra for the design are adequate in the view point of the margin. 2.5 Vertical Ground Motion It seems to be that many evidences of domination of vertical ground motions of this event. One of the reason comes from nonlinear characteristics of soil layer for transmission of S-wave, and linear for that of P-wave. The ratio of peak value of vertical ground acceleration to that of horizontal one was less than a half, 1/2, in general. Also, that of peak velocity and spectral intensity are the same situation. This means that the ratio used for the design is also adequate. Even though, we could find many facts of behaviors showing the vertical ground motion were dominated as mentioned above, and it might be induced by their corelation. Study on the corelation of both vertical and horizontal motions shall be necessary. 2.6 Remarks The analyses of key structures and items in a typical power plant based on the spectra had been made as a reference, and we could not find any critical issue. The report on the new basic proposal prepared by the Central Commission of Disaster Prevention pointed out that all kind of structures, not only nuclear facilities, must be designed in two levels, such as the concept of Sj and $2- The draft of the sped?! committee, the Science Council of Japan recommended that the equalization of the concept of seismic codes, whose number is approximately 40 kinds in Japan, must be made. In 1981, only the concept of zoning had been established except for nuclear power plants, however, now there are opinions that the activity of local active or capable fault must be considered in the codes of others. Also the continuous effort to improve the seismo-techtonic map like Omote map is necessary. Part 3. Future Researches on Seismic Design of Nuclear Power Plants in Japan As a part of the safety study program, the subjects and programs of seismic safety studies by various organizations will be reviewed every five years. For this, a sub-committee was organized under the chairmanship of Shibata, the reporter, and this belongs to the Committee on Nuclear Safety Research, NSC. In this year, next five year projects, 1996 ~ 2000 F.Y. have been reviewed. This job was started in October 1994, therefore, on the way of process the Southern Hyogo-prefecture earthquake occured. As a result, the sub-committee completed their Teport under the condition that the report will be reexamined in next one year, because there might be various topics arising based on facts which were observed in the event. Some of them are as follows : 46 i) How is the distribution of the maximum velocity of ground motions surrounding the fault line? ii) How is the map of the upperbound magnitude through Japan? iii) How is the peak ground acceration of vertical motions affected by local movement of the fault? iv) How is the structural response of buildings, piping systems and equipment to such vertical ground motions, and how is related to the horizontal motions? v) How is the brittle failure of ordinary steel using for piping support and others7 Those subjects would be discussed in future meetings with other subjects newly arising according to the impression of the reporter. And they will be discussed in other fields also. And more over, it might be necessary how the level of the serious but rare event shall be assume for the safety study. Acknoledgement The manuscript for the Proceedings of Post Conference Seminar 16 of SMiRT 13 which was held on August 21 and 22 at Iguazu Argentina was required to he conpleted by the end of November. Therefore, the author trys to rewrite it with new informations and his consideration based on his paper distributed at the CSNI meeting, OECD which was held on November 29 and 30, 1995 in Paris. And also this material was distributed at the IAEA-INS Workshop in Taejon, Korea, on December 6 and 7, 1995. The author greatly appreciate the advices of Dr. Tcbioha, JAERI and Mr. Miller, 9ECD: NEA secretriat, and related Task Groups. 47 Digital Memory Device for 10 second-ahead recording Video i = small articles (books) on "he desk are slightly moving. a = large motions of articles are observed. b = overturn of file-cabinet. c = TVset is droning from a rack. 900 «~:N00C£. AC1. r-MflW .900- Table 1 Sequence ofFailures Observed in Kobe Branch, Nippon Broad Casting Corp. by Video. Table 2 Mechanism of Understanding of Brittle Failures. * Phenomenon is understood for 1) Nuclear P.V. Specialist: It had been known in early '80s. 2) Ship Builders' Specialist: It has been steel doubtful on its mechanism. Some test results are known, and Tech. Comm. in WES. J. has been organized last two years. 3) Structural Engineering Specialist: It is now as a Subject Newly faced on. * Fact NDT Temperature shifts by plastic deformation in some case. Ductility factor, fi = 1.20, may push up the NDTT to Room Temperature in some cases. Temperature of the morning in Kobe, on Jan. 17, 1995, is estimated at 0 ~ 4°C source : Prof. H. Kobayaslii Tokyo Inst. of Tech. 48 T,..i;: :.«^i;; Fig.l Time History observed at Kobe Ocean Meteorogical Observatory. Hyogo 1995/01/17 t-\a - 2.5 i iO"25 dync-cr : 5.9 Dtp: Fig.2. Mechanisms of sequence of Main Shocks from Far-distance Record Analysis (AnaJized by Prof. Kikuchi, Yokohama City Univ.). 49 Unit gal Fi°;-3 Distrubution of P.G.A. 50 ; KSf.WCMIVA > - 1 me . ! SHIN- LJ t-J 30. C "•I i ! i i i J 1Q.C i_ I : : . . , I i--.-s. .1 C.2 0.5 1.0 2.0 5.0 10.0 20.0 SCO 1C0.C 200.0 500.0 DISTANCE fXM3 Fig.4 Attenuation Curve of P.G.A. f,'-:=7.0 2: 0.2 0.5 !.O 2.9 5.0 !0.0 20.0 50.0 1C0.0 200.0 5C0.0 OISTP.vCE (KM) Fi°;.5 Attenuation Curve of P.G.V. 51 Fig.6 Map of Distribution of Maximum Potential Earthquake in Japan DBE Concepl in Japan t-Historical Max. { o* ^ Now) S, S2 1 {CJ Design Basis Earthquake Fig.7 Relation of Si and S2 Design Basis Earthquakes To Probability of Occurrence of Seismic Event 52 9S- 9 iooa 99- 6 - 500 99- 5 1 200 - !00 ar. UJ O UJ UJ S 10 20 SO 100 200 ACCELERATION ( CflL ) 0.996? NON-EXCEEOBNCe 58. 500 I 1 1 1 • J i ' i-^ I i ' ' it 1 I Aoo\ — f— — — I 0 • I 1 0- 95 ~ I I -. 1 ^^ 1 200 CL 9 >"- 0- 8 I 1 ^"^ ' r~ -\ g _ I 1 ^s _ I - jS^ ->^ ^^ 1 1 UJ i > 20 i i_ i_ i i •^%^ i i i t >^ 1 i i 10 l_ I 1 1 - j 1 1 1 - 1 I 1 1 - - i I 1 1 - III!. 1 1 1 I 1 ! 1 , 1 < 10 20 50 100 200 500 r vcoo i Fig. 8 Return Period of Earthquake vs. Estimated Acceleration in Kobe area 53 FREQUENCY 0 (h-0. 05> 2000 1000 - 8 0.01 0.1 PERIIJO (SEC) 0.1 Fig. 10 Comparison of Response Spectrum, Kobe v.s DBE (Scimi-log) 0. 02 PERIOO (SEC) Fig. 9 Comparison of Response Spectrum, Kobe vs DBE (log-scale) PROCEEDINGS OF SMiRT-13-POST CONFERENCE SEMINAR-16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES LATUR EARTHQUAKE AND ITS IMPACT ON THE ASEISMIC DESIGN OF STRUCTURES IN INDIA P.C. BASU XA9952647 Atomic Energy Regulatory Board, INDIA ABSTRACT : The Latur earthquake occurred on September 30, 1995. The epicentre was located near the Killari village of Latur District which is situated in the stable continental region of Southern Peninsular India. The earthquake caused a wide range of damage though its magnitude (MS) was 6.4. Intensive damage survey was carried out and a number of geophysical and seismological studies had been undertaken. It was been concluded from the results, available so far from these studies, that the hypocentre of the earthquake was on the lineament dipping NW-SE. The rock matrix in the hypocentral region was weakened due to the presence of fluid and rupture of this weak region caused the event. The ground motion produced by the earthquake was of complex nature comprising of horizontal and vertical component. The ground acceleration in the epicentral region was estimated as 0.2 g. Latur earthquake raised several issues with respect to aseismic design of structures in India which need further deliberation. These issues are related to seismic zoning of India, determination of design basis ground motion, design/ detailing of structures, etc. INTRODUCTION Latur Earthquake occurred in the early morning of 30th September, 1993. The epicenter of the earthquake is located in the Latur District of Maharashtra State, India. The epicentral area falls in the Southern Peninsular India known to be seismically stable. The event was preceded by several foreshocks and followed by a number of aftershocks. Field studies and damage surveys were carried out by a number of organizations following the earthquake. The Latur earthquake would go down in the history as the deadliest earthquake to strike a stable continental region (SCR). No significant past earthquake incidents are known in the region surrounding Latur and as such no earthquake instruments were installed in this area before this earthquake. However, a number of geological, geophysical as well as seismological studies have been undertaken in the earthquake affected regions following the event. All these studies are yet to be 55 completed but results from some of these studies, have been published. The purpose of present paper is two fold. Firstly, to present an account of the event of Latur earthquake on the basis of data/information collected from the published literature and the interview/discussion with people. Secondly, to deliberate on the impact of this event with respect to the practice followed in India for aseisxnic design of the structures. THE EVENT OF LATUR EARTHQUAKE Main Shock [l TO 6] * Origin time : 22:25:53 GMT, September 29, 1993. 00:03:53 1ST, September 30, 1993 Epicentral : 18.07°N, 76.62°E location Near Killari Village of Latur District of Maharashtra State in India. Hypocentral : < 10 Km (5-10 km) Depth Magnitude : Body wave (Mb) = 6.3 Surface wave (Ms) = 6.4 Richter (Mw) = 6.1 Duration : 30 to 40 seconds Seismic Moment : In order of 10 dyne.cm. (This is about 20% of annual seismic energy released by stable continental region earthquake) (Different values for the above parameters have been published in different literatures) Epicentral Track [5, 14] The topology of the affected area is mildly undulated. The bed rock is covered by black cotton soil mantle having average thickness of 300 mm. There are number of flat topped mounds. The rivers Terana, Manjra and Bhema flow through the affected area. The village Killari which is nearest to the epicentre is near to tributary of river Terana . Fig.2 depicts the epicentral track of the earthquake. * Note: Numerical value inside the square bracket indicates reference number. 56 Foreshocks [1, 6, 7] A number of foreshocks during October-November, 1992 were reported. An earthquake of magnitude 4.0 occurred on 18th October, 1992 in the vicinity of Killari causing minor damages. There were about 25 felt earthquakes in the following week. On November 1 and 2, 1992, four (4) earthquakes of magnitude 2.2 to 3.8 occurred. Felt tremors were also reported in 1962, 1967, 1983 and 1984. aftershocks [6, 7] The main shock was followed by 187 (one hundred and eighty seven) aftershocks, largest being magnitude 4.4 occurred on 30the September, 1993. The aftershock activities decreased rapidly but a shock exceeding magnitude 4.0 occurred on November 12, 1993. The aftershock zone is clustered near the point of confluence of the two tributaries of the Terana river south of Killari village (see Fig. 1). A detail account of the foreshocks and aftershocks are given in reference - 7. Probable cause of earthquake [3,6,7,8,9,10] Indian plate is moving north word direction [9]. Cluster of earthquakes have been observed due to this movement along the main thrust zone below the Himalayan mountain range on the boundary of this plate. Peninsular India is a stable continental region on the Indian plate within which the epicentral area of Latur earthquake falls. The region comprising of the Deccan volcanic province (DVP) and a craton. Eight (8) major earthquakes occurred in the stable continental region of India. Fig.l shows the tectonic map of this stable continental region and the locations of these eight earthquakes. However, no major earthquake event was recorded in the epicentral region of Latur during the historical past. A number of geological, geophysical, seismological studies were started following the earthquake [11]. In some literatures [1&2], question was raised whether the Latur earthquake was induced by reservoir, because of the presence of Terana river in the epicentral track. Rastogi examined this aspects from various considerations and concluded that the Latur earthquake is not a reservoir triggered one [12]. Chetty and Rao studied the lineament pattern of Latur - Osmanabad area [13] and summarised that the area is confluenced by the lineament system as shown in Fig. 3. The results so far obtained from the geophysical studies indicate the presence of fluid in the hypocentral region [10]. It has also been revealed from these 57 investigations that the hypocentre of the Latur earthquake falls on the lineaments dipping NW-SE and passing around the tributaries of Terana river near Killari village; and due to the presence of fluid in the hypocentral region, the rock matrix was weakened and rupture of this weak region caused the event. It also appears from these studies that the event occurred on the plane striking at about 45° NE-SW on fault dipping at about 13 5° NW-SE. Damage Survey The area within 6 km radius of epicenter on either side of the river Terana was almost completely destroyed. The most affected villages were Killari, Talni, Mangrul, Sastur, Hulli, Rajegaon and Yekundi. The extent of damage was gradually lessened with the distance from epicentre. Iyengar et al generated damage intensity map (see Fig.4) in UN scale [5]. The earthquake completely damaged about 19,000 houses and partially damaged 200,000 houses in 67 villages killing about 11,000 persons [ 4 ]. No major failure associated with land-mass such as liquefaction, landslides, subsidence, uplift, etc. except a surface rupturing of about 3 0 mm wide and 3 km long near Killari was reported. Jain [1] and Chetty et al [13] mapped this rupture [see Figs. 5&6]. Strange incidences like smell, sound/noise, unusual behaviour of animals, phenomena associated with the variation of ground water level, etc. were reported [14]. The outcome of the examinations on these incidences are given in references - 15 & 16. The preliminary data obtained from the field reconnaissance of ground water indicates that there was no visible impact of the earthquake on the ground water regime. The gas emanation was reported upto a distance of 2 00 km. The onset of volcanic activities as reason for this gas emanation has been ruled out. The gas emanation in all probability were due to release of trapped gases as a result of earthquake. Subterranean sounds/micro earthquakes and ground cracks were reported in nearby regions within about one month of the main shock. This could be attributed to the stress re-distribution caused by main shock which triggered micro earthquakes. As such, the earthquake did not create significant environment disorder. The structures (buildings, public utilities, industrial facilities, etc.) behaved in expected way during earthquake [4, 5, 14 & 18]. Non-engineered conventional buildings suffered complete collapse to very severe damage making them non-usable. Almost all the conventional buildings of stone wall with mud mortar collapsed in the epicentral area as their resistance 58 against lateral force was very little (see Fig.7). Though the stone wall buildings with sand cement mortar (see Fig. 8) or those with wooden frame to support roof (see Fig. 9) responded in a better way than the stone wall with mud mortar but they also became unusable. The engineered structures responded in expected way and suffered generally acceptable pattern of damage. Performance of reinforced structures, in general, was good except one water tank (see Fig. 10). It was reported that the reinforcement detailing of this tank might not be in accordance with the accepted practice. Iyengar et al observed that the earthquake was felt more widely and severely to the South of epicentre than to the North [5]. They developed the damage intensity map (UN Scale) based on damage survey. An intensity of VIII has been assigned to the mizoseismic area. Another interesting feature is that intensity VII had also been observed in Unrani and Bhosgak which are at a large distance from the epicentral area (see Fig. 4). This may be due to the local geological and sub surface condition. Ground Motion The Latur earthquake was reported to be felt in large area, even at a distance of about 800 km from the epicentral region. This is an interesting feature of an earthquake of magnitude in the range 6.4 and is attributed to two reasons; firstly, the focal depth is shallow and second one is the efficient wave propagation characteristics of the rock strata of the shield region. No strong motion records are available of this earthquake. Based on field survey, interviews of local residents, Sinval et al reported that the ground motion generated by the earthquake was of complex type [14]. The possibility of vertical and torsional component alongwith the horizontal component, generated by the earthquake, was inferred from the information obtained from the field survey, the pattern of damage and deformation suffered by the structures, etc. Sliding of heavy objects were also reported [5,14], Iyengar et al estimated ground acceleration near the epicentral area as about 0.2g based on the information regarding sliding, tilting and toppling of objects (see Table-1). IMPACT ON ASEISMIC DESIGN OF STRUCTURES The Latur earthquake reveals a good deal of information regarding the earthquake potential of the stable continental regions of Peninsular India. The event has raised a number of issues related to the aseismic design of structures in India. These are related to the seismic zoning of India, method to 59 determine the design basis ground motion (DBGM) and design of structures. Some of the present practice related to these activities are examined in the following sections vis-a-vis the various data/information obtained from the field study and other investigation carried out following the Latur earthquake. Present Practice of Aseismic Design in India 1. General Buildings/Structures In India, earthquake resistant design of general buildings/structures are carried out as per "Indian Standard, Criteria for Earthquake Resistant Design of Structures", IS:1893-1984 [17]. In the standard, it has been endeavored to ensure that, as far as possible, structures are able to respond, without structural damage to shocks of moderate intensities and without total collapse to shocks of high intensities. The structures, which are designed as per this standard, are expected to experience more severe ground motion than the one envisaged in design during its life time. In view of this, the standard advocates to adopt the ductile design considerations. The standard divides entire India in five (5) seismic zones based on the probable earthquake intensity (see Fig.11). Table-2 contains the assigned intensity and earthquake potential of these zones. The standard principally deals with the horizontal motion in the detail. However, vertical motion needs to be considered when the stability is main criteria in the design. The standard specifies that the seismic coefficient for vertical motion should be 50% of that for horizontal motion. As per IS-1893, the horizontal seismic co- efficient (a^) , the design parameter of earthquake excitation for which structure is to be designed, is determined from the following expression : ah = 61.a (1) where, the values of £ depend on soil foundation systems and varies from 1.5 to 1.0, while the range of I (importance factor) is given in IS1893 as 1.0 to 3. The value of ground motion factor, xa', varies with seismic zones. (i) Seismic Coefficient Method In this method, xa' is the basic horizontal seismic coefficient (aQ), the values of which for different seismic zones are given in Table-2. 60 (ii) Response Spectrum Method In this method, the structural response is determined by dynamic analysis using response spectrum method based on modal super position technique- The standard provides an average spectra (see Fig. 12) and the horizontal seismic coefficient is calculated from equation (1) for which 'a' is given by a = Fo Sa/g (2) Where, Fo and 'Sa/g' are the seismic zone factor for average spectra (see Table-2) and average spectral ordinate (ref. Fig. 15) respectively. 2. Nuclear Power Plants (NPP) In India, the nuclear safety related structures are designed presently for site specific DBGM. The DBGM are evaluated for two levels of severity. The severe earthquake level is termed as SI (OBE) and the extreme level is designated as S2 (SSE) . The DBGM of a given site is determined following the guidelines laid done in "Safety Guide on Seismic Studies and Design Basis Ground Motion for Nuclear Power Plant Sites", AERB/SG/S11 [18]. In general, peak ground acceleration (PGA) of a site is calculated following deterministic approach and considering the geological, geophysical and seismological information of the region within 3 00 km radius from the site. This region is divided into a number of seismotectonic provinces. The maximum earthquake potential of each fault in a province is evaluated. The maximum earthquake potential of a fault can be determined either from historical earthquake data (adding one intensity to the estimated/recorded intensity of reported earthquake) or by evaluating the maximum earthquake capability of the fault considering its tectonic characteristics. However, the peak ground accelerations of SI and S2 level earthquake in horizontal direction should not be less than 0.05g and 0-lg respectively. Ground acceleration in vertical direction is desirable to be determined following the same procedure as used for determination of the acceleration in horizontal direction. In lieu of this, AERB/SG/S-ll also allows to consider the vertical excitation as 67% of the horizontal excitation. The design response spectrum is derived from an ensemble of acclearograms recorded on similar sites and covering broad range of source and transmission path characteristics. Design time history is generated from this design response spectra ensuring that the time history is compatible to it. 61 Seismic zoning 1. General Buildings/Structures The seismic zone map of IS: 1893 has been developed primarily on the basis of available strong motion seismic data/information. The seismic zone number of a region is the representation of tht size of earthquakes (magnitude or intensity) which could occur in that region. Potential of earthquake primarily depends on the strain of rock. The work of Gaur and others [9] indicates that strain is accumulated in the crust of peninsular India at a slow rate. For rupture of rock, threshold value of strain is in the order of 10 . If the rate of strain accumulation is in the order of 10 to 10~7 (as in the case of Southern part of peninsular India; see ref .9) , the return period of a typical moderate to higher size earthquake in this region may be taken in the order of 10 to 10 years. Therefore, current potential of earthquake in any area of this region depends on the value of accumulated strain till date. Again, if there is any weak spot due to the presence of fluid or due to any other reason in a lineament, the rock mass could rupture at lower threshold value. Bureau of Indian Standard (BIS) has taken up a programme to revise the seismic zoning map of India. The seismic zoning map would be revised considering the earthquake potential of different regions in India and taking into account the geological, geophysical and seismological features of these regions alongwith strong motion as well as micro earthquake data. 2. Nuclear Power Plants Incidentally, no Indian NPP or sites of ongoing nuclear power projects are situated within 300 km radius from the epicentral area of Latur earthquake. Since, the site specific DBGM of NPP-Site is determined conservatively from the maximum earthquake potential of all causative faults (falling on the area within 300 km radius from the site) , change in seismic zonation of India will not have any impact in the engineering of Indian NPP. Ground Motion 1.0 General Buildings and Structures It has been inferred from the field survey that the ground motion generated by Latur earthquake contained horizontal, vertical and possibly torsional component [7, 14]. It may not be possible to infer the existence of torsional motion just on the basis of 62 deformation/damages of structure, but possibility of vertical motion could be inferred with reasonable certainties from these information. IS:1893, principally, deals with horizontal motion in detail and apparently attaches less importance to the vertical motion. The information revealed from Latur earthquake clearly indicates that both of vertical and horizontal motion should be treated with equal importance in aseismic design. The range of ground acceleration (horizontal) calculated is 0.023g to 0.006g considering different values of 5 and I for regions falling in seismic zone-l. The ground acceleration in epicentral area of Latur earthquake has been estimated as 0.2g using the information collected after the event. Therefore, if a structure is designed for earthquake excitation as per IS1893 in the epicentral area of Latur earthquake, then it needs to have ductility factor 8.69 to 33.33 (corresponding to PGA 0.023g to 0.006g) to withstand 0.2g ground acceleration. It is presumed that when a structure is designed and detailed following the codal criteria of ductile design [21] and good engineering practice, the ductility factor of about 5 could be achieved. Demand of ductility factor 8.69 may be acceptable with a view that structure could be allowed to suffer certain level of damage. But demand of ductibility factor 3 3.33 is too high. It indicates that there exists possibility of underestimation of the design value of seismic effects by IS1893 in some cases and there is a need to re- examine this respect. 2.0 Nuclear Power Plants The PGA values are calculated for the areas around Killari village from maximum reported magnitude of earthquake as per AERB/SG/S-11. The maximum magnitude earthquake occurred before the September 30, 1994 event was felt 4.0. For various combinations of epicentral distance and hypocentral depth, PGA values are calculated with this information and are found to be less than 0,2g. The maximum value is 0.104g when epicentral distance of 5 km and hypocentral depth of 5 km are considered. For other combination of epicentral distance and the hypocentral depth, the values are less than O.lg. For the epicentral distance and the hypocentral depth both taken upto 10 km, the calculated PGA values are even less than 0.05 g in some cases. The PGA values were also calculated for different combinations of epicentral distance (upto 2 0 km) and hypocentral depth (20 km) for magnitude upto 7.5. It is found in number of cases the PGA value works out to be less than 0. 2g and in a few cases it is even less than 63 0. lg when magnitude is taken less than or equal to 6.5. In this context, it may be noted, out of eight major earthquakes occurred in the region of peninsular India, magnitude of seven earthquakes were less than 6.5. The magnitude of the largest one was 7.79 which occurred in 1819 at Kutch region. The above results indicate that it may not always be conservative to determine the PGA only on the basis of past earthquake data only, specially, for the site falling in a stable continental region. Design and Detailing of Structure The principal failure modes of concrete structures which are generally addressed in the design of a reinforced concrete structure are flexure, shear and bond, while for steel structure they are flexure, shear, buckling. The joints are also very important aspect for steel structures or pre-fabricated structures for efficient aseismic design. Except flexure, all other failure modes are non-ductile for concrete structures. Structural design should eliminate the possibility of non-ductile failure modes. For this purpose, ductility should be implemented in all design implicitly and when this is warranted, explicit implementation should be made [22]. The behaviour of different engineered brick masonry buildings and RCC structure with aseismic design feature during Latur earthquake again confirms that present state of aseismic design of structure can attribute them the capability to withstand the effect of earthquakes in intended way even if the seismic excitation is more than the one accounted in the design. SUMMARY 1. The Latur earthquake occurred in the stable continental region of peninsular India comprising of the Deccan Volcanic Province (DVP) and a craton. According to seismologist, this earthquake of magnitude above 6.0 in a stable continental region associated with craton is a rare event. 2. The earthquake had shallow focus and while the damage area was comparatively small but the felt area was very large. This is attributed to efficient propagation of seismic wave in the shield region. High number of death toll was due to the fact that the earthquake occurred in the early morning hours when people were fast asleep. Moreover, large damage of buildings is also responsible for this. The buildings/structures 64 responded in expected manner. Most of the buildings in the affected areas were of conventional type construction and were poor in resisting lateral forces and thus suffered severe damage or collapsed. Appropriately engineered structures with earthquake resistant capability withstood the earthquake in intended way. 3. The ground motion produced by the earthquake was of complex nature comprising of horizontal and vertical component. Presence of torsional component in the ground motion was also reported. The ground acceleration in the epicentral region has been estimated as 0.2g. 4. Following conclusions are drawn from the examination of some of the aspects of present methodology of aseismic design followed in India. i) The revision of seismic zoning map needs to be based on the geological, geophysical and seismological information along with the strong motion as well as micro earthquake data. ii) The present" methodology of IS1893 to determine the design parameters of earthquake ground motion for aseismic design of structure may underestimate the value of design ground motion parameter which may not be the intent of IS1893. The method needs to be re-examined. iii) Ductility should be implemented in all design implicitly. When required explicit implementation of ductility is to be made through design methodology. iv) A minimum level of ductile detailing should be specified for the all buildings/structures irrespective of their location of construction with respect to seismic zone. v) Appropriate detailing needs to be adopted, as a first step, to cater for possible torsional motion generated by earthquake. vi) Both the vertical and horizontal excitation needs to be taken care in the design assigning equal importance to each. vii) Evaluation of maximum earthquake potential of faults only on the basis of past earthquake data may not give conservative design basis ground motion of a NPP site in a stable 65 continental crust region. The maximum potential shall also be evaluated considering the capability of the fault from geological, geophysical and seismological information/ data. ACKNOWLEDGMENT Author wishes to express his sincere thanks to Dr. S.K. Guha, Joint Director (Retd.) of CWPRS, Pune, Dr. S.K. Arora, Head, Seismology Section, BARC, Dr. R.D. Sharma, Head, Seismotectonic, NPC and Dr. S.K. Jain of IIT Kanpur, for providing data and information. Author had discussion with a number of persons on Latur earthquake and it is not possible to quote all of their names. Their help and co-operation are also thankfully acknowledged. He is also grateful to Chairman, AERB for allowing him to publish this paper. REFERENCE 1. Jain S.K., Murthy C.V.R., Chandak N., Jain N.K., "The September 29, 1993, M6.4 Killari, Maharashtra Earthquake in Central India", EERI Special Earthquake Report, pp. 1-8 January, 1994. 2. Seeber Leonardo, "Killari, The Quake that Shook the World", New Scientist, 2nd April, 1994, pp. 25-29. 3. Arora S.K., "The Killer Earthquake of Killari", Deep Continental Studies in India, Vol.4, No. 2, September, 1994, Sponsored by Department of Science & Technology, Government of India, pp.2-4. 4. Indra Mohan, Rao M.N., "A Field Study of Latur (India) Earthquake of 30th September, 1993, "Memoir Geological Society of India, No. 35, pp. 7-32, Bangalore, India 1994. 5. Iyengar R.N., Manohar C.S. and Jaiswal, "Field Investigation of the 30 September Earthquake in Maharashtra", Current Science, Vol. Nos. 10th September, 1994, pp 368 - 379. 6. Gupta H.K., "The Deadly Latur Earthquake", Memoir, Geological Society of India, No. 35, pp. 1-5, Bangalore, India, 1994. 7. Baumbach M. , Grosser H. , SCHMIDT H.G., PAULATA, Rietbrock A., Raraakrishna Rao, C.V., Solomon Raju P., Sarkar D. , Indra Mohan, "Study of the Foreshocks and Aftershocks of the Interplate Latur Earthquake of September 30, 1993, India", Memori, Geological Society of India, No. 35, pp. 33-63, Bangalore, India, 1995. 8- Bolt Bruce A, "Earthquakes," WH Freeman and Company, New York, 1988. 9. Gaur V.K. , "Evaluation of seismic hazard in India towards minimizing earthquake risk", Current 66 Science, Vol.67, No.5, 10th September, 1994, pp. 324-329. 10. Gupta H.K., "Latur Earthquake : Some Results of Geophysical Investigations", Deep Continental Studies in India - Newsletter, Vol.4, No.2, September 1994, sponsored by Department of Science & Technology, Government of India, pp. 4-8. 11. Midha R.K., "Latur Earthquake Some Major initiative by DST", Deep Continental Studies in India - Newsletter, Vol.4, No. 2, September 1994, sponsored by Department of Science & Technology, Government of India, pp. 8-10. 12. Rastogi BK, "Latur Earthquake: Not Tiggerred", Memoire of Geological Society of India, No. 3J, pp 1-5, Bangalore, India. 13. Chetty TRK, Rao MN, "Latur Earthquake of September 2 993; Surface Deformation and Lineament Pattern", Geological Society of India, Mcmor 35, Bangalore, India. 14. Sinbhal A, Bose PR, Dubey RN, "Damage Report of the Latur-osmanabad Earthquake of September 30, 1993. 15. Singh V.S, Subrammanyam, Hodlur GK, Angareyulu, "Report on Hydrological Reconnaissance in some EArthquake affected villages in Latur and osmanabad Districts of Maharashtra", Memoir No. 35 of Geological Society of India, pp 131-138, Bangalore, India, 1995. 16. Rastogi BK, Rao MN, "After Effects of Latur EArthquake Smoke/Gas Emanations and sub terraneam Sounds/Microearthquakes", Memoir No. 35, Geological Society of India, pp 139-149, Bangalore, India, 1994. 17. BIS, "Indian Standard, Criteria for Earthquake Resistant Design of Structures", IS:1893-1984 (Fourth revision) , Bureau of Indian Standard. 18. Sinha R, Goyal A, "Damage to buildings in Latur Earthquake", Current Science, Vol. 67, No.5, 10th September, 1994. 19. AERB, "Seismic Studies and Design Basis Ground Motion for Nuclear Power Plant Sites," Safety Guide No. AERB/SG/S-11, Atomic Energy Regulatory Board, India. 20. Guha S.K., Basu P,C, "Catalog of Earthquakes (=> m 3.0) in peninsular India", AERB Technical Document No. AERB/TD/CES-1, 1995, Atomic Energy Regulatory Board, India. 21. BIS, "Indian Standard Code of Practice for Earthquake Resistant Design and Construction of Buildings", IS 4326, 1976, Bureau of Indian Standard. 22. Basu P.C., "Ductility Criteria, A Part of Integrated Design Approach in Proposed REvision of IS456: A Deliberation", paper presented in the seminar on Revision of IS 456, September 4-6, 1992. 67 TABLE - 1 Field Observations of Overturned/Tilted Object [5] Place Distance District Descripti on Horizontal Killari and accelerat ion (km) Taluk to initiate tilting ( g ) Killari Latur Roadside boundary stone near post office 0.18 Ausa Roadside marking stone near main square 0.24 Kavatha <5 Osmanabad Idol in Hanuman temple on 2m high platform 0.16 Umerga Surya idol kept on pedestal in Vishnu temple 0.17 Vishnu idol kept on ground 0.14 Sarwadi 7 Latur Idol in Vishnu temple on 1.5 m high platform 0.13 N i t a n g a Sirur-G 30 Gulbarga Idol in Hanuman temple on 1m high platform 0.16 Aland Idol in Hanuman temple on 1m high platform 0.24 T h a i r 5 0 Osmanabad Idol in Laxmi temple on 1.5 m high platform 0.1 Osmanabad Umerga 25 Osmanabad Steel cupboard in 3rd floor of MSEB quarters 0.25 Ume rg a Kohinoor 50 Basavakalyan Stone block in graveyard on 1m high platform 0.12 Pahad B i da r TABLE-2 Assigned Intensity and Estimated Earthquake Potential of Different Seismic Zone (IS:1893) of India. Seismic Assigned Maximum Basic Seismic zone Intensity Earthquake horizontal zone (MM) Potential seismic Fo (Magnitude) coeff. I V 5.7 0.01 0.05 II VI 6.3 0.02 0.10 III VII 7.0 0.04 0.20 IV VIII 7.7 0.05 0.25 V IX 9.0 0.08 0.40 68 68' 72' 76' 80* 84* BB' 92' 96° E 36* 1 /-" ~v i 1 1 1 - 36' N N [::::::g MssoioicCainoroic volcanism 0 100' 300 500 Km Rill Zonos ( s 1. 1 ..i 1 . i . 1 32' - > ^ - 32' A-1 Dharwar tills A-2 Rilled jone ol alkali J // 8 •*" r plulonilos and carboniles 28* 28' ' DELHI''* 0, ( A-3 Aravalli Delhi rill /•-.~/ A3—," y' '5 N. 7 (Ptecambrian) Plulonic grabens 1 24' J^ 24' i. 1 Godavan 4) T^)CAl **-*-^ I « «k u 7T——• • •*•/ 20* 20* \ III Damodar \ IV Narmada /"/ • 76*1 30. ----N V Cambay 16' \x;:: ilVwo* AABAD TALNI ^ KILLARI Precambrian shear rones 1 MC-ZZ? SASTUR^/*^^-'" x t \ \ {MADRAS:.. "^- - RAJECSAON 1 T1 Archean cralonmpbile bell BANGALORE* JL Inlerlaca; HG; High Giade Tetrain 12' W T1/7 Q i . HLTLLI / T2 Easletn Margin ol Ciddapah Basin \ TOnAMBA / \HG y MUNSAL T3 Sukinda Thutsl 8' Km / 1 1 \iT I Singhbhum shear zones 72* 76* 80* FIR. 1 Generalized tectonic map oj the Stable continental region ojIndia. The big Mar is the epicenter <>J the Latur earthquake oj September 30, 1993. Open star enclosed in solid circles are previous ID ei^ht largest earthquakes /after Johnston {1993)}: I: 1819, M = 7.79; 2: 1803, M = 6.65; 3: 1956, M = 6.-S3; 4: 1927 M = 6.411; 5: 1967, M = 6.3; 6: I93S, M = 6.26; 7: 1918, M = 6.24; 8: 1956, M = 6.05. { Inset is the mcfiiiseisinal area of Lainr earthquake. (Ref-10) TALIII LAMJAH r^i¥— KILLARI HATTAHGA TERNA RIVE HAKHIOAM - JAWAL6A PETH SflUGYI TAU5ISARII HULLI 5ALESA0H RAJE8AOM _ UHAROA YEXUNOI fig 2 Loa t ion of jcmr of the worst »((ect<:d vi )]aje5 devastilcd bf the iitur - Osmimbtd < j r t hqtu k c at September JO. 1993. (Hodi(ied litre Surv-y o^ Jnrfn Toposhect 56 B). Inlet jhovj epicenter of the earthquake on map of Indi2. l'Hj-3 nraiiiii(;e Pjllcm Around Killari Village and Inferred Subsiirlacc Harna^c Jntcustty f/»;»|» in UN Scale 70 BENCH UAEIK MANY fSj (10-20CWI t 50m LEFT OFFSET 11 . 13cm SCAflP -ISon HIGH Figure 5 • Portion of the surface rupture associated with the 1993 Killari earthquake. This rupture was mapped discontmously in an east-west zone over a distance of about 7 km. Barbs indicate the uplifted side of the scarps: the approximate height of the scarps is also indicated. Plus signs indicate uplifted area, circled when localized. Note that a 'transform" fault connects a south- facing scarp in the center of the map with a north-facing, scarp on the right. This 'transform' offsets a field boundary by 53 cm left laterally, which is a minimum measure of the Shortening. The strike of this and other "transforms' in this map is north-northeast, the inferred direction of shortening. (Ref-1) Fig-6 Views Showing Fault Scrap (Courtesy Dr. 5.K. Jain, JIT, Kanpur, India) 71 30-50 Cm TMICX .TUFF ON ROOF (Grey Soil) POINTING WITH SOIL (Grey) Lime/Cement OUTER WALL WALL O-5-2m Thick UNDRESSED STONES(Bosalfs) 7TTV77VV7vv77-rr7 Fig-7 Typical Detail of Conventional Buildings with Stone Wall Fig- g Damage of Conventional Stone Wall Building (Courtesy : Dr. S-K. Jain, IIT, Kanpur, India) 72 Fig-9 Damage of Building with Wooden Frame (Courtesy : Dr. 5-K. Jiiin, NT, KU/I[«JI, India) Fig-10 Views Showing Failure of RCC Circular Water Tank (Courtesy : Dr. S-K. Jain, IIT, Kan pur, India) 73 ,1 i^- II Ilartliquake Zoning Map of India (ReJ. IS 1893) - 07 —> I - - - —- - - ui a - o ... - - - \ - T z - - - o OS - \ ------— - - - i - - G V. DAMPING - - - - - Ill ; — — - z 2 /--? v. i — < \ 1 i / V •Sy. - — - - - U) 7 •— - - - - — f 7 ^< - F x 1 i ' i 1 1 - /" - - - - - 11 1 - - / ,LJ_L . > i J "•—, i ; c| 0 1 -1- 1; r - - — •- - i [ rlr i rri'fr i (H 08 12 IS 10 1 L 28 HAtURAL PERIOD OF viURAtiON (t( 5EC0K0S T 74 l ij;-12 Average Rcsfwnsc Spectra (IS 1893) SESSION II "COUNTRY EXPERIENCE IN SEISMIC RE-EVALUATION PROGRAMME" NEXT PAGEIS) left SLANK 75 XA9952648 1 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES U.S. EXPERIENCE IN SEISMIC RE-EVALUATION AND VERIFICATION PROGRAMS John D. Stevenson J. D. Stevenson, Consulting Engineer 9217 Midwest Ave. Cleveland OH 44125 ABSTRACT: The purpose of this paper is to present a summary of the development of a seismic re-evaluation program for older nuclear power plants in the U.S. The principal focus of this re- evaluation is the use of actual strong motion earthquake response data for structures and mechanical and electrical systems and components. These data are supplemented by generic shake table test results. Use of this type of seismic re-evaluation has led to major cost reductions as compared to more conventional analytical and component specific testing procedures. 1. INTRODUCTION The U.S. nuclear industry has been extremely successful in the development and the limited use various kinds of experience data to include analysis, test and earthquake for the seismic re- evaluation, verification and as necessary upgrading of Nuclear Power Stations. Nuclear power plants prior to 1975, particularly East of the 105° longitude in the U.S. (72 units in all) were designed to seismic criteria which was less well developed and in some cases less stringent than would be the case for nuclear power plants built at the same sites using seismic criteria developed since 1977. As a result, the U.S. Nuclear Regulatory Commission defined an Unresolved Safety Issue A-46 in 1980 which required these older plants (pre 1975) to re-evaluate their seismic design adequacy. This phenomena of increased seismic loading on nuclear power plants is a world wide trend, hence the experience gained in the U.S. could be applicable to many other countries. The program defined in this paper is limited to the USNRC Unresolved Safety Issue A-46 which is concerned with reevaluation of existing nuclear power plants for their original design basis safe shutdown earthquake. However, there is no reason why it could not be applied to other types of nuclear facilities or other type of hazardous facilities in the U.S. or elsewhere. The USNRC has also prescribed a seismic margin or probablistic risk assessment review. This review is applicable to all nuclear power plants in the U.S. which for so called focused and full scoped plant as part of an Individual Plant External Event Evaluation use significantly higher seismic input (i.e. 1.5 to 3.0 times that prescribed in the original design) and could also be applied in other countries where it is desired to quantify the margins or risk associated with the existing earthquake design. Current rigorous seismic design and construction techniques account for about 10 to 14 percent0-2' of the total cost of a nuclear power plant in the U.S. These methods rely primarily on structural analysis for building structures, distributions systems (piping, cable trays, conduit ductwork, etc.) and mechanical components, and on shaker table testing for seismic qualification of electrical components. To apply these conventional seismic qualification techniques to older operating plants and consequential modifications would be extremely expensive between $15,000,000 to $20,000,000 dollars per unit. In Table 1 is a summary of the large number of 77 of mechanical and electrical, distribution systems and components which typically would have been involved for the subset of seismic or safety Category I, Systems and Components, which make up the items required for safe shutdown of a U.S. Pressurized Water Reactor plant of 1000 Mwe size. It should be noted that this is a subset of all the safety related seismic category mechanical and electrical systems and components in the plant. To this list could be added approximately 4 to 6 building structures which also would require some level of seismic re-evaluation. All seismic Category I structures, systems and components to include all those required for postulated accident and external hazard prevention or mitigation systems and components would increase the quantities shown in Table 1 by a factor of 1.5 to 2.0. It was established as part of an earlier SEP program(3) and in NUREG/CR-4334(4) and as part of a NUREG/CR -6239(5) survey of the response of piping in power stations in California that have seen at least 0.2g peak ground acceleration that piping systems designed to U.S. power plant piping standards could withstand without failure zero period ground acceleration ZPGA up to 0.5g. For this reason piping (but not in-line power operated values) was excluded from the A-46 program. For somewhat more obscure reasons, HVAC ducts and their supports (but not fire dampers contained therein) were also excluded. In addition, earlier pilot studies and evaluations (0J> indicated that building structures plus the reactor coolant systems and component designed by the Nuclear Steam System Suppliers in the U.S. would be able to meet the new seismic criteria up to about 0.5g ZPGA without much difficulty hence also have not been included in the A-4-6 seismic re-evaluation effort in the U.S. Recent experience in Eastern Europe suggests that this conclusion may not be valid for reactor facilities that were constructed without meeting the recommendations of the S-l Guides and the S-2 guides of the IAEA(S-9). Recognizing this major NRC concern with regard to older nuclear power plants, starting in 1982, several U.S. utilities banded together to form the Seismic Qualification Utility Group, SQUG, to fund an effort to develop simpler, cost effect and safety neutral ways to seismically verify existing NPP structures, systems and components. They initiated a pilot effort to develop a data base which described and evaluated the seismic behavior in strong motion earthquakes of 8 classes of typical equipment found in nuclear power plants to include high and low and AC and DC switchgear, motor control centers, horizontal and vertical pumps and power operated valves. This effort was completed in 1984, and it clearly demonstrated that such equipment behaved extremely well in strong motion earthquakes for Zero Period Ground Acceleration, ZPGA, up to at least 0.3g, with little or no damage. This was contingent on certain caveats being met, components being positively anchored and spatial interaction with adjacent components and structure evaluated and significant damage precluded. In 1983, a Senior Seismic Review Advisory Panel SSRAP consisting of 5 acknowledged experts in seismic design and evaluations of NPP was formed to review the data being developed by SQUG and their consultants.00' In 1985, the pilot study of 8 classes equipment was enlarged to 20 as shown in Table 2, and the Electric Power Research Institute, EPRI, also began to take an active part in the effort. By 3987, the enlarged study was completed*6' and the NRC issued a formal requirement in the form of a Generic Letter 87-02°I} to resolve the USI-A^46 issue. The SQUG group began the preparation of a Generic Implementation Procedure, GIP, to provide detailed guidance to individual utilities and their consultants based on the class of 20 reports and interactions with the NRC and SSRAP to perform walkdowns, document and evaluate the SSC in the individual plants. The GIP was finalized in 1992, and individual plant walkdowns have been conducted over the past 3 years. It is expected that the walkdowns and reports documenting the results will be completed by most U.S. utilities by the end of 1996. The total cost in terms of manhours per nuclear power plant unit using the GIP methodology and including the evaluation potential malfunction of relays has generally been between 4,000 and 8,000 man hours. Physical upgrades as a result of the GIP process which relies heavily on the use 78 of experience data from both actual earthquakes and generic testing has typically been required in less than 1 to 2 percent of the components surveyed. Total cost per unit to perform the entire A-46 effort including hardware fixes typically has been less than $1,500,000. This compares with engineering manpower estimates alone of 250,000 to 300,000 man hours and an estimated total cost of 15,000,000 to 20,000,000 per unit using conventional seismic analytical and component specific test methods and procedure. 2. SUMMARY OF THE USI-146 GIP PROCEDURE The GIP requires re-evaluation considering four areas: (a) Seismic Capacity versus Demand (b) Caveats (c) Anchorage (d) Interaction as defined in Seismic Evaluation Work Sheet, SEWS. 2.1 Capacity Versus Demand The demand is based on the existing design basis or safe shutdown earthquake in the form of either ground or in-structure response spectra. The seismic capacity of components are defined in the GIP as the Bounding Spectrum as shown in Figure 1 for which the earthquake experience data base is applicable, or the Generic Equipment Ruggedness Spectra from which seismic test data on overall classes of equipment have been gathered. 2.2 Caveats Caveats are developed based on the observations or lack of information contained in the earthquake experiences and the test data base that existed at the time the GIP was prepared (1988- 1992). These are discussed in general in Part II Section 4.3 of the GIP. Individual caveats applicable to specific items of equipment are contained in Appendix B of the GIP. 2.3 Anchorage Anchorage requirements and capacities are defined in the Part II Section 4.4 and in detail in Appendix C of the GIP. 2.4 Interaction Potential interactions of the component with adjacent structures and other. systems and components are discussed in Appendix D of the GIP. Typical SEWS forms for individual classes of equipment from the GIP can be found in Figures 2 and 3. 3. FUTURE OF SEISMIC QUALIFICATION OF STRUCTURES, SYSTEMS AND COMPONENTS IN THE U.S. 3.1 Current Design Codes and Standards The seismic evaluation or base lining of SSC in existing nuclear power plants designed prior to 1975 using SQUG-GIP methodology will be essentially completed by the end of 1996. 79 However, there is the continuing requirement to seismically evaluate and qualify new, replacement and modified SSC in both old (prior to 1975) and newer nuclear power plants. In the U.S. nuclear power plants are typically designed for an operating life of 40 years with life extensions possible up to a 60 year total. During this time period 50 to 70 percent of the original safety related components and distribution systems are expected to be added to, replaced or modified. Such components and systems, particularly in harsh environments which include caustic, acid or wet steam fluids or other corrosive and erosive environments need to be replaced every 10-15 years or so. Seismic qualification procedures for Safety related SSC relied upon, between 1975 and 1987 were typically based on component specific analysis of structures and mechanical components and generic fragility or proof testing of electrical components by their manufacturers. This was a time consuming and expensive procedure typically adding some 12 to 14 percent to the total cost of the nuclear power plant. In 1987, the IEEE-344 Standard was reissued to permit the use of actual earthquake experience data contained as in Section 9 of the standard for reference SSC to be used to qualify candidate SSC based on previous analysis, tests or actual earthquake experience data. This procedure could be used if similarity could be demonstrated between the reference component and the candidate component. For mechanical components a new standard, Qualification of Active Mechanical Equipment Used in Nuclear Power Plants, ASME QME-1-1994 also permitted the use of actual earthquake experience data for operability evaluation as discussed in Attachments A and C to Appendix A of the ASME Standard. Unfortunately to date, neither of these two Standards has received unqualified acceptance by the USNRC. In using experience data for seismic verifications and evaluations it is necessary to establish a similarity relationship between the parent or reference components or system and the candidate component or system. This can be done on generic basis as was the case for SQUG where actual earthquakes have supplied an extremely large data base but with relatively little component specific information or on a component specific basis as is the procedure used by IEEE and ASME Standards. The ASME seismic qualification procedure for operability in ASME QME Section QR, Appendix A, Attachment A is summarized as follows: 1. Functional Characteristics Candidate equipment considered for qualification by similarity shall have similar function/malfunction characteristic to that for the parent equipment for which a database is available. 2. Excitation Characteristic The excitation for the candidate equipment shall be shown to be similar to that for the parent equipment. The parent data may include a composite spectrum that was generated from qualification of several parent equipment items. Specific excitation characteristics to be considered include (but are not necessarily limited to): a. frequency distribution - indicated by amplified region of response spectrum or by power spectral density b. peak amplitude of time history, i.e., excitation ZPA c. maximum amplification factor - ratio of maximum response spectrum value 80 to ZPA. d. time duration - strong-motion portion must be at least 10 sec. e. axes of orientation - must be common for candidate and parent equipment, i.e., careful examination of supports at excitation location is required f. excitation location. 3. Physical Characteristics Physical similarity is determined by those equipment properties which influence its dynamic response. Physical similarity between candidate and parent or reference equipment can be shown by one of several methods, some of which are as follows: a. Essentially identical equipment - Equipment compared by make, model, and serial number, and found to be identical (within deviations associated with manufacturing tolerances) or whose differences are so slight that the dynamic response can be argued to be essentially unaffected. b. Similar modal properties - Equipment whose mass, stiffness and damping properties can be shown to be similar. c. Acceptance criteria - For acceptance criteria, provide comparison of items for both the parent and candidate equipment. If there are significant differences (more than +. 15%) in any one of the items (or sub-items), the effect of the difference shall be evaluated in terms of the following: a) stiffness b) mass distribution c) boundary conditions d) natural frequencies e) damping The different effect on the stresses, deformations, and load capacities (such as stem or shaft buckling capacity, bearing load capacity, etc.) at critical locations shall also be evaluated and shown to be within the allowable criteria limits. d. Similar Critical Transfer Function - The critical transfer function establishes a direct dynamic relationship between the excitation and the critical location where failure or malfunction is being evaluated. It can be established from typical exploratory resonance search data, if available, for a response point near a critical location. When the critical transfer function plot can be established for both candidate and parent equipment, and where this can be shown to be within 20% in amplitude within a designated frequency bandwidth, no further modal characteristics need to be determined. As a result, the equipment is physically similar within the designated frequency bandwidth. 4.0 Concurrence of Excitation and Physical Similarity A valid qualification by similarity requires that the frequency bandwidth within which physical similarity exists for both candidate and parent items shall be concurrent within the frequency band for which the candidate's required excitation 81 spectrum is enveloped by the parent equipment excitation spectrum. Enveloping outside this frequency band is not essential, but cannot be unlimited. The ASME for mechanical components and systems with regard to operability followed the IEEE lead in 1994. To date the use of experience data for the most part has been limited to questions of operability of mechanical and electrical equipment with structural and leak tight integrity as applicable still being addressed by construction codes and standards. It is the author's opinion that for mechanical components and distribution systems in the future, experience data will be used to develop, simplified "design by rule" procedures which will be adopted in the design portion of construction codes (ASME Boiler and Pressure vessel Code Sections ED., VEU and ASME B31.1) for seismic analyses. However, it is highly unlikely that experience data without the express computation of stresses in the component and their supports at least on a generic spacing tables and charts basis will be permitted by these standards. For the design of elevated temperature piping systems with design temperatures greater than about 150° F (80°C), it is highly unlikely that the spacing table and chan methods can be used because of the complication of the computation of restraint of free end expansion stresses. It is also highly unlikely that the structural design codes for concrete or steel structures which are analysis based will be changed significantly as the result of use of current experience data. A program for seismic qualification (functional), STERI since 1992 has been developed for qualification of new, replacement and modification to safety related systems and components in nuclear power plants. This program uses both the SQUG-GIP earthquake experience and past generic testing of components (Generic Equipment Ruggedness Spectra) and a new qualification program called SQRTS. The SQUG-GIP approach has the drawback that to be useful for experience based seismic qualification, the component must be in the data base (i.e. they must have experienced a strong motion earthquake, have been analyzed, or been included in the generic testing. There is much new or replacement equipment particularly those with solid state and digital devices where such data are not available. The nuclear industry in the U.S. and particularly that associated with new or replacement equipment is no longer large enough to support generic testing by individual manufacturers. As a result, individual utilities in the U.S. have banded together to support a particular testing facility to perform seismic qualification testing on individual components as requested by individual utilities with results to be shared by all. This program is called SQRTS and is associated with the seismic qualification of otherwise commercial grade equipment in the U.S.. 4. THE REGULATORY PERSPECTIVE IN THE U.S. For plant designed between 1975 and 1987 in the U.S., the Nuclear Regulatory Commission as part of the individual licensing process was able to secure commitments from the utilities involved to perform individual component specific seismic qualification by test or analysis. In general, no mention was made of the use of experience data. As a result, the NRC has tended to view the use of generic or actual earthquake experience data to seismically qualify SSC as a dilution of this legal commitment. However, they have permitted it to be used for verification of seismic adequacy and in seismic margin evaluations for all nuclear power stations. On the older pre-1975 plants, the USNRC has permitted its use to qualify new equipment since in most cases there were no specific regulatory commitments as to how the seismic qualification would be performed on these earlier plants. It should also be noted that the NRC has not accepted the IEEE-347-87 or ASME QME.QR Appendix A portions dealing with the use of experience data since it would amount to generic acceptance. It is hoped that further research and codification of the use of earthquake experience will permit the USNRC to accept the use of earthquake experience as a valid procedure for seismic evaluations and qualifications in the future. Currently for new or advanced reactor designs in the U.S. there is a major push by the nuclear power industry in that direction. 82 5. FUTURE AND CONCLUSION There are strong economic pressures to increase the use of experience data as the basis of safety neutral seismic qualification of safety related systems and components in U.S. nuclear power plants. Since 1976, the high cost of seismic qualification or design (+12 to +14 percent of total plant cost) which does not take advantage of the use of earthquake experience data and the high cost of document based Quality Assurance (+30 to 35 percent) procedures have resulted in nuclear power being considerably more expensive than other forms of energy for electric power generation in the U.S. There also has been a general surplus of power generation capability and the ability to import electricity from Canada. As a result, this plus the uncertainties of nuclear plant licensing, there have been no new nuclear plant orders since 1976 in the U.S. Significantly, nuclear power stations using essentially U.S. reactor designs are currently being constructed outside the U.S. by using more rational seismic and quality assurance procedures which are still economically viable when compared to the cost of fossil fueled power stations. The U.S. Nuclear Regulatory Commission has moved to simplify the licensing process of new advanced reactor designs but the underlying economic issues of seismic design and document based quality assurance have yet to be addressed in a meaningful way. In the author's opinion, it should be possible by rational use of experience data and "design by rule" produces based on the actual observed damage to mechanical and electrical systems and components in strong motion earthquakes to reduce seismic design and qualification costs to less than 5 percent of plant costs. It should also be possible to reduce the cost of performance based quality assurance to 10 to 15 percent of base plant cost which could once again make nuclear power a viable economical option in the U.S. It must be understood that nuclear power is the only form of renewable (breeder or fusion) source of high density energy in the world. Sooner or later, nuclear power will become the dominate form of power generation in the U.S. as it has in many other countries. If the sooner option is taken, we will be able to preserve our fossil materials as chemical feed stocks for the material needs of future generations rather than squander it as fuel in the near term. We would also be able to address concerns associated with acid rain and the green house global warming effects in a much more ecologically positive manner. 6. REFERENCES (1) Stevenson and Associates, "Evaluations of the Cost Effects on Nuclear Power Plant Construction Resulting from the Increase in Seismic Design Level," NUREG/CR 1508 Prepared for U.S. Nuclear Regulatory Commission, April 1981. (2) Stevenson and Associates, "Differential Design and Construction Cost of Nuclear Power Plant Safety Related Systems," In preparation as a Welding Research Council Bulletin. 1995 (3) Senior Seismic Review Team, "Seismic Review of Dresden Nuclear Power Station-Unit 2 for the Systematic evaluation Program," NUREG/CR 0891, U.S. Nuclear Regulatory Commission, April 1980. (4) Budnitz, R.J., "An Approach to the Quantification of Seismic Margins in Nuclear Power Plants", NUREG/CR-4334, August 1985. (5) Stevenson, J.D., "Survey of Strong Motion Earthquake Effects on Thermal Power Plants in California with Emphasis on Piping Systems", NUREG/CR-6239 Vols. 1 and 2 November 1995. 83 8 (6) EQE Incorporated, "Summary of the Seismic Adequacy of Twenty Classes of Equipment Required for the Safe Shutdown of Nuclear Plants", Prepared for seismic Qualification Utility Group, February 1987. (7) EQE Inc., "Summary of the Seismic Adequacy of Twenty Classes of Equipment Required for Safe Shutdown of Nuclear Plants", EPRI NP7149 March 1991. (8) Safety Guide 50-SG-S1, "Earthquakes and Associated Topics in Relation to nuclear Power Plant Siting", International Atomic Energy Agency, 1979. (9) Safety Guide 50-SG-S2, "Seismic Analysis and Testing of Nuclear Power Plants," International Atomic Energy Agency, 1979. (10) Senior Seismic Review and Advisory Panel, SSRAP, "Use of Experience and test Data to Show Ruggedness of Equipment in Nuclear Power Plants", DE-01921, Sandia National Laboratories, June 1992. (11) USNRC, "Verification of Seismic Adequacy of Mechanical and Electrical Equipment in Operating Reactors, Unresolved Safety Issue A-46", Generic Letter 87-02. 84 99904b Table 1 TABLE 1' List of Quantities of Safety Related Structures Systems and Components Required for Nuclear Power Plant Safe Shutdown MECHANICAL COMPONENTS Components Quantity (Metric) A. Seismic Category I < 1 2.5 cm > 1 2.5 cm > 8 cm Total Piping2,3 1. Containment-Reactor Bldg. a. Instrumentation 2000 2000 tubing b. ASME Class 1 900 1100 2000 c. ASME Class 2 4300 5500 9800 and 3 2. Auxiliary building a. Instrumentation 2500 2500 Tubing b. ASME Class 2 10600 9300 19900 and 3 piping 3. Fuel building a. Instrumentation 160 160 tubing 330 660 990 b. ASME Class 2 and 3 4. Diesel generator a. Instrumentation 330 330 tubing b. ASME Class 2 500 130 630 and 3 5. Intake structure, pipe 160 340 500 B. Valves<2) (Units) 150 1. Motor Operated 2. Hand Operated None in 800 SSEL C. Vertical & Horizontal tanks, heat exchangers and vessels (Units) 1. Column supported 105 2. Skirt supported 45 3. Saddle Supported 10 85 10 Components Quantity (Metric) D. Pumps 60 1. Vertical 2. Horizontal 100 E. Seismic Category I < 12 in. > 12 in. Total HVAC ductwork dia. 30 cm Building 30 cm 1. Containment 1800 800 2,600 2. Auxiliary building 5600 1600 7,200 3. Fuel building 300 200 500 4. Diesel generator 200 100 300 TOTAL 7900 2700 10,600 II ELECTRICAL COMPONENTS Quantity (Metric) A. Seismic < 1 in. 1 in Total Category I 2.5 cm 2.5 cm 4 cm 5 cm 8 cm 10 cm Electrical Conduit 1. Galvanized 1200 9000 6000 12600 5000 8000 41000 rigid 2. Flexible 5300 6600 7000 2500 20400 B. Seismic 8000 8000 Category I cable trays C. Seismic Category I Electrical components 1. Control 90 panels 2. Switchgear 80 3. Transform- 20 ers 4. Motor 60 Control Centers 86 11 Quantity (Metric) 5. Pressure & 200 Temperature Sensors and Transmitters 1. Qualities listed for the Safe Shutdown List are only one third to one half of those typically defined as safety related for a U.S. Nuclear Power Plants. 2. For all piping, assume 40% of the quantities shown are hot (design temperature, >80°C and, therefore, require thermal flexibility analysis and may require the use of constant or variable spring hangers for deadweight and dampers or snubbers to carry seismic loads. 3. Piping and hand operated valve quantities given are for information only. In the U.S. piping and hand operated valves are not included in the Safe Shutdown List because previous evaluations have indicated piping in Nuclear Power Plants including hand operated valves have HCLPF values which exceed 0.5g peak ground acceleration. 87 1 2 99904b Table2.1 TABLE 2 Classes of Safe Shutdown Equipmei 1. Fans 2. Air compressors *> Battery racks 4. Battery chargers and inverters 5. Air handlers 6. Chillers 7. Transformers 8. Vertical pumps 9. Horizontal pumps 10. Motor-generators 11. Motor control centers 12. Low voltage switchgear 13. Medium voltage switchgear 14. Distribution panels 15. Fluid-operated valves 16. Motor-operated valves 17. Engine-generators 18. Instrument racks 19. Sensors 20. Control and instrumentation cabinets 88 13 1.0 5% DAMPING 3 0.8 z o £ 0.6 BOUNDING SPECTRUM UoJ o 0.4 < a. o GROUND ACCELERATION = 0.33 g UJ 0.2 - 0.0 8 12 16 20 24 28 FREQUENCY (Hz) Figure 1.1 Seismic Motion Bounding Spectrum Horizontal Ground Motion 89 14 Status Y N U SCREENING EVALUATION WORK SHEET (SEWS) Sheet 1 of 2 Equip. ID No. Equip. Class 5 - Horizontal Pumps Equipment Description Location: Bldg. Floor El. Room, Row/Col Manufacturer, Model, Etc. Horsepower/Motor Rating RPM Head Flow Rate SEISMIC CAPACITY VS DEMAND 1. Elevation where equipment receives seismic input 2. Elevation of seismic input below about 40' from grade Y N U 3. Equipment has fundamental frequency above about 8 Hz Y N U N/A 4. Capacity based on: Existing Documentation DOC Bounding Spectrum BS 5. Demand based on: Ground Spectra 6RS Amplified (Floor) Spectra AFS Does capacity exceed demand? Y N U CAVEATS - BOUNDING SPECTRUM 1. Equipment is included in earthquake experience data base Y N U N/A 2. Driver and pump connected by rigid base or skid Y N U N/A 3. No indication that shaft does not have thrust restraint in both axial directions Y N U N/A 4. No risk of excessive nozzle loads such as gross pipe motion or differential displacement Y N U N/A 5. Base vibration isolators adequate for seismic loads Y N U N/A 6. Attached lines (cooling, air, electrical) have adequate flexibility Y N U N/A 7. Anchorage adequate Y N U N/A 8. Relays mounted on equipment evaluated Y N U N/A 9. No other concerns Y N U N/A Are the caveats met for Bounding Spectrum? Y N U N/A ANCHORAGE 1. Appropriate equipment characteristics determined (mass, CG, natural freq., damping, center of rotation) Y N U N/A 2. Type of anchorage covered by GIP Y N U N/A 3. Sizes and locations of anchors determined Y N U N/A 4. Adequacy of anchorage installation evaluated (weld quality, nuts and washers, expansion anchor tightness) Y N U N/A Figure 2 Typical Seismic Evaluation Work Sheet for Mechanical Equipment 90 15 SCREENING EVALUATION WORK SHEET (SEWS) Sheet 2 of 2 Equip. ID No. Equip. Class 5 - Horizontal Pumps Equipment Description ANCHORAGE (Cont'd) 5. Factors affecting anchorage capacity or margin of safety considered: embedment length, anchor spacing, free-edge distance, concrete strength/condition, and concrete cracking Y N U N/A 6. For bolted anchorages, gap under base less than 1/4-inch Y N U N/A 7. Factors affecting essential relays considered: gap under base, capacity reduction for expansion anchors Y N U N/A 8. Base has adequate stiffness and effect of prying action on anchors considered Y N U N/A 9. Strength of equipment base and load path to CG adequate Y N U N/A 10. Embedded steel, grout pad or large concrete pad adequacy evaluated Y N U N/A Are anchorage requirements met? Y N U INTERACTION EFFECTS 1. Soft targets free from impact by nearby equipment or structures Y N U N/A 2. If equipment contains sensitive relays, equipment free from all impact by nearby equipment or structures Y N U N/A 3. Attached lines have adequate flexibility Y N U N/A 4. No collapse of overhead equipment or distribution systems Y N U N/A 5. No other concerns Y N U N/A Is equipment free of interaction effects? Y N U IS EQUIPMENT SEISMICALLY ADEQUATE? Y N U COMMENTS Evaluated by: Date: Figure 2 (Cont.) 91 16 SCREENING EVALUATION WORK SHEET (SEWS) Sheet 1 of 2 Equip. ID No. Equip. Class 20 - Instr. & Control Panels & Cabinet Equipment Description Location: Bldg. Floor El. Room, Row/Col Manufacturer, Model, Etc. SEISMIC CAPACITY VS DEMAND 1. Elevation where equipment receives seismic input 2. Elevation of seismic input below about 40' from grade Y N U 3. Equipment has fundamental frequency above about 8 Hz Y N U N/A 4. Capacity based on: Existing Documentation DOC Bounding Spectrum BS 5. Demand based on: Ground Spectra GRS Amplified (Floor) Spectra AFS Does capacity exceed demand? Y N U CAVEATS - BOUNDING SPECTRUM 1. Equipment is included in earthquake experience data base Y N u N/A 2. No computers or programmable controllers Y N u N/A 3. No strip chart recorders Y N u N/A 4. Steel frame and sheet metal structurally adequate Y N N/A 5. Adjacent cabinets or panels which are close enough u to impact, or sections of multi-bay cabinets or panels, are bolted together if they contain essential relays Y N u N/A 6. Drawers and equipment on slides restrained from fall ing out Y N u N/A 7. All doors secured by latch or fastener Y N u N/A 8. Attached lines have adequate flexibility Y N u N/A 9. Anchorage adequate Y N u N/A 10. Relays mounted on equipment evaluated Y N u N/A 11. No other concerns Y N N/A Are the caveats met for Bounding Spectrum? u Y N U ANCHORAGE 1. Appropriate equipment characteristics determined (mass, CG, natural freq., damping, center of rotation) Y N U N/A 2. Type of anchorage covered by GIP Y N U N/A 3. Sizes and locations of anchors determined Y N U N/A 4. Adequacy of anchorage installation evaluated (weld quality, nuts and washers, expansion anchor tightness) Y N U N/A Figure 3 Typical Seismic Evaluation Work Sheet for Electrical Equipment 92 17 SCREENING EVALUATION WORK SHEET (SEWS) Sheet 2 of 2 Equip. ID No. Equip. Class 20 - Instr. & Control Panels & Cabinets Equipment Description ANCHORAGE (Cont'd) 5. Factors affecting anchorage capacity or margin of safety considered: embedment length, anchor spacing, free-edge distance, concrete strength/condition, and concrete cracking Y N U N/A 6. For bolted anchorages, qap under base less than 1/4-inch Y N U N/A 7. Factors affecting essential relays consiaered: gap under base, capacity reduction for expansion anchors Y N U N/A 8. Base has adequate stiffness and effect of prying action on anchors considered Y N U N/A 9. Strength of equipment base and load path to CG adequate Y N U N/A 10. Embedded steel, grout pad or large concrete pad adequacy evaluated Y N U N/A Are anchorage requirements met? Y N U INTERACTION EFFECTS • 1. Soft targets free from impact by nearby equipment or structures Y N U N/A 2. If equipment contains sensitive relays, equipment free from all impact by nearby equipment or structures Y N U N/A 3. Attached lines have adequate flexibility Y N U N/A 4. No collapse of overhead equipment or distribution systems Y N U N/A 5. No other concerns Y N U N/A Is equipment free of interaction effects? Y N U IS EQUIPMENT SEISMICALLY ADEQUATE? Y N U COMMENTS Evaluated by: Date: Figure 3 (cont.) NEXT PAGE(S) l«ft BLANK 93 XA9952649 SMiRT 13 - Post Conference Seminar No. 16 on "Seismic Evaluation of Existing Facilities" A Regulatory View Of The Seismic Re-Evaluation Of Existing Nuclear Power Plants In The United Kingdom By J E Inkester and P M Bradford HM Nuclear Installations Inspectorate Health and Safety Executive St Peter's House Balliol Road Bootle Merseyside L20 3LZ United Kingdom Abstract The paper describes the background to the seismic re-evaluation of existing nuclear power plants in the United Kingdom. Nuclear installations in this country were not designed specifically to resist earthquakes until the nineteen-seventies, although older plants were robustly constructed. The seismic capability of these older installations is now being evaluated as part of the periodic safety reviews which C nuclear licensees are required to carry out. The regulatory requirements which set the framework for these studies are explained and the approaches being adopted by the licensees for their assessment of the seismic capability of existing plants are outlined. The process of hazard appraisal is reported together with a general overview of UK seismicity. The paper then discusses the methodologies used to evaluate the response of plant to the hazard. Various other types of nuclear installation besides power plants are subject to licensing in the UK and the application of seismic evaluation to some of these is briefly described. Finally the paper provides some comments on future initiatives and possible areas of development. Introduction 1 The first electricity-generating nuclear power station to be constructed in the United Kingdom was Calder Hall, a Magnox-type gas-cooled reactor which began operating in 1956 and is still in operation almost 40 years later. In those early days of the UK nuclear programme the installations were not designed specifically to resist earthquakes, since the UK is an area of relatively low seismicity. As modern standard:: developed, however, it came to be recognised that seismic inputs should be taken into account by the design, and the first power reactors to be seismically designed in the UK were the Heysham Stage 2 and Torness Advanced Gas-cooled Reactors (AGRs), which were designed in the 1970s and received consent to begin construction in 1980. In the decade which followed, a policy was adopted of 95 reviewing the safety of all UK power reactors for their longer-term operation, and it was decided to include seismic evaluation, or re-evaluation, in those reviews. 2 This paper describes the history of the approaches and the present position for seismic review of existing nuclear power plants. This is reported within the context of the overall arrangements for the regulation of nuclear safety in the UK. The evaluation of the seismic capability of such installations is now part of a wider programme of periodic safety reviews (PSR) of nuclear power plant. The paper provides some background to these and explains the criteria used to assess the adequacy of the performance of the plant. 3 There is a wide selection of UK nuclear installations other than power reactors - for example, nuclear chemical plant - where a similar lack of knowledge exists regarding the seismic capability of older plant. Brief details are given of one of the assessment processes used to evaluate such plant. 4 HM Nuclear Installations Inspectorate has now had several years' experience of assessing seismic safety cases. At the same time, we are aware of similar initiatives in other countries and have, in fact, had a number of contacts on this subject, either bilaterally or through conferences, seminars and other meetings, such as those arranged by the IAEA. The present seminar provides a timely opportunity for the wider sharing of experience of seismic evaluations of existing nuclear installations. Regulation 5 In the UK the main legislation governing the safety of nuclear installations is the Health and Safety at Work etc. Act 1974 and the associated relevant statutory provisions of the Nuclear Installations Act 1965. Under the Nuclear Installations Act no site may be used for the purpose of installing or operating any commercial nuclear installation unless a nuclear site licence has been granted by the Health and Safety Executive (HSE) and is for the time being in force. HM Nuclear Installations Inspectorate (Nil) is that part of HSE responsible for administering this licensing function. 6 The Health and Safety at Work etc. Act requires the provision and maintenance of plant and systems of work that are, so far as is reasonably practicable, safe and without risks to health. Another way of expressing this is that the risks must be reduced to as low as is reasonably practicable, or ALARP. The legislation places the primary responsibility for safety on the operator (i.e. the licensee) of each installation. It is the duty of Nil to see that appropriate standards are developed, achieved and maintained by licensees, to ensure that any necessary safety precautions are taken, and to monitor and regulate the safety of plant by means of its powers under the licence and relevant regulations. It should be noted that this is a non-prescriptive licensing regime and, in the context of seismic evaluation, Nil does not, for example, prescribe the level of input ground motion nor the spectral shape to be used. 96 7 The Nuclear Installations Act gives the Nil, on behalf of HSE, the power to attach conditions to each site licence in the interests of safety. There are 35 standard licence conditions which are applied to most sites. One of these conditions requires the licensee to produce "safety cases", which consist of documentation to justify safety during the design, construction, manufacture, commissioning, operation and decommissioning phases of the installation. Another licence condition requires the licensee to carry out a periodic and systematic review and reassessment of safety cases. The earliest manifestation of these periodic safety reviews was the series of Long Term Safety Reviews (LTSR) which were carried out for the Magnox gas-cooled reactors after they had been operating for 20 years. We have, however, moved now to the position where periodic safety reviews (PSR) will be required every ten years, and these have already been carried out for some Magnox stations operating beyond 30 years and are currently under way for the AGRs. Seismic evaluation of the installation is one of the topics covered by the LTSRs and PSRs. Safety Assessment Principles 8 Nil assesses the licensees' safety cases for their adequacy, and HSE has published the safety assessment principles (SAPs, Ref. 1) which form the framework used by Nil's inspectors in carrying out this work. For natural hazards, it is recognised that the uncertainty of data may prevent reasonable prediction of design events for frequencies less than once in 10,000 years and, where this is the case, which is generally considered to apply to earthquakes in the UK, the SAPs call for the establishment of a design basis event which is conservatively determined at that frequency. The plant should then be designed to contain or limit the release of radioactivity following such a design basis event, such that there should be no release of radioactivity except in the most severe cases and, even then, no person outside the site will receive an effective dose of 100 mSv or more. 9 The SAPs are, however, "aimed primarily at the safety assessment of proposed (new) nuclear plants, but they will also be used in assessing existing plants." The document goes on to say: "For the assessment of plants which exist today ('old plants') there is a further point to be considered in that the safety standards used in their design and construction may differ from those used in plants currently being designed and built. The existence of such differences has to be recognised by our assessors when applying the SAPs in the assessment of old plants. The ALARP principle is of particular importance to such assessments, and the age of the plant and its projected life are important factors to be taken into account when making judgements on the reasonable practicability of making improvements to those plants." 10 This implies that standards which are lower than the 'modern' standards represented by the SAPs can be accepted for older plants. But there has to be a limit, and HSE has also published a document entitled "The Tolerability of Risk from Nuclear Power Stations" (Ref. 2), in which it is stated that, "We propose to maintain our existing position that a risk [of death] of 1 in 10" per annum to any member of the public is the maximum that should be tolerated from any large industrial plant in 97 any industry with, of course, the ALARP principle applying to ensure that the risk from most plant is in fact lower or much lower." These documents, therefore, give us some basic numerical guidelines, but the essence of our seismic assessments is that we are looking for the licensees to provide a demonstration that the risks from their plant in the event of an earthquake are tolerable, and have been reduced to as low as is reasonably practicable. A more detailed discussion of the application of the SAPs to seismic design can be found in Reference 3. Seismicity of the UK 11 The UK is situated in the intra-plate tectonic region of north-western Europe. The average seismicity is characterised as approximately 0.2g free-field horizontal peak ground acceleration (pga) for an event with a probability of exceedance of 1 in 10,000 per year. The approach used to derive the seismic hazard has developed significantly since the late 1970s, when the subject was in its infancy in the UK. Initially, mainly historical data and seismographs from the west coast of the USA were used to determine the hazard level and the frequency characteristics of the earthquakes used for assessment purposes. The Cornell and Newmark-Hall methodologies were later used to produce a site-specific piece-wise linear spectrum. This spectrum is often designated the PML spectrum, after the initials of the firm, Principia Mechanica Limited, which developed it. More recently, geological and tectonic considerations have been incorporated with strong motion data from intra-plate areas to produce uniform hazard spectra (UHS). The data which are used in this latter approach are believed to be more relevant to the UK. There are no strong motion UK records from which to produce a response spectrum. Consequently, there is debate over whether the low frequency section of the spectrum, in particular, is accurately modelled. 12 The lack of data directly relevant to the UK creates an area of uncertainty. For example, the record of historical earthquakes is only essentially complete above magnitude 4 for the last 200 years. Most seismic events in the UK occur at depths of between 5 and 15 km and it is difficult to associate these occurrences with a particular geological feature. Range of Plant and Licensees 13 The Nil regulates a wide variety of plant types and licensees. These include power reactors, chemical plants, storage facilities, research reactors, process plants and dockyards where the refitting of nuclear submarines (which are themselves exempt from licensing) is undertaken. The commercial nuclear power plants consist of gas cooled reactors, either of the Magnox or AGR type, together with one PWR, Sizewell B, which has achieved its first full-power operation in 1995. These power reactors are currently owned and operated by British Nuclear Fuels pic (BNFL), Nuclear Electric pic (NE) and Scottish Nuclear Limited (SNL). (The British Government has recently announced, however, the restructuring of the UK nuclear power industry, with the Magnox stations remaining in public ownership, while the AGRs and the Sizewell B PWR will be privatised next year.) 98 14 As a result of the non-prescriptive nature of the British regulatory system, the approach to achieving an acceptable level of safety at existing nuclear installations varies between the licensees and the different types of plant involved. The last two AGR stations, Torness and Heysham 2, were seismically designed (although not necessarily to the same standards that would be applied if they were being designed today). Sizewell B was also seismically designed and closely approximates the modern standard. None of these stations has yet reached the date for its first periodic safety review, so no seismic re-evaluation has yet taken place. For the older plant, various techniques have been used to attempt to quantify their seismic capability. Details of these assessments are given in the following sections. Seismic Re-Evaluation Of Existing Plant Overview of the Programme of Reviews 15 BNFL , SNL and NE have all carried out reviews of older plants. The seismic capability of each of the Magnox reactors was assessed in its Long Term Safety Review (LTSR). The purpose of the LTSR programme was to demonstrate that the plants would be adequately safe for at least 30 years operation. (They did not have a 'design life' as such, but their designers are believed to have envisaged a working life of 20 to 25 years.) For the Chapelcross and Calder Hall reactors, BNFL used techniques which were developed during the Seismic Damage Assessment (SDA) of reprocessing plant at Sellafield (see below). From the experience gained in both the LTSRs and SDAs, ways are being developed by the licensees to enhance the methodology of seismic re-evaluation. NE is now carrying out studies to show that its Magnox reactors are fit for continued operation beyond 30 years and BNFL is doing the same for operation beyond 40 years for Calder Hall and Chapelcross. (SNL's Magnox station, Hunterston A, has been closed since its LTSR was carried out.) NE and SNL have also begun Periodic Safety Reviews (PSRs) of their AGR reactors. All these reviews include seismic re-evaluation. NE and SNL's Maqnox Long Term Safety Reviews 16 When the LTSR programme began it was agreed between Nil and the licensees that an assessment against a 0.1g earthquake should be performed. A consideration in choosing this level was undoubtedly that the IAEA guidance for the siting of new nuclear power plants (Ref. 4) recommends that, regardless of any lower apparent exposure to seismic hazard, all plants should adopt a minimum value of 0.1 g peak ground acceleration. Thus, although the LTSR programme was for existing power plants, a certain consistency would be achieved. 17 The LTSR assessments therefore used a ground motion defined by a 0.1g horizontal pga and the PML response spectrum. The intention was to establish that the major structures and the plant used to shut down the reactor, remove decay heat and maintain negative reactivity could survive this motion, and to use this information as a basis for deciding whether the stations were acceptably safe. The assessment should also identify any improvements which were reasonably 99 practicable, which might include improvements giving a capability well beyond 0.1g. Plant improvements have indeed resulted from these reviews, including such things as better restraint of electrical equipment and the installation of tertiary boiler feed systems for decay heat removal. BNFL's Seismic Damage Assessment 18 BNFL has a wide variety of plants ranging from reactors to reprocessing facilities, most of which did not cater specifically for resistance to earthquakes in their design. A seismic damage assessment (SDA) was carried out for the chemical plant at Sellafield, firstly to identify the potential for improvements to the robustness of the installations, and secondly to allow preparation of emergency plans for coping with the consequences of an earthquake. The SDA predicted the likely plant performance at 0.125g, 0.25g, and 0.35g pga (PML spectrum). The 'walkdown' methodology developed in the USA was also used. Many of the techniques in the EPRI methodology for the conservative deterministic failure margin (CDFM) (Ref. 5) were employed. The SDA aimed, however, to provide only a slightly conservative, best estimate of the plant performance and therefore did not actually comply with all the CDFM criteria. For their reactors, BNFL adopted a two-stage methodology. All safety-related plant was shown to be capable of surviving a 0.125g pga event (PML spectrum) and a subset of 'plant essential to safety' one of 0.2g pga. AGR Periodic Safety Reviews Safety strategy 19 NE and SNL have recently embarked on a periodic safety review (PSR) programme for their AGRs. They have proposed the following policy for the integrity of protection:- (a) For any frequent initiating event (more frequent than 10~3 per annum) there should normally be at least two lines of protection to perform any essential function, with diversity between each line; (b) For any infrequent initiating event (less than or equal to 10"3 per annum) there should be at least one line of protection to perform any essential function, and that line should be provided with redundancy. Input motion specification 20 NE and SNL have stated that, for the seismic safety case, the magnitude of the infrequent initiating event should correspond to a severity consistent with a return frequency of 10"4 per annum at the site. The appraisal will examine all essential structures and a single line of protection (including redundancy) to trip, shutdown and cool the reactor. The systems involved have been designated 'the bottom line plant'. 21 The plant which will provide a diverse means of achieving trip, shutdown and post trip cooling against frequent events is called the 'second line plant'. The ground 100 motion specification for the frequent initiating event is 0.1g pga and the PML response spectrum appropriate to the site conditions. 22 Plant whose failure could threaten the defined lines of protection is known as 'related plant'. It will be assessed to the same level as the plant which it could threaten. 23 For the first AGRs to be reviewed, Hinkley Point B and Hunterston B, the 'bottom line plant' assessment ground motion has been pragmatically agreed as the PML hard ground spectrum anchored at 0.14g pga. Since this spectrum envelopes completely the claimed expected 10"4 UHS for Hunterston, and matches that for Hinkley Point closely at the frequencies of most practical importance (1 to 8 Hz), NE and SNL consider this to be a surrogate for a level of ground motion with an expected probability of exceedance of 10"4 per year. Nil will give consideration to this claim when the relevant reports are received for assessment. In subsequent reviews of AGRs, NE intends to provide a uniform hazard spectrum (UHS) for each site at the expected confidence level, with a probability of exceedance of 1 in 10,000 per year. Discussions are ongoing between the Nil and NE as to whether the UHS that they propose are an adequate representation of a 10"4 per year event. Seismic re-evaluation methodology 24 Building response to the input ground motions will be determined using established modelling techniques and soil structure interaction. This will enable f~" secondary response spectra to be generated for use in plant analyses. Two approaches will be used for plant assessment : analysis and 'walkdown1. The 'walkdown1 will make use of the SQUG Generic Implementation Procedure (Ref. 6) and its associated caveats when using earthquake experience data. Analysis will be used whenever the walkdown approach is not applicable or fails to demonstrate that the item can withstand the earthquake. The capacity of the structure and plant items will be determined using design code allowable stresses, strains and deflections in the first instance. Should the determined capacity be inadequate for the proposed functional requirement more detailed calculations may be carried out allowing limited but tolerable damage or inelasticity. 25 NE and SNL believe that the above process is robust. The system caters for uncertainty by specifying an expected level of site specific input motion (approximately 60% confidence level), a median or slightly conservative evaluation of the structure's response and the determination of the plant's capacity using code allowables where possible. The licensees have offered to carry out limited sensitivity studies and to document margins above assessment levels in order to provide Nil with confidence in the methodology. Maqnox Periodic Safety Reviews /-> 26 To establish that the Magnox plants can continue operating safely beyond 30 years (and beyond 40 years in the case of Calder Hall and Chapelcross), Magnox PSRs are being carried out as a development of the LTSR programme. Although two of NE's Magnox stations, Bradwell and Hinkley Point A, have already been 101 cleared for operation beyond 30 years, the methodology continues to be developed. Nil has requested that the licensees' PSRs should show that the 'bottom line plant1 has a safety margin beyond the capacity which was demonstrated in the LTSR against an earthquake ground motion defined by the PML response spectrum anchored to 0.1 g pga. 27 At the present time the method of achieving this requirement is still under discussion with NE. However, NE has proposed to demonstrate that a single line of protection exists against the 10"4 per annum seismic event in an essentially similar but perhaps simplified manner to that for the AGR PSR. NE have suggested that the ground motion input should be a UHS. Regulatory View of PSR Approach Adopted by Licensees 28 The continuing development by the licensees of the methodology for the evaluation of nuclear plant which originally had no designed seismic capability is welcomed. The identification of the 'bottom line plant' and its assessment at the seismic input levels proposed, backed up by the 'second line plant1 at a lower level, should at least enable a judgement to be made that the risks from the plant in the event of an earthquake are tolerable. Sensitivity analyses will provide additional confidence. An appraisal of the margins that exist in the seismic capability of the plant assessed against these events should permit an argument to be developed that the risks have been reduced to as low as is reasonably practicable. This may entail actual plant improvements. Any weak links in items of plant or structure which might cause failure to provide their functional requirements during an earthquake should be identified by this process. If numerical margins are determined, these plant items may then be ranked so as to identify areas where strengthening would decrease risk most effectively. Care must be taken that comparisons between margins are meaningful, e.g. the calculations should be made on the same basis. Additional confidence could be obtained that the risk from the seismic hazard is at an appropriate level if seismic probabilistic risk assessments are carried out. By comparing the seismic risk to the risk from other classes of hazard more effective strengthening of the safety case can be carried out as necessary. At present there is limited expertise in this field within the UK, and this is an area for further development. Future Developments 29 BNFL, NE, SNL and the UK Atomic Energy Authority have established a Nuclear Industry Group on Seismic Methodology, which is reviewing current issues in seismic assessment. This group provides a useful focus for developing the subject within the UK nuclear industry, and Nil is in regular contact with its members. The group considers all hazard or design matters which are common across the industry, and part c* its current programme involves the assessment of existing plant. It has decided to consider the introduction of a generic methodology document with respect to design and assessment criteria for both reactors and chemical plants. It is also arranging for two pilot studies to be carried out on seismic probabilistic safety assessment, one on a reactor plant and one on a nuclear chemical plant. Nil looks forward to discussing the results with the group. 102 30 The Nil will continue to maintain a 'watching brief on developments in seismic assessment in other countries. As part of this, we recently invited two American consultants to the UK to bring us up to date on progress in the USA. They visited briefly a number of nuclear installations and provided information on 'walkdowns1, seismic re-evaluation approaches and simplified seismic PRA. This knowledge was shared with the majority of the licensees in the UK, thus providing them with the opportunity to follow up on some of the techniques for use in their programmes of seismic assessment. 31 The UK has a fairly large nuclear safety research programme. Some of the research is co-ordinated under the auspices of the Health and Safety Commission and is managed by the nuclear industry in consultation with Nil through a series of technical working groups (TWGs). The programme provides for safety issues to be raised by Nil for research to be contracted out to consultants and research establishments by the licensees. The TWG on external hazards currently manages two research projects which have direct relevance to the seismic assessment of existing structures. The first project deals with the seismic performance of masonry panels, which often carry essential systems, or have the potential to damage essential plant if they collapse. The second project is reviewing the work carried out in the USA on seismic PRA and determining its applicability to UK plant. An investigation is also being carried out by one of the licensees into the applicability of American experience data for UK equipment, the results of which will be made available through this programme. 32 Additionally, Nil has an extramural support budget, which is used mainly to buy in specialist advice on a consultancy basis, but which can also provide the means to take part in international collaborations. One such project in which we are currently involved is the Seismic Shear Wall ISP (international standard problem) being co-ordinated by the OECD's Nuclear Energy Agency. Conclusion 33 The non-prescriptive nature of the UK nuclear regulatory system has allowed the licensees to adopt a variety of approaches to the seismic re-evaluation of their plant, much of which was constructed before seismic inputs were specifically considered at the design stage in this country. This flexibility has catered for the wide range of types of nuclear installation and the variety of nuclear licensees. Many of the techniques and methodologies used have been developed from, or have some parallels with, approaches used in the USA. The evaluation processes are still being developed for their application to British plant and there appears to be a move towards consensus among the licensees on the approaches to be adopted. 34 The reviews carried out to date are judged to have allowed the licensees to show that the risks from their plant in the event of an earthquake are tolerable, and have gone some way to showing they are also as low as is reasonably practicable. There is, however, room for further development particularly on the latter point as the techniques become more sophisticated. The reviews currently in progress should improve our understanding of seismic margins. 103 Acknowledgement 35 The authors wish to thank the Chief Inspector of Nuclear Installations of the Health and Safety Executive for permission to publish this paper. The views expressed are those of the authors, and do not necessarily represent those of the Inspectorate. References 1. Health and Safety Executive, Safety Assessment Principles for Nuclear Plants, HMSO, London 1992, ISBN 0-11-882043-5. 2. Health and Safety Executive, The Tolerability of Risk from Nuclear Power Stations, HMSO, London 1992, ISBN 0-11-886368-1. 3. Bye R., Inkester J. and Patchett C, A regulatory view of uncertainty and conservatism in the seismic design of nuclear power and chemical plant, Nucl. Energy, 1993, 32, No. 4, Aug., 235-240. 4. International Atomic Energy Agency, Earthquakes and Associated Topics in Relation to Nuclear Power Plant Siting, A Safety Guide, Safety Series No. 50-SG-S1 (Rev. 1), IAEA Vienna, 1991. 5. EPRI (Electric Power Research Institute) NP-6041 'A Methodology for Assessment of Nuclear Power Plant Seismic Margin1 Project 2722-1, October 1988. 6. 'Generic Implementation Procedure (GIP) for Seismic Verification of Nuclear Plant Equipment', SQUG, Revision 2A, March 1993. c 104 d/5) XA9952650 PROCEEDINGS OF SMIRT13 - POST CONFERENCE SI-MINAR 16 SEISMIC EVALUA TION OF EXISTING NUCLEAR FACILITIES SEISMIC RE-EVA LUA TION OF FRENCH NUCLEAR POWER PLANTS R. ANDRIEU Electricite de France, Direction de I'Equipement, CLI, France ABSTRACT : After a presentation of the seismic inputs which have been taken into account in the design of the French Nuclear Power Plants, the re-assessed values of these inputs are shown. Some considerations about the specificity of the French PWR program with regard to the standardisation of plants are given together with the present objectives of seismic re- evaluations. Finally the main results of the seismic re-analysis being performed for the Phenix Fast Reactor are considered. SEISMIC INPUT ASSESSMENT Since the start, in the seventies, of the extensive French PWR program the anti-seismic rules for French NPP's have been established progressively following a pragmatic approach relying on experience gained through analysis of proposals made in the course of the licensing procedures of the earliest plants. For example the notion of site related design spectra was not considered before 1978. Before this date studies were performed using correlation between intensity and maximum ground acceleration ( 0.1 g for VII MSK and 0.2 g for VIII MSK). The regulatory document RFS l-2c (Fundamental Safety Rule : " Calculation of seismic motions to be considered in safety analysis ") was only issued in October 1981. As far as re-analysis of existing plants is concerned, the approach is similar and the process of gaining experience from actual cases is still underway. The attached table 1 indicates the SSE levels which have initially been considered for the various plants still operating, taking into account standardisation. For CPY type of NPP's standard spectra was EDF 7.73 scaled at 0.2 g ZPA for all standardised buildings ; For P4, P'4 and N4 type of NPP's standard spectra was NRC (RG 1.60) scaled at 0.15 g. It should be noticed that the seismic input taken into consideration in the original design has sometimes been modified during the construction due to complementary studies, and thus assumptions may differ on one site depending on the building considered. This is in particular the case for Paluel NPP where standardised pumping station rafts were designed for an EDF spectra scaled at 0.2 g before the decision to standardise the applicable spectra which became NRC 0.15 g and was used for the remaining standardised buildings. The values given in this table are the final values at the design stage (for horizontal movements). 105 (2/5) For the first built reactors an overall review has already been done including a re-analysis of earthquake hazard at the site which considers the most recent data available on seismicity and tectonics. The corresponding up-dated ground response spectra are shown, when different from the original ones. Following this re-evaluation of the seismic input the site spectra was not in some cases enveloped for all frequencies by the design spectra. For these plants verification of the seismic behaviour has been performed, when considered necessary and on a case by case basis, taking into account the margin in the corresponding frequency domain (if any) due in particular to the ratio between MHPE and SSE. Safety margins have been included in each step of design, but have not yet been clearly evaluated. This is particularly the case of most of the components for which according to codes, the design is controlled by the French equivalent to OBE (half SSE). These design-oriented rules are inadequate for re-analysis purposes. Furthermore one has to keep in mind the French standardisation of plants, (implying for example soil-structure interaction to be made for a large variety of soil conditions), which leads to increased margins. For examples St Alban NPP, for which safety related site buildings, which were initially designed for EDF 0.1 g peak ground acceleration, have been verified for NRC 0.13 g ; Tricastin NPP was initially designed for an EDF spectra scaled at 0.2 g which was revised to a site spectra DSN 0.3 g for which analyses of the behaviour of buildings and components have not shown any necessity to modify the design. Concerning Cruas NPP, EDF made very quickly the decision to install anti-seismic bearings on the raft of the reactor building, as discussions led to the proposal of a site spectra DSN scaled at 0.3 g, instead of the initial EDF 0.2 g. This DSN 0.3 g was used for the design of safety related site buildings and for the verification of standardised buildings. For Creys-Malville Fast Breeder Reactor the initial studies concluded that the plant should be designed with a design spectrum EDF 0.1 g (corresponding to SSE intensity VII MSK, but for reasons of homogeneity with the CPY types of PWR plants, it was decided to add margin and to design the plant with a design spectrum EDF 0.2 g. This appeared to be the right decision as according to the later decrees the MHPE was re-evaluated to VI-VII MSK, thus confirming the use of this design value. SEISMIC RE-ANALYSIS PROGRAM Following these revised ground spectra, although no specific regulations on seismic safety evaluation of existing plants have been issued so far, and the discussions with the French Safety Authorities on this topic have not yet been finalised, some actions have already been undertaken : - For each specific plant or standardised type of plant a collection of floor response spectra per building and per level has been drawn. It includes those initially used in design (when existing) and a complete set of calculated response spectra with the revised seismic input (if necessary). Attention has been drawn to the consistency of the spectra between the various components and building design studies and discrepancies clarified. This new set of floor response spectra should be considered when designing new equipment or structures through a procedure under discussion as it may differ considerably from the initial data. 106 (3/5) - An updated version of the regulatory document RFS l-2c is being discussed with the French Safety Authorities. It may lead to some modifications related to the consideration of site effects, and to calculation methods for soil movements. - Recent legislation and technical rules for anti-seismic design outside nuclear field have been published. Their applicability is being studied with regards to their possible consequences on NPP design. - Walk-downs have been performed on Bugey 2-3 and Fessenheim 1-2 NPP's in order to identify the components which are not seismic resistant and may affect safety components. During these walk-downs report forms have been established which specify the type of action to follow (on site fittings or complementary studies). As far as necessary improvement of design is being studied in particular for increasing rigidity of metallic platforms and for re- qualification of component anchorages. Implementation of the corresponding upgrading measures is planned to be performed during the second ten-year outages (starting 1999). It is also presently considered to complete these walk-downs with instrumentation lines walk- downs (which were not initially checked) and to extend the procedure to the CPY types of NPP's. - Data bases are being established for each plant with consideration of all seismic related data (characteristics of soil, characteristics of structures, modelling and calculation methods used, floor response spectra, qualification spectra, tests...). - It is intended to perform for one plant of 900 MW series a Seismic Margin Assessment based on EPRI recommendations in order to estimate applicability of existing data bases to French NPP with the following objectives : - identification of elements for which design rules and qualification methods bring substantial margins, - identification of elements for which available data are not sufficient to determine existence of margins, - to make specific studies for this second category of elements SEISMIC RE-ASSESSMENT OF PHENIX NPP As far as Fast Breeder Reactors are concerned, a seismic re-evaluation of Marcoule (Phenix) NPP has been undertaken. This plant, mainly composed of steelworks structures, except Reactor Building wich is made of prestressed concrete, was built according to the existing regulations of that period (sixties) without specific requirements due to NPP's except that the plant should operate after an earthquake of intensity VII MSK, and radioactive release should be prevented up to VIII MSK. According to the re-evaluation the MHPE is VII-VIII MSK, which leads to considering distant earthquakes at an EDF spectra scaled to 0.15 g peak ground acceleration and according to RFS l-2c spectrum for close earthquake scaled at 0.2 g. Seismic calculations have been performed this year using recent calculation methods including modelisation of soil-structure interaction, modal analysis and spectral analysis. Stresses have been calculated using 2D anJ 3Dmodel (Hercule). The results of these studies are as follows: - For the Handling Building which is made of concrete for the lower levels and steel works from 8 to 34 m high with precast concrete wall facing, the calculated values showed 107 (4/5) acceleration 3 to 4.5 times higher than those of the initial design. Though the behaviour of the concrete levels is satisfactory this leads to the necessity of complementary bracing frames and reinforcement of facing panels by sprayed concrete, which has to be strongly connected to the existing structure. - For the Steam Generator Building which consists of a 34 m high steelworks structure on a concrete basement calculations performed for both input (EDF 0.15 g and RFS 0.2 g) showed that the EDF spectra was more severe (due to a different shape) and that some members were overloaded by 4 times the admissible values. Some reinforcement will be necessary even when considering behaviour coefficients. The re-assessment of Phenix NPP is still underway and the final upgrading measures are at present being designed. CONCLUSION A seismic re-assessment program is still under evaluation in France. Many actions have been undertaken in order to assess the existing state of the oldest plants with regards to their seismic resistance, but the way these studies have to be completed or used remains to be discussed with the French Safety Authorities, as specific regulations about seismic safety re- analysis of existing plants have not yet been issued in France. REFERENCES J. Betbeder-Matibet and B. Mohammadioun (1986) : Current Practice and Future Trends for Seismic Design and Analysis of French Nuclear Power Plants, Paper presented in the Specialist Meeting Earthquake Ground Motion and Anti-Seismic Evaluation of Nuclear Power Plants, Moscow, March 24-28, 1986. J. Betbeder-Matibet and P. Labbe (1990) : Simplified Seismic Analysis Methods in France, Nuclear Engineering and Design 123, pp 305-312. J. Betbeder-Matibet and B. Mohammadioun (1991): Seismic Re-assessment of French Nuclear Power Plants, Paper presented in the AIEA Meeting Seismic Evaluation of Existing Nuclear Power Plants, Tokyo, August 26-29, 1991. GLOSSARY CPY : Pressurised Water Reactor type 900 MW, DSN : Division Surete Nucteaire (Safety Authorities), FBR : Fast Breeder Reactor, MHPE : Maximum Historically Probable Earthquake, N4 : Pressurised Water Reactor type 1400 MW, NRC : US Nuclear Regulatory Commission, OBE : Operating Basis Earthquake, P4 : Pressurised Water Reactor type 1300 MW first generation, P'4 : Pressurised Water Reactor type 1300 MW second generation, RFS : Fundamental Safety Rule, (French Republic, ministry of industry), 108 (5/5) SSE : Safe Shut-down Earthquake, ZPA : Zero Period Acceleration, FIGURES Commercial SSF. Intensity Ocsign Revised POWER PLANT TYPE Program operation administrcitive ground spectrum ground start-up dute authorisation Site Standard spectrum Building Structures FESSENHEIM 1-2 CPO / 880 MWe 1970-72 1978 VIII MSI EDF0.2g BUGEY 2 to 5 CPO / 880 MWe 1971-74 1979-80 VII MSI EDFO.lg TRICASTIN 1 to 4 CPY/915 MWe 1974-75 1980-81 VIII MSK F.DF 0.2 g EDF 0.2 g DSN 0.3 g GRAVELINES 1 to 4 CPY/910 MWe 1974-76 1980-8! VIII MSK EDF0.2g EDF 0.2 g GRAVEUNES 5-6 CPY/910 MWe 1980 1985 VIII MSK EDF0.2g EDF 0.2 g DAMP1ERRE 1 to 4 CPY / 890 MWe 1974-76 1980-81 VI MSK F.DF 0.1 g EDF 0.2 g St-LAURENT Bl-2 CPY/915 MWe 1976 1983 VII MSK EDF0.1 g EDF 0.2 g BLAYAIS 1 to 4 CPY/910 MWe 1975-77 1981-83 VII MSI F.DF 0.2 g EDF 0.2 g CHINONBI-2 CPY / 905 MWe 1976-77 1984 VIII MSK EDF0.2g EDF 0.2 g CHINON B3-4 CPY/905 MWe 1981-82 1987-88 VIII MSK EOF 0.2 g EDF 0.2 g CRUAS 1-4 CPY/915 MWe 1978-79 1984-85 VIII-IXMSK EDF0.2g EDF 0.2 g DSN 0.3 g PALUEL 1 to 4 P4 / 1 330 MWe 1976-80 1985-86 VII-VIII MSK EDF0.2g NRC0.15g NRC 0.15 g St ALB AN 1-2 P4/1 335 MWe 1979-80 1986-87 VIII MSK NRCO.lg NRC0.15g NRC 0.13 g FLAMANVILLE 1-2 P4 /1 330 MWe 1979-80 1986-87 VIII MSK NRC0.l5g NRC 0.15 g CATTENOM 1 to 4 P4 / 1 300 MWe 1979-84 1987-92 VII MSK NRC0.I5g NRC 0.15 g BELLEVILLE 1-2 P4/ 1 310 MWe 1981 198S-89 VI) MSK NRC0.1 g KRC0.15g NOGENT 1-2 F4/ 1 310 MWe 1981-82 1988-89 VII MSK NRC0.1 g NRCO.lSg PENLY 1-2 P4/ 1 330 MWe 1983-85 1990-92 VII-VI1I MSK NRC0.15g NRC 0.15 g GOLFECH 1-2 P-4/I 310 MWe 1983-86 1991-94 VII MSK NRC 0.15 g NRC 0.15 g CHOOZB1-2 N4 / 1 -455 MWe 1984-87 VII-VIIIMSK Site 0.12 g NRC0.15g C1VAUX N4 / 1 300 MWe 1991-93 VIII MSK NRC0.15g NRCO.lSg MARCOL'LE FBR / 233 MWe 1961 1973 VII-VIII MSK PS 69 EDFO.lSg CREYS-MALV1LLE FBR / 1 200 MWe 1977 1986 VII MSK EDF 0.2 g Table 1 French Nuclear Power Plants Re-assessment of seismic input NEXT PAGE(S) left BLANK 109 XA9952651 NUCLEAR POWER PLANTS SEISMIC REVIEW PROGRAMME IN SPAIN. Consejo de Seguridad Nuclear Area de Emplazamientos J. G. Sanchez Cabanero A. Jimenez Juan 111 ACRONYMS CSNConsejo de Seguridad Nuclear CDFCore Damage Frecuency EPRIElectrical Power Research Institute EQEEarthquake Engineering HCLPFHigh Confidence of Low Probability of Failure IPEEEIndividual Plant Examination of External Events LLNLLawrence Livermore National Laboratory USNRCUnited States Nuclear Regulatory Commission PEDPermiso de Explotacion Definitivo (Final Operational Permit) PSAProbabilistic Safety Analysis PRAProbabilistic Risk Assessment RLEReview Level Earthquake SMMSeismic Margin Methodology SSESafe Shutdown Earthquake USIUnresolved Safety Issue 112 MAIN DATA OF SPANISH NUCLEAR POWER PLANTS UNDER OPERATION NPP Reactor Location (Province) Rated Power Type Origin of Year* (MWe) Technology JOSE CABRERA Almonacid de Zorita 160 PWR (Westinghouse) U.S.A. 1968 (Guadalajara) Sta Ma GARONA Sta Ma de Garona 460 BWR3 MARK I U.S.A. 1971 (Burgos) (General Electric) ALMARAZ1 Almaraz (Caceres) 930 PWR (Westinghouse) U.S.A. 1981 ALMARAZ II Almaraz (Caceres) 930 PWR (Westinghouse) U.S.A. 1983 ASCOI Asco (Tarragona) 930 PWR (Westinghouse) U.S.A. 1983 COFRENTES Cofrentes (Valencia) 990 BWR6 MARK III U.S.A. 1984 (General Electric) ASCO II Asco (Tarragona) 930 PWR (Westinghouse) U.S.A. 1985 VANDELLOS II Vandellos (Tarragona) 1004 PWR (Westinghouse) U.S.A. 1987 TRILLO Trillo (Guadalajara) 1066 PWR (K.W.U.) GERMANY 1988 Year of first connection to the grid LOCATION OF SPANISH NUCLEAR POWER PLANTS IN OPERATION. GEOTECTONIC OVERWIEW. SI A MAMIA GAHOUA •%m&b •>*» }:MU iSA, m K. f ^— mm % &?. ;l$ •V.i'l .••NI ,A«CO 1 -fr x ^ sW! '(AUAIJALAJXRA) !•'**( fy\ '•* ;';'• V"'' i • •• ... . » - j |JT% /vANDril.Oi. II . • i j - •' . • . • • •» |.|E} / OAHnAGONA) OBSERVED 5EISMICITV JO86 CABRERA ^ ^K: '•!;'•.'•• / (aUAOALAJAflA) ' .$•'•-•"•/ • s '...'. .• ••..;• , • ' •. - " • ' A* v" i -• •' X / • HIGH :'.,u'>"-1. ''•• ;• "'. COAENTE* (f Aif.NoiAi •-, \\:'•''•'•• '•• '('' :--»' ' '' Svl; MODERATE & ?y ^ an:F • LOW -'/' WA •1NTERPLATE DOMAIN, II BETICAS. PIRINEOS AND CANTABRIA RANGES. • FORELAND BASINS. GUADALQUIVIR AND EBRO RIVERS. • INTRAPLATE DOMAIN, HERCYNIAN MASSH" 0 CENTRAL SISTLM. IIIB'l IBERICA AND CATALANA RANGES. AFZ -AZORES FRACTURE ZONE o GSR 'GORRINGE SLAMOUNT RIDGE. HAP -HORSESHOr: ABYSSAL PLAIN. tf APF • ALENTEJO-Pl.ASENCIA FAULT. SPANISH NPP's UNDER OPERATION. ORIGINAL SITE SEISMIC DESIGN NPP SITE SSE PGA SSE RESPONSE SPECTRA COMMENTS JOSE CABRERA 0.15g No Spectra Pseudostatic analysis Sta Ma GARONA 0.1 Og J.A. BLUME ALMARAZ 1, II 0.10g NEWMARK One of the diesel buildings is design with R.G. 1.60 Spectra ASCO 1, II 0.13g R.G. 1.60 COFRENTES 0.17g R.G. 1.60 VANDELLOS II 0.20g R.G. 1.60 TRILLO 0.12g R.G. 1.60 RE-EVALUATION OF NPP WITH "EARLY" SEISMIC DESIGN ("Early" means licensed before 12.13.73, the efective publication of 10CFR100) SEP PROGRAM. o The seismic re-evaluation for Sta Ma de Garona and Jose Cabrera plants was developed according to USNRC SEP and USI-A46 methodologies (deterministic approach). Related to PED o There is ongoing a seismic re-evaluation for Almaraz plant according to USA 10CFR100 contents. f ORIGINAL SEISMIC DESIGN ~| RE-EVALUATION NPP SITE PGA RESPONSE SPECTRA PGA RESPONSE SPECTRA JOSE CABRERA 0.15g No Spectra (Pseudostatic) 0.07g USNRC NUREG/CR-0098, Rock a a St M GAROfiA 0.10g J.A. BLUME 0.10g USNRC R.G. 1.60 ALMARAZ 0.10g NEWMARK 0.10g Ongoing INTEGRATED PROGRAM TO IMPLEMENT PSA ANALYSIS IN SPANISH NPP's Spanish PSA Integrated Program was approved by CSN in June 26, 1986. Within this framework, the licensee is obliged to consider External Events [IPEEE] NPP REACTOR YEAR PSA ANALYSIS SCOPE OF EACH SPANISH NPP (*) Sta Ma GARONA 1984 Level 1 PSA, without external events. ALMARAZ I & II 1986 Level 1 PSA, with additional analysis of the reliability of containment systems and fire as an external event. ASCO I & II 1988 Level 1 PSA, with the same coverage as Almaraz but with a new external event: flooding from sources within the plant. COFRENTES 1989 Level 1 PSA, with the same coverage as Asco but with another external event: flooding also resulting from sources outside the plant. JOSE CABRERA 1990 Level 1 PSA, with the same coverage as Cofrentes, and, for the firts time, also Level 2 PSA. VANDELLOS II 1991 Levels 1 and 2 PSA, with the same external events as Jose Cabrera, but analyzing the risks in all operating modes (previously the PSAs only considered operation in power mode), but including one more external event: earthquakes. TRILLO 1992 The same coverage as Vandellos II, but including all of the remaining external events: accidents on contiguous transport routes, airplane crashes, explosions and industrial accidents, atmospheric phenomena, and all the rest which were omitted in earlier selective analysis. Sta Ma GAROftA 1994 The same coverage as Trillo. Future PSAs reviews will have, plant by plant, the same coverage as Sa Ma Garofia plant. 00 IPEEE SEISMIC RE-EVALUATION The earthquake consideration was specificaly reqired by CSN to VANDELLOS II [October 18, 1990], TRILLO [July 18, 1991] and Sta Ma de Garona [December 15, 1994] plants SEISMIC DESIGN RE-EVALUATION NPP SITE SSE PGA SSE RESPONSE SPECTRA RLE VANDELLOS 0.20g R.G. 1.60 Ongoing TRILLO 0.12g R.G. 1.60 Ongoing S. M. GAROISIA 0.10g R.G. 1.60 Ongoing ALMARAZ 0.10g NEWMARK ASCO 0.13g R.G. 1.60 COFRENTES 0.17g R.G. 1.60 JOSE CABRERA 0.07g NUREG/CR-0098, Rock 7 IPEEE SEISMIC SCOPE (I) The seismic review was specificaly reqired by CSN: To: o "Identify potential vulnerabilities and estimate the likelihood of dominant plant secuences that could lead to seismically induced core damage and fission product releases" According to the following minimum requirements: o Determine Seismic Hazard Frequency Distribution Curves based on Uncertainty Analysis and realice Sensitivity Analysis of the Results (ref. USNRC Nureg/CR-2815 -August, 1985- and Nureg/CR-2300 - January, 1983-) o Consider the local tectonic environment and local soil specific response of each plant site o Both, PRA methodology or SMM and binning methodology are considered acceptable (ref. USNRC Nureg-1407 [IPEEE], June 1991) o Perform an independent peer-review of the whole study 119 IPEEE SEISMIC SCOPE (II) Vandellos II plant proposed to CSN that Seismic IPEEE was performed using Seismic Margins Methodology (SMM) As a suggestion of the CSN staff, PWR and BWR Spanish owners groups have jointly developed a single Probabilistic Seismic Hazard Analysis for the all seven Spanish NPP sites Afterwards, they have proposed to CSN a Seismic Categorization (binning and RLE selection) of the Spanish plant sites applying SMM 120 "IPEEE SEISMIC HAZARD STUDY FOR SPANISH NUCLEAR POWER PLANTS" Spanish NPP's owners have presented to CSN to be evaluated the study: "IPEEE Seismic Hazard Study for Spanish Nuclear Power Plants", February 1994 o It consists of several documents: one "Generic Study", and seven "Plant Specific Inputs and Results", one by each site o The study contains two different parts: 1.- A Probabilistic Seismic Hazard Analysis [methodology, input data, modelling techniques, method of eliciting, manipulating, etc.], and 2- Spanish NPP's sites Seismic Categorization [binning and RLE selection] o It has been prepared by Westinghouse Energy Systems Europe EQE International Geomatrix Consultants Inc. With two Spanish experts colaboration 121 10 1.- PROBABILISTIC SEISMIC HAZARD ANALYSIS The used methodology considers only one single expert team. Consensus among team experts has been required Only a few attenuation models, selected by the experts team, are considered Vibratory ground motion is characterizated using espectral accelerations. No local soil conditions correction is taken into account Incertainties are propagated through the logic tree methodology The software used was developed by USA contractors. The validation of this software has not been demonstrated The used methodology is not strictly that of EPRI neither LLNL studies An independent peer-review has not been performed 122 11 2.- SPANISH NPP's SITES SEISMIC CATEGORIZATION o As there are only seven NPP sites in Spain [nine reactors], and only a single expert team has developed the Probabilistic Seismic Hazard Analysis, the NPP's owners propose adapting the NUREG-1407 methodology, to define the Spanish case ranking criteria. D Ranking criteria of Spanish NPP's sites proposed: HCLPF = 0,3g, with Focused Scope Mean annual probability of exceeding the 0.3g Nureg/CR-0098 espectrum is less than 10*4, OR Mean annual CDF* is less than 10~5 Vandellos II, Asco[iand nunits], Cofrentes, Garofia and Zorita plants HCLPF « Reduced scope Mean annual probability of exceeding the SSE espectrum is less than 10"4, AND Mean annual CDF* is less than 10'5 [I and ll units] and TrillO [the last one, conditional on demonstrating a sufficient margin on seismic CDF] Following fragility averaged curves obtained from existinq seismic PRA in U.S.A. and 1 1 65 (B + BU)l seismic hazard curves [HCLPF Capacity = Am • e ' ' * ; BR=0.22, B^.024] 123 12 CSN PRELIMINARY EVALUATION PWR and BWR owners groups have jointly sponsored a single Probabilistic Seismic Hazard Analysis for the all Spanish NPP's sites, and they propose a Seismic Categorization of the plant sites to apply Seismic Margins Methodology o Seismic Hazard. Draft Conclusions: the methodology developed by the USA contractors to perform the Probabilistic Seismic Hazard Analysis, is not EPRI neither LLNL the Analysis has been performed by only one single expert team. So, it is not consistent to consider the incertainties have been fully captured The validation of the software used by the USA contractors has not been demonstrated An independent peer-review has not been performed Seismic Categorization. Draft Conclusions: the adopted methodology does not follow the Nureg-1407. The Spanish NPP's binning is performed using the USA contractors self criteria [not EPRI neither NRC], and these criteria are under discussion in some cases, Seismic Categorization results differ from the seismic design decisions adopted during licensing period 124 13 CSN RECOMENDATIONS The Spanish NPP's owners Probabilistic Seismic Hazard Study: requires an independent and deep-review on methodology, used software, modelling techniques, data management and method of eliciting, in order to decide about its acceptability -* reflects the opinion of only one single experts team and it would be necessary to include more expert opinions to consider the incertainties fully captured The Spanish plant sites Seismic Categorization propose: -* requires an independent and peer-review, because of its methodology does not agree with that of Nureg-1407 To evaluate the Probabilistic Seismic Hazard Study and the Seismic Categorization, the CSN is organizing an "ad hoc" working group which will have the technical assistance of LLNL, that was USNRC adviser about these topics NEXT PAGE(S) i i«ft BLANK I ZZZ I 125 SESSION III "GENERIC WWER STUDIES NEXT PAGE(S) I left BLANK I ••oaanHMMaHM 1 2. I XA9952652 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 on SEISMIC EVALUATION OF EXISTING FACILITIES SEISMIC SAFETY OF NUCLEAR POWER PLANTS IN EASTERN EUROPE Aybars Giirpinar Antonio Godoy International Atomic Energy Agency International Atomic Energy Agency Division of Nuclear Installation Safety Division of Nuclear Installation Safety P.O. Box 100, A-1400 Vienna, Austria P.O. Box 100, A-1400 Vienna, Austria Tel: 43 1 2060 22671 Tel: 43 1 2060 26083 Fax: 43 1 20607 Fax: 43 1 20607 ABSTRACT: This paper summarizes the work performed by the International Atomic Energy Agency in the areas of safety reviews and applied research in support of programmes for the assessment and enhancement of seismic safety in WWER type nuclear power plants during the past five years. Three major topics are discussed; engineering safety review services in relation to external events, technical guidelines for the assessment and upgrading of WWER type nuclear power plants, and the Coordinated Research Programme on "Benchmark study for the seismic analysis and testing of WWER type nuclear power plants". These topics are summarized in a way to provide an overview of the past and present safety situation in selected WWER type plants which are all located in Eastern European countries. Main conclusion of the paper is that although there is now a thorough understanding of the seismic safety issues in these operating nuclear power plants, the implementation of seismic upgrades to structures, systems and components are lagging behind, particularly for those cases in which the re-evaluation indicated the necessity to strengthen the safety related structures or install new safety systems. 1.0 INTRODUCTION The concern on the safety level of existing nuclear power plants in Eastern Europe came into focus a few years ago. One of the major reasons for this concern was the recognition that some site-related external events were not properly considered in the original plant design as well 129 as the need to compare the criteria, standards and methods used to establish seismic safety in eastern European nuclear power plants with those generally accepted in international practice. Seismic safety issues generally involve two major components; those related to the derivation of the design basis parameters (i.e. seismic input) and those involving the seismic capacity of structures, equipment and distribution systems. Regarding the first component, although most Eastern European nuclear power plant sites can be characterized as low to medium seismicity, the deficiency in the geological and seismological databases as well as the methods used in the 1970s for determining the seismic hazard at a specific site, have led to the necessity to implement comprehensive hazard re-evaluation programmes of those facilities. The results of the new studies consistently indicate that the original design basis ground motion parameters had been underestimated, sometimes by a considerable margin. The issues related to the seismic capacity of structures, equipment and distribution systems are even more complex. For WWER and RBMK type nuclear power plants, structures which do not function as a pressure boundary are designed like conventional industrial frame buildings, often using precast concrete elements. Moreover, in WWER 440 and RBMK type nuclear power plants, the 'confinement' concept restricts the pressure boundaries to the lower part of the reactor building. WWER 1000 type plants, however, have a proper structural containment and therefore are inherently more robust for external events. The involvement of the IAEA in the seismic safety issues of Eastern Europe has been substantial through national and regional projects. Seismic safety review missions visited nuclear power plants in Armenia, Bulgaria, Czech Republic, Hungary, Poland, Romania, Russian Federation, Slovakia, Slovenia and Ukraine within the past five years. These countries operate different types of nuclear power plants, i.e. WWER 440/230 (Armenia, Bulgaria, Russian Federation, Slovakia), WWER-440/213 (Czech Republic, Hungary, Slovakia, Russian Federation, Ukraine), WWER-1000 (Bulgaria, Czech Republic, Russian Federation, Ukraine), RBMK (Russian Federation, Ukraine), Candu (Romania) and PWR (Slovenia). 130 The level of IAEA involvement has also varied greatly ranging from minimal in Poland (where the nuclear power programme was abandoned), Russian Federation and Ukraine, to limited in the Czech Republic, Romania and Slovenia, to extensive in Armenia, Bulgaria, Hungary and Slovakia. The extent of the involvement has depended mainly on the urgency of the need as expressed by the host country. The activities related to the assessment and enhancement of seismic safety may be considered within two time frames. The engineering services, i.e. site/plant specific reviews, are short term actions to provide recommendations to the regulatory authority and the nuclear power plant management regarding criteria and methods of assessment and upgrading. There is also the coordinated research programme dealing with the seismic safety of WWER type plants in the medium and long term. This programme is titled, "Benchmark study for the seismic analysis and testing of WWER type nuclear power plants" and involves 25 institutions from 15 countries. Another coordinated research programme on the "Assessment of RBMK type nuclear power plants in relation to external events" will begin in 1997. It should also be mentioned that substantial amount of help in terms of supply of equipment, mainly computer hardware and software for seismic hazard and structural analysis, as well as seismic instrumentation, was provided to Eastern European countries under the scope of national technical assistance and co-operation programmes. These short and long term activities will be described in the subsequent sections of this article with emphasis on the results achieved so far and what remains to be done in order to significantly improve the seismic safety of these nuclear power plants built to earlier standards. 2.0 REVIEW SERVICES A seismic re-evaluation programme for a nuclear power plant has three major components, as follows: 131 (I) the re-assessment of the seismic hazard as an external event; (ii) the evaluation of the plant specific seismic capacity to withstand the loads generated by such event, and (iii) the implementation of upgrades to buildings and components, if needed. Figure 1 shows the general flow diagram for the seismic re-evaluation process, constituted by five major phases, starting with the assessment of the original seismic input and design bases and finishing with the implementation of the upgrades for the structures, systems and components upgrades if required. The IAEA has conducted a substantial number of seismic safety review services to nuclear power plants in 10 East European countries covering 11 sites, the scope of which depended on the stage of assessment and/or upgrading of the specific plant or unit. In most of the cases the process of review started with the assessment of the original seismic input. The interim result of the re-evaluation of the seismic hazard Eastern European nuclear power plants is given in Table 1. The geological stability and the ground motion parameters are assessed according to specific site conditions and in compliance with criteria and methods valid for new facilities, which means in accordance with criteria established by the IAEA Safety Guide 50-SG-S1 (Rev. 1). Therefore, the review level earthquake RLE should correspond to the SL-2 level directly related to ultimate safety requirements, i.e. a level of extreme ground motion that shall have a very low probability of being exceeded during the plant lifetime and represents the maximum level to be used for design and re-evaluation purposes. As established in the above mentioned IAEA NUSS Safety Guide, the recommended minimum level is a peak ground acceleration of 0. lOg for the zero period of the design response spectrum. For the probability of exceedance a typical value of 10' 4/yr is usually used coupled with elastic ground response spectra. Table 2 provides an overview of the IAEA services in relation to seismic safety which were conducted to these plants within the past five years. Each service is designated with a code 132 5 indicating the type of review provided in terms of the stage of the assessment (see Figure 1). Considering that the site related investigations for reassessing the seismic input need a long time for completion (i.e. several years), a conservative preliminary value for the RLE is generally assumed for starting the activities related to the re-evaluation of the seismic capacity and upgrading of plant systems, structures and components. This may be called the interim RLE (iRLE). Another important consideration for re-evaluation purposes only is that if median plus one standard deviation was used for the definition of the peak ground acceleration, a median shaped elastic response spectra as given in US-NUREG/CR-0098, Ref [2], is permitted. 3.0 CRITERIA FOR RE-EVALUATION OF SEISMIC CAPACITY In relation to the second component of the programme mentioned in Section 2, the objective is to enhance the seismic safety in compliance with valid standards and recognized practice, using (a) "as-is" data, i.e. data reflecting the present state of the plant items; (b) more realistic criteria and methods than the ones used in the design process for at least those functions, systems, components and structures required to ensure safe shutdown and to maintain it in safe shutdown conditions, trying to avoid unnecessary conservatism. This is often a subset of the structures, systems, and components important to safety. This practice effectively ensures that a set of "dedicated, earthquake-hardened safe shutdown systems" exist at the plant. Figure 2 provides the flow diagram of the detailed work plan indicating sequence, relationship and interdependence between different tasks. The main steps and criteria used are as follows: 3.1. Identification and classification of seismic safety functions, systems and components The first step is the identification of the functions, systems, components and structures 133 required during and after an earthquake occurrence. For this purpose, the main criteria and assumptions as indicated by international practice are: (a) the plant must be capable to be brought to and maintained in a safe shutdown condition during the first 72 hours following the occurrence of the RLE; (b) safe shutdown means hot or cold shutdown; (c) simultaneous offsite and plant turbine generated power loss occurs for up to 72 hours; (d) loss of make-up water capacity from offsite sources occurs for up to 72 hours; (e) the required safe shutdown systems should fulfill single active failure criterion; (f) the required safe shutdown systems should include one main path and one redundant path; (g) other external events such as fires, flooding, tornadoes, sabotage, etc. are not postulated to occur simultaneously; (h) Loss of Coolant Accident (LOCA) and High Energy Line Breaks (HELB) are not postulated to occur simultaneously. The safe shutdown equipment list (SSEL) is the list of the minimum set of selected equipment required to achieve and maintain those safe shutdown conditions and is the most important result of this step. 3.2 Plant walkdown Emphasis should be given to the collection and compilation of original design basis data and documentation in order to minimize the effort required for the re-evaluation programme. In this regard the seismic plant walkdown has become one of the most important components of the seismic re-evaluation programme for an existing facility, with the main objectives of collection of information on as-is conditions and of assessment of the seismic capacity of equipment. The main focus of the walkdown is on anchorage of the equipment; load path from the anchorage up through the equipment; the equipment structure; and spatial systems interactions. In general, there will be three alternative disposition categories for each structure, system 134 7 and component being evaluated during the walkdown: (1) Disposition 1: a fix is required; (2) Disposition 2: the seismic capacity is uncertain and an evaluation is needed to determine if a fix is required, and (3) Disposition 3: the seismic capacity is adequate for the specified RLE and the items appear to be seismically rugged. The three alternate dispositions are primarily based on judgement and the walkdown teams must be sufficiently experienced in order to make these judgements. Screening guidelines are used to determine if the components are represented by the experience database applicable to the component in question. Unfortunately, most of the components and distribution systems in the WWER type rectors were manufactured by organizations for which seismic and testing experience has not yet been gathered and reviewed on an international scale. Similarity analysis should, therefore, be made. 3.3 Evaluation of seismic margin capacity The concept of High Confidence of Low Probability Failure (HCLPF) capacity is used to assess and quantify the seismic margins of NPPs. In simple terms it corresponds to the seismic input level at which, with high confidence (s 95%) it is unlikely (i.e. < 5%) that failure of a system, structure or component required for safe shutdown of the plant will occur. (a) The first step in the estimation of HCLPF seismic capacity is to develop a clear definition of what constitutes failure for each of the systems, structures and components being evaluated. Several modes of seismic failure may have to be considered. It may be possible to identify the failure mode which is most likely or the most dominant to be caused by the seismic event by reviewing the structure, system, component (SSC) design and to consider only that mode. 135 8 (b) The response analysis for RLE is conducted with the values of appropriate damping ratios, which may be used if the stresses in the majority of resisting building elements for the applicable loading combination are greater than 50% of ultimate strength for concrete or yield capacity for steel (i.e. Stress Level 2). The use of higher damping values, if properly justified and determined, is also permitted. (c) Nearly all structures and components exhibit at least some ductility (i.e., ability to strain beyond the elastic limit) before failure or even significant damage. The inelastic energy absorption ratio, Fu, is related to the amount of inelastic deformation that is permissible for each type of structural element. The additional seismic margin due to this inelastic energy absorption ratio (or ductility) should be considered in any margin review. In most cases, it is feasible to use linear elastic analysis techniques. When linear elastic analysis is applied, the easiest way to account for the inelastic energy absorption capability is to reduce seismic response by the Fu factor. Fu is defined as the amount that the elastic-computed seismic demand may exceed the capacity of the component without impairing its performance. It means that for non-brittle (ductile) failure mode inelastic distortion associated with a demand-capacity ratio greater than unity is permissible. Standard Fu values for different structural systems as being accepted for WWER type plants are determined considering two conditions: (I) the verification of seismic capacity of existing structures and components at WWER type reactors; and (ii) the verification of seismic capacity of structures designed using joint ductile requirements as established in applicable codes. (d) Seismic response of building structures will be evaluated on the basis of dynamic analysis of models of the soil-structure system. In order to develop appropriate structural models special attention is given to (I) structural configuration and construction details (joints, gaps, restraints and supports); (ii) non structural elements, such as masonry or precast reinforced concrete panels that may modify the structure response. Stiffness and strength of such panels, and those of their attachments to the structure, should be accounted for in 136 the formulation of the models; (iii) as-built material properties and dimensions of structural members; (iv) geotechnical data of foundation materials and their potential implications on the necessity to perform soil-structure interaction analysis, for which direct methods are usually being applied. For soil-structure interaction analysis radiation damping will not be limited but resultant composite modal damping should not exceed 20%. (e) Combinations of seismic and non-seismic loads shall be made according to the specific equations (for reinforced concrete structural elements, for masonry walls and precast reinforced concrete panels, component pressure boundaries, supports for piping and pressure components and cable raceways). The reassessed seismic input is defined for each of the horizontal components and the vertical component is assumed as a prescribed ratio of the horizontal input. (f) An earthquake experience and test based judgmental procedure to verify the seismic adequacy of the specified safety-related equipment in operating NPPs using seismic experience methods, was developed in the USA to address regulatory requirements for requalification of older plants. The procedure is primarily based upon the performance of installed mechanical and electrical equipment in conventional plants or other industrial facilities which have been subjected to actual strong motion earthquakes as well as upon the behaviour of equipment components during simulated seismic tests. With a number of caveats and exclusions for excitations below spectra normalized to 0.3 Og and in some cases 0.50g, for the zero period ground acceleration (i.e. ZPGA), it is unnecessary to perform explicit seismic analysis or test qualification of existing equipment to demonstrate functionality after the strong shaking has ended. The existing data base reasonably demonstrates the seismic ruggedness of existing equipment up to these seismic motion bounds. This conclusion should not be extrapolated beyond the classes of equipment existing in the database. (g) The issue of adequate anchorage is perhaps the most important single item which affects the seismic performance of distribution systems and components, which can slide, 137 10 overturn, or move excessively when not properly anchored. Adequate strength of system and component anchorage can be determined by any one of many commonly accepted methods. The load or demand on the anchorage system can be obtained from the floor response spectral acceleration for the prescribed damping value and at the estimated fundamental or dominant frequency of the system or component. A conservative estimate of the spectral acceleration may be taken as the peak of the applicable spectra. This acceleration is then applied to the mass of component or system at its center of gravity. Generally, the four main steps for evaluating the seismic adequacy of equipment anchorage include: anchorage installation inspection; anchorage capacity determination; seismic demand determination; and comparison of capacity to demand. (h) In addition to the inertia effects there may also be significant secondary stresses induced in systems and components by differential or relative anchor motion if the system or component is supported or restrained at two or more points. For supports it is common practice to evaluate such seismic induced anchor motion, where the relative or differential motion of the building structure at the different points of attachment should be input to a model of the multiple supported component or system. Resultant forces, moments and stresses in the support system determined from the seismic anchor motion effects acting alone shall meet the same limits for normal operation plus RLE inertia stresses. 4.0 COORDINATED RESEARCH PROGRAMMES 4.1 Background A coordinated research programme on the benchmark study for the seismic analysis and testing of WWER type nuclear power plants was initiated subsequent to the request from representatives of member states during the IAEA Technical Committee Meeting on the seismic safety of existing nuclear power plants held in Tokyo in August 1991. The conclusions of this meeting called for the harmonization of methods and criteria used in member states in issues 138 11 related to seismic safety. In particular, seismic safety concerns related to WWER type nuclear power plants were expressed. With this objective in mind, it was decided that a benchmark study is the most effective way of achieving the principal objective. Two types of ex-USSR designed WWER reactors (WWER-1000 and WWER-440/213) were selected for the benchmark exercise. Twenty five internationally recognized institutions (public or private companies) from fifteen countries take part in the seismic analysis and/or testing of the two prototypes which have been identified as Kozloduy NPP Unit 5/6 and Paks NPP, representing the WWER-1000 and WWER-440/213 respectively. Four research coordination meetings were held so far, in Paks, Kozloduy, St. Petersburg and Bergamo. Reconnaissance plant walkdowns were performed during the first two meetings for the two selected prototypes. Thirteen volumes of research material has been prepared by the participating institutions. One of the major activities of the program has been the full scale dynamic testing of the Paks and Kozloduy NPPs using blast excitation. 4.2 Prototype Plants Paks NPP Paks NPP comprises four WWER-440/213 units. It is located about 100 kms south of Budapest on the Danube river. In the original design of the plant seismic loads had not been taken into consideration. The seismic input for the plant has been recently re-assessed to be 0.25g having site specific response spectra. A major program of seismic evaluation and upgrading is underway at Paks NPP. The so called "easy fixes" have already been implemented. These mainly include equipment supports and anchorages, as well as strengthening of unreinforced masonry walls with the potential of collapsing on safety related items. 139 12 Structurally, the WWER-440/213 type NPPs lack a containment, i.e. protection from external loads. The reactor building structure is steel frame with infill walls and without proper bracing to resist lateral loads. The monolithic concrete part of the building is in the lower part of the structure and serves as an ultimate pressure boundary for extreme internal loads (Figure 3). Kozloduy NPP Unit 5/6 Kozloduy NPP site has four WWER-440/230 units and two WWER-1000 units. Units 5 and 6 refer to the 1000 MWe units. The site is located north from Sofia and on the right bank of the Danube. The soils can be classified as medium with pockets of looser sands especially under parts of the water intake canals. Originally Units 5 and 6 were designed to 0. lOg. The reassessed seismic design level is 0.20g associated with a wide band response spectrum rich in lower frequencies (mainly due to the Vrancea earthquake source). Although considerable work has been done in terms of re-evaluation and upgrading of the 'easy fixes' type for the smaller units at Kozloduy (these units were not designed for seismic loads originally), so far only a partially completed seismic PSA was performed for Unit 5. Structurally, WWER-1000 units are radically different from the WWER-440 units. The containment structure of the reactor building provides general protection from extreme external hazards (Figure 4). However the adequacy of this protection with respect to site seismicity still needs consideration. 4.3 Participation and tasks In the fourth year of its implementation, the number of participating institutions to the coordinated research program has increased to twenty five, coming from fifteen member states. Each participating institution (generally a public or private company) has a well defined work plan and task. The distribution of tasks is generally made during the research coordination meetings. The areas of interest are grouped in a matrix form and may be related to analysis, testing 140 13 or experience data pertaining to structures, equipment or distribution systems. The application could be either for the Kozloduy NPP Unit 5/6 (i.e. WWER-1000) or the Paks NPP (i.e. WWER- 440/213). Each participating institution identifies the area(s) of interest for the coming year during the research coordination meeting. A typical matrix showing the partition of tasks is given in Table 3. After determining the area(s) of interest of the institutions, a work plan is prepared in terms of concrete tasks, identifying the scope of the task, participating institutions in the performance of the task, coordinator of the task and the date of completion of the task. The following is the summary work plan (titles only) which was prepared in June 1996. Task 1. Safe shutdown systems identification/classification (task completed) Task 2. Design regulations, acceptance criteria, loading combinations (task completed) Task 3. Seismic input, soil conditions (task completed) Task 4. Standards, criteria - comparative study (task continuing) Task 5. Walkdown of reference plants (Paks and Kozloduy Unit 5 (task completed) Task 6a. Dynamic analysis of Kozloduy NPP Unit 5 Reactor Building for seismic input (task completed) Task 6b. Dynamic analysis of Paks NPP Reactor Building for seismic input (task completed) Task 7. Dynamic analysis of Paks NPP structures (benchmarking with results of Task 8) Task 7a. Reactor building (task continuing) Task 7b. Stack (task continuing) Task 7c. Worm tank (task continuing) Task 8 a. Full scale blast testing of Paks NPP (task completed) Task 8b. Full scale blast testing of Kozloduy NPP Unit 5 (task completed) Task 9. Shaking table experiment for selected components (task continuing) Task 10. On site testing of equipment at Paks and Kozloduy NPPs (task completed) Task 11. Previous component data (task continuing) 141 14 Task 12. Experience data from Vrancea and Armenia earthquakes (task continuing) Task 13. Experience data from US earthquakes (task completed) Task 14. Special topic 1 - Assessment of containment dome prestressing of Kozloduy NPP (task continuing) Task 15. Special topic 2 - Assessment of containment dome/cylindrical part for different loading combinations (task continuing) Task 16. Special topic 3 - Stress analysis of safety related piping of Kozloduy NPP (task continuing) Task 17. Special topic 4 - Dynamic analysis of selected structures of Kozloduy NPP (task continuing) Task 18. Paks NPP feedwater line analysis to be compared with testing which was already performed (task continuing) Task 19. Analysis of buried pipelines for KNPP [between DG and spray pools] (task continuing) Task 20. Analysis of buried pipelines for PNPP (task continuing) Task 21. Comparison of beam vs 3D models for KNPP and PNPP structures (task continuing) Task 22. Experience data base (WWER SQUG) initiation (task continuing) Task 23. Consolidation of results and reports (task continuing) Task 24. Dynamic analysis of Kozloduy NPP Unit 5 structures [benchmarking with results of Task 8] (task continuing) Task 25. Comparison of blast and vibrator tests for KNPP (task continuing) Thirteen volumes of research material has been compiled reflecting the results of the completed tasks. These volumes are titled as follows: Volume 1. Data related to sites and plants (Paks and Kozloduy NPPs) Volume 2. Generic material: codes, standards, criteria Volumes 3A, 3B, 3C, 3D, 3E. Kozloduy NPP, Units 5/6: Analysis/testing Volumes 4A, 4B, 4C, 4D. Paks NPP: Analysis/testing Volumes 4A, 5B. Experience data 142 15 4.4 Full scale dynamic test of Paks and Kozloduy NPPs One of the most significant tasks already completed is the full scale dynamic testing of the Paks NPP. The test was conducted by Ismes, an Italian consulting company and Paks NPP with assistance from local contractors especially for the realization of the blasts. The test was performed in December 1994 following a two week preparation period for placing the instruments and recording of smaller test blasts. The blast location was about 2.5 kilometers from the reactor building. Two main blasts were performed with a total each of 300 kilograms of TNT charge. Three free field locations were selected for instrumentation. Two of these had two borehole (@40 meters and 15 meters depth) and one surface recording. About 40 meters corresponds to the depth of the firmer geological formation. An additional (fourth) instrument was located about 12 kilometers away in order to provide some information on attenuation characteristics. A large number of seismometers and accelerometers were mounted in the reactor building (some also in other buildings) to record the structural response. Instruments were also placed on certain heavy components and tanks. Both blasts used a time delay to enhance the duration of the motion so that an adequate time series analysis was possible. In most locations a motion of about 20 seconds was recorded. The records are of very high quality. It should also be noted that all of the instruments functioned as intended. One set of free field recordings have been made available to the benchmark programme participants. Locations and directions of the in-structure instruments have been indicated and the participants have been asked to make a blind prediction of the response recorded at these locations. All the relevant dynamic soil properties and structural properties have been provided to the participants. A similar test was carried out for the Kozloduy NPP Unit 5 in June 1996. The test was again performed by Ismes and Kozloduy NPP as well as local contractors assisted in the test. All instruments, both free field and in-structure, functioned as intended. The results of the test have 143 16 been recently processed. 5.0 CONCLUDING REMARKS A review and comparison of Figure 1 and Table 1, presented earlier reveal some indication of present picture of the seismic safety situation of nuclear power plants with which the IAEA had significant involvement. The first observation from Table 1 is that the reassessment of the seismic design basis has been completed for three of the sites (i.e. Kozloduy, Paks and Medzamor) while for Bohunice and Mochovce this activity is continuing. For all the sites in question, the reassessment has yielded significantly greater RLE values. This, in turn, indicates that for most of the plants, the capacity check yields the result that the plant requires upgrading (i.e. inadequate seismic capacity). The last two columns of Table 1 generally indicates good progress in easy fixes, i.e. mainly supports and anchorages of mechanical and electrical components. For some cases, this included more substantial upgrades involving replacement of batteries and strengthening of unreinforced masonry walls to prevent spatial interaction. Similar progress is unfortunately not the case for structural upgrades or when the installation of additional safety systems were required. Due to bigger funding and longer outage requirements, structural upgrades will probably take much longer to complete. Unfortunately, the overall seismic safety of these NPPs will not have been improved to the target levels, until structural upgrades are implemented. REFERENCES [1] IAEA, Safety Guide 50-SG-S1 (Rev. 1), "Earthquakes and associated topics in relation to nuclear power plant siting", IAEA, Vienna, 1991. [2] NUREG/CR-0098, "Development of Criteria for Seismic Review of Selected Nuclear Power Plants", N.M. Newmark and W.J. Hall, NRC, May 1978. 144 17 [3] Gurpinar, A. and Godoy, A.R., "Seismic safety of nuclear power plants in Eastern Europe", Proc. Tenth European Conference on Earthquake, A.A. Balkema/Rotterdam, 1995. ACKNOWLEDGMENT The authors acknowledge that parts of this article have been taken from technical material contributed by various consultants in the course of the past five years either in relation to IAEA Engineering Safety Review Services for external events or the Coordinated Research Programme on "Benchmark study for the seismic analysis and testing of WWER type NPPs". In particular, the criteria for the re-evaluation and upgrading of existing NPPs were developed by several authors led by Mr. J. D. Stevenson. File: s/godoy/ smirtl3/sl6paper.wpw 145 18 List of Tables and Figures Table 1: Seismic Safety Status of Selected WWER NPPs in Eastern Euorpe Table 2: 5 Year Summary of IAEA Site/Seismic Safety Review Services to Eastern European NPPs Table 3: Partition of Tasks for Participating Institutions Figure 1: Flow Diagram for Seismic Re-evaluation and Upgrading of Existing Nuclear Power Plants Figure 2: Detailed Flow Diagram for the Assessment and Improvement of Seismic Safety Figure 3: Cross Section of Paks NPP Figure 4: Cross Section of Kozloduy NPP, Unit 5 146 19 Table 1. Seismic Safety Status of selected WWER NPPs in Eastern Europe Plant Original Reassessed Capacity Upgrades SDB SDB (RLE) Check to RLE Easy Fixes Structural Kozloduy 440 NED 0.2g Neg. Yes No Kozloduy 1000 O.lg 0.2g PSA (*) No No Bohunice VI NED 0.25g? Neg. Some Some Bohunice V2 NED 0.25g? Neg. Some No Mochovce 0.06g O.lg? No No No Paks NED 0.25g Neg. Yes No Armenia 0.1e/0.2e 0.35s No No No Legend: SDB: Seismic Design Basis NED: No Explicit Design Neg.: Inadequate seismic capacity for the reassessed SDB (RLE) A question mark indicates an ongoing activity with a preliminary indication of the reassessed SDB (RLE) No: The activity has not started yet Incomplete 20 Table 2. 5 Year Summary of IAEA Site/Seismic Safety Review Services to Eastern European NPPs Country Plant Number of services (1990-95) W s SI sc Armenia Armenia - - 5 3 Bulgaria Kozloduy 1-4 1 2 5 5 Bulgaria Kozloduy 5-6 - - 1 2 Bulgaria Belene - 2 2 - Croatia (Site Survey) - 1 - - Czech Republic Temelin 2 4 - - Czech Republic (Spent Fuel Storage) - 1 1 - Hungary Paks - - 6 5 Romania Cernavoda 1 - - 2 Russian Federation (Generic WWER) 1 - - - Russian Federation Smolensk - - 1 1 Slovakia Bohunice VI - - - 3 Slovakia Bohunice V2 1 - 2 - Slovakia Mochovce 1 - 2 3 Slovenia Krsko 1 - 3 1 Ukraine Crimea - - 1 - TOTAL 8 10 29 25 Legend: W: Workshop S: Site Safety Review SI: Review of Seismic Input and Tectonic Stability SC: Review of Seismic Capactiy 148 21 Table 3. Partition of Tasks for Participating Institutions Structures Components Distribution Systems Kozloduy NPP Paks NPP Kozloduy NPP Paks NPP Kozloduy NPP Paks NPP IZSIIS (M) SAGE (B) Siemens (G) Siemens (G)- K-NPP P-NPP (H)- Siemens (G) IZIIS (M) VNIIAM (RF) P-NPP (H) Siemens (G)- Siemens (G) MD (CR) AEP (RF)- WESE (B) CKTI (RF)-SA (CR) WESE (B) CKTI (RF)-SA Analysis CL (BG) P-NPP (H)- BRI (BG)- VNIIAM (RF) SP (CH)-BRI (BG) (CR) Siemens (G) SP (CH) Argonne (US) CL (BG) WESE (B) MD (CR) EQE (US) EQE (US) EQE (BG) Wolfel (G) EQE (USA) CL (BG) K-NPP Ismes (I) IZIIS (M) P-NPP (H) CKTI (RF) CKTI (RF) Testing Ismes (I) P-NPP (H) AEP (RF) VNIIAM (RF) VNIIAM (RF) P-NPP (H) VNIIAM (RF) IZIIS (M) VNIIAM (RF) K-NPP Siemens (G) Siemens (G) AEP (RF)- SA(R) SA(R) EQE (USA) SAS (SR) SAS (SR) Siemens (G) EQE (USA) EQE (USA) AEP (RF) EQE (USA) Siemens (G) AEP (RF) SA (CR) Experience VNIIAM (RF) SA (CR) VNIIAM (RF) VNIIAM (RF) Data WESE (B) VNIIAM (RF) WESE (B) WESE (B) SA (US) WESE (B) SA (CR) SA (US) SA (US) SA (US) 22 Figure 1 Flow Diagram for Seismic Re-evaluation and Upgrading of Existing Nuclear Power Plants ASSESSMENT OF ORIGINAL SEISMIC INPUT AND DESIGN BASIS adequate RE-EVALUATION OF SEISMIC INPUT Specific to the site - RLE ASSESSMENT OF SEISMIC CAPACITY OF THE PLANT TO THE NEW DEFINED RLE adequate DESIGN OF UPGRADES IMPLEMENTATION OF FURTHER UPGRADING ACTION 150 23 Figure 2 Detailed Flow Diagram for the Assessment and Improvement of Seismic Safety GEOTECHNICAL AS-BWLTIDESIGK FUNCTIONS/SYSTEMS DATA DATA COLLECTION /COMPONENTS CLASSIFICATION Tasfc 2 T&sk 4 SSEL STRUCTURE MODEL SOIL-STRUCTURE RESPONSE Task 5 STRUCTURE IN-STRUCTURE INTERNAL RESPONSE FORCES SPECTRA (FRS) T SOIL STRUCTURE DISTRIBUTION SYSTEMS EQUIPMENT: CAPACITY CAPACITY (A) functional; (A) funcfionai EVALUATION EVALUATION (B) structural integrity (0) structural integrity Task 6 Task 7 Task 8 Task 9 1 OK PRJORiTLZATION HIGH LOW FURTHER PRIORITY PRIOKITY ANALYSIS Task 10 Task 10 Task 10 £ UPGRADES DESIGN i' AND IMPLEMENTATION OK 151 FIGURE 3. CROSS SECTION OF PAKS NPP (WITH INDICATION OF INSTRUMENT LOCATIONS FOR THE FULL SCALE DYNAMIC TEST) Gallery Reactor Condensation Tower Building Building FIGURE 4 CROSS SECTION OF KOZLODUY NPP, UNIT 5 (WITH INDICATION OF INSTRUMENT LOCATIONS FOR THE FULL SCALE DYNAMIC TEST) .'.I-; 50 too?:.: I ! SCOSCO . " >?*.(>•< c:: 2.(*m x 'a •a *m m lilii TV M urm l 0 it cfi u iic ou J»o[ I $90 (r) i XA9952653 PROCEEDINGS OF SMIRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES COMPARISON OF EX-USSR NORMS AND CURRENT INTERNATIONAL PRACTICE IN DESIGN OF SEISMIC RESISTANT NUCLEAR POWER PLANTS B .Hauptenbuchner Technical University of Dresden, Dresden, Germany M. David David, Consulting, Engineering and Design Office, Prague, Czech Republic ABSTRACT: Seismic hazard has been estimated according to ex-USSR norms in the original designs of WWER - type Nuclear Power Plants (NPP) in former Soviet Union as well as in all former east European countries. For some steps of the design the national standards has been also taken into account. The original ex-USSR norms and instructions has been several times changed and improved during the time. This contribution is dealing with the development of ex-USSR norms and regulations with the aim to recognise some most important differentiations in comparison with corresponding western or international ones from point of view of civil structures. The understanding of relations of these documents is very important for seismic qualification and upgrading of WWER-type NPPs. The main Soviet/Russian Standards and Regulations related to the seismic design and qualification of NPP structures as SNiP II-A. 12-69, VSN 15-78, SNiP II-7-81, PiNAE G-7-002-86, NTD SEV etc. have been taken into consideration and compared with western or international standards as IAEA 50-SG-S1, IAEA 50-SG-D15, KTA 2201.1-6, ASCE 4-86 etc. The numerical examples of structural seismic qualification has been elaborated according to different standards for better understanding and in order to determine the degree of safety referring to corresponding standards. The authors has tried also to take into account the way of application of ex-USSR norms. The comparison of different norms and regulations has been analysed and corresponding conclusions and recommendations have been derived. These conclusions and recommendations can be helpful by the seismic qualification and upgrading of WWER-type NPPs. 1. INTRODUCTION Many countries, particularly countries with regions with earthquake hazard have, developed standards for seismic design of structures. However, these standards were not sufficient for the construction of Nuclear Power Plants from many reasons. First of all, the required safety of technological systems and structures related to nuclear safety is much higher than the safety required for living houses or ordinary industrial equipment and buildings. Furthermore, these existing standards have not defined methods and acceptance criteria for technological systems and structures of NPPs at all. In connection with the development of NPPs, the corresponding standards have to be also developed. The elaborated standards for design of NPPs in seismic regions have been usually improved step by step in accordance with permanent increase of safety requirements on one site and increase of corresponding experience and knowledge on the other site. 155 The comparison of initial Soviet and current Russian Standards with west standards must be studied in connection with their historical development. 2. REVIEW OF USSR STANDARDS USED DESIGNS OF NAP The historical development of initial Soviet and current Russian Standards can be divided into several periods according to their main characteristics. The most important standards are presented in Table 1. Table 1. Historical Development of Initial Soviet and Current Russian Standards Related to general Safety Issues of NPPs. Period year Standards, Comment Regulations 1st Period up to the end of 70th SNiP II-A.12-69* General seismic years building standard without respect to NPPs 2nd Period up to the end of 80th SNiP II-A.12-69* VSN 15-78 is an years SNiP II-7-81 extension of general VSN 15-78 seismic building OAG 130.03 standards SNiP for RTM 108.020.37-81 NPPs RD 16 20.1-86 3rd Period from the late 80th PiNAE G-5-006-87 years NTD SEV PiNAE G-l-011-89 PiNAE G-7-002-86 NTD SEV 4214-86 RD 16 20.1-86 RD 25818-87 OTT-87 GOST 17516.1.90 GOST 16962.2-90 Note: Thick printed Standards and Regulations refer to earthquake and associated topics in relation to NPP sitting and seismic analysis of NPP civil structures. 3. COMPARATIVE STUDY WITH WEST STANDARDS Following international, westeuropeans and USA Standards and Regulations have been mainly used for comparison with soviet/russian standards: (ref. Table 2.) 156 Table 2. West Standards and Regulations Used for Comparison with Soviet/Russian Standards Standards, Regulations Comments IAEA50-SG-S1 IAEA Safety Guide IAEA50-SG-D15 IAEA Safety Guide: Seismic Design and Qualification for Nuclear Power Plants KTA 2201.1 German Regulations KTA 2201.2 KTA 2201.3 KTA 2201.4 KTA 2201.5 KTA 2201.6 ASCE 4-86 ASCE Standard: Seismic Analysis of Safety Related Nuclear Structures and Commentary. 3.1. General Comparison of Standards and Regulations The main steps of the design of seismic resistant NPPs structures according to different standards will be studied now in order to compare the east and west standards. Earthquake levels: Almost all standards use two levels of earthquake: Despite that the terminology as well as the definition of maximal estimated earthquake is very different, according to the practice east - west. The earthquake with annual probability 10 is assumed as the maximal estimated earthquake in all countries. Some countries ( Germany) use the maximal earthquake only. (Ref. Table 3.). Table 3. Comparison of Earthquake Levels Standard, Regulation Lower Design Earthquake Higher Design Level Earthquake Level IAEA 50-SG-S1 SL-1 SL-2 Initial Soviet VSN 15-78 DE MDE Current Russian PiNAE G-5-006-87 the same the same USA (SRP, RG) OBE SSE Germany KTA 2201.1 - MCE Notes: 1. The SL-1 level corresponds to an earthquake often denoted as Operational Basis Earthquake (OBE). 2. The SL-2 level earthquake corresponds to an earthquake level denoted as Safe Shutdown Earthquake (SSE). 3. Design earthquake (DE) is an earthquake which may occur one times per 102 years. The plant should remain functional during and after DE. 157 4. Maximal Design Earthquake (MDE) is an earthquake which may occur one times per 10 years. 5. The Operating Basis Earthquake (OBE) is an Earthquake which, considering the local geology and seismology and specific characteristic of local subsurface material, could reasonably be expected to affect the plant site during the operating life of the plant. It is the current practice in the US that the OBE is taken one-half of the SCE ground motion. 6. The Safe Shutdown Earthquake (SCE) is that earthquake which is based upon an evaluation of the maximum earthquake potential considering the regional and local geology and seismology and specific characteristic of local subsurface material. It is that earthquake which produce the maximum ground motion for which the safety-related structures, systems and components are designed to remain functional. 7. KTA 2201.1 requires only the higher design earthquake level - Maximum Credible Earthquake (MCE). There is no minimum Peak Ground Acceleration value (PGA) for the higher earthquake in Soviet/Russian standards and regulations for seismic design of NPPs as recommended f.e. by the IAEA document (0,10 g). 3.2. Seismic Categorisation of Structures Seismic categorisation of structures systems and components according to their relationship to safe shutdown of the reactor is assumed in all standards. The Russian standards use more detailed seismic categories, I, la, Ib, III, but there are no substantial differences in the definition of buildings, structures and technology involved in seismic category 3.3. Determination of Seismic Forces 3.3.1. Structural Seismic Analysis according VSN 15-78 Seismic analysis of the structures according to the VSN 15-78 is based on the utilisation of equivalent static seismic forces. These forces are determined by the method of modal analysis of the structure using additional coefficients expressing the features of the structure, of the soil, damping etc. The seismic forces have to be determined according to the relations of the common USSR seismic standard SNiPII-A. 12-69 with some modifications which are described in the VSN-15-78. Seismic forces are calculated according to the formula given in the SNiPII-A. 12-69 and modified by VSN-15-78: s ik - m Qk Ks Pi T| & (1) where it is: m - coefficient according to the VSN 15-78 m = 2,5 for buildings of 1st seismic category m = 1,0 for buildings of 2nd and 3rd seismic category Qk - weight concentrated in the node "k" Ks - seismic coefficient according to the table 6. (3} - dynamic coefficient, Pj = 1/Tj < 3,0 >0,8 for structures with low damping (towers etc.) 158 for structures without effective shearwalls and with the ratio of the height of co- lumns their width equal to 25 or higher coefficient p should be multiplied by the factor 1,5 ,with the ratio equal to 15 or less coefficient p should be not increased. - modal analysis factor - ref. Eq.(2) Modal analysis factor is determined in accordance with the equation: Xi(xk) I Qj.Xi(xj) "Hik = (2) Seismic load will be increased for buildings with the number of floors n > 5 by the factor l+(n-5) but not higher than 1,5 and for the buildings of precast big elements or concrete cast in situ walls by the factor l+0,06(n-5) but not higher than 1,3. Only one factor, higher from the factor for increasing the value B or the factor taking into account the number of floors should be used in the calculation of seismic loads. According to VSN-15-78 furthermore it is necessary to take into account local site conditions. Local site geological and hydrogeological are introduced into consideration by corresponding regulation of site intensity (ref. Table 4.). Table 4. Influence of Local Geological and Hydrological Conditions Site Intensity (ball) Category of the Soil 4 5. 6 7 & I. Category of the soil: 3 4 5 6 7 rock or soil with big pieces and underground water level h=> 15 m II. Category of the soil: 4 5 6 7 8 disintegrated rock with big pieces and underground water level 6 < = h < = 1 Om or sand and clay soil, h >= 8m III. Category of soil disintegrated rock with big 5 6 7 8 9 pieces, underground water level h < = 3 m, or sand and clay soil h < 4 m Seismic intensity is furthermore adapted according to the Table 5. 159 Table 5. Seismic Intensity According Seismic Category of Buildings Seismic Intensity for the Characteristics of the building Calculation (ball) 4 5 6 7 8 NPP buildings and structures seismic category I. and II. 5 6 7 8 9 NPP buildings and structures seismic category III. 4 5 6 7 8 Seismic coefficient should be determined with the respect to the resulting calculation seismic intensity using the Table 6. It should be noted that the seismic intensity used in the calculation can be different for different buildings on the same site Table 6. Seismic Coefficient Ks ( SNiP II-A. 12-69 ) Determined Seismic Intensity 4 5 7 6 8 9 (ball) Coefficient Ks 0,003 0,006 0,025 0,013 0,050 0,100 3.2. Structural Seismic Analysis According PiNAE G-5-006-87 The standard PiNAE G-5-006-87 constitutes an extension of the general Russian seismic standard SNiP II-7-81 The seismic analysis of building and structures of seismic categories I and II according to the standard PiNAE G-5-006-87 can elaborated by the help of static equivalent forces determined by the formula: Sik = Kc-K2Kli;.QK-A-p.-Tiik (3) where it is: K. £ is the coefficient taking into account the specific conditions of the operation of NPP and it is assumed: for structures of the first seismic category 160 K s = 0,625 for structures of the second seismic category Ke = 0,500 for structures of second category but not for storing radioactive products and media K e= 0,300 K2 is the coefficient taking into account the structural systems of buildings and structures, it is assumed in accordance with SNiP II -7-81. For structures with over-crossing walls of in-situ and/or of precast reinforced concrete the value of the K2 coefficient should be assumed as 1, Kv|/, Qk, T]ik, pi, should be determined in accordance with general seismic standard SNiP II-7-81 A is the coefficient the magnitude of which should be assumed according to the table 7: Table 7. Seismic Coefficient A IllliillilllillIllllllIllllllll A = 0,025 0,050 0,100 0,200 3.4. Load Combinations Load combinations together with seismic load for NPP building and structures as prescribed in VSN 15-78 and PiNAE G-5-006-87 are summarised in Table 8. The current load combinations are, in general consistent with IAEA recommendations (50-SG-D15). They are more conservative than the KTA requirements ( KTA 2201.3). Table 8. Seismic Load Combinations for NPP Civil Structures Standard Seismic Category I n m VSN 15-78 NOC+MDE+M NOC+ DE SNiP II-A. 12-69 DA PiNAE G-5-006-87 NOC+DE+DA Ha SNiPII-7-81 161 NOC+MDE NOC+DE lib ANOC+MDE ANOC+DE lib NOC+DE+DA NOC - Normal Operational Condition MDE - Maximal Design Earthquake MDA - Loads due to Postulated Maximal Design Accident ANOC - Abnormal Operational Conditions 3.5. Ductility of Structures Unfortunately there are no detailed instructions in all standards about the utilisation of ductility of structures, despite the fact that the positive influence of structural ductility to bearing capacity of structures in the case of earthquake is well known from the literature as well as from real earthquakes. Most probably the coefficient K in soviet standard PiNAE G-5-006-87 (Ks = 0,625 for Category I structures) refers to the ductility of structures, but on the other hand this coefficient can be used without limits for all structures (ductile and non ductile) according this standard. 3.6. Soil Structure Interaction (SSI) Soil structure interaction is very important for the structures based on the soft soil, but there are no instructions or recommendations in both Soviet/Russian Standards and in Interatomenergo standard referring to this problem. The influence of soil to the seismic response of structures is in Soviet/Russian Standards introduced by the soil quality dependent ground response spectrum and by changing of the seismic intensity of the site (ref. Table 4). 3.7. Damping of Seismic Structural Vibrations Damping of seismic structural vibrations recommended in different standards, east and west, is very similar. Values of damping recommended for most important structures are presented in Table 9. 162 Table 9. Damping Values According to Different Standards (Percent of Critical Damping) liillilllllll :lllilB.BiIl ||f|pPt;llP#I:§;;l iliiiii||p#i:|lilil pilllllllllli! •• :• •••:••-:-:•••:•:•.:•-••: -.-.:•:••••.• •::••••-"•: • : •.•:•• •;• : •:•: • ' Reiforced Concrete 4 4 7 7 Structures Prestressed Concrete 4 ~> 5 Structures Welded Steel Structures t- o 2 4 4 Bolted Steel t- o 4 7 7 Structures 7 Masonry Walls 4. COMPARISON OF SEISMIC STRUCTURAL LOADS ACCORDING TO EAST STANDARDS WITH CURRENT PRACTICE Two soviet standards have been used for the design of seismic resistant NPPs in east Europe in past years, firstly the standard VSN 15-78 and later PiNAE G-5-006-87. The comparison with today practice will be demonstrated by the calculation of seismic load in the node Qk using the both soviet standards. As the representative standard for today practice the American standard ASCE 4-86 will be used. It should be noted that there are no substantial differences between this and other west standards with the respect to demonstrated comparison. The same seismic and geological input will used in all calculations. No Soil Structure Interaction will be assumed in all calculations. The used seismic input: Seismic site intensity : 8 ball Soil category II. Static equivalent force will be calculated for reactor building - (Seismic category I) Calculation according to the standard VSN-15-78 - Seismic intensity 1 = 8 ball (local geological conditions - soil category II ref.Table 4) - Building of the seismic 1st category. Intensity I = 8+1 = 9 163 10 - factor with the respect to number of floors = 1,0 - factor for increasing coefficient p was adopted 1,0 - coefficient for buildings of 1st category m = 2,5 - seismic coefficient Ks for increased Intensity 9 is Ks=0,100 ref. Table 6. - maximal value of pi is for fi > 2,85 Hz, pj = 3,0 Maximal modal force in the node k and for the frequency i is: Sik = 0,75-Qk where it is: Qk weight in the node "k" r]jk modal factor Calculation according the standard PiNAE G-5-006-87 The coefficients of the equation for equivalent static forces according to assumed input are: - Ke= 0,625 conditions of the NPP operation - K-2 = 1,00, structural solution - K \\> = 1,00 type of structure - A = 0,20 for the intensity 8 ball - Pi = 2,7 value for fj > 2,43 Hz Maximal modal force in the node k for the frequency fj is: Sik = 0,3375.QkTlik Calculation according to the standard ASCE 4-86 Seismic and geological input is the same as before but following data must be added. - ground acceleration corresponding to the site intensity 8 ball is assumed to be cca 0,25 g as the maximum. - structural damping is 5 % - the amplification factor for the site-independent response spectra (acceleration) is 2,71, (ref. ASCE 4-86) maximal acceleration spectral value is 2,71 .0,25 = 0,677g for fj in the interval 1,5-10 Hz. Maximal modal force in the node k and for the frequency fj is Sik = Sa.Qk.-pile = 0,677 .Qk.r|ik Comparison of the results The comparison of the modal forces determined using different standards is presented in Table 10. The modal force obtained according to ASCE 4-86 is denoted as 100 % 164 11 Table 10. Comparison of Modal Forces Determined Using Different Standards (ductility factor 1,6 used in the PiNAE - G-5-006-87 only) Standard Modal Force % USA Standard ASCE 4-86 100,00% VSN 15-78 110,70% PiNAE - G-5-006-87 49,85 % * Notice: * Equivalent static force was reduced by the revers value of ductility factor 1/1,6 = 0,625 It should be pointed out, that static equivalent forces according to the standard PiNAE - G-5- 006-87 are reduced with coefficient Ke (Ke = 0,625 for structures, seismic category I) which could be suggested as the reverse value of ductility factor. The coefficient Kg is used for all structures, without respect to their mode of failure. For structures with ductile mode of failure, the ductility factor may be introduced in all calculations. For the reason of comparison of standards, we introduce the same ductility factor in the same way as it is assumed in the standard PiNAE - G-5-006-87 (equivalent static force will be reduced by the revers value of ductility factor 1,6) in all calculations. Comparison of the results is presented in Table 11. Table 11. Comparison of Modal Forces Determined Using Different Standards (ductility factor 1,6 used in all compared Standards ) Standard Modal Force % USA Standard ASCE 4-86 100,00%* VSN 15-78 110,70%* PiNAE - G-5-006-87 79,76%* Notice: * Equivalent static force was reduced by the revers value of ductility factor 1/1,6 = 0,625. 4.0. CONCLUSION The Standard VSN 15-78 and later PiN AE G-5-006-87 (since July 1, 1988) have been used for the design of seismic resistant NPPs in eastern Europe and USSR. These two standards have been analysed in accordance with to day west standards. Significant differences between the Russian and west European standards have been found particularly in the global construction of the standards. On the other hand the main design 165 12 philosophy of all standards is very similar. It is very difficult to compare the main parameters as damping value, load combinations etc. of west and Soviet standards, because different methods for the evaluation of structural seismic forces are assumed. In order to overcome this problem the comparison of resulting seismic forces according to two cited Soviet Standards and the USA ASCE-64 Standard has been elaborated in this contribution. The results of this comparison are presented in the section 4. of this contribution. On the basis of this comparison it can be theoretically concluded that despite some differences the structural seismic forces calculated according to Soviet-Russian Standards are in good relation to that one determined by west standards. It seems that the standard VSN 15-78 was very conservative and the new Soviet-Russian Standard PiNAE G-5-006-87 is less conservative, because it uses the effect of ductility of structures or other effects increasing their bearing capacity without respect to their characteristics. It should be pointed out, that the comparison of calculations according to different standards was elaborated without respect to Soil Structure Interaction, because there are no recommendations or comments to this problem in both Soviet standards. The results of comparison could be very different when SSI will be included, particularly in the case of foundations on soft soil. According to the authors experience, usually the substantial problems have not been caused by the utilisation of Soviet Standards, but mainly by their not correct interpretation in the practice. The improved Soviet seismic Standard PiNAE - G-5-006-87 came too late (valid since July 1988) and not exactly defined methodology have been used for designs of NPPs for a long period. From all above mentioned reasons the comparison elaborated here cannot be generalised and detailed seismic verification will be always necessary in concrete cases of seismic upgrading of NPPs WWER-type. However, despite that the authors of this paper do believe that here presented comparison of east and west standards can contribute to better understanding of initial seismic designs of VVER -type NPPs. REFERENCES [11 David, M., Benchmark Study for Seismic Analysis and Testing of WWER-Type NPPs, Standards, Criteria, Comparativ Study, Prague, 1994 [21 Masopust, R., Benchmark Study for Seismic Analysis and Testing of WWER-Type NPPs, OriginalSeismic Design Data and Application ofSMA and GIP Methodolodogies - Volume 1, Pilsen, 1994 131 Newmark, N.,M., Fundamentals of Earthquake Engineering, Prentice-Hall, Inc. Englewood Cliffs, N. Y., 1980 [41 Birbraer A., N, Sulman, S.,G., Procnost i Nadeznost Konstrukcij AESpri Osobych dynamiceskich Vozdejstvijach, Energoatomizdat, Moskva 1989 151 WolfJ., P., Dynamic Soil-Structure Interaction, Prentice-Hall, Inc.,Englewood Cliffs, N.J. 07632 STRUCTURAL SEISMIC ANALYSIS VSN 15-78 166 1 •20 XA9952654 PROCEEDING OF SMiRT 13 - POST CONFERENCE SEMINAR NO. 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES SEISMIC PRA, APPROACH AND RESULTS Robert D. Campbell EQE International ABSTRACT: During the past 15 years there have been over 30 Seismic Probabilistic Risk Assessments (SPRAs) and Seismic Probabilistic Safety Assessments (SPSAs) conducted of Western Nuclear Power Plants, principally of US design. In this paper PRA and PSA are used interchangeably as the overall process is essentially the same. Some similar assessments have been done for reactors in Taiwan, Korea, Japan, Switzerland and Slovenia. These plants were also principally US supplied or built under US license. Since the restructuring of the governments in former Soviet Bloc countries, there has been grave concern regarding the safety of the reactors in these countries. To date there has been considerable activity in conducting partial seismic upgrades but the overall quantification of risk has not been pursued to the depth that it has in Western countries. This paper summarizes the methodology for Seismic PRA/PSA and compares results of two partially completed and two completed PRAs of soviet designed reactors to results from earlier PRAs on US Reactors. A WWER 440 and a WWER 1000 located in low seismic activity regions have completed PRAs and results show the seismic risk to be very low for both designs. For more active regions, partially completed PRAs of a WWER 440 and WWER 1000 located at the same site show the WWER 440 to have much greater seismic risk than the WWER 1000 plant. The seismic risk from the 1 OOOmw plant compares with the high end of seismic risk for earlier seismic PRAs in the US. Just as for most US plants, the seismic risk appears to be less than the risk from internal events if risk is measured is terms of mean core damage frequency. However, due to the lack of containment for the earlier WWER 440s, the risk to the public may be significantly greater due to the more probable scenario of an early release. The studies reported have not taken the accident sequences beyond the stage of core damage hence the public heath risk ratios are speculative. 1. INTRODUCTION Internal event PRAs have been conducted in the US for a period of about 20 years. The addition of seismic and other external events to these studies followed shortly. In the early 1980's the USNRC sponsored a Seismic Safety Margins Research Program (SSMRP) to study methodology and results for PRAs. Reference 1 is a summary of the overall program and results. Concurrently, several utility sponsored PRAs, including seismic and other external events were conducted. These earlier studies were driven primarily by the regulators to address seismic risk at perceived high risk sites due either to the population density or due to technical issues related to the P:\950000\100\TECHPAPR.DOC/irv 167 2-20 design bases. A PRA Procedures Guide, (Ref. 2) was developed in 1983 to provide guidance on the conduct of PRAs. The guide was focused primarily on internal event analysis with some general guidance on seismic and other external events. In 1988 a Severe Accident Policy (Ref 3) initiated by the US regulators required all operating Nuclear Power Plants (NPPs) to conduct PRAs for internal events. Supplement 4 to the Severe Accident Policy (Ref. 4) required that external events, with emphasis on seismic and internal fire, be included. For seismic events, the regulators allowed the utility to choose between a deterministic seismic margins approach and PRA. For the highest seismic activity sites, PRA was mandatory. Throughout the evolution of PRAs in the US, several Western Countries also commissioned PRAs including seismic events. The methodology for seismic PRAs has stayed basically the same, however, several technical enhancements have evolved. Since the restructuring of the governments in the former Easter Bloc countries, there has been significant concern expressed by people in these countries and in neighboring counties regarding the safety of the Russian designed reactors. To date there have been a few PRAs completed or partially completed to address these safety concerns. This paper will present a summary of the overall methodology for seismic PRA, then summarize seismic PRA activities for Water Cooled Water Moderated Energy Reactors (WWERs) at three sites and compare the results to earlier published PRAs of US Plants. 2. OVERVIEW OF METHODOLOGY The key elements of a seismic PRA are: 1. Seismic hazard analysis - estimation of the frequency of various levels of seismic ground motion (acceleration) occurring at the site 2. Seismic fragility evaluation - estimation of the conditional probabilities of structural or equipment failure for given levels of ground acceleration 3. Systems/accident sequence analysis - modeling of the various combinations of structural and equipment failures that could initiate and propagate a seismic core damage accident sequence 4. Evaluation of core damage frequency and public risk - assembly of the results of the seismic hazard, fragility and systems analyses to estimate the frequencies of core damage and plant damage states; assessment of the impact of seismic events on the containment integrity; and integration of PA these results with the core damage analysis to obtain estimates of seismic risk in terms of effects on public health The process is shown schematically in Figure 2-1. Following is a brief description of the four steps utilized in the PRA process. nt sequences Plant sj/eiy systems B Peak Q:ou"C icce>. BPV fupiuies Fault tree System B failure cue to non-seismic \i seismic inpul failure Figure 2-1: Schmatic Overview of a Seismic PSA Seismic Hazard Analysis Seismic hazard is usually expressed in terms of the frequency distribution of the peak value of a ground motion parameter (e.g., peak ground acceleration) during a specified time interval. The different steps of this analysis are as follows: 1. Identification of the sources of earthquakes, such as faults and seismotectonic provinces. 2. Evaluation of the earthquake history of the region to assess the frequencies of occurrence of earthquakes of different magnitudes or epicentral intensities. 3. Development of attenuation relationships to estimate the intensity of earthquake-induced ground motion (e.g., peak ground acceleration) at the site. 4. Integration of the above information to estimate the frequency of exceedance for selected ground motion parameters. The hazard estimate depends on uncertain estimates of attenuation, upperbound magnitudes, and the geometry of the postulated sources. Such uncertainties are included in the hazard analysis by assigning probabilities to alternative hypotheses about these parameters. A probability P:\950O00\100\TECHPAPR.DOC/irv 169 4-20 distribution for the frequency of occurrence is thereby developed. The annual frequencies for exceeding specified values of the ground motion parameter are displayed as a family of curves with different probabilities (Figure 2-2). 10 0.01 Peak Ground Acceleration (g) Figure 2-2: Typical Seismic Hazard Curves for a Nuclear Power Plant Site Seismic Fragility Evaluation The methodology for evaluating seismic fragilities of structures and equipment is summarized in References 2 and 5. Seismic fragility of a structure or equipment item is defined as the conditional probability of its failure at a given value of the seismic input or response parameter (e.g., peak ground acceleration, stress, moment, or spectral acceleration). Seismic fragilties are most commonly referenced to peak ground acceleration and this definition will be used in this paper. The best estimate of seismic capacity is developed and is defined as the peak ground motion acceleration value at which the seismic response of a given component located at a specified point in the structure exceeds the component's resistance capacity, resulting in its failure. The ground acceleration capacity of the component is estimated using information on plant design bases, responses calculated at the design analysis stage, as-built dimensions and material properties. There are many variables in the estimation of this ground acceleration capacity, thus, the distribution on the capacity is also quantified. Component fragility is described by a family of fragility curves; a probability value is assigned to each curve to reflect the uncertainty in the fragility estimation (Figure 2-3). P:\950000\l 00\TnCHPAPR.DOC/irv 170 5-20 0) (6 LJL "I QL "TO c o c o O SSE Peak Ground Acceleration, g Figure 2-3: Typical Family of Fragility Curves for a Component Analysis of Plant Systems and Accident Sequences Frequencies of severe core damage and radioactive release to the environment are calculated by combining plant logic with component fragilities and seismic hazard estimates. Event and fault trees are constructed to identify the accident sequences that may lead to severe core damage and radioactive release. The plant systems and sequence analyses used in seismic PRAs are based on the PRA Procedures Guide (Reference 2) and can generally be summarized as follows: 1. The analyst constructs fault trees reflecting (a) failures of key system components or structures that could initiate an accident sequence and (b) failures of key system components or structures that would be called on to stop the accident sequence. 2. The fragility of each such component (initiators and mitigators) is estimated. P:\950000\100\TECHPAPR.DOC/irv 171 6-20 3. Fault trees are used to develop Boolean expressions for severe core damage that lead to each distinct plant damage state sequences. 4. Considering possible severe core damage sequences and containment mitigation systems, Boolean expressions are developed for each release category. As an example, the Boolean expression for severe core damage in a prior Probabilistic Safety Study is: MS = 4 + 8 + 10 + 14 + 17 + 21 + (12 + 22 + 26) * 9 (2-1) The numbers represent components for which seismic fragilities have been developed. The symbols "+" and "*" indicate "OR" and "AND" operations, respectively. Plant level fragility curves are obtained by aggregating the fragilities of individual components according to Equation 2-1, using either Monte Carlo simulation or numerical integration. The plant level fragility is defined as the conditional probability of severe core damage as a function of the peak ground acceleration at the site. The uncertainty in plant level fragility is displayed by developing a family of fragility curves; the weight (probability) assigned to each curve is derived from the fragility curves of components appearing in the specific plant damage state accident sequence. Evaluation of Core Damage Frequency and Public Risk Plant level fragilities are convolved with the seismic hazard curves to obtain a set of doublets for the plant damage state frequency, fij> } (2-2) where fjj is the seismically-induced plant damage state frequency and py is the discrete probability of this frequency. Pij = qipj (2-3) fij = If, (a)—-1 da (2-4) da Here, Hj represents the jtn hazard curve, fj the itn plant damage fragility curve; q\ is the probability associated with the i*h fragility curve and p; is the probability associated with the jm hazard curve. The above equations state that the convolution between the seismic hazard and plant level fragility is carried out by selecting hazard curve j and fragility curve i; the probability assigned to the plant damage frequency resulting from the convolution is the product of the probabilities p; and P:\950000\100\TECHPAPR.DOCirv 172 7-20 qj assigned to these two curves. The convolution operation given by Equation 2-4 consists of multiplying the occurrence frequency of an earthquake peak ground acceleration between a and a + da (obtained as the derivative of Hj with respect to a) with the conditional probability of the plant damage state, and integrating such products over the entire range of peak ground accelerations 0 to oo. In this manner, a probabilistic distribution on the frequency of a plant damage state can be obtained. Severe core damage occurs if any one of the plant damage states occurs. By probabilistically combining the plant damage states, the plant level fragility curves for severe core damage are obtained. Integration of the family of fragility curves over the family of seismic hazard curves yields the probability distribution function of the occurrence frequency of severe core damage (Figure 2-4). By extending this procedure, probability distribution functions of the occurrence of different release categories are obtained. 1.0 0.8 /- 2 / Range For o 0.6 O ol / Severe Core C CD -o O _/ — Damage Frequency c o 0.4 /] / o CD L / O X r / LU i c 0.2 / o / ! 0 1 1 10" 10" 10 Annual Severe Core Damage Frequency Figure 2-4: Probability Distribution of Seismically-Induced Severe Core Damage Frequency Public risks in terms of early fatalities, long term adverse health effects, and property damage are evaluated by developing a site-consequence model and inputting the release frequencies calculated above. This analysis would produce seismic risk curves showing frequencies of exceedance at different levels of consequences. Fragility Cutoff Methodology It is not practical to calculate fragilities for all components which are included in the risk modeling. Most components and distributive systems are inherently rugged and can be screened out on the basis that their seismic induced failure rate is low in comparison to the items which will ultimately dominate seismic risk. It is desirable to establish a fragility target above which components exceeding this target may be screened out. l':\950000\IOO\TnCHPAPR.DOC/irv 173 8-20 In developing the target, three variables must be considered; seismic hazard, uncertainty in the median fragility and frequency of failure (potential core damage) relative to that for other events. A fourth variable, consequence of failure is important, but for purposes of establishing a fragility cut off it is assumed that all failures have equal consequence. Parametric studies are usually conducted using the hazard and candidate fragility curves as input variables and examining the resulting failure frequency. Seismic Hazard: NUREG-1407, Ref 6 provides guidance for the US IPEEE program. It states that the seismic hazard must be carried out to 1.5g unless sensitivity studies can show that a lower cutoff is justified. Previous fragility cutoff studies with cut offs at 1.0 and 1.5g have shown that for low capacity components, the extension of the hazard does not make a significant difference but at the fragility level that was ultimately determined to be an acceptable cutoff, there was enough difference between the 1.0 and 1.5g cutoff results that the cutoff decision was based on a 1.5g cutoff. Uncertainty in the Median Fragility: The uncertainty range for fragilities varies with the failure mode. For ductile modes of failure, such as for structures or piping, the margin to failure relative to code allowables is larger than for brittle or functional failure modes but the uncertainty is also larger so that dual criteria must be implemented to establish a minimum value of the median capacity and of the HCLPF. HCLPF is an acronym for high-confidence-of-low-probability-of-failure. It is defined mathematically as 95% confidence of less than 5% probability of failure. The fragility curve is usually described by the median, Am, the randomness, BR, and uncertainty, BTJ, where the Bs are logarithmic standard deviations. For an assumed lognormal distribution the HCLPF may be computed from: HCLPF = Am exp (1.65) (BR + By) For ductile failure modes of flexible systems, such as for structures, the ratio of median to HCLPF is typically about three or greater. For brittle failure modes of rigid equipment or functional failure modes such as relay chatter, the ratio of median to HCLPF can be as low as two. Thus, for the same seismic failure rate, the flexible, ductile items must have a higher median but may have a lower HCLPF than for a non-ductile failure mode. Numerous case studies conducted to determine the fragility level for screening reveal that the seismic failure rate is more sensitive to HCLPF than median. Often it is more convenient to estimate or compute a deterministic HCLPF for making decisions on screening. The final cutoff value for fragility may then be targeted to a HCLPF value, wherein the median value is implied, depending upon the failure mode. Establishment of a HCLPF above the cutoff target is the approach usually used for screening of structures and equipment. P-^OCKXA 100\TECHPAPR.DOC/irv 9-20 Failure Rate Relative to Other Event Failure Rates: Core damage from internal events usually governs the plant risk. Internal event core damage frequencies typically are on the order of 1E-5 or greater. Seismic failures that could contribute more than 10% of this should not be screened out so an approximate target for screening is a seismic failure rate of 1E-6 or less. Surrogate fragilities representing the screening level are then used for the top event in any fault tree where all basic events have been screened out. The screening level is dependent upon the severity of the seismic hazard and the core damage frequency of internal events. All significant parameters must be assessed to arrive at a rational and defendable screening level. 3. CASE STUDIES, WESTERN VS. EASTERN EUROPEAN REACTORS Over the past 15 years, there have been over 30 seismic PRA's conducted of Western nuclear power plants, principally of U.S. design. Reference 7 summarized the author's insights from participation in most of these studies. Since the political restructuring of the former Easter Block countries, there has been much involvement of Western contractors in safety assessments and safety upgrades of Russian designed NPPs. Seismic assessments of these plants has become of particular importance in several countries. In conjunction with seismic upgrades being implemented on several plants, some partial and some complete seismic probabilistic risk assessments have been conducted. In this paper, the seismic risk from four WWER reactors has been summarized and is compared to that for a sampling of U.S. reactors. There are three generations of WWER-the 440-230, 440-213, and the 1000. The 440-230's are the oldest design and the 1000's are the latest. The plants studied and summarized are a 440- 230 and a 1000 design at the same site in Bulgaria for which the seismic hazard is moderately severe, a highly modified version of a 440-230 in Finland, and a new 1000 mw plant under construction in the Czech Republic. The seismic hazard for the latter two sites is very low. The WWER 440s studied had no special seismic design provision, but a number of seismic backfits have been made. The WWER 1000's were designed for seismic loading. For the plant in Bulgaria, the design level was less than current seismic hazard studies would suggest, but as will be shown, there is a large seismic margin in most of the 1000 mw structures and equipment. The first generation of WWER's produced 440 megawatts (WWER-440). The earlier model 230's are of greatest concern due to their lack of containment. Later model 213's incorporated a bubble tower to condense steam released from a loss of coolant accident. This second generation WWER-440 has improved safety features and can mitigate a large LOCA. The WWER-1000 has a post-tensioned concrete containment and is very similar in system design to current generation Western PWRs. P:\9S0000\l00\TECHPAPR.DOOirv 175 10-20 The primary focus is on six reactors located at Kozloduy in Bulgaria, four WWER 440-230's and two WWER-1000's, henceforth are subject to identical seismic hazards. The Loviisa 440 plant in Finland and the Temlin 1000 mw plant in the Czech Republic are in low seismicity regions and only the results are provided for comparison. Probabilistic risk assessment methodology (PRA) was used to assess the risk due to internal events and due to earthquakes. Because of the urgency in assessing the relative risk from various aspects of the design and the external hazards, the initial study of the Kozloduy WWER 440's was abbreviated to a "Top Level Risk Assessment" (TLRA), Reference 8, as opposed to a full PRA. The Top Level Risk Assessment utilizes event tree methodology, but does not include the development of detailed fault trees. The PRA for the WWER 1000's was a full PRA conducted in much greater depth. The Top Level Risk Assessment focused on internal event initiated accident sequences. A very brief supplemental study was conducted to include seismic initiating events. At the time of the Top Level Risk Assessment, the seismic hazards for the Kozloduy site had not been finalized; therefore, an existing seismic hazards study of the Belene site (Reference 9) was used. Both sites are on the Danube River and the hazard is dominated by seismic sources near Vrancea in Romania. The recently completed PRA of the WWER 1000's utilized a site-specific hazard. The two seismic hazards curves are compared in Figure 3-1. Although the hazard used for the earlier Top Level Risk Assessment was greater, the results of the two studies show more than an order of magnitude greater core damage frequency (CDF) for the older WWER 440's, whereas, there is only about a factor of two difference in the hazards, i.e., the probability of exceedance of a given peak ground acceleration is about a factor of two greater. Thus, comparison between the results for the two designs are meaningful. Hazard Curves for NPP KOZLODUY and BELENE - - SELENE v o \ KOZLODUY i PEAK ACCELERATION (G) Figure 3-1: Hazard Curves, Free Field Maximum Acceleration l':W5OU()(>\IOOVn:CIII>Al'R.IXX7irv 176 11-20 Description of Reactors The WWER 440's are designed as dual reactor complexes connected to a gallery complex which is connected to the turbine building. Each structure is, however, on a separate foundation. The reactor primary system and six steam generators are contained within a reinforced concrete confinement which is designed for a one bar internal pressure. Upon exceedance of one bar pressure, relief valves open and vent to the atmosphere. The confinement is sized to contain the release from a 100 mm diameter pipe break, but with an orifice to choke flow. Neither the system design nor the containment can mitigate or confine a larger break. The two later WWER 440's have low pressure safety injection systems (LPSI), but the design basis loss of coolant accident remains a 100 mm line break choked by an orifice. The WWER 1000 has a post-tensioned concrete containment which, unlike western containments, sits on top of the auxiliary building structure. Its safety systems are similar to Western PWRs and can mitigate and confine a large loss of coolant accident. The safety systems differ considerably between the two generations of reactors. In general, the Russian reactors have three train safety systems, as compared to a two-train system in Western PWRs. However, in the two earliest WWER 440's at Kozloduy, the service water system and DC power system consist of three trains shared by two units. The service water system is also shared by non-essential heat loads, thus increasing the vulnerability of the service water system. The WWER 440's do not have a closed loop decay heat removal system. Long-term decay heat removal is by means of heat exchangers on the secondary side that cool water circulated through the steam generators. None of the WWER's have procedures for "feed and bleed" for controlled depressurization of the primary system in the event of a loss of main and emergency feedwater to the steam generators. Most Western PWRs have this capability. The WWER 440's at Kozloduy do not have fast closing main steam isolation valves, thus there is a risk of overcooling the embrittled reactor vessel in the event of a main steam line break. The WWER 440's at Kozloduy do not have power-operated isolation valves for the primary coolant letdown and filtration system, thus increasing the system boundary which could initiate a loss of coolant accident. Table 3-1 compares the above features for the three Russian designs at Kozloduy and modern Western designs. Seismic Design Basis The two earlier WWER 440's at Kozloduy did not have a seismic design basis. Soon after operation, the two reactors experienced an earthquake in 1977 which was centered in Vrancea, Romania. There was no seismic instrumentation at the site, but post-earthquake investigations estimated about O.lg peak ground acceleration. One unit was manually scrammed and the other P:\95OOOO\100\TECHPAPR.IX)C/irv 177 12-20 continued to operate. Post-earthquake damage investigation and analysis revealed that the steam generators had moved about 11 cm and that the primary coolant piping was close to yielding. Some minor structural damage was noted and repaired. The structural design of the WWER 440's outside of the reinforced concrete confinement consists of primarily precast concrete columns, beams and shear panels. The connection details are very weak and brittle, thus most of the essential structures are vulnerable to large earthquakes. There is extensive use of unreinforced masonry walls for interior walls. Some of these walls cracked during the 1977 earthquake. Table 3-1 COMPARISON OF DESIGN FEATURES OF KOZLODUY AND WESTERN REACTORS Kozloduy Kozloduy Kozloduy Function 1 & 2 WWER-440 3 & 4 WWER- 5 & 6 WWER- Western PWRs 440 1000 Containment 1 bar confinement 1 bar Yes Yes confinement Large LOCA No No Yes Yes Mitigation Closed Loop No No Yes Yes Primary System DHR Safety System 3 train AC Power, 3 train 3 train 2 train Redundancy 3 train DC and service water, shared by two units Feed and Bleed No No No Yes (most) Fast Closing MSIVs No No Yes Yes Letdown Isolation No No Yes Yes Steam Driven No No No Yes Emergency Feed Pumps Subsequent to the 1977 earthquake, selected upgrades were made to the primary system, including the letdown filtration system to prevent a loss of coolant accident. These upgrades were designed for a O.lg PGA and were incorporated into units 3 and 4 which were under construction at the time. Some upgrades were made to the secondary system steam generators and DHR heat exchanges, but as was discovered in a walkdown of the 4 units (Reference 10), many of the [CKKA100\TKCI lI'APR.DOC/irv 13-20 upgrades were either not completed or were improperly constructed. Upgrades were not carried to other essential equipment or to structures. At the time of the Top Level Risk Assessment there were many unanchored essential equipments. The two WWER 1000's were designed for O.lg PGA but, in many instances, the design is standards for up to 0.3g PGA, thus there is a large margin in the main structure and in many components. The essential structures are cast-in-place reinforced concrete. A previous assessment of the Belene reactor which was under construction in Bulgaria revealed that they could, for the most part, meet design requirements for greater than 0.2g PGA (Reference 9). Most essential equipment is properly anchored and supported, although a few exceptions were noted during a walkdown in review of Reference 11. Much of the electrical and control cabinets are top braced so that they are not as vulnerable to seismic input motions, as were many earlier U.S. plant installations. Unreinforced masonry walls are not used in areas near essential equipment. Although the current hazard study for the Kozloduy site suggest that the design basis should have been for 0.2g PGA rather than O.lg, in general there is good detail and margin in this 1000 mw design. Risk Models WWER 440 The Top Level Risk Assessment for the WWER 440s focused primarily on internal events risk to identify design vulnerabilities and develop concepts for cost-effective back fitting, including construction of a "bunker system" for safe shutdown. The initial seismic activity was focused on determining the magnitude of seismic backfit required to structures and equipment with and without the bunker system. However, a previous study conducted for the International Atomic Energy Agency (Reference 10) resulted in the development of some seismic fragility curves for selected weak link components in the plant. These fragilities, along with some estimated structural fragilities, were utilized in a simple Boolean expression to approximately quantify the seismic risk in order to have a comparison with the risk from internal events. The simple Boolean utilized was representative of units 3 and 4, but approximately represents all four of the WWER 440 units. The weak links of essential systems were assumed to be the top event in the fault trees representing AC power supply, AC power distribution, emergency feed water and safety injection. The Boolean took credit for a "feed and bleed" capability in the event of loss of emergency feed water, but no procedure for this was in place at the time of the study nor was it certain that the power operated relief valves had the capability of performing this operation. Since failure of major structures lead directly to core damage, the weakest structures were also represented. I>:\950000\IOOVIIX'III'AI'K IXXVirv 179 14-20 The DC power system, control systems, and systems interactions, primarily block walls, were not modeled. DC power appeared to be reasonably rugged except for numerous unreinforced masonry walls. The control system was most vulnerable from lack of anchorage of relay cabinets. Since the service water pump structures and diesel building were not any better than unreinforced masonry walls, the walls were not included in the simple model. Also, it was recognized that a major program for anchorage of equipment would be necessary, hence control cabinet anchorage was not included in this first approximate quantification. The convolvement of the hazard and fragility curves in the Boolean expression resulted in a predicted core damage frequency (CDF) of 1.6 x 10" per year, more than an order of magnitude greater than what would be acceptable in the Western world. This was, however, comparable to the calculated CDF for internal events and the CDF estimated for fire. Using the later site-specific hazard shown in Figure 3-1, the CDF was about 7.5 x 10 , still very high relative to Western reactors. If a complete systems model were used, increasing the number of basic events, the CDF would increase and likely be in the neighborhood of 1 x 10" . Some subsequent studies utilizing detailed fault trees representing the AC power system, service water system, emergency feed water system, low pressure injection system and high pressure injection system were conducted to determine the sensitivity of the reactor to the various sources of failures and to aid in setting priorities for upgrades. The major structures were also included as in the initial quantifications. The service water and diesel building failures were incorporated into the service water and AC power systems fault trees and the turbine building was incorporated into the emergency feed water system fault tree. The fault trees contained the logic for quantifying the top event taking into consideration multiple failure sources for the system such as unreinforced masonry walls and electrical cable systems, as well as essential component failures. Again, the DC system and instrumentation and control systems were not modeled for the reasons previously given. Six accident sequences were modeled which included: • Loss of core cooling Small LOCA • Very small LOCA (normal system leakage) • Large LOCA • Loss of reactor protection systems • Failure of auxiliary building Cases were run assuming no "feed and bleed" capability and "feed and bleed" capability. The Belene seismic hazard, Figure 3-1, was used as was the case for the earlier TLRA model. PiiaSOPOOUOOVTECHPAPR.DOC/irv 15-20 The overall results were not much different than the results using the simple Boolean representing top events. The predicted core damage frequency for the case of no feed and bleed was 1.46 x 10"3 per year, whereas the frequency considering feed and bleed was 1.39 x 10"6 per year. The main difference between the earlier results without detailed fault tree modeling and the more detailed fault tree model was that the earlier model assumed loss of offsite power as a given, whereas in the fault tree model analysis, loss of offsite power was quantified. Fussel-Vessely importance functions indicate the major contributors shown in Table 3-2. Using the order of importance in Table 3-2, one can quickly assess cost-benefit of backfits. Numerous cases were run assuming upgrades to the various systems. Cases were also run to represent system design changes that would resolve the seismic issue without seismic upgrading. In many cases, system changes were more effective than seismic backfits. It was, however, evident that because of the numerous vulnerabilities, major backfits were required to result in a significant reduction in seismic risk. Significant equipment anchorage and block wall reinforcement backfits have been performed in the four units. Much of this backfitting, though, has only brought the plant to a level to validate the assumptions in the simplified SPRA. Without backfitting the major structures or building a dedicated safe shutdown system to replace systems in vulnerable buildings, the risk reduction to date relative to the assumed base case in the simplified SPRA is less than a factor of two. WWER 1000 A seismic PRA for the 1000 mw plant is being conducted by Risk Engineering Ltd. in Bulgaria, Reference 11, with assistance in the development of seismic fragilities by the Bulgarian Academy of Science. A site-specific hazard curve is being used for this study (Figure 3-1). The seismic and internal event PRAs are being conducted simultaneously. Guidance in performing the seismic PRA is obtained from the PRA Procedures Guide, Reference 2, and IAEA Technical Guide, Reference 5. The initiating events considered are: • Reactor Vessel Rupture • Large LOCA • Medium LOCA • Small LOCA • Class 1 Transient, secondary cooling available • Class 2 Transient, secondary cooling not available P:\950000\100\TECHPAI'R.DOC/irv 181 16-20 Table 3-2 PERCENT CONTRIBUTION TO CORE DAMAGE AS DEFINED BY FUSSEL-VESSELY IMPORTANCE FUNCTIONS Failure Contribution Without Contributions with Feed and Bleed % Feed and Bleed % Loss of offsite power 14.2 13.2 400V vital bus 13.2 14.6 6.0 kv/400v transformer 8.77 9.76 Manual Operation of MOV 6.12 6.47 Service water gravity feed tank 6.08 7.40 Turbine building 5.83 3.87 Service water building 4.47 5.47 Deaerator 3.96 3.42 Chemical treatment plant 3.37 <1 Brick wall, 6.0 kv bus 3.12 3.31 Brick wall, cables in SWPB 2.98 3.61 Pressurizer (large LOCA) 2.39 2.56 Cable path 2.39 2.56 6 kv vital bus 2.39 2.56 Auxiiiary building 2.39 2.56 Danube water pump building 1.49 1.84 Regenerative HX 1.44 1.53 Ion Filters 1.44 1.53 Diesel generator building 1.18 1.25 Pipeline, Confinement EWST 1.02 2.21 For each initiating event, the plant is assumed to be at full power and offsite power is assumed to be unavailable. The internal event fault trees were modified to reflect the seismic induced failure modes. Fragilities of the main structure, primary coolant system, and major equipment in the ECCS system were developed from 3D finite element analysis . The main structure response analysis included the effects of soil-structure interaction. Probabilistic response analysis to develop probabilistic floor spectra and structural loads was conducted using the Latin-Hypercube OOOMOOYHXIII'AI'R IXKVirv 17-20 experimental design methodology developed in the Seismic Safety Margins Research Program (SSMRP), Reference 12. Fragilities for other equipment were developed for 38 generic categories of equipment. Generic fragilities were obtained from Reference 13, provided that the walkdowns did not identify any obvious seismic vulnerabilities. For cases where components appeared vulnerable, either specific fragilities were developed or fixes were recommended to correct the potential problem. Preliminary computation of core damage frequency results in a mean value of 1 x 10"5 per year as compared to an internal event value of 3.7 x 10"4. The seismic risk is, however, underestimated due to several factors. First, because of software limitations, the entire earthquake hazard range was not utilized in the computation. The computed value represents the core damage frequency for a single 10"5 nonexceedance probability earthquake which is about 0.26g pga. Second, the risk model, derived from the internal events risk model, considers component failures in redundant trains as being uncorrelated where, in fact, for seismic failures, the redundant trains are mostly correlated. Third, in a few instances, either some backfitting is necessary in order for the generic fragilities to be applicable or in a few instances, the generic fragilities derived for standard equipment in the U.S. plants are not applicable to some unique construction features of the 1000 mw WWER equipment. The author's opinion is that the final results will show a severalfold increase in core damage frequency, but will still be an order of magnitude less than the current status of the adjacent WWER 440s. It is also anticipated that the seismic core damage frequency will be considerably less than the currently computed internal event frequency of 3.7 x 10"4. Reference 14 reports the results of the seismic PRA for Loviisa in Finland. Although the plant has virtually no seismic design and has many seismic vulnerabilities relative to minimum seismic design standards set by the IAEA, the hazard is very low and the computed mean CDF due to seismic events is on the order of 10'7/year. Thus, there is not an incentive to do seismic upgrades just because there are perceived weaknesses in equipment anchorage, etc. A PSA for Temlin in the Czech Republic was just completed. The plant was originally designed for 0.05g, but is being reassessed and upgraded to 0. lg prior to licensing. There are existing components that do not currently meet the O.lg new design basis, however, the seismic hazard is very low and the computed seismic induced CDF for the existing conditions is negligible, as was the case for Loviisa. 4. CONCLUSIONS The seismic risk of two generations of Russian designed reactors located at a moderately seismically active site and at two low hazard sites has been compared and it is shown that for the more seismically active region the seismic induced CDF from the earlier 440 mw units is about an P:\95O(XK)\! order of magnitude greater without conducting significant backfits to the 440 mw units. There have been many backfits conducted to date to increase the seismic capacity of the primary system and supporting equipment, but none to date to essential structures which would now govern the risk. Table 4-1 from Reference 15 compares internal and external event risk ol several U.S. reactors. The highest contribution from seismic events to mean core damage frequency in the U.S. plants in Table 4-1 is about 25% of the total CDF. In this case the mean annual seismic induced CDF is about 6.25 x 10" /year, which is comparable to the estimated CDF for the 1000 mw plant. The internal event CDF for the Kozloduy 1000 mw plant was computed to be about 3.7 x 10"4, thus seismic induced CDF is estimated to be on the order of 20% of internal event CDF. For the Kozloduy 440 mw plant, the initial internal event CDF from the TLRA was on the order of 3 x 10" , or about twice the initial seismic risk. With current upgrades in the 440's, both the internal event and seismic CDF are reduced. No quantification of the current status is available, but the estimated risk from both internal and external events is anticipated to be up to an order of magnitude greater than for the adjacent 1000 mw plants. It can be concluded that simplified studies such as the seismic and internal event TLRA can be a useful cost-benefit tool to set priorities for upgrades. It was in fact extremely useful to reach the conclusion that beyond the initial activities to overcome some system design deficiencies and to strengthen equipment and surrounding masonry walls, it is evident that major reconstruction is necessary to bring the safety of the 440 mw units close to that of current Western designs. The more detailed PRA of the 1000 mw units indicates that, with minor upgrades, the seismic safety of Western plants can be approached. I'AWOOOOMOOVI'IICHI'Al'R IXK.7irv 19-20 Table 4-1 CONTRIBUTION OF INITIATING EVENTS TO MEAN ANNUAL CORE MELT FREQUENCY FOR PUBLISHED PRAs WITH COMPLETE SEISMIC ANALYSIS Contribution % Plant Date Seismic Internal Fire Win Externa Mean annual core d 1 melt frequency 5 Zion 1981 8 85 7 - 6.7x10 " IP2 1983 6 58 10 26 1.4 x 10" IP3 1983 2 88 9 1 1.4 x 10" Seabrook 1983 13 75 11 - 2.3 x 10" Limerick 1983 13 34 53 - 4.4x10 "5 Millstone 3 1984 15 77 8 - 5.9 x 10'5 Oconee 3 1984 25 56 4 5 10 2.5 xlO" Notes: Contributions to core melt is not necessarily indicative of public health risk contribution. Seismic events that initiate core melt accident sequences are generally more likely to also cause damage to containment than other initiating event. Comparison of median (rather than mean) seismic risk to median core melt frequency would indicate in most (but not all) cases lower seismic contribution. P:\950000\l 00\TECHPAPR.IX)C/irv 185 20-20 5. REFERENCES 1. NUREG/CR-4431, Summary Report on the Seismic Safety Margins Research Program, January 1986. 2. "PRA Procedures Guide," NUREG/CR-2300, January 1983. 3. USNRC Generic Letter 88-20, "Individual Plant Examination for Severe Accident Vulnerabilities," 10 CFR 50.54(f), November 23, 1988. 4. USNRC Generic Letter 88-20, Supplement 4, "Individual Plant Examination of External Events (IPEEE) for Severe Accident Vulnerabilities," June 1991. 5. IAEA, "Probabilistic Safety Assessment for Seismic Events," Technical Document TECHDOC 724, October 1993. 6. NUREG 1407, "Procedural and Submittal Guidance for the Individual Plant Examination of External Events (IPEEE) for Severe Accident Vulnerabilities," June 1991. 7. Campbell, R.D., "Insights from Probabilistic Risk Assessments and Seismic Margins Assessments Regarding the Variance of Design Margin for Seismic Event," American Society of Mechanical Engineers, Technology for the 90's, 1993. 8. EQE, Ltd., "Top-Level Risk Study of Kozloduy NPP Units 1 through 4," Report 58-01- R001, April 1992. 9. EQE International, Westinghouse Energy Systems and Geomatrix Consultants, "Seismic Review of the Belene Construction Project (Units 1 and 2)," prepared for Techno-Import- Export, March 1990. 10. Westinghouse Energy Systems International, EQE International, "Seismic Ruggedness Evaluations of Kozloduy VVER Units 1 to 4," prepared for IAEA, November 1991. 11. Risk Engineering, Ltd, "Seismic Probabilistic Safety Analysis for NPP Kozloduy Units 5 and 6," Report RE/0-10-3, February 1994. 12. NUREG/CR-2015, "Seismic Methodology Analysis Chain With Statistics," Vol. 9, 1981. 13. "Handbook of Nuclear Power Plant Seismic Fragilities," NUREG/CR-3558, June 1985. 14. Varpasuo, P., J. Puttonen and M. K. Ravindra, "Seismic Probabilistic Safety Analysis of Loviisa NPP, Unit 1," Proceedings of SMIRT 12, Paper MK05/3, August 1993. 15. Sues, R., P. Amico, and R. Campbell, "Significance of Earthquake Risk in Nuclear Power Plant Probabilistic Risk Assessment," Nuclear Engineering and Design 123, 1990. P:\95iQOOO\P:\95iOOi l OOVTECHPAPR.DOC/irv 186 SESSION IV "ANALYTICAL METHODS FOR SEISMIC CAPACITY RE- EVALUATION" NEXT PAGE(S) I left BLANK I • 187 1 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES SEISMIC DESIGN OF NUCLEAR POWER PLANTS - WHERE ARE WE NOW? J. M. Roesset The University of Texas at Austin, Austin, Texas ABSTRACT: The lack of any significant activity in the design and construction of new nuclear power plants over the last ten years has resulted in a corresponding lull in the basic academic research carried out in this field. While some work is still going on related to the evaluation of existing plants or to litigation over some of them (including some that never became operational) most of it is of a very applied nature and little basic research is being conducted at present. Yet research on earthquake engineering in general, as applied to buildings, bridges, lifelines, dams and other constructed facilities has continued. This paper attempts to look at some of the areas where there were major uncertainties in the seismic design of nuclear power plants (selection of the design earthquake and its characteristics, evaluation of soil effects and soil structure interactions, dynamic analysis and design of the structures), the progress that has been made in these areas, and the remaining issues in need of further research. 1. INTRODUCTION The importance of nuclear power plants and the consequences of a nuclear accident required that they be designed to safely withstand the most severe environmental conditions that could reasonably be expected to affect mem during their lifetime. This led during the 1960s and 70s to an extraordinary amount of basic and applied research on their seismic analysis and design, research which benefited not only the nuclear industry but also the area of earthquake engineering in general. Many significant advances in this field are a direct result of these research efforts and the nuclear industry can be proud of their contribution. Yet the desire to apply new knowledge as fast as it was being generated and the pressure to use the latest state of the art procedures without an adequate amount of time for reflection and validation had some undesirable consequences: further analyses and reanalyses were required in some cases without any real justification based on conclusions or results from papers which addressed very particular cases of limited scope; methodologies that were incomplete and at times incorrect (incorrect for some practical situations) were accepted or even endorsed at one time and then disavowed entirely; and criticism of established and commonly used procedures pointing out their limitations was occasionally considered treacherous and detrimental to the good of the industry. All this resulted in a significant amount of controversy. This controversy and the adversary relationships which existed at one time between the owner / designer / manufacturer team, the regulatory agencies and their consultants and the public at large, who justifiably demanded answers to a number of important questions and access to information, hurt the nuclear industry and resulted eventually in a near complete halt in new designs and construction as well as research. Unfortunately, the conflict has not stopped. It has continued with the adversaries becoming the reactor manufacturers, the architect engineering teams, and the owners who try to recover the extra costs associated with the lack of complete knowledge at the time the design of their plants was initiated and was being carried out, the need for research and development of new methodologies which existed at the time, and the required reanalyses and design modifications. The reduction in activity within an area with the corresponding decrease in the pressure to find instant solutions to new problems for which there is very limited experience provides, or should provide, the opportunity to compare and validate methodologies, to better define their ranges of applicability, to identify areas where additional research is necessary, and to reach a consensus, 189 perhaps, on acceptable solutions and procedures. It is also the time for industry to regroup and prepare itself to be ready when the need for new designs arises again, trying to avoid future controversies. Unfortunately, in the seismic design of nuclear power plants this kind of activity has been rather limited, although significant work has been done in Japan by a number of companies (Kajima, Mitsubishi, Ohbayashi, Toshiba) and in the United States by the Electric Power Research Institute (EPRI) and government organizations such as the Nuclear Regulatory Commission or the Department of Energy, among others. Most of this work is, however, of an applied rather than a basic nature and some controversial issues remain unresolved. Research on earthquake engineering has continued, on the other hand, with applications to building structures, bridges, hospitals, dams, lifelines, and other facilities, using in many cases the methodologies developed for nuclear power plants and extending them. The purpose of this paper is to review some of the major sources of uncertainties which existed in the seismic design of nuclear power plants, discuss briefly some of the work that has been done in these areas, and point out remaining topics in need of further research. The three main areas to be addressed are the definition of the design earthquake (or earthquakes), the effects of the soil (soil amplification and soil structure interaction), and the dynamic structural analysis. 2. DEFINITION OF DESIGN EARTHQUAKE The determination of the design earthquake (or earthquakes) for a nuclear power plant was normally based on a series of extensive seismological and geological studies. Historical records of past earthquakes were carefully reviewed and a seismic history of the region was compiled. Tectonic zones were defined and their seismicity evaluated. Potential active faults were identified. Rates of occurrence of earthquakes of different magnitudes were assigned to all known faults or areas where epicenters of past earthquakes had been located. Finally, attenuation laws were developed which could provide peak values of the ground motions parameters (acceleration, velocity or displacement) for a given earthquake as a function of magnitude and distance. Yet the results of all these studies were expressed in terms of a single parameter, the effective peak ground acceleration. The design earthquake was then specified in terms of this acceleration as a scaling factor and a standard set of response spectra for various values of damping. The same shapes of the spectra were applicable to the different levels of motion (operating basis or safe shutdown earthquakes) and in many cases to the horizontal and vertical components of motion, changing only the scale factors (in some cases a small modification was introduced in the shape of the horizontal and vertical spectra using different scale factors over two ranges of frequencies). This implied that the frequency content of the earthquake was considered independent of its magnitude and source mechanism, the distance from the site to the causative fault, the general topographical and geologic conditions of the region, and in some cases even the local soil properties. These spectra were all that was needed when a modal spectral analysis was to be used to calculate the seismic response. It was common, however, to generate one or more sets of artificial earthquakes (synthetic time histories) for direct solutions of the equations of motion in the time or frequency domains. The response spectra of these artificial earthquakes was supposed to match within certain specified tolerances the smooth target design spectra. An iterative procedure was used typically to achieve this match but a perfect match was very difficult to obtain, particularly for several values of damping. As a result the synthetic motions tended to be conservative, and sometimes substantially so, over some frequency ranges. They also had a higher amount of energy for a given value of the peak acceleration than real earthquakes which do not have smooth spectra. This process could be improved considerably by generating design motions on the basis of physical considerations, accounting for the effects of magnitude, focal mechanism, distance, topography, and soil conditions, not only on the value of the peak or effective ground acceleration but also on the frequency content and duration of the earthquake, simulating a fracture propagating along a rupture zone and following the paths of the waves which are generated from the fault to the site under consideration. A significant amount of research has been conducted over the last 15 or 20 years on this subject and is still continuing today. When combined with probabilistic formulations to account in a rational way for all the existing uncertainties and with statistical data from actual earthquake records it has the potential to be one of the major improvements in seismic design. Ideally one would like to obtain a description of the design earthquake in terms of the types of waves that would be arriving at the site, their amplitudes and their angles of incidence at bedrock as a function of frequency and the duration of shaking. A detailed modeling of a fault with the complete 190 length of rupture and the geological features of the terrain over dimensions of many kilometers may be, however, far too expensive even for present day supercomputers, particularly for practical applications. Studies with less detailed models but accounting for information from actual records can provide, however, valuable insight on the general characteristics of the expected motions at the site (duration and frequency content) for different magnitudes and distances. It is interesting to notice that the use of simple physical models based on wave propagation to simulate earthquake motions had already been proposed by Housner in the 50's, but this approach was discarded and researchers preferred to concentrate on the generation of artificial earthquakes as stochastic processes without any physical basis. In 1969, Rascon and Cornell attempted to revive this line of research by combining it with a probabilistic formulation, but their effort was again an isolated one. It was only in the late 70's that a concentrated attack was launched along these lines and today this is the area where most progress has been achieved. Many are the researchers who have contributed to this progress and no attempt will be made here to trace their individual efforts. One should mention, however, the studies conducted by EPRI (1986) and by the Lawrence Livermore Laboratory (Bernreuter et al, 1985, 1987, 1989) for the Eastern United States and more recently the work conducted at the National Center for Earthquake Engineering Research of the State University of New York in Buffalo (in combination with the Lamont-Doherty ObservaioryX Jacob 1994). Clearly more work remains to be done and there are still uncertainties that cannot be fully accounted for. The discrepancies between the results that would be obtained using the earlier procedures suggested by the Lawrence Livermore National Laboratory study and the EPRI methodology illustrate the fact that the interpretation of the same data by different researchers may lead to different conclusions and that further studies may be needed to reconcile the differences. It appears that the later revised methodology of LLNL is in much better agreement with the EPRI's. In spite of the remaining issues, it is possible today to prescribe design motions consistent with a desired return period (or probability of occurrence) in terms no longer of a single parameter but accounting for duration and frequency content as well as peak ground acceleration. This is a major step forward. A second improvement would be the consideration of more than one design motion for the same return period, corresponding perhaps to different distances and magnitudes (one could have for instance a small, or relatively smaller, but very close earthquake and a stronger but more distant one). When the motions are specified in terms of design response spectra or, even better, in terms of a power spectral density that could be used directly for probabilistic dynamic analyses, this is all that is needed. When using actual earthquake records with characteristics similar to those of the design motions or synthetic accelerograms matching the design spectra it would be necessary to consider several samples for each design earthquake. This has been opposed by industry arguing that it would be far too expensive. This is not necessarily so with present day supercomputers or even workstations, particularly for linear analyses, which are the ones most commonly performed. It is possible then to break the analysis into a series of logical steps, storing the results of each one in permanent files. The next major point of concern in the definition of the seismic input is the location where the design or control motion is specified. This was a subject of considerable debate for many years and although some promising trends are observed it is not yet clear whether the matter has been fully resolved. As has been often stated (Roesset and Kim, 1987, for instance), there are five possible choices: • the free surface of the soil deposit at the site. • a hypothetical outcropping of rock. • bedrock when there is rock at some finite depth at the site. • the elevation of the foundation in the free field (soil deposit without any structure or excavation. • directly at the foundation. If the characteristics of the design motions correspond to some average firm ground conditions and the soil at the site can be classified as such, specification of the earthquake at the free surface of the soil deposit would be the logical choice. This would also be the case if the seismic hazard analyses had already incorporated the effect of the local soil conditions (i.e., if the soil amplification studies for the site had been conducted explicitly by the seismologists, instead of waiting for the geotechnical engineers to do it, or if they had been incorporated implicitly). A more general solution 191 would be to specify the motion at a hypothetical outcropping or rock, accounting for earthquake mechanism and distance but not for local soil conditions. These would be considered in soil amplification studies which could be one-, two- or three-dimensional depending on the topography and stratigraphy at the site and which could be different for various locations at the site. If the site has a well-defined transition between soil and much stiffer rock at a finite depth the specification of the motion at bedrock would be very similar to the specification at rock outcropping. Otherwise this alternative is not very meaningful since the characteristics of the motion at one level within the soil deposit will be a function of the properties of the soil above and below that level. Finding thus a consistent motion at some depth would require amplification studies similar to those performed to determine motions at the free surface from a specified input at a hypothetical rock outcrop. Specification of the design motion at the foundation level in the free field, as required at one time, is the least advisable option and leads to a number of serious inconsistencies if there are various structures with their foundations at different levels. Specifying the control motion directly at the foundation is equivalent to ignoring kinematic interaction effects. It is commonly assumed that this is a conservative assumption, particularly for deeply embedded foundations, but it ignores rotational components of motion and, more importantly, the physical reality. And introducing uncontrolled conservatism, which is hard to quantify, is an undesirable approach particularly when attempting to perform more rigorous probabilistic risk analyses. One of the main problems in deciding the location of the control motion was an apparent confusion for embedded foundations between its specification at the level of the foundation in the free field and the direct specification at the foundation. This confusion was aggravated by a number of papers and studies that attempted to justify the reduction in the levels of acceleration with foundation depth by looking at the motions recorded at different depths in boreholes. A source of concern was the fact that the motion that would be obtained at a given depth in the free field would exhibit a very sharp valley in its spectrum at the natural frequency of the overlying soil mass. This concern would be eliminated by the use of more than one earthquake record as suggested above and primarily by the use of more than one set of soil properties (as has always been done). More importantly, the compatible motions at the foundation level accounting for the excavation will exhibit a clear reduction in the high frequency components of motion but much less sensitivity to specific frequencies than the one-dimensional deconvolution solution. A sixth alternative which has not been included in the above list is the specification of the motion not at the foundation but directly at the base of the structure ignoring therefore not only kinematic but also inertial interaction. This is of course what was traditionally done for regular buildings. 3. EFFECT OF LOCAL SOIL CONDITIONS The effect of the local soil conditions on the characteristics of the earthquake motions at the site (soil amplification studies or determination of site specific spectra) is normally carried out assuming a horizontally stratified soil deposit and vertically propagating seismic waves. This implies that soil properties can vary arbitrarily with depth but remain constant in the horizontal direction and that all points on a horizontal plane experience the same motion at any instant of time. The solution of this problem, whether using a continuous formulation based on one-dimensional wave propagation theory (Roesset and Whitman 1969), or a discrete model (finite differences, finite elements or a physical discretization of lumped masses and springs) as suggested by Seed and Idriss (1969), is very simple and well-known. It can be efficiently performed in a personal computer. When the motion is specified at a hypothetical outcropping of rock (the ideal location) or at bedrock the purpose of the analysis is to compute the consistent motions at the free surface of the soil deposit or at any other elevation in the free field, as well as compatible motions and stresses at other points within the soil profile (the points of contact between the foundation and the soil or points along a lateral soil boundary if the soil structure interaction analyses are to be conducted with a finite element model). If the motion is specified directly at the free surface of the soil deposit but the soil structure interaction studies are going to be conducted using a finite element discretization amplification studies must again be conducted to obtain now compatible motions at the bottom boundary of the domain as well as at the other points mentioned above. In the first case the analysis process is referred to as a convolution, while in the second it is known as a deconvolution. Deconvolution analyses are also performed to determine equivalent soil properties with depth (based on the levels of strain) for soil structure interaction analyses using continuous solutions or even simplified expressions for the foundation stiffnesses. 192 These analyses are normally performed in the frequency domain assuming therefore linear elastic material behavior. It has long been recognized, however, that soil is a highly nonlinear material and that the design seismic motions are likely to induce large strains. Nonlinear soil behavior is normally accounted for in convolution or deconvolution analyses using an iterative linear procedure where the values of the shear moduli and the soil material damping are adjusted at the end of each cycle of analysis based on the strains computed in the previous cycle and curves of modulus and damping versus shear strain characteristic of the material (Schnabel et al 1972). In this way amplification studies are carried out not only to determine compatible motions and stresses at various points within the soil but also to obtain equivalent soil properties to be used in the soil structure interaction studies. This procedure is well-established and unfortunately generally accepted. Yet the accuracy of the results is open to question, particularly when dealing with soft and deep soil deposits. When the motions are followed from the bottom (or rock outcropping) to the free surface the procedure filters out excessively the high frequency components. In fact if this procedure were correct, it would imply that earthquake motions recorded on top of very deep or soft soil deposits should have no energy about 8 or 10 Hz. This simply is not the case. In deconvolution analyses high frequency components increase with depth and the solution eventually becomes unstable. As a result motions specified at the free surface are often artificially modified eliminating all components above 8 to 10 Hz before proceeding with the analysis (the cutoff frequency decreases as the depth of the stratum increases). The main reason for these errors is the assumption of a linear hysteretic damping which is independent of frequency although the amplitudes of the different frequency components are quite different (Roesset, Huerta and Stokoe 1995). A large number of studies have been conducted through the years to assess the validity of the iterative linearization. For most cases the linearized solution overestimates the peak accelerations at the free surface by 10 or 20%, while underestimating displacements and strains, sometimes by as much as 50% (Constantopoulos 1973). Unfortunately, the results of the comparative studies are often contradictory (D'Appolonia 1979, Dames and Moore 1978). Nonlinear analyses in the time domain, which would avoid these problems, can be performed very economically and also in a personal computer, for the one- dimensional convolution problem. For the deconvolution problem the process is at times ill- conditioned because as the solution proceeds down the profile from mass to mass one has to compute first the relative displacements between masses associated with the inertia forces, then obtain the second derivative of these displacements to compute a new acceleration and the increments in the inertia forces. The need to conduct repeatedly the differentiation of the displacements to obtain accelerations is the main source of difficulties. A procedure to perform true nonlinear deconvolution analyses using the theory of characteristics has been recently suggested by Yamada et al (1995). If this method is robust even for deep deposits it would represent a significant improvement and it should be adopted to replace the iterative linearization scheme. The main limitation of true nonlinear analyses is that they cannot provide equivalent soil properties, since the soil properties are changing continuously in time, although one could obtain weighted averages. This implies that all ensuing analyses would have to be conducted with nonlinear models in the time domain. There are a number of areas in which additional research is necessary to resolve outstanding questions in the computation of soil specific motions. The main source of uncertainties is related, however, to the soil properties. To perform soil amplification studies, as well as the soil structure interaction analyses discussed later, it is necessary to know the soil properties in situ, under the existing state of stresses, as well as their variation with levels of strain (variations of modulus and damping with strain for one-dimensional studies and more complete nonlinear constitutive models for more general two- or three-dimensional states of stresses). Traditionally, soil properties were determined through laboratory tests on so-called "undisturbed" samples. Yet, when measurements were carried out in the field the values of the elastic moduli obtained for very low levels of strain in situ and in the laboratory could differ by factors of 2 to 4. It is interesting to notice in this respect that when independent studies were carried out by a number of researchers to reproduce the tests conducted by the EPRI at Lotung, those who used only the data provided from laboratory tests obtained relatively poor agreement with the actual data; those who used the in situ data for the low strain values of the elastic moduli, combined with the variation of modulus and damping with level of strain obtained from laboratory tests, or simply with published standard curves for the type of soils encountered at the site, got invariably a very good match. This indicates that an accurate knowledge of the soil properties in situ, even if they correspond only to very low levels of strain, is or can be more important than a detailed definition of the nonlinear variation of these properties with level of strains. This is again one of the areas where significant progress has been achieved in the 193 last 15 years. In addition to the downhole and crosshole method, which have been in use for some time and can be very reliable (particularly the crosshole test), but are expensive due to the need to have boreholes, the Spectral Analysis of Surface Waves (SASW) Method, which has evolved from the Rayleigh wave technique, provides an accurate, fast and relatively economical wcy of determining the soil properties in situ and their variation with depth over an extended area (Nazarian 1984, Sheu 1987). To account for the uncertainties in soil properties, it was common in the seismic design of nuclear power plants to perform analyses with the best estimate of the soil properties and widi these values multiplied and divided by a factor of 2. These factors were applied simultaneously to all the layers. When the properties of the different strata are relatively homogeneous in the horizontal directions and in situ measurements are available over an extended area, more realistic, and probably smaller, variations can be justified from the data. For soils where the variations in properties in both horizontal and vertical directions are large (such as alluvial deposits), the application of a factor to all the layers simultaneously may not provide an adequate range of variation in the final results, as pointed out by Lysmer (1994). It would be more appropriate in these cases to conduct a number of Monte Carlo-type simulations assuming that the properties of the different layers are independent random variables and statistically interpreting the results. A second source of uncertainty is the angle of incidence of the incoming seismic waves in the underlying rock, as a function of frequency, and the relative amplitudes of the different types of waves (SH, SV, P) propagating through the soil deposit. The formulation to study the amplification of any type of waves by a horizontally layered soil deposit has been available for a long time and the solution of this problem is not more complicated than that of vertically propagating waves (Jones 1970). The main difficulty in considering other types of waves lies in the selection of the appropriate type and this is an area where seismologists can again provide valuable information. In many cases soil profiles are not horizontally stratified: soil layers can be dipping at different angles or have arbitrary geometries with soil properties changing in both the horizontal and vertical directions. In other cases some of the structures may be built at different levels on embankments or small hills with two- or three-dimensional geometries. Amplification studies considering simple two-dimensional geometries (a sloping bottom layer, elliptical sedimentary valleys, etc.) have been carried out for some time. A weighted residual formulation with a collocation minimization criterion was used first in what has been known as the Aki-Larner method (Aki and Lamer 1970). More recently solutions are based on the use of the direct or indirect boundary integral equation (or boundary element) method but a number of other alternatives have been explored (Aki 1988, Bard and Bouchon 1980, 1985, Dravinski 1982, 1983, Papageorgiou and Kim 1992, Sanchez Sesma 1983). The studies conducted on simple geometries have provided considerable insight into the nature and importance of 2D amplification effects for shallow rectangular (or trapezoidal) valleys as well as deep triangular canyons, but most of these studies have assumed homogeneous and linear soil properties for the valley or canyon. 2D or 3D amplification analyses are clearly more expensive and time-consuming than the simple one-dimensional solutions and their use in actual practice (rather than for research purposes) is going to be limited. Even so, when the geometry at the site is clearly two- or three-dimensional, some studies of this kind should be conducted to assess the potential importance of geometric effects. The approximate iterative linear analyses described earlier become even more questionable when dealing with two- or three-dimensional states of stresses. To assess their applicability and to answer some of the lingering questions, it would be necessary to conduct true nonlinear analyses with appropriate nonlinear constitutive models for the soil. A considerable amount of work has been done on the development of plasticity-type models for soils but this is one of the areas where relatively little improvement has been achieved in practical studies and much more remains to be done. Some models, like the multiple yield surfaces model of Prevost (1977), the cap model of Di Maggio and Sandier (1971) (based on the original cam-clay model developed by Roscoe in Cambridge), or the more recent forms of the endochronic models (Systems, Science and Software 1980), would seem to have an excellent potential, particularly when combined with a two-phase formulation to account for the buildup and dissipation of pore water pressures. Even so, it appears that the possibility of having a single model which can correctly reproduce all the features of soil behavior under arbitrary states of stress with a manageable number of parameters is remote. Normally the laboratory tests needed to determine the model parameters are selected so as to reproduce the types of loadings which best simulate the field conditions for each specific problem. Nonlinear two-dimensional and three- dimensional analyses are again expensive and require the use of powerful workstations or even 194 supercomputers (particularly for the 3D case). Their use in practical design applications is further limited by the uncenainties in the soil properties and the cost and labor associated with extracting a sufficient number of undisturbed core samples over an area with large horizontal and vertical dimensions. This is an area in which supercomputers can contribute significantly to basic research in order to validate simpler approximate procedures which can be used in practice. Another problem closely related to the effects of local soil conditions on the seismic motions is the assessment of the liquefaction potential for a site and the estimation of the effects of such liquefaction on settlements and the behavior of the foundations. Although a substantial amount of work has been conducted, and is still going on in this area, most of it has been concerned with establishing correlations between the occurrence of liquefaction (observed in past earthquakes) and a number of different soil parameters (Seed 1979, Seed, Idriss and Arango 1983, Seed and Harder 1990, Leed et al 1984). The results of these studies are supposed to provide a mechanism to determine for a given site and a specific level of earthquake whether liquefaction is likely to occur or not. In most cases, these procedures allow one to conclude that liquefaction will indeed occur, that it will not, or that the site falls in a gray area where some liquefaction is possible. The more important question of what are the extent and consequences of the potential liquefaction is much harder to answer and would again require true nonlinear analyses with an appropriate two-phase constitutive model. This is another area in which some significant progress has been achieved during the last years (Dobry 1995), even if this work has not been associated with the design of nuclear power plants. Worth mentioning is the work that has been conducted in this area at the National Center for Earthquake Engineering Research (NCEER) (O'Rourke 1994). The application of these more sophisticated nonlinear dynamic analyses in practice, considering three-dimensional geometries and states of stress, may be again impractical, not just because of the cost of the analyses but primarily because of the lack of detailed information on the soil parameters over an extended volume. Yet this type of studies for research purposes are necessary in order to validate simpler approximate procedures which can be used for design purposes. Probabilistic formulations are also needed (and are beginning to be developed) in order to account in a rational way for the many uncertainties. 4. SOIL STRUCTURE INTERACTION ANALYSES The degree of sophistication of soil structure interaction analyses for nuclear power plants increased continuously with the development of new formulations and computer programs and the improvements in memory capacity and speed of computation. Yet it appears that in the last ten or fifteen years, in spite of the continued improvements in computational capabilities the trend in research has been towards ignoring the more rigorous methodologies already available in order to develop new, alternative, simplified procedures with different degrees of reliability. Soil structure interaction analyses were initially conducted replacing the foundation by a series of springs and dashpots (and sometimes lumped masses) with their values computed with available formulae for circular foundations on the surface of an elastic, homogeneous and isotropic half space. The viscous dashpots were intended to reproduce the loss of energy by radiation of waves away from the foundation, while the lumped masses, when used, were meant to reproduce the variation of the stiffnesses with frequency. This type of model was known (rather improperly) as the foundation impedance approach. It was intended to reproduce the inertial interaction effects. Kinematic interaction was ignored. When applying this model it was common to impose arbitrary bounds on the effective damping of the combined soil structure system particularly in modal analyses, because the nature and magnitude of radiation damping were not well understood. In some cases, however, the limitations were not imposed if the solution was performed in the frequency domain (indicating again a lack of understanding). The practical advantage of this approach was that most of the computer programs developed for general dynamic analysis of structures allowed one to incorporate these constant masses, springs and dashpots or at least the masses and springs. When a modal analysis was conducted the combined structure-foundation system would not have normal modes in the classical sense (real modes) when the dashpots were included and it was then necessary for a modal spectral analysis to compute equivalent values of modal damping using approximate formulae. All these problems disappeared when performing the analyses in the frequency domain. Unfortunately the normal structural analysis programs did not allow this type of solution. With increasing research (Veletsos and Wei 1971, Veletsos and Verbic 1974, Luco 1974, Kausel 1974), it became clear that the dynamic stiffnesses of a mat foundation are functions of frequency 195 and that the parabolic variation implied by the use of constant springs and masses is only valid over a very small range of frequencies. For the case of an elastic half space or a very deep soil deposit with homogeneous properties, which is not a frequent case in practice, the dynamic stiffness is nearly independent of frequency in the horizontal direction; the variation with frequency is, however, important for the vertical and rotational (rocking and torsional) stiffnesses and it depends on the value of Poisson's ratio (the variation is more pronounced as Poisson's ratio approaches a value of 0.5 corresponding to a nearly incompressible material or in practical terms a saturated soil deposit). The frequency dependence of the foundation stiffnesses is further affected by the variation of soil properties with depth. The existence of a much stiffer, rocklike material at some depth (and particularly at shallow depths) gives rise to marked oscillations around the half space solution corresponding to the natural frequencies of the soil deposit. More importantly, below the fundamental frequency of the soil there is no radiation of waves in the lateral direction and correspondingly no radiation damping (when there is some internal material damping there is also a very small amount of energy leakage and therefore radiation damping even below the threshold frequency but the amount is essentially negligible and it can be ignored for practical purposes). To account more realistically for all these effects as well as for the effect of foundation embedment (both on the stiffnesses and on the kinematic interaction) a number of computer programs specially conceived for dynamic analysis in the frequency domain were developed. Most of these programs performed the analyses assuming linear elastic behavior although some of them applied the iterative linearization to simulate nonlinear material behavior with two-dimensional states of strain. Only a small number of programs carried out the solution in the time domain with nonlinear constitutive equations for the soil. Two main approaches evolved from this research work: the analysis in a single step of the complete soil structure system, often referred to as the direct approach, and a three step or substructure approach (Kausel and Roesset 1994). The three steps are: a) Determination of compatible seismic motions for the foundation (kinematic interaction analysis). If the foundation can be assumed to be rigid, which is normally the case for reactor buildings, these motions will consist of at most six components (three translations and three rotations). For a flexible foundation, as a large mat supporting several buildings, it would be necessary to compute three translational components of motion at a sufficient number of contact points between the foundation and the surrounding soil or between the structure and the foundation. b) Determination of the foundation stiffnesses. For a rigid foundation this implies obtaining the terms of a symmetric 6x6 matrix applying unit harmonic displacements and rotations to the foundation and computing the resulting forces and moments, which will be complex functions of frequency. For a flexible foundation the dynamic stiffness matrix should be in general 3n x 3n if n is the number of contact points between the foundation and the soil or between the structure and the foundation. The terms of these matrices would be the reactions at each one of these points (assumed fixed) when a unit harmonic displacement in each of the three coordinate directions is applied at each point. c) Dynamic analysis of the structure supported on a continuum represented by the dynamic stiffness matrix of the foundation, and subjected to the motions computed in the first step. When each one of these steps is carried out in the frequency domain, the results are in terms of transfer functions. In the last step, which is the most time-consuming if one uses a detailed model of the structure, one can further separate the effect of the structure from that of the foundation through a modal synthesis or a simple condensation procedure. One can obtain in this way a dynamic stiffness matrix relating forces and displacements at the base of the structure which can be coupled directly to the dynamic stiffness matrix of the foundation. In this way, for each different set of soil properties due to imposed variations to account for uncertainties or to the variations caused by the levels of strain induced by different earthquakes, one must only solve a system of 6 equations with 6 unknowns for a rigid foundation or 3n equations with 3n unknowns for a flexible foundation, but the structural analysis needs to be performed only once. In the direct approach the structure is normally modeled through a combination of finite elements and linear members. The soil is discretized using finite elements or finite differences. Since a discrete model is used to reproduce a semi-infinite domain, special attention must be paid to the mesh size and to the boundary conditions imposed at the edges of the domain. The main advantage 196 of this approach is that is would permit a true nonlinear analysis with the complete interaction effects. A rigorous solution would require, however, a fully three-dimensional model and an appropriate set of nonlinear constitutive equations for the soil. In practice these requirements are rarely met. The first programs developed for direct soil structure interaction analyses used a two- dimensional plane strain model with elementary, viscous-type boundary conditions at the vertical edges to simulate radiation effects (for instance the program LUSH) (Lysmer et al 1974). Program FLUSH (Lysmer et al 1975) incorporated consistent lateral boundaries which could be placed directly at the edges of the foundation for a linear solution, and allowed to place dashpots on the sides of a finite width soil slice to simulate three-dimensional behavior. This was still basically a two-dimensional model as far as the structure was concerned and the lateral dashpots did not correctly reproduce the 3D soil behavior. For structures with axisymmetric geometry and horizontally layered soil deposits, a true 3D solution assuming linear behavior could be obtained formulating the problem in cylindrical coordinates and solving separately for vertical and torsional excitations on one hand and horizontal and rocking motions on the other. True nonlinear solutions in the time domain using the cap model were implemented in programs such as TRANAL (Baylor et al 1974) and FLEX (Vaughan 1983). In these cases the lateral boundaries of the finite element region must be placed at a sufficient distance from the foundation to guarantee that the reflected waves have a very small amplitude when reaching back the core region. In the substructure approach the foundation motions and the dynamic stiffnesses can be obtained using a discrete model with finite elements and a consistent boundary or using the boundary integral equation (or boundary element) method. Programs such as TRIAX, developed by Stone & Webster Corporation, used the first approach for structures with axisymmetric geometry whereas CLASSI, developed by J.E. Luco, used the second (indirect boundary element method). In the boundary element solutions the Green functions can be obtained from a continuum formulation, evaluating numerically the integrals in the wave number domain, or from a discrete formulation (Kausel 1981, Kausel and Peek 1982). All the above mentioned programs were developed in the middle and late 1970s and the early 1980s. They could be run in the mainframes available in those times with some limitations on the maximum number of layers or finite elements. With the advent of supercomputers or even with the present workstations the capabilities of these programs can be tremendously expanded. New and more powerful programs such as SASSI (Lysmer 1988) have been developed. The basis of these programs is again the use of a core region modeled with finite elements (or hyperelements) and a semi-analytical representation of the far field. These programs allow one to consider truly three- dimensional effects with a linear elastic solution. Programs such as SASSI would allow the inclusion of a number of effects which are normally ignored, such as layer interfaces which are not horizontal, flexibility of the mat foundation, variable degree of embedment along the perimeter of the foundation, structure-soil-structure interaction (effect of adjoining structures), etc.. A limited amount of work has been done on each of these topics at the research level and it is generally felt that they are secondary effects but this may not be so in all cases. So for instance the effect of the flexibility of the mat has been found to be small when dealing with a reactor building by itself, particularly in relation to global response parameters (Kausel 1974). It could be more significant, however, for other buildings or when dealing with several structures supported on a single, very large mat. The lack of uniform embedment is sometimes accounted for in practice by multiplying the contribution of the sidewalls to the foundation stiffnesses (using for instance Novak's procedure) by the ratio of the perimeter in contact with the soil to the total perimeter. This seems to be a relatively good approximation for the imaginary part of the stiffnesses (representing the radiation damping) but not for the real part (Chen 1984). Interaction between adjacent structures through the soil tends to be important when dealing with a light structure next to a much heavier one, but are considered small for the typical buildings encountered in nuclear power plants (Gonzalez 1977). Additional research to assess more fully the importance of these effects and to better delineate the conditions under which they may be safely neglected is still necessary. The main difficulty in including all or some of these effects in a true three-dimensional solution is the cost of computation, even for a supercomputer. A large number of parametric studies might not be possible in practice but a few studies (or even just one) to assess the magnitude of these effects and the variations in the results with those of simpler solutions may be warranted. True nonlinear effects such as the nonlinear soil behavior under three-dimensional states of strain and with an appropriate constitutive model, or separation effects (sliding and uplifting of the foundation) are still not included in these powerful programs and require a solution in the time domain. Nonlinear soil behavior is normally accounted for considering only the nonlinearities associated with the soil amplification problem 197 10 (using the equivalent properties resulting from the last cycle of the iterative procedure) or implementing the iterative scheme with a 2 or 3D model (which makes its validity much more questionable). Separation effects tend to be beneficial, reducing the base shear and overturning moment, but they may increase vertical accelerations near the axis of the structure and produce additional stresses in the mat (Wolf 1976, 1977, Roesset and Scaletti 1979). These effects are also strongly dependent on the properties and initial state of stresses in the soil. The proper consideration of nonlinear effects is the area where the major contributions to the soil structure interaction problem remain to be done. Even if true 3D nonlinear analyres were not possible in practice (at least in large numbers) and were not justified considering the lack of complete data on the soil parameters needed to fit the constitutive models research to assess more fully the validity of the linear or linearized analyses and the potential errors involved in present day procedures is badly needed. Part of the difficulty in launching this research may be in the selection of an adequate constitutive model out of the various available ones. Most of the research work on seismic soil structure interaction has been concerned with rigid, and in most cases circular, mats resting on or embedded in a horizontally layered soil deposit. This is due to the fact that this is the type of foundation most commonly encountered in nuclear power plants. A number of studies have been conducted to determine the dynamic stiffness and motions of rectangular foundations using boundary elements (mostly surface foundations but in some cases also embedded ones) (Dominguez 1978, a,b). These studies have led to a number of simplified procedures to obtain their stiffnesses from those of an equivalent circular mat or to approximate formulae to compute them directly (Dobry and Gazetas 1985). It should be noticed, however, that these approximate procedures are intended to match primarily the static values of the stiffnesses and that their frequency variation may not be as well reproduced. This is particularly so when dealing with a layered soil deposit where the properties vary with depth rather than an elastic half space (normally considered in these studies). A substantial amount of work has also been done on pile foundations including group effects. A number of rigorous formulations assuming a linear elastic soil and perfect bonding between the pile and the surrounding soil have been developed (Blaney et al 1976). Group effects can be accounted for with some approximations such as enforcing the compatibility of displacements along the axis of the pile rather than along its perimeter for the study of two piles, and enforcing interaction at the pile heads only for large groups (Kaynia and Kausel 1982, Sanchez Salinero 1983). These assumptions are no longer valid when considering closely spaced piles or very large numbers of piles, but are reasonable for many other cases, particularly if the layering of the soil is taken into account properly. Simplified procedures based on an elastic half space again raise questions when extrapolated to realistic soil profiles. The main limitation in the analysis of pile foundations is again the linear assumption. The behavior of the soil around a pile and particularly near the pile head is highly nonlinear. Approximations based on the assumption of a concentric annular cylinder with reduced soil properties are of value to provide a qualitative picture of the phenomenon but cannot provide accurate quantitative results. In reality the properties and width of this annular region should be changing with depth and instead of a single annulus with a sharp contrast in material properties with the surrounding soil a smooth transition in properties should occur (Cheng 1986, Kim 1987). The alternative is the use of P-y and T-z curves, as employed in the offshore industry, which better reproduce the nonlinear soil behavior under static loads (particularly monotonic loads) but which do not account for dynamic effects (Matlock and Reese 1960, Matlock 1970). Clearly much more work remains to be done on this type of foundation. Very little work has been done also on spread or strip footings with or without tie beams between them (Vardanega 1978). This is a frequency type of foundation for regular buildings but not so for nuclear power plants. Nuclear power plants are structures for which soil structure interaction effects may be important and normally beneficial when designing for smooth broad band response spectra. As a result a considerable amount of research was performed in this area in the 1960s and 70s under the sponsorship of the nuclear industry, and a considerable amount of knowledge was acquired through this research. By the time the research on seismic design of nuclear power plants started to wane the area had acquired great popularity and the research has continued during the last years with application to other types of structures (regular buildings, bridges, offshore structures, etc.) or in abstract terms. Unfortunately, most of this research has not been oriented towards the solution of the main outstanding issues and sources of uncertainty. Much of the recent work on dynamic or seismic soil structure interaction has dealt with the derivation of alternative procedures to solve problems for which results are already available with emphasis on obtaining more elegant formulations which can be more economical for the same degree of accuracy than the existing ones (boundary element 198 11 solutions in the time or frequency domains for instance) or simplified models which are more economical but still provide a reasonable approximation, at least for the particular cases studied. In the best cases these alternative formulations have provided a better insight into the problem or the relative importance of various effects. In a number of cases the studies have shown on the other hand a lack of understanding of the phenomenon with models which accounted (sometimes incorrectly) for the real part of the foundation stiffnesses but neglected the radiation damping, or with values of the soil and structural parameters which were totally unrealistic. Thus, in spite of all the research that has been conducted over the last 25 years and which has been not only disseminated in a large number of technical papers but even published in books (Wolf 1985, 1988, 1991). S.D. Werner in a state of the art paper published in 1991 wrote, "The overall engineering community has not had adequate exposure to SSI concepts, procedures, and evaluation of results that would enhance the incorporation of SSI provisions in current seismic design practice. Therefore provisions for workshops, conferences, and publications that present SSI to practicing engineers are encouraged." Even the recent DOE Standard, Natural Phenomena Hazards, Design and Evaluation Criteria for Department of Energy Facilities published in 1994 is rather weak and vague on the question of soil structure interaction. This document states for instance that "The shear modulus and material damping ratio used to evaluate foundation impedance shall be values compatible with the shear strain induced in the foundation medium during earthquake excitation," but it fails to indicate at what point or depth within the foundation medium this shear strain should be calculated, how it should be calculated or whether this is the strain due only to the seismic waves in the free field (in which case it would be the horizontal shear strain supposedly) or accounting also for the presence and vibrations of the structure. It also indicates that "Dynamic Modeling of the foundation medium is generally accomplished using a half space model," and later, "When significant layering exists in the foundation medium, it should be modeled explicitly or its effects considered." There is again no indication as to what constitutes significant layering or how its effects can be considered. A serious controversy existed for a long time between the two general approaches to Soil Structure Interaction analysis. The substructure approach referred to as the impedance approach was not allowed at one time because it was erroneously associated with the use of frequency independent springs and dashpots based on the static solution for an elastic half space. The direct approach which had been the recommended one later became unpopular because of the 2D nature of the solution. It appears that at the present time both might be acceptable but there seems to be a trend towards the simplest possible solutions as suggested by the above quotes. Given today's computer capabilities and the existence of computer programs that can provide at very little cost accurate solutions for a layered half space this trend is somewhat strange. Simplified solutions are of great value to gain insight into the behavior of the physical process, to identify key parameters and understand their effects and relative importance, to obtain preliminary estimates of the response, to assess whether effects can be important and more sophisticated analyses are necessary, and to provide checks to the results of more complicated models. There is, however, a risk of oversimplification. When effects are found to be important and when dealing with structures such as nuclear power plants one should always try to use the most accurate models available at least for a limited number of studies rather than relying exclusively on simple models. 5. STRUCTURAL MODELS The buildings encountered in a nuclear power plant are typically very stiff, massive and extremely complex structures, including many different structural types (thick shells for the containment, frames, trusses, heavy shear walls, thick slabs, etc.). A detailed modeling of any one of these structures requires a large number of degrees of freedom, and these models have been often used. In many cases, however, the seismic analyses were carried out using highly simplified models consisting of close-coupled systems of masses and springs (or equivalent stiffness matrices). These were referred to as "stick" models. Each element connecting two adjoining masses, or its stiffness matrix, was intended to reproduce the combined effects of all the members and walls between the two floor levels where the masses were lumped. The stiffness matrices were often computed with the implicit assumption that the floor slabs were infinitely rigid not only as diaphragms in their own plane but also in bending. The use of simplified models with a reduced number of degrees of freedom is of course common to many structures and not just nuclear power plants but the degree of simplification that was used at times in nuclear structures was unusually large. Thus while these models could be reasonably accurate to estimate the general, or global, features of the dynamic response, such as accelerations at various floor levels, some questions could be raised as to their 199 12 ability to reproduce local response parameters such as stresses or deformations in individual members. The main objective of the seismic analyses of nuclear power plants conducted with these models was in fact in many cases the derivation of floor response spectra for the design of equipment rather than a detailed stress analysis of the structure. The structural models could be even cruder for soil structure interaction analyses where, ironically, a considerable amount of effort might be devoted to the appropriate modeling of the soil while the structure was idealized as a stick with only a few masses or a solid block discretized with finite elements. This situation was more likely to be encountered when the complete analysis was performed in a single step (direct approach) than when using the three step or substructure approach (particularly if the modal synthesis procedure described earlier to reduce the structure to forces and displacements at its base was implemented). It was suggested in fact at one time that soil structure interaction analyses should be performed with a highly simplified model of the structure to determine the motions at the base of the structure and that these motions could then be used as input to conventional dynamic analysis programs which assume the structure on a rigid base. Not only does this approach ignore in most cases the rotational component of motion at the base, since conventional programs cannot take it into account, but the procedure is inadvisable and can lead to serious errors: inconsistencies in the structural models will result in shifts in the peaks and valleys of the transfer functions for the motions at the base of the structure. These transfer functions should exhibit some pronounced valleys at the natural frequencies of the structure whereas the transfer function for the structural response will exhibit peaks at the same frequencies. If these frequencies are not the same in the two structural models (the one used to compute the transfer function for the base motion and the one used later to compute the structural response from this base motion), the amplifications at the resonant structural frequencies will be greatly exaggerated. Even when the derivation of floor response spectra is the main objective of the analysis the stick models can introduce some significant errors. This is so, for instance when considering the vertical accelerations at various points on the floor without accounting properly for the flexibility and dynamic response of the slab. The error would become more significant as the flexibility of the floors increased. It is also common to ignore the equipment entirely in the derivation of the structural model. Uncoupling the equipment from the structure is justified in most cases because the mass of the former is very small compared to that of the latter. The exception is when the natural frequency of the equipment is very close to that of the structure and its mass is not negligible. Yet it is interesting to notice that accounting properly for the coupling between equipment and the structure in a rigorous way is relatively easy and not as laborious or expensive as often believed, particularly for analyses in the frequency domain using the substructure approach. It is only necessary to compute the transfer functions for the displacements at the location of the equipment due to unit forces applied at the same location and the transfer functions for the motions at these points due to unit excitations at the base of the structure. The main requirement for this approach is the availability of sufficient computer memory to store all the required transfer functions. This was a significant requirement at one time but is no longer so with the present cost of memory. Similar sources of errors are introduced when entire components (walls, slabs) are omitted from the structural model and then analyzed independently assuming uncoupled behavior. Even when these simplifications are reasonable in general terms, it is very hard to quantify the magnitude of the errors they can introduce which can become important when attempting to conduct more rigorous analyses to accurately define levels of performance under varying earthquake intensities. A considerable amount of time and effort was normally spent on the analysis and design of piping and pipe supports. A source of difficulty in these analyses was the fact that the structural model was not detailed enough in most cases to properly define the motions of the different supports. An even more serious problem in this case tended to be, however, the lack of consistency and up-to- date files with information on the as-built conditions or the latest modifications and the fact that analyses and redesigns were often performed by different persons with little communication. The various structures present in a nuclear power plant are often connected by a variety of ducts and pipes. The fact that each structure was analyzed independently made it very difficult to perform the seismic analysis of these connecting elements unless the input motions to the various structures were consistent and the transfer functions for the motions at the connection points in the different buildings were stored and available in the same data base. 200 13 All these comments point out the desirability of having a large data base where complete information on the nuclear power plant at any time, the models of the different structures and the soil, the seismic motions used for input, and intermediate results such as transfer functions of different effects, can be stored and retrieved as needed. This would not only affect the consistency and reliability of the seismic analyses but facilitate considerably the evaluation of potential changes and their effects, future redesigns or just checks requested at a later date. These points can become particularly important when the plant has reached its design life but it is desired to maintain it operational and therefore extend its service life. It is interesting to notice that data bases of this nature have been developed and used in other fields (offshore structures for instance). In the nuclear field, as stated in the Introduction, efforts along these lines have been conducted by a number of companies in Japan (Machiba and Sasaki 1990, Kaneuji et al 1990, Satoh et al 1990). A variety of methods have been used in the past to perform the dynamic analysis of the structural or soil structure models. They ranged from traditional modal spectral analyses directly using the smooth design spectra and performing the combination of the modal maxima through a number of different expressions (accounting or not approximately for correlation between the modal responses), to standard modal analyses in the time domain combining the time histories of the responses in each mode, direct integration of the equations of motion in the time domain or solutions in the frequency domain. For the second and third options it is necessary to have available earthquake time histories, whether corresponding to actual earthquake records or generated synthetically to match the target spectra; for the last option one can start with these time histories and obtain their Fourier transform or directly with the Fourier spectrum (or in some cases the power spectrum) of the desired motion. It was not uncommon to have different parts of the plant or different portions of the analyses carried out with different approaches using thus several of these methods simultaneously with very little consistency. Thus for instance the soil structure interaction analyses were carried out in many cases in the frequency domain which had no limitations on the effective damping (by opposition to the modal analyses), while the detailed structural analyses were performed with a modal spectral approach. The fact that different groups within the design team were in charge of different sections of the analysis explains in part the diversity in approaches and lack of consistency. By opposition to regular buildings which are designed on the assumption that they will undergo large inelastic deformations under a severe earthquake the structures in a nuclear power plant were always designed to remain linearly elastic even under the safe shutdown earthquake. A very small amount of inelastic behavior was implicitly assumed in allowing larger values of material damping for the safe shutdown than for the operating basis earthquake but it was supposed to be very small. The use of reduction factors based on a vaguely defined and often meaningless system ductility customary in the seismic design of regular buildings had been, however, wisely avoided. Three new and significant trends have been evolving during the last 15 or 20 years in the seismic design of structures: the design for performance criteria rather than simply for a desired factor of safety (or equivalent load and resistance factors); the incorporation of probabilistic concepts to perform a complete seismic risk analysis accounting for the uncertainties in all the different phases; and the use of reduction factors based on system ductility following procedures similar to those used for conventional buildings (although the allowable reduction factors are smaller). The first two trends represent a desire to obtain more accurate and realistic estimates of the effects of potential earthquakes on the behavior of the structures. The third trend is somewhat contradictory to the others and is in the opinion of the writer an undesirable step backwards. A proper use of performance-based criteria would imply predicting the behavior of the structure not only in the linear elastic range but also as yielding and inelastic action begin to occur and as they progress until a limiting serviceability condition or failure take place, as the loads are continuously increased. This is a clear improvement over a conventional elastic analysis that can only furnish information on the level of earthquake for which yielding would start or a limit analysis that might indicate the level of motion associated with failure. It also represents, however, the need for more rigorous and sophisticated models and analysis procedures. Present models for nonlinear dynamic analysis of structures can predict reasonably well elastic behavior or failure but they are not very good at estimating the extent of cracking, damage or inelastic deformations. The use of probabilistic formulations is the only rational way to assess the effect of the many existing uncertainties on the reliability of the final results. Combined with sensitivity analyses these formulations can help to identify the key parameters or effects controlling the response. These are 201 14 the parameters whose variations have a stronger influence on the final results. Once these variables have been identified, one can concentrate resources on decreasing their uncertainty or modeling the effects more accurately rather than spending time in attempting to reproduce accurately less significant effects. On the other hand, it must be realized that the use of probabilistic techniques will not and cannot by itself eliminate or decrease uncertainties; it can only provide an idea of their importance. The reduction in uncertainties can only be accomplished through improved understanding of the physical processes, more accurate determination of material properties and more accurate analysis procedures. The statement often made that refinement of the models or the methods of analyses used is unnecessary and even unwarranted when using a probabilistic approach since any errors due to inaccuracies will be encompassed within the variations due to the uncertainties is highly misleading. Probabilistic analyses cannot account for systematic errors introduced by inadequate models, and compounding errors and uncertainties may lead to a further decrease in the reliability of the results depending on the circumstances. Probabilistic risk analyses are based on the use of fragility curves which provide the probability of failure of a given component or structural arrangement as a function of the level of earthquake (defined by a single parameter such as the effective or peak ground acceleration). The definition of failure is arbitrary and one could in principle derive fragility curves for different levels of performance. These fragility curves would be applicable to a given type of earthquake (a given frequency content) unless the characteristics of the motion are changed as a function of the level of excitation. The basic question is the degree of accuracy with which these fragility curves can be computed or the accuracy of fragility curves based on highly simplified models. A number of different procedures have been suggested to obtain fragility curves. Some are based on relatively simple semi-empirical procedures, others on more rigorous random vibration analyses. It is not clear at the present time how different would be the curves obtained for a given type of structure using these alternative procedures or even how different the results would be if the same technique were to be used by several independent researchers or structural engineers. Thus, while the methodology to perform seismic risk analyses is reasonably well-established the ability to perform these analyses in actual practice with an appropriate degree of accuracy is questionable. The trend towards the implementation of these methodologies is still a positive one, but the validity and accuracy of the results of these analyses should not be accepted without question. Even more questionable is the use of the rather crude techniques commonly employed for regular buildings in the seismic design of important structures such as nuclear power plants. The use of reduction factors on the design spectra to account for the ability of a structure to withstand inelastic deformations is based primarily on the concept of ductility defined as the ratio of the maximum distortion to the elastic (or yield) distortion for a single degree of freedom elastoplastic (elastic - perfectly plastic) system. The actual reduction factors are generally larger than the allowable ductilities to account for a number of other factors often associated with the assumption that the actual strengths are underestimated or that analyses are conservative. This implies of course that more accurate analyses and estimation of the material strength coupled with these reduction factors would lead to unconservative results. The extrapolation of the behavior of an idealized single degree of freedom system to predict the response of an actual multidegree of freedom structure requires the use of a system ductility, which is defined only in loose terms, but is hard to quantify. The relation between this system ductility and the local ductilities of the different elements is a function of the structural type and many other factors. The use of the system ductility as a measure of damage therefore has very little basis and is inconsistent with the concepts of performance-based design or probabilistic risk analyses. To complicate matters further, the allowable ductilities are typically based on tests on single components subjected to idealized states of stress rather than the actual three-dimensional conditions encountered in a real structure. As a result, the ability to predict required ductilities with any reasonable accuracy using present models and methods of analyses, and even more with code-type procedures or spectral analyses and the possibility of quantifying the amount and type of damage that the structure may undergo for the predicted ductilities are highly questionable. 6. SUMMARY AND FINAL CONSIDERATIONS The major uncertainties in the seismic design of nuclear power plants have always been associated with the selection and characterization of the design earthquake (s), the soil properties to be used in the soil amplification and soil structure interaction analyses consistent with the in situ 202 15 state of stresses, the accuracy of the structural and soil structure models and methods of analysis, and the general treatment of nonlinear effects (nonlinear soil behavior, liquefaction potential, separation effects between the foundation and the soil, and to a limited extent only nonlinear structural response). In spite of the considerable reduction in the research effort related to nuclear power plants over the last 15 years substantial progress has been done in some of these areas. Thanks to the research conducted over the last years it is possible now to define the earthquake or earthquakes associated with a desired level of hazard, or probability of occurrence, in a much more realistic way than before, accounting for the effects of magnitude, source mechanism and distance not only on a single parameter such as the peak ground acceleration, or an effective acceleration, but also on the duration and frequency content. This represents a major breakthrough and an important departure from the traditional obsession with the use of only one variable. Much progress has also been achieved in the determination of soil properties in situ over an extended volume but only for low levels of strain. The reasons for the large differences that were observed between the field and laboratory measurements are now well understood and correction factors which can improve greatly the agreement between the two types of results have been developed. For one-dimensional situations the nonlinear soil behavior can be then inferred from the curves relating modulus and damping to level of strain. For two- or three-dimensional situations it would be necessary on the other hand to derive appropriate nonlinear constitutive models. The in situ determination of the parameters needed to define these models is in need of much more work. The techniques to perform linear one-dimensional soil amplification studies have been well- known for a long time. Nonlinear convolution and deconvolution analyses are still being carried out, however, using an iterative linearization technique which can lead to serious errors for deep and soft soil deposits Major advances have been made in the consideration of 2D and 3D geometries, as often encountered in practice. The assessment of the liquefaction potential at a site and the evaluation of the consequences of liquefaction is another area where the state of the art has improved considerably in the last years, with a much better understanding (after years of controversy) of the physical phenomenon, much more statistical data to validate numerical predictions, experimental data from centrifuge tests, and computational models capable of accounting for pore pressure buildup, nonlinear soil behavior and large deformations and displacements. There are now a number of rigorous formulations available to perform linear soil structure interaction analyses including all major effects. There is also an ever increasing inventory of simplified models and procedures which can be used for preliminary design purposes, to estimate the potential importance of various effects or to check the order of magnitude of the results from more sophisticated models. With present day computational capabilities, however, there can be no justification for the use of these simplified models in the final analyses instead of the more rigorous techniques which are available. The arguments often presented that these rigorous solutions can be extremely expensive are erroneous in most cases. The final area in which very significant progress has been achieved is the development of methodologies for complete seismic risk analyses. This is the only rational way to account for the uncertainties that will always be present and to reach decisions related to safety and performance. It should be noticed, on the other hand, that these methodologies by themselves will not be of real value unless accurate models and data are available to predict the structural performance. This is not yet the case in practice. It is clear that more work remains to be done in each one of these areas. The success achieved over the last years should encourage the industry and other funding sources to continue these lines of research if it is desired to avoid unnecessary controversies when the design of new nuclear power plants becomes a need. The area in which less progress has been done and a considerable amount of research is needed in the evaluation of nonlinear effects. If the unfortunate trend to accept ductility factors in the design of nuclear power plants, by opposition to the linear behavior that was essentially required, is accepted and implemented in design regulations the proper modeling of nonlinear structural behavior will become another major source of concern (and another source of controversy until this research is carried out). 203 16 As important as the improvement in the accuracy of the models and the analysis procedures is the maintenance of up to date files with information on the status of the plant, the analyses, intermediate results, modifications introduced, etc.. This will avoid the inconsistencies, omissions and mistakes which are likely to occur when many different and separate groups are involved in the various phases of the design process. The computer capabilities now available make it particularly easy to develop modern data bases where this information can be stored. 7. REFERENCES Aki, K. 1988. Local site effects on ground motion. State of the art report. "Recent advances in ground motion evaluation." Geotechnical Special Publication, ASCE, ne 20. Aki, K. & K.L. Lamer 1970. "Surface motion of a layered medium having an irregular interface due to incident plane SH waves." Journal of Geophysical Research, ne 75. Bard, P.Y. & M.A. Bouchon 1980. "The seismic response of sediment filled valleys." Parts I & II. Bulletin Seismological Society of America, vrlQ. Bard, P.Y. & M.A. Bouchon 1985. "The two-dimensional resonance of sediment filled valleys." Bulletin Seismological Society of America, n- 75. Baylor J., M. Bienieck & J. Wright 1974. TRANAL: A 3D finite element code for transient nonlinear analysis. DNA 3501F: Weidlinger Associates. Bernreuter, D.L., J.B. Savy, R.W. Mensing, J.C. Chen & B.C. Davis 1985. Seismic hazard and characterization of the eastern United States. Vol. 1 & 2. LLNL-UCID-20421. Bernreuter, D.L., J.B. Savy & R.W. Mensing 1987. "Seismic hazard of the eastern U.S. Comparative evaluation of the LLNL and EPRI studies." USNRC report NUREG/CR-4885. Bernreuter, D.L., J.B. Savy, R.W. Mensing & J.C. Chen 1989. "Seismic hazard characterization of 69 plant sites east of the Rocky Mountains." LLNL, USNRC report NUREG/CR-5250, UCED- 21517. Vol. 1-8. Blaney, G.W., E. Kausel & J.M. Roesset 1976. "Dynamic stiffness of piles." Proceedings 2nd International Conference on Numerical Methods in Geomechanics, ASCE. Chen, Huei Tsyr 1984. "Dynamic stiffness of nonuniformly embedded foundations." Report GR 84-10. Civil Engineering Department, The University of Texas at Austin. Cheng, Fu-Ping 1986. "Dynamic response of circular foundations in an elasto plastic soil medium." Report GR86-3. Civil Engineering Department, The University Texas at Austin. Constantopoulos, I.V. 1973. "Amplification studies for a nonlinear hysteretic soil model." Report R73-46. Civil Engineering Department, M.I.T. Dames & Moore 1978. "Study of nonlinear effects on one-dimensional earthquake response." EPRI Report NP-865. D'Appolonia Consulting Engineers, Inc. 1979. "Seismic input and soil structure interaction." Report prepared for the US. nuclear regulatory commission. NUREG/CR-0693. DiMaggio, F.L. & I. Sandier 1971. "Material model for granular soils." Journal of the Engineering Mechanics Division, ASCE. Vol. 97, n2 EM3. Dobry, R. & G. Gazetas 1985. "Stiffness and damping of arbitrary shaped machine foundations." Journal of Geotechnical Engineering, ASCE. Dobry, R. 1995. "Liquefaction and deformation of soils and foundations under seismic conditions." State of the art paper. 3rd International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics. Vol. JH. St. Louis, Missouri. DOE Standard 1994. "Natural phenomena hazards. Design and evaluation criteria for department of energy facilities." DOE-STD-1020-94. Dominguez, J. 1978a. "Dynamic stiffness of rectangular foundation." Report R78-20. Civil Engineering Department, M.I.T. Dominguez, J. 1978b. "Response of embedded foundations to travelling waves." Report R78-24. Civil Engineering Department, M.I.T. Dravinski, M. 1982. "Scattering of elastic waves by an alluvial valley." Journal of the Engineering Mechanics Division, ASCE, nfi 108. Dravinski, M. 1983. "Amplification of P, SV and Rayleigh waves by two alluvial valleys." Soil Dynamics and Earthquake Engineering, n^. EPRI 1986. "Seismic hazard and methodology for the central and eastern United States." Technical Report NP-4726A. 11 Volumes. EPRI 1993. "Guidlines for determining design basis ground motions." Technical Report TR- 102293. 204 17 Gonzalez, J.J. 1977. "Dynamic interaction between adjacent structures." Report R77-30. Civil Engineering Department, M.I.T. Jacob, K. 1994. "Quantifying seismic hazard and providing realistic ground motions for engineering applications primarily in the eastern United States." Research 1986-1994 accomplishments. NCEER. Jones, T.J. & J.M. Roesset 1970. "Soil amplification of SV and P waves." Report R70-3. Civil Engineering Department, M.I.T. Kaneuji, A., N. Akitomo & Y. Katoh 1990. "Technologies of computer integrated engineering and manufacturing systems for nuclear power plant." Proceedings 1st International Conference on Super computing in Nuclear Applications. Mito, Japan. Kato, M. et al 1990. "Simulation analysis of dynamic characteristics on actual reactor buildings." Proceedings 1st International Conference on Super computing in Nuclear Applications. Mito, Japan. Kausel, E. 1974. "Forced vibrations of circular foundations on layered media." Report R74-11. Civil Engineering Department, M.I.T. Kausel, E. and J.M. Roesset 1974. "Soil structure interaction for nuclear containment structures." Proceedings ASCE Power Division Specialty Conference. Boulder, Colorado. Kausel, E. 1981. "An explicit solution tor the Green functions for dynamic loads in layered media." Report R81-13. Civil Engineering Department, M.I.T. Kausel, E. & R. Peek 1982. "Boundary integral method for stratified soils." Report R82-50. Civil Engineering Department, M.I.T. Kaynia, A.M. & E. Kausel 1982. "Dynamic stiffness and seismic response of pile groups." Report R82-03. Civil Engineering Department, M.I.T. Kim, Y.S. 1987. "Dynamic response of structures on pile foundations." Ph.D. Dissertation. The University of Texas at Austin. Luco, J.E. 1974. Impedance functions for a rigid foundation on a layered medium. Nuclear engineering and design.. Lysmer, J., T. Udaka, H.B. Seed & R.N. Hwang 1974. LUSH. "A computer program for complete response analysis of soil structure systems." Report EERC 74-4. University of California, Berkeley. Lysmer, J., T. Udaka, C.F. Tsai & H.B. Seed 1975. "FLUSH. A computer program for approximate 3D analysis of soil structure interaction problems." Report EERC 75-30. University of California, Berkeley. Lysmer, J. et al 1988. SASSI. "A computer program for dynamic soil structure interaction analysis." Report UCB/GT81-02. University of California, Berkeley. Lysmer, J. 1994. Personal communication. Machiba, H. & Sasaki, N. 1990. "Ioshiba CAE system for nuclear power plant." Proceedings 1st International Conference on Super computing in Nuclear Applications. Mito, Japan. Matlock H. & L.C. Reese 1960. "Generalized solutions for laterally loaded piles." Journal of the Soil Mechanics and Foundations Division, ASCE. Vol. 86 SM-J. Matlock, H. 1970. "Correlations for design of laterally loaded piles in soft clays." Proceedings 2nd Offshore Technology Conference. Paper OTC 1205. Miyamoto, A. et al 1990. "Response analysis of the PWR type reactor building for vertical earthquake motion using supercomputers." Proceedings 1st International Conference on Supercomputing in Nuclear Applications, Mito, Japan. Nazarian, S. 1984. "In situ determination of elastic moduli of soil deposits and pavement systems by spectral analysis of surface waves method." Ph.D. Dissertation. The University of Texas at Austin. O'Rourke, T. 1994. "Soil liquefaction, large ground deformation and earthquake resistant design of lifelines." Research 1986-1994 accomplishments. NCEER. Papageorgiou, A. & J. Kim 1992. "Propagation and amplification of seismic waves in 2D valleys excited by obliquely incident SV and P waves." Earthquake Engineering and Structural Dynamics, n222. Prevost J.H. 1977. "Mathematical modeling of monotonic and cyclic undrained clay behavior." International Journal of Numerical and Analytical Methods in Geomechanics. Vol. 1. n2 2. Roesset, J.M. & R.V. Whitman 1969. "Theoretical background for amplification studies." Report R69-15. Civil Engineering Department, M.I.T. Roesset, J.M. & H. Scaletti 1979. "Nonlinear effects in dynamic soil structure interaction." 3rd International Conference on Numerical Methods in Geomechanics. Aachen, Germany. Roesset, J.M. & Y.S. Kim 1987. "Specification of control motions for embedded foundations." Proceedings 5th Canadian Conference on Earthquake Engineering. 205 18 Roesset, J.M., D.W. Chang & K.H. Stokoe, II 1991. "Comparison of 2D and 3D models for analysis of surface wave tests." Soil Dynamics and Earthquake Engineering V: Elsevier Applied Science. Roesset, J.M. C. Huerta & K.H. Stokoe 1995. "Effect of magnitude and type of damping on soil amplification." 3rd International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics. St. Louis, Missouri. Sanchez Salinero, I. 1983. "Dynamic stiffness of pile groups: approximate solutions." Report GR83-5. Civil Engineering Department, The University of Texas at Austin. Sanchez Sesma, FJ. 1983. "Diffraction of elastic waves by three-dimensional surface irregularities." Bulletin Seismological Society of America, vrl3. Satoh, S. et al 1990. "The use and limitations in using supercomputers in structural design for nuclear power plants." Proceedings 1st International Conference on Supercomputing in Nuclear Applications. Mito, Japan. Schnabel, P.B., J. Lysmer & H.B. Seed 1972. "SHAKE. A computer program for earthquake response analysis of horizontally layered sites." Report EERC 72-12. University of California, Berkeley. Seed, H.B. & I.M. Idriss 1969. "Influence of soil conditions on ground motions during earthquakes." Journal of the Soil Mechanics and Foundations Division, ASCE. Seed, H.B. 1979. "Soil liquefaction and cyclic mobility evaluation for level ground during earthquakes." Journal of the Geotechnical Engineering Division, ASCE, Vol. 105, n- 2. Seed, H.B., I.M. Idriss & I. Arango 1983. "Evaluation of liquefaction potential using field performance data." Journal of the Geotechnical Engineering Division, ASCE, Vol. 97, n2 3. Seed, H.B., K. Tokimatsu, L.F. Harder & R.M. Chung 1984. "The influence of SPT procedures on soil liquefaction resistance evaluations." Report UCB/EERC 84/15. University of California, Berkeley. Seed, H.B., & L.F. Harder, Jr. 1990. "SPT based analysis of cyclic pressure generation and undrained residual strength." Proceedings H. Bolton Seed Memorial Symposium: Vol. 2. Sheu, J.C. 1987. "Application and limitations of the spectral analysis of surface waves method." PhD. Dissertation. The University of Texas at Austin. Systems, Science and Software 1980. "New endochronic plasticity model for soils." EERI Report NP-1388. Vardanega, C. 1978. "Soil structure interaction effects on the dynamic response of shear wall buildings." M.S. Thesis. Civil Engineering Department, M.I.T. Vaughan, D. 1983. "FLEX User's guide." Document UG8298:Weidlinger Associates Veletsos, A.S. &Y.T. Wei 1971. "Lateral and rocking vibrations of footings." Journal of the Soil Mechanics and Foundations Division, ASCE, Vol. 97. Veletsos, A.S. & B. Verbic 1974. "Basic response functions for elastic foundations." Journal of the Engineering Mechanics Division, ASCE, Vol. 100. Werner, S.D. 1991. "Soil structure interaction: the state of practice and recommended research needs." NSF Workshop on Experimental Needs for Geotechnical Earthquake Engineering. Albuquerque, New Mexico. Wolf, J.P. 1976. "Soil structure interaction with separation of base mat from soil (lifting off)." Nuclear Engineering and Design. Vol. 38. Wolf, J.P. 1977. "Seismic response due to travelling shear waves including soil structure interaction with base mat uplift." Earthquake Engineering and Structural Dynamics. Vol. 5. Wolf, J.P. 1985. Dynamic soil structure interaction: Prentice Hall. Wolf, J.P. 1988. Dynamic soil structure interaction in time domain: Prentice Hall. Wolf, J.P. 1994. Foundation vibration analysis using simple physical models: Prentice Hall. Yamada, A., K. Miura & T. Kobori 1995. "Nonlinear analysis method for prediction of base motion." 3rd International Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics. St. Louis, Missouri. 206 . 2 - XA9952656 DYNAMIC ANALYSIS OF WER TYPE NUCLEAR POWER PLANTS USING DIFFERENT PROCEDURES FOR CONSIDERATION OF SOIL-STRUCTURE INTERACTION EFFECTS L. Halbritter, N.J. Krutzik Siemens AG, Power Generation Group (KWU), Offenbach, FRG ABSTRACT: The dynamic response of structures due to seismic loadings is conventionally analyzed in the time domain using modal substructure procedures. This procedure uses frequency- independent parameters to represent the soil underneath the structure. These parameters are tuned to the main frequencies of the soil-structure system. This is a common procedure widely used in the preliminary design of power plant structures and provides conservative results. However, parallel to the rapid progress being made in upgrading the capability of data processing systems, methods and software tools have become available which work only in the frequency domain using complex mathematical models or models in which the soil is represented by frequency-dependent impedances. This complex method also allows realistic treatment of kinematic interaction effects and especially consideration of the embedment parameters of the building structure. The main goal of the study presented here was to demonstrate the effects of different procedures for consideration of soil-structure interaction on the dynamic response of the structures mentioned above. The analyses were based on appropriate mathematical models of the coupled vibrating structures (reactor building, turbine hall, intermediate building structures) of a WER 440/213 as well as a WER 1000 and the layered soil. On the basis of this study, it can be concluded that substructure models using frequency- independent impedances and cut-off of modal damping usually provide conservative results. Complex models which allow the soil-structure interaction effects to be realistically represented (by coupled models of the soil and the structure or by frequency-dependent impedances) provide more accurate results. The advantage of the frequency-domain analysis will be demonstrated and discussed, based on results obtained for the WER 440/213 PAKS and WER 1000 Kozloduy. 1 INTRODUCTION The treatment of soil-structure interaction effects in the analysis of structures founded on the surface or embedded in the soil is still one of the most discussed issues in the field of aseismic design and requalification of operating nuclear power plants. In the course of verifying and upgrading seismic input data for a number of sites with operating nuclear power plants, attention was focused on the conservatism of the design procedures and methods used in the past for the design of these units. Although the building structures generally possess enough capacity reserves to sustain higher loads, the conservative design procedures have to be replaced by more realistic methods due to the vast requalification effort required for components and systems which have to be upgraded or, in extreme cases, strengthened to accommodate higher seismic loading. In the framework of requalifying of a number of operating German-type or WER-type (former USSR) nuclear power plants at sites where the seismic input definitions were increased, investigations were performed using different calculation methods and procedures for representation of the soil. It was observed that the conservatism of various substructure methods 207 -3- {Figure 1) that generally assume frequency-independent soil parameters (indirect method) and solve the equations of motion in the time domain could be reduced significantly by means of a direct method (Figure 2) in which the frequency dependency of the soil parameter is realistically considered (by a complex mathematical model of the soil and the structure or by frequency- dependent soil parameters) and in which the analysis is performed in the frequency domain. Unlike the indirect method (Figure 1), which is an analysis in the time domain, the direct method also allows realistic treatment of kinematic interaction effects and, especially, consideration of the real embedment parameters of the structure. In order to demonstrate the conservatism of the indirect methods, the results obtained for the main buildings of a WER 440/213 and WER 1000 by means of the two methods mentioned above will be compared and evaluated. 2 DESCRIPTION AND IDEALIZATION OF THE STRUCTURES 2.1 WER 440/213 Paks Nuclear Power Station consists of so-called twin units. The main buildings are connected two-by-two on a common monolith basemat of 2-m thickness and have a symmetrical layout (Figure 3). The bottom of the basemat is set at an elevation of -8.5 m. On this foundation (145 m long, 52 m wide) there are two condensing towers with a base surface of 42 x 24 m. The towers rise to an elevation of 50 m and are designed to withstand a 2.5-bar pressure generated by a LOCA. Above the 18.9 m elevation there is a hall used for reactor maintenance and refueling. The floor and wall thicknesses vary from 0.6 to 1.5 m and comply with structural and radiation shielding requirements. On the eastern side of the building, a north-south oriented gallery building of 12 m in width and the turbine hall with a span of 39 m are attached to the reactor building. Both are constructed of steel. On the south as well as the north sides there is also a gallery building which as attached to the reactor building and is supported partly by reinforced concrete pillars and by the reactor building wall. This plant is characterized by structural elements with different stiffness properties, resulting in a complex mixed structure. For this reason an accurate and detailed model was necessary. In order to ensure adequate treatment of interaction effects, all the structures described in the previous section were modeled in only one 3D finite element model (main building, turbine hall and galleries). The finite element model of one unit of the Paks nuclear power plant has 9930 dynamic degrees of freedom. It comprises 1675 nodal points and 2132 trapezoidal elements, 470 triangular elements and 1005 beam elements. A general view of the finite element model is shown in Figure 5. The weight of the structure is composed of the weight of the structure itself, the weight of mechanical and electrical components as well as various live loads. The total weight of the model is approximately 2 200 000 kN. 2.2 WER 1000 The reactor building (Figure 4) is designed as a square, reinforced-concrete structure, each side measuring approximately 67.8 m in length. It is supported on a 2.8-m-thick foundation slab. The bottom of the building foundation is located approximately 7 m below plant grade. In view of the thickness of the foundation slab this results in a total height of approximately 73 m, of which 66 m are above plant grade. The load-bearing and stiffening members mostly comprise walls and floors. Up to approximately the 13.2-m elevation, the walls and floors, which for reasons of radiation shielding as well as structural requirements are of massive design, form a composite system of rigid cells. Above the 13.2-m elevation, the reactor building is subdivided into three structures separated by construction joints: the surrounding building, the prestressed concrete containment and the reactor section. 208 -4- All three structures are supported on a 2.4-m-thick floor. The surrounding building encompasses the containment. Apart from the outer and inner walls, only the horizontal floors make any significant contribution to its structural rigidity. The containment has a diameter of 45 m and a wall thickness of 1.2 m. It is a prestressed concrete structure of cylindrical shape with a spherical dome at the top which is designed with a cylindrical transition. The mathematical model and the degree of discretisation were chosen such that the natural behavior of the structure in the relevant frequency range could be computed with good reliability. Furthermore the number of nodes at which information is needed influenced the modeling. Considering the geometric shape, stiffening and mass distribution of the reactor building under concern, as well as the frequency content of the dynamic excitation, an equivalent beam model was used in the first calculations (Figure 6). The beam model includes the outer structure, the containment, the inner structure and the basement structure as an entirely connected total system. Derivation of the equivalent stiffnesses and masses was performed using a computer on the basis of input data and assumptions defined for each floor and region. 3 METHOD OF ANALYSIS 3.1 Substructure model approach The analyses were based appropriate models for the building structures as well as frequency- independent soil impedances (equivalent springs and dampings). As mentioned above, appropriate mathematical models with an adequate degree of discretization were used. The equivalent springs and damping elements, were derived from the corresponding impedance functions (Figures 7-10) and distributed appropriately at nodes of the discretized foundation of the buildings (Figures 5 and 6). Seismic excitation was represented in both cases by three artificial time histories compatible with the given free-field spectra. At the base of the building, excitations were subsequently applied which represented the deconvoluted motion at the embedment level (Figures 11 and 13) in the translational (horizontal 1 and 2) and vertical directions. For the WER 440/213 (PAKS) the maximum horizontal acceleration was 3.50 m/s2 and the duration of the time history was 15 seconds. In the vertical direction, values scaled down by a factor of 2/3 were used. In the case of the WER 1000 the maximum horizontal acceleration was 2.0 m/s2, the duration of the excitation 60 seconds and in the vertical direction values scaled by 0.5 were used. The dynamic analyses were conducted using the STRUDYN finite element program 171. In accordance with German standards (KTA 2201.4), the analysis was performed assuming a cut-off damping of 15 % and 30 % for the horizontal and vertical directions, respectively. 3.2 Complex model approach The structural models of the building described above were coupled with the mathematical model of the soil to form one complex finite element model (Figures 13 and 14). The numerical computations were performed using the SASSI /6/ computer code which solves the equations of motion in the frequency domain. In SASSI the flexible volume method is adopted for coupling the two structures and allows a horizontally layered halfspace representing the soil region under the foundation to be taken into account. Based on the parameters of the layered soil and the geometric form of the foundations in the first step of the calculation the frequency-independent impedance matrices are computed. The impedance matrix for the soil was determined by inverting the flexibility matrix calculated for all interaction nodes which are common to the structure and soil region. The impedance matrix was added to the stiffness matrix of the structure and this resulted in a final complex matrix for all degrees of freedom needed to describe the behavior of the system. The flexibility matrix was calculated for all interaction nodes using an axisymmetric model of the soil deposit with one core element connected to lateral transmitting boundaries allowing for 209 -5- dissipation of energy as radiation damping. The underlying halfspace was simulated through additional layers with viscous dampers at the lower boundary. To represent the real soil-structure interfaces, the beam model was coupled to a rigid foundation plate having the real length and width of the building. The structural models of the building were then coupled with the mathematical soil models as before. The control motion, i.e. input acceleration time history, was defined for the surface of the free-field. The Fast Fourier Transform technique was applied to obtain the response at any node in the models. The equations of motion were only solved at selected frequencies and the results for all "in-between" frequencies were interpolated. 4 CHARACTERISTIC RESULTS To demonstrate the influence of different assumptions regarding the representation of the soil- structure interaction effects on the dynamic response of the VVER 440/213 and VVER 1000 building structures during earthquake excitation, response spectra obtained for characteristic regions Figures 15 and 16) were computed by both procedures and compared (Figures A1 to A21 and B1 to B18). A comparison of the respective spectra for 2 % shows that there is a significant difference between the dynamic response results obtained by means of the two procedures (substructure and complex models). When a analyzing the response spectra it can be generally observed that there is a slight shift in the fundamental frequency of the building and a reduction in the spectral accelerations when using complex models. As regards the higher damping capacity of a more soft soil, the reduction effects obtained for the WER 440/213 (PAKS site) are higher than for the WER 1000 Kozloduy site). Because of the greater effect of soil damping (material and radiation damping) on the dynamic behavior of the lower part of the building, the reductions are more pronounced at the foundation level than on the upper floors of the building. However, on the upper floors the reduction in the peak frequency range is by about a factor of 3 in the horizontal direction and up to 25 - 50 % in the vertical direction. The reduction in the rigid body part of the spectra is nearly the same. It is evident that the lower structural response obtained (in the frequency domain) using coupled soil-structure models is due to the realistic representation of the soil capabilities as well as consideration of the real damping capacity of the coupled soil-structure system. 5 SUMMARY AND CONCLUSIONS Soil-structure interaction analyses using different analytical approaches and solution procedures were performed for seismic loading and the soil conditions of the main buildings of the nuclear power plants VVER 440/213 and WER 1000. In general, complex model analysis (considering the frequency dependency of the soil parameters) yielded significantly lower accelerations, especially in the frequency range above the fundamental frequency, compared to the accelerations obtained through conventional substructure approaches. This is due to the following influences: - Consideration of the real damping capacity - Consideration of kinematic interaction effects by the complex mathematical model - Consideration of the real stiffness of the foundation plate - Filtering effects of the soil. On the basis of this study, it can be concluded that the substructure model approach using frequency-independent (discrete) stiffnesses, dampings and cut-off of modal damping usually provides very conservative results. On the other hand, the complex model approach, which allows soil-structure interaction effects to be represented more realistically, provides more reliable results. 210 -6- This approach yields a realistic and efficient design which is not only safe but also economical, it may be recommended for use especially in situations requiring verification of existing building structures due to upgraded seismic input data. 6 REFERENCES /1/ Katona, T., Turi, L, Halbritter, A., Krutzik, N.: Experimental and Analytical Investigation of PAKS Nuclear Power Plant Building Structures, 10 WCEE - Madrid, July 1992 121 Ewers, J., Hitzschke, U., Krutzik, N.J., Papandreou, D., Schiitz, W., Time Versus Frequency Domain Analysis of Nuclear Power Plant Building Structures, 12th SMiRT Conference, Stuttgart, August 1993 13/ Ambriashvili, Y.K., Boyadjiev, Z., Krutzik, N.J., Papandreou, D., Schutz, W.: Structural Response Behavior of a Standardized VVER 1000 Nuclear Power Plant Using Substructure and Coupled Models, 10th European Conference on Earthquake Engineering, August 1994, Vienna 141 Main Building Complex PAKS, Structural Dynamic Analysis for Seismic Loading (Time Domain Calculation), Working Report Siemens KWU NDA2/94/E173 /5/ Structural Dynamic Analysis for Seismic Loading of the Main Building Complex PAKS (Frequency Domain Analysis), Working Report Siemens KWU NDA2/94/E0288 /6/ SASSI/Siemens (1981), A Computer System for Analysis of Soil-Structure Interaction, User Manual, Siemens VAX Version 111 STRUDYN, General Computer Program for Linear, Elastic, Static and Dynamic Analysis, User Information Manual, VAX Version 3/1991 Coupled System Coupled Model Soil-Structure Spatial ' Model Soil /Model V7/////////////77/////////S •< • Excitation Excitation Fig. 1 Complex Method of Analysis of the Soil Structure Interaction Effect (Coupled Models) 211 -7- Total System Kinematic Frequency Idealization of Soil-Structure Interaction Dependent ths Total System Impedance Functions M 3C Model Layered • • • Soil M i i ; •;»• a ? < : ; : j Frequency 5 ^ • .• - • i Independent 15; ••'- S 1/////////////// Soil Impedances Excitation XG Fig. 2 Substructure Method of Analysis of the Soil Structure Interaction Effect (Decoupled Models) Longitudinal Cross Section (N - S Direction) Perpendicular Cross Section (E • W Direction) Fig. 3 Constructional Concept of a WER-440/213, PAKS 212 -8- Fig. 4 VVER-1000 Reactor Building KOZLODUY, Characteristic Output Regions Containment 4-5 60 41.40 " " 36.90 36.60 - " 3J.60 33.60 - "28.80 " " 24.60 - - 19.40 20.40 4- 16.80 16JM 13.20 0.00 -<20 -s.eo u.. Fig. 6 VVER-1000 USSR Beam Model of the Reactor Building 213 -9- Turbine Hall Spatial Model Soil Impedances :i2 / >M Fig.5 Substructure Model (Spatial) of an 1300 MW BWR Reactor Building Frequency-Independent Soil Idealization 214 - 10- 5.0OE 7 l.SOE n Direction XX 2.SOE 7 .00 WIN O .00 AVE 4 MAX • -1.50E 11 MIN o AVE A MAX • -2.50E 7 -3.00E 11 -SOOE 7 0 3.0 6.0 9.0 12.0 IS.u -4.50E 11 0 30 6.0 9.0 12.0 15 0 FREQUENCY [HZ] FREQUENCY [HZ] 5.006 7 l.SOE 11 Direction YY 2.S0E 7 Direction Y .00 .00 •l.SOE 11 -2.50E 7 •3.00E 11 I •4.50E 11 ' ST3 ST5 12T5 iTo •5.00E 7 0 3.0 6.0 9.0 12.0 15.0 FREQUENCY [HZ] FREQUENCY [HZ] 4.00E 8 5 00E 10 Direction 2Z Direction Z ,oo| 400E 10 •4.00E 8 2.0OE 10 -8.006 8 .00 •1.20E 9 0 3.0 6.0 90 12.0 15.0 •2.00E 10 0 3.0 6.0 9 0 12 0 15 0 FREQUENCY [HZ] FREQUENCY [HZ] Fig. 7 Reactor Building VVER-440/213 PAKS Impedance Functions (Real Part) for Translational and Rotational Modes 3.80E 11 2.20E 8 Direction X Direction XX 1.6SE 8 MIN o 1.10E 8 AVE 4 5.50 7 0 3.0 6.0 9.0 12.0 15.0 •00 o 3.0 6.0 9.0 12.0 15.0 FREQUENCY [HZ] FREQUENCY (HZ) 2 206 8 5 4.52E 11 OiTKttonY Direction YY 1.65E 8 t 3-02E 11 • 1.S2E 11 1.10E 8 5.50 7 I I -1.48E 11 .00 0 3.0 6.0 80 12 0 15.0 0 3.0 6.0 9.0 12.0 15.0 FREQUENCY [HZ] FREQUENCY [HZ] Z 1.60E 11 DnctfcmZ Direction ZZ 5 1356 11 | 9.00E 10 g 450 10 2 ' 0 30 6 0 S.O 12.0 15.0 00 0 30 60 9.0 12.0 15.0 FREQUENCY [HZ] FREQUENCY [HZ] Fig. 8 Reactor Building VVER-440/213 PAKS Impedance Functions (Imag. Part) for Translational and Rotational Modes 215 -11 - 00 20 a.o IOO 00 20 DIRECT©* X DIRECTION XX Fig. 9 Reactor Building WER-1000 KO2LODUY Impedance Functions (Real Part) DIRECTION X DIRECTION 2 00 ?0 o eo 8 0 DIRECTION XX Fig. 10 Reactor Building WER-1000 KOZLODUY Impedance Functions (Imaginary Part) 216 - 12- 2.0 ; 1.E-1 2.E-1 S.E-1 1.E0 2.60 S.EO 1.E1 2.E1 S.E1 1.E2 f nquency fHZ] Frequency [HZ] Fig. 12 Reactor Building KOZLODUY, Unit 5 Fig. 11 Seismic Input Information for the Site PAKS Response Spectra of Foundation Level (Benchmark 3) Turotn* Building Reactor Building Lay«r«d Soil xwwwwv •WWY 5 ™ 60 MN' 40 m 160MNT1? 5Srr 250 MNfnJ Fig. 13 Mathematical Model for the Coupled System Traverse Direction (Direct Method) G= WOMN/m2 D= 7% p= 1.80 t/nf G= 330 MN/m2 0= 7% v= 0.36 p= ZMt/nf Uyared Soil 0= 7% v= 0.45 p= 2.00 t/m3 Fig. 14 Model of Reactor Building and Foundation (Complex Model) 217 - 13- £§> i i«z 183 18* iesi las 1 |:na IOT IK: HI IB US IS« US IM»W r^g rsi ??r ^ ^ Pi; gj S >^"3*a 3^13 Fig. 16 VVER-1000 MW Reactor Building KOZLODUY, Characteristic Output Regions Fig. 15 Reactor Building VVER JJ40/213. PAKS Characteristic Output Regions Foundation Level -6.50 m -I 1 1 00 20 4.0 60 80 100 120 K.O 16 0 180 20 0 0.0 2.0 4.0 6.0 8.0 100 120 U.O 160 18.0 20 0 Frequency [HZ] Frequency (HZ) Fig. A-1 Reactor Building VVER-440/213 PAKS. Fig. A-2 Reactor Building WER-440/213 PAKS, Comparison of Response Spectra, Comparison of Response Spectra, Foundation Level -6.50 m. Direction X1 Foundation Level -6.50 m, Direction X2 I suasm MODEL • SUBSTR MODEL I COMPLEX MODEL • COMPLEX MO061 1 IS^H 1 1 i i 1 1 1 1 1 00 2.0 40 60 80 100 120 14 0 16 0 180 20 0 0.0 2 0 4.0 6.0 8 0 10.0 12.0 14.0 16.0 18.0 20 0 Frequency |H2] Frequency |H2) Fig. A-3 Reactor Building WER-440V213 PAKS. Fig. A-3a Reactor Building WER-440/213 PAKS. Comparison of Response Spectra. Comparison of Response Spectra, Foundation Level -6.50 m. Direction X3 Foundation Level -6.50 m (Central Point), Direction X3 218 - 14- • SJ8STR MODEL : j • COMPLEX MODEL _-. • .' — =_ H= J OC 2C 40 6C 80 -.CO '2 0 '40 16.0 -8 C 2 00 20 Fig. A-4 Reactor Building VVER-440/213 PAKS. Fig. A-5 Reactor Building VVER-440/213 PAKS, Comparison of Response Spectra, Comparison of Response Spectra, Elevation 0.00 m. Direction X1 Elevation 0.00 m. Direction X2 SjBSTfl MOOEL COMPLEX MOOtL 0.0 2.0 4.0 6.0 8 0 10 0 12.0 14 0 16 0 18 0 20 C 0.0 2 0 4.0 6.0 8 0 10.0 12 0 14 0 16 0 18 0 20 0 Frequency [HZ] Frequency |HZ] Fig. A-6 Reactor Building WER-440/213 PAKS, Fig. A-7 Reactor Building WER-440/213 PAKS, Comparison of Response Spectra, Comparison of Response Spectra, Elevation 0.00 m. Direction X3 Elevation 6.00 m. Direction X1 • SU8STR MOOEL I SU8STR MOOEL . • COMPLEX MODEL > COMPLEJ! MODEL < ° 00 2.0 40 60 80 100 120 140 160 180 20 0 00 20 40 60 80 100 12.0 140 16 0 180 20 0 Frequency |HZ] Frequency |HZ] Fig. A-8 Reactor Building WER-440/213 PAKS. Fig. A-9 Reactor Building WER-440/213 PAKS. Comparison of Response Spectra, Comparison of Response Spectra, Elevation 6.00 m, Direction X2 Elevation 6.00 m. Direction X3 219 - 15- 00 2.0 40 6.0 80 100 12 C US 16! 2 0 t.C 6 0 SO 10 0 12 0 «! 16 0 18 0 20 C Frequency ;^Z] Frequency [HZ] Fig. A-10 Reactor Building VVER-440/213 PAKS, Fig. A-11 Reactor Building VVER-440/213 PAKS. Comparison o< Response Spectra, Comparison of Response Spectra, Elevation 10.50 m. Direction X1 Elevation 10.50 m. Direction X2 -r -r 00 20 40 60 80 100 12 0 140 160 180 20 0 00 40 80 100 12.0 HO 160 20 Frequency [HZ] Frequency |HZ] Fig. A-13 Reactor Building WER-440/213 PAKS. Fig. A-12 Reactor Building WER-440/213 PAKS, Comparison of Response Spectra, Comparison of Response Spectra, Elevation 18.90 m, Direction X1 Elevation 10.50 m. Direction X3 • SUBSTB MODFL • COMPLEX MODEL • SUBSTR MODEL • COMPLEX MOOEL i I I 00 20 40 60 80 10 0 12.0 14.0 16 0 18 0 20 0 0 0 20 40 60 80 10.0 12 0 14 0 16 0 18 0 20 C Frequency (HZ] Frequency [HZ] Fig. A-14 Reactor Building WER-440/213 PAKS, Fig. A-15 Reactor Building VVER-440/213 PAKS, Comparison of Response Spectra, Comparison of Response Spectra, Elevation 18.90 m. Direction X2 Elevation 18.90 m, Direction X3 220 -16- r> • B SU6STR MOOEL O 1 x> ] i I D - *—^-im V -i 1 r J i j- 1 1 \ 1 1 1 1 1 1 "OO 2 0 4 0 6 0 8 0 10.0 12,0 110 16.0 18 0 20 C 00 20 40 60 80 100 120 14 0 160 180 Frequency [HZ\ Frequency (HZ] Fig. A-16 Reactor Building VVER-440/213 PAKS. Fig. A-17 Reactor Building VVER-440/213 PAKS. Comparison of Response Spectra. Comparison of Response Spectra. Top-Barbotage, Direction X1 Top-Barbotage. Direction X2 a suesrR MOOEL • COMPl |—j—| J. O 5 6 0 -| i P i- c * I i rc ° i i o o 7 |I fVJ i 0 1 6 CO w\ o. j j 0.0 2.0 4.0 60 80 100 12.0 14.0 16.0 180 20 0 0.0 2.0 40 6.0 SO 10 0 12.0 14 0 16.0 18 0 20 0 Frequency (HZ] Frequency (HZ] Fig. A-18 Reactor Building VVER-440/213 PAKS, Fig. A-19 Reactor Building VVER-440/213 PAKS; Comparison of Response Spectra, Comparison of Response Spectra, Top-Barbotage, Direction X3 Axis "G"- Cran, Direction X1 B SUESTH MOOEL • COMPLEX MOOEL - — • n = i i 1 i A •* i •— — r 0.0 2.0 4.0 6 0 60 10.0 12 0 14.0 16.0 18.0 20 0 0 0 2 0 4 0 6.0 8.0 10.0 12.0 14.0 16.0 18 0 20 0 Frequency [HZ] Frequency |HZ] Fig. A-20 Reactor Building WER-440/213 PAKS. Fig. A-21 Reactor Building WER-440/213 PAKS, Comparison of Response Spectra Comparison of Response Spectra, Axis "G"- Cran, Direction X2 Axis "G"- Cran, Direction X3 221 - 17- 0 0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 0 0 2.0 4.0 6 0 8 0 10.0 12.0 14.0 16 0 18 0 20 0 Frequency (HZ) Frequency [HZ] Fig. B-1 Reactor Building WER-1000 MW KOZLODUY, Fig. B-2 Reactor Building WER-1000 MW KOZLODUY. Comparison of Response Spectra, Comparison of Response Spectra, Foundation Mat (Center), Level -6.5 m. Direction X1 Foundation Mat (Center), Uvel -6.5 m. Direction X2 SUBSTB MOOEL a • SUBSTR MOOEL MODEL 1 CM 1 • C 01JPLEX1^ODEL - ( 1 ll • il V1 j ' 1 • EV\[ A^ i I X' j -.. i J — -— 0 0 2.0 4 0 6.0 80 10.0 12.0 14.0 16.0 18.0 20 0 0.0 2.0 4 0 6 0 8 0 10.0 12.0 14.0 16.0 18.0 210 0 Frequency [HZ) Frequency (HZ] Fig. B-3 Reactor Building WER-1000 MW KOZLODUY, Fig. B-4 Reactor Building WER-1000 MW KOZLODUY. Comparison of Response Spectra, Foundation Mat (Center), Level -6.5 m. Direction X3 Comparison of Response Spectra, Foundation Mat (Outside), Level -6.5 m, Direction X3 • SUBSTR MOOEL • COUPLEX MOOEt 0 0 2.0 4.0 6.0 8.0 100 12.0 14.0 16.0 18 0 20 0 00 20 4 0 6 0 8 0 10 0 12.0 14.0 16 0 18 0 20 0 Frequency [HZ| Frequency |HZ| Fig. B-6 Reactor Building WER-1000 MW KOZLODUY, Fig. B-S Reactor Building WER-1000 MW KOZLODUY. Comparison of Response Spectra. Upper Level of Base Comparison of Response Spectra, Upper Level of Base Structure (Center), Level 13.2 m. Direction X2 Structure (Center). Level 132 m. Direction X1 222 - 18- O SUBSTR. MODEL . • COMPLEX MODEL 0 0 2 0 4 0 6 0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Frequency (HZ1 Frequency [HZ] Fig. B-7 Reaclor Building WER-1000 MW KOZLODUY, Fig. B-8 Reactor Building WER-1000 MW KOZLODUY, Comparison of Response Spectra, Upper Level of Base Comparison ot Response Spectra, Upper Level of Base Structure (Center), Level 13.2 m, Direction X3 Structure (Outside), Level 13.2 m, Direction X3 • SUBSTR MODEL • COMPLEX MODEL 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 0.0 2.0 4.0 6.0 60 10.0 12.0 14.0 160 18.0 20.0 Frequency {HZ] Frequency [HZ] Fig. B-9 Reactor Building WER-1000 MW KOZLODUY, Fig. B-10 Reactor Building WER-1000 MW KOZLODUY, Comparison of Response Spectra, Reactor Section (Center) Comparison of Response Spectra, Reactor Section (Center) Level 36.90 m, Oirection X1 Level 36.90 m, Direction X2 o — a SUBSTR MOOEL 0 SUBSTR MODEL — " • COMPLEX MODEL ' • COMPLEX MODEL • 0 0 2 0 40 6.0 80 10.0 12.0 14.0 16.0 18 0 20 0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 H.O 16.0 180 20.0 Frequency |HZ] Frequency |HZ] Fig. B-11 Reactor Building WER-1000 MW KOZLODUY, Fig. B-12 Reactor Building WER-1000 MW KOZLODUY, Comparison of Response Spectra, Reactor Section (Center) Comparison of Response Spectra, Reactor Section (Outside) Level 36.90 m. Direction X3 Level 36.90 m, Direction X3 223 - 19- 0.0 2.0 4.0 6 0 8.0 10.0 12.0 14 0 16 0 18 0 20 0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Frequency (HZ] Frequency (HZ] Fig. B-13 Reactor Building VVER-1000 MW KOZLODUY, Fig. B-14 Reactor Building VVER-1000 MW KOZLODUY, Comparison of Response Spectra, Containment (Outside), Comparison of Response Spectra, Containment (Outside), Level 46.80 m, Direction X1 Level 46.80 m. Direction X2 ._ O SUBSTR MOOEl — • COMPLEX MODEL 3 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 160 18.0 20 0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20 0 Frequency [HZ] Frequency [HZ] Fig. B-15 Reactor Building VVER-1000 MW KOZLODUY, Fig. B-16 Reactor Building WER-1000 MW KOZLODUY, Comparison of Response Spectra, Containment (Outside), Comparison of Response Spectra, Outer Building, Level 46.80 m. Direction X3 Level 41.40 m, Direction X1 Q SUBSTR. MOOEL . Q SUBSTR MODEL • COMPLEX MOOEl • COMPLEX MODEL ( u> ; T)9 : ° ii v - 1 1 i ! W 0 0 2.0 4 0 6 0 8 0 100 12.0 14 0 16 0 18 0 20 0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20 0 Ffsqueney [HZ] Frequency )HZ] Fig. B-17 Reactor Building WER-1000 MW KOZLODUY, Fig. B-18 Reactor Building WER-1000 MW KOZLODUY, Comparison of Response Spectra, Outer Building. Comparison of Response Spectra, Outer Building, Levei 41.40 m. Direction X2 Level 41.40 m. Direction X3 224 (]) XA9952657 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES DYNAMIC ANALYSIS OF WWER-1000 NUCLEAR POWER PLANTS Alejandro P. Asfura, Ph.D. EQE International San Francisco, California, USA Marin J. Jordanov EQE International Sofia, Bulgaria J. INTRODUCTION As part of the effort to assess the seismic vulnerability of nuclear power plants in Eastern Europe, a series of dynamic analyses have been carried out for several plants [1. 2, 3]. These analyses were performed using modern analysis techniques, current local seismic parameters, and local soil profiles. This paper presents a compilation of some of the seismic analyses performed for the WWER-1000 reactor buildings at the nuclear power plants of Belene and Kozloduy in Bulgaria, and Temelin in the Czech Republic. The reactor buildings at these three plants are practically identical and correspond to the standard building design for this type of reactors. The series of analyses performed for these buildings encompasses various soil profiles, seismic ground motions, and different soil-structure interaction (SSI) analysis techniques and modeling. The analysis of a common structure under different conditions gives the opportunity to assess the relative importance that each of the analysis elements has in the structural responses. The use of different SSI computer programs and foundation modeling was studied for Kozloduy, and the effects of different soil conditions and site-specific seismicity were studied by comparing the responses for the three plants. In-structure acceleration response spectra were selected as the structural responses for comparison purposes. 2. DESCRIPTION OF REACTOR BUILDING The reactor building for a typical 1000 MW WWER nuclear power plant consists of four distinctive structures, as shown in Figure 1. The base substructure starts from the foundation basemat and rises up to a second concrete basemat, which supports the reactor containment, Hie reactor internal structure, and a peripheral auxiliary building called the "outer building." The substructure is a three-story building, which houses the main control room and the auxiliary systems, and equipment including several large tanks. The building has a square shape of about 70 m by 70 m with orthogonally distributed walls. Most of the interior walls and slabs in the substructure are made of precast concrete panels serving as formwork for cast-in-place concrete. The containment is a post-stressed concrete shell with an 8-mm-thick steel liner. The posttensioncd cables arc anchored in a stiff ring girder at the junction of the cylinder and the spherical dome. The containment thickness is 1.2 m for the cylinder and 1.1m for the dome. The internal structure is a massive concrete structure supporting the reactor vessel, four horizontal steam generators, (he reactor coolant pumps, a prcssurizcr, accumulator tanks, and other auxiliary equipment. The core of the structure consists of a thick cylindrical shield wall around the vessel and two groups of pools and cavities containing the spent fuel. > y.'iOO-200/A) A\iiiin.l)ot/Nov-95 225 (2) PMTOPMQ OTiOErlEHUE 3B3P-iDDD "I *' i-'SSrS W///7///77///77/7//Ay////77/M Figure 1: Reactor Building Cross Section Tlic outer building houses miscellaneous equipment and the main steam piping going to the turbine building. This building is isolated from the containment shell by a seismic gap. Figure 2 shows the structural fixed-base stick model developed for the analyses described in this paper. Table 1 presents the dynamic modal characteristics of the first 15 modes of this fixed-base model. «;5(H)-2C)0/Al1AM«in.dix./Nov-95 226 (3) L CONTAINMENT MODEL 61.06 m t OUTER BUILDING 47.09 m MODEL J> 45.6 m 41.4 m L INTERNAL STRUCTU j MODEL r 36.90 m 33.6 m 29.00 m ki 28.8 m •i I 25.70 m 24.6 m 20.4 m 19.34 m 16.40 m A' „'" 16.8 m • I „'' 13.20 m 13.2,m J»*'. PA [P 12.0 m 6.3 m * Mass Point I Beam Element r'T -0.3 m I Rigid Unk z ; -*-X -7.0 m t! SUBSTRUCTURE MODEL Figure 2: Reactor Building Element Model W(W-2(«VA!1A.vniirt.d<>t7Nov-y.'i 227 (4) Table 1: Reactor Building Model Frequencies and Percent ol" Modal Mass Mode Modal Mass (%) Number Frequency (Hz) X Y Z Description 1 3.91 42.940 2.140 0.001 CB-x 2 3.91 2.145 42.682 0.003 CB-y 3 5.73 33.008 0.004 0.001 OB-x 4 5.80 0.007 35.220 0.004 OB-y 5 6.29 0.744 0.031 0.000 SS-x 6 8.85 0.089 0.008 0.000 IS-x 7 9.58 6.464 0.006 0.002 IS-x 8 9.62 0.011 4.378 0.069 IS-y 9 10.82 0.014 0.057 53.875 CB-z 10 12.01 0.001 6.184 0.133 IS-y 11 12.18 5.008 0.000 0.050 IS-x 12 12.60 0.273 0.003 0.002 IS-x 13 14.65 0.000 0.338 16.608 CB-z 14 14.84 1.328 0.003 0.117 CB-x 15 14.93 0.008 1.057 12.596 CB-z CB: Containment Building SS: Substructure OB: Outer Building IS: Internal Structure 3. DESCRIPTION OF FOUNDATION SOILS The foundation mat for each of the 1000 MW WWER reactor buildings is a 2.8-m-thick, approximately 74 m by 74 m concrete slab. Belene and Kozloduy reactor buildings are founded on a layered, relatively soft soil and embedded about 7 m below grade. Figure 3 shows the best estimate strain-compatible shear wave velocity profiles for Belene and Kozloduy. The reactor building at Temelih is founded on rock-like material with a shear wave velocity of approximately 2030 m/sec. Thus, it was determined that there was no need to include the soil effects in the dynamic analysis for this building. '.)MK).200/AlV\smirul<>e/N<>v-95 228 (5) -Belene Profile : — Koztoduy Profile -so - I 200 300 400 Shear Wave Velocity (Vs) (m/sec) Figure 3: Best Estimate Strain-compatible Shear Wave Velocity Profile 4. DESCRIPTION OF INPUT MOTION For the three plants, the seismic motions used were consistent with the seismic hazard for the site. Figures 4 and 5 show the horizontal and vertical acceleration response spectra at 5% damping for the three plants. For Belene and Kozioduy, the hazard was dominated by the Vrancea source [1,2], and the motioas were anchored to an equivalent peak ground acceleration of 0.2g. Both seismic motioas correspond approximately to a return period of 10^ years. For Belene and Kozioduy, the specified vertical spectra were defined as half of the horizontal spectra. The seismicity at the Temeh'n site was dominated by the local conditions [4]. The intensity and epiccntral distances selected for the site resulted in a design peak ground acceleration of less than O.lg. However, for the Temelfn study presented in this paper, the spectra of an actual earthquake considered to be representative for the site [4] anchored to a horizontal maximum peak ground acceleration of O.lg were used to develop a single artificial earthquake for the analysis. The selected earthquake has its main frequency content coinciding with the main frequencies of the fixed-base reactor building. VS(X)-200/AI'A.'smirt.ilix.7Nt>v-<)S 229 (6) -Belene Spectrum Kododuy Spectrum Temefin Spectrum — 0.4 C .2 I o C.3 Frequency(Hi) Figure 4: Comparison of Horizontal Input Response Spectra, 5% Damping Selene Spectrum Koztoduy Spectrum Temelm Spectrum ass- I s I Figure 5: Comparison oi" Vertical Input Response Spectra, 5% Damping M500-200/Al'A.smirt.ilix;/Nov-9.'i 230 (7) 5. ANALYSIS SSI analyses were performed for Belene and Kozloduy to capture the effects of the foundation soil in the structural responses. For Kozloduy, a series of sensitivity studies were performed to assess (he impact thai SSI modeling parameters and calculation techniques have in the structural response of these plants. Since this plant is founded on rock, a fixed-base time-history analysis was performed for Temelfn. 5.7 Impedance and Scattering Functions Calculation To assess the effect that different SSI methods have in the calculation of the impedance and scattering functions, three different codes were used for Kozloduy. The three codes were SASSI (developed by Professor Lysmer at U.C. Berkeley), CLASSI (developed by Professors Wong and Luco at the University o\~ Southern California), and SUPELM (developed by Professor Kausel at MIT). The versions of SASSI and SUPELM used allow for the consideration of the embedment effects. The version ol" CLASSI used considers only surface-founded conditions. For the calculation of scattering functions, vertically propagating seismic waves were assumed. Figures 6 to 8 show the horizontal, vertical, and rocking impedance functions, respectively. Figure 9 shows the horizontal term of the scattering function. For the SASSI and SUPELM cases, it was considered that the foundation was embedded and perfectly bonded to the soil. For the CLASSI case, the foundation was assumed at a free surface at the bottom of the foundation level. SASSI SUPELM CLASS] Figure 6: CLASSI/SASSI/SUPELM Comparison Horizontal Impedance 9500.20CyAPAsmirt.iloc/Nov-95 231 (8) x io7 Imaginary .4 -.2 -.4 .0 S.fl 10.0 1S.0 20.0 .0 5.0 10.0 15.0 20.0 SASSI SUPELM CLASSI Figure 7: CLASSUSASSySUPELM Comparison Vertical Impedance x io" Imaginary -.1 -.2 -.3 .0 5.0 10-0 1S.0 20.0 .0 S.O 10.0 15.0 20.0 Fr*qu«ncy Ltssni SA.SS1 SUPHLM CI-ASSI Figure K: CLASS I/S ASS 1/SUPELM Comparison Rocking Impedance V5(K)-2OO/A]>Asii)in.d<)c/Niw- 232 (9) Real Imaginary 2.0 1.0 -1.0 • -2.C -.6 .0 5.0 10.0 15.0 20.0 .0 5.0 10.0 15.0 20.0 Fr«qu«ncy SASSI SUPELM CLASS; Figure 9: CLASSI/SASSI/SUPELM Comparison Horizontal Scattering From the impedance figures, it can be concluded that the three programs give similar results in the frequency range of interest. For the horizontal term, SASSI deviates for frequencies beyond those represented by the discretization of the soil layers (about 15 Hz). Also, these figures show that for the calculation of the impedance functions, the effects of embedment are negligible. From Figure 9, it can be concluded that SASSI and SUPELM give very similar results for the horizontal scattering function. The same is true for all other scattering components. Since it was assumed that the seismic waves propagate vertically, the translational scattering functions calculated by CLASSI are constant and equal to 1.0. The comparison of the impedance and scattering functions also shows that possible differences between a embedded and surface founded case for Kozloduy will be mainly due to the deconvolution of the motion and not to the stiffening effect of the lateral soil. Due to the similarities between Kozloduy and Belene soils, this is also true for Belene. 5.2 Effects of Building Wall-Soil Banding To study the effect that the lateral soil, excavated and then replaced by backfill during coastruction. can have on the dynamic response of the Belene and Kozloduy reactor buildings, two extreme cases were analyzed for the Kozloduy foundation model. First, the impedance and scattering functions were calculated with program SUPELM considering perfect bonding between the embedded part of the structure and the soil as it is shown in Figure 10. Then, these functions were calculated, also with SUPELM, considering the embedded part of the structure was completely unbonded from the soil as it is shown in Figure 11. <}MH>-2 233 (10) 2R-2.1M.87 tt ; '.,-.•:BACKFILL'. '//{///////////{///////^^^ SANDY Figure 10: SUPELM Analysis Foundation Model (Bonded Wall) 2R-a»136.»7 ft ./ .. /. •v-'--?:/.:-^:-;.r-J •:^-'/-;•••;'~>^'••*.' Figurc 11: SUPELM Analysis Foundation Model (Unbonded Wall) Figures 12 to 14 show the horizontal, vertical, and rocking impedance functions, respectively, lor llic bonded and unbonded cases. Figure 15 shows the horizontal term ol" the scattering function for these two cases. Some differences exist, but mainly beyond the frequency range of the soil-structure responses. These differences will have only a minor impact in the final dynamic response of (he reactor building. Thus, for any practical effect, the SSI analyses for Ko/.loduy and Bclcnc can be done considering perfect bonding. 2 234 (11) X 10' Imaginary .0 5.0 10.0 15.0 20.0 2S.0 30.0 ' .0 S.O 10.0 15.0 20.0 25.0 30.0 frequency Frequency Bonded Case Unbonded Case Figure 12: SUPELM Bonded/Unbonded Case Comparison of Horizontal Impedance x 10' Real /A \ Ji .0 \ V -.1 -.2 .0 S.O 10.0 15.0 29.0 2S.0 30.0 .0 5.0 10.fl 15.0 20.0 25.0 30.0 Ronded Caw Unbonded Case Figure 13: SUPELM Bonded/Unbonded Case Comparison of Vertical Impedance 'JSOO-SOO/Al'Axmirt.doc/Nov-VS 235 (12) x IC" Imaginary .£ -.2 -.4 ' .0 5.0 10.0 15.0 20.0 25.0 30.0 .0 5.0 10.0 15.0 20.0 25.0 30.0 Frequency Frequency Legend Bonded Case Unbonded Case Figure 14: SUPELM Bonded/Unbonded Case Comparison of Rocking Impedance Real Imaginary 2.0 1.0 -.2 • -1.0 -.4 -.6 -2.0 .0 6.0 10.0 15.0 20.0 25.0 30.0 .0 5.0 10.0 15.0 20.0 25.0 30.0 Froquoncy l,egend Bonded Cast Unbonded Case Figure 15: SUPELM Bonded/Unbonded Case Comparison of Horizontal Scattering 236 (13) 5.3 Effects of Embedment The acceleration in-slruclure response spectra for Kozloduy were calculated for the embedded, perfectly- bonded case (SUPELM) and for the surface-founded case (CLASSI) to quantify the differences in the dynamic responses in the reactor buiiding. Figures 16 to 19 show acceleration in-structure response spectra at 5c/<- damping for selected locations at the containment, internal, outer, and substructure portions of the reactor building, respectively. These figures show that the differences between the embedded and the surface-founded cases are in general of the order of 10% to 15% at some particular frequency ranges. The results for the surface-founded case are, as expected, higher than the results for the embedded case. North-South Direction (X) East-West Direction Ci) 1.2 1.2 1.0 £ 1-0 iI If f r \: \i 1 I ~—i ) .2 J , 10" 10' 10 10" 10 Fr«quancy Vortical Direction 1Z) $ A Embedded Surface Founded 4 ly Notes: 3 RLE Level t-, n Embedded Bonded Wall 2 /' 5% Spectral Damping Acceleration in g's J 1 J n 10* 10 Frsqoaacy (Rxl Figure 16: Comparison of Surface Founded to Embedded Structural Model, Top of the Containment Rcspoasc 237 (14) North-South Direction .8 \ \ I 1 I X s •? 10" •}? 10" 10* 10' rrequency {Hz) Frequency (Hz) 101 Vertical Direction (Z) 3.0 Legend Embedded Surface Founded Notes: RLE Level Embedded Bonded Wall 5% Spectral Damping Accelerations in g's 10' 10 Frequency (Bx) Figure 17: Comparison of Surface Founded to Embedded Structural Model, Top of the Internal Structure Response forth-South Oir«ction 1 j .2 f L 16 1C 10* rre 10* Vertical Direction Notes: RLE Level Embedded Bonded Wall SK Spectral Damping Accelerations In g's 10* Figure 18: Comparison of Surface Founded to Embedded Structural Model, Top of the Outer Building Rcspoase <;5(X)-2(KVAI>A.vinirt. 238 (15) North-South Direction (X) East-West Direction .6 ..•ft fl j 1 \ .« \ ) .2 mm .2 I 10 10 10 10" 10' (Hi) Frequency (Hz) X 10 Vertical Direction 2.5 1 Embedded • •4 Surface Founded I 2.0 Notes: I RLE Level Embedded Bonded Wall 5% Spectral Damping < 1.0 1/ Accelerations in g's 10" 10' 10 rr«qu«ncy (S31 Figure 19: Comparison of Surface Founded to Embedded Structural Model, Top of the Substructure Response 5.4 Effects of Site-specific Conditions To quantify the effects thai site-specific conditions, soil, and scismicily have in the dynamic response of these three "standard" reactor buildings, the results of the dynamic analyses for them were compared. For Belcnc and Kozloduy, the SSI analyses had been performed assuming embedment and perfect wall-soil bonding. For Temelfn. a lixed-base analysis was performed. Acceleration in-structure response spectra at 2% damping are compared in Figures 20 to 23 for the containment, internal, outer, and substructure portions of the reactor building, respectively. The in-structurc response spectra compared in Figures 20 to 23 correspond to the envelopes of the two horizontal directions at the selected locations. The spectra for Temelin were broadened by \5% to cover structural and analysis uncertainties. Belene's spectra were broadened 25% to also cover soil uncertainties. Kozloduy spectra were broadened by 15% to cover only structural and analysis uncertainties, but three soil cases were considered to cover the soil uncertainties. This difference in the treatment of uncertainties docs not prevent a meaningful general comparison between the three sets of results. The comparison in Figures 20 to 23 shows that the acceleration in-structurc response spectra lor Bclcnc and Kozloduy arc similar. This similarity results from their seismic input in the frequency range of interest being comparable, as well as (on average) from the foundation conditions. The in-structurc response spectra for Tcmeh'n arc very different from those for Bclcnc and Kozloduy. As expected, the Uirec plants' spectral peaks occur at different frequencies, and even though the seismic hazard al Tcmclfn is lower than the seismic hazard al Bclcnc or Kozloduy, the in-struclurc response spectra at that plant arc much higher than those at Bclcne or Kozloduy. This difference is mainly due to the dissipation of energy 239 (16) through the soil (radiation damping) for (lie Belene and Kozloduy reactor buildings, which reduces their structural responses, making them much lower than the Temelin responses. This large difference in the in-structure response spectra results in a large difference between the seismic demands for the equipment at the Temelin and the Bulgarian 1000-WWER reactor buildings. 6 51 Kaztoduy Spectrum g Temehn Spectrvm C £ 3- o a 2 / "s \ 1 \ ^ V \ 0.1 10 Fn»q«ncy(Hx) Figure 20: Comparison of Horizontal Floor Spectra; Containment, Elevation 46.8 m - Belene Spectrum Koztoduy Spectrum •Temelrt Spectrum Figure 21: Comparison of Horizontal Floor Spectra; Internal, Elevation 25.7 95OO-2OO/APAsmtn.dix;/N 240 (17) Betane Spectrum Kozloduy Spectrum Tom»in Spectrum c o S 2 Fmq«ncy (Hz) Figure 22: Comparison of Horizontal Floor Spectra; Outer, Elevation 33.6 m Fraqancy (Hz) Figure 23: Comparison of Horizontal Floor Spectra; Substructure, Elevation 13.2 m 6. CONCLUSIONS The resulls presented in this paper demonstrate the obvious importance of properly considering the sitc- spccific conditions, soil, and input in determining the seismic demand on "standard" plants. For the three studied cases, the clTect of the soil and its proper modeling became the most important parameters in ySOO-200/AI>Asmiit.d 241 (18) determining the structural seismic demands, overcoming the higher seismic hazard at the Belene and Kozloduy plants. 7. REFERENCES 1. EQE Engineering/Westinghouse/Geomatrix. March 1990. "Seismic Review of the Belene Construction Project (Units 1 & 2)." 2. EQE International. October 1994. "Structural Response of Kozloduy 1000 MW WWER." 3. Asfura, A. P., J. J. Johnson, and M. J. Jordanov. August 1995. "Dynamic Analysis of Three 1000 MW WWER Reactors in Eastern Europe." 13th SMIRT Conference. Porto Alegre, Brazil. 4. David Consulting. Engineering and Design. "Seismic Hazard Analysis. Ground Response Spectra. NPPTemelfn, Probabilistic Safety Assessment for Seismic Events." 242 XA9952658 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES IN-STRUCTURE SPECTRA GENERATION FOR KOZLODUY NPP, BULGARIA Marin Kostov Central Laboratory for Seismic Mechanics and Earthquake Engineering, Bulgarian Academy of Sciences, Sofia ABSTRACT. This paper is presenting most of the results achieve in the last 4 years during the in-structure response spectra generation of NPP Kozloduy, Bulgaria. A variety of models and techniques have been used for investigation of the soil-structure interaction problems. The experience collected and the problems met are discussed hereafter. 1. Introduction In 1993 a new review level earthquake (RLE) for the site of Kozloduy NPP was determined. The maximum acceleration at the site has been changed to 0.2g , respectively a new free field response spectrum has been developed. For the qualification of equipment and systems the in-structure response spectra have been generated for all six units and for the spent fuel storage building. The methods and the techniques used have been a matter of development and improvements trough the last three years. In the present paper some of the experience collected in this work is presented. 2. Free Field Seismic Characteristics and Local Geology The horizontal free field spectrum is presented in fig.l. The respective vertical excitation is assumed to be 1/2 of the horizontal one. In the site conformation report it has been recommended also the separate consideration of local seismic excitation (magnitude 4.5 at distance 7km) The generalized local geology characteristics are presented in table 1. The site of Kozloduy NPP is characterised by medium to weak soils, mainly loess, sand and gravel. 3. In-structure Spectra for the WER 440-230 structures. The units 1 to 4 of Kozloduy NPP are of type VVER 440-230. Those units are constructed as two couples of tween units with a common turbine hall going trough all four units. Each unit consists of the following parts: the reactor building, the intermediate building and the auxiliary building. The lower part of the reactor building (the compartment) is a rigid structure characterized by massive concrete walls up to an elevation of 10.4m. From that level up to the roof there is a frame structure. The turbine hall is formed by flexible frames both in longitudinal and transversal direction. The connection between the two buildings is an intermediate building. The main part of the critical equipment for which the floor spectra should be computed is located in the reactor building and the intermediate building. The in-structure spectra for the first two units are computed by plane models both in transversal and longitudinal directions (fig.2). 243 Table 1, Soil Characteristics, Kozloduy NPP site LAYER FROM-.. DENSITY POISSON'S S-WAVE P-WAVE TYPE THICK- TO. . . RATIO VELOCITY VELOCITY OF NESS Vs Vp SOIL m m t/m3 m/s m/s 3.0 0.0- 3.0 1.60 0.42 170 470 Loess 4.0 3.0- 7.0 1.60 0.44 175 540 Sandy loess 6.5 7.0- 13.5 1.80 0.41 450 1180 Clayey loess 5.0 13.5- 18.5 2.00 0.45 500 1600 Gravelly sand 3.0 18.5- 21.5 2.00 0.45 500 1600 Compact clay 9.5 21.5- 31.0 2.12 0.45 500 1600 Sand-fine clayey 11.0 31.0- 42.0 2.10 0.47 430 1700 Sandy clay 42.6 42.0- 84.6 1.92 0.45 520 1700 Sand-fine clayey 19.4 84.6-104.0 1.98 0.44 550 1700 Sand-fine clayey 29.0 104.0-133.0 2.01 0.46 450 1600 Sandy clay 18.0 133.0-151.0 1.98 0.44 540 1600 Marly clay 24.0 151.0-175.0 2.00 0.44 580 1750 Marly clay 29.0 175.0-204.0 1.96 0.43 530 1470 Marly clay 20.0 204.0-224.0 1.98 0.37 630 1470 Marly clay 21.0 224.0-245.0 2.00 0.40 680 1700 Clayey marl 20.0 245.0-265.0 1.96 0.40 705 1760 Clayey marl 265.0- 2.00 0.40 >705 >1760 Clayey marl Computer code FLUSH and PLUSH are used for that purpose. Correction is made for accounting the torsional response [1]. The in-structure spectra for unit 3 and 4 are developed on the base of complete 3D FE model (fig.3). Soil-structure interaction is modelled by spring and dashpots. The dynamic analysis is performed by computer code STARDYNE. A comparison between results from unit 1/2 and 3/4 is presented in table 2 [2]. Discussion: The FLUSH /PLUSH/ code is using transmitting boundaries in horizontal direction to account for the radiation damping. FLUSH /PLUSH/ incorporates also the deconvolution of the free field motion. For the dynamic analysis with STARDYNE the deconvolution should be done separately, so the input seismic motion is applied to the foundation. The radiation damping is accounted by an equivalent modal damping. The modal damping is computed using the method of composite damping, i.e. the damping is compiled by weighting the material damping of each structure member by the corresponding modal strain energy. After the composite modal damping is determined, it is kept within the limits, prescribed by KTA 2201. As a matter of fact there is almost no need of cutting the damping because the composite damping is usually under those limits. The exception is the case of extremely weak soils ( 0.5 of the best estimate soil properties). That fact is showing that the soil structure interaction is not of primary importance for that type of building. It is important to stress that the overall dynamic behaviour of those structures is heavily influenced by interaction between structures with very different stiffness. Most of the attached to the reactor building structures have a rotational response around the centre of stiffness of the combined structure. Those 244 effects could be represented and analysed only by quite complicated spatial models. The 2D and especially the stick models could produce in that case a misleading result. Table 2 Maximum Spectral Acceleration Values (5% damping), Sa, and Corresponding Frequencies, f, for Unit 1/2 and Unit 3/4. (extracted from broadened and smoothed response spectra) Level Unit 1/2 Unit 3/4 SaH f SaV f SaT f SaL f SaV f m g Hz g Hz g Hz g Hz g Hz -2.2 1.15 1.9- 0.62 3.75- 3.0 5.25 -2.8 0.9 3 .6- 0.9 2 .6- 0.48 3 . 1- 5 .1 3 .6 5 . 1 2.7 1. 18 2.0- 0.66 3.0- 2.8 6.0 2.3 1.0 3 .1- 1.0 2 .6- 0.49 3 . 1- 5 .1 3 .6 5 . 1 10.5 1.13 10- 0.59 12- 12 15 10. 1 0.99 3 .1- 0.98 2 .6- 0.52 4 . 6- 5 .1 3 .6 8 . 1 6.3 1.28 10- 1.96 10- 12 12 6.0 0.91 3 .6- 1.02 3 .1- 0.90 3 . 6- 5 .6 5 .1 5 . 6 14.5 1.58 10- 2.02 10- 12 12 13.5 1.06 3 • 1 ~" 1.22 2 . 6- 0.70 6 . 1- 5 .1 3 .6 9 . 1 13.5 1.45 o . 1- 1.22 2 .6- 0.54 3 . 1- J 5 .1 3 .6 5 . 1 21.8 1.73 10- 2.05 10- 12 12 18.7 1.32 4 . 1- 2.38 3 . 1- 0.45 3 . 1 6.1 4.1 5 . 1 4. In-structure Spectra for the Spent Fuel Storage Building /SFSB/ The spectra for the SFSB are developed using a 3D FE model, the general view of that model is presented in fig.4. This building is similar to the main building of Unit 1/4. From the base slab up to the level 7.3m there are massive and stiff concrete walls, forming the pools for spent fuel storage. The upper structure is flexible RC frame building. The soil-structure interaction is modelled by springs and dashpots. A typical floor spectra from this building is presented in fig.5. the fundamental frequency is about 2Hz. Discussion: As in the case of Unit 1/4 investigation, the free field motion has been deconvoluted first to the foundation base. The radiation damping is modelled also by means of equivalent modal damping. The later is computed as composite damping. Usually for that structure the composite damping is very low and limitation of the modal damping is not needed 245 5. In-structure Spectra for Unit 5/6, WER-1000 The VVER 1000 reactor building is modelled by detailed 3D FE model. Within the Benchmark Program for Seismic Analyses of VVER-Type NPP's of IAEA investigation of that building are performed also using simple stick models. The comparison between results from 3D FE and the stick model computations are showing very good fit. The reactor building of VVER-1000 is a stiff structure consisting of four main parts: foundation block, containment shell, internal structure, auxiliary building. The joining element of all of them is the massive RC plate at level 13.2m. Practically all parts are integrated in one rigid block. The internal structure is supporting the reactor and the primary circuit. Because of the symmetry only 1/4 of primary circuit is modelled as flexible elements. The other parts are considered only as masses. The complete FE model is shown in fig.6 [3]. The soil is represented by springs and dashpots. the equivalent stiffness and damping are determined by an impedance analyses. The equivalent radiation damping is estimated to be 70% for vertical vibration, in horizontal direction it is respectively 60% of the vertical, for rocking 50% and for torsion - 30% of the damping for vertical motion. The fundamental frequency of that building fixed at base is about 5Hz. The soil stiffness i> changing that frequency to 2Hz. Because of the relatively weak soils the fundamental mode> of that rigid building are governed by the soil-structure interaction. Typical in-structure spectra for that structure are shown in fig.7. As it could be seen onl> a few modes are contributing to the spectral shape. For the aims of a PSA analyses a set of probabilistic in-structure spectra have been generated also [4]. The seismic input has been prepared in terms of uniform hazard free field spectra. The later have been represented by means of a set of time-histories, bearing the same response statistics. A comprehensive multiple time-history analyses based on Latin Hypercube Experimental Design procedure have been applied. Statistics of the in-structure spectra arc derived for several levels of hazard. Typical probabilistic in-structure spectra for Unit 5/6 art- shown in fig.8. Discussion: It is interesting to compare the deterministic and the probabilistic response spectra (10--» annual probability of exceedance). Generally there is good fit in spite of the different input excitation. Comparing the mean plus one standard deviation spectra with the deterministic spectra it could be seen that there are cases were the probabilistic values are overshooting the deterministic one. Because of the conservatism involved in the deterministic spectra computation one should expect actually the revers situation, i.e. the deterministic values being grater than the probabilistic one. 7. Conclusions The results achieved for in-structure generation of VVER-type reactors at the site of Kozloduy are leading to the following conclusions: l.The Kozloduy NPP site is characterised by deep alluvial deposits with relatively low strength. Having in mind that the reactor structures are relatively stiff structures an intensive soil-structure interaction effects are expected. 2.For the structures of the 1000 MW reactors the soil structure interaction is determining the fundamental dynamic behaviour. Most of the in-stricture spectra are influenced by the rocking motion of the rigid structure due to soil compliance. The structure response is relatively insensible to the free field spectrum .shape. There is relatively high radiation damping in all direction of vibration. 246 3.The in-structure spectra of the 440 MW units are quite different in character. Obviously the soil-structure interaction effects are not so important for the framed structures. The response is compiled by participation of many mode shapes. This makes these structures sensible to the shape of the free field spectrum but also to the shape of the time history. It is also important to stress that these structures are very sensitive to vertical vibration what is not the case for the 1000 MW reactor structures. 4.There is an important difference in the response of the 440 MW reactors structures and the 1000 MW units. The response of the small units is governed mainly by the difference in the stiffness between the reactor building and the attached auxiliary buildings and turbine hall. All attached buildings are practically rotating around the stiff reactor structure, because of this there are heavy problems of structural interaction which are highly unfavourable. The response of the 1000 MW unit is quit homogeneous, governed by small number of modes, mainly due to soil-structure interaction. The interaction effects in that case are in favour of the structural safety - they are contributing to mild the dynamic response of the building. 8. Literature: 1. Kostov. M. et al., 1993, Floor Response Spectra Generation for Unit 1/2, Kozloduv NPP. Report EGP, Sofia. 2. Rostov, M. et al., 1994, Floor Response Spectra Generation for Unit 3/4, Kozloduy NPP. Report EGP, Sofia. 3. Kostov, M. et al., 1994, Floor Response Spectra Generation for Unit 5/6, Kozloduv NPP. Riskengineering LTD. 4. Kostov, M. et al, 1994, Probabilistic Safety Analysis of Unit 5/6, Kozloduy NPP, Level 1. Report RE/LTD-10.3., Voll-4, Riskengineering LTD, Sofia. UNIFORM HAZARD SPZCTRA. S'J. DAKPtNC DESIGN RESPONSE SPECTRUM ANWUAl PROBABILITY OF EXCEEBANCE 10" 0.50 0.00 Fig.1. Acceleration response spectra at free field - RLE and probabilistic Sa :' §j§§ S r Bmm = 9 = = =: =iai vm "• " •• Fig.2. Plane, 2D FE Models, Unit 1/2 - transversal and longitudinal model 247 Fig.3. 3D FE Model, Unit 3/4 Fig.4. 3D FE Model, Spent Fuel Storage Building fREOUCNCr /H,/ FREOUCNCr /Mi/ 'RtOOCMOr /Mi/ Fig.5. Floor Response Spectra, SFSB, Level 3.55, n.p.904 248 Fig.6. 3D FE Model, Unit 5/6, 1000 MW ACCELERATION RESPONSE SPECTRA ACCELERATION RESPONSI SPECTRA ACCELERATION RESPONSE SPECTRA ENVELOPES ENVELOPES ENVELOPES OAKPINC: 0.02: O OS; 0.07; D.lO DAMPING: 0.03: O.Ot; 6.07; 0.10 DAUPINC: O.OZ; Q.01; 0.07; 0.10 K0D.U. POINT 2OZB NODAL PCTNT WZ8 NODAL POINT ZOZ& COMPONENT L COMPONENT T COKPONENT V ntSt)tJEHCr /Hi/ niEQUENCy /Hi/ Fig.7. Deterministic Floor Response Spectra, Unit 5/6, centre of dome,n.p.2028 Fig.8. Probabilistic Floor Response Spectra, Unit 5/6, centre of dome, n.p.2028 KEXT (eft BLANK 249 XA9952659 Applications of Seismic Damage Hazard Analysis for the Qualification of Existing Nuclear and Offshore Facilities by P. Bazzurro(1), G. M. Manfredini^2) and I. Diaz Molina*3) ABSTRACT The Seismic Damage Hazard Analysis (SDHA) is a methodology which couples conventional Seismic Hazard Analysis (SHA) and non-linear response analysis to seismic loadings. This is a powerful tool in the retrofit process: SDHA permits the direct computation of the probability of occurrence of damage and, eventually, collapse of existing and upgraded structural systems. The SDHA methodology is a significative step towards a better understanding and quantification of structural seismic risk. SDHA incorporates and explicitly accounts for seismic load variability, seismic damage potential variability and structural resistance uncertainty. In addition, SDHA makes available a sound strategy to perform non-linear dynamic analyses. A limited number of non-linear dynamic analyses is sufficient to obtain estimates of damage and its probability of occurrence. The basic concepts of the SDHA methodology are briefly reviewed. Illustrative examples are presented, regarding a power house structure, a tubular structure and seabed slope stability problem. 1.0 INTRODUCTION The goal of this contribution is to illustrate a methodology to predict the level of seismic response of realistic structures that is hazard-consistent. This means that the response meets specified probabilistic safety goals expressed both in terms of serviceability and ultimate-capacity levels. The behavior of such systems to seismic excitation is usually very complex (multi degree-of-freedom, MDOF) and often involves nonlinearities. The seismic evaluation of such structures cannot involve simply the application of a site-specific Uniform Hazard Spectrum (UHS), i.e., the single-degree-of- freedom (SDOF) linear spectral ordinates for a range of frequencies. However, this UHS, coupled with structure-specific fragility curves, is the current prevailing practice for seismic design and re-assessment of structures. The implicit assumption in such conventional approach is that the response of the MDOF structure has the same hazard of the elastic, SDOF uniform hazard spectrum. This approach may prove to be accurate provided that the linear elastic structural response is dominated by the first mode contribution. For realistic cases when the seismic response goes well beyond the maximum elastic capacity, the accuracy of the conventional approach is questionable, unknown, and surely variable on a case-by-case basis. 0) Senior Project Consultatnt, D'Appolonia S.p.A., Genoa, Italy. (2) Senior Partner, D'Appolonia S.p.A., Genoa, Italy. (3) Director, D'Appolonia, Cordoba, Argentina. 251 The earthquake engineering community in many areas (hazardous facilities, offshore structures, bridges, buildings, etc.) has recently recognized the need of going towards more advanced and modern regulatory assessment criteria based explicitly on probabilistic risk goals and on non-linear, closer-to-failure structural dynamic behavior. Both are necessary to achieve the most effective regulation of safety. Examples include the IPEEE (US NRC, 1991) and the draft DG1032 procedures for NRC; the "1020" natural hazard assessment document (US DOE, 1993) for DOE facilities; the API RP2A guidelines for offshore structures (API, 1993) which include a new chapter for seismic reassessment; the new AASHTO LRFD bridge design criteria; the Japanese design guidelines for RC buildings (PRESSS, 1992). The proposed procedure, called Seismic Damage Hazard Analysis (SDHA) is fundamentally probabilistic, includes the seismology relevant aspects, and is based on structure-specific non-linear dynamic analyses. This type of analyses, given the exponential growth in computational efficiency and the availability of adequate software, can be considered today an available office tool. Published references on SDHA include Bazzurro and Cornell (1992; 1994a and b), Bazzurro et al (1994a and b), Bazzurro et al. (1995). This work is organized as follows. In Section 2.0 the procedure is summarized with more emphasis on clarity of exposure and illustration of ultimate goals rather than mathematical details. Chapter 3.0 presents a made-up example included to show how all the steps of the procedure can be put together to obtain the desired damage/failure probabilities. Real case applications where the methodology has been successfully used are included in Chapter 4.0. Chapter 5.0 serves as conclusion. 2.0 SEISMIC DAMAGE HAZARD ANALYSIS The innovative methodology for the evaluation of the probability of exceeding specified levels of post-elastic seismic damage in realistic, multi-degree-of-freedom (MDOF) structures is briefly summarized in the following (excerpted from Bazzurro et al., 1994a). The proposed SDHA is implemented by combining conventional seismic hazard analysis (SHA) for the site and non-linear structural response to ground motions from different events (e.g., magnitude, M, and source-to-site distance, R, pairs). The driving idea is to make use of the well established probabilistic approach (Cornell, 1968), not for computing the probability of exceedance of ground- motion intensity levels at the site but for assessing the probability of exceedance of a damage state in the structure induced by such seismic loads. The damage measures monitored during the non-linear dynamic analyses are the most appropriate to gauge the performance of the particular structure being investigated (for example, the peak damage suffered in a specified group of columns measured in terms of, say, the maximum of their ductility ratios). The most appropriate damage measures (e.g., ductility ratio, normalized hysteretic energy, etc.) and the location of the damage within the structure vary on a case-by-case basis. Computationally, the calculation of the annual probability of exceedance of a desired level, x, of a post-elastic damage, DM, at a location / (e.g., the specified group of columns) in the structure involves the repetitive calculation of the following conditional probability (Bazzurro and Cornell, 1994a): 252 P[DM,l>x\m,r]=p[Sa{f.€)>Saoujmx{f,S)\'».r] (D where: (1) m and r are realizations of the random variables i\4 and R of the ground motion; (2) Sa _ (/,£) is the pseudo-spectral acceleration (PSA) of the ground motion which induces the specified damage level, DM = x, at location / in the structure. This PSA is associated with a representative frequency, /, and percentage of critical damping, £,. Note that for realistic MDOF systems multiple choices of/ are possible. Directions on "best" choices of a repre- sentative single frequency of vibration can be found in Iwan (1980) and Kennedy et al. (1984). In any case, experience suggest that this choice is not critical: the objective is to measure the "strength" of the ground motion in the frequency range swept by the dominant modes of vibration of the structure during the ground shaking. If the first mode is dominant in the response, then an obvious choice could be the fundamental frequency. If more than one mode is important, an appropriate intermediate frequency may be selected and adopted as the reference frequency/. s tne 0) Sa(f For conciseness of notation, in the following So(f,E) will be referred to as simply Sa and SODM l=x(f, i) as SDM. Both are random variables. It is important to recognize that, for a given structure of deterministic resistance, SDM is not a constant but is a characteristic of the ground-motion accelerogram. Inversely, different accelerograms inducing the same damage DM = x at the same location / in the structure have different values of PSA at the same reference frequency/ Previous empirical studies (Sewell, 1988; variable, SDM shows (1) a very moderate record-to-record variability (coefficients of variation of, say, 0.3 or less) relative to the intrinsic variability of Sa (coefficients of variation, COV's, of 0.6 - 0.8 depending on the attenuation law and on the frequency) and (2) insensitivity in the mean to M and R. In light of the preceding considerations, the seismic damage risk can be computed very simply (Bazzurro and Cornell, 1994a): (1) by replacing SDKi in Equation 1 by its unconditional mean, SDM(=E[SDM\m,r] = E\SDM\), and (2) by inflating the original variability intrinsic in Sa (Seismic Load Variability) in order to account for the comparatively small additional variability in SDM (Seismic Damage Potential Variability). For example, if the maximum ductility ratio, //, experienced in a given group of columns during the ground shaking is chosen as the damage measure DM, and if p is monitored for several accelerograms, then a relationship between fj and SDM (here Sp) can be easily obtained. To do so one may select a suite of real ground-motion accelerograms typical of those recorded in the region of the site and perform a series of non-linear dynamic analyses of the structure being investigated. 253 More precisely, each ground-motion record (or all the three components of each event for 3D structures) has to be appropriately scaled to obtain a new record capable of inducing at location / in the structure exactly the specified damage levei DM =x. The spectral ordinate (at the pre-selected reference frequency, /, and damping ratio, E,) of each scaled record represents a realization of the random variable SDM (i.e., SaDMI_x(f<£))- The value of 5^./ can be obtained by simply averaging all the SDM values of the scaled records. It is important to note that, given the relative small variability observed in SDM, the approximation introduced in the calculation of SD^ by the use of only a few ground-motion accelerograms can be usually reduced to negligible values (again relative to the uncertainty of Sa) with only a sample size of 5-7 records (Bazzurro and Cornell, 1994a). An example of how a typical relationship between SDftl and DM (here //) may look like for a structure behaving dynamically (1) as an elasto-plastic (EP) system, and (2) as a stiffness- and strength-degrading system is displayed in Figure 1. The form of this relationship for such systems has been consistently observed in previous empirical studies (Sewell, 1988; Inoue, 1990; Bazzurro and Cornell, 1994a). Once that the relationship between the structural damage and the ground-motion spectral acceleration has been established, a seismic hazard curve such as the curve shown in Figure 2 may be produced by performing a seismic hazard analysis for the site. This analysis requires the use of an inflated variability. Referring to previously reported COV values of So and SDM, it is clear that the total variability increases by only a small amount (e.g., being a total variability equal to 62 + 0.32 =0.67, the increase is in the order of only 10%). Notice that the seismic hazard curve in Figure 2 was obtained for a reference frequency, /, of 2.0 Hertz and a damping, E,, of 5% of critical. Seismic damage hazard curves yielding the annual probability of exceeding various levels of u. ranging from the beginning of post-elastic response (p=\) to very severe damage (//, say, larger than 4) may be obtained simply by rearranging results in Figures 1 and 2. For the two systems considered in Figure 1 and the seismic hazard shown in Figure 2, the resulting seismic damage curves are depicted in Figure 3. It is important to note that, once the mean and the uncertainty of SDM are obtained through a series of non-linear dynamic analyses of the structure, the computation of seismic damage curve (there could be more than one of such curves for the same structure, since more than one damage measure DM or location / could be selected to gauge the structural damage) is neither any different nor more difficult than the routine computation of conventional seismic hazard curves. The seismic damage curves computed following the methodology just presented do not consider possible uncertainty in structural resistance. In the previous example, the uncertainty in the structural resistance can be reflected in different plausible u. values in the columns that could bring the entire structure to collapse. This type of uncertainty (Resistance Uncertainty) is usually far less important than the variability present in the seismic loads. This uncertainty, however, can be easily included by using the following simplified approach. 254 Let HDi/x) = ?[DMJ > x] = ?[Sa > S^^J be the result of the modified hazard analysis plotted versus damage levels x (see Figure 3). Then assume that on log-log paper the seismic damage curve, HD}Jx), can be locally represented as a straight line, that is: K ) = Kox- > (2) Kj being the local slope estimated from the curve in the region of x equal to the median capacity. Furthermore, assume the capacity, C, of the structure to resist damage (measured in damage terms, such as ductility at location /, not in spectral acceleration terms) is lognormally distributed with median c and standard deviation of the natural logarithm equal to a. Incidentally, note that, still referring to the previous example, the variability on C is a column-to-column variability and has nothing to do with record-to-record variability (such as in Sau). With these assumptions, it can be shown that the probability of exceeding a desired level of structural damage is obtained as: r-2 2 hi a pf=P[DM,l>C)=jHDM{x)fc{x)dx^HDSI(C)e -' (3) is the (lognormal) probability density function of C. Notice that Equation 3 is exact if Equation 2 is exact for all damage values. The exponential term multiplying HDM(C) in Equation 3 is a "correction factor" which scales up the seismic damage curve to account for the resistance uncertainty not included before. This "correction factor" is close to one, provided that K;a is small. To quantify this term note that Kj=l/logl0 x10> where xI0 is the factor by which one must increase DM (here /S) in order to reduce the hazard by an order of magnitude. The value of xl0 is about 2 for the EP system in Figure 3, implying K} = 3.3. Recall that if the whole structure during the ground shaking behaves basically as an EP system, then the relationship between SDM and fj is nearly linearly proportional (see Figure 1). In this case seismic hazard curves and seismic damage hazard curves have the same slope. Therefore the experience gained with conventional seismic hazard calculations can be used to assess x10. In our experience this number ranges from 1.75 to 2.5, leading to K, values of 4.1 to 2.51. The slope K, depends on the site and on the probability of exceedance levels. Thus, assuming resistance variability expressed by COV(C) up to 0.6, the "correction factor" (see Table 1) may increase the exceedance probability by as much as one order of magnitude. To derive the numbers in the table recall that, assuming lognormality for C, the standard deviation of the natural logarithm of C, a, and the COV(Q are related by the following relationship: o2=COV2(C)+1 (4) If the structure behaves as a system with degrading strength and stiffness, then previous results show that SDMvaries with // with relationships similar to the curve with triangular markers in Figure 1. b This behavior can be described with relationship of the form SDKi=afi (with b<\) or 1 SDM =a-(bju+c/ . The parameters a, b and c are constants to be determined case by case. Note that, given the same seismic hazard curve, the slope of the seismic damage hazard curve for a 255 degrading system is less steep (see Figure 3) compared to the seismic damage hazard curve of an EP system, implying a lesser correction factor. This simple but effective methodology, which accounts for the uncertainties both in the loads and in the structural resistance, permits the direct computation of damage and, eventually, failure probabilities of structures. This approach does not include any hidden safety factors. Hence, it is conducive in the effort of reducing possible overconservatism in current practice. This is specially valuable in the reassessment process. 3.0 EXAMPLE This section includes an illustrative, made-up example regarding the computation of the failure probability, Ps, of a hypothetical nuclear power plant safety-related structure with regards to seismic loads. The calculation exploits the proposed methodology. It is assumed that preliminary analyses have shown the following: (1) the structure behaves dynamically as a 5%-damped 2.0 Hertz EP system; (2) the damage suffered in the columns during the ground shaking is responsible for the mechanism which leads the entire structure to collapse. This damage is monitored in terms of ductility ratios and the maximum of them, p, is deemed adequate to describe the severity of the induced damage; (3) the relationship between SDA/ at 2.0 Hertz and the maximum ductility ratio u. follows the straight line in Figure 1; (4) the seismic hazard curve in Figure 2 is available as a result of a seismic hazard analysis performed for the site. Both the seismic load uncertainty and the seismic damage potential uncertainty were already included in the derivation of such a curve. After these premises, the seismic damage curve for this structure is the curve for EP system displayed in Figure 3. Moreover, the median of the resistance capacity, C, of such columns is believed to be JJ = 4 with a COV of 0.4. From the seismic damage curve in Figure 3 it follows that x10 is about 2.0 and, thus, Kl is approximately equal to 3.3. Hence, the correction factor in Equation 3 is equal to 2.2 (Table 1). Finally, substituting these values in Equation 3, it follows that the probability of failure of such a 6 -5 6 structure due to seismic loads is simply P,= 5 x 10" x 2.2 = 1.1 x 10 (where 5 x 10" is HfJ(4) from Figure 3). 256 4.0 APPLICATIONS 4.1 POWER HOUSE This application has been published in this journal (Bazzurro et al., 1996). The SDHA methodology has been used to assess the seismic performance of a power house in the highly seismic island of Java, Indonesia. The historical events occurred in the region around the site are displayed in Figure 4. For illustration purposes the numerical calculations have been performed only on the transversal section of the structure, the weaker of the two sections of the building. The finite element model developed for investigating the non-linear response of this steel structure to severe earthquake is displayed in Figure 5. The software employed for this purpose is the widely distributed program DRAIN-2D (Kanaan and Powell, 1973). The frame was modelled by end moment-buckling elements. These elements consider the interaction between end-moments and axial force, the axial force being determined using the procedure suggested for buckling elements by Jain and Goel (1978). The floor, at elevation 7.95 meters, was considered rigid in its plane. Soil structure interaction was neglected and the columns were modelled as rigidly connected to the ground. The seismic response was investigated using six accelerograms recorded on soil, whose characteristics are included in Table 2. The horizontal component of each earthquake was randomly selected between the two available for each of the six events. These ground motions were chosen in order that: o the values of M and R were in the range of interest suggested by the SHA performed for the site under investigation. Seismotectonic features in the site region are capable of generating earthquakes of magnitude 8 and above. Thus, a 7.6 magnitude event was also included in the analyses. Small, distant earthquake recordings were purposely not selected because they are not threatening for well- designed structures; o the region of primary interest in the (M, R) plane was uniformly covered. Notice that preference in the choice of accelerograms should be given to ground motions that occurred in the area. Past earthquakes recorded at the site must be included, if available. Recall, again, that the choice of the limited sample size (i.e., six records) has been made according to the findings of previous studies. It was not the intent of that work to confirm again the statistical behavior of SDM. Much more robust statistical analyses ofSOM were carried out for a broad variety of SDOF systems (Sewell, 1988, 1992, and 1993; Sewell et al., 1991) and for some realistic MDOF systems (Inoue, 1990: Inoue and Cornell, 1991; Bazzurro and Cornell, 1994a) using much larger ensembles of ground motion recordings (i.e., more than 200 records for SDOF systems and more than 40 for some MDOF systems). The results showed that the uncertainty associated with the 257 estimate of the "true" mean ofSDM can be reduced to negligible values (compared to the uncertainty of Sa) by using a sample size of 5 to 7 ground motions. In fact the COV of the estimator is the COV ofSDM (found usually to be not greater than 0.3) divided by the square root of the sample size. Besides seismic loads, dead loads of both structure and internal equipment, and fifty percent of the design live loads were considered in the analyses. The non-linear dynamic analyses were performed in the time domain using an integration time step of 0.01 to 0.02 seconds and a material-structural damping of 5% before yielding. The damping after yielding is directly accounted for by the hysteretic behavior of the elements included in the model. When the structure is subjected to a sufficiently strong seismic excitation the response of the frame becomes non-linear. When lateral displacements of about 3 centimeters are observed at nodes 5 (or 8), hinges start forming at the base of the main columns (nodes 1 and 4). For extremely severe excitations, hinges develop also at the top and at the bottom of the internal columns (elements 16 and 17), at nodes 5 and 8 in elements 1, 2, 4 and 5, and at nodes 11 and 12 in elements 7 and 8 at the connections with the main columns. In order to quantitatively describe the seismic performance of the structure, .the plastic rotation of hinges in the main columns at nodes 1, 4, 5 and 8 was taken as reference damage measure DM. In fact, the most severe damage is concentrated at such locations. For a plastic rotation of 0.015 radians in hinges formed in elements 1 and 2 at nodes 1 and 5 (or, alternatively, in elements 4 and 5 at nodes 4 and 8) lateral displacements of 15 centimeters at node 5, and relative lateral displacements of 35 centimeters between node 5 and 13 can be observed. When at least one of the hinges reaches such a threshold value, To compute the mean values of SDM for the entire range of plastic rotation from zero to q>fail, six analyses were performed for each record in Table 2 scaled to successively higher values. The values of SDM for any desired damage level (i.e., percentage of plastic rotation at collapse) were computed by linear interpolation. Notice that spectral acceleration SDM was associated with a reference frequency equal 2.0 Hertz. This frequency is a linear combination of the first two modes of vibration which are almost equally important and dominate the lateral response of the frame. About the importance of the selection of a value for such reference frequency recall the discussion in Section 2.0. Once that the mean values and the COV's ofSDM were estimated, the annual probability of structural damage and, eventually, failure were computed in accordance with the methodology presented in Section 2.0, with the exception that the uncertainty in the structural capacity is not included here. The seismic hazard calculations were performed using the program EQRISK (McGuire, 1976). The seismic parameters for the area under investigation were estimated using ISC (1989 and 1994) and NOAA (1989 and 1994) earthquake catalogs. The historical events occurred in the region around the site are displayed in Figure 5. 258 In this case study the equation used to estimate the spectral acceleration at the site is the attenuation law developed by Campbell (1990). If possible, however, the selected attenuation law should include accelerograms of past earthquakes recorded in the area. Two sets of calculation were performed using two different values of the variability of SDM: one as computed from the analysis (i.e., COV equal 0.15 across the entire damage range) and the other augmented up to values (i.e., 0.3) which, in our experience, can be considered conservative. The results are the seismic hazard curves (Figure 6) expressed in direct damage terms. The two curves reflect the different degree of conservativeness and/or confidence in assessing the global uncertainty in the seismic hazard computations. Incidentally, with reference to the methodology previously illustrated and recalling that no capacity uncertainty was considered in this application, the curves in Figure 6 represent HDM(C) in Equation 3. Thus, the annual probability of observing an earthquake at the site causing the yielding of the main power house structure is about 1 x 10"3 to 2 x 10°, whereas the collapse of the building has an annual probability of occurrence of approximately 4 x 10^ to 2 x 10"5 in the two cases. Again, these ranges of probabilities reflect the confidence in assessing the response-based factors involved in the computations. The annual probability of yielding, structural damage, and collapse shall then be compared with target probabilities. For the seismic requalification of this type of buildings, usually classified as non- nuclear safety structures, an acceptable level would perhaps be a probability of collapse less than 10-4 per year. 4.2 JACKET-TYPE OFFSHORE PLATFORM This application appeared in Bazzurro and Cornell (1994b). The structure is the steel, jacket-type, Unocal's offshore platform called Rajah Wellhead (Figure 7) located immediately offshore of eastern Kalimantan, Indonesia (Figure 8). This large 3D-structure and its foundation is a challenging example because it displays an array of types of nonlinearities: steel and soil material nonlinearity as well as softening geometric nonlinearity, both locally (buckling braces) and globally {P-A effect). Rajah Wellhead is a typical jacket-type platform operating in 45 m of water. It has a rectangular base and four legs. The piles are driven to a depth of 107 m below the mud line and grouted inside the legs. The distance between the lower side of the deck and the mean water level is about 14 m. Both the superstructure and the piles are made up of steel tubular members. The uppermost soil layer is a soft clay, whose resistance characteristics are very poor. Non-linear dynamic analyses were performed on a 3D finite-element model developed by means of the computer program Karma (ISEC, 1989). The model includes 451 nodes, at which the structural masses are lumped, 644 elements; and 2010 degrees of freedom. Legs and piles were modeled by large-displacement inelastic beam-column elements with distributed plasticity; diagonal braces were modeled by large-displacement postbuckling elements with degradation of both strength and stiffness of the section. Linear beams were used for the deck and non-linear near-field elements were used for the soil. In particular, the foundation soil was modeled by defining a set of three orthogonal springs 259 (two lateral and one axial) at fixed elevations along the shaft of the pile. The pile-soil deformation was related to the soil resistance in both the lateral and the axial directions by specifying, respectively, lateral p-y and vertical t-z force-deformation curves. These curves, for both virgin and degraded soil, were available as a result of a geotechnical investigation at the site. The analysis permits transition from one state to the other. The non-linear dynamic analyses were performed in the time domain, by direct integration of the equations of motion. The non-linear postelastic behavior of the elements explicitly accounts for the structural damping after material yielding. Earthquake (inertia) loads, dead load of the structure in addition to equipment located on the deck, live loads present on the deck during oil production, and buoyancy loads on the submerged members were included in the non-linear dynamic analyses. The actions of other environmental loads, resulting from wind, wave, and current, were not included. Drag forces acting on the submerged members of the jacket, due to the motion induced by the earthquake, were also neglected. In this case, to evaluate the mean of SDM the five earthquakes in Table 3 were adopted. These magnitude and distance pairs were selected to describe the seismicity of the area around the Rajah Platform site. Beyond using records only from "soil" sites, no modifications were made for the specific local site conditions. On the basis of both modal and preliminary non-linear dynamic analyses, the spectral acceleration SDM was associated to a representative frequency of 0.55 Hertz and a damping of 5% of critical. Again the seismic damage risk for the jacket is evaluated by using the methodology presented before with the only exception concerning the structural uncertainty that was not included in the calculations. Three non-linear types of damage DM are considered in the case of the Rajah Wellhead Platform: two for describing the global performance and one for monitoring the local damage in the piles. The overall postelastic damage of the platform is measured by the global ductility ration fu , based on deck displacement. The two different types of global ductility ratios considered are the following: o Mdispjc which is the ratio of the maximum lateral oscillation amplitude of the deck to the reference deck displacement (7.6 cm). Such reference value was selected because due to the poor strength and stiffness of the uppermost clayey layer, the platform subjected to the action of only the asymmetric state vertical deck and self- weight loads deviates from the upright position and leans towards the negative ^-direction of 7.6 cm, before the lateral loads are applied. Initially, under the action of the lateral loads, all the inelastic events occur only in the soil, and the platform tilts virtually as a rigid body while remaining elastic response analyses. The first inelastic event in the structural frame occurs the foundation soil has already yielded in several places. The inelastic events in the structure are concentrated in the piles below the mud line; o noffJC which is the ratio of the deck displacement permanent offset in the ^-direction at the end of the ground shaking to the reference offset displacement (i.e., 12 cm, see Bazzurro and Cornell 1994b, for details on the criteria employed for the computation of such 260 value). This is an important damage measure, as revealed by preliminary non-linear dynamic analyses. In fact, because the X- direction is more flexible and weaker than the ^-direction and because the structure is initially leaning in the negative X-direction, the platform responds to seismic excitations by progressively increasing its deviation from the vertical position towards the negative X-direction. The situation is more complicated in the case of the local damage that occurs only in the piles at a considerable depth below the mud line (about 40 m). Piles behave as non-linear beam-column elements whose response is governed by a set of multilinear relationships that include axial force- displacement, in-plane and out-of-plane moment rotation, and torque-twist behavior. In these elements, a four-dimensional interaction surface controls the inelastic response, and this fact makes it difficult to postulate a compound damage measure. In this particular case, though, since the damage occurs mainly in a pile at a large depth below the mud line, the axial force is dominant over bending and torque moments. Therefore, the usual ductility ratio JJ1OC, based on the axial deformation of the element, is a reasonably accurate measure of the peak damage sustained by the piles. The seismic hazard computations were conducted using the computer program EQRISK (McGuire, 1976). The study of the tectonic features of the region and the spatial occurrence of historical earthquakes has not yet provided sufficient information for identifying with confidence active faults nearby (say within 100 km). The historical data, however, suggests the partition of the region into the three large seismic zones shown in Figure 8. Within each zone, the seismicity was considered relatively uniform, and figure occurrence of earthquakes was described by a single probability distribution. Seismic events wee modeled to occur as single points of energy release at a random location. The earthquake magnitude distribution adopted for the seismic zones in the present study is the doubly truncated exponential distribution. The equation used to estimate the spectral acceleration (for a frequency of 0.55 Hz and damping ratio of 5%) at the site is the attenuation law developed by Campbell (1990). This equation applies specifically to soil sites. No other provisions were made, however, for possible amplification of the input motions due to the extremely soft soil condition at the Rajah Wellhead Platform site. According to the methodology presented before, the variability in the spectral acceleration Sa was increased in order to account for the variability in both SDM. The seismic hazard curves corresponding to the ductility values associated with the maximum deck displacement in the X- direction (measured by fJdlsp^), the residual permanent deck displacement in the ^-direction ar| (measured by Mofp<)> d the structural damage in the piles (measured in terms of {tloc) are displayed in Figure 9. From this figure it follows that, for example, the annual probability that the structure may experience a final permanent displacement of 23 cm or more in the ^-direction (corresponding to iioffx « 2) is approximately 5.0 x 10^. Moreover, the collapse of the Rajah Wellhead Platform, which is an event that requires the values of ductility ratios (both global and local) equal to or higher than those displayed in Figure 9, has an annual probability of occurrence that is less than 1.0 x 10-3. Again, it is important to appreciate that since no uncertainty in the structural capacity was included in the analysis, the three curves in the figure represent the term HDM(C) for three different damage measures DM. 261 4.3 STABILITY OF A SLOPING SEABED This application has been presented at the Third Symposium on Strait Crossings held in Alesund, Norway, June, 1994 (Bazzurro et al., 1994b). This section, excerpted from the reference cited above, discusses the numerical evaluation of the annual probability of exceeding specified displacement values in subsea slopes of the Messina Strait, Italy. As pointed out by Newmark (1965) the magnitude of slope displacements that develop during an earthquake should be the criterion for assessing the degree of stability or instability of a slope, rather than consideration of the possibility of the factor of safety dropping to unity during the earthquake. The amount of acceptable deformation depends on the specific problem. For the landfalls of a submerged-floating tunnel, as an example, the anticipated slope displacement must be small enough not to induce structural collapse. Depending on the seismic event considered, the operability of the system may also be a concern. In this case relatively large deformations may not be acceptable even if the structural safety is preserved. The slope displacements can be computed by a procedure analogous to that used for analyzing the movement of a sliding block on an inclined plane (Figure 10). The displacements in the soil mass are computed integrating twice the acceleration time history portion exceeding the yield acceleration av, as schematically depicted in Figure 11. The yield acceleration is defined as the acceleration value at the ground surface for which failure occurs (i.e., shear strength exceeded) along a potential sliding plane (see Pelli et al., 1994). The yield acceleration depends on soil strength, slope angle, total and effective stresses at the failure surface level, and depth of the failure surface. In this context, ay is the only parameter representing the "capacity" of the slope. Although crude, this model has been used extensively in the past thirty years as it catches some of the basic physics controlling the stability of slopes and embankments under seismic loading (e.g., Goodman and Seed, 1966; Sarma, 1975; Seed et al., 1985; Lin and Whitman, 1986; Yegian et al., 1991;Baziaretal., 1992). Bearing in mind the premises above, in this context, the damage was assumed to be simply the final permanent deformation of the slope. According to the Newmark's method referred to above, the slope lateral displacements induced by the ground shaking can be found by double integration of the earthquake acceleration time history. Only the spikes above the yield acceleration value, a give contribution to the final permanent deformation. Again, the parameter ay is the only quantity describing the resistance of the slope to the ground shaking. or In this application SaDMHx ( simply SDAf) is the peak ground acceleration (always greater than ay) necessary to induce in the slope a permanent displacement x at the end of the ground shaking. SDM, for any value of x, is a random variable whose mean in this application was evaluated using a large database of 52 strong-motion records (Table 4). The seismic hazard calculations were performed using the in-house modified version of the computer program EQRISK (McGuire, 1976). The equation used to estimate the PGA at the site is the attenuation law developed by Joyner and Boore (1988). The variability of PGA suggested by Joyner and Boore was conservatively increased by 20% to account for the additional variability on SDM (see Bazzurro et al., 1994b, for details). 262 The procedure yielded the seismic damage hazard curves in Figure 12 expressed in terms of permanent displacements for different slopes at the site (i.e., for different ay values). For example a displacement of 0.05 meters is exceeded at the site in a slope having a critical acceleration of 0.3 g on average once every 10,000 years. 5.0 CONCLUSIONS This contribution has demonstrated that a methodology is available to assess the probability of exceedance of damage states induced in real structures by seismic loadings. In the qualification process of existing facilities this is a vital need, that is fulfilled without including any hidden safety factor. The SDHA methodology allows for a realistic evaluation of safety margins. Furthermore, upgraded structural configurations can be quickly analyzed, thus validating the reassessment strategy. 6.0 REFERENCES American Petroleum Institute (API), 1993, "Recommended Practice for Planning, Designing and Constructing Fixed Offshore Platforms", API RP2A-WSD, 20th Edition, Washington, DC. Baziar, M. H. R. Dobry & M. Alemi 1992. Evaluation of lateral ground deformation using sliding block model. Proc. 10th World Conf. on Earthquake Engrg., Madrid, 3: 1401-1406. Bazzurro, P. and C. A. Cornell, 1992, "Seismic Risk: Non-Linear MDOF Structures". Proceedings of ]Qth World Conference of Earthquake Engineering. Vol. 1, pp. 563-568, Madrid. Bazzurro, P. and C. A. Cornell, 1994a, "Seismic Hazard Analysis of Non-linear Structures. I: Methodology", Journal of Structural Engineering, ASCE, Vol. 120, pp. 3320-3344, November. Bazzurro, P. and C. A. Cornell, 1994b, "Seismic Hazard Analysis of Non-linear Structures. II: Applications", Journal of Structural Engineering, ASCE, Vol. 120, pp. 3345-3365, November. Bazzurro, P., C. A. Cornell, D. Diamantidis and N. R. Vaidya, 1994a, "Probabilistic Seismic Requalification of Nuclear Power Plant Structures", Proceedings of ASME PVP-1994 Conference. Minneapolis, MN, June 19-23. Bazzurro, P., Cornell, C. A., Pelli F. and G.M. Manfredini, 1994b, "Stability of Sloping Seabed: Seismic Damage Analysis, Methodology and Application", Proceedings of Strait Crossings 1994. Balkema, Rotterdam, The Netherlands, pp. 821-829. Bazzurro, P., Cornell C. A., Diamantidis D. and G. M. Manfredini, 1996, "Seismic Damage Hazard Analysis for Requalification of Nuclear Power Plant Structures: Methodology and Application", Journal of Nuclear Engineering and Design, 160, pp- 321-332. Campbell, K. W., 1990, Report, "Empirical Prediction of Near-Source Soil and Soft Rock Ground Motions for the Diablo Canyon Power Plant Site", San Luis Obispo, CA, prepared for Lawrence Livermore National Laboratories, Dames & Moore Job No. 10805-476-166. 263 Cornell, C.A., 1968, "Engineering Seismic Risk Analysis", Bulletin of the Seismological Society of America, Vol. 58, No.5. Goodman, R. E. & H. B. Seed 1966. Earthquake induced displacements in sand embankments. J. SoilMech. Found Div., ASCE 92 (SM7): 125-146. Inoue, X, 1990, "Seismic Hazard Analysis of Multi-Degree-of-Freedom Structures", Report No. RMS-8, Department of Civil Engineering, Stanford University, Stanford, CA. Inoue, T. and C. A. Cornell, 1991, "Seismic Hazard Analysis of MDOF Structures", Proceedings of ICASP. Instituto de Ingenieria, UNAM, Mexico City, Mexico. International Seismological Centre (ISC), 1989, "Historical Hypocentre File", Computer File, ISC, Newbury, U.K. International Seismological Centre (ISC), 1994, Updating of the "Historical Hypocentre File", Computer File, ISC, Newbury, U.K. ISEC Inc., "KARMA Computer Program", Documentation, Vol. 1-V, San Francisco, California. Iwan, W. D., 1980, "Estimating Inelastic Response Spectra from Elastic Spectra", Earthquake Engineering and Structural Dynamics, Vol. 8, pp. 375-399. Jain, A. K. and S. C. Goel, 1978, Hysteresis Models for Steel Members Subjected to Cyclic Buckling or Cyclic End-Moments and Buckling, Report No. UMEE 78R6, Dept. of Civil Engineering, University of Michigan, MI. Joyner, W. B. & D. M. Boore 1988. Measurement, characterization and prediction of strong ground motion, Proc. of Earthquake Engrg and Soil Dynamics II, Geotechnical Division, ASCE, Park City, UT., June 27-30. Kanaan, A. E. and G. H. Powell, 1973, Purpose Comuter Program for Inelastic Dynamic Response of Plane Structures, Report No. EERC 73-6, Earthquake Engineering Research Center, University of California, Berkeley, CA. Kennedy, R. P., S. A. Short, K. L. Mertz, F. J. Tokarz, I. M. Idriss, M. S. Power and K. Sadigh, 1984, "Engineering Characterization of Ground Motion - Task 1: Effects of Characteristics of Free- Field Motion on Structural Response", NUREG/CR-3805, Vol. 1, U. S. Nuclear Regulatory Commission. Lin, J. S. & R. V. Whitman 1986. Earthquake-induced displacements of sliding blocks. J. Geotech. Engrg., ASCE 112(1): 44-59. McGuire, R.K., 1976. EQRISK, Fortran computer program for seismic risk analysis, Open File Report 76-67, USGS, Denver, CO. National Oceanic and Atmospheric Administration (NOAA), 1989, "The Entire Earthquake Data Base", Computer File, Boulder, Colorado. 264 National Oceanic and Atmospheric Administration (NO A A), 1994, Updating for PDE Catalog from January 1989 to July 1993, Computer Files, Boulder, Colorado. Newmark, N. M. 1955. Effects of earthquakes on dams and embankments. Geotechnique 15(2): 139-160. Pelli, F., G. V. Vassallo & F. Casola 1994. Earthquake-induced permanent deformations in the landfalls of submerged-floating tunnels in sand. Proc. of the Third Symposium on Strait Crossings, Alesund, Norway. Press (Japan) Guidelines Drafting Working Group, 1992, "Design Guidelines (Draft) for Reinforced Concrete Buildings", The Third Meeting of The U.S.-Japan Joint Technical Committe4e on Precast Seismic Structural Systems, San Diego, California, November. Sarma, S. K. 1975. Seismic stability of earth dams and embankments. Geotechnique 25(4): 743-761. Seed H. B., R. B. Seed, S. S. Lai & B. Khemeneh-pour 1985. Seismic design of concrete faced rockfill dams. Symp. on Concrete Face RockfiU Dams, ASCE: 459-478. Sewell, R. T., 1988, "Damage Effectiveness of Earthquake Ground Motion: Characterizations Based on the Performance of Structures and Equipment", Ph.D. Dissertation, Dept. of Civil Engineering, Stanford University, Stanford, CA. Sewell, R. T.5 G. R. Toro and R. K. McGuire, 1991, "Impact of Ground Motion Characterization on Conservatism and Variability in Seismic Risk Estimates", Prepared for U.S. NRC, Risk Engrg., Inc., Final Report, Golden, CO, May. Sewell, R. T. 1992. Effects of duration on structural response factors and on ground-motion damageability. Proceedings of SMIP92, edited by M. J. Huang, Div. of Mines ad Geology, Calif. Dept. of conservation, Sacramento, CA. Sewell, R. T., 1993, "Impacts of Earthquake Strong-Motion Duration on Inelastic Structural Response Factors and on Ground-Motion Damage Potential", CSMIP Data Utilization Report. Risk Engrg, Inc., Final Report, Golden, CO, May. Trifunac, M. D. and A. G. Brady, 1975, "A Study on the Duration of Strong Earthquake Ground Motion", Bulletin of the Seismological Society of America, Vol. 65, No. 3, pp. 581-626. U.S. Department of Energy (DOE), 1993, "Natural Phenomena Hazards, Design and Evaluation Criteria for Department of Energy Facilities", DOES-STD-1020-92, Washington, D.C., February. U.S. Nuclear Regulatory Commission, "Individual Plant Examination of External Events (IPEEE)", Generic Letter No. 88-20, Supplement 4, NUREG/CR-1407. Yegian, M. K., E. A. Marciano & V. G. Ghahraman 1991. Earthquake-induced permanent deforma- tions: probabilistic approach. J. Geotech. Engrg., ASCE 117(1): 35-50. 265 TABLE 1 CORRECTION FACTOR FOR TYPICAL Kj AND RESISTANCE UNCERTAINTIES VALUES 4 .i 3.3 2.51 0.2 1.4 1.2 1.1 cov(C) 0.4 .5 2.2 1.6 0.6 i: 5.3 2.6 Notes: (1) See Equation 3 for correction factor definition. (2) The value of K; equals 3.3 corresponds to xl0 equals 2.0. (3) COV stands for coefficient of variation. 266 TABLE 2 LIST OF GROUND MOTION RECORDS USED IN THE ANALYSES OF THE POWER HOUSE EARTHQUAKE STATION NAME DATE COMP. DISTANCE MAGNITUDE PGA DURATION No. NAME (Km) (cm/sec2) (sec) 1 Kem County, CA Callech Athenaeum 07-21-52 SOOE 109 7.4 -46.5 30.0 2 Hollister, CA Gilroy, Gavilan 11-28-74 S23E 10 5.2 -94.1 2.8 3 Imperial ValJey, CA El Centre, Array 5 10-15-79 230° 4 6.5 367.2 9.5 4 Parkfield, CA Taft Lincoln School 06-27-66 S69E 105 6.1 11.2 33.4 5 Lima, Peru Inst. Geofisico 10-03-74 N08E 38 7.6 179.0 48.4 6 Loma Prieta, CA Palo Alto Hospital 10-18-89 212° 47 7.0 378.2 13.1 Note: The duration values reported in the table refer to the ground-motion duration which brackets the 90% of the energy released (Trifunac and Brady, 1975). 267 TABLES CHARACTERISTICS OF GROUND MOTIONS INCLUDED IN DATABASE USED IN THE OFFSHORE PLATFORM APPLICATION EARTHQUAKE STATION NAME DATE DISTANCE MAGNITUDE SOIL TYPE No. NAME (Km) (sec) 1 San Fernando, Calif. Wheeler Ridge 02-09-71 82 6.6 Soil 2 Long Beach, Calif. Vemon CMD Building 03-11-33 22 6.2 Soil 3 Borrego Mountain, Cal Edison, Colton 04-09-68 130 6.6 Soil Calif. 4 Kem County, Calif. Caltech Athenaeum 07-21-52 109 7.4 Soil 5 Loma Prieta, Calif. Olema, Ranger Station 10-18-89 136 7.0 Soil 268 TABLE 4 EARTHQUAKE RECORDS USED IN THE ANALYSES OF THE SLOPING SEABED EARTHQUAKE STATION NAME DATE COMP PGA (cm/sec2) No. NAME 1 Kem County, CA Taft Lincoln school 07-21-52 EW 175.9 2 Imperial Valley, CA El Centre, Sta.9 05-19-40 S90W 210.1 3 Kern County, CA Caltech Athen. 07-21-52 SOOE -46.5 4 Lytle Creek, CA Cal Edison, Colton 09-12-70 South -40.2 5 San Fernando, CA Palos Verdes 02-09-71 S25E -40.1 6 Hollister, CA Gilroy, Gavilan 11-28-74 S23E -94.1 7 Kalapana, Hawaii Hilo. Univ Hawaii 11-29-75 N16W -169.9 8 Santa Barbara, CA SB. FreitasBldg 08-13-78 158° 284.7 9 Tabas, Iran Tabas 09-16-78 Long 795.8 10 imperial Valley, CA El Centre. Arr.5 10-15-79 230° 367.2 11 Imp.Vry, Aftershock CA El Centre, Arr.4 lO-i.'-79 140° 229.6 12 Lrvermore, CA Livermore. Hosp. 01-27-80 038° 60.6 13 Parkfield, CA Taft Lincoln school 06-27-66 S69E 11.2 14 ML Hamilton, CA HollA City Hall 04-24-84 271° 71 1 15 Imperial Valley. CA Brawlcv Muni Arpt. 10-15-79 225° 162.2 16 Imperial Valley. CA El Centre. An. 10 10-15-79 050° -168.2 17 San Fernando, CA Fort Tejon, CA 02-09-71 N00E -24.7 18 San Fernando, CA Port Hueneme 02-09-71 S90W -25.2 19 San Fernando, CA Long Beach, CA 02-09-71 N21W -28.4 20 San Fernando, CA Oso Pump Plant 02-09-71 N00E -85.2 21 Borrego MountairuCA El Centre, Sta.9 04-09-68 SOOW -127.8 22 Lima, Peru Inst. Geofisico 10-03-74 N08E 179.0 23 Lima. Peru Dr. Huaco Home 10-03-74 LONG 192.4 24 Kern County. CA S.Barbara Courthse 07-21-52 S48E 128.6 25 Kalapana, Hawaii Punaluu. Hawaii 11-29-75 N16W 77.1 26 Imp.Vly, Aftershock CA Brawlcy Muni Arpt. 10-15-79 225° -35.2 27 Santa Barbara. CA S.Barbara Courthse 08-13-78 090° -284.1 28 Lytle Creek, CA Park Dr..Wrightwood 09-12-70 S65E 139.0 29 Hollister. CA San Juan Bautista 11-28-74 S57E 112.1 30 Loma Prieta. CA Fremont 10-18-89 0° -117,7 31 Loma Pricta. CA Palo Alto Hospital 10-18-89 212° 278.2 32 Loma Prieta. CA SFO Airport 10-18-89 90° -325.8 33 Friuli. Italv Castelfranco 05-06-76 NS 30.4 34 Friuli. Italy Codroipo 05-06-76 EW 85.3 35 Friuli. Italy Conegliano 05-<>6-76 EW -76.5 36 Friuli, Italy Forgaria 05-06-76 NS -42.2 37 Friuli. Italv Maiano 05-06-76 NS -81.5 38 Friuli. Italy Buia 09-15-76 NS 107.9 39 Friuli. Italy Conegliano 09-15-76 EW 19.6 40 Friuli. Italy Forgaria 09-15-76 NS 258 41 _ Friuli. Italv Codroipo 09-15-76 NS 43.2 42 Friuli. Italy Conegliano 09-15-76 EW 30.4 43 Friuli. Italy Forgaria 09-15-76 NS 346.3 44 Friuli. Italy Tarcento 09-15-76 NS 136.4 45 Friuli, Italy Buia 09-15-76 EW -93.4 46 Friuli. Italv Forgaria 09-11-76 EW 112.8 47 Friuli; Italv Tarcento 09-11-76 NS 200.1 48 Friuli. Italy Conegliano 09-11-76 NS 13.7 49 Friuli. Italy Forgaria 09-11-76 EW -228.6 50 Friuli. Italy Buia 09-11-76 NS -229.6 51 Norcia. Italy Bcvagna 09-19-79 NS 37.3 52 Umbria. Italy Peglio 11-29-84 NS 53.0 269 2.4- i i / 2.2- i ' / 2- ! /* 1.8- 1.6- 1.4 / 1 0.8 0.6 / 0.4 1 2 3 4 5 6 7 Ductility ratio [i —•- EP system —A- Degrading system Figure 1: Typical Relationships Between Spectral Acceleration and Structural Damage u.uui: -4 I i > 1 o \ •S 0.0001- o I^ \ I Exc e t o V g 1E-05- \ :_. \_ \ \ i 1E-06 \ 0.1 1 1 Figure 2: Seismic Hazard Curve 270 1 \ o (0 \ J o 1 i-, X LLJ x ~. "o 1E-05: « JQ 1 O IN i 1E-06 10 Ductility ratio (i EP system Degrading system Figure 3: Seismic Damage Hazard Curves 271 4'S A SUMATRA ^© T f\_2L ^LS • JAVA SEA 9TS 1CK'E 105* 106" 107 108* SCALE 0 35 70 14OKm LEGEND 6.00 - t . 49 A 3.00 - 3,49 • 6,50 - 6 .99 O 3.50 - 3.99 7.00 - 7 .49 e U.00 - U . 49 7.50 - 7 .99 • H.5C - H.99 • 8.0C - 8 .49 A 5.00 - 5.49 • 8.50 - 8 .99 SITE SYMBOL 5.50 - 5.99 272 0 Figure 4: Past Seismic Activity in the Region 25.00m 20m LEGEND 1 NODE NUMBER 0 ELEMENT NUMBER Figure 5: Power House Steel Structure, Numerical Model of the Transversal Section 273 (D 0.01 O C CO Q) O 0.001 CD ~ 0.0001 E _Q CO O 1E-05= co D C C CO 1E-06 0 10 20 30 40 50 60 70 80 90 100 % of plastic hinge rotation at collapse LEGEND COMPUTED USING CONSERVATIVE VALUES OF COVS COMPUTED USING ESTIMATED VALUES OF COVS Figure 6: Seismic Hazard Curves in Terms of Structural Damage 274 Waterline level 45 m Mudline level 57 m LEGEND: II Soft Clay 12 m [HI! Dense Sand •1 Stiff Clay 50 m Figure 7: Rajah Wellhead Platform with Foundation Soil Profile (Not to Scale) 275 Celebes Sea 114 120° 122° Figure 8: Location of Rajah Wellhead Platform Site and Seismic Zones Figure 9: Seismic Hazard Curves Obtained for Three Kinds of Postelastic Damage Considered Critical in Rajah Wellhead Platform 276 LEGEND M W-OCK «SS W ILOCH ME 1DM1 R RESISTING FO«CE aiDOCCELEKSTIOH HUE MlSIWtt a SLOTE OXOLE RIGID BLOCK SHQKIHG UEDGE Figure 10: Sliding Block Model LEGEND TltLO OCCElEtOUON ) «CCEl.E««:iON 1I«E KISTOSY INDUCED BT 1ME FMTHOUAKE TIHE Figure 11: Method to Compute Residual Displacement (Modified After Goodman and Seed, 1966) 277 -i»- a=().l • -*- • 1=0 L'g -=- a=0 :l£ -**- a=0.4 =o.5e 5 0.1 •o 1 15^ \ "o 0.01 — — — L- *-> ^ • < _ 1 ——~ ~ •• I..,,, ,~ | 0.001 =—=? i. — £ k. I \ — c. 1 •-*— ——-_ ~tt • • '" '— • - --^ \ 0.0001 '',» 1—1 ^^ —=^— a =^—=s — -= ; 1 •—. IE-OS i— 1 , 0 0.0S 0.1 0.1:••> 0.2 0.25 0.:! O.:Jf> 0 4 0.45 0.5 Permanent Displacement (m) Figure 12: Seismic Damage Hazard Curves in Terms of Permanent Displacement. The a in the legend stands for yield acceleration 278 SESSION V "EXPERIMENTAL METHODS FOR SEISMIC CAPACITY RE- EVALUATION" NEXT PAGE(S) I left BLANK I , 1 279 Full-scale dynamic structural testing of Paks Nuclear Power Plant E.M. DA RIN, F.P. MUZZI ISMES S.p.A., Bergamo, Italy ABSTRACT: Within the framework of the IAEA coordinated "Benchmark Study for the seismic analysis and testing of WWER-type NPP's", in-situ dynamic structural testing activities have been performed at the Paks Nuclear Power Plant in Hungary. The specific objective of the investigation was to obtain experimental data on the actual dynamic structural behaviour of the plant's major constructions and equipment under normal operating conditions, for enabling a valid seismic safety review to be made. This paper gives a synthetic description of the conducted experiments and presents some results, regarding in particular the free-field excitations produced during the earthquake-simulation experiments and an experiment of the dynamic soil-structure interaction global effects at the base of the reactor containment structure. Moreover, a method which can be used for infering dynamic structural characteristics from the recorded time-histories is briefly described and a simple illustrative example given. 1. INTRODUCTION An IAEA Coordinated Research Programme was initiated in the early nineties to assist the countries of Central and Eastern Europe in evaluating the actual safety conditions of their first generation nuclear power plants. This Programme fundamentally aims at providing technical bases to the safety related decisions to be taken by the countries operating the plants, with the consulting assistance of other countries providing technical and financial support. Within the above-outlined context, a full-scale experimental investigation into the dynamic structural characteristics of a typical WWER-type Nuclear Power Plant has recently been performed at Paks in Hungary. Experimental data on the actual dynamic behaviour of the plant's major structures is obviously essential for validating computer models and allowing valid seismic safety analysis to be made. The Paks NPP site has thus been subjected to earthquake-like ground shaking through appropriately devised buried explosions - at a safe distance from the plant - and the dynamic response of the plant's major structures digitally recorded, together with the concurrent free-field excitation. The large amount of experimental data acquired during three successive earthquake simulation experiments is being analyzed for to extracting useful reference information. 281 Figure 1. General view of the Paks Nuclear Power Plant. 2. PLANT AND SITE SHORT DESCRIPTION There are presently four WWER-440 type V-213 reactor units in operation at the Paks NPP. The latter was originally designed in the former Soviet Union, but some adaptations were made by Hungarian design offices. The two first reactor units started commercial operation in 1983 and 1984. In the design stage the seismic hazard of the Paks site was considered to be very low and thus, no special regard was given to possible earthquake actions. Lately however, the seismic hazard of south-eastern Hungary is being revised and it was hence considered important that the seismic safety of the Paks NPP be rationally reviewed. The four reactors of the Paks NPP are arranged as two twins (Figure 1). The main building of each twin houses two reactor units in a symmetrical layout and is made up of a stiff reinforced concrete containment building, that is supported - together with an adjacent condensation tower - on a 2m thick continuous direct foundation slab. The foundation soil is a rather soft one, being composed of alluvial silts, sands and gravels becoming dense at around 16m depth. 3. SIMULATED EARTHQUAKE EXCITATION TESTS The Paks NPP site was thus subjected to the effects of appropriately designed buried explosions, with the object of inducing an earthquake-type excitation of the plant's structures. By transmitting the vibratory energy to the structures through their own foundation soil - as actually occurs during real earthquakes - the full-scale dynamic soil-structure interaction effects are activated and can hence be realistically investigated. 282 Three different successive earthquake simulation experiments were performed at the Paks site, with the whole nuclear power plant under normal operating conditions. The experiments were performed by igniting TNT charges, installed in 50m deep boreholes at an overall horizontal distance of about 2,5km from the 1st unit reactor base centre. • The first of the three experiments was a single blast one, which allowed to evaluate the blast-induced vibrations intensity and to conveniently calibrate the dynamic range of the measurement instrumentation. • Subsequently, two time-delayed multiple blasts were produced, with the object of somewhat lengthening the ground excitation duration. In fact, the duration of real earthquakes is obviously longer than that produced by a single underground explosion; but, even more important in the present context, a higher frequency resolution can be used in extracting structural behaviour information from the experimental records, if the latters are of longer duration. Each one of the earthquake excitation tests comprised a different layout of the measurement instrumentation, for the scope of acquiring a comprehensive experimental data set on the structural response of all the power plant's major constructions. A large number of dynamic transducers were installed at appropriate locations in the nuclear power plant's structures. A series of sensitive velocity transducers (seismometers) were fixed against the reactor building foundation mat; in particular, three vertical and two horizontal sensors were set up around the base of the reactor shaft massive containment structure, as shown in Figure 2, and a number of identical sensors were installed at the upper reactor hall floor level. Figure 2. Measurement stations around the reactor shaft base. For measuring the actual free-field excitation produced by the earthquake simulation experiments, three further seismometers were buried 1m deep into the natural soil at a 120m lateral distance aside the reactor base centre. Moreover, a series of piezoelectric accelerometers were used for measuring the vibrations at the upper levels of the reactor hall steel superstructures and close to the top of the nearby reinforced concrete twin chimneys. For the synchronous recording of all the structural response data, together with the concurrent free-field excitation, use was mad of an advanced multichannel data acuisition and analysis system, developed by ISMES and the hardware of which was set up in a mobile laboratory, parked beside the reactor containment building. This system is capable of simultaneously recording up to 52 signals at a 200Hz sampling frequency, with real-time analog to digital conversion; it is a submodule of "AIACE" (the Advanced ISMES Acquisition, Analysis and Control Environment), which was specifically developed for performing static or dynamic experiments, while providing also ample data analysis capabilities. In the case of time-history data to be collected, 283 the acquisition process can be automatically triggered according to a specified criterion; data from all the connected transducers are fed to signal conditioners which, after on-line A/D conversion drive directly into the computer memory. At the end of the data acuisition process, the collected data are ready for graphical examinations by means of various plotting functions, as well as for applying time or frequency domain signal analysis procedures. 4. EXPERIMENTS PERFORMED AND RESULTS OBTAINED As already outlined above, three different blast-induced ground excitation tests were performed at the Paks NPP site, with the plant in normal operating conditions. During each single experiment 52 digitized response signals were simultaneously recorded at a 200Hz sampling rate. Analogic low-pass filters were used for eliminating the high frequency noise prior to digitizing. A preliminar test was carried out by simultaneously detonating two 50kg charges in 50m deep boreholes at a 2442m distance in the SSE direction from the NS oriented first reactor building. Subsequently, two time delayed multiple blasts experiments were performed with the scope of lengthening the overall excitation duration. The first multiple blasts experiment was carried out by detonating three 100kg charges, with two 1,64sec delays, at practically the same mean horizontal distance from the reactor building than before. A second multiple blasts test was later performed with two 150kg blows and a l,58sec delay. Figure 3 shows the three-orthogonal velocity time-histories that were recorded in the free-field during the triple delayed blasts experiment. » 1i • Ki i tilHSnai Slcngitudiral Figure 3. Free-field response records. Figure 4. Reactor base response records. 284 The Paks NPP site appears to have been significantly excited by the buried explosions; about 20 sec long useful response signals were obtained. The free-field records show two consecutive rather distinct high and low frequency excitation phases, separated by an intermediate interference period. The following maximum peak velocities were recorded in the free-field during the respectively higher and lower frequency excitation phases: • 0,081 and 0,071 cm/s in the horizontal directions, • 0,287 and 0,058cm/s in the vertical one. These values are well below the 0,5 and 0,3cm/s conservative foundation velocity limiting values that are recommended in the DIN4150/3(1983) Standard for preventing any damage to occur in the case of blast induced vibrations in a "particularly sensitive building environment". The corresponding maximum peak horizontal accelerations are close to that of a M.M. grade III intensity earthquake, characterized by maximum horizontal accelerations up to 0,002g. 5. REACTOR SHAFT RESPONSE Figure 4 shows the time-histories that were recorded during the triple blasts experiment at the reactor shaft base (see Figure 2) in the longitudinal, transverse and vertical directions. The reactor shaft base responses (recorded at the reactor building foundation slab level) appear to be significantly lower than the corresponding free- field excitations; with the exception of the lower frequencies vertical vibrations, which show to maintain almost the same amplitudes - however with a slower decay - at the reactor base than in the free-field. Just a slight amplification of the vertical response was measured around the metallical top of the reactor shaft, suggesting that a prevailingly "rigid" vertical response of the latter occured. More detailed observations can be made by comparing the response spectra of the reactor shaft base induced motions to those of the free-field excitation. For that purpose, the 2% damping pseudovelocity response spectra were computed in the 1- lOOHz frequency range for the excitations that were simultaneously recorded in the free-field and at the reactor base. These pseudovelocity spectra can be considered to reflect the amount of energy content that is present in the recorded motions at the various frequencies. The free-field and the reactor base response spectra of the triple blasts ground excitation records are shown in Figures 5 and 6. f / i vertical i 1 • I i - _ i } i il i I V J I- in \ Q J A V FREQUENCY Figure 5. "EOUENCY Figure 6. CHi3 Free-field (above) and reacto* base (below) pseudovelocity response spectra. 285 While in the first diagram the spectra of the corresponding horizontal motions can easily be compared, the second diagram shows the difference in the vertical free-field and reactor base motions response spectra. It clearly appears that: - The spectra computed from the horizontal motions at the reactor shaft basement are both well below that of the corresponding free-field excitations. - The same observation holds for the vertical excitation in the higher frequency range. Around 2Hz however the reactor base vertical motion spectrum exceeds the free-field one; two small peaks are noticed at 1,75 and 2,34Hz. These important observations indicate the activation of favourable dynamic soil- structure interaction effects: the thick reinforced concrete continuous foundation slab of the reactor containment building succeeds in remarkably attenuating the earthquake- like excitation levels: The horizontal excitation energies at the reactor shaft base show to be drastically attenuated over the whole frequency range in comparison to the free- field excitation and a considerable vertical vibration energy cut off is achieved above 3,12 Hz; below the latter frequency, however, the excitation energy of the reactor base is somewhat amplified with respect to the free-field one. 6. CONCLUSIVE CONSIDERATIONS The IAEA promoted dynamic testing investigation of the Paks NPP site by means of buried explosions-induced ground motions has provided a large amount of interesting data on the structural response of the plant's major constructions. The technique used by the Hungarian mining specialists for carrying out the underground explosions actually succeeded in producing an earthquake-like excitation of rather low but quite well measurable intensity. High quality digital data acquisitions were made by means of the ISMES dynamic measurement instrumentation and data acquisition system. A first series of analyses of the experimental data has recently been performed for examining the free-field excitations that were actually produced during the blast- induced ground shaking experiments and interesting information on the actual dynamic soil-structure interaction effects could be infered for low level seismic-like excitation. A further detailed analyses task has still to be conducted for extracting information on the structural characteristics and behaviour of the Paks NPP major constructions. For determining the actual modal characteristics (fn, 286 TWIN CHIMNEYS' TIP MEASUREMENT STATIONS . laterall Figure 7. Energy auto-spectral densities Figure 8. Energy cross-spectral densities of of the twin chimneys' tip transverse and the twin chimney's tip transverse responses longitudinal responses From the above-reported diagrams, it can be concluded that: - The first two longitudinal bending resonance frequencies of the twin chimneys are at 1,97Hz and 3,2Hz, with a further minor resonance frequency located around 4,6Hz; - The first two synchronous lateral resonance frequencies of the chimney stacks are at 2,07Hz and 4,73Hz, while the first alternate lateral motion resonance occurs at 3,37Hz. 7. REFERENCES [/I/] Bendat J.S., Piersol A.G. "Random Data: Analysis and Measurement Procedures", J. Wiley & S. 1971. 8. ACKNOWLEDGEMENTS The positive attitude of the Paks NPP people (in particular: Dr. T. Katona, Chief Engineer and Dr. L. Turi, Head of the Experimental Section) during the preparation and execution of the above-described tests is gratefully acknowledged. Special thanks also to Dr. I. Sziics for the collaboration in the design of the multiple blasts experiments. NEXT PAGE(S) left BLANK 287 XA9952661 C.A. PratO, Address: San Eduardo 151. B°.Jardin Espinosa, 5014 Cnrdoha. Argentina Phone/fax: >54 51 6901II or 644320 FULL SCALE DYNAMIC TESTS OF ATUCHA II NPP T. Konno, Kajima Corporation, Japan S. Uchiyama, Kajima Technical Research Institute. Japan L.M. Alvarez, ENACE S.A., Argentina A.R. Godoy, International Atomic Energy Agency M. A. Ceballos and C.A. PratO, National University of Cordoba. Argentina ABSTRACT: This paper summarizes the main results of a series of dynamic tests of the reactor building of Atucha II NPP performed to determine the dynamic properties of its massive structure deeply embedded in Quaternary soil deposits. Tests were performed under two different types of loading conditions: Steady state harmonic loads imposed by mechanical exciters and impulsive loads induced by dropping a weight on the ground surface in the vicinity. Natural frequencies and mode shapes were identified and the associated modal damping ratios were experimentally determined. Numerical analyses of the reactor building-foundation system by two different F.E. models were performed. One of them, based on an axisymmetric representation of the soil-structure system, was used to simulate the steady state vibration tests and to calculate the dynamic stiffness of the foundation slab and soil layers for comparison with those experimentally obtained. The other, a 3-D F.E. model of the superstructure, was used to assess the natural frequencies and mode shapes obtained from the tests, representing dynamic stiffness of the foundation with stiffness coefficients derived both from the tests and from the axisymmetric F.E. model. Good agreement of the natural frequencies given by two types of tests was generally found, with the largest difference between them in the fundamental frequency of the building. Estimates of modal damping derived from the tests showed significant differences depending on the technique used to calculate them. For the fundamental mode damping was found to be 23 - 42 %, gradually decreasing with frequency to 2 - 4 % for around 10 Hz. 1. INTRODUCTION Presently under construction, the Atucha II NPP is located on the Parana River approximately 120 km north of the city of Buenos Aires. It is provided with a 745 MW pressurized water reactor fueled with natural uranium. The civil construction of buildings was completed at the time the dynamic tests were performed (November 1993). The plant site includes another smaller unit (Atucha I) in operation since 1974. Although the plant site is located in an area of very low seismicity, its reactor building was selected to perform the tests as part of a cooperative international agreement involving the Owner (Comision Nacional de Euergia Atomica - CNEA), the Engineer (Empresa Nuclear Argentina de Centrales Electricas - ENACE S.A.), the Kajima Corporation of Japan and the National University of Cordoba, Argentina. The main motivations for the test programme were: i) Collection of experimental data on dynamic characteristics of full-scale structure systems controlled by foundation soil properties; ii) Assessment 289 of the accuracy of currently used analysis models to capture the main dynamic characteristics of complex soil-structure systems, and iii) To obtain valuable data for seismic qualification of the plant. Of particular interest was to measure damping ratios for all modes with significant modal mass. The soil deposits in which the reactor building is founded and partially embedded are soil layers of silty clays and medium/dense sands. At the time the tests were performed the civil construction was completed but installation of components and systems had not yet started. The reactor building is a massive reinforced concrete construction with a total weight at the time of the test of approximately 125,000. tons. The plant layout and nomenclature is given in Figure I. The base slab of the reactor building provides support to two functional units: i) The containment structure with its internal reinforced concrete structure (UJA), and ii) The external building that supports and surrounds the contaiment structure (UJB). Foundation level is approximately 20 meters below natural grade, with an average soil stress due to permanent gravitational load of approximately 450 KN/m2. Other buildings adjacent to the reactor building (Auxiliary Building - UKA, Service Building - UFA and Administration Building - UBA) rest on independent foundations. Low amplitude dynamic tests performed in the laboratory on cored soil samples obtained from a borehole located adjacent to the reactor building give a soil shear stiffness consistent with shear wave velocities ranging from approximately 280 to 480 m/s for the top soil layer (60 m. thick). 2. DESCRIPTION OF THE TESTS 2.1 Steady state vibration tests In these tests the external load was applied by means of a mechanical exciter installed in areas of massive concrete in order to achieve overall response of the structure. Similar tests of large scale models founded on soils and of full scale structures founded on rock have recently been reported [I], [2], [3], and [4]. The exciter was installed at three different locations in the structure. As shown in Figures 2.a and 2.b, they were selected taking into account that the reactor building structure is almost symmetric in plan about the X axis. Three loading cases, one for each position of the exciter, were carried out in order to generate structural response along the horizontal X and Y directions according to the following programme: Case 1: The exciter was installed as close as possible to the X axis at elevation GL + 18.80 m above ground level, and the force exerted in the X direction. Case 2: The exciter at elevation GL + 0.50 m close to the X axis, and the force exerted in the X direction. Case 3: The exciter at elevation GL + 18.80 m close to the Y axis, and the force exerted in the Y 290 direction. Cases 1 and 2 were intended to provide some redundancy in the test data and to determine the sensitivity of the results to the location of the exciting force, since modal frequencies and shapes are intrinsic properties of the structure independent of the point of application of the load. The exciters were capable of generating a stationary harmonic horizontal force with a controlled frequency ranging from 1 to 20 Hz in increments of 0.10 Hz. The load amplitude vs. frequency is given in Figure 4. The M1K (Muto, Ishii, Kajima) measuring system was utilized to record and to process the data obtained from the tests. Steady state response was measured by means of displacement sensors installed at representative locations indicated in Figures 2.a and 2.b. Other sensors not shown in the figure were located at the neighboring turbine building foundations and at the natural soil surface up to a distance of 200 m from the reactor building. The exciter and a typical displacement sensor are shown in photographs of Figure 3. The sensors can be set to a natural period of either 1 or 7 seconds. They exhibit linear response characteristics above a certain frequency that depends on the natural period selected for the sensor; typically for an instrument set to a natural period of 1 second response is linear above 1 Hz. Since the frequency of the tests was equal or larger than 1 Hz, the natural period of all the sensors was set at 1 second for which their sensitivity is higher than for 7 seconds. At each sensor location the response of the structure to the harmonic load generated by the exciter was defined through the amplitude and phase angle of the calculated Correlation Function between the sensor output and the signal from the exciter force. This procedure, outlined in Figure 5, is repeated for each frequency to define the resonance curves for all sensors. Corrections for the measured phase angle between the exciter force and sensor response were introduced by the M1K system in accordance to the specifications of each sensor. 2.2 Impulsive load tests To complement the test programme using mechanical exciters with stationary harmonic loads described above, a set of impulsive load tests was performed to take advantage of all measuring equipment installed. The excitation was introduced by dropping a concrete weight with a crane on natural soil grade (1.6 tons from about 6 meters) at a distance of approximately 80 m from the reactor building at the position indicated iu Figure 1. The photograph of Figure 3 shows the crane and concrete block used for this purpose. Structural response was recorded at selected locations of the inner concrete structure shown in Figure 2.c, and at the free field on the ground surface near the point of impact as shown in Figure 1. 291 The nomenclature used to designate the sensors used for these tests is given in Figure 2.c. The displacement records were then processed to identify dominant frequencies and damping values. 3. TEST RESULTS 3.1 Steady state vibration tests Typical frequency response curves for the steady state vibration tests are given in Figures 6, 7 and 8. The natural frequencies of the system, as derived from the peaks of these curves are given in Table 1 for the three loading cases indicated before. As expected, it can be observed that the frequencies at which the peaks of response occur are independent of the position of the exciter (load cases 1 and 2). The fundamental mode involves deformations predominantly associated with rocking of the superstructure on the foundation and surrounding soil. A very important feature of the forced vibration tests is the possibility to determine the dynamic stiffiiess (or flexibility) of the foundation on the assumption that the base slab and lateral walls of the buried part of the building remain rigid in the range of frequencies of interest. For each frequency, the exciter load is known and the inertia forces can be calculated from the displacements values recorded at measuring points; amplitude and phase angle (relative to the exciter force) of the displacements at representative locations throughout the structure allow calculation of the real and imaginary parts of the inertia forces in the structure. Figure 9 contains the dynamic stiffness coefficients for rocking and horizontal sway motions, Krr and Khh respectively, derived froni test results for each of the three locations of the exciters. Mode shapes and frequencies of the building were identified from both the peaks of amplitude and phase angles. The frequencies of the first modes are given in Table 1. Damping ratios were calculated for the combined soil-structure system from displacement records by the half power method. At the fundamental frequency the damping ratio was found to be approximately 42 % of critical, whereas for higher modes was in the range of 2 to 2.5 % . Damping of the foundation alone was also derived from the foundation stiffness coefficients. At 2.9 Hz the damping ratio for the sway mode was found to be 54 % and for the rocking mode 30 % . 3.2 Impulsive load tests Typical displacement records from the impulsive load tests are given in Figure 10. Records of the free field vertical displacements of the ground near the point of impact (Figure 11) exhibit peaks of amplitude below 2 Hz, but relatively small variation of amplitude in the range of interest (2 to 12 Hz). Since the dynamic input at the foundation level is not known in these tests, the readings of all 292 sensors were normalized before processing without loss of information. Interpretation of the impulsive load tests was carried out as follows. The significant part of the transient displacement records was sampled with a sequence of windows of 1.3 s duration shifted in time at regular increments of 0.05 s. The domiuant frequencies of the Fourier Transform Amplitude of records can be considered an approximation to the natural frequencies of the system. Alternatively, the natural frequencies can be identified with the minima of the phase angle dispersion coefficients [7J calculated for all sensors using evenly spaced sampling windows. Typical curves representing the variation of phase dispersion as function of frequency are given in Figure 12. Mean values of the natural frequencies derived by this technique are given in Table 1. Modal damping values derived from the impulsive load tests were calculated through the amplitude decay with time for all frequencies identified as natural frequencies. The rate of amplitude decay was obtained by best fitting exponential curves to the time evolution of FT amplitude as shown in Figure 13. Sketches of the mode shapes for the low frequency modes are given in Figure 14. There it is shown that for the natural frequency identified at 5.47 Hz the base slab undergoes negligible rotation and deformations. This frequency is in close agreement with the fixed- base fundamental mode of the complete reactor building structure obtained with a 3-D F.E. model. 4. ANALYTICAL MODELS Several numerical models of the reactor building soil-structure system were developed to provide an assessment of their capacity to predict field data derived from the dynamic tests. One approach was to develop an axisymmetric F.E. model for the structure and foundation soil, analyzed with the ASHSD2 computer program [5] to simulate the steady state vibration tests. A sketch of the model is given in Figure 15. The shear wave velocities of soil strata used for this model ranged from 281 m/s to 482 m/s. Viscous boundary at the base, and transmitting boundary at the sides [6] were assumed in the F.E. mode) shown in Figure 15. A comparison of experimental and calculated response of the structure to exciter harmonic load is given in Figure 16 for various locations in the reactor building. Another approach to represent the the superstructure was to develop a 3-D F.E. model of the reactor building above the fondation slab with approximately 12000 d.o.f. This model of the superstructure above the base slab was analyzed with different support conditions at the base. One of them was to assume all nodes of the base slab to be fixed. Under this assumption the fundamental frequency of the structure is found to be 5.6 Hz, which is in close agreement with experimental results of the impulsive tests, where for 5.47 Hz the base slab appears to have almost no 293 displacements due to rocking and horizontal sway (Figure 14). The 3-D model was also analyzed replacing the fixed-base boundary conditions by spring supports applied at the base slab (assumed to be rigid), while global vertical response of the building was prevented by suitable restrictions. Two sets of spring constants were used: i) Those resulting from the steady-state vibration tests (Figure 9), for which the fundamental frequency of the soil-structure becomes 3.35 Hz, and ii) Those derived from foundation stiffness calculations by F.E. axisymmetric soil model, for which the fundamental frequency of the system is 2.46 Hz. Doth estimates are in the range of measured values (2.7 to 3.5 Hz). The difference between the calculated values of the fundamental frequency can be attributed to uncertainties in the dynamic properties of the soil layers used in the analysis to derive the spring constants, since the fundamental frequency of the structure resulting from tests and analysis for the fixed base condition are in very close agreement (5.47 and 5.6 Hz). CONCLUSIONS Valuable data related to the dynamic characteristics of the Atuclia H Reactor Building were obtained both from the tests and analytical models, and the following conclusions may be derived from them: 1. Natural frequencies of the reactor building were identified by two different types of tests (steady state and impulsive), and good correspondence between them was found within the range of interest. The largest difference between them was in the value of the fundamental frequency of the soil-structure system due to the lack of a single peak of response that could be assigned to this mode. 2. Damping ratios were measured for the relevant modes of the structure. They were found to be largest for the fundamental mode, gradually decreasing for higher frequency modes. Damping ratio of the fundamental mode was found to be higher than damping cut-off limits widely used for design of nuclear power plants in conjunction with modal response calculations. 3. The dynamic stiffness of an equivalent rigid fundation was derived directly from the tests measurements and assessed with results from analytical models. Significant differences between experimental and analytical results attest to the shortcomings of assigning dynamic soil properties for numerical analysis by means of laboratory tests of cored samples. Further analysis using more accurate soil properties at the site remains to be done to finalize the comparison of test results with those of analytical models. 4. Results obtained from the tests can be used for seismic qualification of Atucha II NPP, and may provide relevant information for seismic design of nuclear facilities founded on similar Quaternary soil deposits. 294 5. Further processing of the test results may still lead lo obtain more refined results regarding the dynamic properties of the Atucha II reactor building. ACKNOWLEDGEMENTS This paper is a summary of the main results of the Atucha II dynamic tests presented at 13 SMIRT entitled "Full Scale Vibration Tests of Atucha II NPP," Parts 1 through VIII. The authors would like to acknowledge the contributions of their colleagues: Tsugawa,T., Mnsucln, K. and Maeda, T. of Kajima Corporation, Naito, Y. and Ohno, S. of Kajima Technical Research Institute, Sala, G. of CNEA, and Friebe, T.M. and Capelli, P. of ENACE S.A., Halbritter, A., Krutzik N., and Schiitz, W. of Siemens A.G. and to the organizations participating in this research project. REFERENCES [1] Morishita, H. et al., "Forced Vibration Test of Hualien Large Scale SSI Model," 12 SMIRT, Div. K, pp. 37-42, Stuttgart, August 1993. [2] Kurimoto, O. et al., "Field Tests on Partial Embedment Effects (Embedment Effect Tests on Soil-Structure Interaction)," 12 SMIRT, Div. K, pp. 43-48, Stuttgart, August 1993. [3] Morishita, H. et al., "Study on Vertical Seismic Response Characteristics of Deeply Embedded Reactor Building," 12 SMIRT, Div. K, pp. 61-66, Stuttgart, August 1993. [4] Katona, T. et al., "Dynamic Response of VVER-440/213 PAKS NPP to Seismic Loading Conditions and Verification of Results by Natural Scale Experiments," Proceedings of Seminar 16, 12 SMIRT Post Conference Seminar, pp.535-568, Vienna, August 1993. [5] Ghosh, S. and Wilson, E., " ASHSD2 - Dynamic Stress Analysis of Axisymmetric Structures under Arbitrary loading," EERC Report 69-10, revised September 1975. [6] Berger, E., Lysmer, J. and Seed H.B., " ALUSH - A Computer Program for Seismic Response Analysis of Axisymmetric Soil-Structure Systems," EERC Report 75-31, 1975. [7] Ceballos, M.A., Prato, C.A. and Alvarez, L.M., "Experimental and Numerical Determination of Dynamic Properties of the Reactor Building of Atucha 11 NPP," Proceedings of Seminar 16, 13 SMIRT Post Conference Seminar, Iguazii, August 1995. 295 Impact Point UJA/UJB 0 Reoctor Building UKA UMA Auxiliary Building Turbine Building > a. a> I. ? to' UBA Figure 1. Atucha II Plant Layout Exciter Y 18.8m Excitation 180c X 18.8m Excitation 270° (a)18.8m-lcvcl 90° 180c nr- X 0.5m Excitation (b) ().5m-levcl Figure 2a. Location of sensors and of exciter 297 00 p.u. (Vertical) p.u. (Horizontal) -O- Exciter (+18.8m, +0.5m) X Direction Y Direction Figure 2b. Location of sensors and of exciter H2 Elevotion: + 10.10 m Elevotion: + 18.80 m V1 H6 Elevation: - 18.60 m Elevation: + 0.50 m +27.00 m + 18.80 m +10.10 m + 0.50 m — 6.60 m -18.60 m 977777777777777777777777777777777777, References: Vertical Sensors Horizontal Sensors Figure 2c. Location of sensors for Impulsive load tests 299 Photo I Exciter Photo 2 Silualion of Drop Test Photo 3 Horizontal sensor Figure 3. Forcing Devices and Sensors 300 Exciting force lton LLLL _L 1. 0.0.1 1.0 5.0 10.0 20.0 Frequency [Hz] Figura 4. Exciter force-frequency relation EXCITER CROSS CORRELATION FUNCTION Cl>g(r) AMPLITUDE: A. PHASE: 0 Figura 5. MIK System flow chart for Steady-state vibration test 301 Level: -18.60m - Direction X (Horizontal) 0. 2 -1—i—i—i—I—i—i—r i r Q. 0. 1 0) IM ID I i i i i i i I i i i i 0 i i i i ) B 0 9 0 ro - 9 0 - 1 8 0 >—'—' •' Frequency [Hz] Level:-18.60m - Direction Z (Vertical) 0. 3 0. 2 rsi — 0. I O 2= v 0 i I i i i r i 1 8 0 9 0 TO _l 0} to TO ^: -9 0 Q. - 1 8 0 \ 0 i 5 2 o Frequency [Hz] Figure 6. Response of Base Slab 302 Level: Top - Direction X (Horizontal) ' ' I 1 N E O 0 1111 1 8 0 I I l 7VT 1) 9 0 ra -9 0 - 1 8 0 i i i i y\l li u» i i I t 0 I 5 2 o Frequency [Hz] Figure 8. Response of Top of Pressure Containment Vessel 303 xlO7K (t/m) xlO'°K (t m/rad) 10.0 2.5 R ' j 8.0 - 2.0 - J I 6.0 1.5 1 t "6' ~~ < P 1.0 Im. 0.51- I """ V 1 I 1 1 0.0 0 1 4 5Hz 6 x+18.8 7 10 xlO K (t/m) x!0 Kp(t m/rad) 10.0 2.5 o \ cP 8.0 2.0- .p 6 6.0 1.5- P. 0 Re. o 4.0 1.0- > 2.0 0.51- t.fl. Im rf* 1 *\ ( ; 0.0 0.0 i 5 Hz 6 0 1 4 5 Hz 6 x+0.5 xl0luK (t m/rad) 2.5 B a 2.0 6... P 1.5 Re. 1.0- - c O 0.5- A Im. j O V -10.0 0.0 T Hz 9. Dynamic stiffness of fiindation derived from Steady-state vibration tests Test 1 - Sensor H3 Test 2 - Sensor 113 3 i I 0.5- ,L Jj_ n 0.5- J\ 1 i2 \ Mi 0 0- V A I \ n \. di s ^\ fc -0.5 -0.5- Vil o I ! Norr r 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1 5 2 2.5 3 3.5 4 Time [sec] Time [secj Test 1 - Sensor V6 Test 2 - Sensor V6 i I 0.5 i 0.5- v displ. , disp l 0 «, A (\r-J* \- - j-~u -«. - 0- r m -0.5- -0.5- A 0 V Nor m f 0 0.5 1 1.5 2 2.5 3 3.5 *[ 0 0.5 1 1.5 2 2.5 3 3.5 4 Time [sec] Time [sec] Figure 10. Typical displacement record from Impulsive load tests Test 2 - Sensor VI Tesl 2 - Sensor VI \y V~^^ 0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Frequency (Hz) Figure 11. Free field vertical displacement in Impulsive load tests 305 Sensor H3 Sensor V6 [$\ ft 0.8 J £ 0.6 c '-o S 0-4 a l\i 1 A II x: a 0.2 l|s lilM / 5.60 6.68 8.23V 10.08x7 "^ 0 1 3 4 5 6 7 8 9 10 11 12 0 1 23456789 10 11 12 Frequency [Hz] Frequency [Hz] Average for all sensors 0.8 A ft a °.6i CO T3 ffi 0.4- \ I (0 V/8JJ8 f lv 7.62 10.31 V 0.2 6.49 11.25 2.72 w 0 01 23456789 10 11 12 Frequency [Hz] Figure 12. Recognition of natural frequencies by phase-dispersion Table 1. Measured natural frequencies Natural Frequencies [Hz] Mode 1 2 3 4 5 6 7 8 Impulsive Tests 2.72 4.42 5.47 6.49 7.62 8.38 10.31 11.25 Steady State Tests X 18.80 m 2.9 4.5 5.9 7.3 9.2 10.7 11.5 X 0.50 m 2.9 4.5 5.9 7.3 9.0 10.5 11.4 Y 18.80 m 2.9 4.5 6.2 6.9 7.9 9.1 11.2 Average 2.9 4.5 6.0 6.9 7.5 9.1 10.6 11.4 Table 2. Measured damping values from Impulsive load tests Modal Damping Ratios [%] Mode 12 3 4 5 6 7 8 2.72 Hz 4.42 Hz 5.47 Hz 6.49 Hz 7.62 Hz 8.38 Hz 10.31 Hz 11.25 Hz 22.97 I 14.51 | 12.71 7.69 7.84 8.19 4.87 4.30 306 Sensor 113 - Mode 1 : 2.72 Hz Sensor IB - Mode 2 : 4.42 IIz 3- 3- \ 2.5- — / \\ 2 2- / dis p \ 1.5- \ 1.5- \ 1- 1- 0.5- 0.5 F T o f norm , dls p F T o f nor m c) 0.2 0.4 0.6 0.8 1 ) 0.2 0.4 0.6 0.8 1 Time [sec] Time [sec] Sensor 113 - Mode 3 : 5.47 Hz Sensor 113 - Mode 4 : 6.49 Hz 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 Time [sec] Time [sec] Sensor 113 - Mode 5 : 7.62 Hz Sensor H3 - Mode 6 : 8.38 Hz 3 3 2. 25 2-5 I 2 c 15 E 15 I 1 "o t 05 0 0.2 0.4 0.6 0.8 0.2 0.4 0.6 OB Time (see] Time [sec] Sensor 113 - Mode 7 : 10.31 Hz Sensor 113 - Mode 8:11.25 I Iz 3- — 2.5 2-5- / \^ 2 1.5- 1 5 "~~~-\ i - 1 | 1 I- 1 o f nor m 0 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 Time [sec] Time (sec] Figure 13. Modal damping from Impulsive load tests 307 MODE 1 G-» MODE 2 («.« HI) MODE 3 (s.«7 HI) MODE 4 («•« MODE 5 C7.62 m) MODE 6 (fi-M HI) MODE 7 <'OJ1 HI) MODE 8 (11-25 Hi) Figure 14. Sketch of modal shapes of the Internal structure from Impulsive load tests 308 Elastic properties of soil profile Dcpih Thickness Vs Poisson's Unii Mass (GL-m) (m) (m/scc) Ratio (ton sVm') 0-6 6.0 272. 0.35 0.19 6-13 7.0 229. 0.35 0.19. 13-20 7.0 261. 0.35 0.19 GLOm 20-26 6.0 281. 0.35 0.19 26-33 7.0 308. 0.35 0.19 33-41 8.0 344. 0.35 0.19 41-50 9.0 417. 0.35 0.19 50-60 10.0 482. 0.35 0.19 Elastic properties of the structure materials Young's Modulus Poisson's Unit Mass Thickness Part 2 4 UJ (lAn ) Ratio (ton sVm ) (m) R/B 3.5x 10' 0.2 0.2447 0.6 PCV 2.1x 10' 0.3 0.7970 0.3 Base Mat 3.5x 10s 0.2 0.2447 2.8 Figure 15. Axisytnmedic F. E. Model (c) Top of 1/C X10 "rad/ion (d) Operating Floor (c) Base Mat (0 Rocking of Base Mat Figure 16. Response and phase lag curves by simulation analysis F NEXT PAGEIS)! 8 left BLANK 3 309 EXPERIMENTAL AND NUMERICAL DETERMINATION OF THE DYNAMIC PROPERTIES OF THE REACTOR BUILDING OF ATUCHA II NPP M. A. Ceballos, E. J. Car, T. A. Prato and XA9952662 C. A. PratO, National University of Cordoba, Argentina L. M. Alvarez, EN ACE S.A., Argentina, and A.R.Godoy, IAEA ABSTRACT Determination of the dynamic properties of the reactor building of Atucha II NPP is carried out in order to: i) Obtain valuable information for seismic qualification of the plant, and ii) Assess some procedures for testing and analysis that are used in the process of seismic evaluation of existing nuclear facilities founded on Quaternary soil deposits. Both steady state and impulsive dynamic tests were performed but attention is centered here in the techniques used to determine natural frequencies and modal damping ratios with impulsive tests. Numerical analyses were performed by means of a 3-D model model of the superstructure together with foundation stiffness coefficients derived in a separate paper from steady state vibration tests, and also from analysis with a 2-D F.E. model of the soil layers capable of approximating the 3-D features of the problem. The computed foundation stiffness coefficients are compared both with those obtained from the tests and from an axisymmetric F.E. model; results indicate that foundation stiffness coefficients calculated with F.E. models with soil parameters given by laboratory tests performed on cored samples are significantly lower than those given by the steady state vibration tests. I. INTRODUCTION Seismic evaluation of existing nuclear facilities, currently under way following national and international guidelines, normally requires as a starting point a reliable estimate of the dynamic characteristics of the structure-foundation system, and of the electromechanical systems and components. In contrast to the seismic design of new facilities, their as-built properties can be determined both by tests and analysis, and the results of the two approaches can be used to validate the assumptions adopted for analysis. As part of a seismic evaluation programme, small amplitude vibration tests of full scale structures are one of the alternatives to estimate the dynamic characteristics of structure-foundation systems The size and complexity of nuclear power plants is often such that considerable resources are required to perform these tests. In order to obtain as much data as possible from the tests of Atucha II NPP, several concurrent objectives and redundant paths to achieve them were formulated while planning the test programme. One of the main participants of the Atucha II vibration test programme has performed, cooperated and participated in other similar tests of full scale nuclear power plants and large scale models. For that purpose it has developed measuring techniques, software and equipment to record, process and interpret steady state vibratory tests by means of mechanical exciters, providing the basic 311 experience to achieve the proposed objectives of the test programme. The results of these tests for Atucha II NPP and other plants have recently been reported [1], [2], [3], [4], [6] and [7]. Since a very significant part of the total effort involved in the tests is related to installation and control of the equipment, it was decided to use the same instrumentation to record response of the structure to impulsive loadings. Ground explosions have been used as impulsive excitations to determine dynamic properties of the VVER type nuclear power plants, as part of a comprehensive effort to evaluate seismic capacity of the main building complex [5]. In the case of Atucha II NPP the use of explosives was not acceptable due to congestion of the construction site at the time of the tests. Other methods to generate impulsive actions, such as the sudden release of pressure from pressurized cylinders, airguns or similar devices were also considered, but were later discarded in favor of the simpler alternative of impacting a weight on the natural ground near the reactor building. A 3-D F.E. model of the superstructure was developed and analyzed under different assumptions for the foundation stiffness, and natural modes and frequencies were determined for the soil-structure system. Dynamic stiffness coefficients of the foundation were determined by means of a 2-D F.E. representation of the horizontally layered site, adapted to account for the three dimensional nature of the problem. Results for this model were then compared with those given by an axisymmetric model [6] and by direct reduction from the steady state vibration test results [7]. 2. TEST SEQUENCE The impulsive excitation was introduced by hoisting a concrete block of 1.6 tons with a crane, and dropping it on the ground from approximately 6 m. The magnitude and location of impact was selected and assessed in the field with a series of preparatory shocks to set the scales of the instruments. The location of the point of impact, indicaded in Figure 1, is contained in a meridional plane at 45° with the X,Y axes. Both horizontal and vertical displacement sensors were installed to measure displacements of the concrete structure at the base slab and at higher elevations. The horizontal displacements were measured in the meridional plane of impact. Even though more recording channels were available, restrictions in the sequence of the other tests performed at the time did not allow a more complete instrumentation. Location of instruments is given in Figure 2. After all instruments were in place, preliminary tests were performed to select the scales of the recording equipment prior to the final tests. The complete sequence of main shocks, six in total, was carried out in a few minutes; Absolute displacements of the structure were measured with inertial displacement sensors set at a natural period of 1 second, so that their transfer functions are flat for 312 frequencies higher than 1.4 Hz., with a sensitivity of 1 urn; the lower sensitivity for frequencies below 1.4 Hz did not impair identification of the natural frequencies of the structure, all above 2 Hz. It is of interest to note that, in spite of the large total mass of the structure, the limited energy of the excitation was sufficient to generat" clear and distinctive signals in the displacement sensors, with sufficient amplitude to render acceptable signal to ambient noise ratio. In fact, as shown in Figure 3 for sensors H3 and V6, the recordings of response at any given location for two successive shocks were very similar in amplitude and frequency content. Moreover, as previously pointed out, the frequency content of the incident wave generated in the soil by the impact, as shown in Figure 3 for sensor VI, was uniformly distributed in the range of 2 to 12 Hz. Thus, even though the structure response is strongly affected by deformations of the soil, particularly at the fundamental frequency of the soil-structure system, the peaks in the frequency distribution of response is a direct consequence of the selective amplification of the excitation waves by the natural modes of the combined system. 3. INTERPRETATION OF TEST RESULTS Duration of the significant part of the structural response to the impact loads was typically two to three seconds as depicted in Figure 3. Some background noise was present in the records but its amplitude was not significant relative to the signals due to impact. Absolute displacement records shown in Figures 3 and 4 were normalized with respect to the maximum value at one particular sensor and impact test; this normalization did not imply loss of information since the exact nature, distribution and intensity of the excitation transmitted by the soil to the structure is not known, and is not required for interpretation of the results. The width of the sampling window used to analyze the records was selected according to the following criteria: i) It should be sufficient to capture the lowest dominant frequency of response, estimated at 2.5 Hz, with a minimum of 3 cycles within the sampled segment; ii) The window should allow analysis of the evolution of amplitude and frequency content of response along the 2 to 3 seconds total duration of the records; and iii) A silent zone should be added to all records in order to increase frequency resolution. With these conditions a Hanning window of 1.3 second width was adopted and a final silent zone up to 20 seconds was added to all records. A basic condition to identify the natural frequencies of the soil-structure system from the frequency content of the transient records is that the distribution of amplitude of the excitation with frequency be as uniform as possible within the frequency range of interest. This assumption is reasonably satisfied as indicated by the free field displacements given in Figure 3. A Fourier analysis of the total length of these records indicates that the dominant frequency components were all above 313 2 Hz. The amplitude of the FT of the free-field ground motions measured at a location between the point of impact and the reactor building shows a smooth and almost uniform frequency distribution in the range of interest (2 to 12 Hz) so that the peaks of the frequency distribution of structural response can be attributed to the natural frequencies of the system. The Fourier amplitude spectra of Figure 4 were obtained by sliding the sampling window along the time axis for sensor H3. The lowest frequency peaks of each curve exhibit some frequency drift from 2.5 to 3.2 Hz; this drift is less pronounced for the other peaks. An alternative way to determine the dominant frequencies of response is proposed here through the phase angle of the Fourier Transform of the signals sampled with a series of windows equally spaced in time. A norm of the phase dispersion (PD) of the FT is proposed through the following expression: n- ] n ] v—i ^—i / \2 / \ 2 (sirtp^ - sin + |COS(p.. - COScp, ) m *>ik) > ~2 2_, \ 1J '/ n j=i k=j + l (0 where: /' is the frequency at which PD is evaluated (po is the phase angle at frequency / of the/th sample of the signal (pit is the phase angle at frequency / of the Ath sample of the signal n is the total number of windows considered Notice that PDi varies from 0 (no dispersion at that frequency) to 1 (maximum dispersion at that frequency). The dominant frequencies are associated with the relative minima of PD as shown in Figure 5 for individual sensors (top) or for the summation of the PD of all sensors (bottom). The latter values are given in Table 1 together with the frequencies obtained from the steady state vibration tests [7]. In this procedure dominant frequencies are implicitly assumed to remain constant during the transient process. The authors are not aware of a previous similar application of the phase angle variation with time to define the dominant frequency components. To calculate the damping ratio for each of these frequencies the following technique was implemented. The amplitude of response was obtained for each of the sampling windows, and plotted as function of time. Since the duration of the excitation is very short as shown in the record of the free field on the ground surface, it may be assumed that the amplitude after the first peak of response corresponds to free vibration of the system. The damping ratio is directly related to the rate of decay of the amplitude after the first peak. This technique was assessed by aplying it to calibration signals of known characteristics. These tests showed that this method is reliable when the dominant frequencies follow the pattern given in 314 Table 1, but is Jess accurate when closely spaced frequencies and high damping ratios are involved. Damping ratios found are given in Table 2. Their values vary with location and frequency, from 23 % for the fundamental mode to 4 % for the eighth mode. Figures 6 and 7 contain typical results of this analysis. Figure 8 shows sketches of the mode shapes associated with the frequencies given in Table 1. Horizontal displacements correspond to the meridional plane that contains the point of impact. These results indicate that the first two modes involve dominant rocking motions, while the third mode (5.47 Hz) shows dominant horizontal sway displacements of the superstructure and negligibly small displacements of the base slab. 4. NUMERICAL ANALYSIS The 3-D F.E. model developed for the superstructure, with approximately 12.000 d.o.f, is shown in Figure 9. Most of the structural parts were represented by cuadrilateral elements with membrane and bending stiffness; the base slab and central sections that support the spherical steel containment were modelled with eight-node brick elements. The purpose of this model was to determine the natural frequencies and mode shapes having a modal mass of at least 1% of the total mass of the structure, using different support conditions at the base. One condition was to assume the base slab fixed, while the others represent the foundation by horizontal and rotational springs. Two sets of spring constants were considered; one of them was the foundation stiffness coefficients as derived from the steady state vibration tests [7], while the other was obtained with a 2-D F.E. model of the foundation. In both cases the static stiffness coefficients were used in the analysis since only real-valued eigenvalues of the complete model were to be determined. Even though it has long been recognized that 2-D models have inherent limitations to represent 3-D dynamic behaviour of rigid foundations as discussed in [8] and [9 ], the dynamic stiffness coefficients of the foundation of the Atucha II reactor building embedded in a horizontally layered site were calculated with an alternative plane 2-D F.E. model capable of approximating the three dimensional nature of the problem. Axisymmetric models to represent 3-D behaviour by means of a 2-D description of the soil if the geometry of the foundation and of the lateral boundaries of the soil are axisymmetric, are also widely used for dynamic soil-structure analysis. These two approaches were applied for the case of Atucha II in order to compare the amount of radiation damping of the foundation given by both analysis models with that obtained from the steady-state vibration tests. Figure 10 depicts a F.E. discretization of a vertical plane of the soil at the site. The distribution of soil properties with depth given in the figure was derived by means of small amplitude laboratory 315 tests performed on cored samples. The thickness of the slice of this plane model is taken equal to the diameter of the foundation. To help represent the 3-D behaviour of the slice, two sets of distributed springs and dashpots are added to the lateral planes (parallel to the plane of the model) in order to account for: i) The static stiffness of the 3-D system; and ii) The radiation of energy in the direction normal to the plane of the model. Horizontal and vertical springs are taken so as to reproduce the static horizontal and rotational stiffness of the same rigid foundation placed on the surface of the uniform elastic half space. For the layered site these constants are taken directly proportional to the dynamic shear modulus of the layers. The two sets of springs generate an additional stiffness matrix that turns out proportional to the consistent mass matrix of the plane model, which is superimposed with the standard F.E. stiffness matrix. Horizontal and vertical dashpots are defined as the shear terms of a Lysmer-Kuhlemeyer boundary [10]. These elements generate a damping matrix also proportional to the consistent mass matrix of the plane F.E. model, that accounts for the variation of soil properties with depth. Figure 11 contains the horizontal and rotational stiffness coefficients for the Atucha II model shown in Figure 10; results are given as obtained with both the plane 2-D F.E. solution and with an axisymmetric model of the same system [6]. There it is shown that the real parts are in good agreement, while the imaginary parts exhibit somewhat larger differences. Figure 12 presents a comparison of the stiffness coefficients as derived from the forced vibration tests [7] with those of the plane 2-D F.E. model. There it can be seen that the results of both analysis models underestimate the imaginary part. In addition, a sign inversion of the experimental results suggests the existence of a natural frequency of the soil layers between 5 and 6 Hz which is not picked up by the analyses. These results confirm that the dynamic properties of soil layers as derived from laboratory tests from cored samples may differ from the actual properties in the field. The results of the natural frequencies of the 3-D model of the superstructure with the foundation stiffness as derived from tests analyses is given in Table 3. For the spring stiffness given by tests the fundamental frequency is 3.35 Hz and for the stiffness from analysis is 2.46 Hz. For the structure fixed at the base the fundamental frequency is 5.6 Hz, in close agreement with the tests (5.47 Hz). 5. CONCLUSIONS The series of tests and analyses presented in this paper provide valuable information regarding the dynamic characteristics of the complex soil-structure system of the reactor building of Atucha II NPP. Natural frequencies obtained from impulsive tests are in general good agreement with those 316 derived from steady state tests by means of mechanical exciters. For the fundamental frequency of the structure, results of the impulsive tests show some drift of frequency from 2.5 to 3.2 Hz with a mean value of 2.72 Hz, while the steady state vibration tests give a mean value of 2.9 Hz. Damping ratios extracted from the transient tests by measurements at the base slab and in the reinforced concrete internal structure show a tendency to decrease with increasing frequency. This is attributed to the larger participation of soil deformations relative to structural deformations for lower frequency modes. Estimates of modal damping values show considerable scatter from test to test, situation that is not unusual in attempts to measure damping values of structural systems of comparable complexity. Time and manpower required to perform and interpret the impulsive load tests as performed here, as well as the valuable data obtained from them, indicate that this test method is an expeditive and potentially useful technique to determine dynamic properties of complex structural systems. Analyses of the foundation stiffness performed with a plane 2-D F.E. model have produced a set of dynamic stiffness coefficients in general agreement with those derived by axisymmetric F.E. models, provided that appropriate boundary conditions are applied to account for the three- dimensional nature of the problem. However, when the dynamic properties of soils used for analysis are obtained from laboratory tests performed on cored samples, both types of analysis give results significantly different from those obtained through full scale tests. 6. ACKNOWLEDGEMENTS The authors would like to express their gratitude to the organizations supporting this project: Kajima Corporation of Japan, Comision Nacional de Energia Atomica (CNEA), Empresa Nuclear Argentina de Centrales Electricas (ENACE), Universidad Nacional de Cordoba, and the Science and Technology Research Council of the Province of Cordoba, Argentina (CONICOR). 7. REFERENCES [1] Morishita, H. et al., "Forced Vibration Test of the Hualien Large Scale SSI Model," 12 SMIRT, Div.K, pp. 37-42, August 1993. [2] Kurimoto, O. et al., "Field Tests On Partial Embedment Effects (Embedment Effect Tests on Soil-Structure Interaction)," 12 SMIRT, Div.K, pp. 43-48, August 1993. [3] Morishita, H. et al., "Study on Vertical Seismic Response Characteristics of Deeply Embedded Reactor Building," 12 SMIRT, Div.K, pp. 61-66, August 1993. [4] Tsutagawa, M. et al., "Seismic Verification Program of Nuclear Power Plants Quaternary Deposits in Japan," 12 SMIRT, Div.K, pp. 67-72, August 1993. 317 [5] Katona, T. et al., "Dynamic Response of VVER-440/213 PAKS Nuclar Power Plant to Seismic Loading Conditions and Verification of Results by Natural Scale Experiments," Proceedings of Seminar 16, 12 SMIRT Post Conference Seminar, pp. 535-568, August 1993, IAEA, Vienna. [6] Masuda, K., Maeda, T. and Uci.iyama, S., "Full Scale Vibration Tests of Atucha II NPP: Part IV Numerical Simulation of Steady-State Vibration Response by Axisymmetric FEM," 13 SMIRT, Div.J, Porto Alegre, 1995. [7] Uchiyama, S., Naito, Y. and Ohno, S., " Full Scale Vibration Tests of Atucha II NPP: Part II Interpretation of Tests Results for Steady-State Harmonic Forces," 13 SMIRT, Div.J, Porto Alegre, 1995. [8] Luco, J.E. and Hadjian, AH., "Two-Dimensional Approximations to the Three-Dimensional Soil-Structure Interaction Problem," Nuclear Engineering and Design, 3 1 (1974); pp. 195-203. [9] Wolf, J.P., "Foundation Vibration Analysis Using Simple Physical Models," PTR Prentice Hall, 1994. [10] Lysmer, J. and Kuhlemeyer, R.L.. "Finite Dynamic Model for Infinite Media," Engineering Mechanics Division Journal, ASCE, Vol.95, EM4, 1969, pp. 859-877. 318 UJA/UJB 0 Reactor Building u Figure 1. Atucha II Plant Layout. Location of Impact. 319 H2 Elevation: + 10.10 m Elevation: -i- 18.80 m V1 V2 V3 H6 II5 Elevation: — 18.60 m + 27.00 m + t8.80 m + 10.10 m + 0.50 m 77? - 6.60 m / / / /. ', -1B.60 m References: Vertical Sensors llorizontol Sensors Figure 2. Location of displacement sensors. 320 Test 1 - Sensor 113 Test 2 - .Sensor 113 Test 1 - Sensor V6 Test 2 - Sensor V6 0 0.5 1 1.5 2 2.5 3 3.5 4 0 05 1 1.5 2 2.5 3 3.5 4 Test 2 - Sensor VI Test 2- Sensor VI 0.75 -0.75 0 1 2 3 4 5 6 7 8 9 101112131415 Frequency [Hr| Figure 3. Displacement records of structure response (H3, V6) and free field (VI). Test 1 - Sensor 113 Tcsl 1 - Sensor 113 0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Frequency [Hz] Figure 4. Sampling and evolution of Fourier Amplitude Spectra. 321 Sensor 113 Sensor V6 01 23456789101112 0 1 2 34 5 67 89 10 11 12 Frequency (Hz] Frequency (Hz) Avcr:»gc of nil Sensors 01 23456789 10 11 12 Frequency [Hz] Figure 5. Recognition of natural frequencies by phase-dispersion. 322 Sensor 113 - Mode 1 : 2.72 II7. Sensor 113 - Mode 2 : 4.42 1 Jr. zL 2.5 0.4 0.6 0.8 0.4 0.6 Time [sec] Time Isec) Scnsor 113 - Mode 6 : 8.38 Hz Sensor 113 - Mode 8 : 11.25 11/. zL 2.5" zL 2.5- 1-5- 05 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 Time |sec] Time (sec] Figure 6. Modal damping from sensor H3 Sensor Vf> - Mode 3 : 3.47 11/. Sensor V6 - Mode 4 : 6.49 1 Iz o.a 0.A 0.6 0.8 0.2 0.4 0.6 0.8 Time (sec] Time (sec] Sensor V6 - Mode 5 : 7.62 I \z Sensor V6 - Mode 7 : 10.31 IIT-. 0.4 0.6 0.8 0.2 0.4 0.6 06 Time |secj Time |secj Figure 7. Modal damping from sensor V6 323 MODE 1 MODE 2 («••« "0 '/////A MODE 3 MODE 5 C-62 in) MODE 6 MODE 7 (>°Ji »") MODE 8 ("•" HI) Figure 8.' Sketch of modal shapes of the internal structure 324 Figure 9. 3-D Finite Element Model of Siiperstnictiire 325 Khh [ton/m] x106 Krr [ton.m/rad] x10 20 15 Imag. \ .••" i Reat^ 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Frequency [Hz] Frequency [Hz] 2-D F.E.M. Axisymmetric F.E.M. Figure 11. Foundation StifTness CoefTicicnts from Finite Clement Analysis Khh [ton/m] x 10 Krr [ton.m/rad] x 10 10 o 1 2 3 4 5 6 0 1 2 3 4 5 6 Frequency [Hz] Frequency [Hz] - Experimental 2-D F.E.M. Figure 12. Foundation Stiffness Coefficients Experimental vs. 2-D Finite Element Analysis 326 Natural Frequencies [Hz] Mode 1 2 3 4 5 6 7 8 Impulsive Tests 2.72 4.42 5.47 6.49 7.62 8.38 10.31 11.25 Steady State Tests X 18.80 m 2.9 4.5 5.9 7.3 9.2 10.7 11.5 X 0.50 m 2.9 4.5 5.9 7.3 9.0 10.5 11.4 Y 18.80 m 2.9 4.5 6.2 6.9 7.9 9.1 11.2 Average 2.9 4.5 6.0 6.9 7.5 9.1 10.6 11.4 Table 1. Measured natural frequencies. Modal Damping Ratios [%" Mode 1 2 3 4 5 6 7 8 2.72 Hz 4.42 Hz 5.47 Hz 6.49 Hz 7.62 Hz 8.38 Hz 10.31 Hz 11.25 Hz H 1 22.45 16.50 15.63 9.91 7.52 7.11 6.42 2.25 H2 22.74 12.54 9.99 6.40 7.93 9.46 3.30 2.69 113 23.38 13.92 14.54 10.46 7.48 7.60 2.30 1.45 H4 22.12 10.59 11.47 6.57 5.34 5.29 3.06 3.74 H5 21.53 10.92 9.17 6.85 6.49 6.44 3.75 3.53 116 24.03 14.13 9.23 5.05 5.97 11.54 4.90 6.53 V4 22.14 13.30 19.25 7.81 9.54 10.18 6.53 5.97 V5 22.28 14.69 18.82 10.12 7.34 9.57 7.03 5.72 V6 25.90 18.24 7.43 6.37 11.32 7.21 6.65 5.75 V7 23.11 20.24 11.56 7.38 9.47 7.48 4.79 5.34 Average 22.97 14.51 12.71 7.69 7.84 8.19 4.87 4,30 Table 2. Measured modal damping ratios. Mode Foundation Spring from Tests Foundation Spring from F.E.Model % Modal Mass Frcq. [Hz] Direction % Modal Mass Frcq. IHzj Direction 1 72.1 3.35 Y 70.0 2.46 Y 2 73.5 3.36 X 70.1 2.46 X 3 1.43 4.87 Y 1.7 4.80 Y 4 2.72 5.89 Y 11.3 5.16 X 5 4.43 6.66 X 8.1 5.20 Y 6 2.34 7.10 Y,X - - - 7 17.4 7.15 X - - • 8 19.3 7.30 Y - - 9 1.47 . 8.37 X - - - Table 3. Natural Frequencies by 3-D F.E.Model. NEXT PAGbiS; left BLANK 327 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES XA9952663 SHAKING TABLE TESTING OF MECHANICAL COMPONENTS D. Jurukovski, Lj. Taskov, D. Mamucevski, D. Petrovski Institute of Earthquake Engineering and Engineering Seismology, Skopje, Republic of Macedonia ABSTRACT: Presented is the experience of the Institute of Earthquake Engineering and Engineering Seismology, Skopje, Republic of Macedonia in seismic qualification of mechanical components by shaking table testing. Technical data and characteristics for the three shaking tables available at the Institute are given. Also, for characteristic mechanical components tested at the Institute laboratories, basic data such as producer, testing investor, description of the component, testing regulation, testing equipment and final user of the results. 1 INTRODUCTION The mechanical and the electrical components in a nuclear power plant should be capable of withstanding a pre-established seismic environment. This process is known as seismic qualification Regulatory agencies usually specify the general procedures to follow in seismic qualification According to the existing practice, a regulatory agency might stipulate seismic-excitation capability requirements for the equipment used in the plant, or the regulatory agency might specify the qualification requirements for various categories of equipment. The customer is directly responsible to the regulatory agency for adherence to the stipulations. Therefore, the customer or manufacturer hires the services of a test laboratory which is the contractor for seismic qualification of the equipment in the plant. A basic step in any qualification program is the preparation of a qualification procedure. According to IAEA (International Atomic Energy Agency) specifications (guides), the following types of testing can be applied: 1 Type-approval test (fragility test), 2 Acceptance test (proof test); 3 Low impedance test (dynamic characteristics test), 4 Code verification test. The seismic qualification test is required when failure modes cannot be identified or defined by analysis or earthquake experience. Direct qualification by testing employ type-approval and acceptance tests. Low impedance (dynamic characteristics) tests are normally used to identify similarity or verify or help to develop analytical models. Method of testing depend on required input, 329 weight, size, configuration and operating characteristics of the item, plus characteristics of the available test facility. 2. TESTFNG FACILITIES OF THE INSTITUTE The experimental investigations of the seismic and vibratory withstand of different types of me- chanical equipment were performed in the last fifteen years in the Institute of Earthquake Engineering and Engineering Seismology, University "St. Cyril and Methodius", Skopje. The investigations were performed for known end-user and the selection of the standards, criteria and requirements was made by the Investor, Producer and End-user. Most of the investigation programs were based on the former Soviet Union (GOST and OTT 82/87) standards, but a lot of the tested prototypes were tested based on the state of the an methodologies in the experimental mechanics A lot of designer's dilemmas or mathematical adjustments were solved after experimental investigations. This type of testing required usage of a sophisticated testing equipment, and application of advanced methods. At the Institute of Earthquake Engineering and Engineering Seismology, three shaking tables are installed: biaxial shaking table; uniaxial shaking table; electromechanical shaking table. The short review of the IZIIS testing facilities is presented herewith. 2.1 BIAXIAL SHAKING TABLE (Figure 1) • Size: 5 m x 5 m • Mass of the table: 40000 kg • Mass of the tested specimen. 40000 kg • Height of the tested specimen: 9 meters • Type of existing equipment: Servo controlled electro-hydraulic equipment • Type of vibrations: Random, sinusoidal, artificial wave forms • Dynamic load capacity: 800 kN • Frequency band: 0.1 -70 Hz • Directions (axis): 2, horizontal and vertical • Dynamic performances: Horizontal direction Stroke: ± ! 25 mm Velocity: ±75 cm/s Acceleration: ± 2 g maximum (depending on the mass of the specimen) 330 Vertical direction Stroke: ±60 mm Velocity: ± 50 cm/s Acceleration: ± 1 g maximum (depending on the mass of the specimen) Programming devices: Standard Function Generator Instrumentation Tape Recorder Player Random Noise Generator Digital Computer with D/AC Subsystem Data Acquisition Equipment: Electromechanical transducers: Accelerometers Displacements transducers Strain gages Other transmitters: Voltage output stage ± 5 V Recording equipment: Instrumentation Tape Recorder Digital Computer with A/DC Subsystem Readout equipment: Oscilloscope, Oscillograph, Digital Voltmeter Signal processing equipment: Digital Spectrum Analyzer Digital Computer System 2.2. UNIAXIAL SHAKING TABLE (Figure 2) Size: 1.5 m x 1.2 m Mass of the table: 1050 kg Mass of the tested specimen: 3000 kg Height of the tested specimen. 9 meters (recommended maximum 3 m) Type of existing equipment: Servo controlled electro-hydraulic equipment Type of vibrations: Random, sinusoidal, artificial wave forms Dynamic load capacity: 100 kN Frequency band: 0.1 - 140 Hz Directions (axis): 1, horizontal Dynamic performances: Horizontal direction Stroke: ± 100 mm Velocity: ± 50 cm/s Acceleration: ± 8 g maximum (depending on the mass of the specimen) Programming devices: Standard Function Generator Instrumentation Tape Recorder Player Random Noise Generator Digital Computer with D/AC Subsystem Data Acquisition Equipment Electromechanical transducers: Accelerometers Displacements transducers 331 Strain gages Other transmitters: Voltage output stage ± 5 V Recording equipment: Instrumentation Tape Recorder Digital Computer with A/DC Subsystem Readout equipment: Oscilloscope, Oscillograph, Digital Voltmeter Signal processing equipment: Digital Spectrum Analyzer Digital Computer System 2.3. SMALL ELECTROMECHANICAL SHAKING TABLE Size: 50 cm x 50 cm Mass of the table: 1000 kg (total mass without foundation) Mass of the tested specimen: 50 kg Height of the tested specimen. 3 meters (recommended height 1 m) Type of existing equipment: Three Phase Asynchronous Motor Type of vibrations: Sinusoidal or Sinusoidal Sweeping Dynamic load capacity. lOOkN Directions (axis). 2, horizontal and vertical Dynamic performances: Horizontal direction Stroke. ± 15 mm (max. for low frequency band) Acceleration: ± 8 g maximum (depending on the mass of the specimen) Vertical direction Stroke: ± 15 mm (max. for low frequency band) Acceleration: ± 8 g maximum (depending on the mass of the specimen) Data Acquisition Equipment: Electromechanical transducers. Accel erometers Displacements transducers Strain gages Other transmitters: Voltage output stage ± 5 V Recording equipment: Instrumentation Tape Recorder Digital Computer with A/DC Subsystem Readout equipment: Oscilloscope, Oscillograph, Digital Voltmeter Signal processing equipment: Digital Spectrum Analyzer Digital Computer System 332 The Institute staff have developed the software package for processing control, data acquisition, signal processing and functioning performance monitoring. The different flexible solutions provide a lot of facilities to satisfy all the testing criteria and requirements prescribed in the IAEA regulatory guides, IEEE Standards and recommendations, IEC testing procedures, and a lot of domestic documents related to seismic and vibratory testing of the equipment. 3. SELECTED TESTED COMPONENTS Different types of mechanical components have been tested on the shaking tables installed at the Institute of Earthquake Engineering and Engineering Seismology, Skopje. In this review, a summary on the testing of four types of mechanical components are presented. The selected types of equipment are: • valves; • electromechanical driving gears, • large scale mechanical shutters; • base isolating components. 3.1. LARGE SCALE VALVES Producer. "ENERGOrNVEST" - Sarajevo, Bosnia and Herzegovina Testing Investor: "ENERGOINVEST" - Sarajevo, Bosnia and Herzegovina Description: The tested large scale valves are primarily intended for usage in nuclear power stations produced by the Soviet Union companies. These "open - close" devices are designed for piping installations conducting low pressure fluids (usually water and air). Up to 6 different models were tested. However, some models were tested in two or three versions. The tested basic models were as follows: v JZZ~.T Sluice valve DU-600 Sluice valve DU-400 Valve DU-150 Valve DU-100 Regulatory valve DN100 PN200 Regulatory valve PN25 ND300 Scheme of DU-100 Sluice Valve 333 All the tested models were fully assembled with electromechanical driving gears and associated power transferring mechanisms. The tests were performed for opened and closed valve. The mounting of the valves on the shaking table was made ideally stiff, using two short segments of convenient pipes. Selected regulations: The investigation procedure was conceptualized by the former Soviet Union regulations referred to as OTT-82 and OTT-87 (General Technical Requirements published 1982 and 1987). Testing equipment: All the tests are carried out on the uniaxial shaking table. A digital computer system with D/AC and A/DC subsystem is used for seismic input data pre-processing, process control, data acquisition and preliminary data processing. Up to 32 high speed data acquisition channels are used to collect data on accelerations, displacement and strain on the carrier structure, the supports and the active parts and mechanisms forming the tested assembled specimen. The acquired data processed in the time and frequency domain (Fourier spectra) Test results: All the tests were performed in two orthogonal directions: longitudinal and transversal. In the first phase, the dominant or the first natural frequencies for both directions, at all possible different states of the tested assemblage, were defined. In the second phase, sinusoidal vibrations with duration of 20sec and an acceleration amplitude of 3g were applied in both directions, at all possible states of the tested assemblage. In each case, the tests were performed: • under resonant conditions, if subjected natural frequency was in the range from 20Hz to 50Hz • under frequency of 50Hz , if the subjected natural frequency was higher than 50Hz. If the natural frequency of the tested specimen was lower than 20Hz, the tested specimen does not satisfy the basic criteria and such a specimen was tested again after improvements made in the factory. If the tested specimen lost its function during the testing, such a model had to be modified in the factory and tested again. Accelerations, displacements and strain-stress distribution were measured. The final data processing was performed to help the designers improve the designing procedure. Final user of the results: All the tested products from this series were primarily intended for usage in nuclear power stations and produced by the Soviet Union producers and Eastern Europe co-producers. 3.2. ELECTROMECHANICAL DRIVING ASSEMBLAGES Producer: "ENERGO1NVEST" - Sarajevo, Bosnia and Herzegovina Testing Investor: "ENERGOINVEST" - Sarajevo, Bosnia and Herzegovina Description: The electro mechanical driving assemblages with different power capacity and different models were tested applying similar testing procedures. The following basic models were tested: EMSTN 25-16-C1-P EMSTN 40-16-C1-P EMPN 100-10-C-00/KK-0-6-1-9-00 EMPN 250-10-C-00/KK-0-6-1-9-00 334 EMPN 320-16-C-P EMPN 400-16-C-P EMPN 800-16-C-P EMPN 1000-16-C-P EMPN 2000-16-C-P Simplified Scheme of Electromechanical Driving Gear Some models were tested in two or three versions when the first version did not satisfy the testing criteria. The nominal power of the tested models varied from 25Nm to 2000Nm. All the tested models are primarily designed for usage in nuclear power installations as driving part in switching or regulations valves. Fully assembled driving gears were tested. The tests were performed by simulation of real loads, constant or variable, ranging between 0 to 125% from the nominal (declared) power capacity. Selected regulations: The investigation procedure was conceptualized by the former Soviet Union regulations referred to as OTT-82 and OTT-87 (General Technical Requirements published 1982 and 1987). Testing equipment: All the tests are carried out on the uniaxial shaking table. A digital computer system with D/AC and A/DC subsystem is used for seismic input data pre-processing, process control, data acquisition and preliminary data processing. Up to 32 high speed data acquisition channels are used to collect data on accelerations, displacement and strain on the carrier structure, the supports and the active parts and mechanisms forming the tested assembled specimen. The acquired data processed in the time and frequency domain (Fourier spectra). Test results: All the tests were performed in three orthogonal directions of the tested specimen In the first phase, the dominant or the first natural frequency was defined for each directions and for all the possible states of the driving gear: • out of operation • under operation without load • under operation with a load In the second phase, sinusoidal vibrations with duration of 20sec and acceleration amplitude of 8g applied in all the directions and under all possible states. In each case, the test was performed as follows: • under resonant conditions, if the subjected natural frequency was in the band from 20Hz to 50Hz • under frequency of 50Hz, if the subjected natural frequency was higher 335 If the natural frequency of the tested specimen was lower than 20Hz, the tested specimen does not satisfy the main criteria and such specimen was tested again after improvements done in factory. In case the specimen lost its functioning capabilities during the test, such a model had to be modified in the factory and tested again. Final user of the results: All the tested products from this series were primarily intended for usage in nuclear power stations and produced by tne Soviet Union producers and Eastern Europe co-producers. 3.3 "OPEN - CLOSE" DEVICE FOR LOW PRESSURIZED LARGE SCALE PIPING INSTALLATION MODEL "KZOK - 1200" Producer: "MIN" - Nis, FR Yugoslavia Testing Investor: "MIN" - Nis, FR Yugoslavia Description: The device "KZOK - 1200" (prototype model was tested) is intended for "open - close" functions in low pressurized large scale (diameter 1200 mm) piping installations in nuclear power stations. The main parts of the tested device are shown on the scheme. 1. base 2. "open - close" device body 3 closer 4. gearing device 5. drive Scheme of KZOK - 1200 Prototype Model The "open - close" functions could be performed manually or automatically supported by an electrically driven gear. Selected regulations: The investigation procedure was conceptualized by the former Soviet Union regulations referred to as OTT-82 and OTT-87 (General Technical Requirements published 1982 and 1987). Testing equipment: All the tests are carried out on the biaxial shaking table. A digital computer system with D/AC and A/DC subsystem is used for seismic input data pre-processing, process control, data acquisition and preliminary data processing. Up to 32 high speed data acquisition channels are used to collect data on accelerations, displacement and strain on the carrier structure, the supports and the active parts and mechanisms forming the tested assembled specimen. The acquired data processed in the time and frequency domain (Fourier spectra). 336 Test results: The prototype model was tested in two positions: • opened, free closer • closed, fixed closer The testing procedure was realized in two phases. In the first phase, the natural frequencies were realized for all the six different states (two positions, three directions). In the second phase, the testing was performed by sinusoidal vibration with a duration of 20sec and an amplitude of 3g in the horizontal direction and 2g in the vertical direction. The amplitudes were in the geometrical center of the models' bodies (or more precisely, in the center of the pipe segment). According to the technical requirements, the tests were performed under resonant conditions and consequently, a different frequency was applied in each case. Also, in each case, the most critical frequency was applied since the tested model consisted of three dynamic subsystems. The dynamic behavior of the model was monitored by recording of acceleration and strain time histories. The acceleration time histories are interesting from the driving-gear functioning view point and from the aspect of identification of the dynamic subsystems in the assembled tested model. The strain-stress time histories are important for monitoring the stress distribution at characteristic points and checking of some design considerations and approximations. During the tests, strengthening was performed, especially at the driving gear interconnections Final user of the results: The testing investor requested an answer to the following two problems: • verification of the applied design methodology and checking of the considered solutions and approximations during designing as well as manufacturing of the prototype • checking of the realized model capabilities making a comparison with OTT-82 and OTT-87 criteria 3 4 NEEDLE SHAPED VALVE TYPE Pp 160 Tp 100 Producer: "PRVA ISKRA" - Baric, Belgrade, FR Yugoslavia Testing Investor: "PRVA ISKRA" - Baric, Belgrade, FR Yugoslavia Description: The needle shaped valves type Pp 160 Tp 100 are small size devices with manual control only. In nuclear power installations, these devices are used in small-scale piping installations of secondary systems. The automatic control of functioning is not considered. On the scheme, the main dimensions of the tested specimens are presented. Tested were three identical specimens The tested specimens were mounted on ideally stiff supports (plates mounted on the shaking table). 337 10 4)70 I 26 Scheme of Pp 160 Tp 100 Needle Shaped Valve Selected regulations: The investigation procedure was conceptualized by the former Soviet Union regulations referred to as OTT-82 and OTT-87 (General Technical Requirements published 1982 and 1987). Testing equipment: The tests were performed on the small electromechanical shaking table. This shaking table can generate sinus or sinus sweeped vibration in a frequency band from lHz to 7Hz and from 7Hz to 77Hz. The vibrations can be generated in horizontal and vertical directions Two vibrations can be generated in horizontal and vertical direction. Two accelerometers are used for measurement of input (excitation) vibrations and output (response) vibrations A two channel spectrum analyzer was used for measurement and analysis of vibratory accelerations. Test results: Each specimen was tested in two positions. • opened • closed At each position, the tests were performed in three orthogonal directions. In all the six cases, the same testing procedure was applied. Anticipated in the first phase was definition of the natural frequency. Using the shaking table in the frequency domain from lHz to 70 Hz, no natural frequency was detected. The first natural frequency was detected in all six cases, using pulse excitation and 338 11 spectral analysis of the free damped oscillations of the tested specimen. All the detected natural frequencies were greater than lOOHz. In the second phase, all the tests were performed at a frequency of 20Hz and 30Hz with an acceleration amplitude of 3g in the horizontal and 2g in the vertical direction according to the regulatory criteria and because all the natural frequencies were higher than 3 OHz. Final user of the results: The production of this type of valves was ordered by the Soviet Union manufacturers specialized in equipment for nuclear power stations. 3.5. STOCK BRIDGE DAMPERS FOR HIGH VOLTAGE CONDUCTORS AND DISTRIBUTION LINES Producer: "EMO" - Ohrid, Republic of Macedonia Testing Investor: "EMO" - Ohrid, Republic of Macedonia Description: The stock bridge damper for high voltage conductors and distribution lines was tested in several versions and several sizes. The role of this device is to attenuate the ambient vibrations of high voltage conductors. Areal and spatial dampers were tested. Each version, representing a separate physical model was tested independently, without any conductor. The tests were performed on the small electromechanical shaking table. Some selected models were tested mounted on a conductor. The programmable vibrations of the conductor were generated by the biaxial shaking table. i—i ', , /•• - Simplified Scheme of STOCK BRIDGE Damper Selected regulations: The applied methodology was developed following some industrial practice and standards for testing of such type of an equipment. 339 12 Testing equipment: All the tests with a conductor are carried out on the biaxial shaking table. A digital computer system with D/AC and A/DC subsystem is used for seismic input data pre- processing, process control, data acquisition and preliminary data processing. Up to 32 high speed data acquisition channels are used to collect data on accelerations, displacement and strain on the carrier structure, the supports and the active parts and mechanisms forming the tested assembled specimen. The acquired data processed in the time and frequency domain (Fourier spectra). All the tests without a conductor are carried out on the small electromechanical shaking table. The digital shaking table can generate sinus and sinus sweeped vibration in a frequency band from lHz to 7Hz and from 7Hz to 77Hz. The vibrations can be generated in horizontal and vertical direction. Two accelerometers are used for measurement of input (excitation) vibrations and output (response) vibrations. A two channel spectrum analyzer was used for measurement and analysis of vibratory accelerations. Test results: For each tested model, the natural frequency, damping ratio and vibration shape mode were determined using sinusoidal vibrations. The mechanical impedance was calculated using some empirical formulae. The dynamic behavior of a segment of a conductor without a damper and with a damper was investigated using different type of programmable vibrations. Comparative analyses were performed to identify the influence of the mounted damper. Final user of the results: The testing investor defined two tasks: • creation of a base for mathematical modeling of different types of dampers • preparation of a data base for marketing purposes, especially for export jobs in Iran and other Asian and Arabic countries 3.6. BASE ISOLATION ELEMENT FOR DIESEL GENERATORS AND ENGINES TYPE 3.08 140 000 Producer: "JUGOTURBINA" - Karlovac, Croatia Testing Investor: "JUGOTURBINA" - Karlovac, Croatia Description: The base isolation element is primarily intended for base isolation of diesel generators and engines in the ships. However, it could also be used for other base isolation purposes The element represents a rubber based base isolation product. The minimal number of base isolation elements is four. For testing purposes, a minimal configuration consisting of four identical elements was used Elements of different hardness were tested (45 SH, 55 SH, 65 SH). An ideally stiff frame loaded by a continuous load was used as a tested specimen isolated by four base isolating elements Some tests were performed for each element taken separately and some were executed for the set of four elements. Selected regulations: The producer defined the testing requirements based on Lloyd's Register Regulations Testing equipment: All the tests are carried out on the biaxial shaking table. A digital computer system with D/AC and A/DC subsystem is used for seismic input data pre-processing, process control, data acquisition and preliminary data processing Up to 32 high speed data acquisition channels are used to collect data on accelerations, displacement and strain on the carrier structure, the supports and the active parts and mechanisms forming the tested assembled specimen. The acquired data processed in the time and frequency domain (Fourier spectra) To determine the 340 13 dynamic characteristics, a dynamic electrohydraulic actuator with programmable movement was used. To define the static characteristics, a quasi-static actuator with programmable stroke was used Test results: A static characteristic represents a ratio between load and deflection. In the horizontal direction, a set of four elements connected by a stiff frame was used. The characteristics were defined for three different axial pre-loads: 30kN, 40kN and 50kN per element. The maximal deflection was ±40mm. In the vertical direction, a static characteristic is determined in each element individually, for maximal axial load of 70kN. 3.7. HYDRAULIC. SNUBBER TYPE HA - 500kN Producer: "GOSA" - Belgrade, FR Yugoslavia Testing Investor: "GOSA" - Belgrade, FR Yugoslavia Description: The hydraulic snubbers from the HA series are designed as energy absorbing elements primarily designed for usage in railway transports and some industrial installations. A prototype was tested. The tested specimen is a bi-stable device. In the first state (low velocity movement), the piston of the snubber is free and only friction forces provide some resistance against the external axial load. If the movement becomes with a velocity greater than the critical one, the piston becomes locked and further movement is not possible. Selected regulations: Not specified. Some design considerations, theoretical and empirical requirements and criteria were the basis for the investigation program. Testing equipment: Electrohydraulic dynamic and quasi-static actuators with programmable movement were used for testing of the hydraulic snubbers. Test results: The investigations were performed in two main phases. In the first phase, the methodology for selecting of a set of a valve and spring was experimentally developed. The task was to establish criteria regarding the compatibility between the valve and the spring to obtain a double ended (symmetrical) hydraulic cylinder. In the second phase, the parameters of the snubber were experimentally defined: the friction load, the critical velocity and the nominal load. Final user of the results: The testing investor (GOSA Institute for Research and Development) was the user of all the experimental results. 4 CONCLUSIONS In the Dynamic Testing Laboratory at the Institute of Earthquake Engineering and Engineering Seismology, University "St. Cyril and Methodius", Skopje, Macedonia in the past fifteen years many mechanical and electrical components for NPP as well as for other special facilities have been tested under seismic and various dynamic loads. From the gathered experience and existing world practice, seismic qualification of mechanical and electrical components primarily depends on the three following testing conditions. Definition of the seismic input at the location where the component is installed. The seismic input should be determined to represent actual seismic action from the point of view of frequency range of interest, amplitudes and time duration. The frequency range should bound the most 341 14 influenced frequencies of the components. In our experience, the seismic or dynamic input was determined by the manufacturers or end-users of components. It is obvious that accurate experimental simulation of the determined seismic or dynamic input has to be provided. It means that the simulated input on the shaking table has to be done without frequency or amplitude modifications. The outcome of seismic qualification obtained by testing, is mostly influenced by the criteria of acceptance. The criteria of acceptance should be coordinated decisions from the point of view of definition of functional and structural acceptance. In the existing practice functional acceptance is usually estimated after the testing which means off- line. This should be a joint effort of experts from producers, users, regulatory bodies and testing laboratories. 5. REFERENCES 1. IEEE 279-1971 Criteria for Protection of Systems for Nuclear Power Generating Stations (ANSI/IEEE) (REAFF 1978) (Revision of IEEE Std 279-1968) 2. IEEE 323-1983 Qualifying Class IE Equipment for Nuclear Power Generating Stations (Revision of IEEE Std 323-1974) 3. IEEE 336-1985 Installation, Inspection and Testing Requirements for Power Instrumentation and Control Equipment at Nuclear Facilities (Revision of IEEE Std 336-1980) 4. IEEE 344-1975 Recommended Practices for Seismic Qualification of Class IE Equipment for Nuclear Power Generating Stations (ANSI/IEEE) (Revision of IEEE Std 344-1971) 5. IEEE 3 82-1980 Standard for Qualification of Safety-Related Valve Actuators (Revision of IEEE Std 382-1972) 6 IEEE 420-198 2 Standard for the Design and Qualification of Class IE Control Boards, Panels and Racks Used in Nuclear Power Generating Stations 7. IEEE 467-1980 Standard Quality Assurance Program Requirements for the Design and Manufacture of Class IE Instrumentations and Electric Equipment for Nuclear Power Generating Stations 8. IEEE 497-1981 Standard Criteria for Accident Monitoring Instrumentation for Nuclear Power Generating Stations (ANSI/IEEE) 9. IEEE 600 Draft - Trial Use Standard Requirements for Organizations that Conduct Qualification Testing of Safety Systems Equipment for Use in Nuclear Power Generating Stations (ANSI/IEEE) 10. IEEE 603-1980 Standard Criteria for Safety Systems for Nuclear Power Generating Stations (Revision of IEEE Std 382-1972) 11. IEEE 627-1980 Standard for Design Qualification of Safety Systems Equipment Used in Nuclear Power Generating Stations (ANSI/IEEE) 12. IEEE Std 693-1984 IEEE Recommended Practices for Seismic Design of Substations, The Institute of Electrical and Electronic Engineers, Inc. 13. Nuclear Regulatory Commission, Regulatory Guide 1.100, Revision 1, 1977. Seismic Qualification of Electronic Equipment for Nuclear Power Plants 14. IAEA Safety Guides, Safety Series No. 50-SG-S2 Seismic Analysis and Testing of Nuclear Power Plants, A Safety Guide, IAEA, Vienna 1979 15. IEC Publication 68-2-6 Basic Environmental Testing Procedures; Test and Guidance; Vibration (Sinusoidal), International Electronical Commission, 1982 16. IEC TC 50 SC 50A Environmental Testing; Shock and Vibration Tests, International Electronical Commission, 1982 17. IEEE Std 501-1978, IEEE Standard Seismic Testing of Relays, The Institute of Electric and Electronic Engineers 18 INTERATOMENERGO - Moscow, General Technical Requirements OTT-H2 and OTT-H7 342 15 19. IAEA Safety Series No. 50-SG-D 15 Seismic Design and Qualification for Nuclear Power Plants, A Safety Guide, IAEA, Vienna 1992 20. Atomic Energy Commission, Regulatory Guide 1.60, Design Response Spectra for Seismic Design of Nuclear Power Plants, 1973 21. Atomic Energy Commission, Regulatory Guide 1.29, Revision 1, Seismic Desigti Classification 22. Lloyd's Register festing Manuals Figures 343 PROCESS CONTROL COMPUTER SYSTEM Digital VAX-LAB 4000 Model 200 Z 12bit. 12bit P Magnetic Magnetic Terminal Magnetic CPU Memory 3 D/AC A/DC Printer Plotter Terminal rocessoi Disk Disk Server Streamer 8ch 64ch Magnetic Tape Recorder BIAXIAL SHAKING TABLE Random Noise Generator Analogue Control Read-Out System Selector Function Generator Oscilloscope 4ch Signal Conditioning System Digital Spectrum Analyzer Test Specimen with Hydraulic Power Supply Electromechanical and other transducers Fig I Functional Block Diagram of Biaxial Shaking Table XNV19 H»l (S)30Vd 1X3N ^> n • v; o — 11 3» <~ -o — O C r» 3 •- B> n z 3 o -n (Tl fij 3 ne t c 3-D (D» re in CL o o - TOO re a 00 IV D/AC A/DC -> o o 3 Subsystem Subsystem -1 7> I a 3 O 5' CO o o 03 CO O G 3_ 5' C/3 CD oa to a; a Displacement Transducer * L\ 1 XA9952664 PROCEEDINGS OF SMIRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES EXPERIMENTAL AND COMPUTER ANALYSES OF CONTROL ROD DRIVE SYSTEMS SEISMIC CAPACITY. Victor V. Kostarev, Victor N. Abramov, Alexis M. Berkovski, Peter. S. Vasiliev, Alexander J. Schukin, CKTI-Vibroseism (CVS), St. Petersburg, Russia. ABSTRACT: The experimental and computer analyses of the 1/4 scale Control Rod Drive System (CRDS) model of WER-440 reactor has been carried out. The experimental study has been under- taken on CVS 20 ton's capacity shaking table with modeling operability of CRDS during earth- quake and operational vibration. A special PC computer program has been developed for evaluation of CRDS seismic and vibration margins. The program enables estimation of different nonlinear ef- fects in bearings and gaps of CRDS including shocks and friction that highly influence on dynamic capacity of CRDS. The results of these investigations are presented in this paper. 1. INTRODUCTION ' The WER-440 Control Rod Drive System consists of the control rods and related mechanical components which provide the means for mechanical movement. The main special feature of WER CRDS is the length of over 10 meter's height construction with a number of bearings and gaps with relatively small clearances. Such construction suggests some sensitivity of the CRDS to dynamic and seismic impacts during operational of its safety functions. According to NUREG - 0800 Standard Review Plan, Section 3.9.4 General Design Criteria 26 and 27 require that the CRDS provide one of the independent reactivity control systems. The rods and the drive mechanism shall be capable of reliably controlling reactivity changes either under conditions of anticipated normal plant operational occurrences, or under postulated accident condi- tions. The same idea is included in demands of Russian Code PNAE G-5-006-87, 1987. A positive means for inserting the rods shall always be maintained to ensure appropriate margin for malfunction, such as stuck rods. Since the CRDS is a system important to safety and portions of the CRDS are a part of the reactor coolant pressure boundary (RCPV), General Design Criteria 1, 2, 14, and 29 and 10 CFR Part 50, p. 50.55a, require that the system shall be designed, fabricated and tested to qualify standards commensurate with the safety functions to be performed. This is to assure an extremely high probability of accomplishing the safety functions either in the event of anticipated operational occurrences to withstand the effects of postulated accidents and natural phenomena such as earthquakes. The WER CRD Systems using the above mentioned criteria mean confirmation of extremely high probability of CRDS functioning in modes of AR (normal operational regulation of reactor reactivity) and AZ (safety function in postulated accidents including earthquakes). The main inter- est belongs, of course, to AZ mode when WER CRDS is working in conditions of rod free gravity fall inserting and malfunction and stuck rod can occur due to dynamic impact. 347 A number of WER natural scale CRDS shaking table tests shows the real possibility of stuck rod occurrences during seismic wave excitation [ 1 ]. At the same time it was rather hard to get the general objectives and regularities due to highly complicated structure of WER CRDS and multiple nonlinear effects in CRDS bearings and gaps during dynamic excitation. That is why the new investigation of WER CRDS seismic capacity has been undertaken in the stream of IAEA efforts for Seismic Upgrading of WER type NPPs. The current investigation consists of two main parts: a) Study of WER- 440 CRDS 1/4 scale model seismic and vibration capacity on the CVS 20 ton's shaking table; b) Computer analyses of CRDS dynamic behavior under seismic and vibration loads by means of special developed program for PC and comparative study of experimental and analysis results. 2. EXPERIMENTAL STUDY. 2.1. Description of the Test Rig and Measurement System. The experimental investigations of WER - 440 CRDS 1/4 scale model seismic and vibration capacities have been carried out on the CKTI-Vibroseism 1-D horizontal Shaking Table, specifi- cally designed for all kinds of dynamic testing, including seismic loading of the full scale CRDS for WER-440 and WER-1000 MWt reactors, Figure 1. The CVS shaking table has the following technical characteristics: - maximal dynamic pushing force 120 kN (12tons); - mass of foundation more than 250 tons; - mass of vibration platform 2.0 tons; - dimensions of vibration platform for equipment installation 3.0 x 1.5m; - height of undertable space for samples installa- tion more than 5m; - maximal acceleration of unloaded shaking table (without samples) 15g; - maximal mass of testing equipment with lg ac- celeration level 10 tons; - maximal overturning moment 600 kNm; - frequency range 0 (static), 0.05 - 100 Hz; - maximal amplitude of displacements ±90 mm; -resonant frequency of the shaking table more than 40 Hz; - errors in frequency setting less than 1%; - coefficient of nonlinear distortion in displace- ment setting (0,05-20 Hz) less than 10%. The shaking table Control System consists of PC with Analogue-Digital-Analogue Converters (ADC/DAC) and permits to set the following modes of operation and testing: - sinusoidal sweep excitation; - static displacements of the table; - multiharmonic excitation; - random dynamic and seismic excitation. 348 The vibration measurements during the CRDS testing on shaking table have been carried out by Computer Multichannel Complex "MERA" and "Bruel & Kjer" instrumentation, Figure 2. 2.2. The CRDS Model. The main goal of current investigation was to determine the nonlinear dynamic behavior of CRDS and make an attempt to create reliable analytical model and to verify the computer program and results of analyses on this base. From this point of view the CRDS model has to reflect all im- portant peculiarities and dynamic properties of CRDS that can influence on seismic and vibration capacity of CRDS. To achieve this task is not necessary to make a full copy of CRDS but only to reproduce the main inertia, material, stiffness, gap and other properties of CRDS so as interaction of internal elements. Usually for model seismic testing of structures on shaking tables is used the Simple Geometric Principle. The specific side of the CRDS modeling is the necessity of taking the acceleration model coefficient equal to "1" for modeling the operability of CRDS during accident situations with AZ mode free insertion of the rod (rack in WER case). The tested structure presents the 0.25 scale geometric model of VVER - 440 Paks NPP CRDS ARK like prototype. The main characteristics and model scales are the following: - time 0.5; - material 1 -0; .-linear scale 0.25; - displacement 0.25; - frequency 2.0; - velocity 0.5; -mass 0.015625. 2.3. Methodology. Obviously, the main role in CRDS stuck rod and insertion time in AZ mode under dynamic im- pacts belong to horizontal excitation. The influence of vertical excitation is practically negligible in comparison with horizontal one. That is why the testing of CRDS model has been carried out only under horizontal random dynamic excitation, scaled to demand intensity and duration in accordance with model coefficients of acceleration and time. The Paks NPP Design Spectra (elevation 18 m) has been used for developing of model synthetic TH process for testing and analyses of CRDS. The duration of model excitation was equal to 10 seconds and ZPA level has been varied from 0 up to 0.72g. The Figure 3 shows the four different seismic wave displacements of the model seismic excita- tion on the shaking table. The main criteria of CRDS operability and at the same time the CRDS seismic capacity is the time of CRDS rack inserting (free fall) in AZ mode which is limited for natural scale CRDS from 8.5 up to 12.8 seconds, that corresponds the average velocity of the rack insertion about 300 - 200 mm/s. For CRDS model it means that the time of CRDS rack insertion has to be in limits reduced according to model scaling. 349 So the general goal of CRDS seismic testing is determination of the CRDS insertion time in AZ mode during design earthquake or other accident postulated events, for example high intensity op- erational vibration. The Control and Measurements System of the test rig allows to set the needed input kinematic characteristics of shaking table and to gather, process and analyze all output data such as accelera- tion, velocity and displacement of structure and parameters of CRDS rack insertion under dynamic excitation. 2.4. Results of CRDS testing. The initial stage of experiment was to determine the first natural frequency and damping charac- teristics of the CRDS model. This test has been performed with upper and lower positions of CRDS rack and shows that the first mode and damping parameters are the same for these two cases. The plots of CRDS free oscillations are shown on the Figures 12, 13. The first natural frequency of this system is equal to: /= 4.6 Hz and damping ratio is equal to: k = 0.013 (1.3%). The investigations of influence of seismic and vibration impacts on CRDS rack free insertion in AZ mode have been performed with ten levels of seismic wave accelerations from 0 up to 0.72 g. Five experiments with statistical processing of results have been fulfilled for each level of seismic acceleration . The analyses of results show rather good reproducing and repeatedness of shaking table parame- ters during different experiments, Figure 3. The recording of the upper part CRDS displacement makes clear that the CRDS rack is working like non-linear gap dynamic damper in CRDS bearings, that's why output displacement waves are quite different in the levels and phases in case of the same input parameters of shaking table, figure 4. The sensitivity of CRD system to "rack — housing" in- teractions can be also illustrated by the plots of free rack insertions under the same conditions of the experiment, including the shaking table excitation, Figure 6. The main results of seismic excitation influence on CRDS rack free insertion in WER AZ mode are shown on Figure 6. The plots are illustrated the dependencies of average free fall insertion of the CRDS rack from intensity of earthquake excitation. It is clear that the Mean +/- standard de- viation zone of results becomes more narrow with increasing of seismic impact intensity. It is also possible to conclude that the time of rack insertion mainly is limited by duration of excitation and can achieve several times greater levels in comparison with normal designed time in AZ mode un- der real earthquakes. The investigations of influence of operational vibration on CRDS free insertion in AZ mode have been also performed during the tests on the shaking table. In these tests the frequency of harmonic excitation was tuned according to resonant frequency of CRDS rack to achieve the vibration shock mode in CRDS bearing gaps. The frequency was equal to 37 Hz for model of CRDS that means 18.5 Hz for natural scale CRDS. The results are shown on Figure 7. The resonant of CRDS rack greatly influences on the CRDS free insertion time and depends primarily from the deepness of resonant process of the system "Rack - Bearings". The main experimental results of CRDS dynamic testing under seismic and vibration excitations show that the time of CRDS AZ rack accident insertion like the general criteria of safety functions 350 of CRDS can increase up to 2 - 6 times against design requirements, Figure 8. So, achieving of the CRDS demand seismic and vibration capacities is the real way to upgrade the safety properties of WER-type reactors. 3. ANALYTICAL STUDY OF THE SEISMIC RESISTANCE OF THE CRDS. 3.1 Computer Code SEISM-2000 for the dynamic analysis of the non-linear systems with paratiiet- rically varying characteristics. SEISM-2000 is the computer software program developed for analytical investigation of the non- linear systems whose parameters and characteristics are changed during the dynamic process. The background for developing of SEISM-2000 is the component-element method (CEM) [2] combined with the finite-element method (FEM). To solve the difference equations of motion the direct integration method in terms of central finite differences is used. The program provides the dynamic time-history analysis with finite-element approximation of beam-type elements and systems with concentrated parameters (lumped mass, stiffness, damping, etc.). The library of non-linear elements includes a number of components. Among them are: - elastic spring element; - viscous damper element; - constant friction element; - variable friction element (the reaction force in this element depends from reaction in other de- scribed early element and treated by program as friction force); . - limit stop; - moving limit stop (this element takes into account interaction between moving inserting rod and external construction); - hydraulic damper and snubber, - bearing damper, etc. To create the analytical model of system the program uses finite-element approximation of beam substructures and concentrated parameters, such as lumped masses, stiffness and damping. Then all parts of analytical model incorporate among themselves by component elements. To provide dynamic analysis the recording of time-history acceleration is used as input excita- tion. The input data for the SEISM-2000 are: the geometry and properties of beam substructures, val- ues of concentrated inertia and stiffness parameters, the characteristics of component elements, digital records of input excitation. The dynamic response of system is the result of computer analysis. Among the output results there are following dependencies: "time-displacement," "time-velocity", "time-acceleration" for dy- namic degrees of freedom and "time - deformation - reaction force" for component elements. 3.2 Analytical model of the CRDS. To create the analytical model of CRDS the geometry and properties of experimental 1/4 scale model of WER-440 CRDS have been used. From experimental data the damping ratio for whole construction was determined too. The time of the control rod's insertion is defined by the complex of kinematic parameters of CRDS including dry friction in the bearings of drive mechanism and hy- 351 draulic resistance of moving parts of construction. These peculiarities of the construction have been modeled by means of friction elements and viscous damper element respectively. The analytical model consists of three beam substructures. The first one is modeling the support- ing frame of the CRDS, the second - CRDS housing, in which the control rod is moving. The third substructure is the beam finite-element model of the inserting rod, Figure 9. All three substructures are connected with component elements. For example the elements "moving limit stop" imitate the interaction between inserting control rod and housing (collisions take place in the bearing and housing clearances). Then, as result of these shocks, appear the horizontal reaction, which induces the vertical friction force. The result of this force acting is additional slowing down of rod inserting (stuck rod). All these processes are modeled by component elements "variable friction". The complete analytical model of this system is shown on Figure 10. 3.3 Analytical results and their comparison with experimental data. The seismic excitation in terms of the Response Spectra for Unit 3 of NPP Paks at the elevation of upper block of Reactor Pressure Vessel (RPV) has been chosen for analytical and experimental investigation of the CRDS. The Spectra was modified according to the methodology described in chapter 2.3. On the basis of this Spectra the synthetic accelerogam was generated to carry out the variation analyses, Figure 11. On the first stage of investigation the experimental and computer analysis data have been com- .pared to verify the analytical model and computer code SEISM-2000 itself. The following modes were chosen for comparison: - free insertion of the control rod without external impacts, Figure 12; - free oscillation of the supporting frame, Figure 13; - dynamic behavior of the system under external excitation with magnitude 0.6g, Figures 14,15. The last mode in the further consideration will be named later on like "basic variant". The fol- lowing parameters of system correspond to this mode and analytical model, Figure 10: - the diameter clearance in the upper limit stop: 1 mm; - the diameter bearing's clearances: 0.05 mm; - the diameter clearance in the bottom limit stop (RPV fuel element): 0.5 mm; - the friction coefficient of bearings: 0.15; - the friction coefficient of "steel-steel" pair: 0.5; - start time of the control rod insertion: 1/2 from total duration of seismic excitation. The total time of seismic excitation is 10 sec and there was 4 sec more to record the free oscilla- tions of the system after the end of impact. The results of comparative experimental and computer analyses show appropriate agreements in main dynamic parameters, such as natural frequencies, time of rod insertion, so on, Figures 12 - 15. These results permit to conclude that the analytical model and computer code are verified against the experimental research. The next step of this analytical investigation was to estimate influence of the different parameters of system on dynamic behavior of CRDS. The reason for such investigations is to choose the critical parameters for analytical study of CRDS operability under dynamic impacts. It should be noted that 352 7 for real construction of CRDS it is too complicate to define the exact values of such parameters as clearances, friction, etc.. All analyses have been fulfilled under "basic variant" of CRDS model de- scribed above. Figure 16 shows that the clearance in CRDS bearings is not influence practically on dynamic be- havior of system. On the other hand, varying of clearance in bottom limit stop, that imitate the WER Fuel element, gives the great effect in CRDS rod time insertion, Figure 17. Changing of friction coefficient in CRDS gaps demonstrates practically the linear dependencies between CRDS rod insertion time and friction conditions, Figure 18, 19. The next important result of analytical study is that the CRDS rod insertion time is heavily de- pends on start point of rod insertion during earthquake excitation, Figure 20. The dependencies of CRDS rod insertion path and time from seismic intensity (ZPA) are shown on Figure 21. It is clear from these plots that seismic impact can heavily increase (2-4 times) the main safety characteristic of CRDS. That means the necessity of additional evaluation of these phenomena in seismic design and up- grading of WER. 4. CONCLUSIONS. 1. The complex of experimental and analytical study of the WER-440 Control Rod Drive System .seismic and vibration capacities have been carried out. 2. To analyze dynamic behavior of CRDS the computer code SEISM-2000 has been developed and successfully verified against experimental results. 3. The investigations of CRDS seismic and operational vibration capacities show that dynamic exci- tation can significantly increase the time of CRDS accident insertion and may influence safety of VVER plants. 5. REFERENCES. 1. V. Kostarev. Evaluation of Potential Hazard for Operation of WER Reactor Control Rods under Seismic Excitation, IAEA Report, No. 7448/EN, 1994. 2. S. Levy, J.P. Wilkinson. The Component Element Method in Dynamics. McGraw Hill IBC, 1976. 353 i 1. The general view of CVS Shaking Table with natural scale CRDS. Figure 354 1 - potentiometer 1 2, 3, 4 - acceleration transducer B&K amplifier/integrator PC B&K 2635 CRDS Hydraulic Actuator Shaking Table DAC 77777777 77777777 Table Control System Figure 2. Principal chart of measurement UJ 10 Time, s. Figure 3. Coincidenced Seismic Wave displacements of the Shaking Table (four records). r 1 1 ~ -' ' 6- ! ! ! I 1 ! ! a •>• 2-L J ill 11 . til •i*- !•- li •L- !•• i • Tl 1 IT i 1 5 ' 11 i '111 , 'i _ L_ _ J i -a- \ | T, i i. i l \ I ' l' i[•ITU I-TTV.-I Figure 4. The response displacements of the upper part of CRDS (for cases under the same input excitation of the Shaking Table). 356 1! without excitation under excitation Figure 5. Free insertion of the CRDS Rack in AZ mode (four cases under the same input excitation of the Shaking Table). Mean value 00000 Mean - Std.Deviation Mean + Std.Deviation End of the Table Excitation n—i—i—i—i—i—i—i—;—i i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—;—i—i—i—i—\—i—i Acceleration. G Figure 6. Influence of the CRDS free insertion time in AZ mode from acceleration level of seismic excitation. 357 12 Amp I .= 0. 03 cm Amp I .= 0.02 cm Without Ex c I t & t I o n 10 12 Time, s Figure 7. Influence of the CRDS free insertion lime in AZ mode from amplitude of operationc! vibration. 60-E 7' y I. O 50-E I I a; 40-E I 03 without excitation Q 20- seismic shock — — vibration 37 Hz ( S il n 12 14 18 time, s Figure 8. Influence of the CRDS free insertion time in AZ mode from different types of dynamic excitation. 358 Z pipes 32x2 5 \ pipes 4&j<2 5 4G0 r\ Solid Cirrlc Be*m (0 IZmm] C\ 460 pipe 40x2 460 . -OX 460 '"" pipe 76x2.5 Load Substructure 1 400 -14 Substructure 3 Substructure 2 uo Figure 9. Substructures of the Analysis Model. 14 gap: +/- 0.5 mm. structural friction coefficient 0.15 1 ; • 0 gap: +/- 0.025 mm. structural friction coefficient: 0.15 /////Y//// ////V///// /////// gap: +/-0.C25 mm. structural friction coefficient: 0.15 gap: +/- 0.25 mm, structural friclion coefficient: 0.5 10 Figure 10. Analysis Model of the CRDS. 360 15 6.0- 1.0- o a 3.0- 0) o 1.0- 0.0- I t I r 10 15 20 25 Frequency, Hz Time, sec Figure 11. Response Spectra and corresponding seismic wave of the input excitation for test and analyses. 361 80.0-, 16 60.0- e u a I 40.0- 0) o I—H0, 10 Analysis Experiment 20.0- 6.0- e.ee i.ee 3.ee 4.00 Time, sec Figure 12. Free Insertion of the Control Rod without dynamic excitation Experiment 0.0 e.s i.e I.S 2.0 2.S 3.6 3.6 Time, sec Analysis e.e e.e s.e 3.c Time, see Figure 13. Free oscillations of the 362 CRDS Housing (point 2). 70-i 17 60- Experiment Analysis 8 ? 10 11 12 13 14 15 16 17 18 Time, sec Figure 14. Free Insertion of the Control Rod under seismic excitation (basic variant) Time, sec Analysis 12 13 H 15 Time, sec Figure 15. Response vibration of the CRDS Housing under seismic excitation (basic variant, point 2). 363 18 11 .0-1 10.0- Bearings 0) g 9.0- O •H 8.0- 7.0 I i i i i | i i i i ] i i i i | i i i i | i ' i i | i i i ! |—n—i i | i I"-I—i—pi—i—i—i—i—i—i—i—;—i 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.B3 0.09 0.1l3 Clearance, mm Figure 16. Influence of CRDS insertion me from bearing clearances. 11.0-1 10.0- 0) Upper limit stop B Bottom limit stop 9.0- c o CO £ 8.0- 7.0- 0.0 0.1 0.2 0.3 a.A 0.6 0.6 0.7 e.8 0.9 1.0 Clearance, mm Figure 17. Influence of CRDS insertion time from clearances in the upper and bottom Stops. 364 19 8.0-1 45 s "steel - plastic" pair 6.0- oa 03 W J5S.0- •4.0- ' i i | i r e.ee 0.05 b.10 0.15 0.20 Friction coefficient Figure 18. Influence of CRDS time insertion from friction in bearings. 8.2-1 *-*•***-* "steel — steeP pair emn '• 0.0 Friction coefficient Figure 19. Influence of CRDS time insertion from friction in the bottom part of CRDS. 365 20 Tuna, B*C R«lottv« tlm« of Insertion start against «arthquak« duration. Figure 20. Influence of CRDS insertion from start point of Insertion during earthquake excitation. i - a eoo 1 I!i j n": 3 - e.i2o z Jl I 6 * dl36o ' if I 7 jm i / V' 6 - el72s 5- IP ij r Jl Tine, to Figure 21 Influence of the CRDS insertion time from ZPA of earthquake. 366 XA9952665 PROCEEDINGS OF SMIRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES SHAKING TABLE TESTING OF ELECTRICAL EQUIPMENT IN ARGENTINA CARMONA, Juan S.; ZABALA, Francisco; SANTALUCIA, Jorge; SISTERNA, Cristian; MAGRINI, Marcelo; OLDECOP, Luciano. Institute de Investigaciones Antisismicas Universidad National de San Juan - San Juan - ARGENTINA ABSTRACT: This paper describes the testing facility, the methodology applied and the results obtained in the seismic qualification tests of diferent types of electric equipment. These tests were carried out on a shaking table that was developed and built at the Earthquake Research Institute of the National University of San Juan, Argentine. The equipment tested consist of 500 KV and 132 KV current transformers, a 500 KV voltage transformer, a 145 KV disconnecter and a relay cabinet. The acceleration response of the tested equipment was measured at several locations distributed along its height, and strains were measured at critical points by strain gauges cemented on the base of the porcelain insulator. All the information was recorded with a data acquisition system at a sampling rate of 200 times per second in each channel. The facility developed at this Institute is the largest one in operation in Argentina at present and the equipment tested is the highest, heaviest and more slender one which has been seismically qualified on a shaking table in this country. These tests have been a valuable experience in the field of structural dynamic testing applied to equipment of hydroelectric and nuclear power plants. 1. INTRODUCTION Earthquakes have often affected the electric power systems causing the interruption of power supply to industries and homes, which sometimes has extended during several days. Some components of high voltage substations have shown to be very sensitive to the shaking of strong earthquakes. Measurement transformers,disconnec-ting switches, circuit breakers and other similar substation equipment of 220 KV or larger voltage have been severely damaged on the 1978 Miyagi-Oki, Japan earthquake (Katayama 1980), on the 1987 Bay of Plenty, New Zealand earthquake (Rutledge 1988) and on the 1989 Loma Prieta and 1995 Northridge, USA earthquakes (EERI 1990, 1995). The restoration of the operation of this equipment demands time which produces important economic losses that are larger than the reposition cost itself. Also, some equipment could be essential for assuring the security of power plants. With the purpose of carrying out dynamic tests on full scale substation electric equipment subjected to seismic motion, the Earthquake Research Institute of the National University of San Juan, Argentina, has designed, built and put into operation a shaking table. With this testing facility the seismic qualification tests of two 500 KV measurement transformers and other equipment have been recently performed (Figures 5 to 9). 367 2, THE SHAKING TABLE The shaking table mentioned above has one horizontal degree of freedom and was designed and built keeeping in mind that the electric equipment to be tested are slender and have normally the centre of gravity in such a position that generates a very important seismic overturning moment at its base. The testing facility is located in San Juan City which, with its mild and very dry climate, allowed the facility to be built outdoors. This fact also simplifies the erection of the equipment to be tested which has, in the case of 500 KV current transformers, a total height a little larger than 10 meters. The shaking table motion is produced by a PC computer controlled electro-hydraulic actuator. The operation limits of this facility are shown in Figure 1. The installation has also a data acquisition system to measure and record the cinematic variables which correspond to the motions of the shaking table and the equipment under testing. The whole installation has been designed by the authors of this report, including the frame and supports of the table, the actuator, the transducers and electronic circuits of the data acquisition system and its software. 3, SEISMIC QUALIFICATION TEST MOTIONS The electric and mechanical equipment seismic qualification tests are the experimental approach to demonstrate the ability of the equipment to perform its required functions during and after the occurrence of earthquakes. About the seismic activity in Argentina, the most destructive earthquakes have occurred at the centre of the western part of its territory, (Volponi 1962). For example, in November 23, 1977 the city of San Juan was shaken with Mercalli Intensity VIII by one Ms = 7,4 earthquake from an epicentral area at a distance of 60 km. Figure 4 shows the acceleration record obtained in San Juan city on that occasion with a maximum acceleration 0,17g whereas one Wilmot seismoscope with a period 0,7 sec and 10% of damping located on the same site recorded 0,26g as spectral acceleration. (Carmona 1978). The 500 KV measurement transformers tested will be installed in substations of hydroelectric power plants located at Limay River in Comahue Region, one thousand km. to the south of San Juan City, where the seismic activity is lesser than in this place. To seismically qualify these equipment Hidronor, the owner at the time of these tests, specified the spectral response accelera-tion curve shown in Figure 2 with a maximum value of 0,26 g. The shaking table motion that was specified to fulfill this require-ment was a sine-beat type acceleration. It was specified a sequence of 5 sine-beats separated by quietness intervals, each sine-beat having five complete sine waves with amplitude modulated by a half sine wave and period equal to the fundamental natural period which has the equipment under testing in the direction of the shaking table applied motion (Figure 3). An example of another motion applied with this shakinq table is shown in Figure 12. This accelerogram was derived from the record obtained in San Juan City in 1977 and aplied to a 5 tons structural model. 4, PERFORMED TESTS ON 500 KV MEASUREMENT TRANSFORMERS. To perform the seismic qualification test, the transformers were mounted on the shaking table with its bolted steel tower support rising to a total height of 10,5 m. It is very important to properly reproduce the service condition since the tower support changes the dynamic response of the device (Figures 5-6). 368 The instrumentation includes 6 accelerometer transducers distributed along the height of the device under testing and 2 strain gauges cemented on the base of its porcelain insulator (Figure 9), all of them connected to the data acquisition system in which the information was measured and recorded as digital data at a sampling of 200 times per second in each channel. The tests carried out on the shaking table have had two stages. In the first stage one sloped step and sinusoidal scanning motions were applied in order to determine the natural periods, mode shapes and damping of the electric device under testing. After the identification of the dynamic parameters the required motions for the equipment qualification were applied. The amplitudes were successively increased until the response spectral acceleration of the equipment tested was equal or larger than that specified on Figure 2, which was 0,26g for both transformers as a consequence of their natural frequencies. In the current transformer the maximun response spectral acceleration obtained during the test was 0,30g whereas in the voltage transformer it was 0,38g, both larger than the required value given in Figure 2. Figure 11 shows the acceleration-time curve measured in the upper part of the current transformer during one of the strongest sine-beat shaking table motions and also it is shown with a dotted curve the response calculated using the modes and frecuencies obtained by a minimization output error method (Zabala,1993) . The 2,5 and 3% damping acceleration response spectrum curves of the motion applied are shown in Figure 10. Finally, it must be pointed out that after the shake neither through visual inspection nor through electric measure tests any disturbances or damage on the measurement transformers tested have been detected. In this way, the seismic qualification test of these electric devices has been successfully completed. 5. FINAL REMARKS. It should be pointed out that the testing facility built is at present the largest one of its type in Argentine and the equipment tested is the highest, heaviest and slenderest one which has been seismically qualified on a shaking table in this country. Furthermore, even though this shaking table has only one degree of freedom, the tests performed have been a valuable experience for a better understanding of the seismic behavior of special electric devices. In the near future other degrees of freedom will be added to this testing facility in order to be able to better represent earthquake motions. 6. REFERENCES Carmona, J. S. , et al. 1978. El sismo de Caucete, San Juan, Argentina, del 23 de Noviembre de 1977 y la Seguridad que Proveen las Normas Sismo-Resistentes.- SEMINARIOINTERNACIONAL SOBRE PREPARACION PARA ATENCION DE CATASTROFES. Vina del Mar, Chile. Carmona, J. S., R. P. de Carmona & B. G. de Ugrin, 1988. Millenary Occurrence of Seismic Intensity: Its Evaluation by a Mathematical Model of Mean Seismic Activity.- PROC. NINTH WORLD CONFERENCE ON EARTHQUAKE ENGINEERING : Vol. 2.-65-Tokyo, Japan. E.E.R.I. 1990- Loma Prieta Earthquake Reconnaissance Report, Lifelines.- EARTHQUAKE SPECTRA: Vol. 6 Sup-314.- California - USA. E.E.R.I. 1995 - Northridge Earthquake Reconnaissance Report, Lifelines.- EARTHQUAKE SPECTRA: Vol. 11- California - USA. IEC-50 A - 1983. Guide for Seismic Testing Procedure for Equipments- INTERNATIONAL ELLECTROTECHNICAL COMMISSION.- Geneva- Switzerland. 369 IEEE Std 344-1975 -Recommended Practices for Seismic Qualification of Class IE Equipment For Nuclear Power Generating Stations.- THE INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS- New York- USA. Katayama, T., Y. Masui & R. Isoyama 1980. Restoracion of Lifelines in Sendai after the damage caused by the 1978 Miyagi-Ken-Oki Earthquake.- PROC. SEVENTH WORLD CONFERENCE ON EARTHQUAKE ENGINEERING: Vol. 8-233.- Istanbul. Turkey. Rutledge, A. L. 1988. Earthquake Damage at Edgecumbe and Kawarau Electricorp Substations in the Bay of Plenty Earthquake on 2 March 1987.- BULLETIN OF THE NEW ZELAND NATIONAL SOCIETY FOR EARTHQUAKE ENGINEERING: Vol. 21-N2 4-247. Wellington - New Zealand. Volponi, F. 1962. Aspectos Sismologicos del Territorio Argentino.- ACT AS DE LAS PRIMERAS JORNADAS ARGENTINAS DE INGENIERIA ANTISISMICAS: Tomo 1-51.- San Juan, Mendoza, Argentina. Zabala, F. 1993. Identificacion de modelos matematicos de comportamiento electrico a partir de ensayos en mesa vibratoria. V Encuentro Regional Latinoamericano de la CIGRE. Ciudad del Este. Paraguay. 370 FRECUENCY (C/SEC) 100 50 20 10 5 2 1 0,5 0,2 0,1 0,05 ! o 0 -4 1 ;^-H 1 H~r—i i l-TsrH 1 1" 10° 0,01 0,02 0r05 0,1 0,2 0,5 1 2 10 20 PERIOD (SEC! FIGURE 1. Shaking table operation limits. OISPLACEMENT/Anax T2 | SAC/03) ACCELERATION/A max 1,00 - •-, "max •JM: 0,50 O,8Z| / \ A \ \ 1 \ 0,00 0,0'- P'!I -0,50 -0,024 -1,BO -0,8Jx M 1,2 T(sec) (a) ACCELERATION (b) EMPLACEMENT FIGURE 2. Acceleration response spec- FIGURE 3. Sine-beat type shaking table motion required by trum required by Hidronor. Hidronor. w^^ ACCELERATION 10. Amax = 17.0 cm/sec- Vroax » 30,4 cm/sec- 20- 0- A A VELOCITY 20-- cm DISPLACEMENT 10.. 1.4 cm i I I I • 1 I I t *. | 5 10 20 30 40 50 60 T(sec.) FIGURE4. Acceleration record of San Juan (Argentina), Nov.23,1977 earthquake. 371 FIGURE 5. 500Kv current transformer with its FIGURE 6. 500Kv voltage transformer ready supporting tower. for the test. ~'-:*A". FIGURE 7. 145Kv disconnectors with accele- FIGURE 8. Relay cabinet mounted on the sha- ration transducers. king table. 372 TI500KV WILMOT SEISMOSCOPE 0.4 - 0.2 - 0,0 -, I i i : i i i i ! i i i i 0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 PEP10D (sec.) FIGURE 9. 500Kv current transformer test FIGURE 10. Qualification movements response instrumentation. spectrum. 0.48 -.20- - . -10 " I l.B 2.8 3.8 4.8 SEG FIGURE 11. Acceleration history measured at the top of the 500Kv current transformer. e.B -i.e — 5-6 18. a SEC FIGURE 12. Example of shaking table applied motion. NEXT PAGE(S) left BLANK 373 SESSION VI "CASE STUDIES" NEXT PAGE(S) left BLANK ___ ~"* 375 1 XA9952666 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES DESIGN AND IMPLEMENTATION EXPERIENCE OF SEISMIC UPGRADES AT KOZLODUY AND PAKS NPPs V.Borov, V.Trichkov, A.Alexandrov, M.Jordanov EOE-Bulgaria, Sofia, Bulgaria ABSTRACT: Series of upgrades have been designed and implemented by EQE-Bulgaria at Kozloduy NPP and as a subcontractor of EQE-International - at Paks NPP. Wide variety of facilities have been upgraded, including Electrical Equipment, Control and Instrumentation Equipment, Technological Equipment, Brick Walls and Building Structures. Different design approaches and concepts have been applied in compliance with the specific technological and structural conditions. The effect of the excitation intensity as well as the presence of specific floor response spectra over the upgrading concept and cost is discussed. Specific problems of supporting heavy technological equipment are noted. A practical approach for seismic upgrading of Brick Walls, as well as a tendency for unification of the engineering design is shown. The first completely upgraded Building Structure at Kozloduy NPP is the structure of the Electrical Control Building to the Diesel Generator of the River-bank Pump Station. Specific problems of the implementation of the final upgrading design of the Diesel Generator Building are outlined. 1. FOREWORD Series of upgrades have been designed and implemented by EQE-Bulgaria at Kozloduy NPP and as a subcontractor of EQE-International - at Paks NPP. Wide variety of facilities have been upgraded, including Electrical Equipment, Control and Instrumentation Equipment, Technological Equipment, Brick Walls and Building Structures. The purpose of the seismic upgrade projects has been to implement possible changes immediately in order to improve the seismic safety of the power plants With the exception of the building structures, the overall approach has been increasing the seismic capacities of VVER-440 units by identifying and designing modifications for so-called "easy fixes- All the work on the seismic upgrades in Kozloduy NPP and Paks NPP has started in 1992 and 1993 respectively [1,2] and some final designs and implementations have been completed in 1995. During this period considerable efforts have been made by experts from different institutions, essentially supported by IAEA, to develop more adequate seismic input characteristics for the seismic qualification of elements and components associated with the safe shutdown of the plants. This included development of a Free-field Response Spectrum for Kozloduy and different Floor Response Spectra for both NPPs. On the other hand, there is a considerable difference in the seismic hazard prediction for the sites of Paks and Kozloduy NPPs. being respectively 0.35g peak ground acceleration for the first site and 0.2g PGA for the second one. All this has inevitably affected the 377 qualifications and the design of upgrades, ending sometimes not only with heavier or lighter fixing elements, but even in changing the upgrading concept. Another significant factor affecting the design of seismic upgrades of existing facilities is obviously the free space and the existence of strong structural elements near by, such as reinforced concrete floors, ceilings, walls, etc. It has to be noted, that very often initial concepts, that have seemed possible and reasonable in the beginning, have to be principally changed in the end. The implementation of the projects for seismic upgrading often is impossible because of lack of space, especially concerning the upgrading of building structures. 2. SEISMIC INPUT 2 1. SEISMIC INPUT FOR KOZLODUY NPP In the beginning of 1992 EQE-Bulgaria, under contract with Kozloduy NPP, started the design of high priority short term seismic upgrades for the four VVER-440MW operating units at Kozloduy NPP [3]. At that time the seismicity of the Kozloduy site was still under investigation, so the estimation of the seismic capacity of the equipment was done on the basis of an anticipated earthquake of 0.25g peak ground acceleration and assuming a broad banded response spectrum similar to a US Nuclear Regulatory Commission standard spectral shape [4], Besides, in the design of seismic anchorage upgrades of specific electrical power and C&I cabinets the seismic input loads were to be based on the following equation, defining a static horizontal acceleration component coefficient for 5% damping, Ft, [5]: Fh = (0.8 + 2.0Hs/Hn)g where : Hs = height from the building base to elevation X Hn = total height of the building. The value of was to be reduced by a factor of 0.5, if the fundamental frequency of the upgraded equipment could be shown to be equal to or greater than 8 Hz. The vertical component acceleration coefficient was to be taken as 0.67Fh. The Bulgarian Building Research Institute developed specific site response spectra, approved by IAEA in the end of May 1992 [6]. The seismic evaluation was to be conducted for a safe shutdown earthquake defined as 0.2g horizontal peak ground acceleration with 50% of this value for the vertical component. On this basis a team of Energoproject S.A. lead by Mr. M. Kostov developed floor response spectra for units 1-4 using detailed computer models of the structures [7]. The last seismic qualifications and upgrades of equipment for units 1-4 were conducted using the floor response spectra. 2.2. SEISMIC INPUT FOR PAKS NPP Paks NPP had retained EQE-International and Westinghouse Energy Systems Europe (WESE) to participate in strengthening plant equipment which had insufficient seismic capacity. EQE-Bulgaria took part in the program as a subcontractor of EQE-International. The goal of the program was to determine which components and structural elements associated with the safe shutdown of the plant following an earthquake had seismic capacity less than 0.3g peak ground acceleration and to increase their seismic capacity to withstand a O.35g earthquake (as a minimum). 378 The screening of existing equipment and commodities has been carried out for a 0.3g peak ground acceleration using 50% spectral shape defined in NUREG/CR-0098 [8]. For design of seismic upgrades the more conservative design spectrum has been used. The ground motion was specified as a US NRC Regulatory Guide 1.60 spectral shape anchored to 0.3g [4]. The seismic analysis has been conducted by Siemens [9], They have developed floor response spectra for all locations of equipment in the reactor building and the turbine hall. The analytical technique has followed the guidelines of the US NRC Standard Review Plan [10] for conducting soil-structure interaction. The spectra provided were the result of enveloping all locations on a particular floor with broadened and smoothed peaks. The spectra were to be linearly scaled to 0.35g by multiplying by the ratio of 0.35/0.30 and were to be used in the design of upgrades. 2.3. COMPARISON OF THE SEISMIC INPUT FOR THE DIFFERENT CASES In order to assess the effect of the different seismic input for the separate cases a brief comparison of the inputs is made. For the purpose the different free field spectra for 5% damping are plotted on Figure 1. INITIAL K^ZLODUY jlNPUT (TO THB FORMULA) r r L 0.5 1.0 5.0 10.0 50.0 Frequency (Hz) Fig. 1 Comparison of 0.3g NUREG/CR-0098 spectrum, 0.3g Regulatory Guide 1.60 spectrum, Kozloduy site 0.2g spectrum and Initial input for Kozloduy, 5% damping It is obvious, that the seismic input for Paks is about twice the finally developed input for Kozloduy. On the other hand the initial input for Kozloduy according to the formula is respectively close to the 0.3g RG 1.60 Spectrum. The maximum spectral value of the acceleration on the Kozloduy site has dropped 1.74 times and the maximum horizontal floor accelerations should have dropped about two times and more. In general it might be concluded, that the initial input for Kozloduy NPP site and the one for the Paks NPP site do not differ substantially. The development of site specific response spectrum for Kozloduy, the development of detailed structural models and from there - of floor response spectra, has diminished the input seismic excitation for the structural elements and equipment considerably. 379 3. UPGRADING DESIGN APPROACHES AND CONCEPTS The general approach is to use seismic and testing experience and expert judgement, supplemented by analysis, to demonstrate the seismic adequacy of equipment. The demonstration of seismic adequacy includes the strength of the components and their anchorage as well as the operability. "Generic Implementation Procedure for Seismic Verification of Nuclear Plant Equipment" has been developed by the Seismic Qualification Utility Group (SQUG) in 1991 [11]. It covers twenty generic classes of active equipment as well as cable trays, tanks and heat exchangers. The strength is demonstrated for the most part by providing that the equipment is adequately anchored. Operability is demonstrated by assuring that the equipment is similar to equipment in the experience database and that it meets all the inclusion rules contained in the Generic Implementation Procedure 3.1. CONCEPTS FOR UPGRADING OF ELECTRICAL, INSTRUMENTATION AND CONTROL EQUIPMENT For better understanding of the approach for supporting and upgrading the cabinets it is essential to remind, that initially for Kozloduy the value of horizontal excitation was to be reduced by a factor of 0.5 if the fundamental frequency of the upgraded equipment could be shown to be equal to or greater than 8 Hz. This is completely reasonable and complies well with the character of the standard response spectra , such as RG 1.60 (see the graph on Fig. 1). From experimental data the natural frequencies of such type of cabinets are found to be in the range of 3 to 10 Hz [12]. As the usual structural scheme for the cabinets is a cantilever beam, it is natural to choose an approach to support the cantilever at the top (Fig. 2), thus changing the natural frequency and drifting it in the range of 12 to 30 Hz. . CSd CABINET fn=3-10Hz fn=12-30Hz 3 / I 1 support M M support anchorinc I Fig. 2 Structural schemes for dynamic calculation of cabinets In this way it is considered, that not only the seismic stability of the whole cabinet is provided, but the seismic effect over the electrical and control and instrumentation devices is reduced. In cases with rows of cabinets the supporting at the top is preceded by uniting the separate cabinets . Uniting at the top is usually done with bolts and uniting members, the latter being pieces or continuous element along the whole row. The supporting at the top has different opportunities in accordance with the cabinet location - to be anchored to the ceiling, to a neighbouring concrete wall, to the floor, to a perpendicular neighbouring row of cabinets, by uniting cabinets arranged in "horse shoe" and by vertical supporting frame. Typical details are shown on Figures 3,4,5,6. 380 5 Fig. 3 Uniting of cabinets and supporting to the ceiling 1 - <• Z-'Z 3-3 Uniting of Cabinets and Supporting to the Ceiling and the Floor 381 n n /•/ UNITING MEMEBER «? LB0.8 '"V -r- C10 I i, ANCHORING /•;,£». 6 Vertical Supporting Frame It was found in Paks, that the seismic vulnerabilities of equipment were almost identical to those determined for Kozloduy and that the same "easy fix" backfits would be generally applicable for Paks Practically the same concepts were applied as the initial input for Kozloduy was comparatively close to that assumed for Paks (Fig. 7). 382 Fig. 7 {'-niting of Cabinets and Supporting to the (''eiling in 1'aks M'l' The development of the floor response spectra for Kozloduy units 1-4 gave the opportunity to avoid the necessity to change the system's natural frequency for cabinets located at lower levels Even more - in the "easy fix" calculations for these cabinets the highest spectral values of the acceleration for a certain level was applied, regardless of the natural frequency. The reactions in the anchoring points were calculated to be small enough to allow very simple detailing and constructing. Simple anchoring elements with one or two anchoring bolts were applied (Fig. 8). A-A CABINET COLUMN V- Fig. H Anchoring Detail At the same locations very simple unification at the top was applied, often comprising of direct bolting one cabinet to the other. 3.2 TANKS AND OTHER TECHNOLOGICAL EQUIPMENT Thin-walled tanks and similar equipment had been supported in Kozloduy and Paks in a similar manner. They are usually supported at the top and anchored at the bottom. Supporting of vertical tanks at the top if possible is preferred because of the much greater mass compared to the cabinets For the lightest tanks and in cases, where the seismic excitation is shown to be weak enough, only strengthening of the legs and improvement of the anchoring was applied (Fig. 9). 383 /'"/#. 9 Improvement the supporting of tanks in I'aks NPP As the upgrades are usually designed as steel supports, elements and anchorings, the technological and structural restrictions for minimal size do not allow taking advantage of lower excitations and realising any considerable material economy. For control of the expenses for upgrades the backfit concept is of the greatest importance. Sometimes there are special problems with supporting heavy technological equipment, especially of such, included in the primary circuit. Equipment of this type is predominantly made of stainless or other special alloyed steel and welding to its corpse is not only difficult, but very often forbidden. All this requires a special approach for the support concept and design. Interesting examples are some tanks in Paks and Kozloduy NPPs (Fig. 10,11) and the pressurisers in Kozloduy NPP units 1-4 (Fig. 12). ID Seismic supporting of n Steam iiciieralor Illnutluwn Separator lank in Ko:hxliiy 384 /'/«. / / Seismic: supporting of a Steam (ieneralor Blowilown Separator lank in Kuzlodtty - detail The pressurisers are supported to the neighbouring reinforced concrete walls, by means of special elements (Fig. 12). It would be interesting to mention, that all the connections are dismountable. 12 Seismic supporting of I'ressunser to the wall in Kozloduy NI'P 385 10 3.3 SEISMIC UPGRADING OF BRICK WALLS. Masonry brick walls are classified as major sources for seismic interactions. The seismic interactions may result from the falling of one item which is not essential (brick wall) into an essential item. That is why if the neighbouring equipment is in the safe shutdown path, the brick walls are to be reinforced and supported. A practical approach for seismic upgrading of brick walls is adopted. The main concept is to support the walls by means of vertical standard channel steel, anchored to the ceiling, the floor and the wall itself. The calculation procedure is reduced to determining the distance between the vertical supporting profiles, the last depending on the mortar tensile strength, the wall height and the seismic input. Typical details were developed for anchoring the steel channel to the floor (Fig. 13) and to the top according to the type of ceiling (Fig. 14,15). ANCHORING TO FLOOR X BOLT HILTI Fig. 13 Anchoring the steel channel lo the floor It should be noted, that a significant economy had been realised by applying testing procedure of HILTI HIT C20 anchor bolts [13]. The application of this procedure allows the reduction of the safety factor v from 3 to 1.25 thus reducing the number of anchorings more than two times. This is important especially for anchorings of the channel to the brick wall, where their quantity is very Fig. 14 Anchoring of the supporting steel channel to the ceiling - variants 386 11 ROOF DETAIL T1 REINFORCE SEAM DETAIL T4 UNITING MEMBER IH2 10.100 BOLT ROOF DETAIL T9 / REINFOftCEB BEAM I / < / I / ' / 1 / HILTI BOLT ' / 'v/ I / uoa8 / U12 .!_,. 'A/ 1 ^ n | U12 I U12 ^^ / I .1 j i 1 1 II *" ~25O0 —4= 20 •/ILJLJ /'7g. /5 Anchoring of the supporting steel channel to the ceiling - variants 387 12 3.4 SEISMIC UPGRADING OF BUILDINGS EQE-Bulgaria under contracts with Kozloduy NPP has elaborated projects for seismic upgrades of a series of building structures including: Diesel Generator Building 2, Central Pump Station 2, Autonomous Pump Station, Electrical Control Building to the Diesel Generator Building of the River-bank Pump Station, Diesel Generator Building of the River-bank Pump Station and as a subcontractor of EQE International - Spent Fuel Storage Building. The design process itself is difficult enough and there are a lot of problems related to it, that will not be considered in this paper. As the process of implementation of these projects has already started it is better to point out some specific problems that have appeared. The general idea in the seismic upgrading of buildings is to construct new bearing elements to take the seismic forces from the top of the building to the foundations such as: strengthening of roofs and floors, vertical steel X braces and outer steel braces, new shearwalls, in cases - new additional foundations. The first completely upgraded building structure at Kozloduy NPP is the structure of the Electrical Control Building to the Diesel Generator Building of the River-bank Pump Station [14]. The upgrading of the building of the diesel generator itself is almost completed. These buildings are typical for the Kozloduy site, erected of precast reinforced concrete elements. The general upgrading approach for such buildings is by placing vertical steel X braces between the reinforced concrete columns, spanning from the roof to the basement floor (Fig. 16). Fig. 16 New vertical steel X braces in the Diesel Generator Building 388 13 At some places, if possible, casting of new reinforced concrete walls is applied. The braces are anchored to the concrete columns or to the concrete floors. Typical detailing is shown on Fig. 17 /-/». / 7 Detail of a new vertical brace anchoring to column and floor In order to avoid insufficient anchoring, the attachment of the bracing members to the existing concrete columns is done by drilling the columns and placing uniting rods threaded on both sides. By means of nuts the ending plates of the bracing members are fastened to the column faces (Fig. 18). EXISTING RE-BARS ENDING PLATE VERTICAL BRACE PASSING THREADED RODS EXISTING CONCRETE COLUMN WITH'NUTS"""" "" /•is,'. Jfi Detail of brace anchoring lo column Specific problem of the implementation of the final detailed design is placing the uniting rods passing through the concrete columns. This is so because of the production displacements of the reinforcing bars from their prescribed positioning. Sometimes the necessary passing rods are to have greater diameters and conflict with the re-bars of the column are possible because of uncertainties and admissions. Finally new position of the bracing has to be chosen, which is not always possible. Moving slightly the bracing axis from the column axis is not usually a serious structural problem, but 389 14 is a serious implementation problem if the bracing member comes into conflict with neighbouring cable trays, cabinets, tubes or ducts of different type. The moving of the neighbouring device is not always possible not to mention the additional expenses. The best decision in such case is to move the brace to another span between the columns if it is structurally possible. Unfortunately it happens that the whole upgrading concept has to be changed, followed by new calculations and design respectively requiring additional time and efforts. Generally the application problems are connected with absence of free space to place the upgrading members. Moving of existing ducts and equipment usually comes out to be more expensive and time consuming than the very upgrades. All this brings up again the importance of precising the seismic excitation and limiting the conservatism to a reasonable point in order that smaller upgrading elements could be applied without safety loss. 4. CONCLUSION The implementation of the programmes for improving the seismic safety of VVER-440 type units in Eastern Europe aims to increase the reliability of safety systems and reduce the influence of the external events in the total risk of nuclear accidents. The re-evaluation of the design basis and design implementation from the point of view of the modern criteria for designing and construction of NPPs [15], show the leading influence of earthquakes in the global nuclear risk. Substantial investments were directed to seismic upgrades. The approach for their realisation includes three stages. I and II stage - short-term programmes realised during regular annual outages including the seismic upgrade of technological, electrical and C&I equipment, cable trays, pipelines and brick walls which are source of risk for such equipment; III stage - long-term programmes including seismic upgrade of building structures. Stages I and II are almost completed in Kozioduy NPP with the exceptions of some complex pipelines and cable trays. It is possible to perform a risk reduction assessment of Kozioduy NPP based on probabilistic safety analysis (PSA) of a Top Level Risk Study (TLRS), realised by EQE-Bulgaria [16]. According to general data the realisation of stages I and II has reduced the nuclear risk of an earthquake with a factor of 7 to 10 and the influence of the earthquake as an initial external event from 50% (or dominant) to 20-25%. The realisation of the stage III will reduce the influence of the seismic event below 10% which means that it will not predominate in a quite lower level of risk. In spite of all the specific problems and difficulties of the implementation of seismic upgrades of nuclear facilities in Eastern Europe the work is proceeding and there is a considerable effect already. REFERENCES [1] "Interim Report on Seismic Evaluation of Kozioduy NPP, Units 1-4", EQE-International, June. 1992 [2] "Report on Seismic Assessment Walkdown of Paks NPP", Robert Campbell, EQE-International. Prepared for IAEA, Project Number INT/9/122-04, 1993 [3] "Short Term Seismic Upgrading of Kozioduy NPP, Units 1 - 4", EQE-Bulgaria Reports No 0202-D-02xx, 0202-D-03xx, 1992-1993 390 15 [4] U.S. AEC Regulatory Guide 1.60 "Design Response Spectra for Seismic Design of Nuclear Power Plants", Revision I, December 1973 [5] "Terms of Reference and Technical Specifications for Seismic Upgrading Design of Kozloduy NPP Unit 1 and 2", IAEA, Item HB of WANO Programme, 1992 [6] "Seismic Safety Review Mission on Design Basis Earthquake for Seismic Safety Upgrading of Kozloduy NPP (2nd Mission)", Final Report, Project. BUL/9/012-14 of IAEA, Sofia, Bulgaria 26-29 May, 1992 [7] M.Kostov, et a!., "Floor Response Spectra for Units 1&2 at Kozloduy NPP", Final Report, Vol.3, Energoproject, October 1992 [8] N.M.Newmark, W.J.Hall, "Development of Criteria for Seismic Review of Selected Nuclear Power Plants", NUREG/CR-0098, May 1978. [9] Halbritter, Dr., "Paks II - Main Building Design Response Spectra for Seismic Loadings", Siemens Work-Report, KWU/R324/92/E019, 1992 [10] US NRC, "Standard Review Plan for the Review of Safety Analysis Reports for Nuclear Power Plants," NUREG-0800, 1989 [11] "Generic Implementation Procedure (GIP) for Seismic Verification of Nuclear Power Plant Equipment", Rev.2, SQUG, 1992. [12] A.P.Kirillov, Yu.K.Ambriashvili, "Seismic Stability of Nuclear Power Stations", IAEA-TC- 472.2, Vol.2, IAEA, Vienna, 1989 [13] EQE-Bulgaria Technical Document No. 0106-TI-007: "Recomendations for Implementation and Testing of Anchor Bolts HILTI HIT C20", 28 March 1994 [14] "Seismic Upgrading of DGS Building of the River-bank Pump Station", EQE-Bulgaria, Report No. 0210-D-02xx, 1993 [15] "Design and Evaluation Guidelines for Department of Energy Facilities Subjected to Natural Phenomena Hazards", R.Kennedy, S.Short, et al., UCRL-15910, June 1990. [16] "Assessment of Activities Carried Out for Implementation of the Programme for Upgrading of Operational Reliability and Safety of Units with VVER-440 (V230) Reactors at Kozloduy NPP", EQE-Bulgaria Report No. 0230-R-001, 13 February 1995 NEXT PAGE(S) left BLANK I —" " 391 XA9952667 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES SEISMIC UPGRADING OF WER 440-230 STRUCTURES, UNITS 1/2, KOZLODUY NPP D.Stefanov, M.Kostov, H.Boncheva, G.Varbanov Central Laboratory for Seismic Mechanics and Earthquake Engineering, Bulgarian Academy of Sciences, Sofia ABSTRACT: The purpose of this paper is to present final results from a big amount of computational work in connection with the investigations of the possibilities for upgrading of VVER 440-230 structures, units 1/2, Kozloduy NPP. 1. Introduction The first NPP's with VVER-type reactor structures have been designed and built without consideration of seismic influences and on the base of simplified calculation. It's the case with NPP Kozloduy. In connection with change of site seismic characteristics and safety demand the necessity rise of checking up and ensuring of technology systems seismic resistance of the existing 440-MW VVER-type units in Bulgaria. 2. Description of the existing structure The Kozloduy NPP consists of four Units of type VVER-440/230 and two units of type VVER-1000. Units 1 and 2 are of the first type and they are constructed as twin units, i.e. Unit 2 is a mirror image of Unit 1 with a temperature expansion joint in between. The layout of Units 1 and 2 is schematically shown in Figure 1. The expansion joint is in axis 22. The main building is composed essentially of two parts - the reactor building (between rows C and D - Figure 2) and the turbine hall (between rows A and B). Between them is located the longitudinal intermediate part. Next to the reactor building are the smaller buildings of the ventilation centre and the control rooms connected to the reactor building. The reactor building consists of many massive reinforced concrete walls, shells and slabs irregularly distributed. The roof structure is constructed by steel trusses mounted on reinforced concrete columns. The turbine hall is a regular frame structure - longitudinal RC frames and steel trusses in transverse direction with hinge joints between them and the RC columns. The later are founded on separate foundations. The roof is made of prefabricated RC panels. The turbine hall structure is divided in two equal parts by an expansion joint of 5 cm in axis 12. The longitudinal intermediate building is constructed mainly by precast RC girders and floor panels. This part connects the reactor building and the first part of the turbine hall. The second part is independent. 3. The three dimensional model of the structure Previous investigations (1,2,3,4) proved the necessity of creating a complex 3D model of the main structure. The entirely different dynamic behaviour of the reactor building and the turbine hall lead to some spatial effects in the seismic response of the structure. The two parts of the turbine hall structure (separated by an expansion joint) have also different behaviour. 393 Unfavourable torsional effects appear and dominate the seismic response. As a result some structural elements (beams and columns in the turbine hall structure) as well as the longitudinal intermediate building will be overloaded during on earthquakes. That is why a detailed three-dimensional finite element model of the soil-structure system has been used. The seismic input motion at the foundation level is computed by deconvolution of the design "free field" motion represented by three components of a generated acceleration time history. The local geological conditions are taken into consideration. The design "free field" spectrum and the response spectrum of the N-S component at foundation level are shown in Figure 3. In the model the soil is represented by springs and dashpots corresponding to its stiffness and damping characteristics. The spatial structural model consists of 3-D beam elements with 6 degrees of freedom at each node and 3-D rectangular hybrid finite elements with 5 degrees of freedom per node. All columns, longitudinal beams and girders and roof trusses are modelled as beam elements. The roof and floor panels, slabs and shells are modelled with rectangular elements. 4. Investigation of the original structure Dynamic and static analyses of the original structure are performed using program STARDYNE. The verification of the mathematical model is made using the results from a full-scale test. The first mode of vibration is shown in Figure 4 and the fourth mode - in Figure 5. The main characteristics of the response are: - the response of the independent part of the turbine hall building (between axes 1 and 12) is primarily in transverse direction. The rotation of that part can be clearly seen. - the response of the other part (between axes 12 and 24) is predominantly in longitudinal direction. - the intermediate building is loaded in an unfavourable way because of the different stiffness of the reactor building and the turbine hall. - the displacements of some control nuda! points (at the ends of the turbine hall "tail") are larger than the permissible ones. Several variants of combination of the internal forces due to the static and dynamic loading are performed in order to get the most unfavourable loading condition. The position of the crane is changed in different places of the turbine hall. The bearing capacity of all structural elements is checked. The girders at the upper levels in the longitudinal frames of the turbine hall are assessed that they could not resist the respective forces. The bearing capacity of almost all columns in raw "A" and some of columns in row "B" is found as insufficient. The final conclusion is that the structure should be upgraded. 5. Investigation of the upgraded structure The basic idea of the structure upgrading consists in an increasing of the stiffness by adding of additional elements - mainly diagonal bracing for the RC frames, steel stretch bars, girders, etc. The distribution of those elements in the existing structure is of great importance because the stiffness concentration should be avoided. The places of the additional elements should be in accordance with the technological requirements too. The stiffness of the existing structure could be increased also by strengthening of the structural elements cross section. The existing rigid structural elements (e.g. the stair-cases) should be used to support the more flexible parts of the structure. Different variants of strengthening are analyzed (5). For each separate case a capacity checking of the bearing structural elements is performed. The displacements of some nodal points are controlled also. 394 Several alternatives of upgrading are investigated in order to find the optimum solution. The final variant (Figure 6) incorporates the following additional elements and connections: 1. Steel bracing diagonals 2L 125/125/10 in the spans of the RC frames shown in Figure 6. 2. Steel stretch bars 130 along axis 10 (from row"B" up to the end span of diagonals) at four levels in control room. 3. The girders at level 18.70 m in whole row "A" and in the "tail" only of row "B" are strengthened connecting the two girders in a "box-like" cross section. 4. All columns in row "A" and the columns in the "tail" of row "B" are strengthened in the upper part (over the crane path) by steel plates with dimensions 20/2 cm placed in the corners of each column. 5. There are stiff beam connections between two adjacent columns (forming the expansion joint) at all levels of girders in longitudinal frames. 6. In the intermediate building (between rows "B" and "W") in the spaces between levels 20.80 m and 28.40 m "K-bracing" 2L 125/125/10 are put. 7. There is a steel connection (2L 125/125/10) between turbine hall, row "B" at level 28.40 m and the reactor building, row "W" at level 33.60 m. 8. There are two inclined connections (2U20) between control rooms and the frame in row "B". 9. Additional steel girders are put between the control room and the longitudinal frame of the turbine hall. The dynamic analysis of the "upgraded model" shows that the added stiffness is sufficient - the displacements of some characteristic nodal points are smaller than the allowable ones. The reduction of the displacements is shown in Table 1. The first mode of vibration is shown in Figure 7. The fundamental period is reduced considerably. A capacity check of all structural elements is performed according to Bulgarian Code for design of concrete and reinforced concrete structures (Sofia, 1988). The checking is done for eccentric compression or eccentric tension about the two principle axes of the cross section. As a criterion for safety of the elements is assumed the ratio of the bending moment of maximum internal forces to the moment of the external forces. A ductility factor is used element by element to account the nonlinear capacity of the RC members. The factor used for columns is 1.25 and for girders 1.5 respectively. Several positions of the crane are investigated. First the crane is put in axes 20,21,22. The result is overloading of the columns at these axes. This is the reason to put "K-bracing" in the intermediate building. The second position of the crane is in axes 10,11 and 12. This place is a typical one because there is an expansion joint between the two parts of the building. The third position of the crane is in axes 5,6 and 7 (middle of the "tail" structure). Looking at the shear force values in all elements there are not considerable problems. Nevertheless additional connection are needed between the top of columns (level 28.40) and steel trusses. 6. Optional upgrading of the structure The idea is to use the possibility of fixing the parking position of the crane and perform upgrading for that particular load case. In that way a considerable amount of upgrading structures could be saved. The new option is described as it follows: 395 1. Crane in the turbine hall is located in axes 12 13 and 14. This is one usual parking position of that crane. 2. Static and dynamic calculation of the complete building is performed only for that particular crane position. As a result of this analyses the following reduction of the already proposed upgrades can be done: - remove the upgrades of the RC columns above level 18. m in row "A" and "B" everywhere except row"A", axes 13 and 14. - remove the "K-bracing" in axes 22,21,20 and 19. A precondition for acceptance of that optional proposal is the elaboration and the strict control for implementation of an operational procedure to use the crane in turbine hall. The procedure should minimize the crane stay in other positions and should assure the parking position as described. 7. Conclusion The main building of Unit 1/2 is a complicated structure - there is a great variety in height and stiffness of the different structural parts. The analyses of the seismic behaviour have shown weak elements that should be improved. This is achieved by strengthening of the building. There are two main requirements - decreasing the large displacements and increasing the bearing capacity of some structural elements. In the case of Unit 1/2 the structure is upgraded by adding additional stiffness but also by connecting stiff structural elements with more flexible parts in order to redistribute seismic forces. As a whole the structural stiffness is increased and the displacements due to seismic response are decreased. The bearing capacity of the upgraded structure is checked, the dynamic behaviour of the upgraded building satisfies the requirements for seismic safety of critical facilities. The seismic upgrading of existing NPP is usually very complicated and expensive. The engineering solution is to find the optimum variant between the economy and safety of the upgrading and to satisfy these conflicting requirements. 8. Acknowledgements This work is performed in cooperation with Energoproject PLC, Sofia. The authors are expressing deep acknowledgement to Dr. Arturo Ordonez, Empresarios Agrupados, Spain, for the highly professional support during the investigation process. 9. References 1. Kostov M. et al. Floor response spectra, Unit 1/2, Kozloduy NPP, Final report C31 - 128, Energoproject, Sofia, 1992. 2. Kostov M. et al. Seismic capacity of turbine building, Final report, Task 2-3. Six month WANO Program, Item HB. Energoproject/WESI/EA. 1992. 3. Kostov M., D.Stefanov, H.Boncheva Response and capacity evaluation of Unit 1-2, Kozloduy NPP, SMiRT-12 Post Conference Seminar No.16, Vienna. 1993. 396 4. Rostov M., H.Boncheva, D.Stefanov, Main features of the Units 1-4 building complex, Kozloduy NPP in respect to seismic safety., SMiRT-12 Post Conference Seminar No. 16, Vienna. 1993. 5. Kostov M. et al. Design seismic upgrades for the turbine building. Task 3-6. Six month WANO Program, Item HB. Phase 3, Energoproject/WESI/EA. 1994. TABLE 1 NODAL POINT DISPLACEMENT (m) ORIGINAL AND UPGRADED MODELS Nodal Along X-direction Along Y-direction point Original Upgraded Original Upgraded model model model model Row "A" , Level 28.40 797 .3642 .0374 .1293 .0294 792 .3387 .0362 .1294 .0294 Row "B" , Level 28.40 803 .3643 .0374 .0661 .0271 796 .3387 .0336 .0662 .0271 Row "A" , Level 18.70 594 .2719 .0287 .0863 .0304 397 s .« \i I u—c—u u "\i'ii u-u—u u u u u u Figure 1. Main building, Layout Figure 2. Main building, cross section 0.50 0.00 PERIOD /•/ Figure 3. Acceleration response spectrum at foundation level and design spectrum, N-S component 398 ORIGINAL STRUCTURE'S MODEL Figure 4. First mode of vibration, Tl=1.3915 s ORIGINAL STRUCTURE•S MODEL Figure 5. Fourth mode of vibration, T4=0.5377 s 399 UPGRADED STRUCTURE ' S MODEL Figure 6. General view UPGRADED STRUCTURE • S MODEL Figure 7. First mode of vibration, Tl=0.5815 s 400 1 XA9952668 PROCEEDINGS OF SMiRT 13 - POST CONFERENCE SEMINAR 16 SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES SEISMIC UPGRADING OF PIPING SUPPORTS IN VVER 1000MW Martin F. Schmidt Stuessi & Partner, Engineering Consultants, Zuerich, Switzerland ABSTRACT: Russian VVER type nuclear power plants were initially designed for a horizontal earthquake peak acceleration of O.lg. Since a level up to 0.2g is now considered more realistic, a reassessment of the seismic capacity of safety relevant piping systems and supports is of great importance. An improved behaviour can be obtained by adding more or better support devices. In the current paper, two types of inexpensive and easily implementable motion limiting devices are compared. 1. INTRODUCTION The improvement of the support configuration for an existing piping system requires an optimal choice of the number, position and type of additional support devices. In the current work, two easily implementable motion limiting devices are examined and compared: viscous dampers and gaps. An important issue is the tradeoff between stresses in pipings and support loads. 2 MOTION REDUCTION BY GAPS The introduction of gaps is an inexpensive and effective method to reduce seismically induced large amplitude motion while allowing free thermal expansion and free low-amplitude vibrations. The energy accumulation of a resonant mode is limited mainly by perturbing the phase lag between excitation and response such that a resonant behaviour is not possible (Messmer, 1993). A sample piping system has been used to numerically investigate the behaviour of a piping subsystem with motion reducing gaps. The system shown in figure 1 consists of a straight pipe of 10m length, simply supported at both ends. The cross sectional properties and overall dimensions correspond approximately to segments of the emergency feedwater system of VVER 1000MW reactors. In a first step, the system was excited by a sine ground acceleration input with 0.5g amplitude and a frequency near the resonant frequency of the pipe without gap support. The response has been analyzed during 15 seconds starting from a motionless initial state. Without gap, a maximum amplitude of 559mm is reached. In the following, this amplitude was limited by introducing a symmetrical 20mm gap with a finite stiffness. The non-linear support stiffness is shown in figure 2. Several choices of the gap stiffness have been tested. In figure 3, the maximum stress in the piping system and the maximum gap reaction and displacement are plotted against the relative gap stiffness for a 20mm symmetrical gap. 401 Figure 1 - Sample piping subsystem with gap support Relative gap stiffness Force 48 El I3 Displacement Figure 2 - Non-linear stiffness of the gap element 10 Maximum displacement (cm) 8- -B- Maximum bending stress (daN/mm2) Maximum gap reaction (kN) 10 20 30 40 50 60 Relative Support Stiffness K Figure 3 - Maximum gap reaction and maximum stress in pipe as a function of the relative stiffness K 402 From figure 3, it can be seen that the stress in the pipes decreases for increasing support stiffness up to a relative stiffness of about 20 and then increases again. This is due to the apparition of high frequency motion with very hard spring characteristics. For a relative support stiffness between 5 and 50, the variation of the maximum stress in the piping system varies little. Since low values for the gap stiffness lead to lower gap support reactions, a value of 5 for the relative stiffness parameter would be a good choice in the present example. 3 APPLICATION OF REALISTIC EATH-QUAKE MOTION TO GAP AND VISCOUS DAMPER SUPPORTS A typical earth-quake motion in a horizontal axis at building level 30m of the Kozloduy lOOOMW reactors has been applied to the sample system described in section 2. The acceleration input is shown in figure 4. The system response has been obtained for the system without support, with a gap of 20mm and a stiffness of 200'000 N/m (relative stiffness 8.7) and with an idealized viscous damper of 1500 kg/s. The characteristic results are reported in table 5. The displacement of the beam central section is shown in figures 6 to 8 for each of the three configurations. —i —I C .00 1.00 10.00 •5.CO 20.00 25.00 30.00 Time (s) Figure 4 - Acceleration input at level 30m Configuration Maximum Stress Maximum Maximum stress in pipe reduction displacement reaction force (N/mm2) factor (mm) (N) Without support 52 1 54.4 — Gap 27 0.52 26.3 1251 k=200'000 N/m Viscous damper 22 0.42 21.5 268 c=1500 kg/s Table 5 - Characteristic response values 403 Figure 6 - System without additional support - Displacement at central section (in m) —I —i —I 1 0.00 5.CG 10.CO i i . C 0 20.00 25.00 20.00 Tine (s) Figure 7 - System with gap - Displacement at central section (in m) —I —i —I —I coo 5.00 10.00 IS.CO 20.00 25.00 30.00 Tirr.e (s) Figure 8 - System with viscous damper - Displacement at central section (in m) 404 From the results presented above, the following conclusions can be drawn: • Both gaps and viscous dampers allow an efficient reduction of the stress in pipes • Viscous dampers lead to lower support reactions for a comparable reduction of the pipe stresses. Energy accumulation is reduced by an increased modal damping factor. From figures 7 and 8, it is evident that the average vibration amplitude is much lower for the system with viscous damper although the peaks are of comparable magnitude. The main advantage of the gap solution is the very low cost. In fact, a stiffness such as the one used above can be realized easily by a typical steel construction of the support (cantilever beam, beam frame, etc.). As general guideline, gap motion reduction is appropriate for light structures (low reaction forces, high bending stresses) whereas viscous dampers are useful in systems where high reaction forces are the dominant problem. 4 PRACTICAL CONSIDERATIONS FOR VVER 1000MW REACTORS Actual calculations on the 1000MW VVER plants are currently just starting such that only few results are available. Preliminary results indicate that most of the safety relevant piping subsystems (systems required for the safe shutdown of the reactor after an earthquake) receive rather low seismic stresses. This is due to the fact, that the piping subsystems are rather short and have large cross-sections. For the VVER primary sytem under study, upgrades, if any, will most probably be oriented toward reduction of the reaction forces at supports or strengthening of the support structures. Considering rhe results of section 3, an application of viscous dampers could be of interest if large amplitude resonant motion occurs. 5 CONCLUSIONS Motion reduction by gaps and viscous dampers in seismically excited piping systems has been studied by numerical examples. Both devices allow an efficient reduction of pipe stresses and both are inexpensive and easy to implement. Gaps are probably the most efficient means for the motion reduction of smaller diameter piping subsystems. Though the implementation of gaps or viscous damper elements is straightforward, their analysis needs some additional consideration. In fact gaps are highly non-linear elements, and localized viscous dampers result in complex mode shapes. In the current study, calculations were performed by direct time integration. If modal superposition or response spectrum analysis are to be used, linearization procedures can be applied (see for example zum Felde & Haas or Tang, Jaquay & Larson). REFERENCES zum Felde, P.& Haas, E. 1987. Consideration of local damping mechanism in modal FEM analysis of piping systems. In SMiRT 9 - Transactions of the 9th International Conference on Structural Mechanics in Reactor Technology, Lausanne. RotterdanrBalkema Messmer, S. 1993. Repeated impacts in a piping system under seismic excitation. In SMiRT 12 - Transactions of the 12th International Conference on Structural Mechanics in Reactor Technology, Stuttgart. Elsevier Tang, H.T. & Jaquay, K.R: & Larson, J.E. 1987. Simplified nonlinear dynamic piping analysis methodology development. In SMiRT 9 - Transactions of the 9th International Conference on Structural Mechanics in Reactor Technology, Lausanne. Rotterdam:Balkema NEXT PAGE(S) I left BLANK | — 405 XA9952669 METHODOLOGY AND RESULTS OF THE SEISMIC PROBABILISTIC SAFETY ASSESSMENT OF KRSKO NUCLEAR POWER PLANT M.K. Vcrmaut, Ph. Monette IVestinghoitse Energy Systems Europe. Brussels. Belgium 20. Bvd. Paepsem. 1070 Brussels. Belgium lei. 32-2-5568625; fax. 32-2-5568758 R.D. Campbell EQE International. Irvine. CA. USA ABSTRACT A seismic IPEEE (Individual Plant Examination for External Events) was performed for the Krsko plant. The methodology adopted is the seismic PSA (Probabilistic Safety Assessment). The Krsko NPP is located on a medium to high seismicity site. The PSA study described here includes all the steps in the PSA sequence, i.e. reassessment of the site hazard, calculation of plant structures response including soil-structure interaction, seismic plant walkdowns. probabilistic seismic fragility analysis of plant structures and components, and quantification of seismic core damage frequency (CDF). Also, relay chatter analysis and soil stability studies were performed. The seismic PSA described here is limited to the analysis of CDF (level 1 PSA). The subsequent determination and quantification of plant damage states, containment behaviour and radioactive releases to the outside (level 2 PSA) have been performed for the Krsko NPP but are not further described in this paper. The results of the seismic PSA study 407 indicate that, with some upgrads suggested by the PSA team, the seismic induced CDF is comparable to that of most US and Western Europe NPPs. 1. INTRODUCTION This paper describes the seismic Probabilistic Safety Assessment (PSA) performed for the Krsko plant. Krsko is a Westinghouse 2 loop PWR. The safe shutdown earthquake level specified for design is 0.3g PGA, with Reg.guide 1.60 design response spectra. The seismic PSA is one of the options for performing a seismic Individual Plant Examination for External Events (IPEEE), i.e. examining NPPs for beyond design basis loadings. A seismic margins assessment is another alternative. However, for Peak Ground Accelerations (PGAs) above 0.5g, a PSA is the only acceptable method. The PSA study was conducted in strict accordance with the criteria specified by the USNRC for the evaluation of NPPs for beyond design basis events and was reviewed by the IAEA. The seismic PSA described here is limited to the analysis of core damage frequency (CDF)(level 1 PSA). The subsequent determination and quantification of plant damage states, containment behaviour and radioactive releases to the outside (level 2 PSA) have been performed for the Krsko NPP but are not further described in this paper ([4]). The different sections of this paper describe the successive building blocks of the seismic PSA study. Obviously, no detailed methodology descriptions can be provided for each of the sections. This paper intends to illustrate the application of the seismic PSA methodology to the Krsko plant, using the Krsko specific assumptions, inputs and results. 408 2. SEISMIC HAZARD A site-specific seismic hazard analysis was prepared for the Krsko NPP site by the University of Ljubljana Institute of Structural and Earthquake Engineering. The process of the Krsko NPP site seismic hazard analysis is not further described in this paper; only results from the process, to be used in the Krsko seismic PSA, are included here. The hazard analysis has resulted in the determination of probabilistic hazard curves and uniform hazard response spectra. The probabilistic hazard curves expressing frequency of exceedance as a function of PGA are shown in Figure 2.1. PGA is the motion input parameter in terms of which seismic fragilities (see sections 3 and 4) are most commonly expressed. Figure 2.2 shows the probabilistic response spectral accelerations corresponding to a uniform hazard of 10000 years. These spectral shapes, referred to as Uniform Hazard Spectra (UHS) were used in the calculation of soil structure interaction and building response analysis (section 3). Local earthquakes. Accelerographs installed in the buildings of the Krsko NPP and surroundings have recorded several small magnitude local earthquakes in the past. All records demonstrate very short duration of strong ground motion (less than 1 second). The input energy of such ground motion is very small. High frequencies are clearly predominant in the response spectra for the local earthquakes (sharp peaks occurring in the frequency range 11-12Hz). Accelerograms, obtained at the foundations of buildings simultaneously with the free field motion, are systematically much smaller than at those on tne surface. Studies following a 1989 local earthquake at Krsko NPP have aimed at numerically simulating this reduction in acceleration. According to 409 experience, such ground motions do not damage buildings and equipment located in the buildings. Rather, the concern is restricted to functional failures of devices such as relays which are sensitive to high frequency albeit small displacement motion. Theoretically, these frequently occurring small local earthquakes could affect the shape of the UHS and the seismic hazard curves, leading to a higher seismic risk. However, acceleration spectra, traditionally used for design of structures and used for the fragility analyses of this Krkso seismic PSA study, do not provide any information on the duration of ground motion and do not take into account this parameter, which is of great importance as a measure of input energy. It was therefore believed not correct to combine the influence of strong earthquakes with 'standard" characteristics (larger magnitude earthquakes from distant sources), and of weak local earthquakes with short duration and predominant high frequencies. The characteristics of the first type are defined by spectra obtained by the probabilistic seismic hazard analysis as described above (results shown in Figures 2.1, 2.2). The second type corresponds to small local'earthquakes. An idealized spectrum (see Figure 2.3) is used to represent the latter. It was concluded from the seismic hazard analysis that peak ground accelerations greater than 0.5g were not expected from local earthquakes. However, data were insufficient to develop a probabilistic description of the hazard due to local earthquakes. The approach taken in the Krsko PSA study is to assume that a local earthquake PGA of 0.6g will not be exceeded in less than 10000 years, making the local earthquake PGA hazard comparable to the PGA hazard for distant earthquakes. Structural response studies performed on an equivalent basis for the local and the distant earthquakes show that response spectra for the distant earthquakes generally exceed the corresponding spectra for local earthquakes. Exceptions where the local earthquake spectra exceed the distant earthquake spectra are limited to some higher building elevations, and to a 410 narrow frequency band around I l-12Hz. Given the non-damaging character of local earthquakes, only the impact of the local earthquake spectra on the relay seismic capacity evaluations is considered to be of any significance. 3. SOIL STRUCTURE INTERACTION AND BUILDING RESPONSE ANALYSIS For seismic PSA purposes, it is of fundamental importance to obtain realistic estimates of structural responses to the postulated seismic events. In general, floor response spectra and structure member forces developed for the Final Safety Analysis Report (FSAR) are considered to be conservatively biased. Hence it was decided to generate new seismic structural responses using current state-of-the-art techniques, and to avoid any intentional bias in the analysis with respect to soil-structure modeling. In order to generate seismic results in a form convenient for the development of structural and equipment fragilities (section 4), a probabilistic approach was adopted. The structures included in the study were the Main Complex (MC), the Diesel Generator Building (DGB) and the Essential Service Water Intake Structure (ESWIS). The MC is formed by the reactor building, intermediate building, control building, fuel handling building.auxiliary building and component cooling building. Since all the buildings at the MC are on a common foundation, the analyses were performed considering all of them. The objectives of this part of the study were twofold: • To estimate median structure forces and the variability about the median for all major structures of interest, for input to the seismic fragility analysis of these plant structures (section 4). 411 • To develop probabilistic floor response spectra in all major structures for use in the seismic fragility analysis of equipment located within the plant structures. The approach to probabilistic response analysis was to perform multiple deterministic SSI analyses using the methodology described herebelow. Input motion and SSI parameters (structural frequency and damping, and soil shear modulus and damping) were sampled following the Latin Hypercube Sampling method. As a result of multiple deterministic analyses using the sampled input values, distributions were obtained of the analysis results - i.e. loads in structural elements and in-structure response spectra. These distributions are then described by the median (50th percentile) values and the variability (represented e.g. in the 84th percentile curve). As both the seismic hazard and the structure/component fragility curves (consistently) use the PGA as the reference seismic input parameter, SSI and probabilistic structural response analysis were performed for a reference PGA value. However, direct scaling of results from one earthquake level to another is not strictly correct due to nonlinearity in soil behaviour. Also, due to the complexity of the structural model and the probabilistic (multiple time history) analysis method used, a single level of earthquake was desired rather than multiple earthquakes. Past studies have shown that the greatest risk comes from earthquakes 2 to 3 times the SSE. For the Krsko PSA, 2 times the SSE level was chosen as the level that would challenge the weaker elements of the plant which would govern risk. For those components with much higher capacity, scaling the response for an input of 2 times the SSE would tend to be conservative since higher input levels that would challenge these components would result in more attenuation in soil-structural amplification. 412 In probabilistic response analysis, the characteristics of the free-field ground motion is defined by the shape of the median uniform hazard spectrum (UHS) corresponding to a return period of interest. For the Krsko PSA, the UHS shape corresponding to the 10000 year return period was used (see Figure 2.2). The elements of the SSI and probabilistic response analysis are outlined below. The approach is based on work performed under the SSMRP (Seismic Safety Margins Research Program, [2|). Analysis results are also provided below. • Specifying the free-field ground motion. Since the SSE level for Krsko is 0.3g PGA, the median UHS shape for the probabilistic analyses was anchored to a PGA of 0.6g. To perform the probabilistic analysis, an ensemble of 30 earthquakes was developed to capture the randomness of the seismic input. The median (50% non exceedance probability - NEP) matches the median UHS, and the 84th percentile (84% NEP) of the spectra matches the 84th percentile of the UHS, as is shown on Figure 3.1 (UHS spectra anchored to 0.6g PGA). To account for the effects of deconvolution in the SSI analysis of the Main Complex, the motion at the embedment depth of this structure was determined by deconvolving the surface ensemble of the time histories, using soil properties compatible with the other analysis steps. For comparison with Figure 3.1. Figure 3.2 shows the comparison between the 50% and 84% NEP of the deconvolved spectra with the 50% and 84% NEP UHS. • Development of the soil models, i.e. defining the soil profile and performing the site response analysis. For the low strain soil properties and the dynamic soil properties, best estimate values were obtained from previous studies. A site response analysis was performed for the 0.6g PGA level to establish median strain compatible soil properties. For the probabilistic SSI and 413 response analysis, the distribution of soil parameters was required. A lognormal distribution was taken for each parameter (soil shear modulus and soil damping), with a coefficient of variation based on previous work and expert judgement. • Calculating the foundation impedance functions and wave scattering effects. The high strain soil properties obtained above were used to develop impedance functions for the three structures (MC, DGB, ESWIS). • Determining the fixed-base dynamic characteristics of the structure. Structural models developed for the original Krsko design analyses (and reported in the FSAR) are representative of current procedures, and may be considered as best estimate models for the purpose of this study. SSI effects were incorporated using foundation impedance functions to replace the soil springs representing the supporting soils flexibility in the original design analysis. As for the soil properties, the structural frequencies and structural damping are probabilistic parameters which were assigned lognormal distributions and typical coefficients of variation representing all modelling and random uncertainty in the estimation of the median values. The largest variabilities for the Krsko analyses are in the soil parameters. • Performing the SSI analysis, i.e. combining the previous steps to calculate the response of the coupled soil-structure system. The SSI and structural response analysis results of interest include peak accelerations, maximum member forces, and floor acceleration time histories. These quantities are needed for downstream fragility development. Floor acceleration time histories computed for each of the 30 simulations performed were post- processed into 5% damped floor response spectra. For each location, the spectral accelerations were fitted to a lognormal distribution and the median and 84th percentile values were 414 extracted. An example comparison between the calculated median in-structure response spectra and the Krsko FSAR design spectra is given in Figure 3.3 (the example applies to the Main Complex). The most notable difference between the FSAR and the median centered spectra is the frequency at which the spectral peak occurs. This shift can be explained through a shift in the dominant frequency of the SSI and structural response, which is caused by the lower median soil stiffness properties corresponding to the 0.6g PGA earthquake level which is higher than the 0.3g PGA SSE level used in the FSAR. Local earthquakes. As indicated in section 2, a distinction was made between low energy local earthquakes and large magnitude distant earthquakes. Deterministic SSI and structure response analyses were performed for a representative local earthquake which was determined to be an approximate 84th percentile amplification from recorded close-in earthquakes. Response was compared, on an equivalent basis (a median response analysis of a 84% NEP close-in free field input response spectrum), to results from the response analyses (84% NEP response to 50% NEP free field input) for distant earthquakes. Analysis showed that the local earthquake frecfield motion is attentuated considerably. Indeed, the system (soils + structure) frequency is not in the amplified portion of the input spectra of the local earthquake. In contrast, the distant sources with low frequency cause significantly- higher response. Therefore, it is generally seen that the floor response spectra from the distant earthquakes envelop the corresponding floor response spectra calculated for the local earthquake. However, for a limited number of locations at higher elevations in the buildings. 415 the local earthquake did induce higher response than the distant sources, in the limited 11-12Hz frequency range. However, based on the fact that local earthquakes, with the very short duration and small energy input are deemed not to cause damage to structures and not to structurally fail equipment, the exceedances of the local over the global response spectra are only taken into account in the seismic capacity assessment of relays (section 6). 4. FRAGILITY ANALYSIS Background on the probabilistic seismic fragility curve representation and development is provided in [1]. 4.1 Seismic Walkdowns Past experience in conducting seismic PSA and seismic margin assessments has shown that the walkdown is generally a very beneficial task in a seismic IPEEE. A walkdown conducted by experienced engineers is valuable in order to identify any potential seismic vulnerabilities and, using knowledge regarding the performance of structures and equipment in strong motion seismic events, screen out the inherently very rugged components and assemble data on the components for which plant specific fragility curves will be developed (to a required level of detail as described in section 4.2). The Krsko PSA seismic walkdown scope of survey included: • structures (MC, DGB, ESWIS) • safe shutdown equipment including support systems: pumps, tanks, heat exchangers, diesel generator system, batteries, HVAC, electrical cabinet 416 • piping and piping components (per P&IDs): support configurations. 11/1 issues, valve operator proximity, ... • cable trays: sample • instrumentation and tubing (per P&IDs) The general observation from the walkdown was that the design was conservative and that the plant was quite rugged. Potential vulnerabilities of a few items of equipment were observed. These included: • poor anchorage welding on few electrical cabinets • low bending capacity of support legs of one tank (the corresponding low seismic capacity was confirmed from the fragility calculations of the tank) • control room ceiling support required reinforcement Fixes were recommended for the above issues, as they were easy fixes which would increase the seismic capacity of the components to a generaiiy adopted screening level (as per section 4.2). Seismic-fire and seismic-flooding interaction waikdowns were also performed, in order to identify potential seismic sources of fire and flooding respectively. 4.2 Screening Level for Seismic Fragility Analysis From the safe shutdown equipment lists and following the seismic waikdowns of the plant, 37 equipment items (i.e. individual components or groups of components) were retained for fragility analysis, in addition to the essential structures (MC, DGB, ESWIS). 417 However, from a verification of the impact on CDF. it was determined that structures and components could be 'screened out' if their High Confidence of Low Probability of Failure (HCLPF) was about 0.74g PGA or greater or their median capacity was about 2.0g PGA or greater. If it could be determined through a conservative analysis that the screening level was exceeded for a structure or component, no further detailed fragility analysis would be performed, and the conservatively low screening capacity level wouid be assigned to that structure or component. The impact of this conservatism on the resulting CDF is marginal. 4.3 Seismic Fragilities of Plant Structures and Equipment Seismic fragility curves were calculated for plant structures and components. The seismic fragility of a component is defined by a curve that gives the conditional probability of failure as a function of the reference seismic input motion parameter (PGA in the case of the Krsko study). Randomness and uncertainty are tracked throughout the fragility analysis and incorporated into a family" of probabilistic curves ([ 1]). Sources of plant documentation to support the fragility analyses included original design analysis, seismic qualification reports, plant drawings as well as data and notes on expected limiting failure modes collected during the walkdowns. The determination of the seismic capacity of plant structures normally requires the evaluation of a number of parameters such as strength, inelastic energy dissipation, response characteristics, ... including the determination of median factors and associated variabilities. For all the Krsko structures, the single strength parameter could be demonstrated to be sufficiently high for the screening seismic capacity level to be met, without further detailed 418 evaluation of inelastic energy dissipation nor evaluation of the variabilities associated with the various parameters contributing to the seismic capacity. Similarly, the seismic capacity of the majority of plant safe shutdown components was found to exceed the screening level. Provided the fixes of seismic vulnerabilities identified during the walkdown are implemented, the following is the small list of components for which capacities were calculated to be below the screening level. Component Median capacity (g) HCLPF (g) Condensate storage tank 0.78 0.31 DG control cabinets 1.25 0.46 Refueling water storage tank 1.11 0.48 Battery chargers 1.59 0.58 DG fuel oil tank 1.64 0.67 It should be noted that the calculated seismic capacities of the DG control cabinets and the Battery chargers are based on the design qualification level. It is expected that higher seismic capacities could be demonstrated if qualification test reports to such levels were available. 4.4 Screening of Soils Stability Issues The evaluation of the potential for soil liquefaction is a requirement of the IPEEE. The soil stability evaluations were therefore performed and concluded that the HCLPF was in excess of 0.7g, which is consistent with the screening level adopted for the structures and components. 419 The following list summarizes the credible soil related issues for which an evaluation or verification was performed: • liquefaction potential for yard area soils supporting ESW piping and electrical duct bank • settlement of soils underlying the safety related plant structures • lateral earth pressure on partially buried safety related buildings as well as stability of the essential service water (ESW) pumphouse and intake structure against sliding • stability of the river bank slope at the ESW building and the potential impact of its failure on the intake structure 5. RISK QUANTIFICATION The frequencies of core damage are calculated by combining the component and structures fragilities described in earlier sections, with the plant logic. Event and fault trees are constructed to identify the accident sequences which may lead to core damage. The risk quantification process described here is limited to the calculation of core melt frequency (level 1 PSA). The subsequent determination and quantification of plant damage states, containment behaviour and radioactive releases to the outside (level 2 PSA) were performed for the Krsko NPP but are not further described in this paper ([4]). The Krsko NPP SPSA was performed in such a way as to employ much of the work done in the internal events analysis of the Krsko Individual Plant Evaluation (IPE). That is, the event trees and fault trees developed for the internal events analysis would at most need to be modified to address the specific aspects of the plant or systems response to a particular seismic event. 420 Scismicallv Induced Initiating Event Determination and Frequency Calculation Seismically induced initiating events considered in the CDF quantification are outlined below. The list of initiating events is constructed based on the following process: 1. A choice is made of buildings, structures and equipment used to determine the plant status following the seismic event. 2. Given the failure of each of the items listed in step 1, the plant disposition is defined. Failures with similar results are grouped together into failure groups. 3. A hierarchy among initiating events is developed. The order of the hierarchy is such that, if one initiating event occurs, the occurrence of other initiating events further down the hierarchy are of no significance in terms of plant response. 4. The conditional probability of failure for each failure group is determined from the fragility curves of the components in the failure group. The failure groups defined for the Krsko seismic initiating events are described below. • Break beyond ECCS capacity. This failure is assumed to lead to direct core damage, and is a function of building or SG/RJPV support failure. • Large primary pipe break. This failure is a function of the RCS equipment supports. • Medium primary pipe break. This category includes all pipes of sufficient size to produce a medium LOCA event. The probabilities are estimates based on calculations for appropriately sized piping calculated in the SSMRP Zion analysis [3]. 421 • Small primary pipe break. This category includes all pipes of sufficient size to produce a small LOCA event. The probabilities are estimates based on calculations for appropriately sized piping calculated in the SSMRP Zion analysis [3J. In addition, the failure mode of the reactor coolant pumps which leads to damage of the seals causes a leakage equivalent to the small LOCA. • Emergency service water pumphouse failure. This failure leads to loss of the ESW systems, leading eventually to the loss of component cooling water system heat removal ability. • Secondary side pipe break. • ATWS due to control rod insertion failure. • Loss of off-site power. Note that since the switchyard ceramic conductors have a high probability of failure during a seismic event, loss of off-site power was also considered to be combined with all other initiating events. A generic seismic fragility based on US electrical grids was used for the Slovenian electrical grid. Seismic Event Trees and Fault Trees For each initiating event, an event tree models the plant system performance, and hence the accident sequences leading to different plant states. The event trees developed for the Krsko NPP internal events IPE were used as the basis for the seismic event trees. In general, modifications to internal events analysis event trees are simplifications as branches where failure is assumed to occur with certainty following the occurrence of an earthquake, can be eliminated. The assumed failure is caused by seismic failure of a system which is not 422 seismically qualified (such as the instrument air system causing valves to go to a fail-safe position). The seismic fault trees are defined by the seismic event tree nodes and those components and support systems which are required for successful] system operation represented by the event tree node. The seismic fault trees are put in parallel with the internal events analysis fault trees, i.e. random and seismic fault trees are combined in the CDF quantification process. Several assumptions affecting the construction of seismic fault trees are: • Similar redundant components, generally located in close proximity, simultaneously fail with a probability equal to that of one component. This (conservatively) removes train redundancy while simplifying the seismic fault trees. • Off-site power is assumed not to be recoverable within 24 hours. Events which take credit for system recoveries are generally not possible within the first 24 hours after an earthquake. ' • Operator actions required within 10 minutes after the occurrence of the earthquake are assumed to fail. • Systems which are not classified as seismic category I are conservatively assumed to fail at any seismic activity level (e.g. instrument air) Seismic Hazard Intervals For the CDF quantification, the range of PGA of interest was split into a number of intervals: seismic interval: 1: 0.15-0.25g 2: 0.25-0.35g 423 3: 0.35 - 0.50g 4: 0.50-0.70g 5: 0.70-0.90g 6: 0.90-1.10g For each interval, the median PGA was selected to represent the interval. The corresponding frequency of occurrence was set equal to the frequency of occurrence of PGA values within the interval, as obtained from the seismic hazard curves (see Figure 2.1). Core Damage Frequency Quantification Results Compilation and quantification of the fault trees and event trees leads to insight in the most important core damage sequences, and the most important core damage cutsets i.e. the components whose failures contribute the most to core damage. Figure 5-1 illustrates the distribution of core melt frequency contributions from the different seismic intervals. A review of CDF quantification results leads to the following observations. • The significant contributors to core melt from the first two seismic intervals (PGA < 0.35g) are DG random failures combined with the loss of off-site power. • For higher PGA levels, seismic failures of components begin to appear in parallel with the random failures and loss of off-site power. The significant seismic failures of components involve the diesel generator control panel, the battery chargers, the condensate storage tank, and the refueling water storage tank. As indicated in section 4, the failure probabilities of battery chargers and DG control panels assumed in the analysis are 424 considered to be conservatively low as they were based on limited seismic qualification documentation. • The station blackout initiating event represents more than half of the total seismic core damage frequency. Therefore, if plant modifications are made as a result of the seismic PSA, they should focus on improvements that lower the contribution to core damage by station blackout. Also, from the level 2 analysis, the plant damage states which represent the station black-out sequences contribute by far the largest frequency to containment failure and containment bypass ([4]). 6. RELAY EVALUATION A relay chatter evaluation was also performed as part of the Krsko IPEEE. The purpose of the evaluation was to verify the capacity of relays against chattering, and/or the acceptability of relay chatter in a seismic event. A progressive screening of relays based on at least one of the following criteria was performed: 1. The best estimate seismic capacity of the relay is higher than the screening level of 2.0g PGA. which is consistent with the screening level adopted for plant equipment. 2. Relay chatter which occurs does not affect the ability to achieve and maintain safe shutdown. Progressive screening was applied to extensive lists of relays, switches, contactors and breakers available from plant equipment databases. Screening was performed in any sequence which would permit rapid elimination of groups of relays from the lists. Summarized, the screening was based on the following: 425 • solid state relays and some contacting devices, such as mechanically actuated contacts, are considered seismically rugged. • seismic capacity against chattering is determined for relay types for which test data, generic industry data, ... are available. Relay demand was calculated from floor spectra and cabinet amplification. As explained in section 2, the local earthquake floor response spectra were taken into consideration as well as the distant earthquake response spectra. Of the probabilistic relay capacity description, only the median (best estimate) capacity value was calculated for the relay types and compared to the 2.0g PGA median capacity criterion. A significant number of relays could be screened out as the criterion was met. • relays whose change of state can be tolerated as having no adverse impact on "safe shutdown (no spurious seal-in or latch occurs which would prevent the system from performing its safe shutdown function, or prevents control resets and operational control such as pump restart from the control room or other normal point of control), and/or which can be reset by operator action (within a reasonable time, assumed for the purpose of this evaluation as 30 minutes to 1 hour, and according to existing procedures and based on accessibility of required indications) are screened out. 7. CONCLUSIONS The study was performed in accordance with the criteria specified by the USNRC and was reviewed by the IAEA. During the seismic PSA, a thorough walkdown was conducted. The plant appeared to be rugged and generally designed with ample seismic margin. Some miiior design errors were 426 noted and some walkdown observations indicated potential vulnerabilities of a few items of equipment, for which easy seismic improvements were proposed. With the few upgrades suggested above, the seismic induced core damage frequency is comparable to that of most US and Western Europe NPPs. Valuable insights are obtained from the process of modelling the plant and quantifying the core damage frequency. If plant modifications are considered, they should be benchmarked against the insights. 8. REFERENCES [ 1 ] IAEA-TECDOC-724, October 1993, "Probabilistic safety assessment for seismic events" [2] NUREG/CR-20'15. Vol.9, Lawrence Livermore Laboratory. September 1981, '"Seismic Safety Margins Research Program. Phase I Final Report - SMACS - Sismic Methodology Chain with Statistics (Project VIII)" [3] NUREG/CR-4550. also SAND86-2084. Vol.3. Rev. 1. Part 3, December 1990. "Analysis of Core Damage Frequency: Surry Power Station, Unit 1 External Events". M.P.Bohnatal. [4] "KRSKO SEISMIC LEVEL 11 PRA". P.N.Shah, R.Prior. F.P.Wolvaardt. R.Bastien. presented at the 3rd Regional Meeting on Nuclear Energy in Central Europe, Portoroz, Slovenia, September 1996. 427 Figure 2.1 Krsko NPP Site Fractile Hazard Curves 428 2 TO §* N» I NEK PSHA 94 / ALL TEAMS / ft UNIFORM PSA FOR 10 000 YEARS a s 1.6 -i x 1.5 - / 1 4 - / s 1 3 • / N s. 1 2 - / •O N \ n 1.1 - r> s bfiO.9 - // \ •—'0.8 - l\I J *** V C/3 3 0.7- — (^ 0.6 • S, •Si 0.5 -Iw 0.4 • «-, 1 . •— 0.3 - J —— 0.2 • •—J 1— p 0.1 - 0 • 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 T(s) 7 — 15% —50% —85% ao'. o (srti C rt u D O g O CL. — OJ *-• CL 2- CO a c ^^ OO S« ecira . :oj e s o 8.5. e 1 S c: s E G. o •o o o Jc CO 5-S 3 ika : on S voa/¥s Figure 2.3 Statistics of Six Recorded Local Earthquakes and Proposed Local Earthquake Response Spectrum 430 30 **1 a ** 3.0 ft j^ ft r» 2 ft m M ft cr ft" o o S' 2= Frequency (Hz>» JO ft Legend: Notes: ENSEMBLE 50% NEP UHS anchored to 0.6g pga "I ENSEMBLE 84% NEP Spectra calculated at 5% damping UHS 50% NEP Spectral acceleration in g's ? UHS 84% NEP S' o. C •O c a 1 j o 1 1 • -a JJ o a) o nj y'J/l 3 •a v 1 q V u 01 'Xv u i^. ^-^-— s \ a< w 5u3 <*> ion CD a a S3 o o 4P -0 O ^* o J2 in CO Vc 0) o o V o o w . i wQ uQ Figure 3.2 Response Spectra of the Ensemble of Deconvolved Time-histories vs. 10000 Years Return Period UHS 432 cH aur e c •&i a a 1 u (0 -U "O J Qi TJ xJ V to 4J r-i <6 3 --( O 3 ^H O ^! ^__ • • ^ CD « K / j N / X o & V a) ^^ -—V ^\\ \ \ W \ 01 ^ tn \ men Hi o ~ § a; -H en « Figure 3.3 Comparison of FSAR vs. Median Probabilistic Response Spectrum (Reactor Containment Base, East-West Translation) 433 O a ero o 69 Selimlc Hazard Frequency Core Melt Frequency 2\ & 1 JE-05 3' I 0E 05 »0E-Ofl 6 0ECW < 0E 06 i 2 OEM OOE-00 3*5 mil.1 2 3 Safimli:4 Inlarvt ! 5 6 .> 6 Soltmlc Interval Conditional Core Melt Probability Haz. Freq. Cond. Core Melt Core Melt Freq Int 1 2 54E03 3 914E03 9 942E06 Int 2 7 32E-04 1 274E02 9 326E06 Int 3 3 11E-04 3 406E-02 1 059E05 Int 4 8 20E05 1 628E01 1 335E05 Int S 2 00E05 5 803E-01 1 161E05 Int 6 3 60E-06 9 262E01 3 334E06 >6 1 40E06 1 OOOE*00 1 400E06 Total 5 96E05 > t Stlamlc Inttrvll SMiRT-13, POST CONFERENCE SEMINAR NO. 16(ATS) "SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES' IGUAZU, ARGENTINA, 21 - 23 AUGUST 1995 TIMETABLE Monday. 21 August 1995 08:30 - 09:30 Registration 09:30-10:15 Opening Session (Joint session for Seminars PCS No. 2-10-16) Opening by D. BENINSON, Chairman of ENREN Presentation of Keynote Paper: M. ROSEN (IAEA) 10:15-10:30 Coffee break 10:30-12:30 SESSION I: "Earthquake Experience" 10:30-11:10 1.1 - J.J.JOHNSON 11:10-11:50 1.2 - H. SHIBATA 11:50-12:30 1.3 - P. BASU 12:30-14:30 Lunch break 14:30-16:50 SESSION II: "Country Experience in..." 14:30-15:10 II. 1 - J. STEVENSON 15:10-15:30 II.2 - J. INKESTER 15:30-15:50 - F. HENKEL (Germany) 15:50-16:10 II.3 - R. ANDRIEU 16:10-16:30 P. MONETTE 16:30- 16:50 III.l - A. GURPINAR (1st part-seismic safety) 16:50-17:10 Coffee break 17:10 -18:10 SESSION III: "Generic WWER Studies" 17:10-17:30 III. 1 - A. GODOY (2nd part-technical guidelines) 17:30-17:50 III.2 - M. DAVID 17:50-18:10 III.3 - A. GURPINAR(3rd part-benchmark programme) 435 SMiRT-13, POST CONFERENCE SEMINAR NO. 16 (ATS) "SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES" IGUAZU, ARGENTINA, 21 - 23 AUGUST 1995 TIMETABLE (continuation) Tuesday. 22 August 1995 08:30-10:50 SESSION TV: "Analytical methods ... "(cont'd session III) 08:30-09:10 III.4 - R. CAMPBELL 09:30-10:10 IV. 1 - J.M.ROESSET 10:10-10:50 IV.2 - N. KRUTZIK 10:50-11:10 Coffee break 11:10-12:10 SESSION IV: (continuation) 11:10-11:30 IV.3 - A. ASFURA 11:30-11:50 IV.4 - M. KOSTOV 11:50-12:10 IV.5 - I. DIAZ MOLINA 12:10-13:30 Lunch break 13:30-17:00 Visit to Iguazu Falls 17:30 -19:10 SESSION V: "Experimental methods ... " 17:30-17.50 V.2 - C.PRATO 17:50-18:10 V.3 - C. PRATO 18:10-18:30 V.4 - D. PETROVSKI 18:30-18:50 V.5 - V. KOSTAREV 18:50-19:10 V.6 - J. CARMONA 20:00 - Official Dinner for all participants to the 3 Seminars 436 SMiRT-13, POST CONFERENCE SEMINAR NO. 16 (ATS) "SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES" IGUAZU, ARGENTINA, 21 - 23 AUGUST 1995 TIMETABLE (continuations Wednesday. 23 August 1995 08:30-10:00 SESSION VI: "Case Studies" 08:30 - 09:00 VI. 1 - V.BOROV 09:00 - 09:20 VI.2 - D. STEFANOV 09:20 - 09:40 VI.3 - M. SCHMIDT 09:40-10:00 VI.4 - P. MONETTE 10:00-10:30 Coffee break 10:30-12:30 SESSION VII: "Panel discussions" 12:30-14:00 Lunch break 14:00 Visit to Itaipu Dam (Brazil) - (optional) NEXT PAGE(S) left BLANK 437 SMiRT-13, POST CONFERENCE SEMINAR NO. 16 (ATS) "SEISMIC EVALUATION OF EXISTING NUCLEAR FACILITIES" IGUAZU, ARGENTINA, 21 - 23 AUGUST 1995 LIST OF PARTICIPANTS 1. ALVAREZ, Luis ENACE Angel Gallardo 391 (1A) Buenos Aires Argentina Tel: +54(1)856-7387 Fax: 2. ANDRIEU, Roger EDF 35-37, Rue Luis Guerin - B.P.1212 69611 Villeurbanne Cedex France Tel:+33(72)824051 Fax: +33(72)824010 3. ARAKELIAN, Frederick Republic of Armenia Ministry of Energy & Fuel Armenergy Seismicprojects Institute Yer. HPS-2, Razdan Canyon, Yerevan 375015 Armenia Tel: 7(8852)580 649 Fax: 7(8852)151 805 4. ASFURA, Alejandro EQE International 44 Montgomery St., Suite 3200 San Francisco, California 94104 U.S.A. Tel: +1(415)989-2000 Fax:+1(415)362-0130 5. AVILES, Alejandro N.A.S.A. Central Nuclear Embalse Embalse - Cordoba Argentina Tel: +54(51)244577 Fax:+54(51)244577 6. BASU, P. C. Atomic Energy Regulatory Board Vikram Sarabhai Bhavan, 4th Floor N. Wing Anushaktinag., Bombay-400 09 India Tel:+91(22)5562310 Fax: +91(22)5562343 439 7. BOROV, Vinsent EQE Bulgaria S.A. Chr. Smirnenski 1, 11th Floor 1421 Sofia Bulgaria Tel:+359(2)660417 Fax: +359(2)650039 8. BRUSCHI, Silvestro Techint Wineberg3015 1636 Olivos - Pcia. de Bs. As. Argentina Tel:+54(1)790-9685 Fax:+54(1)318-4745 9. CAMPBELL, Robert EQE International 18101, Von Karman, Suite 400 Inving, CA 92715 U.S.A. Tel:+1(714)833-3303 Fax:+1(714)833-3392 10. CAR,Eduardo University Nac. de Cordoba C.C. 916 5000 Cordoba Argetina Tel: 54(51)60-3800 Fax:+54(51)60-3800 11. CEBALLOS, Marcelo Univ. Nac. de Cordoba C.C. 916 5000 Cordoba Argentina Tel: Fax: 12. CHIVERS, Terry Nuclear Electric pic Berkeley GlosGLl 3 9PB Berkeley Technology Centre United Kingdom Tel: +44(14)53812257 Fax: +44(14)53812659 13. DAVID, Milan David Design Eng. & Cons. 14700Praha4 Ke Krci 7 Czech Republic Tel: +42(2)870375 Fax: +42(2)878674 440 14. DIAZ MOLINA, Ivan D'Appolonia Arturo M. Bas 73 Piso 2 5000 Cordoba Argentina Tel:+54(51) 254225 Fax:+54(51)254225 15. ENGSTLER, Susanne Preussenelektra, Headoffice Tresckowstr. 3 30457 Hannover Germany Tel: 0511/4394178 Fax: 0511/4394187 16. FA JFAR, Peter Univ. of Ljubljana Jamova 2 SI-61000 Ljubljana Tel: +386(61)268562 Fax: +386(61)272696 17. GIULIANO, Alejandro Inst. Nac. prev. sism. Roger Balet 47 - Norte 5400 San Juan Argentina Tel: +54(64)234463 Fax: +54(64)234463 18. GODOY Antonio IAEA Division of Nuclear Safety P.O. Box 200 A-1400 Vienna Austria Tel: +42(1)2060 26083 Fax:+43(1)20607 e-mail: [email protected] 19. GURPEVAR, Aybars IAEA Division of Nuclear Safety P.O. Box 200 A-1400 Vienna Austria Tel:+43(1)2060 22671 Fax:+43(1)20607 e-mail: [email protected] 20. HAUPTENBUCHNER, Barbara Technische Universitat Dresden Mommsenstr. 13 01062 Dresden Germany Tel:+49(351)463 4641 Fax:+49(351)463 7108 441 21. HENKEL, Fritz-O. Woelfel Beratende Ing. P.O. Box 1264 D-97201 Hoechberg near Wuerzburg Germany Tel:+49(931)49708-0 Fax:+49(931)49708-15 22. INKESTER, John E. N.I.I. St. Peters House, Balliol R., Bootle Merseyside L20-3LZ United Kingdom Tel:+44(151)951-4000 Fax:+44(151)922-5980 23. INOUE, Norio Tohuku University Aramaki Aobaku Sendai 980-77 Japan Tel:+81(22)217-7872 Fax: +81(22)217-7873 24. JOHNSON, James J. EQE International 44 Montgomery Street, Suite 3200 San Francisco, Ca 94104 U.S.A. Tel:+1(415)989-2000 Fax:+1(415)433-5107 25. KAMIMURA, KazuDori Nuclear Power Engineering Corporation Shuwa-Kamiyacho Bldg., 2F 3-13 4-Chome Toranomon Minato-Ku, Tokyo 105 Japan Tel: 81-3-3434-4551 Fax:81-3-3434-9487 26. KONNO, Takaaki Kajima Corporation 6-5-30, Akasaka, Minato-Ku Tokio 107 Japan Tel:+81(3)5561-2111 Fax:+81(3)5561-2345 27. KOSTAREV, Victor V. CKIT - Vibroseism Atamanskaya 3/6 St. Petersburg 193167 Tel:+7(812)277-2940 Fax: +7(812)395-1338 e-mail: [email protected] 442 28. ROSTOV, Marin Central Laboratory' for Seismic Mechanics and Earthquake Engineering Acad. G.Bonchev Str. Block 3 1113 Sofia Bulgaria Tel:+359(2)713-3303 Fax:+359(2)268-8951 e-mail: [email protected] 29. KRUTZIK, Norbert Siemens AG, NPP Engineering Berliner Str. 295-303 P.O. Box 101063 63010 Offenbach a. Main Germany Tel: +49 69 807-3355 Fax: +49 69 807-4822 30. LAPAJNE, Janez Geophysics Survey of Slovenia Kersnikova 3 61000 Ljubljana Slovenia Tel:+386 (61) 1320283 Fax:+386 (61)1327067 31. LLOPIZ, Carlos R. Invap S.E. Granaderos 897 P.l-Of. 4 Mendoza Argentina Tel: +54 (61) 253455 Fax:+54(61)253455 32. MASUDA, Kiyoshi Kajima Corporation 1 -2-7, Akasaka, Minato-k4 Tokio 107 Japan Tel:+81 (3)3403-3311 Fax:+81(3) 3470-1444 33. MONETTE, Philippe Westinghouse E.S. Eur. Boulevard Paepsem, 20 B-1070 Brussels Belgium Tel: +32(2)556-86-24 Fax: +32(2)556-87-58 e-mail: Monette.Ph%[email protected] 443 34. NAITO, Yukio Kajima Technical Res. Inst. 2-19-1, Tobitakyu, Chofu-Shi Tokio 182 Japan Tel:+81(424)85-1111 Fax:+81(424)89-7116 35. PETROVSKI, Dimitar Inst. Earthquake Eng. P.O. Box 101 91000 Skopje Rep. of Macedonia Tel:+389(91)111-344 Fax:+389(91)112-163 36. PRATO, Carlos A. University Nacional de Cordoba C.C. 916 5000 - Cordoba Argentina Tel:+54(51)60-3800 Fax:+54(51)60-3800 [email protected] 37. PRATO, Tomas A. University Nacional de Cordoba C.C. 916 5000 - Cordoba Argentina Tel:+54(51)60-3800 Fax:+54(51)60-3800 38. QUEVAL, Jean-C. CEA CEN Saclay Drn/Dmt/Semt/Emsi 91191 Gifs/yetteCEDEX France Tel: +33(1)6908-6652 Fax:+22(1)6908-8331 39. ROESSET, Jose M. The University of Texas at Austin Austin, Texas 78712 U.S.A. Tel: 512-471-4927 or 471-8482 Fax: 512-471-8477 e-mail: [email protected] 40. ROSEN, Morris IAEA P.O. Box 200 A-1400 Vienna Austria Tel:+43(1)2060-22700 Fax: +43(1)2060-2948 444 41. SASAKI, Youichi Nuclear Power Eng. Corp. Shuwa-Kamiyacho Bldg. 2F 3-13, 4-Chome Tor., Minato-Ku Tokyo 1 Japan Tel:+81(3)3434-5695 Fax:+81(3)3434-9487 42. SCHMIDT, Martin Stussi & Partner Steinwiesstr. 30 8032 Zurich Switzerland Tel:+41(1)262-4224 Fax: +41(1)262-4228 43. SHIBATA,Heki Yokohama National University Dept. of Mech. Eng. Tokiwadai, Hodogaya Yokohama 240 Japan Tel: +81(45)335-1451 Fax: +81(45)331-6593 44. STEFANOV, Dimitre Central Lab. Seismic Mech. Acad. G. Bonchev Str. Block 3 1113 Sofia Bulgaria Tel:+359(2)703107 Fax: +359(2)700226 45. STEVENSON, John D. Stevenson & Associates 9217MidestBlvd Cleveland Ohio 44125 U.S.A. Tel: +1(216)587-3805 Fax:+1(216)587-2205 46. TADRA, Takaomi Nuclear Power Eng. Corp. Shuwa-Kamiyacho Bldg. 2F 3-13, 4-Chome Tar., Minato-ku, Tokyo 1 Japan Tel:+81(3)3434-5695 Fax:+81(3)3434-9487 47. UCHIYAMA, Shoji Kauima Technical Res. Inst. 2-19-1 Tobitakyu, Chofu-Shi Tokyo 182 Japan Tel:+81(424)85-1 111 Fax:+81(424)89-7116 445 48. VARPASUO, Pentti Ivo International Ltd. Rajatorpantie 8, Vantaa Fin-01029 Ivo Finland Tel:+358(0)8561-2223 Fax:+358(0)8561-2223 49. YANG, Kejian Tohoku University Faculty of Engineering Aoba, Aramaki, Aoba-ku, Sendai Miyagi, Japan, 980-77 Tel/Fax: 81-22-217-7873 50. YUN, Choul-Ho KINS P.O. Box 114, Yusong Taejeon Korea Tel:+82(42)868-2615 Fax: +82(42)861-2535 51. ZABALA, Francisco National University of San Juan Av. del Libertador 1290 5400 San Juan Argentina Tel:+54(64)228123 Fax: +54(64)213672 52. ZARATE, Stella M. ENREN Av. Libertador 8250 1429 Buenos Aires Argentina Tel: +54(1)704-1494 Fax:+54(1)704-1181 446>