Seismic Isolation Technology March 18-20,1992

Home , Seismic base isolation, Vibration control

International Atomic Energy Agency IWGFR / 87 International Working Group on Fast Reactors

XA0055376

IAEA Specialists' Meeting on Seismic Isolation Technology March 18-20,1992

Proceedings

GE Nuclear Energy

6835 Via Del Oro, %, 3 1/36 San Jose, California 95119

93-077-01 International Atomic Energy Agency IWGFR/87 International Working Group on Fast Reactors

IAEA,Spe^^|Meeting on Seismic Is6itti^OTechnology MarchvlS-20,1992

Proceedings

GE Nuclear Energy

6835 Via Del Oro, San Jose, California 95119

93-077-01 IAEA Specialists' Meeting on Seismic Isolation Technology San Jose, California, March 18-20, 1992

FORWORD

The International Atomic Energy Agency held a Specialists' Meeting on Seismic Isolation Technology in San Jose, California, on March 18-20, 1992. Twenty-three experts from seven countries participated including V. Arkhipov, as the IAEA representative.

The objective of the meeting was to provide a forum for review and discussion of seismic isolation technology applicable to thermal and fast reactors. The meeting was conducted consistent with the recommendations of the IAEA Working Group Meeting on Fast Breeder Reactor-Block Antiseismic Design and Verification in Bologna, October 1987, to augment a coordinated research program with specific recommendations and an assessment of technology in the area of seismic isolation.

Seismic isolation has become an attractive means for mitigating the consequences of severe earthquakes. Although the general idea of seismic isolation has been considered since the turn of the century, real practical applications have evolved, at an accelerating pace, over the last fifteen years aided by several key developments: (1) recent advances in hardware developments in the form of reliable elastomer bearings, (2) development of reliable analytical methods for the prediction of dynamic responses of structures (3) construction of large bearing test machines and large shake tables to simulate earthquake effects on structures for validation analytical models and demonstration of performance characteristics, and (4) advances in seismological engineering. Although the applications and developments of seismic isolation technology have mainly benefitted commercial facilities and structures, including office buildings, research laboratories, hospitals, museums, bridges, ship loaders, etc., several seismically isolated nuclear facilities were implemented: the four 900 MWe pressurized water reactor units of the Cruas plant in France, the two Framatome units in Koeberg, South Africa, a nuclear waste storage facility in France and a nuclear fuel reprocessing plant in England.

The scope of this specialists' meeting was to review the state-of-the-art technology related to the performance of seismic isolator elements and systems, performance limits and margins, criteria for the design, fabrication, testing of seismic isolation elements and systems, the capabilities of analytical codes and models and status of validation.

The presentations provided by the participating countries indicated that seismic isolation technology has sufficiently advanced to make it an attractive feature in advanced nuclear power stations for mitigation of severe earthquakes. Indeed, advanced reactor concept evaluation and studies in Canada, Europe, Japan and the USA include horizontal seismic isolation and in some cases a combination of horizontal and vertical seismic isolation.

The development of seismic isolation elements/bearings seems to be progressing towards standardized designs. The testing programs indicate high quality and consistency in the bearing manufacturing process. Significant progress has been achieved in providing reliable

-n- bonding of elastomer layers and steel laminations that is stronger than the rubber itself. Another important aspect is the demonstrated long term durability of steel-laminated elastomer bearings under sustained loading conditions.

The development of design codes and standards for seismic isolation is proceeding independently in the countries applying this technology consistent with the individual frameworks of regulations. In a future meeting it would be of interest to compare key elements of the design codes and standards in particular as they relate to safety aspects.

In this document summaries of the individual meeting sessions and recommendations are provided. Also included are all the papers presented.

The meeting was hosted by General Electric Company at the company offices in San Jose, California. In conjunction with the meeting, a visit of the Earthquake Engineering Research Center of the University of California at Berkeley was arranged.

Emil L Gluekler Meeting Chairman

-in- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, California, March 18-20, 1992

Table of Contents

Page

1. Forword i i

2. Meeting Agenda 4

3. List of Participants 9

4. Session Summaries and Recommendations

Session 1: March 18, 1992 18 Seismic Isolation Technology Review Chairmen: J. W. Herczeg, US-DOE, USA J. K. Biswas, AECL, Canada Session 2 and 3: March 18-19, 1992 21 Seismic Isolation Element and System Performance Chairmen: A. Martelli, ENEA, Italy J. M. Kelly, UC Berkeley, USA

Session 4: March 19, 1992 2 3 Design Criteria and Standards Chairmen: Ch. Coladant, EdF, France R. Seidensticker, ANL, USA

Session 5: March 19, 1992 2 7 Seismic Isolation System Design and Analysis Chairmen: S. Aoyagi, CRIEPI, Japan M. Olivieri, ANSALDO, Italy

5. Meeting Summary and Recommendations 31

Meeting Chairman: E. Gluekler, GE, USA

-1- PAPERS

Page

Session 1: Seismic Isolation Technology Overview

1. J. Betbeder - Matibet, EdF; P. Bisch, Sechaud et Metz; F. Gantenbein, 33 CEA; Tentative Provisions for Seismic Design of Base Isolated Buildings in France."

2. A. Martelli, ENEA; F. Bettinali, ENEL, "Status Report on Activities on 56 Seismic Isolation in Italy."

3. S. Aoyagi, CRIEPI; H. Shibata, University of Tokyo, "A Broad Review 75 of the Status of Seismic Isolation Study in Japan."

4. J. M. Kelly, University of California at Berkeley, "The Current Status 89 of Seismic Isolation Technology in the United States."

5. L Lin, CIAE, "Some Progress on Seismic Isolation Technology in 112 Building Structure in China."

Sessions 2 and 3: Seismic Isolation Element and System Performance

6. K. Ishida, H. Shiojiri, T. Mazda, Y. Ohtori and S. Aoyagi, CRIEPI, 116 "Recent Results of Seismic Isolation Study in CRIEPI - Tests on Seismic Isolation Elements, Vibration Tests and Observations."

7. G. Bonacina, ISMES; F. Bettinali, ENEL; A. Martelli, ENEA; M. 143 Olivieri, ANSALDO; "Experiments on Seismic Isolation in Italy."

8. F. F. Tajirian, Bechtel; E. L Gluekler, GE; P. Chen, ETEC; J. M. Kelly, 163 UC Berkeley, "Qualification of Seismic Isolation Bearings for the ALMR."

9. R. W. Seidensticker, Y. W. Chang, and R. F. Kulak, ANL, "Summary of 177 Experimental Tests of Elastomeric Seismic Isolation Bearings for Use in Nuclear Power Plants."

-2- PAPERS (Continued)

Page

Session 4: Design Criteria, Standards, Methods

10. H. Shibata, University of Tokyo, "Some Difference of Concepts 199 between Design Guideline for FBR Base Isolation System and Aseismic Design Guideline of LWR in Japan."

11. M. Olivieri, ANSALDO; A. Martelli, ENEA; F. Bettinali, ENEL; G. 209 Bonacina, ISMES; "Development of Guidelines for Seismic Isolation in Italy."

12. K. Ishida, H. Shiojiri, M. Moteki, CRIEPI; H. Shibata, T. Fujita, 221 University of Tokyo, "Recent Results of Seismic Isolation Study in CRIEPI - Design Method."

13. R. Kulak, ANL, "Technical Specifications for the Successful 230 Fabrication of Laminated Seismic Isolation Bearings."

Session 5: Seismic Isolation Design and Analysis

14. J. K. Biswas, AECL, "Study of Seismic Responses of Candu 3 Reactor 241 Building Using Isolator Bearings."

15. S. Kitamura, M. Morishita, K. Iwata, PNC, "3D-Seismic Response of 261 a Base-Isolated Fast Reactor."

16. F. F. Tajirian, Bechtel, "Seismic Analysis for the ALMR." 278

17. F. Bettinali, ENEL; A. Martelli, ENEA; G. Bonacina, ISMES; M. 294 Olivieri, ANSALDO; "Numerical Activities on Seismic Isolation in Italy."

18. H. Shiojiri, K. Ishida, S. Yabana, K. Hirata, CRIEPI, "Recent Results 307 of Seismic Isolation Study in CRIEPI - Numerical Activities."

19. T. Sano, G. DiPasquale, ENEA-DISP; E. Vocaturo, ENEA; "Linear 325 Analysis for Base Isolated Structures."

-3- IAEA Specialists' Meeting on Seismic Isolation Technology San Jose, California, March 18-20, 1992

Wednesday, March 18 Thursday, March 19 Friday, March 20

Session 1, Session 3, Visit of Earthquake Seismic Isolation Technology Seismic Isolation Elements Engineering Research Center, Overview and Systems (Completion) University of California - Berkeley (9:00 to 12:00) (9:00 to 10:00) (9:00 to 12:00)

Session 4, Design Criteria & Standards (10:00 to 12:00)

LUNCH LUNCH

Session 2, Session 5, Seismic Isolation Elements Seismic Isolation Design and Systems and Analysis (13:00 to 16:30) (13:00 to 17:00) Summary 16:00 to 17:30

DOE/GE Hosted Dinner at Mirassou Winery, IAEA Reception (18:00)

ELG93-006 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLA TION TECHNOLOGY MEETING PROGRAM

March 18.1992 Welcome, Introduction. Meeting Objectives IAEA. US-DOE, GE

Session 1i March 18.1992 SEISMIC ISOLATION TECHNOLOGY OVERVIEW

Session Chairmen J. W. Herczeg, US DOE J. K. Biswas. AECL Canada

Summary of Main Features of Base Isolation in France Ch. Coladant EdF, France

Status Report on Activities on Seismic A. Martelli. ENEA. Italy Isolation in Italy F. Bettinali. ENEL, Italy

A Broad Review of the Status of Seismic Isolation S. Aoyagi, CRIEPI. Japan Study in Japan

Status of Seismic Isolation Technology in the US J. M. Kelly. UC Berkeley. USA

ELG93-009 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY MEETING PROGRAM

Session 2 and 3: March 18-19,1992 SEISMIC ISOLATION ELEMENT AND SYSTEM PERFORMANCE

Session Chairmen: A. Martelli, ENEA, Italy J. M. Kelly, University of California, Berkeley, USA

Recent Results of Seismic Isolation Study in CRIEPI - S. Aoyagi, CRIEPI, Japan Test on Isolation Elements, Vibration Tests and Observations cr> Experiments on Seismic Isolation in Italy G. Bonacina, ISMES, Italy i F. Bettinali, ENEL, Italy A. Martelli, ENEA, Italy M. Olivieri, ANSALDO, Italy

Qualification of High Damping Isolation Bearings F. F. Tajirian, Bechtel, USA fortheALMR E.LGIuekler,GE,USA P. Chen, ETEC, USA J.M. Kelly, UC Berkeley, USA

Summary of Experimental Tests of Elastomeric R. W. Seidensticker, ANL. USA Seismic Isolation Bearings for Use in Nuclear Power Plants -

ELG93-009 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY MEETING PROGRAM

Session 4; March 19,1992 DESIGN CRITERIA. STANDARDS, METHODS

Session Chairmen R. Seidensticker,ANLUSA C. Coladant, EdF, France

Some Difference of Concepts between Design Guideline Professor H. Shibata, forFBR Base Isolation System andAseismic Design U Tokyo, Japan

Development of Guidelines on Seismic Isolation in Italy M. Olivieri. ANSALDO, Italy AMartelli.ENEAJtaly F. Bettinali, ENEL, Italy G. Bonacina, ISMES, Italy

Recent Results of Seismic Isolation Study in H. Shiojiri, CRIEPI. Japan CRIEPI - Design Method

Technical Specifications for the Successful R.F. Kulak, ANL, USA Fabrication of Laminated Seismic Isolation Bearings

ELG93-009 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY MEETING PROGRAM

Sessions: March 19,1992 SEISMIC ISOLATION SYSTEM DESIGN AND ANALYSIS

Session Chairmen S. Aoyagi. CRIEPI. Japan M. Olivieti. ANSALDO, Italy

00 Study of Seismic Responses of CANDU3 Reactor Building Using J.K. Biswas, AECL, Canada Isolation Bearings

3-D Seismic Response of a Base Isolated Reactor S. Kitamura, PNC. Japan

Seismic Analysis for the ALMR F.F. Tajirian, Bechtel, USA

Numerical Activities on Seismic Isolation in Italy F. Bettinali. ENEL, Italy A.Martelli,ENEA,ltaly G. Bonacina, ISMES. Italy M. Olivieri, ANSALDO. Italy

Recent Results of Seismic Isolation Study in CRIEPI - Numerical Activities H. Shiojiri, CRIEPI. Japan

ELG93-009 IAEA SPECIALISTS1 MEETING ON SEISMIC ISOLATION TECHNOLOGY GENERAL ELECTRIC COMPANY, SAN JOSE, CALIFORNIA MARCH 18-20,1992 Meeting Participants (shown in picture from left to right)

J. Biswas, AECL, Canada R. Kulak, ANL, USA T. Kanazu, CRIEPI, Japan D. J. Lacey, Nil, United Kingdom Professor H. Shibata, University of Tokyo, Japan F. Bettinali, ENEL, Italy M. Olivieri, ANSALDO, Italy S. Kitamura, PNC, Japan G. Bonacina, IMES, Italy A. Martelli, ENEA, Italy J. Herczeg, DOE, USA V. Arkhipov, IAEA, Vienna Mr. Yoshida, Toshiba, Japan E. Gluekler, GE, USA H. Shiojiri, CRIEPI, Japan S. Aoyagi, CRIEPI, Japan M. Patel, GE, USA Ch. Coladant, EdF, France R. Seidensticker, ANL, USA R. Jetter, ETEC, USA F. Tajirian, BNI, USA

-10- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose. California

LIST OF PARTICIPANTS

Names of Participants Telephone (Fax) Sign-In (Registration)

IAEA Mr.V.Arkhipov Division of Nuclear Power (43-1-234564) International Atomic Energy Agency Wagramerstrasse 5 P.O. Box 100 A-1400 Vienna Austria

CANADA Mr. J. Biswas 416-823-9040 Atomic Energy of Canada Limited (416-823-6120/8006) Sheridan Park Research Community 2251 Speakman Drive Missisauga, Ontario L5K1B2

CHINA Mr. L Luan 9357434 China Institute of Atomic Energy (222373 IAE CN) P. O. Box 275 (64) 102413 Beijing P. R. China

FRANCE Mr. Ch. Coladant Electricite1 de France/SEPTEN (33-78944798) 12/14 Arenue Dutrievox 69628 Villeubanne CEDEX France IAEA SPECIALISTS1 MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, California

UST OF PARTICIPANTS

Names of Participants Telephone (Fax) Sign-in (Registration)

GERMANY Mr. U.F.J. Graubner 49-9131-18-3561 Siemens AG-KWU (49-9131-18-3687) Hammerbacherstr 12+14 P. 0. Box 3220 8520 Eriangen Germany

ro INDIA Mr. R. S. Soni 5563060 X2587/3501 BARC (91-22-5560750) Reactor Engineering Division j Bombay-400085 India

Mr. A. S. Warudkar NPCII W/A India

ITALY Mr. G. Bonacina 36-307611 ISMES (36-307710) Viale Giulio Cesare, 29 A 24100 Bergamo Italy IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, California

UST OF PARTICIPANTS

Names of Participants Telephone (Fax) Sign-in (Registration)

ITALY Mr. F. ENEL-BBI Hydraulic and Structural Research Center Milano, Italy

i CO Mr. A. Martelli 51-498468 ENEA (39-51-498639) via Martiri di Monte Sole, 4 40129 Bologna Italy

Mr. Olivieri 39-10-6558100 ANSALDO S.p.A.-Divisione Ricerche (39-10-444666) CorsoPerrone118 16161 Genova Italy

JAPAN Prof. Heki Shibata 81-3-3402-6213 Institute of Industrial Science (81-3-3402-5078) University of Tokyo 22-1 Roppongi 7 1 Minato-ku Tokyo 106, Japan IAEA SPECIALISTS1 MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose. California

UST OF PARTICIPANTS

Names of Participants Telephone (Fax) Sign-in (Registration)

JAPAN Mr. S. Aoyagi 81-3-3216-6601 Central Research Institute of (81-3-3216-2735) Electric Power Industry 1-6-1 Ohtemachi \ Chiyoda-ku Tokyo 100, Japan

Mr. H. Shiojiri 03-3201-6601 Central Research Institute of (03-3216-2735) Electric Power Industry 1-6-1 Ohtemachi \ Chiyoda-ku Tokyo 100, Japan

Mr. S. Kitamura 292-67-4141 O-arai Engineering Center (292-67-7173) Power Reactor and Nuclear Fuel Development Corporation \ 4002 Narita, o-arai-machi Ibaraki-ken 311-13 Japan

Mr. T. Kanazu, CRIEPI c/o EPRI (USA) P.O. Box 10412 \ 3412 Hillview Avenue Palo Alto, CA 94030 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, Califomia

LISTOFPARTICIPANTS

Names of Participants Telephone (Fax) Sign-in (Registration)

JAPAN Mr. Yoshida 045-770-2406 Toshiba Corporation (22587) Isogo Engineering Center 8, Shinsugita-cho, Isogo-ku Yokohama 235, Japan

UNITED KINGDOM Mr. D. J. Lacey 051-951-3522 Nuclear Installations Inspectorate (051-922-5980) L St. Peters House Stanley Precinct Bootle - Merseyside U.K.

USA Mr. J. Herczeg 301-903-9887 (Host, of US Team Leader) (301-903-4211) U.S. Department of Energy Office of Nuclear Energy Washington, DC 20545

Mr. E.L Gluekler 408-365-6515 (Meeting Organizer and Chairman) (408-365-6564) GE Nuclear Energy 6835 Via Del Oro San Jose, CA 95119 IAEA SPECIALISTS' MEETING ON SEISMIC ISOIATION TECHNOLOGY San Jose, California

LIST OF PARTICIPANTS \

Names of Participants Telephone (Fax) Sign-in (Registration)

USA Prof. J. M. Kelly 510-231 -9480 University of California at Berkeley (510-231 -9471) Earthquake Engineering Research Center Richmond, CA 94804

Mr. R.W. Seidensticker 708-252-4492

I Argonne National Laboratory (708-252-4978) 9700 South Cass Avenue Argonne, IL 60439

Mr. R.F. Kulak 708-252-4681 Argonne National Laboratory (703-252-4978) 9700 South Cass Avenue Argonne, IL 60439

Mr. F.F. Tajirian 415-768-1010 Bechtel National, Inc., (415-768-3588) 50 Beale Street San Francisco, CA 94119-3965 IAEA SPECIALISTS' MEETING ON SEISMIC ISOIATION TECHNOLOGY San Jose, California

LIST OF PARTICIPANTS

Names of Participants Telephone (Fax) Sign-in (Registration)

USA Mr. R. Jetter 818-586-5282 Energy Technology Engineering Center (818-586-5118) Rockwell International Corporation P.O. Box 1449 Canoga Park, CA 91304

Mr. E. Rodwell 415-855-2767 Electric Power Research Institute (415-855-1026) P.O. Box 10412 Canoga Park, CA 91304

Mr. M. Patel 408-365-6421 GE Nuclear Energy (408-365-6564) 6835 Via Del Oro San Jose, CA 95119 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

Session 1: March 18, SEISMIC ISOLATION TECHNOLOGY

Session Chairmen: J. W. Herczeg, US DOE J. K. Biswas, AECL, Canada

The following is a summary of the "Seismic Isolation Technology Overview" session of the IAEA sponsored "Specialists' Meeting on Seismic Isolation" March 18-20, 1992. The first session focused primarily on the main features, activities and status of Seismic Isolation. Represented countries included, France, Italy, Japan and the U.S. The Chinese delegation submitted a summary paper, but could not attend.

Summary of Main Features of Base Isolation in France Ch. Coladant

A review of base isolation in France focused on three types; (1) laminated elastomer bearings, (2) sliding bearings, and (3) viscous (hysteretic) dampers. Manufactures included the ERA Company, FREYSSINET and CIPEC specializing in elastomer bearings, EDF-SPIE- Batignolles for sliding bearings, and the GERB company for viscous dampers. Each company specializes in a specific application area which are listed as follows:

ERA Co-"GAPEC System" - Rad-Waste, Power Circuit and Multi-Story Building, dwellings, Hysteretic Dampers FREYSSINET and CIPEC - "CRUAS" - Nuclear Power Plants "LaHague" - Storage Pools 7000 - 17,000t - for Reprocessing Plants EDF-SPIE Batignollos - Sliding bearing; Koeberg NPP(S.A.) GERB Co. - Viscous Dampers

A summary of recommended rules for design of seismic bearing was presented. These rule include:

- General Requirements: Access for inspection, isolation joints between adjacent structures.

- Response Analysis: Aging characteristics, maximum stiffness for acceleration, minimum stiffness for displacement.

- Safety checks for laminated bearing - limitation of horizontal shear strain.

-IS- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary Session 1: (Continued)

Qualification and Tests - Include dynamic tests at design frequency and shear strain.

Specific questions addressed the use of sliding bearings in France and aging. Currently no sliding bearings are being used in France. Aging of laminated isolation bearing increased the shear modulus. A ten week test at 70*C yielded a maximum of 50% change in G modulus; sometimes damping increased, and sometimes it decreased.

Status Report on Activities on Seismic Isolation in Italy A. Martelli

A status report of seismic isolation in Italy was provided by A. Martelli. The first use of seismic isolation was for the SIP buildings in Ancona. In a number of new buildings and to retrofit an old church, seismic isolation using isolation bearing has been used or planned. Interests are expressed also for the chemical industry and the nuclear field.

A working group called GLIS with representation of different experts is working on seismic isolation and dissipation system. A specialists' meeting was held in 1990 in Italy and International Meeting in 1991. Design guidelines for isolation systems are under preparation. Research and development including tests on rubber bearings, numerical analysis using finite element methods, development of methodologies for seismic qualification and tests on actual buildings are underway. Work on improving bearing characteristics using modified compounds and changed geometry and attachment systems is proceeding. In situ snap back testing and shake table testing has provided valuable data. International collaboration with U.S. DOE, GE, Japan, France, the Russian Federation and several European partners is continuing.

A Broad Review of the Status of Seismic Isolation Study in Japan S. Aoyagi

A broad review of seismic isolation in Japan was presented by S. Aoyagi. Programs for the development of seismic isolation are managed by a number of organizations such as CRIEPI, JAPAC, NUPEC and PNC. The leading construction companies played an important role in the development of the technology. Laminated rubber bearings, lead rubber bearings, high damping rubber bearings and sliding bearings have been used in actual buildings. Over a ten year period, 63 base isolated have been constructed. Monitoring of responses of isolated structures to actual earthquakes has been undertaken to collect data. Recommendation for Design of Base Isolated Buildings was published in 1989 by the Architectural Institute of Japan.

-19- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

Session 1: (Continued)

For application of seismic isolation to Fast Breeder reactors (FBR), testing and analysis of bearings are underway. Tests include verification tests, including aging tests and failure test with small scale bearing models. Numerical analysis of bearings using the finite element programs MARC and ABAQUS have been carried out. At first, a scale model of the FBR plant with isolation bearings was tested on a shake table. Secondly, models with a high center of gravity and low center of gravity were tested on the shake table. Numerical analyses to simulate the test results were performed. Several studies to estimate earthquake loads on seismically isolated structures are in progress. Isolation is also considered for LWR's in Japan. NUPEC will perform dynamic tests on seismically isolated computer systems. MITSUBISHI has studied spent fuel racks using isolation systems. For vertical isolation, a new system using air springs is being considered.

It is hoped that research and development will provide increasing confidence in using seismic isolation for nuclear facilities.

Status of Seismic Isolation Technology in the US J. M. Kelly

Conclusions of the US status on Seismic Isolation are as follows:

Damping mechanisms, especially if hysteretic and non-linear, produce high accelerations in the higher modes. The isolation process can occur in the absence of damping. Damping is needed only to suppress resonance at the isolation frequency. The apparent beneficial nature of damping has been formed since most dynamic analyses were conducted on single degree of freedom systems for which increased damping reduces accelerations and displacements. In large structures with many higher modes of damping, especially non-linear, hysteretic damping will lead to increased accelerations due to the increased response of the higher modes. Control of displacement using the hardening effect of the elastomer produced by crystallization at high strains is a more effective and reliable approach and will lead to more economic designs of isolation and structure.

-20- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

Session 2 and 3^ March 16-19,1992 SEISMIC ISOLATION ELEMENT AND SYSTEM PERFORMANCE

Session Chairmen: A.Martelli, ENEA, Italy J. M. Kelly, University of California, Berkeley, USA

Sessions 2 and 3 were held in the afternoon of March 18 and the morning of March 19. They were chaired by Dr. A. Martelli of ENEA, Italy, and Prof. J.M. Kelly of the University of California at Berkeley, USA.

Four papers were presented in these sessions, by Japan (1), Italy (1) and the USA(2). They provided the state-of-the-art and results obtained to date of the experimental work carried out in these countries, to support seismic isolation development, in particular for nuclear applications, and more generally for non-nuclear uses also.

The papers showed that experimental evaluation methods (tests of rubber specimens, individual bearings, isolated structure mock-ups and actual isolated buildings) are similar in Japan, Italy and the USA, and that complementary and consistent results have been obtained to date. More precisely, the Japanese paper entitled "Recent Results of Seismic Isolation Elements, Vibration Tests and Observations," presented by S. Aoyagi of CRIEPI, described tests performed for three types of bearings (natural rubber, lead plug and high damping rubber bearings). These concerned both single isolators in large and small scale, and shake table experiments of isolated structures, together with measurement of seismic response on isolated buildings.

The Italian paper entitled "Experiments on Seismic Isolation in Italy," presented by G. Bonacina of ISMES, dealt with tests performed in this country by organizations participating in the development of seismic isolation. The effort included the testing of high damping rubber specimens and bearings, as well as full and 1/4 scale isolated structure mock-ups with rigid masses, and two actual isolated buildings using such bearings (SIP buildings at Ancona, a three-story isolated house at Squillace, together with an adjacent, conventional equal house).

-21- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

Session 2 and 3: (Continued)

The US papers entitled "Qualification of High Damping Isolation Bearings for the ALMR" and "Summary of Experimental Tests of Elastometric Seismic Isolation Bearings for Use in Nuclear Power Plants," presented by F. F. Tajirian of Bechtel and R. W. Seidensticker of ANL, described tests performed on high damping rubber bearings and isolated structure mock-ups at EERC and ETEC, together with measurements of seismic responses of the Sendai building in Japan, in support of nuclear reactor projects like PRISM and NPR and in the framework of a cooperation between USA and Japan. Both high and medium shape factor bearings were tested, the latter with both high and low shear modules.

The papers showed that most topics of interest for the experimental demonstration of seismic isolation performance have been addressed: accurate data were obtained for the variation of horizontal stiffness and damping with displacement, failure modes, creep effects, dynamic excitation effects, etc. In particular, large safety margins were found for design displacement with respect to failure, for dowelled and especially, bolted bearings. Natural and accelerated aging tests have been performed.

The single bearing test data were found consistent with the results of tests on isolated structures. All measurements provided excellent information for computer code validation and improvement of design guidelines. Comparison among the different bearing types considered in Japan showed similar performances.

The importance of measurements and tests of actual buildings was stressed, to also provide a very clear demonstration of the adequacy of seismic isolation in particular with regard to improvements of the seismic safety of structures.

-22- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

Session 4: March 19,1992 DESIGN CRITERIA, STANDARDS, METHODS

Session Chairmen: Ch. Coladant, EdF, France Ralph Seidensticker, ANL, USA

1. Some Difference of Concepts Between Design Guideline for FBR Base Isolation System and a Seismic Design H. Shibata

This Paper presented modifications needed in Japan to its existing Design Criteria for LWR plants to allow use of base seismic isolation.

Professor Shibata listed several areas needing attention:

1. Need for acceptable standards for the isolation system. 2. Need to consider other events such as fire, flood, tsunami, strong wind, and other factors. 3. External events, aircraft crash, etc. 4. Internal missiles. 5. Protection from sodium and other chemical reaction products. 6. Environmental conditions. 7. Reliability. 8. Accessible for inspection and removal and test. 9. Lightening protection.

Prof. Shibata presented possible "zipper-like" failures (i.e., cascading failure of bearing resulting in torsional effects, etc). He simulated the phenomenon using Monte Carlo methods and indicated that results will be published later. Other investigators have not been able to substantiate this effect experimentally for the systems of interest. The idea of including backup systems to enhance the system reliability did not appear attractive. Simple systems are generally preferred.

The estimation of the probability of larger than S2 earthquakes and the definition of hazard curves are under discussion in Japan. There are currently differences in interpretation.

-23- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY Summary

>||slon4: {Continued)

2. Development of Guidelines on Seismic Isolation in Italy M. Olivieri

The initial set of guidelines was focused on high damping rubber laminated bearing. The guidelines are intended to promote the wide use of base isolation in Italy. The presentation addressed: 1. Ground Motion 2. Overall Seismic System Performance Requirements. 3. Design Requirements for Individual Isolation Devices. 4. Qualification of Isolation Systems Acceptance Testing Reliability Safety Margins Monitoring Considerable information was presented on design requirements, analytical requirements and qualification of systems, and individual devices. Seismic safety and monitoring were also addressed. The author said more work is planned to quantify some parameters and to extend the guidelines to other types of isolators. Considerable discussion took place on the need to shutdown the reactor for seismic events exceeding the OBE. There was general agreement that this requirement could be removed, since it is an operational safety question not related to seismic isolation.

The need for vertical and horizontal restraint was discussed. It seems that vertical "fail-safe" systems are required. Horizontal restraint system may be needed. A question was raised concerning the relationship between proposed guidelines and the Eurocode 8. At present Eurocode 8 does not address base isolation. Efforts will be made to collaborate as needed.

The issue of testing full-size bearings arose. Clearly, this is important, but limitations of test facilities are a problem. The monitoring of seismic isolation was discussed. In addition to the usual instrumentation, direct measurement of relative displacements during seismic events was emphasized.

-24- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY Summarv

ion.4: (Continued)

On the question relating to the number of time histories to be used in the analyses, the author replied that at least four time histories are required for both horizontal and vertical directions, unless one set of time histories completely enveloping the design response spectra can be used.

The coverage of sliding devices and low damping isolations concepts in the design guidelines was discussed. At this time the guidelines are limited to high damping laminated rubber bearings. In the future, if interest arises, other systems will be addressed. In fact, the Commission of European Communities has approved a plan to extend the guidelines to this and other areas. 3. Recent Results of Seismic Isolation Study in CRIEPI - Design Method H. Shiojiri This paper addressed the activity to develop guidelines and requirements for the use of seismic isolation in FBR plants. Present plans are to issue a draft of these requirements in 1993 and to make the final release in 1994. One of the goals is to achieve equivalent safety in seismicaiiy isolated plants and in conventional plants. The presentation included consideration of:

1. Seismic Isolation System Design 2. Design Basis Earthquake Ground Motion 3. Isolation Element Design 4. Building and Structure Design 5. Equipment and Piping System Design The paper emphasized load combinations and definitions of limit states to assist in isolator design and establishment of appropriate safety margins.

A question was raised as to whether it was intended to design the isolated structure as if it were not isolated. The reply was, no; the same margins will be provided for both isolated and non-isolated structures. Benefits are expected for seismic isolation.

-25- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY Summary

[Session 4: (Continued) |

4. Technical Specification for the Successful Fabrication of Laminated Seismic Isolation Bearings R. F. Kulak

This paper described considerations which must be addressed to achieve high quality laminated elastomeric bearings. The need to balance performance and prescriptive requirements was stressed. The paper also addressed the need to achieve a clear understanding between the designer and fabricator on terminology and technical requirements. This included the specification of a minimum ultimate shear strain.

Topics presented included the following: 1. Codes and Standards 2. Elastomeric Specimen Testing 3. Bond Strength 4. Fabrication Process 5. Quality Control 6. Use of a "Proof of Process" Assembly 7. Tests on Completed Bearing 8. Need to use zero axial load with horizontal shear tests

The discussion addressed the "proof of process" assembly, (i.e., dry assembly without bonding). This approach helps to qualify the process and to obtain a layer of vulcanized rubber for measurements (thickness, hardness, tensile strength, etc.).

Regarding the testing of the raw rubber material, the author provided a description of the current practice which includes sampling before batch processing, from the batch and also during the actual process.

A question was raised as to whether fatigue tests were required for acceptance testing. The reply was, no, they are not. Based on laboratory tests of bearings it was concluded that well bonded bearings will have more than adequate fatigue resistance.

-26- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY Summary

$i March 19,1992 SBSMIC ISOLATION SYSTEM DESIGN AND ANALYSIS

Session Chairmen: M. Olivieri, Ansaldo, Italy S. Aoyagi, CRIEPI, Japan

Five paper were presented in this session by Japan (2), Canada (1), USA (1) and Italy (1). They provided the state-of-the-art on the numerical methods available and under development to compute the dynamic behavior of isolation bearings and the seismic response of isolated nuclear power plants and other non-nuclear buildings. In general, both simple and detailed models to characterize the behavior of single isolators and of isolated structures were addressed. Simplified models were generally found adequate in reproducing the behavior of isolated systems within the design conditions.

Rather refined models were generally found necessary to model the isolators. Various types of models were presented, some of them able to reproduce the typical stiffening effects occurring at very large displacement/strains induced by beyond SSE excitations. Both, general purpose codes and special purpose codes were used. Bilinear, trilinear and other special models were presented, including a possible remeshing scheme for very large distortions.

Considerations were also given to the design margin evaluations including the characteristics of the design ground motions for extreme cases. Comparisons of numerical results with these available from several types of experiments tests were presented (static and dynamic tests of simple isotopes, shake table test of structure mockups, snap back tests of full scale isolated structures). Satisfactory agreement was generally found, with some limited exceptions. Future plans to improve the agreement and to validate numerical bearing models, were described. Some additional details on the individual presentations are as follows:

1. Study of Seismic Responses of Candu 3 Reactor Building Using Isolation Building J. K. Biswas Mr. Biswas introduced the history of CANDU, the Canadian Seismic Zoning Map, isolation experiences in Canada, and the CANDU 3 Nuclear Power Plant. CANDU 3 is a nuclear power plant with 450 MW electrical output. During the conceptual development of the CANDU 3, various design options including the use of seismic isolation have been considered.

-27- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

[Session SMContinued) 1

Mr. Biswas described the initial selection of an isolator for CANDU 3, it's characteristics, as well as the reactor building model and analysis methodology. Response displacements and floor response spectrum were reported.

There was a comment to point out the importance of the maintenance of bearings in the severe (cold) environment. In addition, it was asked whether the Guillotine pipe break effect was considered as a possible consequence of differential motions of the isolated and non-isolated part of the structure.

2. 3D Seismic Response of a Base Isolated Reactor S. Kitamura

Mr. Kitamura from PNC introduced a seismic isolation study conducted at PNC. He described the scope of parametric studies, including the evaluation of generic restoring force characteristics. A two-lumped mass model and four different restoring models were shown in detail.

Mr. Kitamura described the verification tests, i.e., shake table tests. Three kinds of isolators, (elastomers with steel dampers, elastomers with lead core, and high damping elastomer) were considered in these tests. The presentation also addressed evaluation of the response of a 1000 MW class FBR plant. There was a question on how to compare the numerical result and experimental result of simulation tests.

3. Seismic Analysis for the ALMR F. F. Tajirian

A non-linear analysis of the ALMR (PRISM) reactor was presented, focused on evaluating the effects of the high stiffness typical of High Damping Laminated Rubber Bearings (HDLRB) at small and very large displacements, on the calculated responses.

Synthetic acceleration time histories, compatible with R.G. 1.60 response spectra were used scaled at 0.3, 0.75 and 0.075g (low level), and to large earthquakes of 0.75g and 2g. Equivalent linear models were generally found adequate for isolated structures within the seismic design range.

Another objective was to quantify margin beyond SSE with models treating the stiffening of rubber at large displacements.

R. G. 1.60 spectra were found comparable with the tentative Japanese spectra for long period motions; EPRI spectra would be lower (higher in the high frequency region).

-28- IAEA SPECIALISTS1 MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

Ulllon 5: (Continued)

It was noted that long period motions are more predictable and with lower extreme values than high frequency inputs.

For the evaluation of beyond design basis conditions, the characteristics California site and magnitude, were considered assuming a point source at 9 km distance. Results obtained with site specific excitations were compared with R. G. 1.60 scaled to 2.0 g zero period acceleration and were found conservative.

4. Numerical Activities On Seismic Isolation In Italy F. Bettinali

Simplified bearing models included equivalent damping models and hysteretic models. Detailed bearing models were implemented in ABAQUS and are being refined by use of an elastic plastic model for steel and a hyperelastic model for rubber. Models of isolated structures are based on simple bearing models. Both one degree-of-freedom systems and finite elements models were developed for the superstructures.

Comparison with test data demonstrated the applicability of single bearing test results for the prediction of the response of an isolated structures.

5. Recent Results Of Seismic Isolation Study In CRIEPI H. Shiojiri

Different models were considered. Comparison with test results were presented for different level of excitation, (Sj) up to 2 g's. A model with multiple pad modeling was presented to evaluate rocking effects.

Three-dimensional models of isolators were presented:

Numerical analysis of rubber bearings with ABAQUS and MARC predictions were found to compare well.

The effect of meshing was studied and found not to critical.

A one layer model can be sufficient to provide acceptable horizontal results, but several layers are needed for predicting the vertical response.

Constitutive equation were studied and improved to take into account the compressibility for calculation of vertical response. For very big distortions of the finite element mesh, remeshing seems necessary.

-29- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

ISesslon £ (Continued)

For the modeling of buildings, floor flexibility can be important, but this applies mainly to conventional buildings.

The activities conducted so far indicated that adequate results can be obtained with available computer codes, when good parameters for the constitutive equations are used.

-30- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY Summary

March 19,1992 MEETING SUMMARY DISCUSSION AND RECOMMENDATIONS

Session Chairmen: E. Gluekler, GE, USA S. Aoyagi, CRIEPI, Japan

The Specialist's Meeting concluded with a summary discussion of key aspects of seismic isolation and expected future developments. The meeting participants were in agreement that the meeting was useful and timely and contributed significantly to advancing the understanding of the programs in progress. The full range of technical issues was openly discussed. In this summary, the topics addressed in the summary are presented along with recommendations for follow-on actions.

In general, seismic isolation for nuclear structures is not different from seismic isolation that was developed and qualified for other industrial and commercial applications. However, the characterization of safety margins of seismic isolation systems, as well as the understanding of failure modes as input to residual risk determinations are of additional importance in reactor applications. For seismically isolated nuclear power plants, the safety characteristics will still be dominated by nuclear reactor safety considerations. In general, it is accepted that seismic isolation systems can be made with large margins beyond the design basis earthquakes such that failures would happen in the superstructure before elastomer bearings fail.

There is general consensus in the various countries on the requirements for qualification and verification of seismic system responses. They include the characterization of the long term reliability of the system, the validity range of the system and failure modes, and establishment of appropriate analysis rules. In addition, the validation of analytical models for evaluation of physical phenomena and dynamic responses with bearing and system tests including shake table tests is required. Maximum and minimum values need to be established for design parameters. Acceptance tests will be required for seismic isolation bearings before their installation in facilities.

Several types of seismic isolation systems could be considered. The general trend is towards the use of simple, predictable, low or no maintenance systems. It was recognized that sliding systems can always carry the vertical load and perform an isolation function, even if sliding on a concrete surface would be considered. For nuclear structures sliding systems are not favored. The key issue is that accelerations are governed by friction and high frequency energy content can be transmitted into the nuclear structure. Certainly, sliding surfaces could be designed and fabricated - potentially economically - with low friction coefficients, but uncertainties in response

-31- IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY

Summary

I MEETING SUMMARY DISCUSSION AND RECOMMENDATIONS (Continued) | predictions including the maximum expected sliding distance, may be a problem. In the US, it would be expected that designers have to consider displacement restoring devices to obtain acceptance of sliding systems by regulatory agencies.

It was pointed out by CRIEPI that lead rubber bearings have approximately the same structural and safety performance as high damping rubber bearings. Some meeting participants questioned the need for the parallel development of two systems. However, this is more an economics than a technical issue. The merit of the lead rubber bearing is that damping can be adjusted by changing the diameter of the lead plug. CRIEPI sees some advantages of the lead rubber bearing in terms of simplified analysis, and less materials development and characterization. Natural rubber bearing could also be used in combination with separate damping elements. There was no consensus that the separation of isolation and damping functions would provide a better design than achievable with simpler systems - such as the high damping rubber bearings system - that combine these functions.

There was no agreement on the failure characteristics of lead plug bearings; in particular it was not generally accepted that the lead plug has no effect on the failure. Although all bearing types available may qualify for general commercial/industrial application, the questions that should be asked is: What is best for nuclear application?

The development of seismic isolation design guidelines has made significant progress. For the specification of meaningful safety criteria some statistical testing will be required. Common data bases and testing approaches were seen as a benefit if implemented. The seismic design guidelines development has proceeded in Italy, Japan, and the USA with some collaboration between Italy and the USA, and Japan and Italy. Since the design objectives for seismic isolation of nuclear reactor are similar in all countries, if would be fruitful to compare requirements developed in the framework of existing regulations in each country and to move towards a consolidation and harmonization in key areas; foremost in the area of safety design. This could also be the topic of a follow-up IAEA meeting to be considered possibly for 1996.

-32- PAPERS

33-34 XA0055377

TENTATIVE PROVISIONS FOR SEISMIC DESIGN CF BASE ISOLATED BUILDINGS IN FRANCE

Jacques BETBEDER-MATIBET (EDF/SEPTEN) Philippe BISCH (SECHAUD et METZ) Francoise GANTENBEIN (C.E.A.-C.E./SACLAY - DMT/SEMT/EMSI)

Abstract

This paper presents Chapter 22 of the AFPS (French Earthquake Engineering Association) Recommendations, devoted to aseismic bearing pads. This chapter gives general rules applicable to any kind of pads, and specific rules for hooped elastomer pads. It contains specifications for general or detailed arrangements when aseismic bearing pads are used, calculation procedures (in particular, a simplified method), and a technical validation procedure for new designs of pads.

1 - INTRODUCTION

The aseismic regulations are presently being modified in France both from the point of view of technical procedures and that of the legal framework. At the technical level, to which this paper is limited, the complete revision of the procedures presently in force (1969 Aseismic Regulations revised in 1982) has been underway for several years and is now being finalized by the issuing and the completion of new texts.

-35- This revision is judged to be indispensable for taking into account the growth in knowledge and to correct insufficiencies in the 69-82 Aseismic Regulations. This has been done by inserting new chapters, in particular, on soils and foundations, industrial installations and special pads.

The new texts resulting from this work are being printed in the form of Recommendations of the French Earthquake Engineering Association, intended to serve as a basis for future regulations. The first nine chapters, which describe the field of application, and give general design and calculation procedures and procedures for soils and foundations, were published in 1990 in the first volume of these Recommendations (Ref. 1). A second volume is in press; it contains, in particular, chapter 22 on special pads, the subject of this paper. The inclusion of this chapter shows the interest that there has been for several years in France in this technique of aseismic isolation, as the first French work in this field goes back to 1977 (Ref. 2-9).

With respect to the definition of seismic load, the AFPS Recommendations have retained the option of the ground elastic spectra. These differ depending on the nature of the terrains,

which are divided into four categories, SQ (rock), S1 (good

resistance), S2 (medium resistance) and S3 (low resistance). To take the seismic load into account the behavior coefficient q

is used. This coefficient is comparable to the Rw factor in the American Codes UBC 1988 and SEAOC 87-88, to the structural type factor S of the 1984 New Zealand Standard NZ 4203 and to the factor q of the Eurocode project Ne 8. The AFPS Recommendations have, therefore, as in other modern aseismic codes, specified the accepted level of inelastic deformation as a function of the type of material and the lateral resistance of the structure.

2 - GENERAL FORM OF THE DOCUMENT

The present paper is a report of the work of the group which has the responsibility of Chapter 22, on aseismic pads, of the AFPS Recommendations.

-36- The resulting text was established starting with a project of Mr. DESPEYROUX for Eurocode 8, amended by the expertise of various participants in the group, in particular, Messieurs BISCH, COLADANT, DELFOSSE, DOURY, GUERAUD and SOULOUMIAC, then debated and revised during discussions that took place in the AFPS Scientific and Technical Committee (STC). Thus, after completion, and approval by the STC, Chapter 22 has been included in volume II of the AFPS Recommendations.

Although the text was established in the framework of aseismic isolation of buildings (standard or nonstandard), the majority of the recommended procedures are fairly directly adaptable to the case of other structures.

The chapter, in principle, is aimed at all types of pads designed to ensure some extent of aseismic isolation. General procedures have therefore been included that make it possible, for all types of pads, known or still to be designed, to obtain behavior in conformity with the defined aseismic objective, while retaining approximately uniform safety. Such an undertaking is not easy, since it is always difficult to derive the fundamental procedures that can be applied to the systems yet to be invented from existing systems.

It is found that the aseismic pads most often now employed (Ref. 10 - 15), particularly in France, are the hooped elastomer pads, similar to those used for supporting the decks of bridges. Thus, for pragmatic reasons, procedures belonging to this type of pads have been inserted in the text. It is understood that other types of pads require their own procedures, which are not found in the text, as it is impossible to be exhaustive. Under these conditions, it was feeled best to define a qualification procedure, which can lead to contractual acceptance. This course of action makes it possible to leave open all innovation possibilities.

-37- The chapter 22 consists of three parts, given in detail in the following sections : a) The conditions of use and the general procedures ; b) The calculation procedures and verifications ; c) The validation procedures.

3 - REVIEW OF THE OPERATING PRINCIPLE OF PADS

The pad systems used in France up to now are based on the principle of flexibility with respect to the horizontal forces acting on the structure, so as to :

- increase the period of the fundamental mode to obtain a reduced acceleration response,

- make the higher mode responses insignificant by concentrating practically all of the mass in the mass of the fundamental mode.

Let us start with a very simple model with two degrees of freedom (Fig. 1) where : k is the stiffness of the pad, m the mass of the first floor, M the mass of the superstructure K the stiffness of the superstructure.

Figure 1

It can be easily shown that in practical cases (where M is considerably greater than m) the following approximations hold

-38- T,1 ,"4-d , ^ Ta

T? +~r r M 1 . ^J 2 1 I" 1'2 Tf

Xl ~ 1 + c

/U ^ l m I =1 2 ^1 " M (1 + c*')

T1 and T2 are the periods of the two modes, Ta the calculated period for a rigid block of mass m + M placed on the pads (T_ =

2TTV(M + m)/k), Tf the period calculated for the superstructure on a fixed base (T^ = 2T1YM/K), x^ the displacement of m in the fundamental mode (in setting the displacement of M equal to 1), JJ^ the modal mass of the fundamental divided by the total mass

m + M and *( the quotient Ta/Tj.

It can be seen that when this parameter e( is significantly greater than 1 (which is generally the case), that practically all of the mass is concentrated in the fundamental mode (JJ-^ very close to 1, in general greater than 0.99), which has a period

slightly greater than Ta, and that the deformation associated with this mode is produced, essentially, in the pads (x^ close to 1) .

-39- In general, the value for T^ is chosen fairly large, of the order of 1 to 2 s, so as to favor a reduced acceleration response; these values for the period are found in the 1/T region of the elastic spectra (Fig. 2). By setting the constant value of the pseudo-velocity which corresponds to this region by V, it can then be shown, with the same degree of approximation as in the preceding formulas that :

1 V T max 2 It Figure 2 X r\i— 1 max ~ 2TI V T

(VT7

being the maximum (as given by the quadratic

combination) of the displacement x of the base, XJnax the maximum of the displacement X of the superstructure and A the maximum relative displacement X - x between the superstructure and the base.

It can be seen that the response is determined by the two parameters Ta and c< = Ta/Tj ; the following table shows the variation of the response as a function of o<:

-40- 2TL x 271 X 2l\ A max max V T V T a V T a a 1 0.707 1.414 0.707 1.5 0.832 1.202 0.370 2 0.894 1.118 0.224 2.5 0.928 1.077 0.148 3 0.949 1.054 0.105 3.5 0.962 1.040 0.078 4 0.970 1.031 0.061 4.5 0.976 1.024 0.048 5 0.981 1.020 0.039

- Table 1 -

The character of the response and, in particular, the low value of the deformation A of the superstructure is retained as long as o( is large enough (greater than 2) . For the values of ft< approaching 1, A increases rapidly ; this observation will be used later (discussion of the behavior coefficient in section 6).

In the standard cases (c< definitely greater than 1), it can be seen that the objectives stated at the beginning of this section are attained, i.e., :

- a value of,the period of the fundamental mode (that is chosen on the basis of the parameter Ta) located in a slightly amplified region of the acceleration spectrum (typically 1 to 2 seconds).

- a very preponderant fundamental response mode and for which practically all of the deformation is concentrated in the pads.

These observations remain valid for real models of buildings (models with several degrees of freedom), even if some couplings (particularly between translation and rotational motion) can then occur and reduce the effectiveness of the isolation effect obtained by the pads.

-41- In addition, a viscous damper can be associated with the spring k where the viscosity constant can be controlled. As the motion of the superstructure is practically rigid, the essential part of the energy absorbed by the damper is linked to the motion of the spring. Nevertheless, as the damping is increased, the force increases in the damper. If the total force on the system, spring plus damper, in the case of shock is considered, it is at a minimum for a critical percentage of damping of the order of 25 % and increases progressively when the percentage is increased beyond this. It is not of interest, therefore, to have too much damping and in practice it is limited to values of the order of 30 %.

An ensemble made up of springs, that are sufficiently flexible to impose the extensive participation of a first mode with a long period, and dampers, is thus capable of acting as a partial isolator with respect to the seismic stress.

The aseismic pads most used in France at the present time can be grouped into two categories:

a) Hooped elastomer pads with elastomer compositions that allow large distortions under dynamic loading. b) Springs, which give well controlled stiffness in all directions.

There are, nevertheless, other types of pads: pendulum systems, rollers, balls, etc..

The natural damping of the elastomer pads is of the order of 7 %. That of the springs is very low. These can be associated with :

a) controlled dry friction systems (Coulomb friction), obtained by the contact of two plates. Note that if the coefficient of friction is low, as in the case of teflon plates, the resisting force may be insufficient and the displacements too large.

-42- b) viscous or equivalent damping systems.

4 - GENERAL ARRANGEMENTS AND PROCEDURES

4.1. These arrangements aim at ensuring the correct operation of the ensemble of the pads by avoiding :

- The creeping of the pad units, except in the case of controlled sliding plates, when these exist. (The elastomer pads must remain compressed).

- That, in the case of pads including a sliding system, the movable component can escape from the plate : for this a border around this plate equal to 1.2 times the maximum calculated amount of sliding is foreseen and it is recommended that the concrete is at the same level as the plate.

- That the ensemble of the block consisting of the superstructure above the pads does not impart a shock to neighboring infrastructures and superstructures. To increase the margin of safety, the opposite phase motions of two blocks and the periods close to each other have to be taken into account.

UKA <^J>3<«"

- Figure 3 -

-43- 10 - That the fluid lines connecting the block on the pads to the exterior can be damaged due to displacements of the block : they must be designed to resist the calculated displacements overestimated by 20 %.

4.2. In addition, the pads are positioned at various points between the infrastructure and the superstructure. If the infrastructure or the superstructure is too flexible locally, the pad system can be stressed non-homogeneously, changing the dynamic response. It would be possible to take account of these phenomena by an extremely accurate calculation of the structure (finite elements). Nevertheless, neither current practice (for ordinary buildings) nor the standard methods given in the Recommendations can take such phenomena into account.

floor slab It is, therefore, necessary to control the deformations of the

column or aseismic pads by eliminating the local supporting bearing pad deformations that are not at all, wall or are poorly, modeled. To do this two slabs that are sufficiently mat or beam rigid are positioned as near as possible on both sides of the pads. In the case where one of the rigid - Figure 4 - slabs cannot be placed in the immediate proximity of the pad (the case of basements), the deformation of the vertical components between this slab and the pad is limited to l/20th of the deformation of the pad.

In addition, measures are taken to avoid the buckling of the rigid planes under the effect of the horizontal stresses : - minimum thickness of the floor slabs - limiting the inclination of the beams.

4.3. In some cases it has been considered, to solve the possible disconfort problems with respect to the wind, to use "fusible" systems, i.e., systems which transmit the horizontal force up to a certain value, then release the displacements beyond this. In

-44- 11 practice, the sequence of the releasing of the forces cannot be controlled, and unforeseeable rotational movements of the superstructure can result. This is why these measures are prohibited. On the other hand, units with progressive stiffness without any discontinuity are authorized.

4.4. Finally, when the pads are liable to ageing, their inspection and their possible replacement must be prescribed.

5 - DYNAMIC RESPONSE

The dynamic response of the system has to be studied following the methods stated in the section basic to all the Recommendations (spectral modal method or chronological analysis). When the behavior of the system cannot be represented linearly, which is, in particular, the case when sliding plates are used, a nonlinear chronological analysis must be used.

In taking account of the flexibility of the pads with respect to the horizontal stresses, special attention should be paid to the torsion movements. These are naturally due to the eccentricity of the horizontal projection of the center of gravity with respect to the center of stiffness of the pads. In addition, it is necessary to take account of the contractual eccentricity of the masses, defined elsewhere.

The values of the dynamic properties (stiffness, damping, friction...) to be introduced in the calculation are those which give the most conservative results ; for example, these are the maximum values of the stiffness when the stresses are sought, but the minimum values when the displacements are calculated. In the standard cases, the use of average values is accepted.

In the case where it can be taken that the first mode of the superstructure is comparable to a pure horizontal translation, a simplified method of calculation can be used: in this case, the acceleration is uniform over all the height and equals where T^ is the period of the first mode. This horizontal

-45- 12 acceleration produces, outside of translational motion, a torsion movement which must be taken into account in the calculations for the pads and the displacements.

In order that such a calculation can be used, the underlying hypotheses must be effectively verified:

a) the building must have a sufficiently uniform lateral resistance and the vertical loads must be transmitted directly from the columns and the supporting walls to the pads, to avoid deformations which invalidate the hypothesis that the superstructure cannot be deformed.

b) The vertical and the rotation rocking stiffness of the pads and the underlying foundation are high enough to prevent the horizontal acceleration of the masses from affecting the rocking;

thus, for the soils that are too flexible (S2 and S3) the simplified method has been ruled out and conditions for the relative stiffness between the horizontal and vertical translations imposed.

c) In order to have a pure translation, the superstructure must not have too much deformation on bending, when subjected to horizontal acceleration. This is shown by the fact that the period of the superstructure fixed at its base is short with

respect to Tlfi.e., that the parameter 0( = Ta/Tf introduced in section 3 must be sufficiently large, in practice, greater than 3.

6 - BEHAVIOR COEFFICIENT

In numerous past constructions, in the nuclear industry, obviously, but also for ordinary buildings, the structures have been calculated in the elastic domain, i.e., with an elastic response spectrum and a behavior coefficient equal to 1. Taking into account the good understanding that it is possible to have of the dynamic behavior of a structure placed on aseismic pads, this elastic approach has great advantages with respect to safety and allows minimizing of the repairs after an earthquake. It also

-46- 13 makes it possible to minimize the effects of the earthquake on the installations and equipment inside the building, which can be an essential safety element, or on a strictly economic level when critical installations are involved : nuclear facilities, hospitals, communications centers, etc.

On the other hand, when ordinary buildings are concerned, it is not possible to hope that the cost of the infrastructure and the pads will be compensated by a rebate on the installation, and the net advantage in terms of safety results in a surcharge that the prime contractors are not ready to accept in the vast majority of cases. It is thus a question of putting the structures on aseismic pads under safety conditions equivalent to those of the same building without pads.

This is why it has been agreed to calculate the dimensions of the superstructures using a behavior coefficient, as in the case of a building without pads. However, as indicated at the end of section 3, in the comments on Figure 1, it has appeared prudent to limit the value of the behavior coefficient for the flexible superstructures (frames), i.e., those where the effects of plasticity, that are implicitly accepted by the use of the behavior coefficient, may decrease the ratio o( = Ta/Tf in the zone (<>( between 1 and 2) where the deformation of the superstructure increases rapidly. The choice of the condition q < 2/3

Nevertheless, in the ordinary cases, the essential part of the deformation occurs in the pads. It is desired that these work in the elastic range so that they continue to operate properly during an earthquake; the dynamic response of the overall system is then essentially that of an elastic oscillator, except in the nonlinear cases mentioned above. The response spectrum to be used

-47- 14 for the motion of the ensemble is, thus, the elastic spectrum. The last calculation serves to determine the displacements of the overall structure and the stresses in the pads and the infrastructure.

This necessity of using a twofold calculation does not result in a very significant increase in the work required, particularly in the simplified method. It is compensated by the gains in material, which can be substantial in the superstructure.

7 - VERIFICATIONS

The verifications are carried out at the Ultimate Limiting State, in an accident situation, the resistance checks are made with the normal values of Ym (Jrm is taken as 1.5 for the elastomers).

An important verification is that of the buckling resistance, which is expressed by the equation:

*s pu < pc

where Pc is the critical load, obtained either by a representative calculation or by tests.

For the hooped elastomer pads, Vs is taken equal to 3, and Pc can be evaluated from the equation:

A Pc = 4 GAS id h where A is the nominal area of the pad G is the dynamic shearing modulus d is the sum of the hoop thickness and the elastomer thickness for one layer h is the total height of the pad S is a shape factor

-48- 15

* In the case of circular pads:

S = D 4e

* In the case of rectangular pads:

S = ab 2e (a + b) where:

D, a and b are, respectively, the diameter and the sides of the hoops e is the thickness of an elastomer layer.

This theoretical equation gives values in agreement with experimental results.

For the hooped elastomer pads, a special verification is made of the distortion : the tests show that the ultimate distortion depends on the vertical load, and more specifically on p =

Pu/Pc. A safety range has been determined (

-49- 16

UISTORTION a = 0,9

a diameter h elastomer total V» thickness

0.1S

Variation range for

1 ra (no WO 90 ($90 -k

- Figure 5 - - Figure 6 - Experimental results for ultimate Curve of the recommendations distortion (DELFOSSE)

Obviously, analogous work has to be carried out for other types of supports. For example, curves determined by GERB for its spiral springs are given below.

-50- 17

D average coil diaaeter d wire diaaeter i number of active tuz-ns G torsion »odulus of the material L - Lo - fp faeigr-.t of the spring under the load P fp vertical deflection under the vertical load P fq transverse displacement under the action of the transversal force Q cp • P/fP vertical stiffness cq > Q/Fq horizsntal stiffness Re • cq/cp stiffness ratio

- Figure 7 - Stiffness and buckling safety coefficient (GERB reports)

It should be noted that the structures on aseismic pads, having a fairly long period, can exhibit a response to the wind, and it is necessary to thoroughly verify their behavior under the effect of this loading. To avoid possible undesirable vibrations due to the wind, dampers positioned in parallel with the aseismic pads can be used.

8 - TECHNICAL VALIDATION FOR A SYSTEM OF PADS

As other designs of aseismic pads will be developed, a procedure should be specified for the technical validation of any new system. The chapter 22, therefore, specifies :

-51- 18 a) The desired objectives, i.e. :

- to prove the conformity of the system with aseismic requirements ; - to prove the long-term reliability (with respect to the resistance and the dynamic properties) ; - to determine the validity range of the system and its failure modes ; - to define the rules of analysis which are representative of the behavior. b) The methods required : - models which describe the physical phenomena and allow us to reproduce the behavior ; - tests for validating the behavior during an earthquake ; - qualification tests.

From the ensemble of these procedures a validation file has to be be established. This will also include a guide for the users (construction details, calculation method...)

This purely technical procedure can possibly be used as a basis for an administrative or contractual validation procedure, which is outside the scope of the chapter (for example, the CSTB Technical Note procedure).

9 - TESTS

There are two types of tests required : a) the qualification tests, required for the systems with a new design, or when the changes of material or dimensions of existing systems do not allow extrapolation of the characteristics from the previously determined values. The qualification tests are used, in particular, to establish the minimum and maximum values of the design parameters. They must,

-52- 19 therefore, be carried out under conditions close to those for the use of the pads (in particular, with respect to the vertical loads and frequencies).

Annotations to the procedures are given for the statistical treatment of the results. b) The acceptance tests for checking the characteristics of the pads before their installation.

- Figure 8 - Shearing/distortion cycles at 1 Hz of an elastomer pad (from EDF/SEPTEN)

-53- 20

10 - CONCLUSIONS -

Regulations (Ref. 16 - 18) on earthquake resistant design of buildings have so far been mostly based on experience.

It may therefore look premature to make an attempt towards regulating base isolated structures for which experience from actual earthquakes is currently very limited.

The discussions within the AFPS working group have shown that the key issues for a safe aseismic design of such structures were reasonably well identified, thus enabling the group to draft recommendations which have been presented in this paper. It is expected that these recommendations, even before their transformation into an official French Standard, will provide useful guidance for designers.

REFERENCES - [1] Recommendations AFPS 90 - Presses des Ponts-et-Chaussees, 1990.

[2] Genie parasismique : V. DAVIDOVICCI et al - Presses des Ponts-et-Chaussees, 1985.

[3] Seismic isolation from idea to reality - Earthquake Spectra (special issue) Vol. 6 - n* 2, May 1990.

[4] Dynamic behavior of nuclear power plant founded on sliding elastomer bearing pads : K. UCHIDA, K. EMORI, K. MIZUKOSHI, Y. TAKENAKA, J. BETBEDER-MATIBET, J.P. NOEL-LEROUX, P. UHRICH - Architecural Institute of Japan - 1985 - Papers 2255-2256.

[5] Aseismic foundation system for nuclear power stations : C. PLICHON, F.I. JOLIVET - Mech. E. Conference on engineering design for earthquake environments, 1978.

[6] Protection of Nuclear Power Plants Against Seism : C. PLICHON, et al. - Nuclear Technology, Vol. 49, 1980.

-54- 21

[7] Seismic foundation system for nuclear power stations : F. JOLIVET and M. RICHLI - Trans Fourth Conf. Structural Mechanics in Reactor Technology, San Francisco, Vol. K, n" 9/2, 1977.

[8] Wood framed individual houses on seismic isolators : G.C. DELFOSSE - Proc. Conf. Natural rubber for earthquake protection of buildings and vibration isolation - Kuala Lumpur, Malaysia, 1982, pp 104-111.

[9] Earthquake protection of a building containing radioactive waste by means of base isolation system : G.C. DELFOSSE and P.G. DELFOSSE - Proc. Eighth World Conference on Earthquake Engineering, San Francisco, California, Vol. 5, pp 1047-1054, 1984.

[10] Appareils d'appui en elastomere frette - SETRA - Bulletin Technique n" 4, 1974.

[11] Elastomere vulcanise, adherise ou non a des frettes (metalliques ou autres) - Projet de Methodes d'Essai N° 20 - Laboratoire Central des Ponts-et-Chaussees, 1988.

[12] Appareils d'appuis en elastomere frette utilises en genie parasismique : Cahier de Regies Techniques EDF 20 C.001.01, 1985.

[13] Seismic isolation using sliding-elastomer bearing pads, R. GUERAUD, J.P. NOEL-LEROUX, M. LIVOLANT, A.P. MICHALOPOULOS, Nuclear Engineering and Design 84, 1985, pp 363-377.

[14] Les fondations antisismiques de la centrale nucleaire de Cruas-Meysse : J.C. POSTOLLEC - Notes du Service Etude Genie Civil d'EDF-REAM, 1983.

[15] The Gapec System : A new highly effective aseismic system : G. DELFOSSE - Proc. Sixth World Conference on Earthquake Engineering, New Delhi, India, Vol. 3, pp 1135-1140, 1977.

-55- XA0055378 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, California, USA, March 18-20, 1992

STATUS REPORT ON ACTIVITIES ON SEISMIC ISOLATION IN ITALY A. Martelli (ENEA-RIN, Bologna, Italy) F. Bettinali (ENEL-CRIS, Milano, Italy)

ABSTRACT The development of seismic isolation and its application to structures other than bridges were started in Italy in 1988. Considerable efforts are being devoted to this technique, both because it can already be widely used in civil buildings (where it is particularly attractive for constructions that are critical for emergency and disaster planning), and due to the very promising perspectives for application to the industrial plants. In particular, ENEA is also quite interested in verifying the applicability of seismic isolation to the high risk plants, including the innovative nuclear reactors. The correct development of seismic isolation, for a future wide use in all the domains of interest - including high risk and other industrial plants - requires that a sufficient number of applications to civil buildings is undertaken, so as to improve the knowledge on the design and behaviour of isolated structures. It also requires seismic monitoring of isolated constructions. This is the reason why all the ongoing studies in Italy - including those of ENEA and ENEL - are based at present on applications to civil buildings. To the aforesaid aims, R&D work is also needed: such a work, together with the experience acquired on actual isolated buildings, is essential to set up adequate design rules. On the other hand, development of design rules must be carried out in parallel, in order to determine the features of the necessary research activities. Until now, our development work has been focussed on the high damping steel-laminated rubber bearings, which have been adopted for most isolated buildings in Italy. It consists of: [a] the set-up of proposals for design rules and guidelines; [b] experiments on bearing materials, individual bearings, isolated structure mock-ups, and actual isolated buildings; [c] development and validation of simplified and detailed numerical models of bearings and structures. Furthermore, support is being provided to the designers of isolated buildings and studies are being started for the development of new and optimized bearing materials, analysis of different bearing types, seismic monitoring, and applications to energy production plants. Finally, tests have been planned for interface piping. Due to the high complexity and considerable cost of development work on seismic isolation, national and international collaborations are essential to optimize such studies: national cooperations have been established in Italy among several organizations, universities and companies, forming the Working Group on Seismic Isolation (GLIS); international collaborations exist with the USA and Japan. This paper summarizes the main features of the above-mentioned activities. More details on the studies [a] to [c] have been reported by separate technical papers.

-56- 1. INTRODUCTION Considerable efforts are being devoted in Italy to seismic isolation development and application. The interest in this technique is rapidly increasing in our country, both because its maturity is already evident for a wide-ranging use in civil buildings, and due to very promising perspectives for industrial plants (Chang et al. [1]): the National Utility (ENEL) aims at evaluating the applicability and usefulness of seismic isolation to its energy production plants; the Italian Agency for the New Technologies, Energy and Ambient (ENEA) is also quite interested in verifying the possibility of using this technique to the high risk plants, including the innovative nuclear reactors (Martelli [2] and Martelli et al. [3]). For these applications, its advantages appear such as to warrant further work to resolve some of the outstanding technical problems. The correct development of seismic isolation, for a future wide ranging use in all the domains of interest - including in the high risk plants -, requires that a sufficient number of applications to civil buildings is undertaken, so as to improve the knowledge on the design and behaviour of isolated structures. To this aim appropriate seismic monitoring of isolated constructions is also essential. This is the reason why all the ongoing studies in Italy - including those of the Department of Innovative Reactors (RIN) and the Directorate of Nuclear Safety and Health Protection (DISP) of ENEA - are based on applications to civil structures. Obviously, a great care must be devoted to these applications: this is also made necessary by the lack of design rules, in Italy, for this innovative construction technique (for buildings, these must be defined by law, in our country). To this aim, R & D work is also needed: such a work, together with the experience acquired on actual isolated buildings, is essential to set up adequate design rules. On the other hand, development of design rules must be carried out in parallel, in order to determine the features of the necessary research activities. Due to the high complexity and considerable costs of development work on seismic isolation, national and international collaborations are essential to optimize such studies. This paper (which is a shorter and updated version of that presented by Martelli et al. [4] at the Post-SMiRT Conference Seminar of Nuclear and Non-Nuclear Structures of Nara, Japan, in 1991) ex- plains the national framework in which seismic isolation activities are being performed in Italy, cites applications existing in our country, summarizes the state-of-the-art of activities in progress and the work planned, and mentions international collaborations. More details on the guidelines development, experimental tests, and numerical activities have been reported by the separate technical papers of Bonacina et al. [5], Olivieri et al. [6] and Bettinali et al. [7], respectively, presented to this meeting. Further information on numerical studies in progress at ENEA-DISP has also been given by the paper of Sand et al. [8]. 2. NATIONAL COLLABORATIONS IN THE FRAMEWORK OF GLIS At present, the R&D work on seismic isolation is mainly being performed in Italy by ENEA, ENEL, ISMES and ALGA: these were the main promoters (in 1989) of the National Working Group on Seismic Isola-

-57- tion ("Gruppo di Lavoro Isolamento Sismico", or GLIS), which already includes representatives of all the organizations, universities, companies and designers who are dealing with the new technique. At the end of 1991, GLIS was formalized as a group of the Italian National Association of Earthquake Engineering (ANIDIS). 2.1 Objects of GLIS activities The objects of GLIS activities include - but are not limited to seismic isolation. Indeed, they are: (a) seismic isolation and energy dissipation systems, as well as other innovative systems that are capable of reducing the seismic risk of structures (for instance, active control); (b) the devices forming such systems; (c) the materials of such devices; (d) structures (buildings and plants) using the above-mentioned systems; (e) inner components of such structures and components at the interface between isolated and non- isolated structures, or independently isolated structures (interface components). 2.2 Purposes of GLIS The purposes of GLIS are the following: (a) to allow for the co-ordinated development of numerical and experimental studies concerning the above-mentioned systems, devices, materials, structures and components; (b) to identify and propose the studies necessary to integrate those in progress; (c) to promote new applications to buildings and plants; (d) to provide technical support to the designers of structures adopting the above-mentioned systems (data useful for the definition of such systems, co-operation in the design of systems and structures, determination of seismic input, etc.); (e) to promote activities for the verification of the above- mentioned systems and the structures using such systems (cooperation in acceptance tests, integration of these tests when needed, in-situ tests of structures); (f) to coordinate the development activities in progress on codes and standards, concerning the above-mentioned systems, devices, materials, structures and components, and to promote new activities, according to the needs that will be identified; (g) to provide support to the State institutions that are charged with the approvals of structure designs, for the design verification; (h) to provide a correct information on the developments (at both national and international levels) of the aforesaid activities, by also taking advantage of the international agreements that involve GLIS members; (i) to organize - in the framework of information activities national and international conferences, specialists meetings and workshops, and to keep adequate contacts with mass-media; (j) to evaluate the possibility of extending the methodologies under study to other fields (for instance, control of non-seismic vibrations), and in the case of interest, to promote activities in such fields also.

-58- 2.3 Members and organization of GLis At present, GLIS has 77 members (March 1992). These are grouped into the following seven cathegories: (1) representatives of the organizations that promoted GLIS constitution (ENEA-RIN, ENEA-DISP, ENEL and ISMES) and well-known experts; (2) representatives of other companies and organizations working at the development of the techniques mentioned in Sect. 2.1; (3) university professors specialized in civil engineering; (4) university professors specialized in mechanical, nuclear and industrial engineering; (5) representatives of State Departments and governments of Italian Provinces; (6) designers of isolated structures and representatives of builders of structures using base isolation or other innovative systems; (7) representatives of shops fabricating the devices and materials mentioned in Sect. 2.1. GLIS activities are coordinated by a Council, which designates the Chairman. This Council is formed by one member for each of the organizations forming the first cathegory, three experts of the same cathegory, and one member for each of the other cathegories, with the exception of university professors specialized in civil engineering, who have two members. The representatives of each cathegory 2 to 7 are selected by the members of such a cathegory, within themselves. Such a type of organization should be capable of ensuring the achievement of GLIS purposes, by allowing - at the same time - a correct representation of all the parts interested in the development of the new techniques. The first Council was formed in December 1991; Dr. Martelli of ENEA was elected as first chairman. GLIS activities are performed by subgroups, each dealing with a specific subject. 2.4 Present activities of GLIS GLIS has been very active since its foundation, i.e. before becoming a group of ANIDIS also. This will be made evident by the next sections and by Refs. [5] - [7], where co-operations in progress for the various studies and promotion of new applications are clarified. In addition, it is worthwhile mentioning that GLIS has already orga- nized a very successful national Workshop on Seismic Isolation at Bologna in May 1990, and was one of the main promoters of the equally successful International Meeting on Earthquake Protection of Buildings, held at Ancona in June 1991 (Ref. 9). Finally, particular attention has already been paid by GLIS to the information of public opinion - through adequate contacts with mass-media -, both on the occasion of the above-mentioned meetings and at the time of in-situ tests of isolated buildings (such as those described by Bonacina et al. [5] and Sect. 6.1), which had a large echo in Italy, thus strongly contributing to the promotion on seismic isolation in our country (these tests were shown by the national and several private TV channels and were reported in detail by several important newspapers, magazines and technical journals). With regard to the activities that have already been undertaken

-59- in the framework of ANIDIS, it is noted that the first, most urgent subgroups were formed by GLIS Council in March 1992. These are: (a) design guidelines; (b) experimental tests; (c) numerical analyses; (d) seismic input; (e) development of energy dissipation systems. 3. SYSTEMS CONSIDERED Studies in progress in Italy on seismic isolation of structures are based on high damping steel-laminated rubber bearings (HDLRB). All the HDLRBs used in isolated buildings and for research purposes in Italy have been fabricated by ALGA (Milano). other bearing types and energy dissipation systems have mainly used up to now for bridges only (Sect. 4). The choice of HDLRBs for the seismic isolation studies of structures was due to both the very promising features of these isolators for high risk plants also (including innovative nuclear reactor projects such as PRISM, see for instance Kelly et al. [10]), and the fact that they had been selected for the first group of large isolated buildings that have been constructed in Italy: the SIP Administration Center at Ancona (Sect. 4). All the bearings of such buildings (Figs. 1-3) had been subjected to very detailed acceptance tests; furthermore, the perspective of performing in-situ experiments on one of these buildings had been judged extremely attractive (Sect. 6.1). HDLRB is the bearing type that has also been used or considered up to now for all the other isolated buildings in Italy, with the only exception of the very first application: the Fire Command and Control Facility at Napoli (Sect. 4). It is also the bearing type that has been chosen for the first application to high risk non- nuclear plantsin Europe: some liquid gas storage tanks to be constructed in Greece by Dyckerhoff & Widmann, for which isolators are being fabricated in Italy by ALGA. Anyway, studies for the development of new materials for high damping bearings, including compounds based on synthetical polymers, are also being started (Sects. 7 and 8). Furthermore, analysis of other isolator types and energy dissipation systems has also been started or planned (Sects. 5 and 8). 4. APPLICATIONS OF SEISMIC ISOLATION IN ITALY In addition to several bridges, applications of seismic isolation in Italy mainly concern both some important public buildings, which are critical for civil defence, and some first industrial structures. More precisely (see Ref. [9] and Sparacio et al. [11]), isolation has been or is being adopted in Italy for: (a) the Fire Command and Control Center Facility at Napoli (a suspended steel structure supported through neoprene bearings at the top of large reinforced concrete columns), which was the first pio- neering application of seismic isolation in our country; (b) five large buildings (seven floors, 25 m height, 70,000 to 78,000 kN weight each), owned by SEAT, STET Division, designed by Giuliani [12], which were recently completed at Ancona (those forming the new Administration Center of the National Telephone Company - SIP - for the Marche Province); (c) a three-story reinforced concrete house that was recently completed at Squillace (Calabria);

-60- Figure 1: Plastic model of the SIP Center at Ancona.

rJ-il J

Figure 2: View of a row of isolators at the base of SIP buildings (March 1990) .

Figure 3: High damping steel-laminated elasto- .••.,",--.-. •,•- -.if- . mer bearings used in SIP buildings at Ancona, during construction (March 1990) .

-61- (d) standardized precast telephone switch houses, which have already been fabricated, to be isolated in seismic areas; (e) a second building at Ancona (owned by the Ministry of Defence), which is under construction; (f) the four-story Operating Center of the Traffic Police at Napoli, which has already been commissioned. Furthermore, among the various further applications that have been planned, it is worthwhile citing: (g) a new hospital building to be soon constructed at Castelno- vo ne Monti (Reggio Emilia), for which base isolation has been recommended by ENEA and other GLIS members, including the Ambient De- partment of the Province Emilia-Romagna and the National Group for Earthquake Defence (GNDT) of the National Research Council - (CNR), after the modification of the original, non-isolated design was judged feasible at a limited cost by GLIS experts (see Sect. 8); (h) a small masonry church at Frigento (Avellino) that should be retrofitted by use of seismic isolation to verify the applicability of this technique to historic monuments; (i) some new University buildings in seismic areas. As to buildings critical for civil defence, the possibility of isolating hospitals is considered particularly attractive (following the experience of California), according to the need of ensuring full hospital operation after strong earthquakes and the need for a better protection of both the structure and the quite expensive equipment contained: this is the reason for the great interest in the application at Castelnovo ne Monti. Also with regard to applications to industrial structures, we remind that the reasons which led Giuliani [12] to isolate the buildings of the SIP Center at Ancona were not only savings in construction costs, but especially, achievement of a much higher confidence to prevent earthquake-induced damages of highly costly inner electronics. Similar reasons and the possibility of taking advantages of the economical benefits of standardization led Giuliani [13] to also design the aforesaid precast telephone switch houses. The perspective for a better protection of energy production plants - through seismic isolation and/or energy dissipation - has also been judged very attractive by ENEL (Sect. 8). Furthermore, following the aforementioned application in Greece, these systems may be appropriate to reduce the seismic risk of chemical plants, which may be particularly sensitive to earthquakes, but which are frequently not designed to withstand large earthquakes in Italy, although they are located in highly seismic areas also. For some of those existing, isolation (and/or energy dissipation systems) may be also considered to seismically retrofit them. Seismic retrofitting by use of isolation may be advantageous for other industial plants also, and even for some historic and artistic monuments, to make them resistant against possible seismic damage. It is evident that such an application may be justified in Italy, where so many historic monuments are located in seismic areas, especially for those that have already been severely damaged by earthquakes: this is the reason for the great interest in the projects of Sparacio et al. [11], concerning the above-mentioned church at Frigento and other applications. Finally, isolation of small buildings and houses in highly seismic areas - such as that at Squillace - may be also justified, because of the better protection of human life, the structure and

-62- its other contents. In addition, according to the Japanese experience, houses such as that mentioned above (which is adjacent to an equal, non-isolated house - see Bonacina et al. [5]) form a very good tool for improving the knowledge on the behaviour of isolated structures, when subjected to in-situ tests and extensively monitored to record real earthquakes (see Sects. 6.1 and 8). 5. DESIGN GUIDELINES The first activities on seismic isolation were initiated in Italy by ENEA-RIN - with the co-operation of ISMES and GE Nuclear Energy, and the support of experts of ENEA-DISP and Bechtel Nation inc. - in 1988. They concern the preparation of a proposal for design guidelines for nuclear power plants using HDLBRs (Martelli et al. [14] & [15]). A first revision of the document is being prepared and will be soon published: it accounts for both comments received and the first results of R&D studies in progress in Italy and the USA (Olivieri et al. [6] and Bonacina et al. [5]). These activities were recently extended - as part of a cooperation with the Italian Standard Authority (UNI) - to other antiseismic devices, for application to civil buildings and non-nuclear plants. A co-operation has also been started by ENEA, ENEL and ISMES with the National seismic Service. Furthermore, extension of the guidelines document [14] to nuclear reactors using bearings different from the HDLRB has also been planned, under the sponsorship of the Commission of the European Communities (CEC): this work will be performed by ENEA, with the cooperation of ALGA, ISMES, ANSALDO and the Nuclear Engineering Laboratory (LIN) of the Bologna University. More details on guidelines development have been reported to this meeting by Olivieri et al. [6]. 6. R&D WORK IN PROGRESS In 1989, R&D work was also undertaken by ENEA, ENEL, ISMES and ALGA. This work takes avantage of national co-operations in progress in the framework of GLIS activities, as well as that of international collaborations (Sect. 9). It concerns both experimental and numerical studies of isolators, bearing materials and isolated structures. 6.1 Experimental analysis Static and dynamic tests have been presented at this meeting by Bo- nacina et al. [5]. A review on these experiments was also recently published by Martelli and Castoldi [16]. Tests have been performed on HDLRBs in various scales (with respect to SIP building bearings), rubber specimens and structures isolated by means of such bearings. These are mainly being performed at ISMES and also, at present, at the ENEA/ANSALDO Centre of Boschetto (Genova). Tests on rubber specimens and bearings (Figs. 4-7) have already provided important data (vertical and horizontal stiffnesses, damping, creep, temperature, aging and scale effects, failure modes, etc.), necessary for the development and validation of numerical models, comparison with the test results of isolated structure

-63- I

Figure 4: view of the SISTEM test machine after recent modifications performed to allow for tensile vertical loads and better guide of vertical jack; full- & half-scale modified SIP-type isolators being tested (possibility of attachment with central dowel with or without bolting; modified compound); static vertical compression test of single 1/2 scale modified isolator on modified SISTEM). Figure 5a: View of the SISTEM test machine before recent modifications (2 horizontal actuators in "push-pull" positions).

Figure 5b: Dynamic test of single full scale SIP-type isolator mounted on a roller slide (nonmodified SISTEM machine) .

-65- I-

I en ••!!

i. Figure 7: 1/2 scale bearing vzithout lateral rubber cover, displaced at 260% shear strain during a static failure test (nonmodified SISTEM test machine). Figure 6: Test of a pair of superposed full-scale isolators (non-modified SISTEM test machine). mockups and actual buildings, and development of design guidelines. Dynamic experiments of structures concerned both full-scale and scaled isolated structure mock-ups (Figs. 8-10) and actual isolated buildings (one of those forming the SIP Administration Center at Ancona, see Figs. 11-15; the isolated house at Squillace, Calabria, together with that adjacent, non-isolated house). Both snap-back tests and forced excitation experiments were performed, to large displacements (85 mm for the full scale mock-up of Fig. 8, 36 mm for the 1/4 scale mock-up of Fig. 10, 107 mm for the SIP building of Fig. 11). Forced excitation tests were both sinusoidal and (on the 1/4 scale mock-up of Fig. 10) seismic, with one- and multi-directional simultaneous excitations. Test results have already demonstrated the adequacy of seismic isolation and have provided data useful for the comparison with single bearing test results and validation of numerical models for the analysis of isolated structures. 6.2 Numerical activities The numerical activities which are in progress in Italy in the framework of the seismic isolation studies have been presented to this meeting by Bettinali et al. [7] and Sano et al. [8]. They mainly concern the definition of models for bearings and isolated structures, and their use for test design and the analysis of experimental results. Simple bearing models have been set up, and the development of finite-element (f.e.) three-dimensional (3D) and 2D axisymmetric models is in progress. Simple models have been based on the results of single bearing tests: models formed by a spring in parallel to a viscous damper, where both horizontal stiffness and viscous damping vary with displacements, have been developed by ENEA. Models based on hysteretic damping have also been developed by DISP and ISMES. Detailed bearing models (Fig. 16) include separate elements for the rubber and steel plates. A 3D model has been implemented by ENEA in the ABAQUS computer code. Linear elastic calculations have been performed with this model. The implementation of an elastic-plastic model for steel is also being completed, together with that of a hyperelastic model of the rubber, based on tests on specimens. Detailed bearing models will be validated based on measured data. They will be used for bearing design and the analysis of the effects of defects: two bearings with artificial defects were fabricated to this purpose also (Bonacina et al. [5]). As to the isolated structures, finite-difference programs were set up for the analysis of such structures in the case that they can be represented by sets of one-degree-of-freedom oscillators. The program ISOLA includes the aforementioned simple bearing model of ENEA, where both stiffness and damping depend on displacement and the effects of viscous creep are accounted for. A similar program has been based on the bearing model developed at ISMES. These models have been successfully used to analyse the experimental results concerning both isolated structure mock-ups and actual isolated buildings, based on the single bearing test data for both horizontal stiffness and damping. It is also noted that Sano et al. [8] have shown that, at less than 120% shear strain, an elastic model of bearings can be appropriate to estimate displacements and accelerations of isolated

-67- Figure 8: A view of the 9,500 kN inertial mass of the ISMES multiexcitation rig, supported by six isolators.

Figure 9: Isolator supporting the 9,500 kN mock-up, deformed before release in a snap - back test.

I .tlfl.-1,.1'

! ••, I:' i- .

••••• /;/

- Jt

Figure 10: A view of the 394 kN isolated mock-up, mounted on the MASTER shake table of ISMES.

-68- i med for the structure also, and sophisticated of the

7. DEVELOPMENT OF HDRB BASED ON SYNTHETIC ELASTOMERIC MATERIALS

f°r HDLRBs' "search

be industrialized in the medium period" (Martem e? al

8. EXTENSIONS OF THE WORK

at rovi are lready^ndefltudmgmmmv flfP ^.the knowledge on topics that

-69- NEXT PAGE(S) left BLANK mass of the SCORPIUS shake table, immediately after the conclusion of the single bearing tests with such attachments (these are in progress - see Bonacina et al. [5]). (3) Continuation of the experimental analysis of natural aging effects (in co-operation with SEAT - see Bonacina et al. [5]). (4) Extensions, mentioned in Sect. 5, of the development of design rules and guidelines (including those for isolated nuclear reactors), to bearings different from the HDLRB (including vertical isolation). (5) Detailed numerical and experimental studies on the applicability of seismic isolation to electric substations, large industrial plants and high power transformers, which have already been commissioned by ENEL to ISMES. (6) Application of seismic isolation to the Hospital of Castelnovo ne Monti, in the framework of a cooperation between the competent hospital authorities, the Province Emilia-Romagna, ENEA and GNDT-CNR (Sect. 4), and that of a project on seismic protection of hospitals which has been proposed to the CEC (this application may be followed by in-situ tests and seismic monitoring). (7) evaluation of the applicability of seismic isolation to an ENEL building at Terni. (8) Experimental analysis of interface piping at LIN Laborato- ries, in the framework of a cooperation between ENEL and ENEA and that of a specific project which has been proposed to the CEC. (9) Optimization of design and performance of HDRBs (geometry, rubber materials, reinforcement plates, attachment systems) and evaluation of the benefit of seismic isolation on structures' design, for which a specific project is being prepared. (10) Analysis of the behaviour of isolators different from the HDLRB and energy dissipation devices. Some of these activities will form an extension of the promotion work in progress, in the framework of GLIS activities. 9. INTERNATIONAL COLLABORATIONS Work will take advantage of collaborations of ENEA with international partners, which have recently been activated: these are part of the Co-operation Agreement on Research and Development of Energy between the Italian Ministry of Industry, Commerce and Handicraft and the U.S. Department of Energy (which extends the collaboration existing with GE Nuclear Energy), the Agreement of Scientific and Techological Co-operation betweeen Italy and Japan, a recent agreement with TsNIISK (USSR) and a new agreement that is being defined between ENEA and the French CEA. In addition, cooperation with several partners of other Euro- pean countries have been started in the framework of projects proposed to the CEC. 10. CONCLUSIONS This paper has provided an overview on the activities in progress in Italy for seismic isolation development and application. It has shown that considerable efforts are being devoted to this aim in our country, by clarifying the national collaborative frameworks, mentioning the various applications, summarizing the main features and results of design rules and guidelines development and R & D work in

-71- ^B^^^^^^^^E^^^^B^^^^^^^^^»^^^J t iMJsJin^^^S^^^^W^^^^^^^^^^^^^^^^^^^^^^^^^^^^^BiSiifSW^^^^^K^My-'^l

^^^^B^^sB^mSBBB^S^mI^^^^^^BHB sB^^^^^ts^SHHI^HSi^ J^-**-

JM ro i

Figure 15: View of a transducer indicating the displacement reached at the base in the most severe snap-back test of the SIP building.

I - •;

Figure 14: In-situ experiments of the SIP building: view of a hydraulic jack and a snap-back mechanism in position. progress, and outlining the new activities planned. More details on R&D work and guidelines development have been provided by separate technical papers, presented to this meeting. The great importance of international and national collaborations, for optimizing seismic isolation studies, has been stressed. Those existing have been mentioned. As to national collaborations, the main features, purposes and activities of the Working Group on Seismic Isolation (GLIS) have been outlined. The formation of similar national groups might be of interest for other countries, as well, so as to allow for a better coordina- tion of seismic isolation activities at national level and for more complete international collaborations. ACKNOWLEDGMENTS The authors thank G.C. Giuliani and all the collegues of ENEA, ENEL, ISMES, ANSALDO, ALGA, LIN and other national organizations involved in the GLIS collaborative activities, for their contributions to the work described in this paper. They also thank collegues of other countries, especially E.L. Gluekler of GE Nuclear Energy, F.F. Tajirian of Bechtel National Inc., J.M. Kelly of the University of California at Berkeley, and Y.W. Chang of Argonne National Laborato- ry, who have provided important cooperations for such a work.

REFERENCES [1] Y.W. Chang, T. Kuroda and A. Martelli, Preface: Overview and summary of First International Seminar on Seismic Base Isolation of Nuclear Power Facilities, in: Proc. First Int. Post-SMiRT Conf. Seminar on Seismic Base Isolation of Nuclear Power Facilities, San Francisco, CA, USA, ANL Report, Argonne National Laboratory, CONF- 8908221 (1989) pp. 1-8; Nucl. Engrg. Des. 127 (3) 233-237. [2] A. Martelli, Some remarks on the First Int. Seminar on Seismic Base Isolation of Nuclear Power Facilities, Energia Nuclea- re, 3 (1989) 71-74. [3] A. Martelli, P. Masoni, M. Forni, M. Indirli, B. Spadoni, G. Di Pasquale, V. Lucarelli, T. Sand, G. Bonacina and A. Castoldi, ENEA activities on seismic isolation of nuclear and non-nuclear structures, in: Proc. First Int. Post-SMiRT Conf. Seminar on Seismic Base isolation of Nuclear Power Facilities, San Francisco, CA, USA, ANL Report, Argonne National Laboratory, CONF-8908221 (1989) pp. 57- 73; Nucl. Engrg. Des., 127 (3) (1991) 265-272. [4] A. Martelli, F. Bettinali, T. Sand, G. Bonacina, E. Gian- greco and A. Marioni, Overview on the activities in progress on seismic isolation in Italy and future R&D work, Proc. Post-SMiRT Conf. Seminar on Seismic Isolation of Nuclear and Non-Nuclear Struc- tures, Nara, Japan (1991); to be published in Nucl. Engrg. Design. [5] G. Bonacina, F. Bettinali, A. Martelli and M. Olivieri, Experiments on seismic isolation in Italy, Paper presented to this meeting. [6] M. Olivieri, A. Martelli, F. Bettinali and G. Bonacina, Development of guidelines for seismic isolation in Italy, Paper presented to this meeting. [7] F. Bettinali, A. Martelli, G. Bonacina and M. Olivieri, Numerical activities on seismic isolation in Italy, Paper presented to this meeting.

-73- [8] T. Sand, G. Di Pasquale and E. Vocaturo, Linear analysis for base isolated structures, Paper presented to this meeting. [9] Earthquake Protection of Buildings, CREA ed., Ancona, Italy, 1991. [10] J.M. Kelly, F.F. Tajirian, E.L. Gluekler and V. Veljovich, Performance margins of seismic isolation bearings, in: Proc. 1990 Int. Fast Reactor Safety Meeting, Snowbird, Utah, USA, Vol. Ill (ANS, 1990) pp. 393-412. [11] R. Sparacio, A. De Luca and G. Serino, New applications of seismic isolation to civil constructions in Italy, in: Proc. Int. Post-SMiRT Conf. Seminar on Seismic isolation of Nuclear and Non- Nuclear Structures, Nara, Japan (1991); to be also published in Nucl. Engrg. Des. [12] G.C. Giuliani, Design experience on seismically isolated buildings, in: Proc. First Int. Post-SMiRT Conf. Seminar on Seismic Base Isolation of Nuclear Power Facilities, San Francisco, CA, USA, ANL Report, Argonne National Laboratory, CONF-8908221 (1989) pp. 220-245; Nucl. Engrg. Des. 127 (3) (1991) 349-366. [13] G.C. Giuliani, Design and construction of prefabricated isolated buildings, in: Proc. Int. Meeting on Earthquake Protection of Buildings, Ancona, Italy (1991), pp. 191/C-201/C. [14] A. Martelli, P. Masoni, G. Di Pasquale, V. Lucarelli, T. Sano, G. Bonacina, E.L. Gluekler and F.F. Tajirian, Proposal for guidelines for seismically isolated nuclear power plants - Hori- zontal isolation systems using high damping steel-laminated elastomer bearings, Energia Nucleare, 1 (1990) 67-95. [15] A. Martelli, M. Forni, M. Indirli, P. Masoni, B. Spadoni, G. Bonacina, G. Di Pasquale, T. Sano and E.L. Gluekler, Development of design guidelines for seismically isolated nuclear reactors and R&D work performed by ENEA, Nuclear Technology, 97 (1992) 153-169. [16] Martelli, A. & A. Castoldi, Seismic isolation of structures, In: J. Donea & P.M. Jones (eds.), Experimental and numerical methods in seismic engineering (1991), pp. 351-377. Dordrecht/ Boston/London: Kluwer Academic Publishers for the CEC.

Figure 16: 3D model of SIP-type Figure 17: 3D model of the SIP bearings; first mode of the un- building; first vibration mode loaded bearing. of the superstructure.

-74- A Broad Review of the Status of Seismic Isolation Study in JAPAN

Sakae AOYAGI Dr.-Ing. Senior Research Engineer Central Research Institute of Electric Power Industry

Heki SHIBATA Dr. Eng. Professor Institute of Industrial Science University of Tokyo

ABSTRACT XA0055379

In Japan, studies on seismic isolation technologies have been extensively performed by a several organizations for the last decade, in order to apply them to fast reactors and thermal reactors. These programs have been managed by CRIEPT11 , JAPC*21 , NUPEC*3) , PNC"11 the electric utilities and so on. Japanese major reactor manufacturers and construction companies have been participating in each program. Consequently the base isolation technologies in Japan have been well - developed and are changing their stage from the basic studies to the integration for actual nuclear application. In this paper, the background, the current status and future perspective on the seismic isolation studies conducted by a several Japanese organizations are concisely described.

1. BACKGROUND

Prior to describe the current status of seismic isolation studies in Japan, it is worth briefly recalling the research and development activities in the related field.

Japanese major construction companies and reactor manufactures have been playing an important role in seismic isolation studies. Their contribution is outlined as follows.

*1) Central Research Instiute of Electric Power Industry *2) Japan Atomic Power Company *3) Nuclear Power Engineering Center *4) Power Reactor and Nuclear Fuel Development Corporation

-75- In Japan, the base isolation concept have been studied for quite a long time. Since Kozo Kawai's paper appeared in 1891, a wide variety of base isolation concepts have been reported. However, research and development activities on base isolation technologies started in the late 1970s since laminated rubber bearings were introduced from Europe. Since the early 1980s, practical applications have become common, which accelerated research and development efforts. This was made possible by the maturity of basic technologies in response to demand from the construction market, and a result of the comprehensive development of relevant technologies. Leading construction companies played a major role in this development process, combining extensive development activities with experimental studies on isolation devices, the development of analytical methods, evaluation of earthquake ground motions, investigation of maintenance procedures, and the trial construction at their own facilities. The major isolation devices which have led to the practical implementation are the laminated rubber bearings using natural rubber, which are commonly used with some kinds of dampers, the lead rubber bearings, high damping rubber bearings, and sliding bearings.

Over the ten years through 1991, 63 base isolated buildings have been constructed in Japan, and recently a tendency has developed toward making application to larger buildings. In a number of the base isolated buildings, monitoring sensors for earthquake response were installed and verification studies based on the data obtained have been conducted continuously, with the expectation that the results would contribute to the further development of the isolation technologies. One of the most significant progress in technology aspects in this decade is that the examination guidelines for building permits and approvals by the administrative office were formulated in 1989 by the Building Center of Japan under the Ministry of Construction, and the Recommendation for Design of Base Isolated Buildings was published in 1989 by the Architectural Institute of Japan.

In the field of nuclear power facilities, research and development on the base isolation technologies by construction companies began in mid-1980s. Needless to say, applications for nuclear power field require highest possible safety and reliability. Based on the base isolation technologies accumulated by construction companies through their construction experiences of general buildings, research has been conducted, being sponsored by PNC and the electric power companies etc.

-76- Major accomplishments have been gained, such as those for adequate safety allowances, technologies to predict dynamic responses, the assurance of safety over the lifetime of plants, and the manufacture of large isolation devices. Concerning earthquake ground motions, data have been accumulated from earthquake observation sites around the nation.

Japanese reactor manufactures, now being involved in the CRIEPI's MITI-sponsored seismic isolation program (1987-93), took notice of the strategic importance of seismic isolation technologies in early years. With an ultimate goal of trimming/simplifying structures and facilitating plant design standardization, HITACHI Ltd. for example, performed a feasibility study on the application of seismic isolation to FBR (1984-86) under the sponsorship of MITI. The study covered a wide spectrum of technical issues including the development of isolators and nonlinear analytical method, the development of a seismically isolated FBR plant concept and its economic evaluation, and the confirmation of floor response reduction effect, thereby paving the way for large-scale seismic isolation research programs.

Under these situations, both authorities and the electric utilities paid great attention and had interest to the effectiveness of the application of seismic isolation technologies to important nuclear facilities, and some big projects were mapped out and started in the mids of 1980's, some of which are still going on.

2. CURRENT STATUS

2.1 FAST BREEDER REACTOR

FBR is operated at high temperature, which causes severe thermal loads in the reactor vessels and other major components. Therefore, the thickness of reactor vessel, for example, should be minimized to limit the thermal stress taking advantage of low operation pressure. This is, however, a disadvantageous features on the seismic design of the reactor vessel, requiring so called "harmonization" of the design for thermal loads and that for seismic loads. The seismic isolation concept is one of the most effective solutions to overcome this technical problem.

-77- [CRIEPI : Ref.(l)~ (8)] CRIEPI has been managing the verification test program of FBR seismic isolation systems since 1987 under the sponsorship of MITI. This is a seven year research program being devided into phase I (87- 89) and phase II (90-93), and CRIEPI is now 5 years into this program which is currently well on schedule. In carrying out this CRIEPI's MITI-sponsored program, an advisory committee has been established in CRIEPI in 1987, headed by one of the authors, H. Shibata. Four reactor manufacturers, six construction companies and three rubber makers have been taking part in this.program. This research program mainly focuses its target on the establishment of the horizontal base isolation technologies of reactor buildings, in which a several large scaled tests such as the failure modes tests of various isolators, the shake table tests of various types of isolated structures have been carried out. The primary objectives of these tests are to establish the design criteria of isolators and isolated structures as well as to assure their long term integrity. Furthermore, earthquake motion with long period components have been studied in order to prepare an appropriate method to determine design earthquake motions. Based on these accomplishments, methodologies for the isolation design of FBRs will be proposed by March 1994. The major contents of the program are summarized as follows.

[1] TESTS AND ANALYSIS OF ISOLATORS (TESTS) One of major studies on isolators is the deformation characteristics tests and the failure tests (breaking tests) of large scaled elastomeric rubber bearings with a tentative specification for nuclear use. Two-dimensional testing apparatus which is capable of applying both horizontal static load and vertical static load was introduced for this purpose in CRIEPI in 1988. Three kinds of test models with different sizes have been prepared. By using this testing apparatus and these models, more than 70 failure tests have been carried out until 1991 under a several load conditions. Based on the accumulated test results, the envelope curve for the failure has been developed so far. Isolators for nuclear use should have longer servicebility, therefore, the aging tests simulating the service life of the plant has started in 1991. The objective of the tests is to examine the influence of the aged deterioration of lead rubber bearings and high-damping rubber bearings on the mechanical characteristics of these bearings. The accelerated aging test equipment was introduced for this purpose.

-78- The bearings for the tests were subjected to heat - accelerated deterioration on the assumption that such acceleration will occur in accordance with Arrhenius's law. After this procedure, deformation characteristics tests and failure tests have been carried out to examine the changes of horizontal stiffness , equivalent damping constant, breaking strain and so on. The comparison with the test results of non- deteriorated test models is now being performed. All these accomplishments will be evaluated and reflected on the design method described later.

As for the dynamic characteristics of isolators, dynamic tests using relatively small scale models have been planned, and will start toward the end of 1992. The dynamic two-dimensional test equipment for this purpose has already constructed in 1991.

(ANALYSIS) Numerical analysis of laminated elastomeric rubber bearings using the general purpose finite element programs MARC and ABAQUS have been carried out, assuming that the rubber is isotropic, hyperelastic, and nearly imcompressible. As for the relationship between the shear restoring force and shear displacement, the numerical results agree well with the experimental ones up to large shear strain if the constitutive equation derived by material tests-based data was employed. For the advanceed numerical analysis method for isolators, especially for the prediction of vertical displacement under horizontal load, a special computer program which takes into account the compressibility of rubber has been developed by CRIEPI.

[2] TESTS AND ANALYSES OF SEISMICALLY ISOLATED STRUCTURES (TESTS) As far as the shake table tests are concerned, two sort of tests have already been conducted in the program. First, shake table tests on seismically isolated FBR plant model were carried out in 1989. The model is three - stories steel frame structure weighing about 20 tons. This is supported by 9 laminated rubber bearings which are reduced to a scale of 1/15. The notable feature of the test is that the shear and axial forces acting on the rubber bearing during the excitation were directly measured by component force transducers. The fundamental dynamic characteristics of isolators and isolated structure, and the effectiveness of the base isolation system were verified through the design basis excitations.

-79- Secondly, models having two different height of gravity center with 8 isolators were made, and the shake table tests were carried out in 1991. These tests covered the following cases. (CASE A) Shake table tests under design basis seismic motion. The objective of the tests is to investigate the detailed behavior of isolated structures and to prepare the practical design method. An analysis method was also verified through this test. (CASE B) Shake table tests under beyond design basis seismic motion. The objective of the tests is to investigate the ultimate characteristics at extremely large amplitude as well as to develop reliable analytical methods. Especially the dynamic failure process and the effect of the height of gravity center on the dynamic behavior has been investigated in detail through the tests.

(ANALYSIS) Three kinds of hysteresis of isolators, Overlay model, Ramberg- Osgood model and Harding-Drnevich model, have been used to analyze and to compare to the results of the shake table test under design basis seismic motion. The analytical result with Ramberg-Osgood model and Harding-Drnevich model have shown good agreement with the test results.

[3] EVALUATION OF EARTHQUAKE LOAD FOR SEISMICALLY ISOLATED STRUCTURE.

Several methods to predict the earthquake ground motions with long periods components have been examined to confirm their applicabilities to the seismic design of FBR. A tentative design spectrum was also proposed for the research program. As for the tentative design spectrum, several studies on the estimation of velocity response spectra were performed with various methods using strong motion accelerograms on bedrocks. Based on these results, and from an engineering standpoint, the tentative spectral value for Si earthquake motion is set at 100 cm/sec in the range of 2 to 10 seconds in period. This value is almost equal to the velocity response spectrum value derived from analytical studies. As for short- period domain, a guideline of the design base spectra for light-water reactors (so-called Osaki Spectra) was applied. Since there is a difference between the two spectral values around the period of 2 seconds, the tentative spectrum is defined by combining and interpolating these two spectra.

-80- [4] PSA METHODOLOGY DEVELOPMENT AND APPLICATION TO THE ISOLATION SYSTEM

Seismic PSA methods (Probabilistic Safety Assessment) for nuclear power plant have been developed in the past decade. Especially in the U. S.A., a number of utility - based seismic PSAs using the so called "simplified method" for LWR have been conducted so far. In the case of isolated structures, it is difficult to extrapolate the response for earthquake motion with arbitrary intensity from the response under design earthquake motion, because the isolated structures behave inelastically under design earthquake motion, and furthermore "hardening effect" of rubber bearings have to be taken into account irt the ultimate state. A method proposed by CRIEPI, which is based on regression analysis using observed and synthesized earthquake motions is developed and usedL In this analysis, the effect of hardening of rubber bearings and the non-linearity of the super structure, as well as the "group effect of rubber bearings", were taken into account. Thus, so far, it has been made clear that the isolated structures are highly reliable against earthquakes, comparing to the non - isolated nuclear structures (i.e., probability of failure is quite small). Im the future scope of this study, improvement of the accuracy in the fragility estimation and the reflection of the reliability analysis te» the design specification are to be included.

[5] PROPOSAL OF DESIGN METHOD FOR SEISMIC ISOLATION SYSTEM.

With the assistance of professors from a several universities and? engineers from electric power companies, a tentative draft has been developed by integrating the findings of various studies completed SOB far. The proposal of design method scheduled in 1993 will consist of the? following three parts.

Volume I Safety Design Principle of Seismic Isolation Systems for FBRs. Volume II Seismic Design Principle of Seismic Isolation Systems for FBRs. Volume IH Design Procedures of Seismic Isolation Systems for FBRs.

In volume I , basic principles of the safety design for FBRs will be

-81- described through examining the applicability of the current safety design principles to seismic isolation systems. In volume n , basic principles of the seismic design of isolation systems for FBRs will be described including various criteria such as allowable limits. In volume HI , a design manual will be presented. These descriptions will be summarized on the basis of various findings including full-scale tests of isolators, scaled shake table tests of isolated structures, and a number of analytical evaluations etc.

[JAPC : Ref.(9)] JAPC has an interest in the application of seismic isolation technologies to fast breeder reactors in the future, and recently began the feasibility study on isolated FBRs with close information exchange with CRIEPI's research program. CRIEPI's major role is the verification of the fundamental technologies of seismic isolation concept and the summarization of the design methodology. On the other hand, JAPC has a complementary program, namely, they have been carrying out the feasibility design work of an isolated nuclear reactor building. With the aim of examining the feasibility of practical application of seismic isolation technologies, JAPC has designed a top-entry loop type FBR using horizontal seismic isolation bearings, and cost saving effects has mainly been evaluated and technical issues to be solved for practical application have been identified. The technical and economical assessment for the seismically isolated plant was made in comparison with a preliminary conventional seismic design. The pipework interfaces were especially investigated to confirm that the practical seismic analysis methods using enveloped response spectrum gave conservative seismic loads to the structures and they were appropriate in evaluating the pipings running between the isolated and non - isolated part receiving different excitation at each end. In particular, the measure of absorbing the relative displacement of main steam pipework by means of looping pipework system has been investigated and confirmed as feasible. In parallel with these design studies, JAPC has been performing basic R&D, such as the development of steel dampers and the dynamic rupture test of the multi-arrayed laminated rubbers.

[PNC : Ref.(lO)] PNC started its research and development program of seismic isolation technologies in 1983. In this program, a three dimensional response analysis methodology for the base isolation system was

-82- developed based on mathematical models concerning the restoring force characteristics of several types of isolation bearing. The method was verified by a series of shake table tests using a small scale model and then applied to assess the seismic response of a conceptual base isolated fast reactor. Recently PNC has embarked on a new research program, in which possibility of a vertical isolation of reactor components, in addition to horizontal base isolation, will be pursued.

2.2 LIGHT WATER REACTORS

LWRs have been well developed and standardized, but it is still important and valuable to consider the application of the base isolation technologies to LWRs, especially from the viewpoint of one of the solutions for the enlargement of nuclear construction sites.

[ELECTRIC UTILITIES : Ref.(ll)] TEPCO and the other electric utilities had been studying on the applicability of seismic isolation technologies to LWRs until 1990. A joint study had been carried out for 6 years (1985-90) by ten electric power companies, three reactor manufacturers and five construction companies with the objective of establishing a design method for base isolated LWR plants. This study consists of three phases with each phase being for a period of 2 years. In phase I (1985-86), analytical studies of base isolated structures, trial design studies of base - isolated buildings and loading experiments of base isolation devices were carried out. The possibility of application of base isolation systems to LWR buildings was also evaluated. In phase II , (1987-88), focusing on the evaluation of technical feasibilities and reliabilities of base isolation systems, studies were made such as two-way loading experiments of base isolation devices and shake table experiments on reduced base isolated structure models. In phase! (1989-90), a draft of the design guide on base isolated structures was prepared based on these studies.

The composition of the draft guide is given as follows. (1) Principle of Earthquake Resistant Design (2) Input Earthquake Motion for Design of Base - isolated Structures (3) Design Techniques for Base-isolated Apparatus (4) Design Techniques for Base-isolated Reactor Buildings (5) Design Techniques for Equipments and Piping Systems

-83- (6) Quality Control and Maintenance Control of Base Isolation Devices

[NUPEC] In NUPEC, a research program of computer floor isolation sponsored by MITI started in 1988. This program will continue until 1993. In this test program, improved and standardized 1000 MWe class LWR plant located in high seismic zone was selected as a reference power plant. Test models are composed of computer system [computer, process input and output units, auxiliary memory units], main control panel, operator cosole [display, printer etc.] and seismic isolation devices. Vertical and horizontal synchronized excitation which is larger than Sz earthquake motion will be applied to the test model (up to the functional limit of the table). Through this test, the structural and functional integrity of isolated the computer system will be verified.

2.3 OTHER RESEARCH AND DEVELOPMENT

[MHI : Ref.(ll)] From 1985 to 1987, base isolation of the PWR spent fuel storage rack was studied by the Kansai Electric Power Company, INC. and Mitsubishi Heavy Industries, Ltd, using a 1/3 reduced scale model. In this study, a series of seismic tests were conducted using a three-dimensional shake table. A sliding-type base - isolation system was employed for the prototype rack considering environmental conditions in actual plants. Non - linear seismic response analysis was also performed, and it was verified that the prototype of a base - isolated spent-fuel storage rack had a sufficient seismic safety margin for design seismic conditions.

[IHI : Ref.(12)] Though the laminated rubber bearing serves as an effective seismic isolator to the horizontal direction, because of its high vertical stiffness, which is unavoidable to obtain the large load capacity, it can not reduce the response under vertical seismic motion. Certain areas of high technology activities may call for isolation from vertical as well as horizontal seismic excitation. Therefore, the development of three-dimensional isolator with large load capacity is still needed. A new three-dimensional isolator has been proposed by IHI, where a high-pressure air spring is mounted on a laminated rubber bearing. The vertical flexibility is induced by the air spring, while the horizontal

-84- one is primarily by the laminated rubber bearing. To examine the effectiveness of the above concept, shake table tests of experimental models were performed, where a concrete block weighing about 24 tons was mounted on the four isolators. For suppressing rocking motions, an inclined support method was also tested. It is reported that the results were satisfactory and the effectiveness of new isolators has been confirmed.

[CONSTRUCTION COMPANIES : Ref.(ll)] The current status of isolation-related R&D being conducted by construction companies can be summarized as follows. Concerning the development of isolation devices, intense competition among companies is almost over and a focus has been placed on achieving advanced isolation performance with highly sophisticated technologies. Kajima Corporation, for example, has been developing laminated rubber bearings with thick rubber layers being effective for not only earthquake excitation but also vibration caused by motor traffic and other sources, as well as special lead rubber bearings being able to provide protection not only from large earthquakes but also from the more frequent smaller earthquakes. Moreover, they have been developing fireproof covers for rubber bearings, which are necessary for the effective application of isolating layers to residential and parking spaces. These devices have already been used in an actual building. As the number of isolated structures increased, verification data has been accumulated from in-site vibration tests as well as from continual earthquake observations. Earthquake monitoring systems have been installed in almost one-third of the base isolated buildings existing in Japan. Partly due to the fact that the nation is subjected to frequent earthquakes, numerous records of building response to actual earthquakes have been acquired. Records of response to major earthquakes are compiled regularly by the Building Center of Japan and released to the public. The highest intensity recorded until now was 5 (in the Japanese Meteorological Agency's seismic intensity scale). Regarding research on methods of designing and analyzing base isolation structures, intensive efforts are being directed at methodology for evaluating the ultimate characteristics of base isolation structures in the field of nuclear power, where parametric tests and shake table tests are conducted under severer conditions than expected to occur.

-85- 3. FUTURE PERSPECTIVE

As briefly mentioned above, a several research programs on seismic isolation technologies have produced a variety of accomplishments. As for the base isolation technologies of the nuclear reactor building, a great number of data assuring the reliability and the integrity of isolators and isolated structures have been accumulated, that enable the safety evaluation from the viewpoint of seismic margin, and suggest the high possibility of standardization of the nuclear power plant. Consequently, the development of the base isolation technologies in Japan are changing their stage from the basic studies to the integration for actual nuclear application.

Future developments activities in base isolation will include, for example, diversified technologies to satisfy a wide range of requirements, construction cost reduction, the completion of design methodology to apply the horizontal base isolation to actual plants, and the development of three-dimensional base isolation systems which can reduce vertical earthquake motion in addition to horizontal motion. Needless to say, the study on earthquake motion considering long period components should be continued further.

In addition, it seems that there still exist a big barrier to employ seismic isolation technologies in FBR plants. What is it ?. It is not an easy question. Some say that there are no experience in nuclear field in Japan, others say that a kind of demonstration tests using proto type models are needed before introduction. Accumulated accomplishments can offer evidence to overcome these opinions, but more persuasive evidence should be persued. One of the answer will be in the "climate" of nuclear technologies. For example, FBR should employ innovative technologies to achieve the goal of economy, but at the same time, from the viewpoint of "nuclear safety", it should follow, to some extent, the conventional safety concept accumulated in the history of construction, operation and performance of LWR. Therefore, a kind of verification on "comparison of safety" between isolated whole plant and non - isolated plant will be needed further. The public perception for nuclear power and the safety "climate" have been totally and irrevocably changed by Chernobyl accident. For this reason, especially for FBRs which is expected to be the new energy source of coming century, an internationally acceptable safety concept

-86- will be required. Seismic isolation concept is not an exception. A requirement for more detailed and deeper discussion on the safety assessment will come about in the near future. In the case, international collaboration will become particularly important to obtain the public perception for the safety of seismic isolation technologies. It is hoped that the research and development activities in the safety area will be cooperated internationally and will provide increasing confidence in using seismic isolation to nuclear facilities.

ACKNOWLEDGEMENT

This paper was made based on the discussions with JAPC, NUPEC, PNC and CRIEPI Staff members etc. The authors are very grateful to members of each organization. The authors especially would like to express thanks to Mr. TAKABAYASHI and Dr.-ENG. MIZUKOSHI from KAJIMA Corporation, Mr. YOSHINARI from HITACHI, Ltd., Dr.-ENG. TOKUDA from IHI, Mr. ITOH from MHI Ltd., Mr. AMADA form JAPC, Mr. TANAKA from NUPEC and Dr. IWATA from PNC for their kind and deep discussions. Special thanks are given to Mr. HAGIWARA from Abiko Lab. CRIEPI, for his review on whole expressions and descriptions of the paper.

(REFERENCES)

(1) ISHIDA, K. et.al, "Failure Tests of Laminated Rubber Bearings", 11th SMiRT Vol.K TOKYO (1991)

(2) MAZDA, T. et.al, "Test on Large-Scale Seismic Isolation Elements Part 2 Characteristics of Laminated Rubber Bearing Type" 11th SMiRT Vol.K TOKYO (1991)

(3) SHIOJIRI, H. et.al, "Numerical Method for Laminated Elastomer Bearings" 11th SxMiRT Vol.K TOKYO (1991)

(4) ISHIDA, K. et.al, "Shaking Table Test on Base Isolated FBR Plant Model Part 1 Shaking Table Test Results" 11th SMiRT Vol.K TOKYO (1991) (5) ISHIDA, K. et.al, "Shaking Table Test on Base Isolated FBR Plant Model Part 2 Simulation Analysis" 11th SMiRT Vol.K TOKYO (1991)

-87- (6) YABANA, S. et.al, "Response of Base Isolated Structure During Strong Ground Motions beyond Design Earthquake", 11th SMiRT Vol.K TOKYO (1991)

(7) ISHIDA, K. et.al, "Elastic-Plastic Analysis of Base Mat Concrete for Base Isolated FBR", 11th SMiRT Vol.K TOKYO (1991)

(8) ISHIDA, K. et.al, "A Study on Analytical Methods to Evaluate Sloshing Phenomina of Base Isolated LMFBR", 11th SMiRT Vol.K TOKYO (1991)

(9) WATANABE, Y. et.al, "Study on Practical Application of Seismic Isolation to Top-Entry Loop Type FBR Plant" FAST REACTORS AND RELATED FUEL CYCLES'91 KYOTO

(10) MORISHITA, M. et.al, "Investigation on Base Isolation for Fast Breeder Reactor Building", 10th SMiRT Vol.K 2 ANAHEIM (1989)

(11) FUJITA, K. et.al, "Seismic Testing of the Base-Isolated PWR Spent- Fuel Storage Rack" JSME International Journal Series! , Vol.33, No.3 (1990)

(12) TOKUDA, N. et.al, "Shaking Test of 3-Dimensional Isolator by Using Air Spring and Rubber Bearing, ASME-PVP vol.127 (1987)

-88- XA0055380 The Current Status of Seismic Isolation Technology in the United States

James M. Kelly Professor of Civil Engineering University of California at Berkeley Earthquake Engineering Research Center Richmond, California 94804

Abstract

Seismic isolation is at the present time in a very active state of development. Many new types of isolation systems are being explored and elastomeric isolators, the system which has been employed on almost all isolation systems completed to date, continue to undergo improvements. At least one dozen large projects, either new or the retrofit of existing buildings, have been completed and design studies are underway for at least another one dozen large projects. A large experimental research project for isolators with nuclear reactor application has been carried out over the past few years at EERC. This program has involved shake table testing and the testing of full-scale and model isolators. A wide variety of isolators have been tested including low-shape factor, moderate-shape factor, and very high-shape factor elastomer bearings. The range of elastomers that have been tested include low-damping, high-damping, and very low-modulus compounds. Full-size and model isolators have been tested to failure in several failure modes and the safety margins for isolation systems have been established. The test results have shown that properly designed and manufactured isolators for nuclear reactor applications can sustain levels of loading beyond any possible seismic input and demonstrate that failure of an isolation system cannot occur before failure of the isolated structure. Thus, the use of isolation can only have beneficial contributions to the protection of nuclear facilities, internal piping, and equipment. The presentation will review the latest developments in the implementation of base isolation and describe the results of the test program for its application to nuclear facilities.

-89- The Current Status of Seismic Isolation Technology in the United States

James M. Kelly Professor of Civil Engineering University of California at Berkeley Earthquake Engineering Research Center Richmond, California 94804

Introduction

At the present time, the concept of seismic isolation is in a very active state of development. Ten years ago the use of base isolation to reduce the response of structures to seismic activity was not regarded as a realistic approach and was met with great skepticism by the engineering community. After a slow start, the concept is gaining widespread acceptance. The extent of this acceptance can be gauged by the large number of journal articles, technical reports, workshops, and symposia devoted to the topic. At least one dozen large projects, either new or the retrofit of existing buildings, have been completed and design studies are underway for at least another one dozen large projects. It is now generally accepted that a base-isolated building will perform better than a conventional building in moderate and strong earthquakes. A number of buildings so designed has experienced such earthquakes and their response has confirmed these expectations. The major benefit of isolation in such cases is to reduce damage to contents and sensitive internal equipment and in many buildings, such as computer manufacturing facilities, emergency preparedness centers, and hospitals, the reduction of damage to equipment is of sufficient importance to justify the increased initial cost of isolated construction. Nuclear power plants are another class of building in which the reduction of response of internal equipment is of primary concern. The analysis of equipment and piping systems in nuclear plants for seismic loading is one of the most expensive parts of the design process. The analysis is complicated by the fact that with conventional construction, the higher levels of the plant have amplified accelerations and the time histories at each level can be very different. Generally, it is necessary to analyze piping and equipment using multiple support response spectrum analyses. Base isolation reduces drastically the amplification of acceleration at the higher levels of the plant and would permit the use of simpler design methods for piping and equipment as well as eliminating the need for seismic restraints such as snubbers. As the popularity of base isolation has grown, so too has the volume of research devoted to developing new types of isolation systems. The system which has been

-90- -2- employed on almost all isolation systems to date involves the use of elastomeric isolators; current research continues to improve the performance of these isolators. A large experimental research project for isolators with nuclear reactor application was carried out over the past few years at Earthquake Engineering Research Center (EERC). This program involved shake table testing and the testing of full-scale and model isolators. A wide variety of isolators were tested including low-shape factor, moderate-shape factor, and very high-shape factor elastomer bearings. The range of elastomers that were tested included low-damping, high-damping, and very low-modulus compounds. Full- size and model isolators were tested to failure in several failure modes and the safety margins for isolation systems were established. The test results show that properly designed and manufactured isolators for nuclear reactor applications can sustain levels of loading beyond any possible seismic input and demonstrate that failure of an isolation system cannot occur before failure of the isolated structure. Thus, the use of isolation can only have beneficial effects in the protection of nuclear facilities, internal piping, and equipment. This paper will review the latest developments in the implementation of base isolation and describe the results of the test program for its application to nuclear facilities.

--91- -3-

Base Isolation in 1992

New types of isolation systems have been proposed that are based in some way or another on sliding systems. Sliding elements and sliding isolation systems have been tested in the bearing testing machines at EERC and in model systems used in shake table testing. As compared to elastomeric systems, sliding systems have both a number of advantages as well as some drawbacks. The major advantage is cost. Slid- ers are widely used for both bridges and to reduce shrinkage and thermal stresses in large reinforced-concrete structures; thus, the manufacturing base for sliding elements is very wide. Many companies make sliding bearings and the industry is highly com- petitive. Thus, the cost of a sliding bearing is low compared to that of an elastomeric bearing. The cost factor between the two depends on detailing and size, but might be of the order of 1/2 to 1/3. The other advantage the sliding bearing has is its low height as compared to an elastomeric bearing. With one notable exception, the Kroe- berg Nuclear Facility in South Africa isolated on sliders of a lead-bronze alloy in contact with stainless steel plates, sliding bearings are usually made of teflon pads filled with fiberglass bearings on stainless steel plates. The coefficient for the rate of movement that results from thermal expansion or shrinkage of concrete is around 5%. However, this coefficient is very sensitive to velocity at pressures below around 2000 psi, and increases to around 20% at the velocities of seismic loading. The velocity sensitivity makes such a system extremely difficult to analyze and there have been attempts to eliminate, or at least reduce, the velocity sensitivity by increasing the pressure on the teflon pad. At pressures around 10,000 psi, the coefficient of friction remains of the order of 5% at seismic velocities and is not very velocity sensitive. The difficulty in such a strategy is that the bearing plate must be designed to sustain this pressure with large displacements and not a lot of data is available on the long-term behavior of the teflon-stainless steel surface under these high pressures. Such an approach is unlikely to be acceptable, or to pass code requirements, until these questions are answered. Another problem in using sliders, and only sliders, in an isolation system is that there is no effective restoring force, and thus the design factor for the displacement will become extremely large. Since this displacement can be in any horizontal direction, the diameter of the stainless steel bearing plates and its support system must be made very large; in addition, the superstructure components bearings on this must be designed for large moments caused by these large displacements. It is possible to introduce a restoring force capability in several ways. The sliding bearings can be combined with elastomeric bearings. The combination of sliders and elastomers was originally proposed by the author in 1978 [1] as a way to make use of the best features of both types of isolator. The use of sliders gives a system

-92- -4- with a long period without running the risk of instability in the elastomeric bearings. The rubber bearings control the displacement by providing a centering section; additionally, they control torsion and if the displacements exceed the design level they produce stiffening action. This type of system was studied experimentally on the shake table and its performance was excellent. The results are published in a number of articles [2] [3]. A system of this type was used in the seismic rehabilitation of a Univer- sity of Nevada building, the Mackay School of Mines in Reno, Nevada [4]. The retrofit of this building is due to be completed in 1992. At the present time, retrofits projects constitute a large proportion of the base isolation projects that are under design or are being proposed, and sliding systems have been proposed for several. The use of base isolation for retrofit generally involves a brittle and weak structure, i.e., an unreinforced masonry building or a reinforced- concrete building of an early design that does not include the type of detailing of the reinforcement that will ensure ductile performance. Base isolation will lower the force demand on the structural system and impart a certain degree of energy absorption to the structure. However, if a sliding system is used for retrofit it is absolutely essential that the force which causes the slider to break be predictable. If the break-away force increases over the years of quiescence, the possibility exists that the structure could be damaged before the isolation system begins to move. If a weak, brittle structural system begins to deteriorate above the isolation system, it may never be able to produce the necessary force to cause the isolation system to start to slide and the building will act as if it were unisolated; the mitigating effects of the isolation system will not be achieved, thus negating the whole point of the retrofit. There are, as of 1992, at least 3 retrofit projects that have been completed or under construction; two using lead-rubber bearings, and as mentioned above, one using the combined high-damping rubber and slider system. Several other large retrofit projects in the northern California area are in the design phase, including Oakland City Hall, Hayward City Hall, and San Francisco City Hall. The United States Government buildings in San Francisco are to be retrofitted with base isolation (and at least one state building).

-93- -5-

Recent Experimental Studies of Elastomeric Isolator Performance

As a result of an interest to use base isolation for nuclear power plant facilities, a large program of bearing testing has been carried out at EERC over the past five years. The funding of this testing program came from several companies involved in the developing of liquid-metal fast-breeder nuclear systems and from the U.S. Department of Energy through the Argonne National Laboratories. A large number of different types of bearings were designed and procured from several bearing manufacturers in the United States, United Kingdom, and Japan. These were tested at EERC to determine their mechanical behavior and to study their failure mechanisms. The results of this test program have been published in a series of EERC technical reports. As an example of the type of results obtained during this program, it is useful to review the tests of four bearings obtained from Bridgestone Corporation in Japan. These bearings were models of the Bridgestone high-damping rubber bearings used in several large isolation projects in Japan, including the Tohoku Power Co. Office Build- ing. The results of the test program are given in detail in Ref. [5]. The rubber compound used in the bearings is designated KL401, which is a high-stiffness, high-damping rubber with around 31% carbon filler. The rubber has a shear modulus of around 0.86 MPa (125 psi) at 100% shear strain and an equivalent viscous-damping factor of around 15%. The dimensions of the bearings are shown in Fig. 1. The bearings are very flat with a width that is about four times the height between the end plates. The shape factor is 30, which for these bearings is very high. Four bearings were tested under horizontal shear in displacement controlled cycles from ±5% up to a maximum of ±350%. The test program was identical for all bearings with one variation, the vertical pressure was different for each bearing and ranged from zero pressure through 3.45 MPa (500 psi), 6.90 MPa (1000 psi), to 10.34 MPa (1500 psi). After the cyclic tests, each bearing was loaded monotonically to failure at the same level of vertical load. The effective stiffness and the equivalent viscous damping are the characteristics of most interest to be determined from the dynamic tests. The effective stiffness is computed from the secant measured from peak to peak in each hysteresis loop. The equivalent viscous damping is computed from the area of the hysteresis loop. At each level of maximum strain there is a certain amount of softening between the first cycle and subsequent cycles. The effective stiffness and damping for the first and fifth cycles are given in Table 1. The reduction in the damping ratios in the higher strain cycles is a consequence of the strong hardening of the material when the strain exceeds 200%. The damping factor quoted is based on modelling the bearing as an elastic and linear viscous element. This model predicts that the energy dissipation is quadratic in the displacement

-94- -6- and the effective stiffness is independent of the displacements. In the actual component, the energy dissipation is not quadratic but varies roughly as displacement to the power 1.5, and the effective stiffness increases at the higher strains. Both factors act to reduce the damping factor. This reduction, however, is of very little significance to the response of a system using the isolators, since at these large strains it is futile to predict the dynamic response of the isolated structure by linear modelling. The important aspect of the bearing behavior is the energy dissipation which continues to increase while the strong hardening of the elastomer eliminates the possibility of resonant response. The relationship between the hysteresis loops generated at the various strain levels indicate it might be useful to use one model to analyze the response at the design level; for example, for strains that do not exceed 200%, and a second model for extreme events at strains above that. It should be emphasized, however, that the dynamic tests were carried out within a time frame of a few hours and that the large strain cycles followed many cycles at lower levels. Thus, the results at the larger strains were obtained from bearings that underwent many tests in rapid succession. The response of a bearing might be somewhat different if it were to be tested to ± 350% strain initially. Isolation bearings are generally used at pressure levels ranging from 5 to 7 MPa (700 to 1000 psi). Bearings with low shape factors and the range of height to width ratios that result from the selection of a low pressure tend to be somewhat sensitive to vertical load due to the fact that the vertical load is a significant fraction of the bearing buckling load. For bearings of the Bridgestone type, which have very large shape factors and a squat aspect ratio, the buckling load is so much larger than the design vertical load that the influence of vertical load on the horizontal characteristics through the stability of the bearing is negligible. However, the vertical pressure does play a role in the horizontal response and in these bearings the effect must be through an interaction between pressure and shear in the elastomer. This is a reasonable conjecture since for such large shape factors it has been shown [6] that the normal assumption of incompressible material behavior does not hold. Thus, the vertical load on the bearing produces a volume change in the material which could interact with the shear behavior to modify the horizontal response. The dynamic tests were carried out at four levels of vertical pressure, 0, 3.45, 6.90, 10.34 MPa (0, 500, 1000 and 1500 psi). The bearing stiffness at the various peak strains were computed using peak-to-peak measurements in the resulting hysteresis loops. The stiffnesses are given in Table 1. At smaller strain levels ofless than 200%, there is no identifiable effect of pressure on stiffness; above 200% the stiffness is smaller at the higher pressure but the effect is small and can be ignored. The pressure, however, has a very definite effect on the damping. It is obvious from the hysteresis loops that their area for fixed strain increases with increasing pressure, leading

-95- -7- to much larger damping factors. The damping factors for each pressure level and each peak strain level are given in Table 1. The damping values over the range of strains up to 200% varies very little with strain and averages 12% for 0 MPa (0 psi), 12% for 3.45 MPa (500 psi), 13.6% for 6.90 MPa (1000 psi) and 15% for 10.34 MPa (1500 psi). These are very high values for damping. This is especially when it is recalled that in the high-damping, natural-rubber compound bearings commonly used in the United States [7], the damping drops steadily with strain. The fact that the Bridge- stone compound retains its high values of damping over such a wide range of strain is a significant advance in rubber technology. Furthermore, the fact that the higher pressure can generate higher damping factors with no detrimental effects would encourage the use of higher bearing pressures. Since the horizontal natural frequency of an elastomeric isolation system is governed by the ratio of pressure to shear modulus, using a higher pressure could allow the use of a stiffer elastomer. Since it is generally easier to increase the damping in the rubber by increasing the stiffness, the result is an increase in damping both from the increased pressure and from the increased stiffness. Each bearing was loaded monotonically to failure at a rate of 6.35 cm/min. (2.5 in/min.). The response of the bearings during this test is shown in Fig. 2 which gives the shear stress as a function of shear strain to failure for the four bearings. The stress-strain curve for the bearing with zero-vertical load lies below the others but continues to a displacement which is larger than the others. The failure of this bearing was also different from the others in that the force rose continuously to the point of failure; whereas for the loaded bearings, the curve near failure was more rounded, increasingly so with increasing pressure. In the heaviest loaded bearing, the curve reached a maximum and began dropping before the rubber failed. In the bearing with 6.90 MPa (1000 psi) pressure, the peak load in the bearing developed at 507% strain, and the rubber failed at 516% strain. For the bearing with 10.34 MPa (1500 psi), the peak load developed at 473% strain and the rubber failed at 513% strain. The failure was instantaneous in the bearing with zero pressure; in the bearings it was more gra- dual. It is interesting to note that the displacement at failure exceeded the diameter of the bearing in the case of the bearings with 0 and 3.45 MPa (0 and 500 psi) vertical pressure. The results show that the stress-strain relation of the elastomer is trilinear. An initial stiffness of around 2.276 Mn/m (13 kips/in.) applies for a strain of up to around 15%, the bearing has a stiffness of 350 kN/m (2 kips/in.) up to a strain of around 250%, and the third segment of the force-displacement curve has a stiffness of 2.276 Mn/m (13 kips/in.) from 250% to failure. The implications of these Tesults for the analysis of buildings using this type of isolation system are several. The initial stiffness is significant for estimating the

-96- -8- response of the isolated building to wind load and ground borne vibrations from traffic and other sources of low-level vibration. For seismic loading at design levels the initial stiffness can be ignored and the system analyzed as a linear system with viscous damping. For an estimate of the response under earthquake loading beyond the design level, the system can be analyzed using a bilinear model with the second stiffness much larger than the first. The analysis will be nonlinear and a realistic modeling of the energy dissipation can be made by combining viscous damping and hysteretic damping. The strongly-stiffening character of the force-displacement curves also has implications for design. For example, for the bearing with 6.90 MPa (1000 psi) pressure the force level at 200% strain which could safely be considered as the design level is around 44.5 kN (10 kips) and at the maximum of the force-displacement curve it is 275.7 kN (62 kips). Thus, if a superstructure is designed just to be at yield level at the design strain of the isolators, the superstructure would have to be able to sustain a base shear of six times the yield level in order that the isolation system would fail before the superstructure would collapse. Although this is possible, it is highly unlikely. By using this kind of approach to isolation design, the bearing designer can be fairly confident that the isolation system will not be the weakest link. The failure mechanism of the bearings at all levels of pressure was tearing of the rubber. No evidence of bond failure was observed.

-97- -9-

Design Process for Elastomeric Bearings

The critical factor in the design of a bearing is the bond between the rubber and the steel shims. In the ideal case, the bonding compound should be sufficiently strong and the workmanship of the molding process reliable enough to insure that failure is always by tearing of the rubber and not delamination by bond failure. In the United States, concern about the quality of the bonding has meant that the maximum design shear strain has been 100% and this has led to bearing designs that are rather tall, and consequently somewhat unstable and prone to roll-out. Experience with Japanese designs of isolators such as those manufactured by Bridgestone Corporation which tend to use design shear strains of 200% and even in some cases 300%, suggests that a better proportioned isolator will result. It will be more squat and less subject to instability and roll-out, and in a sense, makes more efficient use of the highly intrinsic strength of the elastomer by making material failure the mechanism of failure rather than buckling or roll-out which are overall rather than local modes of failure. If the bond strength can be guaranteed, then the elastomer can be relied on to sustain shear strains of 450%-500%. The computation for roll-out suggests that roll-out of dowelled bearings or severe tension in bolted bearings will not develop if the displacement is less than 90% of the plan dimension. Accordingly, the diameter of the circular bearing should be about five times the rubber thickness. If the maximum shear strain is 200%, then the rubber thickness should be half the design displacement. The design process for a bearing with this configuration can be made fairly automatic in the following way. The design process for elastomeric bridge bearings is governed by a number of code specifications that reflect the fact that the loads to which the bearings are subjected are well-defined and happen on a regular basis. If these provisions were to be applied to elastomeric bearings for isolation, they would result in unnecessarily conservative designs. In the design of seismic isolation bearings for buildings, it has to be recognized that codes such as the Uniform Building Code (UBC) 1991 impose very severe seismic loading on the isolation system and that this loading may be interpreted as ultimate state loading and that the isolators should be designed reflecting this. The point is that the conservatism is already incorporated in the specified seismicity of the site and need not be further increased by over conservative design of the isolators, and further, that the extreme loads to which the isolator may be subjected will occur, if at all, no more than once or twice over the lifetime of the structure. The preliminary design of a bearing in an isolation system begins with the determination of the load to be carried by the bearing. In most buildings, the design load at each column (based on dead load plus fixed partitions, equipment, furniture, etc.) can vary quite widely. It will generally be necessary to minimize the number of different types of isolators, and thus, the first decision to be made will be how many different

-98- -10- bearing types to design. Once this decision is made, the design load for each bearing type can be selected to minimize the variation of load on that type. After the design load is selected, the design specifications will fix the following quantities:

fH • horizontal frequency or T« horizontal period

fv • vertical frequency Ymax " maximum permissible shear strain D m design displacement (from response spectrum or SEAOC formula) W « isolation load

Two safety factors will also be needed. The first is the safety factor against buckling: this should be based on dead load plus live load on the bearing. The second is the safety factor against roll-out which should be based on the minimum load on the bearing.

The design quantities to be selected are a or , tr t, n, G, and h, where

a >= plan dimension of a square bearing <3> = diameter of a round bearing

tr = total rubber thickness in the bearing / • thickness of individual layer n m number of layers G •= shear modulus of elastomer h « total height of the bearing br p » W/A = vertical pressure on bearing

In almost every case, a design spectrum will be a constant velocity spectrum and thus, at least for preliminary design, the design displacement D will vary linearly with the period T. We can begin then from the formula D - kl (1) where k with units of length per second will depend on the site, the soil, and the damping in the system. Next, the average load to be carried by each type of isolator can also be assumed to be specified. The maximum shear strain, y, in the isolator may be specified by code but it is possible that its choice may be left to the designer. With W, k, and y known, the choices available to the isolator designer are aspect ratio and the shear modulus of the elastomer. A compact design for the isolator will mandate that the diameter of will be related to the total elastomer thickness, f, through

<£ e \tr, with X being at least 4 and preferably around 5. The shear capacity of high- strength natural rubber compounds is at 500% and with an aspect ratio of 5, this can

-99- -11 - be developed while some overlap of the top and bottom still exists to carry the vertical load. The other design parameter of interest is the vertical pressure p. This is related to the other quantities through the fact that the period T is given by

Now

r ' Y so that with Eq. (1) we have

8 ys ' ' 8 ys which can be written in the form

(2)

The area A of the isolator is —O2 and thus 4

Jf_ k 4 r 7 g-yG-T (3) Since t. • — - — we have Y Y

This can be solved for T if the other quantities are specified. However, since 3> is related to T through . . XD TJcT 4> Xf . <=«(6) Y Y we can solve directly for the isolator diameter in the form

m Wk 4 , Y2 \ 3 2 Y g Y^ n k\ )

When O has been computed we can easily compute

-100- -12-

D = Y'r

and the base shear coefficient C «= ^— becomes by virtue of Eq. 2. P

The range of the parameters that appears in this sequence of equations is quite wide. The aspect range should be around 4 or 5 and the maximum shear strain could vary between 1 and 2. The variation in the shear modulus will depend to some extent on the choice of maximum shear strain, but will vary over a factor of 3 from minimum to maximum. At a shear strain of 2, the range of shear modulus can be from .035 to 1.03 MPa (50 to 150 psi). The solution of the design equation for isolator loads varying from 0.07 to 20.69 MPa (10 to 3000 kips) is shown in Fig. 3 with the corresponding pressures shown on the graph. It is clear that there is a wide range of possibilities for any value of W. If the pressure is too low, which would lead to high values of C, it is possible to increase G; and if too high, say above 13.79 MPa (2000 psi), it can be reduced by reducing G or increasing X. In fact, the manipulation of Eqs. 3 and 4 gives p explicitly as

In both this equation and that for <£, the quantities k and y come in as the ratio of one to the other. The value of k represents the seismicity of the site. A large value of k might represent a degree of conservatism in assessing the possible ground motion. On the other hand, the value of y represents the degree of conservatism assumed for the strength of the elastomer. If a large value of k is used to represent a conservative approach to the seismicity, a large value of y is appropriate to avoid compounding factors of safety. Bearings with a large aspect ratio k will be very stable. The critical buckling stress ocri{ is given [8] approximately by

where

Ps »

-101- -13-

and

h2 For moderate values of shape factor S - —t, say between 4 and 16, the bending

stiffness (EI)eff of a single pad is given by <£'W - f £< • "fr • t and 2 Ec s6GS

Taken together these give

4 tr with P~^f .£. 2 2 2 (2K)2 * (2K)2 k (2K)2 * ^

We have the safety factor against buckling, S.F., given by

s 2 2 2 SF ^^EiL.^L. • C *) - * > P 4 lr g Y Sk2 g

2 2 g At y tr Ag Y '

This leads to the useful result that once the design criteria X, k, y are specified, the only design quantity that appears in the formula for the safety factor is t, the thickness of an individual layer. The safety factor will be greater than 1 for values of t less than a certain thickness given by S.F. a 1 if 2 2 V2"K3 X * Ag Y2

-102- -14-

For X = 5, k - 4, y - 2, this gives f s 72.9 mm (2.83 in.). Since / is the thickness of an individual layer, which will be at most around 12.7 mm (0.5 in.), this indicates that the safety factor will be at least 6 and possibly more. This result verifies that the bearings designed by this approach will be very stable against buckling. The advantage of using this type of design method is that overall behavior of the bearing will not be the critical factor in defining the bearing capacity. The bearing capacity will be controlled solely by the strength of the elastomer and by the shear capacity of the steel-rubber bond. Both of these quantities can be verified and controlled using small test specimens, thus reducing the cost of verifying the bearing capacity.

-103- -15-

Conclusions

The results of the test program on the mechanical and failure characteristics of elastomer bearings has confirmed the extremely high quality that can be achieved for such isolators. They have been shown to be capable of extremely large shear strains before failure even under high levels of vertical pressure. Additionally, because a) the failure mechanism is only slightly affected by pressure, and b) the stiffness is unaffected by pressure while the damping is increased by pressure, these unique characteristics can be used in the design process, leading to the implementation of smaller isolators than those currently in use. The damping in the bearings was found to be in the range of 15% of critical damping. However, damping is a secondary issue in isolation. It is not generally realized that the effect of an isolation system of the elastomeric type is the result of the difference between the fixed-base frequency of the superstructure and the overall frequency of the isolated building. When the ratio between these is large the participation factor of the higher modes that involve structural deformation is low. This means that the displacement that results from the earthquake input is almost entirely in the isolation system. Also, if there is large energy in the earthquake input at these higher frequencies, this energy is not transmitted into the structure as it is in conventionally based structures. This effect is achieved even in the absence of damping. Damping is only needed to counteract the possibility of resonance at the isolation frequency. In fact, damping can be viewed as a contaminant of the isolation process The more damping there is in the isolation system - and especially if this damping is produced by nonlinear mechanisms such as mechanical dampers - the more energy leaks into the higher modes thus counteracting the beneficial effects of the isolation system. The search to find damping mechanisms to accompany isolation systems has been a misplaced effort. Much more important in designing practical isolation systems is the consideration that the elastomer hardens very strongly after a level of shear strain has been exceeded. In these tests, the stiffness increased by a factor of six beyond 250% shear strain. Thus, if 200% shear strain were taken as the nominal design level, at which level of base shear the superstructure is just at the yield point, then the base shear must be increased by at least a factor of six before the isolation system were to fail. This means that the failure mechanism for the entire structure will occur in the superstructure and not in the isolators. Conceptually, the collapse of an isolated structure is no different from that of a conventional structure with, however, the proviso that the level of earthquake impact that produces the failure must be much greater for the isolated structure owing to the large displacement capacity of the isolation system. The strong-hardening response of the system beyond the design level dictates that if the system exceeds this level, the period shortens significantly and the system becomes non-resonant.

-104- -16-

The test results have shown that properly designed and manufactured isolators for nuclear reactor applications can sustain levels of loading beyond any possible seismic input and demonstrate that failure of an isolation system cannot occur before failure of the isolated structure. Thus, the use of isolation can only have beneficial contributions to the protection of nuclear facilities and their internal piping and equipment.

-105- -17-

References

[1] Kelly, J.M, "Aseismic Base Isolation," Shock and Vibration Digest, 14, 5 (1982).

[2] Kelly, J.M. and Chalhoub, M.S., "Earthquake Simulator Testing of a Combined Sliding Bearing and Rubber Bearing Isolation System," Report No. UCB/EERC- 87/04, Earthquake Engineering Research Center, University of California at Berkeley, Berkeley, California (1990).

[3] Kelly, J.M. and Chalhoub, M.S., "Sliders and Tension Controlled Reinforced Bearings Combined for Earthquake Isolation," Journal of the International Asso- ciation for Earthquake Engineering, Wiley-Interscience, 19, 3, 333-358 (1990).

[4] Way., D.J. and Howard, J., "Seismic Rehabilitation of the Mackay School of Mines with Base Isolation," Earthquake Spectra, 6, 2, 297-308 (1990).

[5] Kelly, J.M., "Dynamic and Failure Characteristics of Bridgestone Bearings," Report No. UCB/EERC-91/04, Earthquake Engineering Research Center, Univer- sity of California, Berkeley (1991).

[6] Chalhoub, M.S. and Kelly, J.M., "Effect of Bulk Compressibility on the Stiffness of Cylindrical Base Isolation Bearings," International Journal of Solids and Structures, 26, 7, 743-760 (1990).

[7] Tarics, A.G., Way, D.J., and Kelly, J.M., "The Implementation of Base Isolation for the Foothills Communities Law and Justice Center," Technical Report, RTA, San Francisco (1984).

[8] Kelly, J.M. and Koh, C.-G., "Effects of Axial Load on Elastomeric Isolation Bearings," Report No. UCBIEERC-86112, Earthquake Engineering Research Center, University of California at Berkeley, Berkeley, California (1987).

-106- -18-

Table 1

Bridgestone Bearing Test Horizontal Shear Test

BEARING ID: HR03O-4 EERC ID: No. 3 VERTICAL PRESSURE: 0 PSI

File No. 901023.22 901023.23 901023.24

Y<*> 10% 25% 50% 75% 100% 100% 150% 200% 250% 200% 250% 300% 350%

%) (»« cy) 1158 9.22 7.28 5.89 5.14 4.86 4.53 4.41 4.62 3.28 3.84 4.5 4.49 Kips in (5th cy) Hil 8.75 6.54 5.45 4.67 4.61 3.94 3.80 3.74 3.13 3.48 3.52 3.62

P (wcy) 1225 12.24 12.01 11.95 12.09 13.18 12.17 11.39 9.97 13.64 10.02 8.74 7.89 (%)(5lhcy) 1339 13.14 12.06 12.09 13.86 12.01 12.70 11.61 10.61 13.22 10.47 9.4 8.4

WD (lstcy) 0 1 4 6 10 11 21 35 50 31 43 62 70 Kips-in (5th cy) 0 1 3 6 11 10 19 30 43 29 39 52 60

G (lstcy) 311 228 180 145 127 120 112 109 114 81 95 111 111 Ksi (5th cy) 277 216 162 135 115 114 97 94 93 77 86 87 90

BEARING ID: HR030-3 EERC ID: No. 2 VERTICAL PRESSURE: 500 PSI

File No. 901023.16 901023.17 901023.18

Y(=O 10% 25% 50% 75% 100% 100% 150% 200% 250% 200% 250% 300% 350%

p (lstcy) 10.22 12.91 12.6 12.15 12.8 1339 12.85 1111 10.59 14.2 10.68 9.06 8.42 (%)(5lhcy) 11.21 11.65 11.87 12.77 14.03 13.56 13.55 1189 11.88 14,49 113 9.98 9.06

WD (lstcy) 0 1 4 7 11 12 24 38 53 33 44 63 77 Kips-in (5th cy) 0 1 3 7 11 11 22 34 48 31 42 56 67

G (lstcy) 324 253 192 159 137 130 121 114 114 84 94 110 111 Ksi (5th cy) 314 232 175 143 123 122 103 93 92 79 85 89 89

y (*): Shear Strain (%) K(tfn : Effective Stiffness (kips in) p : Equivalent Viscous Damping (%) WD : Energy Dissipation (kips-in) G : Shear Modulus (psi)

-107- - 19-

Table 1 (Cont.)

BEARING ID: HR030-2 BERC ID: No. 1 VERTICAL PRESSURE: 1000 PSI

File No. 901023.04 901023.05 901023.06

Y(±) 10% 25% 50% 75% 100% 100% 150% 200% 250% 200% 250% 300% 350% Kw) 12.66 9.61 736 5.89 5.08 4.87 4.58 4.53 4.83 331 3.91 4.65 4.75 Kips m (5th cy) 1Z71 8.6 632 5.41 4.72 4.55 3.9 3.77 3.84 3.13 3.54 3.7 . 3.71

P (1'tcy) 13.71 12.99 12.48 13.24 14.01 15.45 13.86 1X24 10.57 15.68 10.99 9.11 8.29 (%)(5thcy) 13.73 11.08 14.07 15.1 14.91 15.03 14.66 13.02 1131 15.25 11.74 10.18 9.46

WD (lstcy) 0 1 4 7 11 13 24 38 55 50 45 64 78 Kips-in (5th cy) 0 1 3 7 11 12 22 34 48 31 43 57 70

G (lstcy) 313 237 182 145 125 120 113 112 119 82 97 115 117 Ksi (5th cy) 314 213 156 134 117 113 96 93 96 77 87 92 92

BEARING ID: HRO3O-S EERC ID: No. 4 VERTICAL PRESSURE: 1500 PSI

File No. 901023.10 901023.11 901023.12

Y(-) 10% 25% 50% 75% 100% 100% 150% 200% 250% 200% 250% 300% 350%

15.76 11.04 8.56 6.13 5.09 4.66 438 4.28 433 2.97 3.52 4.15 4.2 Kips in (5th cy) 8.45 632 5.08 4.48 436 3.74 3.46 3.44 2.77 3.14 3.23 3.19

P (l«cy) 12.71 14.52 1434 14.84 16.39 17.13 15.77 14.06 12.81 18.58 13.24 10.85 9.92 (%)(5thcy) 14.64 14.75 14.52 15.97 16.52 16.16 16.96 15.46 13.67 17.61 14,11 12.4 11.59

WD (151 cy) 0 1 4 8 13 13 26 41 60 37 50 70 85 Kips-in (5lh cy) 0 1 4 7 12 12 24 36 50 33 48 62 75

G (lstcy) 389 273 211 151 126 115 108 106 107 73 87 103 104 Ksi (5th ey) 279 209 156 125 111 108 92 86 85 68 78 80 79

1 ( x ): Shear Strain (%) K(,ff) '• Effective Stiffness (kips in) p : Equivalent Viscous Damping (%) wD : Energy Dissipation (kips-in) G : Shear Modulus (psi)

-108- HR030Z

Horizontal Characteristics

I o I RUBBER 2x22=44 STEEL .8x21=16.8

FOR Earthquake Research Engineering Center DWG NO, HR03QZ-1 (CUE 1990. 8.10 APpmmo 1 CHtCtlD nmsio IPPHOIlt CNICXf0 FLANGE SS4KJIS G 3101) STEEL PLATE SPCCUIS G 3H1 ) 21 0.81 RUBBER KL401 22 21 BRIDGESTONE e»n • o. I*K( j QUA*rirr TOKYO JAPAN

Fig. 1: Design Details for Test Bearings BRIDGESTONE-BEARING TEST SHEAR STRESS-STRAIN DIAGRAMS

BEARING ID: HR030-4 HR030-3 HR030-2 HR030-5 VERT PRESSURE: 0 psi 500 psi 1000 psi 1500 psi STRAIN RANGE : 561 % 550 % 507 % 473 % Shear Stress/Shear Strain 1.2

1.0 s h e a 0.8 r S t 0.6 r e s s 0.4 K S I 0.2

0.0 100 200 300 400 500 600 700

Shear Strain %

Fig. 2: Stress-Strain Curves for Bearings in Failure Tests -22-

k = 4 X-5 V-

/ / / / b / 20 - 2 - 8 O) / / / / CO S- 'iliii ^) o /;

i / i / /$ k

/ '<; f7 * 1 1 i 10 V / / i - 1 4 i 9 / // - 0.9 3.6 i i 8 - 0.8 3.2 i

10 50 100 500 3000

log10(Axial Load) kips

Fig. 3: Design Chart for Bearing Diameter as A Function of Axial Load

-in- XA0055381

•"S> Some Progress on Seismic Isolation Technology in Bulding Structure in China • _ .^w. .»,.

Luan Lin Mechanics Group, FBR, CIAE

Isolation technology for equipment has been developed considerably in China. Appropriate codes and design manuals have been used by China designers. In the recent ten years, the tests, analysis methods and application of seismic isolation technology, especially in civil engineering, have been in continuous progress.

For any earthquake motion, the response of a building structure is determined by the amount of energy 'fed in and dissipated. So there are two basic objects in seismic isolation design. The first, introducing sufficient extra damping into structure to reduce dynamic amplification. The second, detuning the structural dynamic properties with respect to the frequency content of input ground motions to reduce energy fed in. To achieve these objects, various kinds of vibration control systems have been developed, such as: base isolation system, frictional brake system, energy absorber, etc. One most important of them is the base isolator, which has been recognized as an excellent protecting technique against the effects of earthquakes.

By using hysteretic or frictional base isolator, the building will act as a free standing structure under the excitation of minor earthquakes and winds. As earthquake excitation increases, the stiffness of isolator soften tends to act as a mechanical fuse, protecting the super—structure from damage. The potential reduction in the seismic forces can consequently reduce the cost and construction time. In China, this technique has been used in the seismic.

When earthquake occured in Tang Shan in 1976, masonry block buildings, in which the reinforcement was not carried through to the foundation, performed better than those in which the reinforcement was carried through to that. In a structure which performed well in the earthquake, a horizontal crack was observed at the foot of wall and a

-112- residual displacement of about 6 cm occured. As a result of these observations, the approach adopted in China is to set a separation layer under the floor beams above a wall foundation. A thin layer of special screen sand is laid on the sliding surface and the building is constructed on it [1]. Four demonstration buildings have been built in China using this technique [2]. Graphite has also been used as a sliding material and a building has been constructed with such sliding base in Xi-Chang city, Sichuan province. Some vibration tests and analyses of masonry models have been made [3]. Graphite and polytetraflourethelene are selected as sliding materials. Some model experiments of base brick walls have been done by means of sand slide measure [4], Some model tests and analyses of reinforce concrete frame structures with sliding base isolation measure has been made. Steel plated laminated rubber pad base isolators also have been investigated in China. Some of them have been used as the bearings of highway bridges. Lead plug systems has been investigated and used as the base isolators of high voltage breakers in the region of high seismicity of China.

Besides base isolators, frictional dampers have been used in the cross bracing of frame structures and in the construction of some buildings. Many experiments and analyses for high building models by using vertical frictional joints on partitioning walls in R. C. frame structures has been made.

The optimal design of such isolation systems is depended on the development of new materials and new technologies. Also it is necessary to develop analysis methods to design such systems. Owing to the nonlinearity of these systems, this is a difficult task. In Institute of Me- chanics of China Academy of Sciences Prof. Tian and his cooperators have done some research work in this field. In 1982, they suggested a method to analyse a large structure with artificial damping [5] and a method to calculate the frictional base isolation problem. In 1984, they developed a method to calculate the stochastic response of a structure with frictional joints [6]. In 1986, Prof. Tian in cooperation with Hong Kong University made some analyses and did experiments on infilled frame models with sliding bases [7] [8]. In 1987 Prof. Tian suggested a

-113- method for the vibration isolation of a super-structure of an ocean platform [9-13]. Model test results showed that the peak response can be reduced about 5% by using a sliding base isolation with heavy damping. In 1989-1990, Prof. Tian, in cooperation with Hong Kong Polytechnic, made some analyses and did experiments on frame models with nonlinear isolators. They have developed a method for the dynamic analysis of large structure with locally strong nonlinear elements by applying the impedance method and incremental harmonic balance method. Through experiments and analyses, they suggested that the wire-rope isolator may be used as a mechanical fuse in seismic design instead of the dry friction device. It has the advantages of easy to repair and easy to be controlled during and after an earthquake excitation.

Seismic isolation technology has not been applied to construction of nuclear power plants yet. The basic reason is that reactor can be, generally, placed in low seismicity area. The second one is that introducing seismic isolation technology into reactor raises construction cost. It was suggested to use this technology for Guangdong NPP and we intended also to have a demostration seismic isolation system at this site. But it was estimated the cost will be increased for more than 20 million dollars. Therefore, this intention has not been realized.

There is a plan of developing Fast Reactor Technology under the framework State High-Technology Program in China [14]. The China FBR Research Centre had its foundation stone laying ceremony in Beij- ing in Nov. 1990. As a first step, it is planed to design and construct a 25 MWe (65MWth) Experimental Fast Reactor, The conceptual design was completed just at the end of last year. We are now setting about the investigation of seismic isolation technology for fast reactor. We would like to cooperate with scientists of foreign countries in this area.

Acknowledgement The author wishes to express his appreciation to Prof. Q. L. Tian of Institute of Mechanics of China Academy of Sciences, Prof. M. Xu of CIAE. They offered their papers and manuscripts. The author wishes also to thank Prof. Z. Y. Zhang who critically read this manuscript.

-114- Reference [I] L. Li, "Base Isolation "Measures in Aseismic Structures", Proc. U.S.-P.R.C. Bilateral Workshop on Eerthquake Engineering. Harbin, China 1982. [2] L. Li, "Base Isolation Measures for Aseismic Buildings in China", Proc. 8th World Conf. on Eerthquake Engineering. San Fransisco, California, V. 6. Pg 781-798, 1984. [3] G. B. Shui, Report of Xian Building Developing Corporation. [4] Y. L. Lou & R. F. Wu, Report of Dalian University, Departement of Engineering Mechanics. [5] Q. L. Tian, et al. , "Dynamic Analysis of a Large Structure with Artificial Damping", Shock and Vibration Bulletin. USA 52 pt. 4 1982. [6] Q. L. Tian, et al. , "Stochastic Dynamic Analysis of a Structure with Frictional Joints", Shock and Vibration Bulletin. USA 54 pt. 3 1984. [7] Q. L. Tian, et al. , "Dynamic Response of Infilled frames Incoporating a Sliding Base Device", Proc. Instn. Civ. Engng. pt. 2 1986. 81 Mar. [8] Q. L. Tian, et al. , "Dynamic Analysis of a Structure with Sliding Base", Shock and Vibration Bulletin. USA 57 pt. 3 1987. [9] Q. L. Tian, et al. , "Vibration Isolation for the Super-structure of An Ocean Platform Structure", Proc. of the National Conf. of Earthquake Engng 1987 Nov. Harbin, China. [10] Q. L. Tian, et al. , "Steady Dynamic Response Analysis of a Large Structure with Frictional Joints", Proc. of the 1st ASIA Vib. Conf. 1989 Dec. Shenzhen, China. [II] Q. L. Tian, et al. , "Dynamic Analysis of a Structure with Frictional Base Isolator", Proc. I.C.V.P.E. 90 1990 June. Wuhan, China. [12] Q. L. Tian, et al. , "Vibration Control of Frame Structure by Using Wire-rope Isolator", J. of Mechine Strength, 1991. No. 4. [13] Q. L. Tian, et al. , "The Dynamic Analysis of Structure with Dry Frictional Joints", J. of Engng Mech. 1991. Vol 9, No. 3. [14] M. Xu, "The Status of Fast Reactor Technology Development in China", The 24th Annual Meeting of IWGFR, Tsuruga,

-115- XA0055382

Recent Results of Seismic Isolation Study in CRIEPI - Tests an Seismic Isolation Elements, Vibration tests and Observations -

Katsuhiko ISHIDA Dr. Eng., Hiroo SHIOJIRI Dr.Eng. Taiji MAZDA, Yasuki OHTORI Abiko Research Laboratory, Central Research Institute of Electric Power Industry.

Sakae AOYAGI, Dr.Ing., Head Office, Central Research Institute of Electric Power Industry.

L Introduction Seismic isolation is expected to be effective in raiding reliability during earthquake, reducing cost, enlarging siting and promoting the design standardization of electric power facilities. In Japan,seismic isolation has been applied to several office buildings and residences. However, it is considered that more research and demonstration are needed to verify the reliability and effectiveness of seismic isolation for important facilities such as Fast Breeder Reactors. In Central Research Institute of Electric Power Industry (CRIEPI),research programs are being conducted In the preliminary study of isolation concepts for FBR, the horizontal base isolation of buildings were selected as the subject of detailed investigation. The laminated rubber bearings were considered to be best suited for the isolation system. Tests on large scale models of rubber bearing and vibration tests of base isolation system have been conducted under contract with Ministry of International Trade and Industry. The earthquake response observation of isolated buildings have been conducted in the joint studies with the companies which constructed the buildings. In this paper, the results of the researches are described.

2. Static Test of large-Scale Laminated Rubber Bearings One of most critical issues in the evaluation of the reliability and effectiveness of the seismic isolation system is to clarify the characteristics of seismic isolation elements. They may be size dependent,and the size of the model should be as close to the prototype as possible. The limit state of rubber bearing have to be elucidated for the evaluation or safety margin of the seismic isolation system, but the failure tests results of rubber bearing were limited. It is intended to conduct tests on large-scale seiHtnic isolation elementB for FBR so as to demonstrate the safety against seismic load. The tests on the elements were conducted according to the following procedure. 0 A survey of past tests on seismic isolation elements was made to select necessary test

-116- (p A survey of past tests on seismic isolation elements was made to select necessary test items and to develop a test program. ® A specification of a laminated elastomer bearing for base isolation of FBR building was examined. (§> A static two-dimensional fracture test equipment which is capable of breaking large scale (200t class) bearing was fabricated. <§> Scale law was confirmed by conducting tests on full-sale and reduced scale models using the equipment and by testing cut-out samples from the elastomers of different scales. © The failure loads and displacements were determined from the test results of reduced-scale models using the test equipment, while load-displaoement pftiffHnryftrip? for design loads are determined using full-scale models.

(1) Test specimens Natural rubber bearing(NRB), lead rubber bearing(LRB) and high damping rubber bearing(HRB) are regarded as most promising for base isolation of buildings. Pull-scale models and two kinds of reduced scale models were submitted for the test Table. 2-1 shows design specification of each full-scale model and Fig. 2-1 shows the detail of the models. Table2-1 Table i Design Specifications of Laminated Rubber Bearing

t NRB LRB HRB Par»att«r ^*"* *'^^^^>^ Uasiai *«i«bl fo (loaf) 5 00 5 0 0 500 lateral frt«. Hor. fH 0.5 0. 5 0. 5 (Hi) Utl. ft t 0 to to 6.04" Sori&f Const. Hor. Kll 5.0 4 6.04' <4.*e>"" (tonf/oi) tart. KV a o 5 7 (057 t 05 7 ftorisontal ftisp. liaair SO 5 0 to Inikioi 10 0 1 00 1 00 Charactcrliclci liatiaator t I' s :•• Slittr Strasitt Q4 (lonf) (S I)1"

N HB:lilin! l»M«r l»rl>i * Unt »ttil» .r rrtk*r lli.H LRB :Uid I.Hnr i*irU< • » U«r itnla *t rrttor lttl(t»»lc>

H R B :Hi«h «>a»l>t l.»k.r lMrl>i * • * llmr itr.U #f niUr lHKltitlc)

•lutltr

MRB LRB HRB hrmttr ' -^^.^ •iuattr (•) i too 1(00 1 4 t 0 miikt (•) 44 0 «* 0 « i 0 ItlckMU «f liMtr Unl <•) 11.5 t « Ik »f liMtr (kMti 1 * IS S 1 Ikiekatit «I IIMI H«l« (a) 4.S ».« t.t h or ttMi n«t.« 1 t J 4 i 0 Hardaua Njm af labk«r 4 0 4 0 • 0 Fig. 2.1 Details of Laminated Rubber Bearing

-117- (2) Static two-dimensional fracture test equipment The static two-dimensional fracture test equipment is designed to grasp the lead-displacement relationship and deformation capacity, etc in detail It is among the vcrld largest static test equipment capable of loading in horizontal and vertical directions at the same time. The equipment Is composed of X-axis and Y-axis actuators giving static lead/displacement in the shearing direction and the compreasive/tensile direction on a bearing and a slip table attaching the bearing guiding each directional travel and reaction moment It is high pertdraance equipment minimizing mechanical friction and rotation by large capacity static pressure bearings providing smooth traveL The equipment allows vertical load of reactor etoxa^re born by the bearing as well as vertical load and horizontal defonnation generated on the bearingduring earthquaketo be reproduced at the same time.

Photo 2-1 Outline of Facility Specification of Equipment Horizontal Vertical Load ±600 tonf ±600 tonf Stroke ±600 mm ±350 mm Velocity 0.5 cm/sec 0.5 cm/sec

-118- (3) Test method 1) Horizontal stiffness test The objectives of these tests are to estimate the horizontal stiffness and damping of the bearing and to confirm the similarity between a full-scale model and reduced scale models. The tests were performed under the condition of low frequency cyclic loading. (less than 0.01Hz) Four cycles of sinusoidal horizontal displacement were applied under constant vertical load The amplitude of horizontal displacement varied from shear strain of rubber ± 25% to shear strain of rubber ± 200%. And the amplitude of vertical loads varied from -20% to •200% of design vertical load, and the record of third cycle was adopted as basic data.

lor. UriP

Kl •

W- — Kl (HI1 h« ••r. Hit. I hH- 77w

I ) IKI

Hor.load F> f«M-0d*

Fig. 2.2 Definition of Characteristics

2) Vertical stiffness test The objectives of these tests are to evaluate the vertical stiffness and damping of the bearing and to investigate the similarity between a full-scale model and reduced scale models. The tests were performed under the condition of low frequency cyclic loading in the same way as the horizontal stiffness tests. In the first test, four cycles of sinusoidal

-119- vertical load were applied superposed on initial static vertical load with shear strain kept constant The amplitude of sinusoidal vertical load was ± 50% of design vertical load. The initial vertical loads varied from 50% to 150% of the design vertical load, and the constant shear strain,

3) Ultimate Test The objectives of these tests are to evaluate the strength of the bearings and deformation characteristics around the breaking paint and to oonfinn the similarity law of two reduced scale models around the breaking point. At first, four cycles of sinusoidal horizontal displacement was applied horizontally to

reduced scale models under design vertical load Po. The amplitudes of displacement were ± 300%and ± 400% in shear strain of rubber. After that,horizontal displacement was

applied under design vertical load Pc up to break. In addition to these tests,horizontal breaking tests under various vertical load and vertical breaking tests with some offset displacement were conducted using LRB. The vertical load P was varied from +100%(tension) to -1000% (compression),and the offset displacement was 0% and 200% in shear strain of rubber.

(3) Test results 1) Horizontal stiffness test Fig. 2-3 shows the horizontal load-horizontal displacement curve of each bearing, Fig. 2-4 shows the relationship between the normalized horizontal spring constant and shear strain of rubber. From these results, natural rubber bearing were determined to behave almost lineally within shear strain of rubber 200%. The horizontal spring constant obtained by the tests agree approximately with the design value. Dependency on shear strain of rubber and vertical load are hardly recognized. Horizontal load-displacement relationship of lead rubber bearing can be approximated by bi-linear modeL The area enclosed by the hysteresis loop of the lead rubber bearing is larger than that of natural rubber bearing due to the absorption of energy by the lead plug. The horizontal spring constant of LRB depends on shear strain, and decreases as shear strain increases, but the dependency of the constant on vertical load is hardly recognized. In the case of high damping rubber bearings,the spring constant agree approximately with the design value over shear strain 100%. Deformation characteristics of the high damping rubber bearing has remarkable dependency on shear strain, as shown in Fig. 2-3. Horizontal spring constant decreases as shear strain increases. Furthermore, horizontal spring constant of high damping rubber bearing changes corresponding to the maximum

-120- shear strain experienced.

2) Vertical stiffness test Fig. 2-5 shows the relationship between vertical load and vertical displacement, and Fig. 2-6 shows the relationship between normalized vertical spring constant and sheer strain of rubber. From these results, each bearing was determined to behave almost lineally, and it was proved that the vertical spring constant agrees approximately with the design value when there is no shear strain. However, the spring constant of each bearing decreases as shear strain increases, and is about 50% of the design value at shear strain of 200%.

3) Ultimate test Fig. 2-7 shows the horizontal hysteresis curve of both reduced scale models with under design and over design loading condition. As the test results show, the effect of hardening appears clearly at shear strain of rubber above 300%. Fig. 2-8 shows the relationship

between shear stress and shear strain of rubber under design vertical load Po up to the breaking paint . From these results, it is clear that the fracture characteristics of two reduced scale models agree well with each other. Consequently, the validity of simlarity low was confirmed. The fracture characteristics of natural rubber bearing and lead rubber bearing are similar with each other, and the effect of lead plug isn't recognized. Breaking point of both bearings are approximately between shear strain of 400% and 500%, and between shear stress of 40kgf/cm- and 70 kgf/cm-. Breaking paint of high damping rubber bearing are approximately between shear strain of 450% and 550% ,and about shear stress of 60kgf/cm;. Fig. 2-9 shows shear strain coresponding to horizontal breaking point under varied vertical compressive or tensile stress o. Fig.2-10 shows tensile strain coresponding to vertical breaking point under offset shear strain of rubber y. The dependency of horizontal breaking point on vertical compressive or tensile stress are hardly recognized in these loading condition.

(4) Summary of results The results obtained by tests may be summarized as follows. 1) Static characteristics of full-scale laminated rubber bearings under the condition of ordinary temperature were made clear. These characteristics agreed approximately with the value calculated by design formulas. 2) Fracture characteristics under design vertical load were clarified by the tests using reduced scale models. Consequently, validity of similarity low was confirmed.

-121- 3) Fracture characteristics of reduced ecale LRB model under varied vertical load clarified. was

see soo 200* toe J I lee ioo I e • -too • I Vtrllut MftwtM) I ttrtiul UtACawtaat) -100 K: tui$> *anieal tmi infauraa -JOOl -J00 -400 -2 0 0 200 400 *ar. tita.to tar. »lta.(aa) • )«•• Fig. 2.3 Horizontal Hysteresis curve of Full-Scale Models

l.S 0 «r.i. ••• | 2.0 \ LS. *——» '• Sl.O 1 1 6 M ,n Jos • H-f. > o o.s

0 SOO SO 100 ISO 90 TOO 1 SO 100 Stoa' Strain af tiiM

100 too COO V A 0* 00* 200* i 1 on Jo* —e 0*— 00* soo soo SOK'IOOK ! 1 f 400 / / / J00 1 300 — _J y ]seo / JA- y I 2 00 J200 — — A y y 100 1 00 I 00 • .. OHaaia. (-) «trt. »ii>. (•») «trt. tiw. (•»> >> MX* I) Lilt •I HR« Fig. 2.5 Vertical Load-Vertical Disp. curve of Full-Scale Models

l.S l.S l.S 1 >*^. '1 —__ 41.0 •—«,

, •—1 I ~~—— jo.s Jos

SO 100 ISO 200 SO 100 ISO 100 ioo iso see Skt" Strait •( Mtar r.tt) OffMt «»ir Itraia •( Uttr t.(J) Mlaat Saatr Strata af Maar T.OD IUII IIMI Fig. 2.6 Relationship between Normaalized kH and Offset Shear Strain of Rubber 7 -122- 1*0 JOOJ-

-100

-2S0 -ISO 100 400 (00 -•00-400-tOO 0 tOO 400 COO -Voo-400-JOC tar. •(••.(•) toe 4oe «oo •>kM Fig. 2.7 Horizontal Hysteresis Curve of both Reduced Scale Models

100 1 1 1 1 1 .100 — l/». It M.«.l l..- — 1/1. It «•«•! • 0 fe / 40 ho i o 100 200 100 400 S00 tOO 10 0 2 06 TO 0 4 0 100 200 100 400 800 000 Vx«r Strain ll • itnlaaf Mtar rOl f tuWff TU) Strait •( bkktr rd] I) MM* *t Fig. 2.8 Relationship Between Shear Stress x and Shear Strain of Rubber v

ZP so f.: Stliin Vtrtiol

Po 25

0 100 200 300 400 500 -25 -p • s Stair tirtin >f b*hr T(8 -2P. -50 » • 400 -3P -75 •J -4P, -1 00 o ~ 200 -5P • I -125 •2 100 -6Pc -150 1 0 0 1 00 200 300 400 500 -7 P. -175 OffHt «*••' ttriln >f Mttr T.W -200 -BP Fig. 4,-10 Tensile strain of rubber e in vertical breaking point under offset -9P. - -225 shear strain of rubber y

-10P. -250

Fig. 2.9 Shear Strain of Rubber 7 in Horizontal Breaking Point under Varied Vertical Stress a

-123- 3. tynamic Test of Small Laminated Rubber Bearing Models The effects of loading rate and loading path were examined using smaller scale models. (l)Test Specimen 1/15 scale models of NRB, LRB and HRB were used for the test (Fig3-1) Similarity law is shown in Table 3-1. In the case of NRB and LRB, the similarity conditions are perfectly satisfied with respect not only to mechanical properties but also to geometrical details. In the case of HRB, the geometrical similarity conditions are not satisfied because in its design more stress was placed upon the similarity of mechanical properties. Table 3-1 Similarity law

Prototyp* Model

L»nghi 1 1/15 Force 1 I/IS' Strtss 1 1 Strain 1 1

U^£^A1U I 1 1 LU

Fig. 3.1 Detail of specimens

(2) Test Equipment Fig. 3-2 shows the test equipment. The specimen was mounted on a table which can slide horizontally. The upper flange of the specimen was fixed to a loading block which can move vertically but not rotate. Dynamic actuators were attached to the table and the block. Load cells were arranged so that measurements may not include friction force of the equipment.

Fig. 3-2 Testing apparatus -124- (3) Test Method A loading scheme was prepared to investigate the large-deforaation hysteretic behavior under various combinations of ahear and axial forces, and is shown in Table 3-2, where the following definitions are introduced. Axial strain t = 6 v/h Axial stress o = P /A

Shear strain y = 6 H/h Shear strain x = Q /A

where 6 v is the axial displacement, 6 « is the lateral displacement, P is the axial force, Q is the shear force, h is the total rubber thickness, and A is the cross sectional area (excluding the rubber cover and a lead plug). The loading program consists of tensile tests and shear tests. Most of the specimens were loaded statically and cyclically, while some of the specimens were loaded dynamically and/or monotonically to evaluate the effect of strain rate and cyclic loading. In the case of the Table 3-2 Loading scheme

•>ENSILE TEST SHEAR TCST CTMMH ROW MM. 1— - (WTt 4 UMNI CAlt cue OK wm r fwnvw 7 1

• - MK/te UIHJIUIl,

• •OOT8MC * - amum. Ul^l^B.

• CVCLJC - • .•INt CVCUC

t.UHi cvcuc - •.•IN Crcuc

W • ••IHJ CYCUC • - MlHi cvcuc

m Ml* CVQJC % - MW crcuc

w •••IK* Crtuc M. - MlHi crcuc

••••Ha crcuc

- H M". cvcuc

H - HI t.«IM. crcuc

tensile tests, the offset shear strain was selected as a test variable. In the case of the shear tests, the axial stress or the axial strain, which was kept constant during shear loading in each test, was selected as another test variable. A half-cyclic sinusoidal wave of gradually increasing amplitude was used as a tensile strain control signal for the tensile test. On the other hand, a sinusoidal wave of gradually increasing amplitude was used as a shear strain control signal for the shear test In the case of the monotonic loading, specimens were loaded at a constant strain rate. The strain rate of dynamic tests was determined so as to be roughly equal to that of the prototype rubber bearing. In other words, the scale factor for strain rates is the unity for the reduced scale model

(4) Test Results l)Relationship between Lstress and strain, and effect of test variables The tensile stress - strain curves under pure tension shown in Fig.3-3 represents

-125- the effects of strain rate and cyclic loading. Fig. 3-4 compares tensile stress - strain envelopes obtained at the different offset shear strain levela likewise, the shear stress - strain curves under the design axial stress shown in Fig. 3-5 represents the effects of strain rate and cyclic loading. Fig. 3-6 compares shear stress - strain envelopes obtained at the different axial stress or strain levela

Shear stress-strain relationship (See Fig. 3-5): Each type of specimen shows a low shear stiffness within about 200% shear strain and hardening phenomenon beyond this strain leveL The hardening slopes of NRB and LRB are rather steep, while that of HRB is relatively gentle <§) Effect of axial stress and strain on shear behavior (See Fig. 3-6) : In most cases, shear stress-strain curves of each type of bearings are very close to one another in spite of a variety of axial stress and strain condLtiona In the case where a high tensile stress or strain is imposed, however, a little higher stiffness and a little earlier failure can be seen compared to the other cases. © Effect of strain rate (See Fig. 3-3 and Fig. 3-5): The stress- strain curves and the failure limits of NRB and LRB are almost independent of the strain rate. On the other hand, those of HRB depend greatly on strain rate. In this case both stress and failure strain level increases at high strain rate.

-126- y*% etnuaw MONOTtMC tnts MONOtOMC tntieetcuc .••«

M MC ) LRB (c)HRB Fig. 3-3 Tensile stress - strain curves (the effect of strain rate and cyclic loading)

a r»»PK m y a

(MM m m m m m TCNM.t tntUN ( I',) (a) NRB (b)LRB fc)HRB Fig. 3-4 TensJte stress - strain envelopes (the effect of offset shear strain)

(a) NRB

-127- (D Effect of cyclic loading (See Fig. 3-3 and Fig. 3-5): Regardless of the specimen type, the cyclic loading decreases stress and increases failure strain level to some extent. 2) Failure limit ~~ The maximum •Sresses and the maximum strains are plotted in an axial stress- shear strain p^e. All the specimens is not necessarily fail at the maximum points, but the failure paints were not so far from the maximum points. The characteristics of these plots can be described as follows : At high tensile stresses, the maximum tensile stress gradually decreases as the maximum shear strain increases. In the range from small tensile stress to relatively large canpressLve stress, however, the maximum shear strain level remains almost constant. In this range, the maximum shear strain level ranges from 500X to 600% regardless of specimen type. In most cases, the failure occurred in a rubber layer near the flange, not at an insert plate surface 3) Comparison between NRB and LRB LRB has the same configuration as NRB except for a lead plug. Comparison in the stress - strain relation and failure limits are shown in 9 rof h* These graphs demonstrate that there is no noticeable difference between NRB and LRB. It should be noted that there is no contribution of a lead plug to tensile behavior and only slight contribution to large - deformation shear behavior. Furthermore, a lead plug has no bad influence on failure

(5) Summary of Results Three different types of rubber bearing - natural rubber bearing (NRB), lead-rubber bearing (LRB), and high-damping rubber bearing (HRB) - were tested up to failure undo? various combinations of axial and shear forces. The test results are summarized as follows: 1) Offset shear strain greatly affects tensile behavior and failure. 2) Axial stress or strain does not so greatly affect shear behavior and failure. 3) The characteristics of NRB and LRB are almost independent of strain rate, but those of HRB depend greatly on it. 4) Cyclic loading affects large-deformation behavior and failure to some extent. 5) Failure limits which include the interaction between axial stress and shear strain were quantitatively evaluated. 6) A lead plug of LRB has no bad influence on failure.

4. Shaking Table Test on Base Isolated FBR Plant Model (1) Shaking Table Test under Design Base Seismic Motin l)Outline of Test Model Shaking table tests were conducted to examine the performance of seismic isolation systemes under design base seismic motions. Photo 4-1 shows the experimental model which represents the dynamic characteristics of the reactor building for FBR. The model was a

-128- 4.1 Experimental Model

3-story steel frame with a mass of 20 ton, and supported by nine laminated rubber bearings. Similatity law and design specifications are shown in Table 4-1 and 4-2. The model was designed so that the axial stress of the laminated rubber bearing and the acceleration of ground motions were equal to those of the prototype. Nine bearings were installed under the columns of the superstructure. Three kinds of rubber bearings, Le., natural rubber bearing (NRB), lead rubber bearing (LRB) and high-damping rubber bearing (HRB) were used in the test. Four elasto-plastic steel dampers were used along with NBR to absorb the energy of building motions. Each bearing is a 1/15 scaled model of the prototype bearing. Table 4-3 shows the types of tests. The tentative design wave (NS.EW.UD), El Centro 1940 (NS,EW,UD) and Hachinohe 1968 NS were used for the input ground motion. These waves were reduced in time to 1/ J~ 15 according to the similarity law. The acceleration on each floor, horizontal and vertical displacements between the shaking table and lst floor of the superstructure, and the shearing and axial force acting on each bearing were measured. The shearing and axial force were measured by component force transducers and leveling devices were installed under component force transducers to equalize the initial loads of the nine bearing as shown in Photo 4-2. Table.4-1 Similitude

Scale of Similitude Displacement L L./L.-X- 15 Acceleration o a,/a.« 1 Frequency f f ,/f .-X*"1'-0.258 Stress a o,/am=1 Strain T T./T.-l Force F F./F.^X* «225

-129- Table 4-2 Design Specifications

Steel temper hbbtr Burin KB 1*8 m loading WriBhl 2.» tf Yield force 0.19 tf Horizontal ferine Const. 0.336 tf/oa Yield Oitelactrant O.Krm Itatural Frequency l.MHi OiantUr tie* QUaster 107 en Kan Ungth 97 n» Rubber layers O.*0mx» i.JtmxU

Photo 4.2 Isolator and Component force Transducer Table 4-3 Type of Test tapliwle lou

tofora'ahakim Utel last Static Cyclic Loadiat i?5r»is.o so.oa» J5.0 JO.0 45.0 S2.ta* •fur ahakia* tabU Uat mu acoalentioo: Kudos Vtve Shaking VhiU loiae UGal Oaitca lue brtoquake: tanuti»e taiin ***W II Castro 1940 (IS N UP) Mb 19W IS Saiuic Have S, S.M.5S.) boool O.LE. Tanutive Bafifo Nave IS

2) Test Results Response waves of each floor's acceleration and relative displacement ie., the bearing's deflexion between the table and the superstructure under the tentative design wave NS (SI), are shown in Fig.4-1. Each floor swayed slowly without a phase difference among floors, and the maximum acceleration was reduced from 337 Gal to 217 Gal on the 1st floor and 280 Gal on the roof. The maximum relative displacement in the isolation layer was 1.13 cm which corresponded to a maximum shear strain of 75 percent. The hysteresis loops of the total shearing force by the static cyclic loading and seismic shaking test are shown in Fig.4-2. The shape of the static loop is constricted a little in the middle compared with that of the dynamic loop, but it can be said that the dynamic loop is almost equal to the static one. The floor response spectrum on the 2nd floor is shown in Pig.4-3 The results for an

-130- > I. 300 RF ACC

250-f *"*•• •"••"•'••' •*<"•' "W'l"1 *•

-It Fig. 4.1 Time Histories for Tentative Design Wave NS(S!>

SUlle Cnllc ( »'»*• I fcl—lc IMilW >wl«n •»«»

I * 5 I

-SOiro tisrtK(M»t 50im Fig. 4.2 Difference In Staric and Dynamic Loops (LRB)

Frequency(Hz) 0.2

0.6 1 S Period(Sec) Fig. 4.3 Floor Response Spectra of Model Structure i isolated model with three kinds of devices are compared with those for a non-isolated modeL In the case of the isolated model, the response spectrum in the range of the period lower than 0.2 sec, are reduced and the maximum value of acceleration spectra was mitigated below one half. It can be seen that seismic isolation systems remarkably reduce response acceleration compared to a non-isolated building. The maximum acceleration ratio on each floor and the relative displacement in the isolation layer under multi-directional shaking by tentative design wave are shown in Fig.4-4. The response acceleration on the roof for the simultaneous NS(X) and UD(Z) waves

-131- is enhanced slightly compared with uni-directional excitation. However, it can be said that there is no significant increase in the response under the multi-directional shaking. Although the relative displacement under three-directional (XYZ) shaking by El Centre wave is 15 percent in excess of that under uni-directional (X) shaking, the deflexion scene to be almost the same between uni-directional and three-directional shaking by the tentative design wave

Hir«ttji» — •»— n-Mnctlx —— Tentative »t»i«f> I«»e(S,) ' C-linttlw ——- Tantttiv* •••lfn l«v«tSj) » W-tlmtlw —•- El Ctntro US i-iilti tii'j'.f i./

i^v. 1 l ir u lit Input lave Deretion Acceleration QJRelatlve Displacement Ratio on each Floor (LRB)

Fig. 4.4 Difference in Multi-Directional Shaking The maximum response acceleration on me £uu ixum w -«»«*- *-•- Jwn in Fig. 4-5. It is observed that the maximum reponse acceleration increases more rapidly as the intensity of the input acceleration increases beyond that of 1.9 times the tentative design wave SI. The reason is considered to be that the response of the superstructure is affected by the tensile force acting on the bearings and the hardening of the rubber. The relationship between the axial force and shearing strain of the bearing (A3) under 2.3 times the tentative design earthquake is shown in Fig.4-6 with the results of rupture test on the bearings. The bearing has enough safety margin for the failure even under this extreme seismic load.

3) Summary of Results The results from the shaking table test under design earthquakes are summarized as follows: (D The hysteresis characteristics of bearings obtained from the shaking table test are almost equal to those in the element test (g) The base isolation systems can remarkably mitigate the seismic load compared with non-isolated buildings. (§) Under the tentative design earthquake, the isolation devices show good performance in deflexion and sustainment.

-132- *«»ult el tun.rtil Tt»t • Static Cyclic Lastfinf 1000 O Static ttonotonic Loadint cc if M fl Uynaaic loftoienic tasoiftg

e B coo e e

I e : 200 > < 0 1.0 1.6 JO 2.8 1.0 • -««-l«0-40r-100-100-10l 0 U0 ilO 100 400 ill ill HOT LEVEL (XS1) *> Sbctrinf Striin (I) Fig. 4.6 Axial Force Ratio( x 2.2ton)-. Shearing Strain Relationship Fig. 4.5 Maximvun Acceleration of 2nd Floor, l on Rupture to Each Input Level

Fig.4.6

(2) Shaking Table Test on Ultimate Behavior of Seismic Isolation System 1) Outline of The Test It is very important to evaluate the aseismic margin of a base-isolated nuclear reactor building to ensure safety against earthquakes. In order to achieve it, the ultimate behavior of base isolation system, must be completely understood. Although the amount of data for the ultimate behavior of laminated rubber bearings through rupture tests of bearings have gradually increased recently, there have been few tests for the total base isolation system. In order to investigate the response behavior of seismic isolation system with laminated rubber bearings at extremely large amplitude, shaking table test using small-scale model were conducted During the test, particular attention is paid to the base isolation bearing, which plays an important role in the base isolation system. Increasing the input acceleration gradually, the response behavior of the model up to the vicinity of the ultimate state of base -isolation bearings is investigated. The same similarity law as was employed for the shaking table test carried out at design level input, is adopted which requires the f ollowings. (D The stress of the laminated rubber bearings be equal to that of the prototype. (g) The amplitude of the input acceleration of the shaking table be equal to that of the prototype. ® The geometry of the bearing be similar to that of prototype and the scale ratio be 1/15. The superstructure with a weight of 17.8 tons is made of reinforced concrete .and can be consided to be a rigid body. In order to investigate the influence of the overturning moment of the superstructure upon the ultimate response behavior of the laminated rubber bearings, models with high and low height of gravity center were adopted Fig.4-7 shows the dimensions of the models. The ratios of the height of the gravity center to the model width

-133- were 0.5 and 0.25. Lead-rubber bearings(LRB) with the same materials, size and geometry as employed for the shaking table test at the design level input, were used The number of the elements is 2 rows x 4 units=a The LRB specimens were 1/15 reduced scale of prototype LRB of which rated capacity is 500 tans. _-concrete ^-Concrete

unit:* 1000 1000 8 8 B 8 s g & a Arranjaanet of Rubber Bearinfs K ELEVATION Test Model with Lew Center of Gravity

Fig.4-7 Test Model Teit Hodel with Hifh Center of Gravity

The same seismic motion as used in the test at the design level input (Tentative design response spectrum, and the phase of La Union ground motion are utilized) was employed as SI wave. However, due to the capability of the shaking table, the component with the period longer than 3 second were omitted to bring the response displacement of LRB close to the ultimate level. Tab. 4-4 shows cases for tests, At first, excitations were carried out using random waves of white noise in order to investigate the basic characteristics of the base isolation system. Then, seismic wave excitations were applied The amplitudes of each run was increased gradually from the level of tOSl considering the safety or the performance of the shaking table. In order to investigate the behavior of the test model and the bearing, the following items were measured by accelerometers, displacement gauges, and component force transducers. CD The axial and shear forces acting on the bearings. (g) Relative displacement of the shaking table and the lower part of the test model (D Absolute acceleration of the shaking table and the test model Table.4-4 Test Case Test fedel Center of grevit; input rave input input /Test aodel width level direction A 0.25 imadoa nve SOfftl X.U tentative atve 1.0S1- Z B 0.5 raodoa rave GOftl l.X.1 tentative rave LOS1~ I rudoi rave SOfal C 0.25 tentative m»t 11 Centro mn LOS1~ LXY.XZ Htchinobe mn -134- 2) Test Results Fig. 4-8 shows the hysteresis of total shear forces and the displacement of eight LRBs in case A. Those hysteresis indicate that a slight hardening tendency of bearings begins at the level of 3.0S1 and distinctive hardening feature can be seen at 4.0S1. Stable hysteresis can be seen even when the input level reaches 6.0S1. In Fig. 4-9, the shear strain-axial stress hysteresis of A4 laminated rubber in case A for each excitation level, are shown together with the rupture curve of LRB obtained by element tests. The response reaches the rupture curve at the level of 5.0S1, and it exceeds the rupture curve at 6.0S1. Fig.4-10,4-11 show the horizontal and vertical response acceleration at the center of gravity induced by each excition leveL At the level of 2.5S1 or less, the influence of the height of the gravity is not recognized. However, at the level of 2.5S1 ~ 3.0S1, the horizontal response acceleration of the test model, of which the center of gravity is high, become slightly larger than that of the model with low center of gravity, while beyond the level of 3. OS, the horizontal response acceleration of the model with low center of gravity is larger than that of high center of gravity model- As for the vertical response acceleration which is more influenced by the rocking vibration, the response value of the model with the high center of gravity is remarkably higher than that of the model with the low center of gravity, when input motion exceeds the level of 3.0S1. The tests were carried out increasing the input level by 0.25S1 for each case until LRB were ruputured Table 4-5 shows the rupture state for case C. At the level of 6.60S1 after 6.75S1 excitation, A4 was ruptured first and then four pieces of LRB were fractured by the following excitations in case C. Only one piece was ruptured at each excitation leveL No evidence was observed that the rupture of one piece of LRB would cause a chain-reaction rupture of other pieces. Fig. 4-12 shows the time history wave and the deformation of LRB at A4. Main characteristics were as follows. (I) Local fracture was found before the complete rupture of each specimen. © When the local rupture occurred in the four cases mentioned above, compressive axial force was acting on the LRB. After this, at the point where the compressive stress of the rubber turned to tensile stress, the entire section of the rubber was ruptured © There was one case that the rubber was ruptured when the tensile force was acting on the LRB. @ All of these LRB were ruptured after they exceed the nipture curve several tunes. Fig. 4-13 shows the variation of the predominant frequency of the superstructure corresponding to the rupturing of LRB. It can be considered that the predominant frequency tend to decrease due to the decrement of the horizontal stiffness caused by the cyclic

-135- excitations and the rupture of LRBs.

3) Summary of Results Shaking table tests on ultimate behavior of BRJsmic isolation system was carried out As the results of the tests, several datas were obtained concerning behavior of LRB and superstructure when base isolation system reached the vicinity of the ultimate state. Furthermore , the behavior of Hmamc isolation system in the process when base isolation elements were successively ruptured, were observed.

a.0 si) 5 I toe too

HflUGl SISP <-) AVBUGt BISP («) Fig. 4.8 Hysteresis of Laminated Rubber Bearing in Case A

(3.0 SI) .(i.esi) . let. ! i i ctct 0>taticil rupture point

-100 -*--400 fTrT' *"" TIOC SBEAR SHAH (V (to $i) r '?.?•

-400 0 SHEAR STRAIN It} of fte IRES'in Case in Case A Compared with the Rupture point Fig. 4.9 Hysteresis

ACCUMAtH* «0«A»ltT>X> ••"*** 1 i i i i r A'Tcst Hoiitl •i^hUib Ctnter of Cr»vity_ OTot lod

>wo| • r

IJOOt

IOOOI-

IMP vmi (xso Fig. 4.10 Comparison of the Maximum Horizontal Accelerations at the Center of Gravities of the Two Models -136- RESPONSE ACCELERATION ICKAYMT- it ••• • AIS

Atal Mil »llkM*b •fill Mtl til > imtmUr •>« i* fir MOO —

MOO * s — .. f « tseo

• •00 • < I m ^ 1 IIP UVtl (XSO Fig. 4.11 Comparison of the Maximum Vertical Accelerations at the Center of Gravitieis of the Two Models

SHEARING DISPLACEMENT

AXIAL FORCE

Fig. 4.12 Time Histories for LRBs of A4

Table.4-5 Process of the Rupture for Each Excitation in Case C

txciuton Location of order ruptured rubber 6. MSI 6.2SS1 6.50S1 8 8 8 S 6.75S1 6.60S1 U 2 2 8 2 .MSI M .MSI .00S1 Al .00S1 A3

I • J.J i i ' •

I I 01 ...... | .. j

| : I.IU, I.HI. l.M, t.MS, t.m, I.M, I.MI, U 14 II U iciutiS artar Fig. 4.13 Predominant Frequency of the Superstructure Corresponding to the Rupturing of the LRBs in Case C

-137- 5. Earthquake Response Observation of Base iaolated Building The performances of the three base isolated building has been examined to provide the proving data of the reliability of the seismic isolation system. The main features of the buildings were listed in Table5-L Free vibration tests, forced vibration tests, earthquake response observations as well as simulation analysis of the response were conducted. Through the vibration test, the vibration characteristics of the buildings were confirmed. It was also confirmed that the acceleration responses of the building were strongly influenced by the frequency contents of the earthquakes, and that response accelerations were in general greatly reduced due to seismic isolation delvicea The observed earthquake response of one building is described in some detail below. Table 5-1. Base Isolated Buildings

Cmtruetar tfclfht tooUUoB »nia»

m1 Obaun zso 1300 a taNw feu-li* • CorpormUoo ton Stwl tar S«<»r (Nrled 2 me)

3 1W0 TTO m> RubMr lMrlS( • Ctapuu Me •tnl tor DMpn- (Srlod >

TotUhm 5 9»r Kit/ •Kb to*!* Ccnatructloc Un kttv turtle*

(1) Base Isolated Building It is a four-story reinforced concrete building as shown in Fig5-1. The seismic isolation devices are shown in Fig 5-2 and Fig 5-3. The average horizontal stiffness of the rubber bearing adopted for this building is 0.82ton/cm.The stiffness of the damper before yielding is about 2.0ton/cm and the yielding displacement is about 30mm. The design horizonal natural frequencis of this building are 0.71Hz(1.4sec) and 0.48Hz(2.1sec) cxrresponding to pre-yielding and post-yielding stiffnesses, respectibvely. (2) Observed Earthquakes More than thirty earthquakes have been observed since September in 1986 in the building. Epicenter of each earthquake is shown in Fig5-4. Observed earthquakes are and characterized as shown in Table5-2. Fig5-5 shows the averaged acceleration fourior spectra on the basemat of each qroup earthquakes.Each spectrum is normalized such that a maximum spectrum value be 1.0. Low frequency components below LOHz of qroup in are larger than that of group I and H. Table5-2 Classification of Obaarbad Earthquake Group Epieantral No. Typa Diatanc* Foeal Dapth Hagnitudt I N«ar 0 ~ 6 Oka 50 ~ 80k« — 6.0 II Int«r««diat« 50 ~ 160k* 40 ~ 80km ~ 6. 0 III Far 150km ~ 0 ~ 50ka 6.0 ~ 7.0 Magnituda of Japan tfatsorological Agancy

-138- YtJir. :U*ineted Rubber Bearing Steel X-dir.

Flg5-1 Section and Plan of Buildino

700 600

Rubber Sheet t7. OX 14(«88.0) Steel Plate tl 2X13 (=4). 6) Fig5.2 Beerino

Ow.o O«S.0 OIM.0 eiO.O

FIg5.4 Epioenter of Observed Earthquakes

-139- 0.6 0.6 X-direction Y-

0-1 0. S 1 5 10 20 0.1 0.5 1 5 10 20 Frequency (Hz) Frequency (Hz)

Flg5.5 Normalized Acceleration spectra on Basrat

(3) Responce of the Building Fig5-6 shows the ratio of the maximum acceleration(lst floor /basemat). It can be seen that the horrizontal acceleration ratio for both qroup I and n earthquakes are less than 0.5, that is , the acceleration of the building is reduced sufficiently.But,in case of group ID earthqakes, acceleration is not so reduced This is because they contain much low frequency componens as is shown in Fig5-5. The average ratio is around 0.6 in both X and Y-direction. On the other hand,the vertical acceleration ratio depends little on the amplitudes of ground motion and the location of epicenter, and the average ratio is 1.24. Fig5.7 shows the relationship between the maximum relative displacement(besemat-lF) and the maximum acceleration and velocity on the basemat Strong correlation is observed between displacement of the isolation device and maximum velocity. This suggests that deformation of the isolation device can be predicted by the maximum velocity of input earthquakes.

2.0 .X-direction ^'2.0 Y-direction 2.0 Z-direction o 1.5 1.5 • IS i ;i.o

T 0.0 0.0 0.0 20 40 60 50 100" 200 6 20 » 40 lev Act (Base) (sal) Uax Ace (Base) (sal) law Ace (Baae) (sal)

Flg5.6 Relationship between Acceleration neapome Ratio of First Floor and Uaximn Acceleration on

so o O Group I r*0.S4 • Group 1 20 A Group IH O Others 10 r : Coefficient of Correlation Qe o Q 0 20 40 (0 1 2 3 * S lfax.Acc. (Baaonat) (oal) Max. Velocity(Basemt) (cVaec)

Fig5.7 Relationship between Ifexiiun Displacement (1F-Baaonat)and Maxiiui Acceleration and Velocity on Besmrt -140- It must be noted here that the displacements of the isolation device in these earthquakes are less than yielding displacement of the dampers of 30mm so dampers did not work as damping devices. The eccentric distance between the stiffness center of the isolation devices and the gravity center of the upper structure is about 30cm. Fig5-8 shows the time histories of the translational displacement and torsianal angle around the gravity center in the case of Dec.17,1987 eerthquake.The maximum horizontal displacement due to torsional notion is about 3.0mm at the edge of the building which constitues 15X of the recorded displacement This is not so large as to make much influence on the safety of the bearings. And, if needed, the torsional behavior can be reduced by adequate arrangement of the isolation devices. Fig5-9 shows the time history of the relative torsional angle between the first and the 5 roof floor. Maximum torsional angle of 7x10- rad is much smaller than that of the first a floor of 3x10- ra& This means that the torsional deformation of the isolated building concenrtate at the isolation devices and that of the upper structure is very smalL

2.0C MAX 1.69X10

-2.00 0.0 25.0 50.0 75.0 100.0 (sec) (a) Translaticnal Displacement / 3.00 MAX -2.96X10"' UI0*'> HAX 7.44XI0*1 8.00

25.0 50.0 75.0 100.0 0.0 (sec) (b) Torsional Angle Fig5.9 festive Torsional Anple Fig5.8Translationel Displacement and Torsional Angle between First Floor and Roof Floor at the Center of Gravity on First Floor (4) Summary of results ^Acceleration response during earthquake is strongly influenced by the frequency characteristics of the ground motion, especially by the intensity of the component around the first natural frequency of the base isolated building. However.for any earthquake, high frequency component of the acceleration in the building is sufficiently reduced in comparison with that of the ground surface. 2)The maximum relative displacement of the isolation device is almost in proporsion to the maximum velocity of input earthquake. 3)Torsonal deformation of the base isolated building is concentrated at the isolation devices and that of the upper structure is very small.

6. Conclusions The performance of seismic isolation elements and base isolated buildings up to the failure of elements are confirmed through large scale element tests and shaking table tests. The existing basei solated building haven't experienced strong earthquakes, but the

-141- observed responses of the buildings indicates that the base isolated buildings behaves as designed.

(Acknowledgements) Tests on rubber bearings and shaking table tests are a pert of a research project ~ Verification Test of Seismic Isolation for Past Breeder Reactor " sponcered by Ministry of International Trade and Industry, and were conducted under the guidance of Prof H. SHIBATA, Prof T. Fujita and other members of the Advisory Cannrittee,

(References) (l)MAZDA,T,edaL " Test on Large-Scale Isolation Elements " , 10th SMiRT VoLK ANAHE3M(1989)

(2)MAZDA,T,edal, " Test on Large-Scale Seismic Isolation Elements Part 2 Static Characteristics of Laminated Rubber Bearing Type " , Uth SMiRT VoLK TOKYO(1991)

(3)ISHIDA,K,etaL " Failure Tests of Laminated Rubber Bearings " , Uth SMiRT VoLK TOKYO(1991)

(4)ISHIDA,K,et.al " Shaking Table Test on Base Isolated FBR Plant Model Part 1 Shaking Table Test Results " , 11th SMiRT VoLK TOKYO(1991)

(5)The Special Scientific Events Subcommittee of the Executive Committee for SMiRT 11, " Demonstration Test of Seismic Isolation System For FBR " in Seismic Isolation and Response Control For Nuclear and Non-Naclear Structures " SMiRT 11 TOKYO(1991)

(6)KAWALN,et&aL " Earthquake Response Observation of Isolated Building " 10th SMiRT VoLK, ANAHEIM.

(7)SHI0JIRI, H,etal, " Seismic Isolation Study In CRIEPI " .SeismicShock, and Vibration Isolation 1990, ASME.

-142- IAEA SPECIALISTS1 MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, California, USA, March 18-20, 1992

EXPERIMENTS ON SEISMIC ISOLATION IN ITALY XA0055383 G. Bonacina (ISMES S.p.A., Bergamo, Italy) F. Bettinali (ENEL-CRIS, Milano, Italy) A. Martelli (ENEA-RIN, Bologna, Italy) M. Olivieri (ANSALDO-RICERCHE, Genova, Italy)

ABSTRACT Static and dynamic tests have been performed in Italy on high damping steel-laminated elastomer bearings in various scales, rubber specimens and structures isolated by means of such bearings, in the framework of studies in progress to support seismic isolation development. Tests on rubber specimens and bearings have already provided important data (vertical and horizontal stiffnesses, damping, creep, temperature, aging and scale effects, etc.), necessary for the development and validation of numerical models, comparison with the test results of isolated structure mockups and actual buildings, and improvement of design guidelines. Dynamic experiments of structures concerned both full-scale and scaled isolated structure mock-ups and actual isolated buildings (one of those forming the SIP Administration Center at Ancona, an isolated house at Squillace, Calabria). Both snap-back tests and forced excitation experiments were performed, to rather large displace- ments. The latter were both sinusoidal and (on a 1/4 scale mock-up) seismic, with one- and multidirectional simultaneous excitations. Test results have already demonstrated the adequacy of seismic isolation and have provided data useful for the comparison with single bearing test results and validation of numerical models for the analysis of isolated structures. This paper reports the main features and results of tests performed or in progress. Further tests planned have been mentioned in the Status Report. Numerical analysis of measured data and guidelines development have been discussed in separate technical papers. 1. INTRODUCTION Martelli & Bettinali [1] have explained that considerable efforts are being devoted in Italy to the development of seismic isolation and its application to both civil buildings and industrial constructions. To the aim of demonstrating the adequacy of this technique to improve the seismic safety and performance of structures, and that of supporting the development of appropriate design guidelines for

-143- isolated constructions (Olivieri et al. [2]) R&D work was undertaken in Italy by the National Agency for New Technologies, Energy and Ambient (ENEA), the National utility (ENEL), ISMES, ALGA and ANSALDO-RICERCHE in 1989. The Italian R&D activities are focusing at present on the use of the high damping steel-laminated elastomer bearings (HDRBs): they are considered very innovative and adequate for a wide-ranging application (including to high risk plants), and have been adopted in most isolated buildings in Italy. Work takes avantage of national cooperations in progress in the framework of the National Working Group on Seismic Isolation (GLIS), as well as of international collaborations. It concerns both experimental and numerical studies of isolators, bearing materials and isolated structures. The ongoing experimental campaign includes tests at ISMES on single isolators and isolated structure mock-ups, as well as in-situ experiments of isolated buildings and tests on rubber specimens. Experiments on single isolators, rubber specimens and an isolated mock-up are also starting at the ENEA/ANSALDO Center of Boschetto (Genova). After some remarks on bearings used, this paper (which is an updated summary of those of Martelli & Castoldi [3] and Forni et al. [4]) presents the main features and results of tests performed or in progress. Further tests planned have been mentioned by Martelli & Bettinali [1]; numerical analysis has been discussed by Bettinali et al. [5]. 2. BEARING FEATURES All the HDRBs adopted in the isolated buildings in Italy and those used in our tests have been fabricated by ALGA. Most of our tests have been based on one (the 500 mm diameter) of the two HDRB types used by SEAT in the five isolated buildings of the Administration Centre of the National Telephone Company (SIP) at Ancona, which are at present the most important application of seismic isolation in Italy. This choice was mainly due the possibility of performing in-situ experiments on one of such buildings, and thus, to compare the results of tests performed on single bearings to those obtained for a real isolated structure (Sect. 5.2.1). Furthermore, detailed acceptance tests performed by Bonacina et al. [6] on the SIP bearings had already made very useful data available. The basic parameter for the fabrication of the SIP bearings (Fig. 1) was the horizontal displacement of 144 mm which was assumed by Giuliani [7] for their design at 100% a {o = shear strain), i.e. at a displacement equal to the total rubber height. Several SlP-type bearings, in both full and reduced scales were fabricated and tested (28 full size isolators, together with 22, 20 and 40 in 1/2, 1/3 and 1/4 scales, respectively). These correspond to the original attachment solution used for bearings in the SIP buildings, where the steel-end plates are provided with a machined groove to restrain them. However, 48 modified isolators (half in full scale, half in 1/2 scale) were also fabricated and are being tested: these are provided with attachment devices - designed by ENEA - that also allow for dowelling or bolting them, and modified rubber materials, as well (see also Sect. 3); furthermore, they are characterized by a central hole (which enabled a better centering of steel plates and better

-144- Figure la: Sketch of the HDRBs Figure lb: Sketch of modified of the SIP buildings at Ancona SIP - type bearings designed by [diameter = 500 mm or 600 mm; ENEA [diameter = 500 mm; total total height = 204 mm ; total height = 202 mm; total rubber rubber height = 144 mm (11 in- height = 132 mm (again 11 inner ner sheets 12 mm thick and 2 sheets 12 mm thick but no cover cover sheets 6 mm thick); total sheets); total steel height = steel height = 60 mm (10 inner 70 mm (again 10 inner plates 3 plates 3 mm thick and 2 outer mm thick, 2 outer plates 20 mm plates 15 mm thick); lateral co thick); lateral cover width = ver width = 10 mm; no central 10 mm; central hole diameter = hole; no holes for bolting the 30 mm (74 mm for outer plates); isolators; bearing attachment 8 holes for bolts at r = 200 mm to the structure and basemat from bearing center, 6 holes at provided by a machined groove r = 70 mm; possibility of cen;t of the steel end plates]. ral dowel and dowel + bolts].

a.DO

7.58

c an. 1 3.DO V

• * 2.58 •

0.88 -28.ee -5.ee ie.ee 25.88 4f» T . Figure 2: Temperature effects on stiffness, measured in specimen shear tests of the SIP HDRB compound (•).

-145- bonding) and by the absence of upper and lower rubber cover (Fig. 1) • Finally two bearings (one for both aforementioned scales) were fabricated with various artificial defects concerning bonding, so as to enable the validation of three-dimensional bearing models (Betti- nali et al. [5]) and check of non-destructive analysis methods. 3. TESTS ON RUBBER SPECIMENS Tests on rubber specimens are being performed by ENEA in cooperation with ALGA and ANSALDO, with the aim of improving fabrication processes, controling bearing quality and determining rubber properties. In particular, shear tests on rubber specimens preceded all experiments on bearings and isolated mock-ups. These were carried out for all bearing batchs, to mainly measure the shear modulus of elasticity (G). Two G values were determined: G-, which corresponded (as required by existing national codes for rubber supports of bridges) to deformations from 0 to 60% a, and G2, which corresponded to deformations from -100% a to +100% a. It was found that G_ is about 40% lower than G1. This result is consistent with the decrease of bearing horizontal stiffness by increasing displacements (see Sect. 4). The scattering of G data was rather small (Forni et al. [4] and Martelli et al. [8]). It is noted that some differences were detected for G. with respect to values measured for the (nominally equal) rubber of the actual bearings used for SIP buildings (the latter being about 10% higher, on the average). The scattering of G data was also more limited than that found for such bearings. These results are certainly partly due to modifications in the fabrication process, which was completely automated after the acceptance tests of SIP building bearings only (Martelli et al. [8]). Tests on specimens have also been performed for new rubber materials: the aim (which has been achieved) was to define three HDRBs with improved rubber-steel bonding and different compounds, to be later tested (Sect. 4). With respect to SIP HDRBs, the first compound has a higher (10%) ultimate tensile strength and equal creep, while the second has a better (15%) elongation (for both, the condition of a damping decrease lower than 20% with respect to SIP HDRBs - see Sect. 4 - has been respected); the third is a very soft compound (SHORE A3 equal to 30-35, against a value of 60 for SIP- type HDRBs), provided by Malaysian Rubber Producers Association to ALGA. Further tests on the new compounds aimed at defining a hyperelastic model of the rubber to be implemented in the ABAQUS computer code for the detailed analysis of isolators (see Bettinali et al. [5]). More precisely, following tests were performed: (a) dynamic tests for the determination of damping; (b) quasi-static and sustained compression experiments (7 days long) of cylindrical specimens; (c) static tests, with tensile loads on ring-type specimens and sustained tensile loads on dog-bone-type specimens; (d) quasi- static, sustained (96 hours long) and dynamic shear tests; (e) threeaxial compression tests to evaluate rubber compressibility. Experiments on specimens formed by the compound used in SIP HDRBs have also been performed by ALGA at CERISIE to define the accelerated aging methods to be used in the analysis of aging effects

-146- on bearing response (Sect. 4). The extrapolation of test data to the assumed normal temperature (T = 30 C) - by use of the Arrhenius law - led to the following results: (a) compression tests, performed at constant deformation and temperatures equal to 50 c, 70 C and 90 C, indicate a relaxation of pressure equal to 50% of the initial value after 142 years at TN; (b) elongation tests, performed in air at the aforesaid temperatures and 600% elongation, indicate a residual deformation of 60% after 44 years only, at T , due to more severe oxigen attack (a longer time is expected in vacuum conditions); (c) shear tests, performed at 90 C and 110 C, indicate an increase of G of 33% after 110 years, at T . Finally, tests are in progress at the Boschetto Centre for a first analysis of both temperature (Fig. 2) and accelerated aging effects (see also Sect. 4): it has already been demonstrated that temperature does not produce any permanent modification of compound features and that it has a non-negligible effect on horizontal stiffness at the low values (Fig. 2). 4. TESTS OF ISOLATION BEARINGS Tests of single SlP-type bearings were defined according to the guidelines document of Martelli al. [9]. They were started by ENEA at ISMES in June 1990, to determine vertical and horizontal stiffnesses, damping and failure modes, as well as the effects of bearing scale, dynamic excitation, creep, vertical load variation and natural aging (Martelli et al. [8]). Natural aging studies will be continued in co-operation with SEAT (within GLIS activities) using bearings placed inside the Ancona buildings, under the actual compression load and close to those in operation, so as to subject them to the actual aging conditions. The following experiments have been completed: (a) tests for the evaluation of static vertical stiffness; (b) cyclic tests for the evaluation of static horizontal stiffness; (c) sustained compression tests for the evaluation of creep effects; (d) tests for the static evaluation of the effects of vertical load variation on the horizontal stiffness; (e) three sets of static tests for the evaluation of natural aging effects on stiffnesses and damping; (f) sinusoidal horizontal excitation tests at fixed frequencies; (h) a quasi-static cyclic failure test. Tests for the evaluation of accelerated aging and temperature effects will be also performed on bearings (in addition to those on specimens, see Sect. 3) at the Boschetto Centre. Moreover, experiments were performed on bearings used in an isolated house at Squil- lace, Calabria (see Sect. 5.2.2). Finally the experimental analysis of the effects of bearing attachment and compound features has been started, and tests on isolators used in other buildings have been planned, in the framework of the promotion activities of GLIS. For bearing tests, use was made of the SISTEM machine, which had been designed and fabricated by ENEA: it allows for static and dynamic testing of both single bearings and a pair of superposed isolators to rather large displacements, with one-directional (ID) and 2D simultaneous, horizontal excitations under vertical compression load, or also - after recent modifications - tensile load (see Martelli & Bettinali [1] and Fig. 3). Test results so far obtained (Forni et al. [4]) have already provided very useful information to improve the knowledge on isola-

-147- SISTEM (Seismic isolator TEst Machine)

TECHNICAL CARD (January 1992}

OVERALL DIMENSIONS: 5.5 x 3.5 x 2.5 m WEIGHT: 100 kN MAXIMUM DIMENSION OF THE ISOLATORS TO BE TESTED: Height (mm) Diameter (mm) - Single *: 360 700 - Sandwich: .. 310 700 VERTICAL LOAD (one actuator) **: Force (kN) Stroke (mm) Frequency (Hz) - Static (compression): ... 3000 90 0 - Static (traction): 1000 90 0 - Static (traction)*: 500 90 0 - Dynamic: 1000/3000 +-45 0-5 -Dynamic*: 500/3000 +-45 0-5 HORIZONTAL LOAD (two actuators) ***: Force (kN) Stroke (mm) Frequency (Hz) - Static* (Push-pull): 400 350 0 - Static* (Push-push): 480 350 0 - Static (Push-push): 480 1000 0 - Dynamic* (Push-pull): 400 +-350 0 - 5 - Dynamic* (Push-push): .... 480/320 +-350 0-5 - Dynamic (Two directional): 240/160 +-500 0-5 HYDRAULIC CIRCUIT: - Capacity: 1000 1/min - Pressure: 210 bar

* By use of an horizontal roller slide (friction coefficient* 0.003, stroke= +- 350 mm). ** Vertical and horizontal loads can be simultaneous. *** The actuators can be used in the same direction and same sense (push- push), in the same direction and opposite sense (push-pull) and il two normal directions.

Figure 3: Technical card of the SISTEM test machine after recent modifications.

-148- tor behaviour, test procedures and design guidelines. Figs. 4 and 5 (where data obtained for the actual bearings mounted at SIP building base are also reported for a matter of comparison) show that the simplified formulas suggested by Martelli et al. [9] to calculate vertical and horizontal stiffnesses (starting from bearing geometry and rubber properties) are reasonably accurate: this result will allow future acceptance tests to be limited to a much lower number of bearings, with considerable economic advantages (tests on specimens are obviously rather less expensive); it also enables the simplification of complicated studies such as those of temperature and aging effects, for which a large use of tests on specimens can thus be made (Sect. 3). However, the application of such formulas requires - like in our case - a good knowledge of rubber properties. In particular, a correct measurement of G is essential. Also, the spread of G data must be limited as much as possible: this makes it necessary to improve the bearing fabrication process. Such an improvement, together with that of the characteristics of bearing materials, may enhance bearing performance, and thus safety margins with respect to the design displacement. The dynamic similitude was sufficiently well respected by the response of scaled bearings (Figs. 4 and 5): this permitted a reliable use of data obtained on scaled isolated mock-ups, such as those described in Sect. 5.1.2. Bearing damage started at about 160% a in the failure test performed to date (which concerned one isolator only, without lateral rubber cover, see Martelli & Bettinali [1]), but no collapse nor overturning occurred even at 260% shear strain (Fig. 6). Furthermo- re, damage might have been caused by some initial defect, due to bearing reworking, performed to eliminate the lateral rubber cover. Anyway, further failure tests have been planned on bearings with better bonding features and improved geometric details of steel- plates. (It is noted that tests carried out by the Karlsruhe Univer- sity on the bearings to be installed at the base of the liquid gas storage tanks in Greece mentioned by Martelli & Bettinali [1]), which are characterized by such improvements, were successfully performed at 200% a without any bearing damage before a very large number of cycles.). Horizontal stiffness was only slightly affected by vertical load variation; it correctly decreased considerably by increasing displacement, to 50% a; then it remained quasi-constant (10% decrease) to 100% a; finally, it increased slightly to the excitation level at which bearing damage started (Fig. 6). Damping nature (Figs. 6 and 7) was largely non-viscous; equivalent viscous damping was similar for the various bearing scales; it decreased by increasing excitation, to the excitation level at which damage started (it remained larger than 10% of the critical in dynamic tests). The effects of dynamic excitation on damping were found more important than those on horizontal stiffness (on which it had no effects at 50% a and increased the quasi-static value by less than 10% at 100% a). Creep effects due to vertical load were small (7%-8% of the dead load deflection). No effects of natural aging have been found, yet, after 16 months. Finally, the first results of tests in progress on the modified

-149- — Average measured values Figure 4: Static verti- Maximum and minimum measured values cal stiffness versus 0 Theoretical data 1 SIP bearing tests bearing diameter. 750-

300 450 Isolator diameter (mm)

™—• Mean measured values Figure 5: static hori- —- Maximum and minimum measured values zontal stiffness versus 0/ Theoretical values 9 SIP bearing tests bearing diameter at the 1050. vertical design load in the range of -100% a to +100% o. 700.

350- s o

0 150 300 isolator diameter {mm)

CN/mm] B Figure 6: Horizontal 1 stiffness k, [•] and 21 fraction [%] of critical i 1 900 equivalent viscous 1 damping R [o] measured in the failure test, > eoo 1 versus applied displacement (• = stiffness . 700 1 measured in previous 1 i 1 tests). IS . i 1 1 600 V 13 . 500

11 . S / 400 S s o 9 .

300 """—°" • 7 .

10. 30. 50. 70. 90. 110. 130. 1S0. 170. maximum displacement [aim]

-150- \ Measured top view • V * 0.3 Hz Pos. 1 15 * \ - 0.6 Hz \ 0.9 Hz «\ \ _. - Maximum 14 y \ \ \ and minimum v • \ values 13 \ ' \ \ \ N o\\ \ \ 12 N \\ "\ SIP test N 11 x 0.1 Hz o 0.3 Hz N\. 10 — 0.6 Hz • 0.8 Hz

10 30 50 70 90 110 130 Maximum displacement (mm) en Figure 7: Fraction [%] Pos. 3 of critical equivalent viscous damping B for full scale bearings measured in dynamic tests at fixed frequencies, versus applied lateral view displacement.

Pos. 3 Figure 8: Sketch of the 9,500 kN isolated structure mock-up. bearings mentioned in Sect. 2 (more precisely, those provided with a central dowel as attachment system), show a reduction of vertical stiffness by about 20%, due to the combined effects of lower values of shape factor, cross-section area and total rubber thickness. The- se results have confirmed the validity of the aforesaid simplified formulas to compute bearing stiffnesses. They have also stressed the adequacy of the single-dowel attachment system for minimizing the deformations of bearing end-plates. 5. TESTS OF ISOLATED STRUCTURES 5.1 Mock-up tests The mock-ups used in the laboratory tests performed to date were such as to only reproduce the mass of actual structures, being characterized by very large stiffness. However, experiments on more realistic mock-ups of isolated structures have also been planned. Tests have already been performed on a full-scale mock-up and a 1/4 scale mock-up; experiments are also beginning on a 1/2 scale mock-up. 5.1.1 Tests of a full-scale mock-up. The first laboratory tests were performed in August 1990. They made use of the inertial mass of the ISMES multiexcitation rig, supported at the base by six full- scale bearings (Fig. 8). The use of this mass (which weights 9,500 kN) allowed for a close approximation of the actual design vertical load (1,600 kN) that is prescribed for each of the 500 mm diameter bearings which support the five isolated buildings of the SIP Admi- nistration Center at Ancona (see Sect. 2). This mock-up was subjected to snap-back tests (making use of collapsible devices), where the initial displacements were equal to 13 mm, 30 mm, 65 mm and 85 mm (Fig. 9). The largest value corresponded to about 60% c. This value - such as to hinder sliding of the mass (which could not be attached to the steel-end plates of bearings) - was sufficiently close to the design value (144 mm = 100% a) as to provide reliable information on the behaviour of isolated structures and demonstrate the adequacy of the snap-back mechanisms, for their subsequent use in the in-situ tests of the SIP building (Sect. 5.2.1). The campaign was concluded by a second experiment at 65 mm displacement to verify the reproducibility of test results. 5.1.2 Tests of a 1/4 scale mock-up. A mock-up supported by four 1/4 scale SIP-type bearings was also fabricated and tested (Fig. 10). Its weight was 394 kN, which correctly provided a vertical load per isolator equal to about l/16th of that present in the tests of the 9,500 kN mock-up. This mock-up was subjected to both snap-back and forced excitation experiments on the six-degrees-of-freedom shake table of ISMES (MASTER) in February 1991 (Figs. 11 and 13). All tests were performed with response displacements gradually increasing to 100% o (9, 18 and 36 mm, corresponding to 25%, 50% and 100% a, respectively, were applied for snap-back tests). Forced excitation tests consisted of both ID sinusoidal experiments and seismic tests with ID, 2D and 3D simultaneous excitations (for the two horizontal directions and the vertical): the latter

-152- 400 Figure 9: Displacement time-history measured in the most severe snap- back test of the 9,500 kN isolated structure mock-up (X = direction of the initial displacement, Y = transverse direction, Z = vertical -eoo - direction).

120

7 2 120

MASS Figure 10: Sketch of the MASTER 394 kN isolated structure mock-up.

-153- 4U- 394 kN ISOLATED MOCK-UP Figure 11: Displacement snap-back test from 36 mm 30- time-history measured in \ the most severe snap- 20- back test of the 394 kN \ isolated structure mock- 10- up (direction of the initial displacement). \ \ r \ 0 • _L \ y \ / -10- 0.61 V '0.57 0.55 0.51 0.46 0.43 -?n- 2 time[s]

0.6- 394 kN ISOLATED MOCK-UP Fig. 12: Acceleration San Rocco NS seismic test at 0 dB time-history measured on 0.4 c\T the 394 kN isolated structure mock-up in the San Rocco NS test at 0 db.

shaking table mock-up

2 time [s]

394 kN ISOLATED MOCK-UP Fig. 13: Acceleration Tolmezzo WE seismic test at 0 dB time-history measured on the 394 kN isolated structure mock-up in the I Tolmezzo tests at 0 db.

shaking table mock-up (1D WE) mock-up (2DNS+WE)

4 6 8 10 time (s]

-i 54- corresponded to records of actual Italian earthquakes (1976, Friuli and 1980, Irpinia) for rigid, medium and soft soil conditions (records at San Rocco, Tolmezzo and Calitri, respectively). Very large amplifications of the actual ground motion were found necessary to reach 100% o for medium soils (a factor 2.5), and especially, rigid soils, while margins were obviously rather reduced - although existing (12%) - for soft soil. 5.1.3 Tests of a 1/2 scale mock-up. Tests on a third mock-up are beginning (March 1992) at the Boschetto Centre. The campaign will consist of both snap-back and forced excitation tests on a mock-up formed by the inertial mass of SCORPIUS shake table. This mass weights about 1,600 kN, thus, it will be supported by four 1/2 scale SIP bearings. SIP-type, dowelled and bolted isolators will be used (as usual, all of them will be previously characterized at ISMES on the SISTEM machine of ENEA, see Sect. 4). Forced excitation will be provided by the shake table, on which various rigid masses will be mounted. Snap-back tests will be performed with initial displacements increasing to about 150% a (this is possible because the mass has been equipped so as to allow for bolting the end plates of bearings). 5.2 In-situ tests of actual buildings In-situ forced-excitation and snap-back tests were performed by ISMES, mainly with ENEL funding, on one of the SIP buildings at An- cona, in September and October 1990 (Fig. 14). These are seven-floor buildings, 25 m high, each weighting 70,000 to 78,000 kN. Starting in June 1991, experiments were also carried out by ISMES, on behalf of ENEL, on both an isolated and a non-isolated houses at Squillace, Calabria (Fig. 15). These houses are four-story reiforced concrete space frame structures: apart from the isolation system, they are characterized by identical sizes, mechanical properties and construction methods (see Forni et al. [4]). All the buildings tested will be provided with a seismic monitoring system. Further tests may be carried out on other new and exixting isolated buildings, in the framework of the promotion activities of GLIS (Martelli & Bettinali [1]). 5.2.1 Tests of the SIP building. The SIP building of Fig. 14 weighted about 63,000 kN at the time of tests, because it was practically still "nacked". Test feasibility had been demonstrated by the results of GLIS collaborative activities, concerning numerical pre-test analysis of the building (Bettinali et al. [5]), and experiments on the 9,500 kN isolated mockup (Sect. 5.1.1). Furthermore, tests took advantage of the results of a numerical study, concerning the propagation of vibrations through the soil due to the collapse of snap-back mechanisms, which had demonstrated the absence of damage induced to the surrounding houses (Forni et al. [4]). Forced vibration tests were carried out using a 100 kN two- eccentric mass mechanical vibrator installed on the building roof (Martelli & Bettinali [1]). For snap-back tests, use was made of hydraulic jacks to displace the building and collapsible devices to release it, similar to the laboratory experiments of the 9,500 kN mock-up. These devices were provided with explosive bolts, to ensure

-155- J J J

SECTION

_JL®_ JS> _js J 5 © ® ®.

ol.J

I j i © ©

Figure 14: View of the PLAN SIP building subjected to in-situ tests. Figure 15: Sketch of the isolated and nonisolated houses at Squillace. simultaneous release at all loading positions (Martelli & Bettinali [1]). Four snap-back tests were carried out by applying initial displacements of 7.5 mm, 37 mm, 70 mm and 107 mm; the latter (corresponding to about 75% a, see Fig. 16) was sufficiently close to the design value. A final test was repeated by ENEA at 70 mm displacement with a very detailed instrumentation of the building (Fig. 17). The in-situ experiments had been preceded by detailed acceptance tests of the actual bearings (Bonacina et. al. [6]) and some further characterization tests of one of such bearings on the SISTEM machine (Martelli et al. [8]). 5.2.2 Tests on the houses at Squillace. Both houses at Squillace were subjected to forced excitation tests (by use of a mechanical vibrator located on the roof) and ambient vibration measurement (wind- and train-induced microtremors). Forced excitation was provided in the two main directions at different amplitudes. Prior to the in-situ tests, some bearings had been characterized dynamically on the SISTEM test machine (Sect. 4). 5.3 Test results Tests of isolated structure mock-ups and actual isolated buildings provided essential information on the behaviour of isolated structures and isolation systems, and for the assessment and validation of calculation procedures, in particular for the verification that single bearing test data are applicable to the analysis of isolated structures (see Bettinali et al. [5]). Tests were also extremely important to demonstrate the excellent performance of seismic isolation to both the technical milieu and public opinion: a careful inspection of the SIP building after tests' conclusion absolutely excluded any damage of the structure and the few brick wall partitions that were already present. 5.3.1 Results of snap-back tests. In snap-back tests, the motion in the initial displacement direction lasted a very few seconds for both the building and the mock-ups (about 3s), and consisted in three appreciable cycles only (Figs. 9, 11, 16 and 17). The mock-up responses indicate a quasi exclusively horizontal translation mode in that direction after mass release; for the building, some translation in the normal direction was due to the structure asymmetry. Residual displacements of some millimeters were always detected at test conclusion; these were partly recovered within some hours; their values seem not to be related to those of the initial displacement, but appear dependent on the deformation history and rest intervals to which the isolators were subjected. Reproducibility of test results was successfully verified for both the mock-ups and the SIP building (Fig. 17). The first response frequency (f^) increased considerably during motion, as displacement amplitude decreased (Figs. 9, 11, 16 and 17); for the building, the f.. values were larger than those measured for the 9,500 kN mock-up, due to the lower mass per isolator and the presence of mostly 600 mm diameter isolators. This behaviour is consistent with the non-linear correlation that exists between bearing elastic forces and displacements (Sect. 4).

-157- JACKS REACTION WALL Figure 16: Displacement time-history measured in w the most severe snap- w back test of the SIP ^OS 1000 • COS 4 building (x = direction • VIBRATOR of the initial displace- ( ON THE ROOF ) ment, Y = transverse di- POS 1 rection, Z = vertical direction). .L,

-158- •0.0 -

ao.o -

• 010 1202 s.eoe «ow TIME ( • 1 _> A A 0.0 - / ^

•IS - \ \J V (MR V

ieoe

Figure 17a: Displacement Figure 17b: Displacement time-history measured in time-history measured in the first snap-back test the second snap-back of the SIP building at test of the SIP building 70 mm initial displace- at 70 mm initial displa- ment (X = direction of cement (X = direction of the initial displace- the initial displacement, Y = transverse di- ment, y = transverse direction, Z = vertical rection, z = vertical direction). direction).

-159- For similar reasons it was impossible to define an unique damping value for each entire test. The assumption of equivalent viscous damping (J3 ) and use of the logarithmic decrement technique led to fi values that vary from 17% of the critical (at the largest cycle) to 20% for the full-scale mock-up, and between 16% and 20% for the building. These values are rather larger than those obtained in the single bearing tests, due the strong increase of fi with decreasing displacement during each test and the hysteretic nature of energy dissipation (Bettinali et al. [5]). 5.3.2 Results of shake table tests. Similar to the snap-back tests, the ID horizontal seismic tests of the 1/4 scale mock-up showed that the motion was correctly mainly limited to the excitation direction. The shaking table and the mass acceleration time-histories obtained by applying the San Rocco NS record at 0 db (unsealed acceleration values) are shown by Fig. 12. The mass acceleration was obtained by averaging the two acceleration signals recorded on the mock-up in the excitation direction, which were almost identical. A significant reduction of the motion through the isolation system was obtained: peak acceleration changes from 0.667 m/s on the shake table to 0.102 m/s on the mock-up (85% decrease). In Fig. 12, the frequency filtering effect caused by the isolators is evident: the San Rocco record has a very short predominant period, but the mock-up responds according to its free vibration period. Similar results were obtained for the Tolmezzo WE record at 0 db (Fig. 13). Despite the longer predominant period of the input signal, a large reduction of the motion amplitude was still present: the peak acceleration reduced from 3.85 m/s on the shake table to 0.425 m/s on the mock-up, thus by 90%. Fig. 13 also shows the mock-up acceleration time-history measured in the previous direction under the 2D excitation of the shake table, corresponding to the simultaneous application of both horizontal components of the Tolmezzo records. The mock-up response is very close to that obtained during the ID test: this indicates that 2D interaction effects on isolation bearings are very small. Similar results were obtained in all the other 2D and 3D tests. An immediate consequence of this result is that - in the absence of eccentricity between the superstructure center of gravity and the center of stiffness of the isolation system - 3D seismic analysis of a structure isolated through HDLRBs might more easily reduced to three ID analyses, one for each of its principal directions. The Calitri records are characterized by longer predominant periods, because of the soft soil conditions at the recording location. When this earthquake is applied to the table, the isolation system is less effective in reducing the input motion transmitted to the structure (see also Sect. 5.1.3). For the Calitri WE record at 1 db (acceleration values of the shake table amplified by 12%), the mock-up responded with a peak acceleration of 0.851 m/s , against a value of 2.09 m/s at the shake table. This corresponds to a reduction of about 60%, which is smaller than those obtained for the other two earthquake records, but still significant. Finally, as to the effects of input acceleration level on response amplitude, it is noted that peak response acceleration varied almost linearly with the scale factor applied to each input acceleration record in the various tests concerning each of the three eathquakes (despite the strongly non-linear behaviour of bearings).

-160- 5.3.3 Results of forced-vibration in-situ tests. Forced vibration tests of the SIP building allowed for the dynamic characterization of the superstructure: it was found - among other results - that modes corresponding to its elastic deformations are located above 5 Hz (Fig. 18), in agreement with pre-test analysis (see Bettinali et al. [5]); furthermore, it is worthwile noting that the relatively large frequency corresponding to rigid body modes (about 2 Hz) was due to the rather small excitation level, at which isolation is not effective, yet. Finally, tests on the two houses at Squillace confirmed the large, beneficial effects of seismic isolation (Fig. 19). 6. CONCLUSIONS Tests performed on high damping rubber specimens and the corresponding isolation bearings have already provided useful information to understand the behaviour of such bearings, enable the numerical analysis of isolated structures, and improve design guidelines for such structures. The experimental analysis of the effects of temperature and accelerated aging and the execution of further failure tests will complete the information required. Further tests on SIP-type bearings with modified compound and dowelled and bolted attachments will enable the application of improved bearings to important public and industrial structures, and will provide data to begin a detailed study for optimizing bearing performance. The experiments carried out on structure mock-ups and actual buildings, isolated by means of high damping elastomer bearings, clearly demonstrated the adequacy of seismic isolation to guarantee the integrity of the structures and their contents. The results of such tests provided excellent data for the characterization of isolation systems, comparison with the single bearing experiments and validation of numerical models for the analysis of isolated structures. The dependence of bearing stiffness on displacement was confirmed, together wth the largely nonviscous nature of damping. REFERENCES [1] A. Martelli and F. Bettinali, status Report on activities on seismic isolation in Italy, Paper presented to this meeting. [2] M. Olivieri, A. Martelli, F. Bettinali and G. Bonacina, Development of guidelines for seismic isolation in Italy, Paper presented to this meeting. [3] Martelli, A. and A. Castoldi, Seismic isolation of structures, in: J. Donea & P.M. Jones (eds.), Experimental and numerical methods in seismic engineering (1991), pp. 351-377. Dordrecht/Boston /London: Kluwer Academic Publishers for the CEC. [4] M. Forni, M. indirli, A. Martelli, B. Spadoni, F. Bettina- li, G. Bonacina, G. Pucci, G. Serino, G.C. Giuliani and A. Marioni, Most recent results of R&D studies in progress in Italy on seismic isolation, in: Proc. of the Int. Post-SMiRT Conf. Seminar on Seismic Isolation of Nuclear and Non-Nuclear Structures, Nara, Japan (1991); to be also published in Nucl. Engrg. Des.

-161- [5] F. Bettinali, A. Martelli, G. Bonacina and M. Olivieri, Numerical activities on seismic isolation in Italy, Paper presented to this meeting. [6] F. Bonacina, M. Torda, G.C. Giuliani, M. Forni, A. Martel- li, P. Masoni and B. Spadoni, Experimental work in Italy on seismic isolation - Activity program and some first results, Proc. 8th ASME- PVP Conf., Nashville, Tennessee, USA (1990), PVP-Vol. 220, pp. 29- 44. [7] G.C. Giuliani, Design experience on seismically isolated buildings, in: Proc. First int. Post-SMiRT Conference Seminar on Seismic Base Isolation of Nuclear Power Facilities, San Francisco, CA, USA, ANL Report, Argonne National Laboratory, CONF-8908221 (1989) pp. 220-245; Nucl. Engrg. Des. 127 (3) (1991) 349-366. [8] A. Martelli, M. Forni, M. Indirli, P. Masoni, B. Spadoni, G. Bonacina, G. Di Pasquale, T. Sand and E.L. Gluekler, Development of design guidelines for seismically isolated nuclear reactors and R&D work performed by ENEA, Nuclear Technology, 97 (1992) 153-169. [9] A. Martelli, P. Masoni, G. Di Pasquale, V. Lucarelli, T. Sand, G. Bonacina, E.L. Gluekler and F.F. Tajirian, Proposal for guidelines for seismically isolated nuclear power plants - Hori- zontal isolation systems using high damping steel-laminated elastomer bearings, Energia Nucleare, 1 (1990) 67-95.

2 5 VELOCITV/FOACE I BASE SLAB I (a) MODULUS (ms" /N)*10~ 0.8 \ \ \ 1 HA5E-C OUTEI BU1LDI \ /I 0.4 \ ft \ ' \ 11 \V/ / / V. / t . A / 0.0 -r **' CONV. FOUNDED BUILDING (b) PHASE LAG

-180.0 + n 150

Figure 18: Frequency response of the SIP building in forced excitation tests. 12 16 20 Frequency (Hz)

Figure 19: Transfer functions measured on the Squillace buildings.

-162- XA0055384

QUALIFICATION OF HIGH DAMPING SEISMIC ISOLATION BEARINGS FOR THE ALMR

F. F. Tajirian E.L.Gluekler W.P.Chen J. M.Kelly

Bechtel National, Inc., GE Nuclear Energy, The Energy Technology Department of 50 Bcale Street, 6835 Via del Oro, Engineering Center, Civil Engineering, San Francisco, CA 94119 San Jose, CA 95153 P.O. Box 1449, University of California, U.SA. U.SA. Canoga Park, CA 91304 Berkeley, CA 94720 U.S.A. U.S.A.

ABSTRACT

The Advanced Liquid Metal Reactor (ALMR) seismic isolation system consists of high damping steel-laminated elastomeric bearings. This type of bearing is used worldwide to isolate buildings and large critical components. A comprehensive testing program has been developed to qualify the use of this system for the ALMR. The program includes material characterization tests, various scale bearing tests, full-size bearing tests, shake table tests, and long-term aging tests.

The main tasks and objectives of this program are described in the paper. Additionally, a detailed assessment of completed ALMR bearing test results will be provided. This assessment will be mainly based on half-scale bearing tests performed at the Earthquake Engineering Research Center (EERC) of the University of California at Berkeley and at the Energy Technology Engineering Center (ETEC). These tests were funded by the U.S. Department of Energy (DOE). Both static and dynamic tests were performed. Bearings with two types of end connections were tested: dowelled and bolted.

The parameters examined will include the vertical, horizontal stiffness and damping of the bearings under different loading conditions up to failure. This will determine the available margins in the bearings above the design vertical load and horizontal displacement Additionally, the self-centering capability of the bearings after an earthquake will be addressed. On the basis of these findings, recommendations can be made if necessary, to improve current manufacturing procedures, quality control, and procurement specifications.

1.0 INTRODUCTION

Seismic isolation is used around the world to mitigate the damaging effects of earthquakes on buildings. It is now generally accepted that seismic-isolated buildings (excluding flexible high-rise structures) will perform better than a conventional building in moderate and strong earthquakes [11]. The most commonly used isolation device has been the steel-laminated elastomeric bearing.

-163- The ALMR incorporates seismic isolation in the reference design to support plant standardization, enhance plant safety margins, permit siting in zones with higher seismicity, and to potentially reduce plant costs. The selected isolation system consists of steel-laminated elastomeric bearings using a high damping compound [1]. Details of the specific compound used were presented elsewhere [12,17]. This particular isolator was selected for the ALMR following a review of available hardware because it is a simple design and its dynamic response, especially at extreme loading, is easier to characterize than some of the more nonlinear systems. Furthermore, it has sufficient inherent damping to eliminate the need for additional energy-absorbing devices which can complicate the design and the system response. Several investigations have shown that equipment response in isolated buildings is minimized when elastomeric bearings with no add-on damping elements are used, and that when frictional or elasto-plastic dampers are incorporated, they inevitably cause high frequency response and increased accelerations in equipment and reduce the effectiveness of isolation [4,6,7,8, IS]. Additionally, the bearings selected are self-restoring even after the application of very large displacements, making them effective during the foreshock, main shock, and after shocks. Furthermore; bearing mechanical properties are unaffected by cycling.

The isolated ALMR structure and components weigh 13,000 kips and are supported on 25 seismic isolation bearings. The average load per bearing is 525 kips.

2.0 ALMR SEISMIC ISOLATION BEARING DESIGN

The seismic design basis for the ALMR is a design safe shutdown earthquake (SSE) with a maximum horizontal and vertical acceleration of 0.3 g anchored to a design earthquake that envelopes the NRC Regulatory Guide 1.60 spectra. The selected criteria is expected to cover over 80 percent of potential nuclear sites in the U.S. excluding California. Options for siting in higher seismic zones, with design earthquakes exceeding 0.5 g, were investigated and were found acceptable.

The main parameters that govern the design of the bearings are the required horizontal and vertical stiffness and the allowable shear strains for different earthquake levels. The frequency goals are a horizontal frequency of 0.75 Hz and a vertical frequency of 20 Hz. The allowable shear strains during an SSE is 50 percent The maximum horizontal displacement in the bearings for this condition is 7.5 in. The bearing is required to have a horizontal displacement margin beyond SSE of at least 3. The bearing shall have a demonstrated vertical load margin of at least 10.

The resulting bearings have a diameter of 52 in. and a total height of 23.1 in. and consist of thirty layers of rubber 0.5 in. thick bonded to 29 steel shims. The steel shim thickness is 0.125 in.

3.0 ALMR SEISMIC BEARING QUALIFICATION PROGRAM

Testing has been completed on a large number of elastomeric bearings of various designs and using different high damping rubber compounds. These tests have been performed at

-164- two facilities, the Earthquake Engineering Research Center (EERC) at the University of California, Berkeley and at the Energy Technology Engineering Center (ETEC). The bearings tested were half-size ALMR bearings. The results of these tests are summarized in this paper. Numerous other bearing configurations have been tested in support of other nuclear programs. The results of these tests have been presented elsewhere [10,11, 14,18]. Plans for additional ALMR specific tests are summarized below:

Elastomeric Characterization Tests

Elastomeric specimens made up of the same compound as used in the ALMR bearings will be tested to characterize aging effects, temperature effects, and other environmental effects such as responses to low gamma radiation environment A program to expose elastomeric specimens to low gamma radiation up to a target dose of 107 rads was initiated in the EBRII sodium purification cell. The maximum accumulated dose expected during the 60 year plant lifetime is less than 5.25x10^ rads. Based on a recent study [20], which reports that high damping rubber bearing horizontal stiffness is affected only after the dose of 5x10^ rads is exceeded, the ALMR radiation field is not expected to cause problems.

Full-Size Bearing Tests

Five full-size bearings will be tested to evaluate design stiffness and damping properties. The results of these tests will be compared with the half-scale tests to examine the validity of extrapolating scaled tests results. These tests will also be used to demonstrate the required minimum beyond design margins and self centering capability.

System Tests

Shaketable tests will be performed at EERC using a simple sub scale representation of the ALMR seismic platform and systems supported by the platform. The model will be subjected to various input earthquake time histories including motions representative of the SSE and beyond design earthquakes. Responses of key components will be measured and compared with predicted responses computed using the selected system response computer programs. These comparisons will verify the computational tools and demonstrate their effectiveness in predicting the response of isolated structures under different earthquake conditions.

Aging Assessment Tests

The ALMR has a design life of 60 years. Long term performance data will be collected from tests on sub-scale ALMR bearings. The bearings will be initially subjected to benchmark vertical load and horizontal displacement tests. Subsequently, the bearings will be stored under constant vertical compressive loads in an environment which simulated both the radiation and temperature environments predicted for the reference standard plant. Periodically, the bearings will be removed, inspected for general condition, and retested to monitor creep effects, and any changes in stiffness and

-165- damping characteristics. After testing, the bearings will be returned to the aging environment for another residence period. In addition, the effectiveness and validity of accelerated aging tests, such as ones performed in Japan [13] will be investigated.

4.0 ALMR BEARING TESTS

A program for testing half-scale ALMR bearings at EERC and ETEC has been completed. The dimensions of the bearings tested are compared with the full-scale ALMR bearings in Table 1.

Table 1 Comparison of Test and Full-Size Bearing Configurations

CHARACTERISTIC FULL-SCALE TEST BEARING Outside diameter (in.) 52 26 Thickness of end plates On.) 2 1 Number of rubber layers 30 30 Thickness of rubber layers (in.) 0.5 0.238 Number of steel shim plates 29 29 Thickness of shim plates (in.) 0.125 0.105 Diameter of shim plate (in.) 46 23 Total bearing height (in.) 23.13 11.81

Eight half-scale bearings were tested at EERC. These tests were quasi static and were performed simultaneously on four bearings. The first set of four used dowel type connections for transferring lateral loads. This was the preferred type of connection in the U.S. at the time, and was selected for the reference ALMR design. Upon completion of the tests, it was decided to convert the remaining four dowelled bearings to bolted bearings by threading the dowel holes and bolting the new end plates to the bearings. The test series had the following objectives:

• Evaluation of vertical stiffness of bearings. • Evaluation of horizontal stiffness and damping, and the influence of vertical load on these characteristics. • Identification of failure modes under axial load and combined axial and shear loads. • Investigation of the effect of end plate connection on the performance and stability of bearings under extreme loads.

The equipment used at EERC and the important test results have been previously summarized [9,16,17].

Nine half-scale bearings similar to the ones tested at EERC were also tested at ETEC [2,3]. Five new bearings were manufactured and tested in the dowelled configuration. Additionally, the four bearings which were converted to a bolted configuration and tested at EERC were retested by ETEC. The main difference between

-166- the EERC and ETEC tests was that the later were performed on individual bearings. In addition to the above objectives the ETEC tests were performed to determine the following effects:

• Examine the effect of loading frequency and number of cycles on the horizontal stiffness and damping characteristics of the bearings. • Investigate the failure modes of the bearings under combined axial and horizontal load (These tests could not be performed at EERC due to limitations of the test equipment). • Investigate the self-centering capability of bearings after a large displacement • Develop data from dynamic tests to compare with the results of static tests performed at EERC.

Axial Load Tests

Several axial load tests were performed at EERC. The tests confirmed the high vertical stiffness of the bearings. Additionally, it was demonstrated that the margin of safety against buckling of a single bearing was 28. Details of these tests were summarized in [17].

Combined Axial Shear Tests

A series of horizontal static and dynamic tests were performed at ETEC. The tests were repeated for different axial loads which were kept constant during the tests. These loads were 14 kips, 70 kips, 140 kips (design load for half-scale), and 210 kips. The dynamic tests were performed for three frequencies; 0.5 Hz, 0.75 Hz, and 2.0 Hz for three horizontal displacements corresponding to a shear strain of 14 percent, 55 percent, and 100 percent The main results from these tests are summarized in this section.

Figure 1 shows the hysteresis loops for one of the dowelled bearings. The axial load for these tests was 140 kips, and the loading frequency was 0.75 Hz. Figure 2 shows the hysteresis loops for a bolted bearing at 100 percent strain for the four axial load levels. In general it can be concluded that in the range of loads used in the tests, the magnitude of the axial load had a negligible effect on the horizontal bearing stiffness and damping. Figures 3 and 4 show the variation of horizontal stiffness and damping for the dowelled and bolted bearing with axial load and loading frequency. The results are plotted for a strain of 14 percent It can be seen that the effects of both axial load and frequency are small. In general both stiffness and damping increase slightly when the frequency is raised from 0.75 Hz to 2.0 Hz. However, the effect of increasing frequency from 0.5 Hz to 0.75 Hz did not give consistent results, indicating that there may be some other factors influencing the results.

-167 8 6 1.0 0 E 1.0 0 9

28 B A 0 ? FOBC E e o UJ kg. u '

o 1 / r JO . 0 i s i.-t OISPLfiCEMENT

Figure 1 Hysteresis Loop for Cyclic Shear Test, Dowelled Bearing Axial Load = 140 kips, Frequency = 0.75 Hz

OISPLfiCEMENT Figure 2 Hysteresis Loop for Cyclic Shear Test, Bolted Bearing Axial Load = 14,70,140, and 210 kips

-168- DOWELLED BEARING

17-

16-

E 050 Hz S 0.75 Hz 15- / • 2.0 Hz

14- /

13- c rlr IJ'I- 14 70 140 i210 AXIAL LOAD (WPS) i

(a) Horizontal Stiffness

DOWELLED BEARING

15-r

£13-

14 70 140 AXIAL LOAD (KIPS)

(b) Damping Figure 3 Effect of Axial Load and Cyclic Frequency Constant ShearStrain = 14 Percent

-169 BOLTED BEARING

13 70 140 210 AXIAL LOAD (WPS)

(a) Horizontal Stiffness

BOLTED BEARING 15f

14-' - 13-'

12-

*mmm BBS H 11-' ii m R^ l.fc 1 10- <• •, • », •• ™,' 14 70 140 210 AXIAL LOAD (WPS)

(b) Damping Figure 4 Effect of Axial Load and Cyclic Frequency Constant Strain = 14 Percent

-170- - BOLTED STATIC •PREDICTED •BOLTED 0.75 Hi

•«• - 0OWELLEO0.75 —• Hz

•+• -4- •+• 20 40 60 80 100 120 140 160 160 200 SHEAR STRAIN (%)

Figure 5 Variation of Horizontal Stiffness with Strain Axial Load = 140 kips

• MATERIW. TESTS • BOLTED 0.7S Hz DOWEU£D0.7SHi - EEFC DOWELLED STATIC

20 40 60 80 100 120 160 SHEAR STRAIN (%)

Figure 6 Variation of Horizontal Damping with Strain Axial Load = 140 kips

Figures 5 and 6 show the variation of horizontal stiffness and damping with strain. These plots include the averages obtained from the ETEC dowelled and bolted tests, as well as the EERC test results. It should be noted that the EERC tests were obtained from static tests. Static tests were also performed at ETEC. The static and dynamic stiffnesses

-171- at 100 percent shear strain and 140 kip axial load are compared in Table 2. Figure 5 also includes the predicted horizontal stiffness based on coupon tests and Figure 6 includes the damping obtained from coupon tests. In general the bearing dynamic damping is higher than the static damping and both are lower than the damping obtained from coupon tests.

Table 2 Comparison of Static and Dynamic (0.75 Hz) Stiffnesses Axial Load-140 kips

DOWELLED BOLTED STATIC TESTS 9.1 8.2 DYNAMIC TESTS 8.8 8.1

The effect of cycling was also investigated by testing some bearings up to SO cycles at 0.75 Hz and a displacement corresponding to 100 percent strain. The horizontal stiffness and damping values were relatively unchanged after the third cycle. Fatigue tests performed on other similar bearings are reported in [10,14] show that the bearings can withstand over 1120 cycles of ISO percent strain before failure.

Self-Centering Tests

One concern with seismic isolation systems has been whether they will remain effective after a large earthquake such as the SSE and continue to provide protection from aftershocks. To satisfy this condition, it would be necessary to demonstrate that the bearings self-center after imposing a large displacement while the design axial load is maintained. To simulate this condition a stack of two dowelled bearings was placed in the testing machine, and the vertical design load was applied and maintained throughout the test. The bearings were then displaced horizontally, and after reaching the required displacement, the actuator was retracted and the residual displacements in the bearings were measured. These tests were performed for two displacement levels: 4.0 in. (SS percent shear strain) and 7.2 in. (100 percent shear strain). These were representative of an SSE and a beyond SSE event The initial residual displacements from these tests were 0.3 in. and 0.6 in. After approximately 200 seconds the displacements were further reduced by SO percent These tests were only performed for the dowelled bearings. However based on tests done in Japan on bolted bearings, it is expected that they will perform as well as the dowelled bearings. These tests demonstrate a major advantage that these type of bearings have over systems which use hysteretic dampers such as lead plugs. Tests have shown that the hysteretic elements would prevent the bearings from self-centering [5].

Shear Failure Tests

For these tests, the bearings were subjected to increasing horizontal displacements up to failure while maintaining a constant vertical load. Failure for the dowelled bearings was defined as roll-out, the displacement beyond which the applied horizontal load started

-172- decreasing with increasing displacement For the bolted bearings, failure was defined as physical rupture. Two dowelled bearings, and three bolted bearings were tested. For the dowelled bearings, the tests were repeated three times with an axial load of 140 kips, 210 kips, and 560 kips. The bolted bearings were tested under 14 kips, and 140 kips. The results of these tests are summarized in Figure 8. Roll-out for the dowelled bearings was observed at about 200 percent shear strain. This was independent of the magnitude of the vertical load. No visible damage was observed at die end of these tests. It should be noted that if lower vertical loads are applied, roll-out will occur at a smaller strain. For example in the tests performed at EERC, when the bearings were tested with a 70 kip vertical load, roll-out was observed at 160 percent strain. For the bolted bearings, the average failure strain was 255 percent It should be noted that these bolted bearings had been previously tested at EERC to 200 percent strain. The hysteresis loops from the failure tests for both types of bearings are shown in Figure 9. For comparison purposes the hysteresis loops obtained from the tests, but for a different axial load are shown in Figure 10.

300-

250-

\m _ _ D Bolted No. 1 -200- I 1 H Bolted No. 2 M 150- H Bolted No. 3 5 ui i 1 D Dowelled No. 1 1W) S " 1 I • Dowelled No. 2 50-' 1 0-r 1 14 140 210 560 AXIAL LOAD (KIPS)

Figure 8 Summary of ETEC Shear Failure Tests

Tension Failure Test

One bearing was tested in tension to failure. Failure was initiated at a load of 165 kips and a displacement of 1.2 in. (2.4 in. in full-scale). The load translates into a tensile stress of 400 psi based on the steel shim diameter. The test was continued up to a displacement of 3.0 in. At this displacement the tensile force was 50 kips. Analysis results show that the maximum uplift in a corner bearing with a 2.0g input earthquake is less than 0.2 in. [19] which is considerably less than the tensile strain required to cause failure.

-173 s e _—-* — i. ^- ' 2g ^^

* / oow ILLED 55

CD • \

E I D L, U. it / M 1 1 o / BOLTED g / i

-ISO.0 0 [/ •zo -ie -is -\\ -12 -10 -B 01SPLRCEMENT

Figure 9 Comparison of ETEC Failure Hysteresis Loops

0. 2. 4. 6. 8. 10. 12. 14. 16. DISPLACEMENT (INCHES)

Figure 10 Comparison of EERC Large Strain Hysteresis Loops

-174- 5.0 SUMMARY AND CONCLUSIONS

The EERC and ETEC test results and observations confirm the design characteristics of the ALMR seismic isolation bearings. Additionally, the performance of the bearings for both dowelled and bolted configurations exceed the goals for design margins. The following are some relevant observations:

• In the range of axial loads that were investigated, the horizontal stiffness and damping were not appreciably affected by changes in the axial load • The effect of loading frequency on the horizontal stiffness and damping is small. Consequently, when full-size bearings are tested, the design characteristics can be evaluated from static tests. • There were some differences between die EERC and ETEC tests. These may be attributed to the following reasons: • The EERC test results were the average of four bearings; the ETEC test were for individual bearings. • The bearings tested at the two facilities were from two lots manufactured at different times. • The particular testing methodology and instrumentation used at each tests facility. • The ETEC test results display larger than acceptable variation in stiffness between bearings of the same type. It is expected that this variation will be much smaller if current bearing purchase specifications are used. For example, in a recently completed non-nuclear application, 64 high damping elastomeric bearings of the same size were procured from a U.S. manufacturer using the revised specifications. Tests performed on these bearings demonstrated that the stiffnesses of all the bearings were within ±10 percent of the mean stiffness. • Properly manufactured bolted bearings have higher horizontal displacement margins than dowelled bearings. Furthermore, the increase in horizontal stiffness at high strains observed in bolted bearings is a desirable feature for reducing displacements for beyond design earthquakes.

REFERENCES

1. Derham, C. J., Kelly, J. M., and Thomas, A. G., "Nonlinear Natural Rubber Bearings for Seismic Isolation," Nuclear Engineering and Design, 84(3): 417-428,1985. 2. Energy Technology Engineering Center, "Seismic Isolation Test Results Part I: PRISM ALMR Dowelled Bearings," Report to US. Department of Energy, February, 1991. 3. Energy Technology Engineering Center, "Seismic Isolation Bearing Test Results Part II: PRISM ALMR Bolted Bearings," Report to US. Department of Energy, March, 1991. 4. Fan, F.-G., and Ahmadi, G.," Responses of Equipment in Base-Isolated Structures Under Earthquake Ground Motions," Seismic, Shock, and Vibration Isolation, PVP Vol 200, ASME 1990.

-175- 5. Fujita, K., ct al., "Dynamic Characteristics of Elastomer with Lead Plug," Seismic, Shock, and Vibration Isolation, ASME PVP-Vol. 181,1989. 6. Fujita, S., et al., "Earthquake Isolation Systems for Buildings of Industrial Facilities Using Various Types of Damper," 9th World Conf. in Earthquake Engineering, Tokyo, Japan, August, 1988. 7. Kelly, J. M., and Tsai, H. G, "Seismic Response of light Internal Equipment in Base Isolated Structures," Report No. UCBIEERC- 84/17, Earthquake Engineering Research Center, University of California, Berkeley, 1984. 8. Kelly, J. M, Buckle, 1G., and Tsai, H. C, "Earthquake Simulator Testing of Base - Isolated Bridge Deck," Report No. UCBIEERCSSI09, Earthquake Engineering Research Center, University of California, Berkeley, 1986. 9. Kelly, J. M.f Tajirian, F. F., Gluekler, E.L. and Veljovich, W., "Performance Margins of Seismic Isolator Bearings," Int. Fast Reactor Safety Meeting, Snowbird, Utah, August, 1990. 10. Kelly, J. M., "Mechanical and Failure Characteristics of High Damping Natural Rubber Isolation Bearings from Four Different Test Programs," Post-SMiRT 11 Conference on Seismic Isolation of Nuclear and Non-Nuclear Structures, Nara, Japan, August, 1991. 11. Kelly, J. M., "The Current Status of Seismic Isolation Technology in the United States," IAEA Specialists'Meeting on Seismic Isolation Technology, San Jose, CA, March, 1992. 12. Kulak, R. F., and Hughes, T. H., "Mechanical Characterization of Seismic Base Isokrion Elastomers," Trans, of 11th SMiRT Conference, Vol. K2, Tokyo, Japan, 1991. 13. Nakazawa, M., et al., "Study on Seismic Isolation of LWR Plants (Durability Tests of Laminated Rubber Bearings)," Trans, of 11th SMiRT Conference, Vol. K2, Tokyo, Japan, 1991. 14. Seidensticker, R. W., et al., "Summary of Experimental Tests of Elastomeric Seismic Isolation Bearings for Use in Nuclear Reactor Plants," IAEA Specialists' Meeting on Seismic Isolation Technology, San Jose, CA, March, 1992. 15. Skinner, R. T., Robinson, W. H., and McVerry, H. R, "Seismic Isolation in New Zealand," Proc. of First Int. Seminar on Seismic Base Isolation of Nuclear Power Facilities, Post SMiRT 10, San Francisco, CA, 1989. 16. Tajirian, F. F., Kelly, J. M., and Gluekler, E. L., 'Testing Of Seismic Isolation for the PRISM Advanced Liquid Metal Reactor Under Extreme Loads," Trans, of 10th SMiRT Conference, Vol. K2, Anaheim, CA 1989. 17. Tajirian, F. FM Kelly, J. M., and Aiken, L D., "Seismic Isolation for Advanced Nuclear Power Stations," Earthquake Spectra, Vol. 6, No. 2, EERI, May, 1990. 18. Tajirian, F. F., et al., "Elastomeric Bearings for Three-Dimensional Seismic Isolation," Seismic, Shock, and Vibration Isolation, ASME PVP - Vol. 200, 1990. 19. Tajirian, F. F., "Seismic Analysis for the ALMR,"/AEi4 Specialists'Meeting on Seismic Isolation Technology, San Jose, CA, March, 1992. 20. Yoneda, G., et al., "Experimental Study on Radiation Resistant Properties of Seismic Isolation Elements," Trans, of 11th SMiRT Conference, Vol. K2, Tokyo, Japan, 1991.

-176- The submitted manuscript has been authored by a contractor of the U. S. Government under contract No. W-31-109-ENG-38. Accordingly, the U. S. Government reuins • nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for U. S. Government purposes.

Summary of Experimental Tests of Elastomeric Seismic Isolation Bearings for Use in XA0055385 Nuclear Reactor Plants

R. W. Seidensticker, Y. W. Chang, and R. F. Kulak

Argonne National Laboratory Argonne, Illinois USA

Abstract

This paper describes an experimental test program for isolator bearings which was developed to help establish the viability of using laminated elastomer bearings for base isolation of nuclear reactor plants. The goal of the test program is to determine the performance characteristics of laminated seismic isolation bearings under a wide range of loadings. Tests were performed on scale-size laminated seismic isolators both within the design shear strain range to determine the response of the bearing under expected earthquake loading conditions, and beyond the design range to determine failure modes and to establish safety margins. The test program was funded by the U.S. Department of Energy (DOE) and the National Science Foundation (NSF).

Three types of bearings, each procured from a different manufacturer, have been tested: (1) high shape factor-high damping-high shear modulus bearings; (2) medium shape factor-high damping- high shear modulus bearings; and (3) medium shape factor-high damping-low shear modulus bearings. These bearings were designed by ANL and made according to ANL specifications. ( All of these tests described in this report were performed at the Earthquake Engineering Research Center (EERC) at the University of California, Berkeley, with technical assistance from ANL. In addition, tests were performed by ETEC on two of the type (1) bearings cited above. Those tests will be reported by others at this workshop.

The tests performed on the three types of bearings have confirmed the high performance characteristics of the high damping-high and low shear modulus elastomeric bearings. The bearings have shown that they are capable of having extremely large shear strains before failure occurs. The most common failure mechanism was the debonding of the top steel plate from the isolators. This failure mechanism can be virtually eliminated by improved manufacturing quality control.

The most important result of the failure test of the isolators is the fact that bearings can sustain large horizontal displacement, several times larger than the design value, without failure. The results of these test programs should give the designer confidence that base isolated structures

-177- can be designed and built with more than adequate safety margins. Their performance in moderate and strong earthquakes will be far superior to conventional structures.

I. Introduction and Scope

This paper presents an overview of the results of a number of tests of elastomeric seismic isolation bearings conducted over the last three years. These tests were conducted as part of the U.S. Department of Energy (DOE) program to develop seismic isolation technology for use in nuclear facilities, and also as part of a joint U.S./Japanese program to study seismic isolation. The joint program was conducted by Argonne National Laboratory (ANL) and the Shimizu Corporation of Japan. The ANL work on this joint program was funded by the National Science Foundation (NSF).

The goals of these tests are to determine the performance characteristics of isolator bearings under a wide range of loadings, to determine failure modes and to establish safety margins. During the course of conducting these tests, three types of elastomeric bearings were tested. The results of these tests are presented along with conclusions relating to bearing design and manufacture, acceptance testing and margins of safety against failure.

II. Background

ANL has been deeply involved in the development of seismic isolation for use in nuclear facilities for the past decade. Under funding and direction of DOE, ANL has participated in the development of the methodology needed to design and evaluate usefulness and effectiveness of seismic isolation for advanced liquid metal-cooled reactors (LMRs). As part of that program, tests were conducted on elastomeric bearings for DOE by the Earthquake Engineering Research Center (EERC) at the University of California, Berkeley, California and at the Energy Technology and Engineering Center (ETEC) near Los Angeles, California. The results of those tests are reported in other papers at this workshop and elsewhere.

More recently, ANL had a major role in evaluating the use of seismic isolation for the DOE New Production Reactors (NPR), and also worked with Shimizu Corporation on the joint U.S./Japanese program to study seismic isolation of a full size test building facility in Sendai, Japan.

As part of these programs, ANL procured over fifty elastomeric isolation bearings for use both in the Sendai test building facility and for individual bearing tests. The results of these individual bearing tests are summarized in this paper.

Three types of high-damping rubber bearings, each procured from a different manufacturer, have been tested: (1) high shape factor-high shear modulus bearings; (2) medium shape factor-high shear modulus bearings; and (3) medium shape factor-low shear modulus

-178- bearings. The bearings were designed by ANL and made according to ANL specifications. All tests described in this report were performed at the EERC with technical oversight by ANL. The tests at EERC were conducted on either the Single Bearing Test Machine, or the Large Scale Test Machine. Descriptions of these test facilities are given in Refs. (1) and (2). The bearing configurations, test plans, and test results are described in the following sections. This material is based on detailed test reports prepared by EERC.

III. High Shape Factor-Hieh Shear Modulus Bearings

A. Background Information

The high shape factor-high shear modulus bearings are made of a high damping- high shear modulus rubber compound. They have a high shape factor value of 24. The bearings were designed and constructed as "dowelled" bearings to avoid tension in the rubber.

ANL purchased sixteen (16) bearings for the joint ANL/Shimizu program. Eight bearings were shipped to Japan for installation at the Sendai test facility and for performance of laboratory tests at Shimizu Corporation; four bearings were sent to EERC for testing to determine their static and dynamic characteristics; two bearings were sent to ETEC for dynamic tests; and two bearings were retained by ANL for archival purposes.

B. Description of Bearings

Figure 1 shows the configuration of the Sendai bearing. The overall diameter of the bearing, including a 3/4 inch protective cover layer, is 20 inches. The bearing has 1 inch thick end plates located at its top and bottom with four drilled holes in the top plate used for dowel pins.

C. Test Plan

The ANL test program at EERC was specifically designed to replicate insitu field tests conducted in Japan on the full size building after the bearings were installed. The bearings were tested in the Large Scale Bearing Test Machine at EERC.

The Sendai seismically-isolated test building weighs a total of 255.4 metric tons and is supported by six bearings. It is estimated that the four corner bearings carry a load of 37.8 metric tons each, and the two middle bearings carry 52.3 tons each. These loads correspond to 83.2 kips and 115.1 kips, respectively.

The four bearings were placed in the test machine with the vertical load set to 83.2 kips. Each horizontal displacement cycle (±3.15%, ±6.25%, ±12.5%, ±25%, ±50%, +75% - 50% shear strain) was repeated three times and data collected for all three cycles. The vertical load was increased to 115.1 kips and the test sequence repeated. The bearings were then deformed to

-179- displacement cycles of ±100% and ±125% at the larger vertical load. It was observed that one bearing was showing signs of distress at this point. This bearing and its partner on that side of the test machine were immobilized by a strong-back, and the testing was continued with displacement cycles corresponding to ±150%, ±200% shear strain. No further evidence of damage was visible, and it was concluded that the two tested bearings performed satisfactorily to ±200% strain. (The distressed bearing was cut in half after the test and it was found that the bond between the rubber and the top end plate had failed.)

D. Test Results

Test results are given in detail in ANL/Shimizu report, ANL-002 [Ref. 1]. Figures in this section are taken directly from Ref. 1. Major test results are force-displacement loops and damping values. The force-displacement (hysteresis) loops were recorded for the bearings at the two vertical loads, 83.2 kips and 115.1 kips. The results for the 115.1 kip vertical load is shown in Fig. 2. Figure 3 shows shear modulus as a function of strain obtained from the third cycle of the bearing tests and from the 0.75 Hz coupon tests provided by the elastomer compound supplier. The damping values obtained from the bearing and coupon tests are given in Fig. 4.

As can be seen from Fig. 3, at 50% shear strain the shear modulus of the rubber is about 120 psi. With a building weight of 255.4 metric tons, the bearing's frequency at 50% shear strain is 0.79 Hz. This is higher than the design frequency of 0.75 Hz. In fact, the frequency of 0.75 Hz would not be achieved for the Sendai bearings until the shear strain reached around 100%.

E. Conclusions

The results indicate that the bearings are somewhat stiffer than anticipated in the design. The system frequency at 50% strain is 0.79 Hz. For small displacements generated by minor earthquakes, the frequency will be even higher. For example, at around 10% strain, the response will be of the order of 1.2 Hz. The damping in the bearings is also consistently high. Over the range of strains, from 3% to 50%, the average damping is 15.5%. Thus, for small to moderate earthquakes, structures using these high-shape factor bearings will not experience significant reductions in horizontal seismic loads.

The bearing that showed early signs of distress experienced a bond failure that occurred at the top end plate. While observations of the type of failure show that the failure is not sudden, they also highlighted the difficulty of detecting a poor bond in bearings. The testing procedure of 50% shear strain (at 100% of design vertical compressive load) as an acceptance test did not reveal the presence of a poor bond. The test procedure was modified for later procurements to require the test to subject the bearings to large horizontal displacements while the vertical load on the bearing is reduced to essentially zero, thus eliminating the frictional effect which may mask defects in the bonding.

-180- 5 IV. Medium Shape Factor-High Shear Modulus Bearings

A. Background Information

The motivation to design and test medium (or moderate) shape factor bearings include a desire to achieve a bearing design that facilitates fabrication to high quality standards and has the highly predictable behavior needed for nuclear applications. Also, this particular test program was very important to the DOE's seismic isolation program, because it provided data on a range of shape factors lying between the earlier bearings of GE's PRISM and Rockwell International's SAFR reactors tested at EERC, which were high and low shape factor bearings, respectively.

Four types of rubber bearings were manufactured. The bearings were designed as "bolted"-type bearings (i.e., the bearing is able to develop tensile stresses in the rubber during horizontal deformations). The four types of isolators have two different rubber compounds and two different shape factors, 9 and 18.

Each type of bearing was subjected to seven different types of tests, making a total of 28 bearings tested. Each type of isolator was subjected to five different failure tests in which combinations of the axial and the shear loads were varied in order to obtain an insight into effects of axial-shear load interaction. Fatigue tests were carried out at two different shear strain levels: one at 100% and the other at 150%.

B. Description of Bearings

The four types of bearings used in this test series were designated as types 1, 2, 5 and 6. The dimensions and overall arrangement of bearing type 1 is shown in Fig. 5. (Details of the other bearing types are similar.) A summary of the dimensions and properties of the four types of bearings is given in Table I. As shown in Table I, two different rubber compounds were used (designated here as A and B), and two different shape factors were used, 9 and 18. The overall height of the bearings varied slightly to maintain the overall rubber thickness at 3.75 inches.

The compound used in types 1 and 2 bearings, designated A, has a nominal shear modulus of 120 psi at 100% shear strain; the compound used in types 5 and 6 bearings, designated as B, has a shear modulus of 150 psi at 100% shear strain. The nominal design vertical load for the bearings was 62.5 tons. These isolators were designed to provide a horizontal frequency of 0.50 Hz at 100% shear strain.

C. Test Plan

The seven tests performed on each type of bearing were fatigue and combination failure tests. The purpose of the fatigue tests was to study (1) the resistance of the isolators to a long time duration of cyclic dynamic loads, (2) the loss of stiffness under cyclic loading, and

-181- (3) the failure mechanisms. The purpose of the combination failure tests was to gain insight into an axial-shear load interactions.

There were five combination shear-axial load failure tests performed on each type of bearing for a total of 20 tests. The five tests are: (1) pure compression failure, (2) shear failure with a compressive design vertical load, (3) shear failure with no axial load, (4) shear failure with a tensile design vertical load, and (5) pure tension failure.

The original test program for the fatigue tests was to test each type of bearing to a cyclic displacement corresponding to 50% and 100% shear strains until failure while carrying the design load of 62.5 tons (125 kips). However, after the first fatigue test was performed, bearing type 5 resisted 2880 cycles without a sign of failure. A decision was made to modify the test program to shear strain levels of 100% and 150% and only bearing types 1 and 5 were tested. Thus, there were a total of five fatigue tests.

D. Test Results

Test results important to establish dynamic characteristics and failure mechanisms of the bearings are summarized below.

1. Fatigue Tests

During the five tests, only the two bearings subjected to cycles at 150% shear strain failed. One bearing tested in fatigue at 150% shear strain sustained up to 1120 cycles before failing. Another bearing sustained 515 cycles due to a fissure caused by the expulsion of an air bubble entrapped in the bearing. Figures 6 and 7 are photographs of the two bearings after failure. (Note that Fig. 7 shows only the lower portion of the failed bearing after removal from the test machine.)

The shape factor has no apparent influence on the mechanical behavior of the isolators when subjected to cyclic loading. Similarly, the type of rubber compound had no apparent influence on the mechanical behavior of the isolators when subjected to a fatigue test.

2. Combination Failure Tests

The combination failure tests gave a variety of results which depended to a large extent on the quality of the bond between the steel plates and shims and the rubber layers. For example, some of the isolators carrying high compressive load and tested in shear did not fail, but when the axial load was removed and tested again in shear, they did fail. This suggests that weak bonding was the cause of the failure.

In general, for tests of shear failure with compressive vertical load, the maximum shear strain reached by the isolators surpassed 250% shear strain, and all tests surpassed 175% shear strain.

-182- As expected, bearings with higher shape factor resisted higher compression loads than those with low shape factor bearings, since bearings with higher shape factor require more layers and consequently more steel shims which resist most of the vertical load. The compression tests gave consistent results.

It was possible to test all bearings to failure during the test of shear failure with no axial load. This test was useful in determining the quality of the bonding between the steel shims and plates with the rubber layers.

The results of the tension failure tests were varied, with the tensile strength ranging from 111 psi up to 586 psi. The variability in tensile strength is due to variability in bond strength between the steel plates and shims and rubber layers.

E. Conclusions

These bearings performed well during the fatigue tests. The bearings tested in fatigue at 150% shear strain sustained up to 1120 cycles before failing (equivalent to 30 minutes of very severe ground shaking). These results are more than satisfactory when considering that no earthquake will generate so many cycles all with large displacements. For example, an earthquake of two (2) minutes duration (which is very high) would have about 200 cycles, only a fraction of which cycles would be of the higher displacement motion.

The compression tests revealed that the mechanism of failure is yielding and tearing of the reinforcing steel plates. The plates are loaded by surface shear stresses as they act to prevent barrelling of the bearing under the vertical load. It was demonstrated that the elastomer and the rubber steel bond are able to sustain shear stresses that are large enough to yield the steel in a type of all around tension. The vertical pressure on the bearing, at which the steel reinforcing plates failed, depended on the shape factor but was in every case many times larger than the design vertical pressure.

The program has shown that adequate bond can be achieved if correct quality assurance procedures are followed during the molding and vulcanizing of the bearings. It has also been shown that bonding cannot be determined by tests with a horizontal shear strain of 100% under the presence of any vertical compressive load. One must test under zero axial load or under a slight tension load (to minimize pseudo-beneficial effects of friction).

V. Medium Shape Factor-Low Shear Modulus Bearings

A. Background Information

After experiencing 37 minor to moderate earthquakes in 1-1/2 years, ANL and Shimizu Corporation decided that the high shape factor-high damping isolation bearings installed in the Sendai test facility should be replaced by a set of medium shape factor-low shear modulus

-183- 8 - high damping isolation bearings. This was done to achieve improved response of the building over a wide range of small to moderate earthquakes.

A new compound had been developed which had a shear modulus about half that of the shear modulus of the high damping rubber used in the original test bearings. In July 1990, ANL purchased ten bearings using the newly developed compound. Eight bearings were shipped to Japan in October 1990: six for installation in the Sendai test building and two for laboratory tests in Japan. The two remaining were sent to EERC for laboratory testing.

B. Bearing Configuration

These bearings are characterized as having medium shape factor-high damping-low shear modulus. These bearings have a diameter of 14.75 inches and are 8.1 inches in height. They have 12 layers of 0.394 inch thick rubber and 11 steel shims which are 0.128 inches thick. Details of these bearings are described elsewhere at this workshop (see paper by R. F. Kulak, "Technical Specifications for the Successful Fabrication of Laminated Seismic Isolation Bearings"). With a nominal vertical load of 92.6 kips corresponding to 621 psi vertical pressure, these bearings provide a horizontal frequency of 0.40 Hz at 100% shear strain.

C. Test Plan

The test plan consists of three types of tests: (1) horizontal tests, (2) vertical tests, and (3) failure tests. All tests were carried out on the Single Bearing Test Machine at the Earthquake Simulator Laboratory of EERC.

In the horizontal tests, the bearing was subjected to two sequences of horizontal displacement cycles. Each sequence included three cycles of displacement of the strain levels:

Sequence 1: ±5%, ±10%, ±25%, ±50%, ±75%, ±100%

Sequence 2: ±125%, ±150%, ±175%, ±200%

In the vertical tests, the bearing was subjected to five cycles of vertical loading centered around 500, 625, 1000, and 1500 psi of vertical pressure. The variation in the vertical load in each of these tests corresponded to about ±100 psi. This particular test series is close to the limits of the test system capability at EERC.

The horizontal tests in Sequence 1 were carried out at frequencies of 0.1 Hz and 0.5 Hz. The horizontal tests in Sequence 2 and the vertical tests were carried out only at a frequency of 0.1 Hz.

After the completion of the horizontal and vertical tests a failure test was performed. In the failure test, the bearing was loaded monotonically in shear at a rate of 2.5 in/min.

-184- D. Test Results

Test results are discussed in detail in Ref. [2]. Tables II gives the effective stiffness and equivalent viscous damping for tests performed at a frequency of 0.5 Hz. The force displacement plot for Sequence 1 tests at frequency of 0.5 Hz with 1000 psi vertical pressure is shown in Fig. 8.

Table III gives the stiffness values of the bearing for five cycles around 500 psi, 621 psi, 1000 psi, and 1500 psi. The variation of pressures is ±100 psi. The hysteresis loops are shown in Fig. 9 where they are superimposed on the monotonic loading curve. As can be seen from Table III, the stiffness value varies with the pressure level around which the cycles are centered. The stiffness values are higher at high pressure levels. Figure 9 shows that the slopes of the hysteresis loops due to pressure variation are much higher than the tangent stiffness of the monotonic loading curve.

In the failure test the bearing was loaded monotonically at a rate of 2.5 in/min. In all three trials, the bearing did not fail when the actuator reached its maximum displacement. Figures 10 and 11 show the deformation of the bearing at shear strain of 200% and 300%, respectively. The maximum shear strain obtained in the test was 415%, with a maximum shear stress of 355 psi. This is equivalent to a maximum displacement of 19.6 inches with a maximum shear load of 52.9 kips.

E. Conclusions

The results of horizontal shear tests show that the shape of hysteresis loops change with the magnitude of vertical pressure loads. The loops change from narrow-elongated to wide- elliptical as vertical pressure increases. The results of vertical tests show that the bearing stiffness varies with the pressure level around which the cycles are centered and they are much higher than the tangent stiffness of the monotonic loading curve. The most important results of the failure test is the fact that the bearing was able to sustain very high horizontal shear displacements while under the design vertical load.

V. Summary

The three test programs summarized in this paper have confirmed the high performance characteristics of the high damping-high and low shear modulus elastomeric bearings. The tests have shown that these bearings are capable of experiencing extremely large shear strains (several times the design shear strain level) before failure occurs. Failure mechanisms of these bearings are strongly influenced by the quality of the bond between the steel plates and shims and rubber layers. Bearings tested in fatigue at 150% shear strain sustained up to 1120 cycles before failing. This is roughly equivalent to about 30 minutes of extremely strong ground shaking. These results are more than satisfactory when considering that no earthquake will generate so many cycles with such large displacements.

-185- 10

VI. Acknowledgments

The authors wish to thank Professor James M. Kelly, Dr. Ian D. Aiken, and Messrs. Wes Neighbors, Donald Clyde, Edgardo Quiroz, and Ivo Van Asten of the Earthquake Engineering Research Center (EERC) for performing and analyzing the bearing tests and preparing test reports. These tests were performed as part of work in the Engineering Mechanics Program of the Reactor Engineering Division of Argonne National Laboratory under the auspices of the U.S. Department of Energy, Contract No. W-31-109-Eng-38. The funding of the experimental tests was partially provided by the National Science Foundation under NSF Agreement No. CES- 8800871.

VII. References

1. James M. Kelly, Ian D. Aiken, and Donald Clyde, "Performance Evaluation of ANL Sendai Bearings," ANL/Shimizu, ANL-002, September 1989.

2. James M. Kelly, "Mechanical Characteristics of Low Modulus High Damping Natural Rubber Isolators for a Base Isolated Demonstration Building," ANL/Shimizu, ANL-004, June 1991.

-186- Table I. Dimensions and Properties of Medium Shape Factor Bearings

MEDIUM SHAPE FACTOR BEARING TEST BEARINGS DIMENSIONS AND COMPOUNDS TOTAL RUBBER THICKNESS: 3.75" SINGLE STEEL SHIM THICKNESS: 1/16" STEEL SHIM RUBBER BEARING SHAPE COMPOUND DIAMETER AREA (IN2) LAYERS THICKNESS LAYERS I TYPE FACTOR TYPE 00 1 14" 153.94 3/8" 9 A I 9 10 2 14" 153.94 19 3/16" 20 18 A 5 12" 113.10 11 5/16" 12 9 B 6 12" 113.10 23 5/32" 24 18 B Table II. Equivalent Stiffness and Viscous Damping of Low Shear Modulus Bearing for Tests Performed at 0.5 Hz Frequency

HORIZONTAL SHEAR TEST FREQUENCY: 0.5 Hz SHEAR STRAIN PRESSURE1 FILENAME DYNAMIC PSI (5-100%) PROPERTIES 5% 10% 25% 50% 75% 100%

K(ea) (Kips in) 3.64 3.14 2.61 2.21 1.97 1.83 0 910325.10 P (%) 13.69 11.72 9.66 8.82 8.27 8.02 G (psi) 115 100 83 70 62 58

These pressures are the target pressures, actual values are indicated in the plots. Table III. Stiffness of Low Shear Modulus Bearing at Various Pressure Loadings SENDAI - BEARING TEST COMPRESSION TEST STIFFNESS (Kips in) PRESSURE1 AP FILENAME PSI PSI 1ST CYCLE 5TH CYCLE HERTZ 910325.15 500 ±100 1114.11 1114.12 12.10 910404.04 500 ± 100 916.88 914.16 10.99 910325.16 621 ± 100 1151.32 1149.26 10.84 910404.05 621 ± 100 1059.80 1054.51 10.56 910325.17 1000 ± 100 1697.81 1657.7 10.47 910404.06 1000 ±100 1621.35 1588.43 10.21 910325.18 1500 ±100 3878.29 3820.09 13.03 910404.07 1500 ± 100 2545.18 2635.64 10.56

These pressures are the target pressures, actual values are indicated in the plots.

-189- • 11 UNC-2A through - Cover layer deep in top plate 11" dia 18.5" dia Shim

20" O.D.

2" dia hole filled with 62 duro natural rubber

33layers© 0.191" 62 duro natural rubber

14Ga Shim (A570 Gr40) Typ.32 l"xl8.5"dia(A36)

13/ 1 /16" Detail B

Fig. 1. Design Detail for Sendai Bearing

-190- 5RIJDM - 11RA RINGS

static vcrUlcnl lond - 115.31 klps/bearlng

IlorlzonUnl. J.ontl/d'o£.IcctJ.on 60

o a

-20

-40

-60 -n -6

0 l:iplnc;omrsiil: J.nchns

Fig. 2. Force-Displacement Loops for Bearing, Vertical Load = 115.1 kips S A; in "°! cs r- in • © CO n S o 4J *— o 0) CN 4) H P i-t .s 0

c: tr a H § t H CO u - o W D •0 a O O © 1 2 n H CO % "^ w Crf CO Cn I w C M IS £-* a) i o m CO

CO bb

o o o o o o o o o cs o

-192- ANL-SENDM BEARING TESTS

DAMPING FACTOR vs. SHEAR STRAIN Bearing and Coupon Tests

o Coupon Test (0.005 Hz)

32 Coupon Test (0.75 Hz) D A V 3rd cycle (P-115.1k) M P I 3rd cycle (P-03.2k) N 24 G F A C T 0 16 R

50 100 150 200 250

STRAIN -

Fig. 4. Damping Factor vs. Shear Strain for Bearing and Coupon Tests T-MUNS-28 I" DEEP (12 PLACES) 3/«"-llUNC-!S 9/s" DEEP (4 PUCES)

10 D1JL B.C. II CIJL B.C

• END PLATE- 14" 0.0. •

-BEARING- 16* O.D. —

I OIA. KOLE.HLLEO END PLATE (2RE0.I* ni ) ' RUE8ER

I 1 \ 1 t I-l/S"

7-1/16" » 3/a* LATt 4-5/16' 9 1/16' JKIU5 I V/////A/M\ Tl 1/4" / 1/4'R

SHOWN OUT OF SECTION A-A POSITION FOR CLARITY

ARGONNE NATIONAL LABORATORY DATE: February 22, 1990 DWG. NO.: NPR-RAS-001-B V o, ^

Fig. 5. Design Details for Test Bearing Type 1

-194- Fig. 6. Bearing Type 1.07 - Fatigue Test at 150% Shear Strain Test No. 1

Fig. 7. Bearing Type 5.07 - Fatigue Test at 150% Shear Strain Test No. 2 SENOM-DEAnillG TEST SHEAR DIAGNOSTIC TEST S% - 100% STRAIN Strain Stiffnoao Shoar Modulus Damping 5% 5.39 Kipa/in 171 pel 17.81 % FILENAME I 910325.13 lot 3.82 121 23.03 25* 2.29 72 27.77 VERT PRESSURE I 1000 PSX 50% 1.56 49 27.49 FREQUENCY! 0.5 HERTZ 75% 1.2B 41 26.68 100% 1.16 37 25.54 Horizontal Load vn Dioplacomont 10

L o a d en K i I P S

-5

-10 -6 Dioplacomont Inchon

Fig. 8. Force Displacement Plots for Sequence I - 1000 psi Vertical Pressure at Frequency 0.5 Hz SENDM -BEARING TEST COIIPnESSIOH TEST Fllonnmai 910404.03 Max.Vert.Loadi 232.36 Xipn

Vortical Load vn Vertical Deflection 250

1500 p$I

200

150 _ J000 psi o a d I I—• to K 625 psl I I S

500 psl

0.0 0.1 0.2 0.3 0.4

Dofleotion lnclten

Fig. 9. Superimposition of Cyclic Vertical Tests on Monotonic Vertical Compression Test - for Low Shear Modulus Bearings Fig. 10. Bearing Appearance at 200% Strain

Fig. 11. Bearing Appearance at 300% Strain

-198- XA0055386

IWGF2P 4IWGFR San Jose, USA March 1992

DRAFT

Some Difference of Concepts between Design Guideline for FBR Base Isolation System and Aseismic Design Guideline of LWR in Japan

SHIBATA, Heki Professor-Dr. Institute of Industrial Science, University of Tokyo 22-1, Roppongi 7, Minato, TOKYO 106, JAPAN

after April 1, 1992 Faculty of Engineering Yokohama National University 156 Tokiwadai, Hodogaya, Yokohama 240, JAPAN

This draft was prepared for the 4th Int. Working Group for Fast Reactor, IAEA. Because of the limitation of the avairable time for the author, neither reference, figure nor table is attached. However, the figures and so on intended to be prepared by the author is conceptual ones, and the most of originals appear in the reports by CRIEPI during this meeting or in the future. Or they were presented in other previous papers by the author, or in JEAG 4601-1984, 1987 or 1991 as mentioned in Section 1.

Abstract

This paper deals with the concept and the relation of "the Base Isolation System and FBR" to the Safety Criteria and the Guideline of the Aseismic Design of LWR in Japan. The Central Research Institute of Electric Power Industries have been working for FBR last several years. The author has been contribute to their works, and this is one of the subjects. He described his own idea obtained through the cooperative work with CRIEPI.

-199- §1 Introduction

It is not clear for the author whether ot not the design guideline should be prepared before to start the design of some actual systems. The draft of the aseismic design guideline for nuclear power plants was prepared in 1960 before to start the design of LWRs, Tsuruga #1, JAPCO, and Fukushima #1, TEPCO. However, the guideline had never prepared until the Guideline of Aseismic Design of NPPs, JEAG-4601-1987 by Japan Electric Association in 1987. Even though the first JEAG-4601 was published in 1970, it was a text book style, and it didn't clearly indicate the design practice. And the Guideline for the Design Basis Earthquakes was issued by the Nuclear Safety Commission in 1981. In this paper, the author describes the concept and main idea on the design of the base isolation system and the aseismic design of base isolated nuclear power plants including equipment and distribution systems. First of all, some new items should be added to the safety criteria for light water plants. And it is important how to categorize the base isolation system according to the concept of "Factor of Importance" which has been used for the aseismic design of LWR. The third is the design spectrum. In Japan, Ohsaki spectrum is commonly used for LWRs and other nuclear facilities, however, the region of longer period component, that is, T = 1 sec or -longer is considered to be poor comparing to those of the actual ground motions. Ishida, CRIEPI, introduced a modified standard spectrum to their guideline. The author tries to discuss this subject. Add to these three subjects, he is briefly reviewing the difference of the details between the guidelines for FBR and LWR. The guideline for LWR has been prepared by the committee of the Japan Electric Association, which has been chaired by the author for these several years, as JEAG 4601-1987 and 4601-1991.

§2 Design Criteria in General

The design criteria of nuclear power plants are clearly established. As far as those of structural parts of LWR, they are described in the Design Criteria of LWR, which was issued by the Nuclear Safety Commission in 1978 and remodified in 1990. If we try to transfer them to those for FBR, it may be necessary to be modified added to those for LWR as follows: i) The base isolation system and their elements must be designed, fabricated, and inspected based on the related standards and guidelines. And all materials must be selected from those as listed in the related standards. ii) The whole system should be designed as the safety functions of the supported structure and the base isolation system must be kept against all kinds of natural events including seismic forces induced by the design basis earthquakes of level S, and S2. Other events, flood, tsunami, strong wind, freezing, snow, ground slide and so on must be considered. For the base isolation system, underground water also must be considered.

-200- iii) As the external events, aircraft crashing, failure of upper stream dam, explosion must be considered. Also, the access of the external persons must be controlled to avoid their action to destroy the base isolation system. iv) Internal missile must not fail the safety function of the base isolation system. Also the effects of failures of equipment and piping systems including their secondary effects such as fires, flood, chemical attack and mechancial shocks must be considered. v) The base isolation system must be protected against sodium and chemical product from sodium in the case of sodium leakage or flood. Also the effect of the chemical product in heat exchanger caused by the leakage of water and/or steam from their pipes must be considered. vi) Against the dangerness of fires, the base isolation system must be designed in consideration with the following three measures: the prevention, detection and extinguishment of fires. Materials of elements of the system should be selected from those which are not burnable or fire proof as well as possible. For their maintenance, the care must be taken to avoid the fire, such as, on electric power supply system and other fire sources. vii) Design and treatment of their elements must be done in consideration with their environments, such as temperature, moisture, salt powder and radiation under normal operating condition and abnormal conditions. If the control of their environment is done by air conditioning systems, their design or the effect of their failure must be considered according to the criteria. viii) The elements must be reliable, and this must be ensured at all times through by design, fabrication, inspection and test. ix) The base isolation system can be accessable to inspect and to test them during operation and inoperation conditions. x) They must provide the adequate measure to prevent the failure of control system and instrumentation system by thunder lightening, and to prevent the fire and other unsafe events caused by thunder lightening directly and/or indirectly. xi) The base isolation system and the distribution system, bridged between the isolated part and unisolated part, must be able to be inspected, maintained and repaired, or to be replaced if necessary. Even though, the eleven new items have been raised, there might be some unknown problems as well as the methods of analysis, design and fabricating the system itself. We must study on the practical model and method for hazard analyses of the whole system and its elements. For example, it is not clear for us the effects of the following natural or external events: i)aircraft crash, ii)wild fire, iii)thunder-lightning, iv)chemical explosion and v)flood. The direct effect of those is mainly fire. The flood may bring rock, sand and soil surrounding the system and bury it. These events are not considered enough in the criteria above, but it is clear that there are some difference between an ordinary designed NPP and the base-isolated NPP. However, it should be noticed that those

-201- differences don't come from the differences between FBR and LWR.

§3 Factor of Importance

In Japan, all items in NPP are categorized according to the factor of importance for their seismic design. This concept was originated by Housner or USAEC in late 50's, but now popular. Three or four categories have been applied to those in Japan since the second plant, that is the first LWR. Supporting structures are divided into two categories, that is, a direct supporting structure and an indirect structure. A typical direct supporting structure is a support for the primary coolant pipings. We usually call "a piping system", and this includes not only pipings themselves, but also hangers, snubbers, rods and supporting structures up to anchoring devices. Those, which consist of the system, are categorized as the same categories as As class of the primary coolant piping system. A typical indirect supporting system is the main reactor building. Many important items are involved in this building. As a rule, the main reactor building is not necessary to categorize in this sense, but it is recommended that it is categorized as the same category to the highest category of involved equipment and/or piping systems in the building. But, according to the situation, it may be deducted to the lower category, if the condition in the view point of the safety may allow it. Back to the base isolation system, it is considered to be an indirect supporting structure, except that for locally supporting a component like a computer. In the case of a floor isolation system, it is also categorized as an indirect supporting structure in principle. If we defined all panels and racks as one control system, and they are supported by one base isolation floor, we may say them as "a system", and the base isolation system would be a direct supporting system. But this definition usually may bring a confusion, because items supported by this floor are those categorized to various categories. The use of a base isolation system, which is discussed and planned to be used at this moment, is only for the main reactor building, and some computers and control devices including operators. Those are categorized As class or at least A class. In such cases, the base isolation system should be A class or As class. However, there is a discussion whether or not superstructure, especially, the main building is stronger than the system, as discussed later. In the sense of the categorization, both should be equally strong to the design basis earthquake.

§4 Design Basis Earthquake

The guideline for the aseismic design of nuclear power plants was issued officially by the Nuclear Safety Commission in 1981. According to this, the Ohsaki spectrum has been emploied to define the design basis earthquake in principle. This chart of response spectrum was prepared by Ohsaki and others based on -202- approximately fifty strong motion records originally, and modified in some extent later. However, it is rather poor in the region of longer period range, that is, longer than 1 second. Ishida, CRIEPI, tried to strengthen this region based on the knowledge of engineering seismology in some extent. In two cases, we observed the rich component in longer period region 2 ^ 10 sec in Japan. One is the case of the huge earthquakes which were tf=8 or higher. And, if the local strata condition has eigen periods in this region of the period , then the strong displacement waves were observed in a particular period(s). Typical examples were reported in the Keihin area, in the south of Tokyo, and in the Niigata Tohkoh area, in the east of City of Niigata. Dynamic characteristics of the upper soil layers, which are several thousand meters in this case, mainly governs their eigen periods in the longer period range as we concern in these regions. Several practical techniques to estimate those characteristics have been established in the field of engineering seismology. The author considers that there is no new problem to define the response spectrum in this region anymore. However it should be mentioned that there were some misunderstanding on the evaluation of the effect in this range because of the dynamic characteristics of accerographs which had been widely used in 1970's, and some engineers have the tendency that they evaluate too low, maybe only one tenth of the actual value. The author establish the following relation for the MITI Notice #515 on the "Seismic Design Guideline of the High Pressur Facilities" is 1981:

D = 60 cm T > 7.5 sec ) 9 (1) V = 50 kine T < 7.5 sec )

where D is the design basis ground displacement, and V is the deseing basis ground velocity.

§5 Design Principle of the Whole System

Hereafter, the author discusses mainly on a main reactor building supported by multi-rubber pad system. Therefore, a main building, a base isolation system and equipment and piping systems are the three main items concerned. The details of each will be discussed in later. Most significant discussion is how to make the balance of the strength of the main building and the base isolation system. According our design practice, most of design engineers try to design that the building is stronger than the base isolation system. For the S2 earthquake, the probability of failure of the base isolation system is lower than that of of the main building in 10~5 i> 10~° estimated, by the group of CRIEPI. The fragility curve of the base isolation curve is not steeper compare to that of the main building. This means that against the far severe earthquake, the main building may be weaker than the base isolation system. This relation is ideal, because the capability of supporting

-203- the super-structure may be remained after the failure of the base isolation system, especially rubber pads, according to the authors opinion. However, there is another result that the fragility curve of the multi-pad system would be very steep because of the zipping failure effect, which observed in anchor bolt failure' in the fields of previous earthquakes. The model above mentioned is the simpler than multi-pad system, but it should be mentioned of that the degree of the steeper fragility curve depends on the result of the quality control of pads and other elements such as anchoring devices. The study on the fragility curve of these items should be continued to establish the design concept. However, the capability of supporting the loads from supper- structure of the failed pads should be carefully studied to solve this problem. Some results of the mid-scale testing in Japan beautifully showed the possibility to realize this concept. The design against the shock of breakage of pad and other elements, and also that of the over deflection would be not necessary, according to the author's opinion. There are two design principles. In the case of two elements series, one design concept or the design criteria of the stronger element should be governed by the strength of the weaker one. It is typically applied to a case of a design of a nozzle and vessel. As a concept of the design of the base isolation system, the base isolation system is stronger than the super-structure, because the function of supporting the main building is significant. This is one idea. However, the supporting function could be remaining after some failures of its elements, then the base isolation system may be weaker than the super-structure at the design level S2 or higher like 1.5 S2. It should be noticed that the relation "weaker" or "stronger" is not absolute in this discussion. It means that their fragility curves may cross around this design levels. Now it is clear that the concept, like a nozzle and vessel, is not employed in the case of the design of the base isolation system. One of the reasons is that their strengths are relative, and the other reason is the remaining functional capacity. And also, it comes from the design basis earthquake S2, that is, the upper bound earthquake. In Japan, the probability of exceeding S2 earthquake is zero in principle.

§6 Design of Base Isolation System

The design of rubber parts is the most significant part. It is based on the CRIEPI's research. Two levels should be considered, that is, for S, earthquake and S2 earthquake. Two design limits are considered relatively to two levels of the design basis earthquakes. The relation of shear strain to resistant force is linear and these stress hardening type one. The first point for S1 should be stayed in the linear range, and the second point for S2 should be stayed in the stress hardening range, but lower than its ultimate strength. The margin may be decided based on the variation coefficient of their ultimate strengths. However, it should be noticed such relation is also

-204- function of its vertical load and the pattern of the stress hardening curve, which may change in the relation to vertical loadings, and the vertical loading is also variable by the rocking type response of the super-structure. In general, the margin would be 1.5 ^ 2.0 against statical vertical load, however, more precise analysis may be necessary. The draft of CRIEPI's guideline is more concervative compare to the author's one. They take the design limit for S2 is at the first point for its elastic limit. And that for S2 is lower than this elastic limit with enough margin. The author feels the necessity of further discussion on this problem. For the detailed design of the whole system and the base isolation system, it is recommended to use the response simulation. For this simulation, the lumped mass model may be employed by expressing each story of the building as one mass- spring system as well as the base isolation system. However, for the design of the base isolation system, their elements, multi-pad model may be necessary to evaluate their vertical load change caused by the rocking response of the super-structure. It is also significant for the probabilistic failure analysis of pads, and the gradient of the fragility curve is affected by the way of modelling of pads. The nonlinear characteristics of pads are one of keys to predict the behavior of the whole system. Two points should "be considered, one is how to express the relation of the reaction force to the horizontal displacement. There are several models are proposed by several research groups. The results may be not significant compare to the uncertainty of ground motions for future earthquakes. However, to the difinit input motions, there are some differences including the effect of induced higher harmonic motions. The second one is the effect of vertical dynamic load caused by the rocking of the super-structure. This is significant for the failure analysis. The most adequate type of models is different from the type of base isolation pad and the combination of damping device. Base-plate and anchor bolts are also very important for the function of the base isolation system. The author believes that their design criteria near to those of the class I vessel or piping should be applied. At least their allowable stress limit should be lower than those of ordinary structural elements. The criteria of steel plates in a pad have not been discussed in detail. For S,, it should be remained in elastic limit. But for S2, it is not clear whether or not it still should be remained in elastic limit. If it deforms plastically, the pad can't work normally anymore, and their replacement is required. It, maybe, depends on an individual situation of the plant. A damping device is divided to three types mainly, that is, elasto-plastic deformation of metalic material, hydraulic damper and fliction damper. The design criteria of latter two types are clear, that is, they should be normal even for S2. The criteria of metalic type should be considered in two points. On its fatigue failure, it should be discussed whether or not several after shocks of S-, level after one S2 shock should be taken into account. And also the same type of the evaluation for S

-205- earthquakes. The level of SQ earthquake is around 90 ^ 120 gal defined at the free field of supporting soil or rock. This concept of fatigue analysis may be applied to other elements. The allowable stress of plastic part of the metal has not been established as well as lead for damping device. Their fatigue analysis is important as mentioned, but it is difficult to be defined by the concept of plant conditions like "Plant Condition D". At least the margin to its ultimate strength and also its fatigue process should be considered. Also the characteristic of hardening and the condition of its brittle failure should be considered. If we use ordinary material in an industrial standard, usually there is no detailed definition for its behavior in higher level strain as using for damping device. It is necessary to design the energy absorbing portion and other portions separately and the latter should be designed as a supporting structure. The allowable stress of their portions should be the same to that of ordinary structural element. This is the same to the concept of other elasto-plastic damping devices for nuclear pipings. Also, the inspection method of the energy absorbing portion should be established. The author feels the necessity of the inspection on its grain size level including impurity of matrix.

§7 Design of Building Structure

The basic method of the structural design of the superstructure is not different from that of un-isolated building. As the author discussed in Section 5, there are several choices in the relation of the strength of the super-structure to that of the base isolation system. The author recommends the strength of the super-structure is stronger than that of the base isolation system on the basis of detailed design. It should be mentioned that the horizontal force induced by the response has longer period components of motions compare to that of ordinary ground motions. It depends on the designed eigen-period of the whole system, however, the force will work on super-structure and so on as a static force in general. This effect is more significant to equipment and piping systems, especially a main reactor vessel. Also, the effects of higher modes of the whole system should be evaluated, especially to a light damped, flexible structure. The relation of a pad arrangement to the wall arrangement and the load distribution should be evaluated by an adequate method. The detailed design practice of the basement slab of the super-structure, and also that of the lower basement mat have not been established. In a case of softer foundation soil, 7s=l,000 m/sec or lower, it should be considered to the design of the lower basement mat like Winkley model, according to the author's opinion. Because the equivalent shear velocity of the ordinary reactor building may be 1,000 ^ 1,500 m/sec. Design criteria and the way of analysis of a supported super-structure are not different to an ordinary unisolated nuclear building. Based on the level of input acceleration or velocity at the upper level of the base isolation system induced

-206- by the design basis earthquakes S2 or S^ or SQ, buildings and others should be designed. Their allowable limits may be the same to those of unisolated nuclear structure, the author recommends. However, there are some discussions, and the most elegant evaluation may be a comparison on their fragility curves in various earthquake levels.

§8 Design of Equipment and Distribution System

As their structural design, it is not much difference to those in unisolated buildings, except force works longer duration like a static force. In the design of a shell or a rod, the buckling should be examined carefully. If the after-buckling behavior is considered for the design, this fact is more significant. Connecting distribution systems should be designed as enduring to the differential movement of the both systems. For this, again, there is the same type of discussion to the previous one, that is, whether or not the design displacement is the maximum limit of the displacement of the system, or the response to S^ earthquake with an adequate margin. In principle, the author believes that it should be designed based on the response induced by the design basis earthquake. On the other hand, the categorization of induced stress of the system, especially of that for a piping system, should be the primary stress. The reason is clear, that is, the strain (stress) caused by their differential motion is almost static one, and for the tensile strain, it is easy to exceed the elongation limit of the material.

§9 Quality Control and Inspection

The quality control of rubber pads has not established well. At this moment, we try to test at the time of receiving them. However, on tire of automobiles and aircrafts, the techniques of the quality control is most advanced. Therefore, manufacturers of rubber pads may overcome these difficulties easily in the near future. The evaluation of degrading of rubber material is also established and the acceleration test is commonly done. Even though the standard surveillance test should be established for the system, it may be no farther difficult problem. They will meet to the safety requirement of nuclear power plants soon.

§10 Concluding Remarks

There is no much difference between the design guidelines of LWR and FBR. The main difficulty of the design of FBR is the sloshing phenomenon. The base isolation system can not work for it, and in some cases, the system may amplify it. We need further research how to overcome the sloshing phenomenon. Within, one or two years, we are going to complete the final draft, it may be applicable to both LWR and FBR. And it is not

-207- so much difference of criteria for base isolation floor. The actual systems will be used more often and earlier than base isolated reactor buildings. Already, an example is built in the field of nuclear facilities. The concepts, which the author described above, have been obtained through the discussions in the meetings with CRIEPI and Japan Atomic Power Company. And also the informal research group sponsored by Mitsubishi Heavy Industry Co.. The author expresses his gratitude to the related members of these committees, especially Mr. M. Motegi, CRIEPI, who has been discussing this subject with the author. The author heartily appreciate the support of Professor Emelitus N. An, University of Tokyo for this research.

"§11 Reference" and Tables and Figures have not been prepared.

-208- XA0055387 IAEA SPECIALISTS' MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, California, USA, March 18-20, 1992

DEVELOPMENT OF GUIDELINES FOR SEISMIC ISOLATION IN ITALY M. Olivieri (ANSALDO-RICERCHE, Genova, Italy) A. Martelli (ENEA-RIN, Bologna, Italy) F. Bettinali (ENEL-CRIS, Milano, Italy) G. Bonacina (ISMES S.p.A., Bergamo, Italy)

ABSTRACT The first activities on seismic isolation that were performed in Italy concerned the preparation of a proposal for design guidelines for nuclear power plants using the high damping steel-laminated elastomer bearings (HDLRBs). They were jointly initiated by ENEA-RIN and GE Nuclear Energy in 1988, with the co-operation of ISMES and the support of experts of ENEA-DISP and Bechtel National Inc.. The features of the guidelines proposal were outlined at the First Post-SMiRT Conference Seminar on Seismic Base Isolation of Nu- clear Power Facilities (San Francisco, 1989). The full text of the document was published in the Journal "Energia Nucleare" in 1990, in a tentative form, to allow for a broad review. A summary of the main items - together with some first results of R&D studies performed in support to guidelines development - was also reported in a paper which was recently published by the Journal "Nuclear Technology" (February 1992) . A first revision of the document is being prepared and will be soon published: it accounts for both comments received - for instance, by the American Society of Civil Engineers (ASCE), ENEA-DISP and the Malaysian Rubber Producers' Association (MRPRA) - and the first results of R&D studies in progress in Italy and the USA. These activities have recently been extended - as part of a cooperation with the Italian Standard Authority (UNI) - to other antiseismic devices, for application to civil buildings and non-nuclear plants. A co-operation of ENEA, ENEL and ISMES has also been started with the National Seismic Service to help it in the assessment of national regulations. Furthermore, extension of the aforesaid guidelines document to nuclear reactors using bearings different from the HDLRB has been planned, under the sponsorship of the Commission of the European Communities: this work will be performed by ENEA, with the cooperation of ALGA, ISMES, ANSALDO and the Nuclear Engineering Laboratory (LIN) of the Bologna University, and the support of experts of the French CEA. This paper reports the main features of the aforesaid activities, pointing out the need for R&D to support guidelines development.

-209- 1. INTRODUCTION The Italian Status Report, presented to this Meeting by Martelli & Bettinali [1], has explained the reasons for the considerable efforts that are being devoted by the Italian Agency for the New Technologies, Energy and Ambient (ENEA), the National Utility (ENEL), ISMES, ANSALDO-Ricerche and other members of the National Working Group on Seismic Isolation (GLIS) to the development of guidelines for seismically isolated constructions. We remember that one of the purposes of GLIS is to provide support to the state Institu- tions that are charged with the approvals of structure designs, for the design verification (Martelli & Bettinali [1]). The Italian Status Report [1] has also stressed the intercorre- lations existing between guidelines development and R&D work (such as the experimental activities reported by Bonacina et al. [2] and the numerical studies presented by Bettinali et al. [3]), and the need for performing both activities in parallel. This paper presents the main features of guidelines development, pointing out the need for R&D to support it, and mentioning the national and international collaborations that are in progress or planned on this subject. It forms an updated version of the paper published by Martelli et al. [4] in the Journal "Nuclear Technolo- gy". 2. FIRST PROPOSAL FOR DESIGN GUIDELINES FOR NUCLEAR REACTORS ISOLATED BY HDLRBs The first activities on seismic isolation development were initiated in Italy by the ENEA Department of Innovative Reactors (RIN), in 1938. They concern the preparation of a proposal for design guidelines for nuclear plants using High Damping Steel-Laminated Rubber bearings (HDLRB), and are being performed with the cooperation of ISMES and GE Nuclear Energy and the support of experts of the ENEA Directorate for Nuclear Safety and Health Protection (DISP) and Bechtel National Inc.. The reasons for the choice of these isolators have been outlined by Martelli & Bettinali [1]. The aforesaid guidelines document was prepared taking into account the most recent information on seismic analysis of nuclear reactors in general and the state-of-the-art of engineering design of isolated structures. It mainly deals with items different from non-isolated systems. Focus is on requirements, analysis methods, qualification procedures and monitoring of isolation devices and systems. Proposals for design requirements and analysis methods for isolated structures, systems and components are also included. Although particular attention is paid to the case of LMRs (due to their sensitivity to earthquakes), guidelines also aim at fully covering other types of nuclear reactors that are isolated by means Of HDLRBS. The features of the guidelines proposal were outlined by Mar- telli et al. [5] at the First Post-SMiRT Conference Seminar on Seis- mic Base Isolation of Nuclear Power Facilities (San Francisco, 1989). The full text of the document was published by Martelli et al. [6] in the Journal "Energia Nucleare" in 1990, in a tentative form, to allow for a broad review. A summary of the main items together with some first results of R&D studies performed in support to guidelines development - was also reported by Martelli et al. [4]

-210- in a paper which was recently published by the Journal "Nuclear Technology" (February 1992). It will be shown below that some safety factors to be used in the design, some test parameters and some details of qualification procedures are not defined, yet (these are indicated as TBD, i.e. "To Be Determined"). Indeed, the exact definition of these items requires specific R&D work, such as that described by Bonacina et al- [2] and Bettinali et al. [3]. Furthermore, some other items - such as, for instance, low frequency effects - need to be better precised. Some remarks on the different sections of the proposed guidelines are reported below. The most important items are better specified by Figs. 1 to 4. There, reference is made to the published document of Martelli et al. [6] for details, and to well-known U.S. reports, such as those of the Nuclear Regulatory Commission (NUREG Reports). 2.1 Definition of Ground Motions Reference is made to existing codes for Safe Shutdown Earthquake (SSE) and Operational Basis Earthquake (OBE). It is specified that special attention must be paid to the low-frequency range (0.1 - 1 Hz), because of the effects on isolated structures. This implies the use of site specific ground motions and the corresponding response spectra (Fig. 1).

2.2 Design Requirements and Analysis Methods for Isolated Buildings and Isolation Support Structures (Fig. 1) It is stressed that design of isolated structures must rely on displacement. The methods to be used for determining the reference displacement are prescribed. Requirements are provided for the structural elements located above and below the isolation interface; it is specified that they shall be rigid in the horizontal plane. Safety factors are also provided against overturning of the supported structure, together with requirements to avoid isolator uplift. The features and safety functions of ultimate horizontal and vertical restraint systems (fail-safe systems) are pointed out. It is clarified when the horizontal system is required. The methods to be adopted for defining the gaps that shall be present between isolated and non isolated structures or independently isolated structures are also provided. These gaps shall be equal to relative displacements times a safety factor. The exact value of this factor is still TBD: it has to be determined by experimental tests as a function of the strength reserve of isolators and SSE return period. Finally, rules are given for the inspectability and replacement capability of isolators, soil-structure interaction analysis, and design analysis methods. As to the last item, prescriptions concern the use of time-histories, applicability of simplified methods, and need and features of parametric calculations to be performed by varying soil, isolation and structure stiffnesses. We note that the use of simplified methods is only permitted in particular cases and requires precise justification.

-211- Fig. 1. Design guidelines for nuclear reactors isolated by HDLRBs: definition of ground motion; design and analysis methods for isolated building and isolation support structure.

SSE and OBE defined in free- Attention to the reliabil- Design response field condition at the surface ity of records in the fre- spectra a those of the ground by site-specific quency range 0.1 to 1 Hi specified by Regula- design response spectra and in the filtering process tory Guide 1.60 for and time histories the frequency <1 Hz Definition of ground motions if Determination of response spectra and artificially generated time histories according to the proposed Rev. 2 to Sees. 2.5.2 and 3.7.1 of NUREG-0800 DR = reference displacement = spectral displacement of the building calculated as a single degree of freedom linear system plus contribution of eccentricity Design verification based on the Design basis design displacement D = Safety f factor • max(DR,Dc) DC = value of displacement calculated by dynamic analyses

three-dimensional effect included in the analyses Linear model of structure and soil should be used

Nonlinear model of isolators may be used, if necessary A global model of the soil, the isolation system, and the For linear and nonlinear model, structure shall be used use of three contemporary components of input motions

For linear model it is also possible to use three components of the motions applied one at a time and to combine the effects by the SRSS method

At least four different Time history total dura- Strong time histories for both tion &10 times the first motion Time-history analyses horizontal and vertical natural period of the part directions isolated structure fc6 s

Use of response spectrum techniques is not recommended

Parametric evaluation shall be performed by varying soil stiffness, isolator stiffness, and structure stiffness

Rigid in the horizontal plane Sufficiently rigid out of to reduce differential Two horizontal dia- their plane to reduce the phragms above and displacements in the below the isolators superstructure and in differences of isolators' the foundation vertical displacements

Vertical backup system Horizontal restraint system shall be provided to guar- Structural elements not necessary if isolators Ultimate restraint systems antee the transfer of the vertical loads to the foun- can support the structure dation in any condition up to TBD times the SSE

Gaps between adjacent structures > Safety factor • Relative design displacement

'212- Fig. 2. Design guidelines for nuclear reactors isolated by HDLRBs: design and performance requirements for overall seismic isolation systems; design requirements and analysis methods for isolated structures, systems and components; design requirements and analysis methods for interface components.

Horizontal isolation frequency in the range 0.5 to 1 Hz

Minimum damping of 10% at the design frequency

A restoring capability so that the maximum offset after SSE = 10% maximum displacement

Maximum variation of stiffness and Design requirements for damping due to environmental isolation systems condition and aging = TBD% of the mean value

Vertical stiffness high enough to reduce any amplification of vertical components

Eccentricity limited to minimize torsion

Minimize global structure rocking and local uplift

Designed to withstand one SSE and at least two OBEs

Secondary structures and components analyzed using the motions (time history or floor response spectra) calculated at their supports by means of the "global model" Analysis methods for isolated of the isolated building structures, systems. and components

Static analysis performed if component horizontal natural frequencies lie in the floor response spectrum zone where little amplification is present

Attention to possible sloshing effects

D, = gaps required in the same location at the foundation level

Safety-related components shall be D2 = displacement of the point Design requirements for capable of absorbing a relative of the structure to which the interface components displacement D, = D, + D + D component is attached, 2 3 relative to the base

D3 = deformation of supporting component if present

-213- 2.3 Design and Performance Requirements for Overall Seismic Isola- tion Systems (Fig. 2) The horizontal displacement to be used for the design of the isolation system is defined. The safety factor to be adopted for assessing its value shall be equal to that (TBD) related to gaps. The design case for vertical loads is also defined, and requirements are provided with regard to horizontal and vertical stiffnesses, damping, self-centering, wind and small earthquake resistance. 2.4 Design Requirements and Analysis Methods for Isolated Structu- res, Systems and components Rules are provided as to how to determine Floor Response Spectra (FRS) and to analyse systems and components located inside isolated structures (Fig. 2). For the FRS determination, parametric time- history analyses are required, by varying the stiffness parameters of both the structure and isolators. For the design of components, simplified static analysis is usually permitted. Analysis of sloshing and other low frequency effects requires specific R&D to be better precised. 2.5 Design Requirements and Analysis Methods for Interface Compo- nents Displacements to be accomodated by the components and systems that cross the isolation interface are defined as functions of gap values and structure flexibility (Fig. 2). Effects to be accounted for in the analysis are stressed, and the need for adequate qualification, especially for piping expansion joints, is pointed out. This may require tests, to be performed in the correct pressure and temperature conditions. 2.6 Design Requirements for Individual Isolation Devices Requirements are provided with regard to the vertical load capacity and design load, the maximum horizontal displacement capacity and design displacement and stability, the determination of vertical and horizontal stiffnesses and buckling load, the assessment of stiffness - strain relationship and damping (Fig. 3). In particular, it is specified that the isolator characteristics (stiffness, damping and buckling load) to be used in the design shall be based on specific test data or validated finite-element methods. Simplified formulas - such as those provided in appendix to the document - may be used if demonstrated adequate. As to R&D needs in support to these prescriptions, we note that experimental tests were judged necessary to determine the (TED) safety factors related to the vertical load, buckling, and combined effects of vertical load and horizontal deformation. To allow for the assessment of the total rubber thickness of bearings, experimental work was also judged necessary for the exact definition of the (TBD) value of shear strain (horizontal displacement divided by the total rubber thickness) that shall correspond to the SSE. Further requirements concern the effects of cycling and related degradation, environmental effects, creep effects, aging effects, self-centering capability, uplift and rocking, and design tolerances

-214- (Fig. 3). For the exact definition of some of these requirements, R& D work was again judged necessary: tests were needed to fix the TBD values of the parameters related to environmental effects (temperature, radiation, ozone attack, fire), to determine the isolator life -time, and to define the maximum offset permitted for each isolator after an earthquake. With regard to the latter item, appropriate test features had also to be identified. 2.7 Qualification of Seismic Isolator Bearings Static and dynamic tests identified as necessary for the qualification of single bearings and isolation system are reported. These are consistent with the ongoing experimental campaign of ENEA-RIN reported by Bonacina et al. [2], which also aims at confirming the adequacy of the test series and defining the exact values of some test parameters (see also Fig. 4). 2.8 Acceptance Testing of isolation Devices Tests required to provide the quality control of bearings are outlined. These also include checks on the external geometry, some distructive controls, and tests to confirm the isolator characteristics, performance and integrity. The criteria that determine the number of isolators to be tested are given, together with those to be adopted for rejecting bearings - in the case that controls are out of tolerances - and those to be used for bearing identification. 2.9 Seismic Isolation Reliability Requirements concern the Quality Assurance (QA) program, bearing life-time and in-service inspection. Use of three-dimensional (3D) finite-element (f.e.) models is specified to identify, assess and if necessary, correct isolator weaknesses (Fig. 4). 2.10 Seismic Safety Margin Assessment Methods to be used to ensure an adequate level of seismic safety and to identify if necessary, feed-backs on the design are provided. The process consists of the following steps: (a) selection of the earthquake assessment level; (b) technical QA (design organization, design methods, codes and standards, etc.); and (c) safety margin assessment. The safety margin assessment derives from that proposed for the existing commercial Light Water Reactors as a simplification of Pro- babilistic Risk Assessment (Fig. 4). For those plants, however, construction details are well known, while for future isolated plants there is a lack of construction and operational experience. This makes it necessary to carry out the assessment for isolated plants at both design and construction stages: the first step allows for a check of plant design and if necessary, feed-backs to improve it, while actual margins may only be provided by the second step. The process makes use of the best available plant information at the time of the analysis. It permits detailed analysis to be avoided for the safety-related components and systems for which previous experience or collective opinion of appropriate experts

-215- Fig. 3. Design guidelines for nuclear reactors isolated by HDLRBs: design requirements for individual isolation devices.

The stiffness and damping values used in the design and analysis shall be based on specific test data of prototypes

Isolator shall have a minimum safety factor of 3 for vertical loads in its laterally undeformed state

The isolator shall provide a safety factor (TBD) for the maximum horizontal displacement

The isolator shall be stable under TBD times the vertical load at a horizontal displacement that is TBD times the design displacement

Horizontal stiffness and damping shall not change more than 10% from the mean value of all isolators Design requirements for individual isolation devices The isolators must have adequate fatigue life (TBD cycles at the TBD strain level)

The isolators shall be designed to withstand environmental effects (TBD) such as temperature, radiation, ozone attack, and fire

The creep deformation shall be limited to not more than 20% of the dead load deflection

The design life of the isolators shall be equal to that of the plant

For the design conditions, resultant tensile loads are not permitted

-216- Fig. 4. Design guidelines for nuclear reactors isolated by HDLRBs: qualification of seismic isolation bearings; acceptance testing of isolation devices; seismic isolation reliability; seismic safety margin assessment; seismic monitoring.

At least three extra prototypic full-scale bearings for each different type present in the system must be constructed for qualification tests; these will not be mounted in the reactor

Static and dynamic tests shall be performed

Bearings If necessary, tests shall be repeated before and after artificial aging

Horizontal stiffness and damping shall be assessed with dynamic cycling at 0.5, 1, and 1.5 times the first natural frequency of isolated structure, at different levels and Qualification combinations of horizontal displacement and vertical load

Unless previous detailed experience is available on the selected isolation system, shaking table or other acceptable dynamic tests are required on scaled isolated structure mockups

Isolation Multifrequencial simultaneous three-dimensional tests system should be performed by means of shaking table

Snap-back tests on the real structure are recommended

Acceptance testing i The isolators shall be subjected to the tests and isolation devices controls reported in Ref. 6

Design, material procurement, manufacturing, qualification, and acceptance tests of the bearings shall be carried out according to a quality assurance program, as specified in 10CFR50 Appendix B

Finite element analysis for the seismic isolators to identify, assess, and correct, if necessary, the weaknesses and vulnerabilities to external challenges, including failure of interfacing components, functional testing, storage handling, packaging, transportation, and maintenance Reliability Seismic isolation analysis reliability Sensitivity of reliability performance to testing, in-service J inspection, and maintenance strategies

In order to assess whether or not significant changes in In-service isolators' properties have occurred inspection program The isolators shall be subjected to the tests and controls reported in Ref. 6

Reference to the proposed methodology of NUREG/CR-4482 is made; only Seismic safety margin differences related to the utilization of isolation system are considered

The reactor shall be shut down in the case of Seismic safety system earthquakes exceeding the OBE; to this aim, the plant shall be equipped with an adequate safety system Seismic safety and monitoring systems This system shall provide sufficiently reliable information to the operator in the case of a Monitoring system large earthquake, with regard to both its level and the structure response

-217- shows that there is a high confidence of low probability of failure at an earthquake level larger than that selected. 2.11 Seismic Monitoring The need and required features of a detailed seismic monitoring system, capable of recording earthquake motions during time in the free-field and on the structure, partly in short time, are stressed (Fig. 4). It is pointed out that displacements between the structure base and isolation system support base must be recorded, in addition to accelerations at various elevations. 3. REVISIONS OF THE GUIDELINES DOCUMENT OF REF. [6] The guidelines document of Martelli et al. [6] will be periodically updated by ENEA and GE, to include comments and to reflect the advances of seismic isolation technology development. A first revision of the document is being prepared and will be soon published: it accounts for both comments received - for instance, by the American Society of Civil Engineers (ASCE), ENEA-DISP and the Malaysian Rubber Producers' Association (MRPRA) - and the first results of R&D studies in progress in Italy and the USA. 4. EXTENSION OF THE GUIDELINES DOCUMENT OF REF. [6] TO OTHER BEARING TYPES The extension of the guidelines document of Martelli et al. [6] to nuclear reactors using bearings different from the HDLRB will be also soon initiated, within the studies sponsored by the Commission of the European Communities (CEC). This work will be performed by ENEA, with the cooperation of ALGA, ISMES, ANSALDO and the Nuclear Engineering Laboratory (LIN) of the Bologna University. It will consist in the following activities: (a) Revision of the document of Ref. [6], based on comments of European organizations and an updated analysis of the state-of-the- art on the design of isolated nuclear and non-nuclear structures in Europe. (b) Extension of the document to: (bl) other horizontal isolation systems of interest for the European Projects (neoprene bearings, sliding devices, etc.); (b2) the other types of horizontal isolation systems of general interest (other elastomeric isolators, including lead plug and low damping rubber bearings, etc.); and (b3) three-directional isolation. (c) Identification of items to be precised through further R & D and specification of the related necessary work. Activities will take into account the other available proposals and recommendations for design guidelines for isolated structures, and will take advantage of the co-operations existing between ENEA and other national and foreign organizations. 5. GUIDELINES DEVELOPMENT FOR NON-NUCLEAR ISOLATED STRUCTURES Development of guidelines for isolated structures is also in progress in Italy for non-nuclear structures, according to the increasing interest in these applications and the increasing number of civil buildings being isolated in Italy (Martelli & Bettinali

-218- [1]). More precisely: (a) The preparation of guidelines concerning the isolation bearings and energy dissipation systems that have been judged of interest for buildings and bridges is in progress in the framework of the activities of specific groups established by the Italian Standard Authority (UNI). ENEA, ISMES, ANSALDO-Ricerche and other members of GLIS participate in these activities. (b) A co-operation of ENEA, ENEL and ISMES was started with the National Seismic Service to help it in the assessment of national regulations (the design of constructions is regulated by law in Italy). The preparation of two guidelines documents (the first for bearings, the second for isolated buildings) is already in progress: these should be used by the designers of isolated buildings in the transition period preceding the promulgation of the necessary law, to get the approval of the High Council of Public Works which shall examine each plan for an isolated building in that period. (c) A specific subgroup of GLIS was very recently formed, with the task of collecting, analysing, commenting and if necessary, integrating the available guidelines documents and regulations (Mar- telli & Bettinali [1]). The aim of this subgroup is not to further duplicate works in progress in different frameworks, but to compare and homogenize these works. 6. COLLABORATIONS The importance of national collaborations for the development of guidelines for isolated structures are made evident by the previous sections. As to the existing international collaborations, it is worthwhile stressing that with the USA (especially with GE - see Sects. 2 and 3) and that with Japan. In the first meeting concerning the Project on Seismic Isolation in the framework of the Agreement of Scientific and Techological Co-operation between Japan and Italy, held in August 1991 at Tadotsu, it was agreed that the Japanese experts will provide comments to the proposal of Martelli et al. [6], and documents of common interest will be exchanged. Furthermore, the work to be performed with the CEC sponsorship will strengthen cooperations between Italy and other European countries: in particular, a collaboration between ENEA-RIN and the Department of Mechanical and Thermal Studies (DEMT) of the French Commissariat a l'Energie Atomique (CEA) is foreseen. 7. CONCLUSIONS Work in progress in Italy for the development of design guidelines for seismically isolated structures has been outlined. In particular, the main features of a proposal concerning nuclear reactors, isolated by high damping steel-laminated elastomer bearings, have been recalled. Items that needed R&D work to be precised have been stressed. It has been mentioned that a first revision of the aforesaid document is in progress, to include comments received and to account for the results of the ongoing R&D work, and that this work is being performed in cooperation with GE Nuclear Energy. Extensions of the guidelines document to other bearing types, in the framework of a project sponsored by the CEC, have been cited, together with a collaboration with Japan and the work in progress in Italy for non-nuclear applications, as part of the activities of the

-219- Working Group on Seismic Isolation (GLIS) and to support the Italian Standard Authority and the National Seismic Service for the assessment of national rules. REFERENCES [1] A. Martelli and F. Bettinali, Status Report on activities on seismic isolation in Italy, Paper presented to this meeting. [2] G. Bonacina, F. Bettinali, A. Martelli and M. Olivieri, Experiments on seismic isolation in Italy, Paper presented to this meeting. [3] F. Bettinali, A. Martelli, G. Bonacina and M. Olivieri, Numerical activities on seismic isolation in Italy, Paper presented to this meeting. [4] A. Martelli, M. Forni, M. Indirli, P. Masoni, B. Spadoni, G. Bonacina, G. Di Pasquale, T. Sand and E.L. Gluekler, Development of design guidelines for seismically isolated nuclear reactors and R&D work performed by ENEA, Nuclear Technology, 97 (1992) 153-169. [5] A. Martelli, P. Masoni, M. Forni, M. Indirli, B. Spadoni, G. Di Pasquale, V. Lucarelli, T. Sand, G. Bonacina and A. Castoldi, ENEA activities on seismic isolation of nuclear and non-nuclear structures, in: Proc. First int. Post-SMiRT Conf. Seminar on Seismic Base Isolation of Nuclear Power Facilities, San Francisco, CA, USA, ANL Report, Argonne National Laboratory, CONF-8908221 (1989) pp. 57- 73; Nucl. Engrg. Des., 127 (3) (1991) 265-272. [6] A. Martelli, P. Masoni, G. Di Pasquale, V. Lucarelli, T. Sano, G. Bonacina, E.L. Gluekler and F.F. Tajirian, Proposal for guidelines for seismically isolated nuclear power plants - Hori- zontal isolation systems using high damping steel-laminated elastomer bearings, Energia Nucleare, 1 (1990) 67- 95.

-220- XA0055388

Recent Results of Seismic Isolation Study in CRIEPI -Design Metnod-

Katsuhiko ISHIDA, Dr. Eng., Hiroo SHIOJIRI, Dr. Eng, Masafumi MOTEKI, Abiko Research Laboratory, Central Research Institute of Electric Power Industry.

Heki SHIBATA, Dr.Eng. Professor., Takafumi Fujita Dr. Eng.Professor Institute of Industrial Science, University of Tokyo.

L Introduction Japan's seismic isolation technology has mainly developed on the basis of base isolation of non nuclear building structures. In recent years, there are some studies on applying this seismic isolation technology to nuclear power plants, especially to Fast Breeder Reactors(FBR).To realize this, further verification will be required with respect to the reliability of seismic isolation technology. The research project called " Verification Test of Seismic Isolation for FBR (consigned to the Central Research Institute of Electric Power Industry), commenced in 1987 sponcered by the Ministry of International Trade and Industry to obtaitn the technical data. Aiming at completion in March 1994, efforts are being made to draft " FBR Seismic Isolation System Design Methods " which will summarize the results of research in this field (Table 1). Research items are as follows. 1) Assessment of the characteristics of large seismic isolation elements 2) Assessment of the dynamic vibration characteristic of seismic isolation systems 3) Assessment and study on appropriate seismic isolation structures 4) Setting of design basis earthquake ground motions for seismic isolation design 5) Assessment of the reliability of seismic isolation systems 6) Development of seismic isolation design procedures

2. Seismic Isolation System Design Considerations As a general rule, buildings and structures incorporating seismic isolation must be supported on bedrock. Also, as a general rule, the structures supported by seismic isolation elements shall be designed sufficiently stiffer than the isolation devices. Meanwhile, in order that the sedsnricaUy isolated nuclear power facilities may have the same degree of safety as those of conventional nuclear power facilities, the buildings, structures, equipment and piping systems which have identical functions shall follow basically conventional seismic design considerations, and the seismic isolation elements and their accompanying new structures shall be designed in accordance with technical standards which

-221- assure the same degree of reliability as that of conventional nuclear power facilities,

3. Considerations on Design Basis Earthquake Qround Motion When determining the design basis earthquake ground motions, the short period components of earthquake ground motions shall assessed in conformity with design earthquake ground motion assessment methodology for conventional power reactors, while the slightly long-period components of a seismic ground motion shall be assessed by analysis considering the behavior of active faults and ground structures. When assessing the slightly long-period components of design basis earthquake ground motion, two different types of ground motion Si and S2 shall be selected according to the degree of its intensity.

4. Seismic Isolation Element Design Considerations Seismic isolation elements must fulfil their functions with ample safety under all possible conditions of loads during their service life.

A. Allowable limits of seismic isolation elements From the standpoint of ensuring the integrity of the functions of the seismic isolation, the allowable limits of seismic isolation elements shall be set in consideration of the uncertainty of ultimate behavior, aging of seismic isolation elements and other effects. B. Seismic isolation element durability assessment Seismic isolation elements must sufficiently resist deterioration due to various environmental effects.

5. Building and Structure Design Considerations Buildings and structures must have ample safety under all loads during their service life. A. Integrity assessment for buildings and structures The integrity of buildings and structures shall be assessed appropriately in consideration of the behavior of seismic isolation systems during earthquake, and other loads under all possible conditions.

B. Structural planning for buildings and structures To realize seismic isolation function in seismic isolation systems, a relevant structural planning shall be formulated with respect to the layout of buildings and seismic isolation elements and each part around the seismic isolation layers.

-222- 6. Equipment and Piping System Design Considerations Equipment and piping systems shall be constructed such that no major accident is induced under all possible conditions of earthquakes.

A. Seismic classification of equipment and piping systems The seismic classification of equipment and piping systems shall follow the definition indicated in the Technical Guidelines for Seismic Design of Nuclear Power Plants, (hereinafter called the Seismic Design Guidelines).

B. Seismic design assessment method for equipment and piping systems 1. In seismic design of equipment and piping systems, the structural integrity shall be confirmed based on the principle of seismic design assessment methods shown in the Seismic Design Guidelines. 2. Design seismic force The considerations for design seismic forces for equipment and piping systems shall follow the conventional standards. However, the horizontal static seismic force applied to equipment and piping in the reactor buildings incorporating seismic isolation shall be determined based on the characteristics of seismic isolation systems.

C. Load combinations and allowable limits of equipment and piping system. Seismic Design Guidelines shall be followed However, in the case of connection piping for which assessment of stress due to relative displacement during earthquake has not been required,structural integrity shall be assessed by applying criteria for higher seismic class requiring assessment of stress due to relative displacement.

7. Conclusion and Remarks Based on these basic seismic isolation design considerations, CR1EPI drafted FBR Seismic Isolation System Design Methods in 1990 in accordance with the items shown in Table 1, fully considering consistency with the Seismic Design Guidelines and other standards. CRIEPI is presently revising the contents of these methods to make them more widely acceptable. This research was supported by Ministry of International Trade and Industry, Japan. a REFERENCES 1. K.ISHIDA, et-al: " SEISMIC ISOLATION TEST PROGRAM FOR FBR- INTRODUCTION ON DESIGN AND TECHNICAL GUIDELINES(DRAFT)- " ,FR'91 KYOTO(1991)

-223- r»bi»i Main Contents of FBR Seismic Isolation System Design Methods

Chapter Section Subsection Ha In Contents

1. 8»9ic design Principle I.I Seismic isolation system The sane degree or reliability •• that at bedrock-iupportetf.earthauake-rasiatant design considerations nuclear power racititles shall be ensured is • general rule.

1.2 Seismic isolation eleaent 1.2.1 Basic principle Functions during service lite shall be naintained. demign considerations

1.2.2 Functions of seismic Support and reduction or seisaic forces or structures, tysteas and equipment isolation elements which are important to maintain sarety.

1.2.3 Allowable Iimits The uncertainty or critical characteri»tics. aging and other afreets shall be taken into consideration.

1.2.4 Durability as9e99Ment Resistivity against deterioration shall be ensured.

1.3 Building and structure 1.3.1 Basic principle ftaple sarety shall be ensured.

1.3.2 Integrity assessment Appropriate assessment of integrity under all possible conditions or load.

1.3.3 Structural planning

1.4 Equipment and piping 1.4.1 Basic prtnolple system design load. consideratIons 1.4.2 S«i3»ie cussif icition Aseisaic Design Guidelines of Nuclear Power Plant shall be roitoued. 1.4.3 Aseisaic design Horizontal static seisaic forces to be applied to etuipaent and piping shall assessaent nethod9 be deterained based on the characteristics of aeisale isolation aysteas. Assessaent ot stresses due to relative displacement or connection pipe during permissible littits earthquake.

2. Design basis earthquakes Z.I Basic principle Short-period components shall conform to those for light water reactors. Regarding long-period components. e99essaent shall be conducted by aean9 of analyses considering the behavior of hypoeentral faults, etc.

ro t r»»ie i Main Contents of FBR Seismic Isolation System Design Methods

Chapter Section Subseotion Main Contents

2. Design basis earthquakes 2.2 Evaluation or design base When assessing the slightly long-period component or design basis earthquake earthquakes ground notion, tuo different type* of earthquake notions S, and S, 9hall be selected according to the degree or Its intensity

3. Structural punning or Structural planning in which the safety or all seisaie isolation systems such as seismic isolation systeas equipment, buildigs. seismic Isolation eleaient*. etc. is determined synthetically.

4. Structural planning or 4.1 Structural planning or 4.1.1 upper buildings Highly rigid construction as a general rut*. Clearcut fr»ae. seismic isolation elements buildings and structures buildings and structures 4.1.2 Layout or elements Uniform) load distribution. Layout on rigid foundation. Suppression of eccentricity between the center of rigidity of seismic isolation elements and the center or gravity or upper structures. Maintenance and inspection. Replaceable when required.

Prevention of herafut subsidence, inclination, etc. Securing or appropriate slabs and seispic isolation rigidity. Checking or ground pressure. Prevention or Hotting. element pedestals

4.1.4 Other members during earthquake. 4.2 Design or selsiito 4.2.1 Scope or application Laminated rubber bearing with natural rubber, laminated rubber bearing with lead Isolation elements plug. h>gh-d»»plng laminated rubber bearing, and steel dapper.

4.2.2 laninstod rubber bearings Applicable asterials. Constants or rubber materials and allowable stresses.

4.2.3 $teel dappers Applicable aaterials. Design conditions. Design aethods. S.I Horizontal seiMie S.I.I assessment or input Conversion rro« reterence* earthquake notion into input earthquake potion based or seismic isolation response ant lysis seismic notions on appropriate pethods. Sufficient attention shall be paid to the effect of a elements, buildings and slightly long-period component. structures 5.1.2 lode)1 ing of buildings Characteristics peculiar to selsaic isolation structures (lengthened period, and structures concentration of stress in seisaie Isolation layers, non-linearity or se'spic isolation layers, etc.) shall be taken into consideration.

I ro rv> • i Main Contents of FBR Seismic Isolation System Design Methods <3'6>

Chapter Section SubseotIon Main Contents

5. Seisaic r«sponc« analyses 5.1 Horizontal seisaic S.I.3 Model ing of seisaie Modeling of hysteresis characteristics by acceptance tests. Consideration of Seismic isolation response analysis isolation layers or the errect of manufacturing errors, aging and environmental «1»Mnt9. buildings and conditions. Integration or seisaic isolation eleaents. structures 5.1.4 Analysis methods Appropriate seisaic response analysis.

5.2 Vertical seismic response 5.2.1 Assessment or input Vertical reference seisaic motion shall be converted appropriately. analysis

5.2.2 Modeling or buildings Consistency with horizontal d»naaie analysis model. and structures

5.2.3 Modeling or seisaic Modeling by single or multiple vertical linear spring constants and internal isolation layers viscous damping constans. Consideration or manufacturing errors, aging and evironmental conditions.

5.2.4 Analysis aethods Appropriate seisaic response analysis.

6.3.1 rhil t i-direot ionel input Shall bs considered appropriately. be oonsldorod

6.2 Static seisaic forces 6.2.1 Basic considerations for Suppression against dynamic seisaic forces. The response properties of seismic

roI ro en i table i Main Contents of FBR Seismic Isolation System Design Methods

Chapter Section Subsection Main Contents

6. loads on 3ftisate isolation 6.3 Design seismic forces Dynaaic seisaic force calculated in Chapter 5 and atatte aeisalc aotion elements, buildings and calculated in Chapter 6.2. whichever is greater. structures •.4 Load combination* Shall be considered appropriately in referent to the loads shown in Chapter 8.1.2. •'

7. Intersrity esessaent of Stresses or derogations which develop I* seisale Isolation layers, buildings, seismic isolation systems structures, equipment and piping shall not exceed the allowable Halts established in Chapters 8 and 9. Also.analysis shall be conducted to ascertain that an appropriate safety aargin exists.

S. Integrity asesament or 8.1 Prineiple of Integrity 8.1.1 lntegrity assessment Integrity shall be judged with respect to all the single units or seissie seismic isolation elements, assessment consideretins for isolation eleaents. buildings »nd 9tructur«s saisnic isolation elements

S.1.2 Integrity assessment Assessment of the intensity or stress and strain, strength and stability. considftrations for buildings and structures

S.S.I Sattlng tha restoring seismic Isolation elements force character ist ics acceptance tests. of seisale Isolation eleitants

8.2.2 Cross-sectional forces For calculation, the top and bottoa foundation slabs shall be assumed to be acting on single unit of rigid. seisaic isolation elenents

8.2.3 Integrity assessment for Allowable limits and sarety margin* concerning the integrity of seismic the shearing, tensile and isolation elements according to load condition.

of laninated rubber bearings

Damping performance under allowable limit to be set shall be secured. daapers

I rsj r*bi« t Main Contents of F B R Seismic Isolation System Design Methods

Chapter Saotlon Subsection Wain Contaats

8. Integrity assonant or 8.3 Intagrity assessaent or 8.3.1 Assessaent or the Shall conrora to Os-class facilities shown in Aseisaic Design Guidelines. seisaic isolation eleaents, buildings and structures intensity or stress and buildings and structures strain

8.3.2 Assessaant or strength Approprleto sarety aargin and saisaic isolation structural characteristics shall be considerd.

8.3.3 Assessaant or stability The contact between aottoa foundation stabs and ground and resistant aginst sliding shall be con'iraed.

9. Design procedure of 9.1 Basic deaingn principle 9.1.1 Flow or sarety assessaant Saae as conventional earthquake-resistant plants. Appropriate stress assessment equipaent and piping systea shall be conducted on connection piping in reference to relative displaceaent during earthquake.

9.1.2 Aseisaic priority Aseisaic Design Guidelines shall be followed. classification

9.1.3 Aseisaio design Dseisaic Design Ouidelines shall be followed. For horizontal static assessaent aethods considered. 9.1.4 Load ooablnations and aseisaio Design Guidelines shall be followed. Appropriate standards permissible Units shall be applied to connection piping in reference to relative dtsplaeeaent during earthquake.

9.2 Structural planning 9.2.1 Structural plan for Construction thet can accoaodate relative displaceaent during earthquake. connection piping, etc. 9.3 Selsaic response analysis 9.3.1 Floor response spectrua Expansion or floor response speetrua obtained In consideration or seisaic isolation eleaent*s aanuracturing errors and aging and the effect of environaenta1 conditions.

9.3.2 Nodeling Response analysis of connection piping. Sloshing response analysis. Confiraition of the effect of torsional response.

IB. Ouatity control and 11.1 Ouility asauranca Observence of quality assurance guidelines for nuclear powerplents (JE«O 41B1- aeinte nance or seisale I98S) isolation eleeents IB.* Acceptance tests or Rubber aaterial tests. Execution or laainated-rubber eleeent characteristics eleaent characteristics acknowledgeaent tests. 10.3 Quality control standards Scope or application. Basic consideration, duality control in aanufaeturing or seisaic isolation processes. Product inspection. Forwarding docuaents. eteaents

IB.4 Construction aethods tB.4.1 Construction Methods Prevention of haraful residual deforaation and soatter In the axial force of seisaio isolation eteaents.

Construction control plan considering types of setsalo isolation aleaents and

IB.5 naintanaca or tB.S.I Objectives Ensuring durability and iaproving reliability. saisaic isolation el««ents

r\» 00 Table i Main Contents of F B R Seismic Isolation System Design Methods (B--8)

Chapter Section SubsectIon Haln Contents

IB. Oueiity control and 18.5 Maintenance or seisaic 19.5.2 Scop* of application Daily inspection. Periodic inspection. Inspection incase or abnormalities. •aintenance of seisnic isolation elements Assessment and recording. isolet ion elenents IB.5.3 Daily inspection Checkino of cracks, derornatlon. corrosion, looseness and other conditions.

1B.S.4 Periodic inspection Inspection by sight. Study of dynasie observation reoords. Dynasic characteristics tests as reanired or coating of steel parts.

10.S.S Inspection in case of Functional cheeks due to occurrence or disasters. Counteraeasures against abnormalitits abnormalities.

IS.5.6 Treatment of inspection Clarirication or execution syste*. results

10.6.7 Recording Recording and preservetion of inspection results.

10 XA0055389

TECHNICAL SPECIFICATIONS FOR THE SUCCESSFUL FABRICATION OF LAMINATED SEISMIC ISOLATION BEARINGS

R. F. Kulak Argonne National Laboratory Argonne, Illinois U.S.A.

ABSTRACT

High damping steel-laminated elastomeric seismic isolation bearings are becoming a preferred device for isolating large buildings and structures. In the United States, the current reference design for the Advanced Liquid Metal Reactor uses laminated bearings for seismic isolation. These bearings are constructed from alternating layers of rubber and steel plates. They are typically designed for shear strains between 50 to 100 percent and expected to sustain two to three times these levels for beyond design basis loading considerations. The technical specifications used to procure these bearings are an important factor in assuring that the bearings meet the performance requirements of the design. The key aspects of the current version of the Technical Specifications are discussed in this paper.

INTRODUCTION

The use of seismic base isolation has been increasing rapidly for such critical facilities as computer centers, medical centers, and emergency control facilities. Base isolation has been employed in many buildings in the United Kingdom and Europe to control unwanted vibrations from subway systems. In New Zealand, road and rail bridges have been the most common structures isolated. Italy used seismic isolation on several bridges, civil buildings and industrial structures. They are devoting considerable effort to this technique. Base isolation has also been used in Greece, the former USSR, and China. The Western United States has several buildings that employ seismic isolators. Japan has been the most aggressive in adapting isolation to their structures. Now, they have more than fifty buildings that use or plan to use seismic isolation. Seismic isolation is beginning to be accepted for use in nuclear facilities as evidenced by its employment in two French designed nuclear power plants: Cruas-Meysee, France and Koeberg, South Africa.

The laminated elastomer bearing is emerging as a preferred device for large buildings/

-230- structures (i.e., with no more than eight stories in height). Bearings that were used in the United Kingdom have experienced more than 30 years of service and have performed well. Laminated bearings have been used for both new construction and retrofitting existing structures. For example, the Salt Lake City and County Building, which was completed in 1894, was retrofitted with laminated elastomer bearings. The United States and Japan are seriously considering this design strategy for some of their future plants. The Advanced Liquid Metal Reactor (ALMR), which currently is the reference design in the United States, uses laminated bearings.

Currently, there are two designs for the laminated elastomer bearing. One design relies on a central lead plug (cylinder) to provide damping, and the other uses a special rubber compound to dissipate energy. As part of the U.S. program to evaluate the performance of these bearings, Argonne initiated an experimental test program. The program, however, only considered the highly damped rubber type of bearing because of time constraints. More than fifty bearings were purchased and tested. Several different designs were tested to evaluate the effects of shape factor, shear modulus, damping and mounting connection. The bearings were purchased from several manufacturers. The U.S. Department of Energy (DOE) and the National Science Foundation (NSF) funded the program.

Early in this research effort, it became clear that the technical specifications used to procure the bearings were critical in obtaining high quality bearings that would perform as designed during an earthquake. The experience gained has shown that a balance between a prescription based and a performance based specification was optimum. Under specification can lead to an inferior bearing and over specification can lead to a bearing that cannot be made or one that is unnecessarily expensive.

Recent experience has shown that the specifications must consider the manufacturing process and the type of bearing testing equipment available at the plant. For example, the design frequency of a typical isolated structure is between 0.4 and 0.8 Hz. Bearing manufacturers, however, do not have test machines that can operate in this frequency range. Typically, manufacturers perform Stiffness tests at a Mountlna Connection Cover Rubber frequency of 0.005 Hz. The value of the bearing stiffness called for in the technical specifications must be adjusted to compensate Bulk for the much lower testing speed. Also, the Rubber paper presents other significant findings. vSh!m Plates LAMINATED ELASTOMER BEARINGS 7 The laminated bearing (Fig. 1) is End Plate constructed from alternating thin layers of elastomer and metallic plates (shims) that are Fig. 1. Typical Steel-Laminated Elastomer bonded together during the vulcanization Seismic Isolation Bearing.

-231- -3- process. The elastomer is usually a carbon filled natural rubber that exhibits damping when subjected to shear. Recently, some blends of natural and synthetic rubbers have appeared. A thick plate is at each end of the bearing for mounting to flange plates that, in turn, are attached to the basemat and upper structure. In the United States, two methods have been used to attach the bearing to its flange plates: bolts and dowels. Recent tests results show bolting is the better of the two. A cover layer (jacket) of rubber encases the bearing to provide protection from environmental factors. In some designs, this cover layer is a special blend of rubber, and in others it is the same rubber as the bulk rubber of the bearing.

CODES AND STANDARDS

A set of codes and standards for structural steel, rubber and quality control and inspection are given in the American Society for Testing and Materials Standard1 (ASTM), The British Standard2 (BS) and the Military Specification and Standards 3 (MIL). Generally, these standards (Table 1) are followed in the steel-laminated elastomeric seismic isolation bearing specifications. As seen from the table, these codes and standards pertain to structural steel, rubber, quality control and inspection. In areas where more specific guidance is required, explicit requirements and procedures are given in the Technical Specification document itself. Some of these specific areas are discussed in this paper. Note, a standard specifically for steel-laminated seismic isolation bearings does not exists. ASTM D 4014 only provides guidance for steel-laminated elastomeric bridge bearings. These bearings do not see the strain levels experienced by seismic isolation bearings during an earthquake.

BEARING DESIGN PARAMETERS AND MECHANICAL PROPERTIES

Perhaps the most important aspect of the Technical Specifications is the specification of the bearing design values. In effect, these are the performance goals the manufactured bearings must meet. Values for the following quantities must be given: dead load, design load, design frequency, design shear strain, ultimate shear strain, stiffness and damping. The elastomer is not required to meet a specific hardness, but hardness measurements must be taken and the value reported.

Stiffness and damping are two important quantities that determine the response of an isolated structure. Because several different definitions for these quantities are being used by designers and isolation bearing manufactures, it is necessary to be very explicit in defining these parameters in the Technical Specifications.

Shear Stiffness

The shear stiffness of the isolator is a quantity that governs the fundamental horizontal frequency of a base isolated system. Because of the relatively high shear modulus of the steel, it is the shear modulus of the elastomer that determines the shear stiffness of the bearing.

-232 -4-

Table 1. Reference codes and standards used for laminated seismic isolation bearings.

Structural Steel ASTM A36 Specification for Structural Steel ASTM A570 Specification for Hot-rolled Carbon Steel Sheet and Strip, Structural Quality Rubber Strength and Elasticity ASTM D395 Test Methods for Rubber Property - Compression ASTM D412 Test Methods for Rubber Properties in Tension ASTM D429 Test Methods for Rubber Property - Adhesion to Rigid Substrates ASTM D518 Test Method for Rubber Deterioration - Surface Cracking ASTM D573 Test Method for Rubber Deterioration in an Air Oven ASTM D1149 Test for Rubber Deterioration - Surface Ozone Cracking in a Chamber (Flat Specimens) ASTM D1229 Test Method for Rubber Property - Compression Set at Low Temperatures ASTM D1415 Standard Test Method for Rubber Property - International Hardness ASTM D2137 Test Methods for Rubber Property - Brittleness Point and Flexible Polymers and Cooked Fabrics ASTM D2240 Test Method for Rubber Property - Durometer Hardness ASTM D4014 Specification for Plain and Steel-Laminated Elastomeric Bearings for Bridges with Annex (Al-Determination of Shear Modulus) BS903 Methods of Testing Vulcanized Rubber: Part A15 - Determination of Creep and Stress Relaxation Quality Control and Inspection MIL-I-45208A Inspection System Requirements

-233- -5-

Extension Fig. 2. Cord Shear Modulus as Defined in Fig. 3. Definitions of Shear Moduli. ASTM D 4014.

However, there are several different definitions being used for this quantity. ASTM D4014, which was established for steel-laminated elastomeric bridge bearings, defines a cord shear modulus at approximately the 50 percent strain level (Fig. 2). Since current practice uses design shear strains up to 100 percent, a strict adherence to this standard would not provide any useful information. Also, a "cord" modulus is not much use to a designer of seismic isolation systems. Rubber compounders use storage, loss or complex moduli, and seismic isolation system designers uses an effective modulus. The difference in these moduli is shown below. The storage modulus, G', is given by

t(Ymax) " G' = (1) " Yn

an are tne where Ymax d Ymax maximum positive and negative shear strains, respectively, that occur s during a complete hysteresis loop, and T(Ymax) i defined to be the shear stress at Ymax effective shear modulus, Geff, is given by

Ax '''m (2) Ay" " Yn

where xBm and Tmax are the maximum positive and negative shear stresses, respectively. Figure 3 shows the shear moduli. Note, for typical low strain (e.g., 50 percent) hysteresis loops Eqs. 1 and 2 can give values that differ by 12 percent. Since the values needed for Eq. 2 can be determined with greater accuracy, this definition has been chosen to define the bearing stiffness. Note, values for the storage, loss and complex moduli are not specified in the document, however, these quantities must be measured and their values reported.

-234- -6-

Stress Stress

Strain Strain

Fig. 4. Energy Dissipated During a Cycle Fig. 5. Stored Energy During a Cycle Shown Shown as Shaded Area. as Shaded Area.

The definition chosen for the shear modulus must be explicitly spelled out in the technical specifications. Otherwise, the compounder may choose a shear modulus definition other than the one intended by the designer and produce an elastomer with a shear stiffness outside the design range.

Damping

Damping is another quantity used to characterize elastomers. Like stiffness there are several terms used to describe damping: loss angle, loss tangent, damping ratio, percent of critical damping and effective damping ratio. In our specifications the effective damping ratio was chosen as the measure of damping. The effective damping ratio, Y\ is defined by

r\ = (3)

Here Us is the energy stored during a cycle and UD is the energy dissipated during the cycle. Figures 4 and 5 illustrate these quantities.

ELASTOMER SPECIMEN TESTING

The material property specification for the elastomer and the testing for those properties

-235- -7- is an important part of the process for producing a successful bearing. Values for the following properties are given in the specifications: (1) minimum elongation at failure, (2) .,,™, ^,«.,.,„ , ^ „ „, „ minimum tensile strength, (3) Fig. 6. ASTM D4014 Style Four-Bar Shear Specimen. effective shear modulus at the design shear strain and design frequency, (4) minimum damping ratio at the design strain and design frequency, and maximum damping ratio at the design strain and design frequency.

The determination of the shear response of the elastomer requires the use of a special test specimen. A test procedure and a suggested test specimen for shear tests are included in ASTM D 4014. This procedure only considers non-reversed cyclical loading, that is from zero to the target strain. The suggested specimen has two sets of rubber pads attached to four steel bars ( Fig. 6). To obtain shear data used for earthquake type loading, fully reversed cyclical shear testing must be performed. Experience has shown that the four-bar specimen becomes unstable during fully reversed cyclical testing. Engineers at LTV Energy Products Co. (Arlington TX) developed a three bar lap shear specimen (Fig. 7) that retains stability during reversed cyclical loading. There are two rubber pads in the specimen each being, nominally, 1 x 1 x 0.2 inches. ANL currently uses this design for all its elastomer testing and requires this type of specimen in r- the Technical Specifications. V////////A

Recent experience has r W///////A shown that the specifications must consider the manufacturing process Fig. 7. LTV Style Three-Bar Shear Specimen, and the type of bearing testing equipment available at the plant. For example, the design frequency of a typical isolated structure is between 0.4 and 0.8 Hz. Note, bearing manufacturers do not have test machines that can operate in this frequency range. Typically, manufacturers perform stiffness tests at a frequency of approximately 0.005 Hz, which is two decades lower than current design frequencies. Figure 8 shows the variation in shear modulus with frequency at a shear strain level of 100 percent. The results were obtained from tests performed at Argonne using LTV style three bar lap shear specimens. The value of the bearing stiffness called for in the technical specifications must be adjusted to compensate for the much lower testing speed. In the United States, a bearing testing machine with the capacity to test medium sized bearings at the design frequency is available at the Earthquake Engineering Research Center (EERC) at the University of California at Berkeley and a large capacity machine is available at the Energy Technology Engineering Center (ETEC) at Rockwell International Corporation.

A requirement of the specifications is to submit plots for all stress-strain tests to the buyer. These plots are used to verify that the elastomer meets the specified values at the design strain and design frequency. They also provide additional data on the variation of the mechanical

-236- -8- properties over the testing range. 200 Assuring that the bond strength between the rubber and steel is adequate, is one important goal of the specifications. The standard adhesion strength tests (ASTM D429 Method A) and peel strength tests (ASTM D429 Method B) are required. In addition, the specifications require that shim plates be removed from the production line and used in the bond tests. The specifications require the .01 .1 removal of the "worst" looking shim plate Frequency, Hz from the production batch for use in a peel strength test. By following this procedure a Fig. 8. Variation of Shear Modulus With check on the production process is obtained. Frequency at a Strain Level of 100 Percent.

FABRICATION

The bearings are molded as a unit and vulcanized under heat and pressure. The manufacturer must provide complete and detailed process control procedures and specifications for buyer approval before fabrication.

The manufacturer must mold a special proof-of-process bearing to verify the vulcanization process. The test bearing is molded without the use of bonding agents. After vulcanization, the steel and rubber layers are separated, and the rubber layers inspected to evaluate the cure throughout the bearing. Hardness readings (Durometer/IRHD) must be taken across the top and middle elastomer layers. A diametric line (Fig. 9) is divided into eleven (11) intervals and readings taken at the center of the intervals. The readings should be reported. The hardness readings can be related to the shear modulus, and the manufacturer can then judge if the Elastomer vulcanization process produced the desired Layer shear modulus throughout the bearing. All layers should be visibly examined for defects Measurement (e.g., porosity). Each layer should be Locations measured to check the layer thickness for uniformity. Production bearings (i.e., with bonding agents) will be made only after a satisfactory vulcanization process has been verified.

Fig. 9. Location of Hardness Measurements on Laminates of Proof-of-Process Bearing.

-237- -9-

COMPLETED BEARING TESTS

A series of tests are performed on the completed bearing. These include a sustained compression test, compression tests and combined compression and shear tests. The sustained compression test subjects the bearing to 1.5 times its design load for 24 hours. Visual inspections for failures, such as debonding of rubber-to-steel and surface cracks, are required during and after the tests. Note, for bearings with thick cover layers, it may not be possible to detect debonding from this test.

The most valuable tests performed on the completed bearing is the combined compression and shear tests series. The first test in this sequence is a compression shear test performed with the bearing loaded to its compressive design load and subjected to five complete reversal loading cycles to plus and minus the design shear strain. The test is performed at 200 sec/cycle (0.005 HZ). This test determines the shear stiffness of the bearing. However, as stated earlier, the value of this stiffness must be adjusted to account for the slower testing rate.

In earlier versions of these specifications, the above shear-compression test was the only test required. Because of Argonne's experimental testing program of elastomeric seismic isolation bearings, it was learned that the above test may not detect poor bonding. The large compressive loading generates enough friction to preclude poor bearing performance and prevent detection of faulty bonding. This can be viewed as a positive safety feature for seismic isolation systems that are designed to have all bearings in compression during an earthquake. However, the manufactured bearings should be properly bonded and the following test was added to the specifications to assess bond integrity. Immediately following the successful completion of the above test, the combined compression-shear test is repeated with the vertical load reduced to zero. The manufacturer must visually inspect the bearing during and after the test, and compare the load-deflection plots for discrepancies between the test with the design axial load and the test with zero axial load.

The bearing manufacturer must submit to the buyer certified reports of the results of all proof testing and other data to show that they have meet the performance specifications prior to bearing shipment. The bearing must be protected from damage during shipment to the final destination.

CONCLUSIONS

Technical Specifications for the procurement of steel-laminated elastomeric seismic isolation bearings have evolved from a set of pre-existing codes and standards from the following sources: the American Society of Testing Materials (ASTM), The British Standard (BS) and the Military Specification (MIL). Since these codes and standards were not specifically written for laminated seismic isolation bearings, they had to be supplemented with additional guidelines. These additional guidelines came from discussions with bearing manufactures, bearing designers

-238- -10-

and bearing research engineers.

Some key findings of the research are summarized below. It was found that cyclical testing of the elastomer and bearing is needed to obtain performance characteristics during earthquake type motions. A special testing fixture, which maintains stability under cyclical reversed loadings, is required for elastomer specimen testing. Definitions for stiffness and damping must be clearly stated to assure that the rubber compounder and bearing designer are using the same terms. The first bearing must be molded without the use of bonding agents so that a post mortem examination of the elastomer layers can be performed to validate the vulcanization process. The test machines that bearing manufacturer's use to proof test completed bearings operate at frequencies that are several orders of magnitude lower those found in the earthquake spectrum. Thus, the testing speed of the proof test must be considered when evaluating the bearing for acceptance. The completed bearing must be tested in shear under zero or, perhaps, some tensile load to detect faulty bonding.

The research that has lead to these technical specifications is still underway. It is expected that some additional guidelines will be added in the future. However, these will be in a fine tuning category. These Technical Specifications bridge the gap between the performance goals of designers and the constraints imposed by the manufacturing process.

ACKNOWLEDGEMENTS

The author wishes to acknowledge the following people with whom he had helpful discussions during the performance of this work: Jack Rainbolt, Gary Whightsil and Mike Hogan of LTV Energy Products Co. in Arlington TX; Vince Coveney of Bristol Polytechnic in England; Keith Fuller of Rubber Consultants in England; Steve Bennett of Furon-Structural Bearing Division in Athens, TX; Jim Kelly of the Earthquake Engineering Research Center at the University of California at Berkeley; Vince DeVita and Paul Chen of Energy Technology Engineering Center in Canoga Park, CA; Fred Tajirian of Bechtel National, Inc. San Francisco, CA; Stan Sitta, Elastomer Consultant in Wheaton, IL; and Ralph Seidensticker, Yao Chang and Tom Hughes of Argonne National Laboratory.

This work was performed as part of the work in the Engineering Mechanics Program of the Reactor Engineering Division of Argonne National Laboratory under the auspices of the U.S. Department of Energy, Contract No. W-31-109-Eng-38.

REFERENCES

The following are global references for the specific codes and standards cited in the text.

1. 1991 Annual Book of ASTM Standards, ASTM, 1916 Race Street, Philadelphia, PA 19103-1187 USA.

-239- -11- 2. British Standards Institute, England.

3. Military Specifications and Standards Service, Information Handling Services, 15 Inverness Way East, P. O. Box 1154, Englewood, CO 80150 USA.

-240- XA0055390

STUDY OF SEISMIC RESPONSES OF

CANDU3 REACTOR BUILDING

USING ISOLATOR BEARINGS

J. K. Biswas Seismic Specialist AECL CANDU Sheridan Park Research Community Mississauga, Ontario, Canada

FOR PRESENTATION AT THE IAEA SPECIALISTS' MEETING ON ISOLATION TECHNOLOGY AT SAN JOSE, CALIFORNIA, MARCH 1992

gauas/userc/btewas/ieismic -9/11 iaeafisolation/ fc^x 82-03-13 ABSTRACT

Seismic isolator bearings are known to increase reliability, reduce cost and increase the potential sitings for nuclear power plants located in regions of high seismicity. High seismic activities in Canada occur mainly in the western coast, the Grand Banks and regions of Quebec along the St. Lawrence river. In Canada, nuclear power plants are located in Ontario, Quebec and New Brunswickwhere the seismicity levels are low to moderate. Consequently, seismic isolator bearings have not been used in the existing nuclear power plants in Canada. The present paper examines the effect of using seismic isolator bearings in the design for the new CANDU3 which would be suitable for regions having high seismicity.

The CANDU3 Nuclear Power Plant is rated at 450 MW of net output power and is a smaller version of its predecessor CANDU6 successfully operating in Canada and abroad. The design of CAND U3 is being developed by AECL CANDU. Advanced technologies for design, construction and plant operation have been utilized. During the conceptual development of the CANDU3 design, various design options including the use of isolator bearings were considered. The present paper presents an overview of seismic isolation technology and summarizes the analytical work for predicting the seismic behavior of the CANDU3 reactor building.

A lumped-parameter dynamic model for the reactor building is usedfor the analysis. The characteristics of the bearings are utilized in the analysis work. The time-history modal analysis has been used to compute the seismic responses. Seismic responses of the reactor building with and without isolator bearings are compared. The isolator bearings are found to reduce the accelerations of the reactor building. As a result, a lower level of seismic qualification for components and systems would be required. The use of these bearings however increases rigid body seismic displacements of the structure requiring special considerations in the layout and interfaces for interconnected systems. The relative advantages and disadvantages of the potential use of isolator bearings in CANDU3 are discussed.

-242 INTRODUCTION Seismic isolation is a remarkable advancement in earthquake engineering which is gaining international acceptance. It can be used to reduce the seismic loads on nuclear power plants and other structures specifically in locations of high seismicity. Today's nuclear power plants and other critical structures are designed to accommodate the effects of high levels of seismicity to meet safety and licensing requirements. Reduction of seismically induced loads results in an economic structural design. When using isolation bearings, the secondary components and systems located in the plant would require a lesser degree of seismic qualification. It results in cost savings in two ways. Firstly, it avoids strengthening and implementations of costly design features which are otherwise necessary. Secondly, due to the decrease in the seismic response levels, a relatively simple analysis procedure can be employed to predict the behaviour of secondary components. For example, supported systems will behave in a linear elastic way and the need for non-linear analysis considering impact, sliding, gaps, etc. can be reduced if not avoided. Isolation devices that are developed today can increase the safety margin of structures. It is claimed that the behaviour of an isolated structure can be predicated with a higher degree of accuracy. The properties of bearing materials are tested and controlled during the manufacturing process whereas the soil properties are subjected to a higher level of uncertainty. Isolation bearings increase the potential of using a standard design in different sites without undertaking costly redesign work. To date many buildings around the world are designed and built using isolation devices. At present there are more than 18 buildings on isolation bearings constructed in Japan [1] and many more have received construction permits. The Foothill Community Centre in San Bernardino County, California is the first building in the United States built with isolation bearings in 1986. In many buildings and bridges built in the U.S.A. the use of isolation bearings have reduced the seismic loads. Bearings are often used for retrofiting existing installations to higher seismic requirements or rehabilitating structures damaged by earthquakes. In New Zealand a number of industrial buildings and 37 bridges are built using isolation bearings. In a number of other countries, isolation devices have been used for buildings, bridges and heavy equipment. In Japan, U.S.A, United Kingdom, Italy and many other countries, research work including testing of bearings and monitoring of seismic responses of demonstration structures are being undertaken with the object of understanding the behaviour of various isolation systems.

gauss/useis/bswaifceterritf laea/isolaiion/ -243- 82-03-13 The adaptation of seismic isolation bearings for nuclear facilities has been slow. The Cruas in France is the first nuclear power plant built with isolation devices in 1982 [2]. At present mere are a total of six base isolated standard PWR units. Four units are located in Cruas, France and two units in South Africa [3]. There is one fuel processing plant in England that uses isolation devices. Many studies and conceptual designs have included the use of base isolation using bearings. A nuclear waste storage facility has been built in France using seismic isolation bearings [4]. Isolation bearings have been used in the conceptual design of the 1500 MW Liquid Metal Fast Breeder Reactor (LMFB). Countries interested in the development and use of this technology are sponsoring joint research activities. The U.K., EPRI and the Japanese Central Research Institute of Electric Power Industry (CRIEPI) have undertaken a joint international program for advancement of this technology. As part of this program, the technical feasibility of selected isolation systems have been evaluated for application in large Liquid Metal Reactors (LMR) and the European Fast Reactor (EFR) plants. Analytical work and testing of representative sample bearings for two U.S. compact LMR concepts have been reported [5].

Nuclear power plants built in Canada are strengthened to carry loads due to the design basis earthquake. Since the first use of isolation bearings, AECL CANDU has maintained an active interest in this technology. During the conceptual design of the CANDU3 nuclear power plant, the use of isolation bearings are considered. The present paper deals with an analytical study to predict the effects of using isolation bearings for the CANDU3 reactor building.

CANDU NUCLEAR PLANT Thirty three CANDU nuclear power plants have been built or are under construction in Canada and abroad. These plants use the CANDU (CANadian Deutrium Uranium) power reactor technology. The power rating of these plants ranges from 203 MW to 881 MW. There are twenty nuclear power plant units located in Ontario. The latest of which is a four unit station at Darlington having the highest capacity of 881 MW per unit. The CANDU6 plant is rated for 670 MW of electric power. One CANDU6 plant has been built at Gentilly, Quebec and another at Point Lepreau, New Brunswick both of which are operating. CANDU6 plants have been built and are operating in Argentina and South Korea. At Cernavoda in Romania a five unit CANDU6 nuclear

gauM/utert/bkwu/stkfric/ 9AA- laea/Isolation/ -fc«tt 92 -OS-13 generating station is under construction. Design and construction of the second power plant unit in South Korea is continuing. CANDU power plants in Canada and abroad built to date are of fixed base design. These plants are strengthened to sustain the effects of the design basis earthquakes (DBE). The requirements of seismic design is followed in Canada according to the guide-lines given in the Canadian National Standards [6] . High seismic activities in Canada mainly occur in the western coast, the Grand Banks and in regions of Quebec along the Saint Lawrence river. In the Charlevoix region of Quebec, high level seismic activity occurs very frequently. In other parts of the country, the seismicity level is low to moderate. Nuclear power plants in Canada are located in regions where the seismicity level is low to moderate. The highest level of seismic acceleration for CANDU plants built to date is 0.2 g ground acceleration. Studies have been undertaken for siting CANDU plants in other countries having high seismicity levels of 0.5 g ground acceleration.

CANADIAN EXPERIENCE Experience of using isolating devices in Canada is somewhat limited. A low level of seismic activity in most parts of the country did not necessitate the use of these devices. Moreover, there has been an initial reluctance in adopting a new technology until it has performed satisfactorily. The first use of isolation devices in Canada has been for a coal ship loader at Prince Rupert in British Columbia. The benefit of friction dampers for framed structures was analytically investigated [7]. Such a system of friction dampers has been used in the new design of the library building for the Concordia University in Montreal [8]. By installing friction dampers in steel cross bracings, a large amount of seismic energy could be dessipated mechanically. The use of these devices in the braced frame structure resulted in cost savings by eliminating the shear walls that would be required otherwise. A tuned vibration control device has been installed in the CN tower located in Toronto to reduce the effects of wind induced vibrations. Specially designed devices are used in a number of high-rise buildings in Toronto to reduce subway vibrations and noises. In CANDU plants, heavy equipment such as diesel generators are mounted on vibration isolating devices. These devices consist of coil springs of specific characteristics to eliminate amplification of machine vibrations. Due to the low stiffness, these springs are also useful to provide some reduction of seismic loads.

gauu/uurs/btewas/telerriG' o /I C taea/isolaMon/ -CtS- 82-03 13 Currently AECL CANDU is developing the design of CANDU3 nuclear power plant rated at 450 MW of net electric output. This is a smaller version of its predecessor CANDU6 which is successfully operating in Canada and abroad. In CANDU3 advanced technologies for design, construction and operations have been utilized. The design uses the concept of modular construction to reduce schedule and cost. During the conceptual development of the CANDU3 design various options including the the use of isolation bearings were considered. A study of the effect of seismic isolation bearings on the responses of the CANDU3 reactor building is presented in this paper.

SEISMIC ISOLATION SYSTEM The cross section of the CANDU3 reactor building is shown in Figure 1. The CANDU3 reactor building weighs approximately 450,000 KN and has a height to diameter ratio of about 1. For such a short stubby structure, the use of a seismic isolation device is known to be useful. For the CANDU3 plant a design basis earthquake with a ground acceleration of 0.3 g is considered. The ground response spectra for the design earthquake are shown in Figure 2. It is evident from the spectra that to obtain the benefits of base isolation the horizontal frequency has to be lowered below 1.5 Hz. The target design frequency of the isolated structure in the horizontal direction is chosen to be 0.6 Hz. To provide seismic isolation, a number of isomeric bearings having high shape factors are chosen. The seismic excitations in the vertical direction is generally low and therefore is not of concern. The configuration of the proposed bearing is shown in Figure 3. The bearing is circular in shape, 0.5 m in height and has a diameter of 1.27 m. The end plates are bolted to the concrete slab for shear transfer. The chosen bearing has 24 shim plates impregnated in the elastomer. The material of the bearing will have to be selected to meet the requirements of the analysis. The bearings will be located between a lower slab and an upper slab on top of concrete pedestals as shown in Figure 4. This arrangement provides space needed for inspection and replacement of these bearings during the 100 year life of the plant. The horizontal stiffness (kh) and the vertical stiffness (kv) of one bearing are estimated using the following commonly used formulae:

kh = GAt/nt (1)

kv = Ec As / n t (2)

eauu/uurt/bkwaa/»itmic/ lMa/i*olallon/ G2-03-13 Where G is the shear modulus of elastomer

Ec is the effective compression modulus

At is the total area

As is the shim area n is the number of layers t is the thickness

The effective modulus Ec used in the computation of the vertical stiffness depends on the shape factor. The shape factor depends on the geometry and is defined by the the ratio of the loaded area to the area free to bulge. The shape factor of the chosen bearing is obtained as 23 using the following expression for a circular bearing:

S= d/4t (3)

Where d is the diameter at the shim

Using equations 1 and 2 the horizontal and the vertical stiffness of one bearing are computed to be 2800 KN/m and 3000000 KN/m respectively. Some degree of uncertainty exist in the stiffness estimation of these elastomeric bearings using these formulae. To obtain better estimates of stiffness properties, finite element analyses of these bearings are necessary. Additionally testing of samples to define the properties of elastomer may have to be undertaken. Such works will be needed before adopting the system. The present approximate estimates of bearing stiffness are considered appropriate for a conceptual study. From testing of similar bearings it is known that the properties of these bearings are nonlinear [8]. It has been found that initially the horizontal stiffness decreases with shear strain. At very high levels of shear stain the stiffness increases. The vertical stiffness of such bearings increase slightly with loads. For the purpose of this study the characteristics of these bearings are assumed to be linear.

SEISMIC ANALYSIS As shown in Figure 1 the CANDU3 reactor building consists of an internal concrete structure and equipment modules enclosed by a cylindrical concrete containment structure with a dome at the top. The internal structure houses the

flaus6/usersA>tswas/S9ism&' taaa'isotaion/ -247- reactor and various components and systems necessary for the operation of the plant. For seismic analysis, the reactor building has been represented by a mathematical model consisting of lumped masses and beams as shown in Figure 5. In this model the beams represent the stiffness of different structural parts of the building and the masses are lumped at a number of key locations. The isolation bearings are represented by one set of springs having equivalent stiffness values. The effect of all bearings are combined into one set of springs in the following way:

Kh = N kh (4)

Kv = N kv (5)

2 Kr = Kv D /16 (6)

2 Kt = Kh D /8 (7)

Where Kh is the horizontal stiffness

Kv is the vertical stiffness

Kr is the rocking stiffness

Kt is the torsional stiffness N is the number of bearings D is the diameter of the base slab

A total of 200 bearings are assumed for the study. The spring properties of the isolation system are computed using equations 4 to 7. In the present analysis, the effect of soil-structure interaction is ignored. The isolation system produces a low frequency dynamic response much lower than the soil-structure interaction frequency. Consequently, the response of the structure is predominantly influenced by the stiffness of the isolation system. The effect of soil structure interaction for the hard soil condition is small and and can be neglected. The analysis is done using the modal time-history method. A response compatible time-history generated using methods suggested in Ref. 10 has

Sauu/uMrt/bkwas/telsmic -248- taM/lsolaiJcvt' 92*3-13 been utilised. The computer program STARDYNE has been used for the analysis. The study is done for two cases with different values of damping. The first case uses a damping value of 8 % which is applicable for rubber bearings. The second case uses a damping value of 15 % to represent the damping value of a high damping elastomer. The additional damping value can also be obtained by using specially designed damping devices. High damping elastomeric bearings and a variety of damping devices have been used in actual buildings constructed in Japan [1].

DISCUSSION OF RESULTS From the analysis the horizontal frequency of the isolated structure is obtained to be 0.61 Hz. The horizontal response acceleration along the height of the building is plotted in Figure 6 and compared with that for the fixed base structure. Substantial reduction of seismic responses are noted for the isolated building for both cases. For the low damping case (Case 1) the acceleration at all points has a constant value of 0.31 g. For the high damping case (Case 2) the building acceleration at all locations is 0.22 g and is lower than the ground acceleration. The horizontal displacement of the building along the height of the structure is plotted in Figure 7 and compared with that of the fixed base structure. The displacement of the isolated structure is 175 mm to 200 mm which is much higher than the displacement of the fixed base structure (15 mm). Such increased displacement is characteristic of an isolated structure. The design of the secondary components governed by the floor response spectra. The floor response spectrum in the horizontal direction for the internal structure at elevation 122.5 m in the reactor building is shown in Figure 8 along with the floor response spectrum for the fixed base structure. The horizontal floor response spectrum at the top of containment is given in Figure 9. For the base isolated building the floor response spectra are substantially lower than those for the fixed base structure for both cases. It is also noted the seismic responses do not vary with the height of the structure. At the DBE level, these bearings are subjected to shear strains of 60 %. Such bearings are known to sustain shear strains in excess of 300 %.

gauss/users/blswas/telsnnic/ -249- taaa/isolaiion' 82-03-13 CONCLUSIONS Based on the study of seismic responses of the CANDU3 reactor building, the following conclusions can be drawn: • The use of isolation devices for a CANDU power plant structure can be considered as an alternative to strengthening and designing for the seismic loads. Large reductions of seismic loads is possible by using such a system. • Isolation devices reduce the accelerations and floor response spectra drastically. The seismic requirements of systems and components, in the frequency range between 3Hz to 33 Hz are substantially lower. • The use of isolation devices increases the displacement of the structure by a large amount. Displacements of the order of 200 mm will require special considerations for interconnected systems such as the steam mains and other pipes connected to the BOP. • To obtain full advantage of the base isolation, most buildings of the plant should be located on a common mat. This will result in less number of cases to design systems to accommodate large relative displacements of the buildings. • Considerable advancement of isolation technology has been achieved in recent years. Further R and D work on properties of the elastomeric bearing is needed. Bearing characteristics need to be expressed in simple design formulae which can be utilised for the analysis and design work.

The isolated system has not been adopted for CANDU3. The added complexity in the design to accommodate large displacements, the penalty to the construction schedule and the added cost of these isolation devices are some of the deterrents in adopting the base isolated system. Further studies to evaluate the economic benefit of using such systems would need to be explored. More experience from the behaviour actual isolated structure subjected to real earthquakes is also needed. Application of this technology, perhaps on an installation on a large common mat would not be ruled out for future consideration.

gaust/uMitftitwat/salsmis' « c A iaaafoolaibn/ -tOU- REFERENCES 1. Kelly, J. M., 1988 Base Isolation in Japan 1988, Report No. UCB/EERC-88/20, Earthquake Engineering Research Centre, California Berklay, U.S.A. 2. Jolivet, F., and Richli, M., Aseismic Foundation for Nuclear Power Stations, Transactions of the Fourth International Conference of Structural Mechanics in Reactor Technology, Paper K 9/2, August 1977. 3. Kircher, C.A., et al, Overview of Seismic Isolation and Application to Nuclear Facilities, Proceedings of the Third Symposium on Current Issues Related to Nuclear Power Plant Structure, Equipment and Piping, Dec 1990, Orlando, Florida. 4. Buckle, I. G., and Mayes, R.L., Seismic Isolation History, Application and Performances - A World View, Earthquake Spectra, Earthquake Engineering Research Centre, Volume 6, Number 2, May 1990. 5 Tajirian, F. F., Kelly, J. M., and Aiken, D., Seismic Isolation for Advanced Nuclear Power Stations, Earthquake Spectra, Earthquake Engineering Research Centre, Volume 6, Number 2, May 1990. 6. Canadian National Standard, CSA CAN3-289.3-M81, Design Procedures for Seismic Qualification of CANDU Nuclear Power Plants, Canadian Standard Association. 7. Pall, A. S., Seismic Response of Friction Damped Braced Frames, ASCE Journal of Structural Division, Volume 108, St. 9, June 1982. 8. Pall, A. S., Verganelakis, V. and Marsh C, Friction Dampers for Seismic Control of Concordia University Building, Proceedings of the Fifth Conference on Earthquake Engineering, July 1987, Ottawa, Canada. 9. Tajirian, F. F., Elastomeric Bearings For Three Dimensional Isolation, ASME PVP Conference, June 1990, PVP-200, Nashville, Tenessee. 10. Aziz, T. S., and Biswas, J. K., Spectrum Compatible Time-Histories, Third Canadian Conference on Earthquake Engineering, June 1979, Montreal, Canada.

gauuAiMttJbiswu/stitmic' iwa/isoiaiiojV REACTOR BUILDING.

HEAT TRANSPORT PUMP ANO MOTOR

REACTIVfTY MECHANISMS DECK CABLE GALLERY EL. 113.50 \\

CORRIDOR

EL. 107.50

CORRIDOR..

EL 100.00

PENETRATION BAN0

FIGURE 1 CANDU3 REACTOR BUILDING-SECTION

-252- 10000" •o «* •• 4* to

/ Damping Factor /C

I en

e.el >« ") to to to M>O

FIGURE 2 GROUND RESPONSE SPECTRA 1.27 m

• INS 25 LAYERS o OFELASTOMER ID

SHIM PLATES

FIGURE 3 PROPOSED ISOLATION BEARING UPPER SLAB

PEDESTAL

FIGURE 4 BEARING ARRANGEMENT AT THE BASE

-255- Containment Structure

Internal Structure

FIGURE 5 REACTOR BUILDING MODEL

256- 7 uol y

30 /

x 20 ISOLATED CASE1 UJ X ISOLATED CASE 2 7 - FfXED BASE 10

0.5 1.0 1.5 2.0 ACCELERATION (C)

FIGURE 6 RESPONSE ACCELERATION PLOT

-257- UO-i

30 li ..... , T^D! ATFfl ffiT 1 i ISOLATED CASE2 x 20-1 o .FfXED BASE X

10 •i i . i 50 100 150 200 DISPLACEMENT (mm)

FIGURE 7 RESPONSE DISPLACEMENT PLOT

-258- FIGURE 8 FLOOR RESPONSE SPECTRUM AT THE INTERNAL STRUCTURE

ACCELERATION(G)

t r 1 i 3 - I 1 \ 1 1 \ J j \ 2 - 1 v

r • • 01 v f - "• / f /

•-~>—

i i 1 1 1 i i 1 0.1 1 10 FREQUENCY(HZ)

ISOLATED CASE 1 ISOLATED CASE 2 FIXED BASE

itt.§ FIGURE 9 FLOOR RESPONSE SPECTRUM AT THE CONTAINMENT STRUCTURE

ACCELERATION(G) 8

t 6- t r ( i I \ 5- 1 4- o u • 3- I \ 2- s /"

\ •i * - v •« - ~ -« — •> ... y 1 - _ _ ' 0.1 1 10 FREQUENCY(HZ)

ISOLATED CASE 1 ISOLATED CASE 2 FIXED BASE

14*.• M IAEA Specialists' Meeting on Base Isolation Technology XA0055391

Mar. 18-20, 1992, San Jose, U.S.A.

3-D SEISMIC RESPONSE OF A BASE-ISOLATED FAST REACTOR

S.Kitamura, M.Morishita, and K.Iwata

Power Reactor and Nuclear Fuel Development Corporation, Japan

ABSTRACT

This paper describes a 3-D response analysis methodology development and its application to a base-isolated fast breeder reactor ( FBR ) plant. At first, studies on application of a base-isolation system to an FBR plant were performed to identify a range of appropriate characteristics of the system. A response analysis method was developed based on mathematical models for the restoring force characteristics of several types of the systems. A series of shaking table tests using a small scale model was carried out to verify the analysis method. A good agreement was seen between the test and analysis results in terms of the horizontal and vertical responses. Parametric studies were then made to assess the effects of various factors which might be influential to the seismic response of the system. Moreover, the method was applied to evaluate three-dimensional response of the base-isolated FBR.

-261- 1. Introduction

It is getting recognized that the application of a seismic isolation system to a nuclear plant seems promising to enhance its seismic safety. Especially for an FBR plant, where the components are designed of thin walled structures and hence relatively vulnerable to a seismic event, a reduction of seismic loads with use of a base-isolation system is effective. With these considerations as background, Power Reactor and Nuclear Fuel Development Corporation started its research and development program for seismic isolation technology in 1983. The purposes of the program are; - to examine feasibility of the isolation system to an actual design taking the severe seismic design requirements in Japan into account, and - to develop reliable evaluation methods for the seismic response of a base-isolated structure which can be used in design and safety assessment. The program is made up of three phases as follows; Phase-1 Parametric study to identify appropriate restoring force characteristics of isolation systems and to assess the effects of various factors, Phase-2 Analysis method development and its verification tests, Phase-3 Application of the method to a base-isolated FBR plant.

2. Application of base-isolation to an FBR plant

Fig.1 illustrates generic restoring force characteristics of an isolation system. The reduction of seismic response of the upper structure is achieved through lengthening the system period by rubber bearings and absorbing vibrational energy by various types of dampers. A trigger function against a strong wind or a small earthquake and a stopper function preventing an excessive displacement of the system may be added. There are three factors which determine the characteristics of the system as follows; f : basic isolation frequency which is related to the secondary

stiffness by the equation f = (1/2 ;r )V Ki g/W, where K2 is the secondary stiffness, W is the total weight of the upper structure, and g represents the acceleration due to gravity, a' ratio of the secondary stiffness to the first stiffness, &: ratio of the yielding force of the hysteretic damper to the total weight of the upper structure.

-262 In order to identify a range of appropriate restoring force characteristics of the system, a parametric survey was made under several conditions of soil properties and ground motions. The isolation system was modeled as a bi-linear type spring which will be described in the next section. Table 1 lists the response acceleration, the response displacement, and the shear force coefficient which is the ratio of the shear force acting on the top of the isolation system to the total weight of the upper structure. Through these analyses, following knowledge was obtained. The response acceleration and the shear force coefficient reduced well when the basic isolation frequency f is 0.5 to 1.0 Hz. As for a hysteretic damper, it is effective for reduction of the response displacement to select the ratio of second stiffness a of 0.1 to 0.5 and the ratio of yielding force $ of 0.1.

3. Analysis method

3.1 Modeling of the isolation system Four types of mathematical models for the restoring force characteristics of the isolation system were used to develop the seismic analysis method. These include the bi-linear and tri-linear expressions, the Ramberg-Qsgood type functional expression, and the equivalent linearization model, as illustrated in Fig.3. The bi-linear model is made up of a linear spring simulating the horizontal stiffness of the rubber bearings and an elastic-plastic spring for the restoring force characteristics of the hysteretic dampers. In the tri-linear model, to approach the characteristics of the model to that of the real system, tri-linear expression was used. For the Ramberg-Osgood model, the restoring force characteristics of the hysteretic damper is modeled more realistically by the following equation; <5 = a +b F*"1 where F, <5 denotes the restoring force and the displacement respectively and a , b, and c are constants. The constants are determined so that the initial stiffness at 6 = 0 and the tangent stiffness at 6 = U6 , , where 6 r is the yielding displacement, are coincident with desired ones. And, for the equivalent linear model, the restoring force characteristics of the hysteretic damper is approximated by one equivalent linear spring with an equivalent viscous damping. The equivalent stiffness of the spring is determined from the secant stiffness of the bi-linear model of which energy is equal to that of the maximum response displacement obtained from linear response analysis of the upper structure using the first stiffness. The equivalent

-263- damping ratio is calculated from the ratio of the dissipated energy of the bilinear model to the elastic energy of the equivalent stiffness as illustrated in Flg.3.

3.2 Verification tests A series of shaking table tests using a small scale model consisted of an upper structure and various base-isolation devices was performed with the purpose of verifying the above mentioned analysis method. As for the isolation system, a rubber bearing with a steel hysteretic damper, a lead rubber bearing, and a high damping rubber bearing were used. The upper structure was designed with a steel frame structure as shown in Fig.il. Its weight was about 1,600 kg. A primary natural frequency of the model was approximately 12 Hz in the case of no base-isolation. Before the verification tests, experimental studies on vibration characteristics of the the isolation system were done. The purposes of the tests were to investigate the basic response characteristics, such as natural frequencies, damping factors, and transfer functions, using the harmonic excitation and to understand the seismic response characteristics using the observed waves and artificial waves. In the seismic response tests, the effect of the intensity and duration of the input wave, and the influence of multi- axial input were assessed. For example, Fig.5 compares the maximum response acceleration and floor response spectrum between one-directional and three- directional excitation. The influence of orthogonal input on one direction is small and the vertical motion input hardly affects the horizontal response.

3.3 Simulation analysis One-directional simulation analyses were performed with a three-lumped-mass model illustrated in Fig.6 inputting the E-W component of the Hachinohe Harbor record in Tokachi-oki Earthquake 1968 shown in Fig.7. Values of the stiffness and damping of the upper structure were obtained from resonance tests. For the base-isolation system, four types of models which were described in the previous section were used. The constants which determined the restoring force characteristics of each model were obtained by a curve fitting method through static loading tests of the isolation devices. In order to Judge the validity of simulation analyses, comparisons of the time histories of acceleration and displacement and the floor response spectra are shown in Fig.8. From the figure, in the case of the steel damper, it can be noted that the analysis results using the bi-linear model tend to estimate the response a little larger. On the contrary, the results using the equivalent

-264- linear model tend to estimate a little smaller. The results using the tri- linear model and the Ramberg-Osgood model are in good accordance with the test results. These tendencies can be seen in the cases of the lead rubber bearing and the high-damping rubber bearing. In either case, the validity of the analysis method is verified through these studies.

Three-directional simulation analyses were also executed. Here, an equivalent linear model was used for modeling the restoring force characteristics of horizontal component of the isolation system. The vertical and rotational stiffness of the device were also considered. Fig.9 shows a comparison of the time history of acceleration, the orbit curve, which is a trace of two-directional motion of the upper plate of the bearing, the maximum acceleration distribution, and the floor response spectrum between the test and the corresponding simulation analysis. A good agreement was seen in terms of both the horizontal and vertical responses. It was found possible to simulate the horizontal and vertical responses by using an equivalent linear model.

4 Analysis of the actual FBR plant

4.1 Parametric studies on effects of various factors Parametric studies were made to assess the effects of various factors which might be influential to the seismic response of a base-isolated structure. A base-isolation system which had the restoring force characteristics based on the analysis mentioned in the second section was designed for a 1,000 MWe class loop-type FBR. Earthquake response analyses were performed with a multiple- lumped-mass model as shown in Fig.10. An artificial ground motion, based on a modified Osaki spectrum in which the long period components were enhanced, was chiefly used here. The response acceleration spectrum of the generated ground motion is shown in Fig.11. The leading results through these analyses were as follows. Fig. 12 shows the relation of the maximum input acceleration and the maximum response displacement to compare the effect of the soil properties. The fact that the difference of the response between the soft rock site ( Vs = 700 cm/sec ) and the hard one ( Vs = 1500 cm/sec ) is small suggests us a possibility of adopting a site independent standard seismic design of the base- isolated FBR plant. In order to grasp the effect of soil-structure interaction, three types of models were considered. That is the sway-rocking model, the lattice model, and

-265- the FEM model illustrated in Fig.13. Fig.11 compares the results of maximum response accelerations. In the case of the base-isolated structures, the effect of the interaction was insignificant, as opposed to an ordinary one. The effect of variability of the characteristics of the isolation system was investigated. Here, the variability of the stiffness of the bearings and the steel dampers was assumed ± 20 % and that of the yielding displacement was assumed ± 10 % as illustrated in Fig.15. Fig.16 shows the upper and lower bounds of the responses owing to the variability of the characteristics. The variability of the response is almost the same degree as that of the stiffness. Then a possibility of the torsional vibration due to the eccentricity of the center of the stiffness was examined. Fig. 17 compares the response displacement as a function of the eccentricity between the center of gravity and the edge of the isolation device. From these analyses, the effect of the variability of the characteristics of the system is so small that we can disregard it in the design analysis.

4.2 Evaluation of the vertical response of an FBR plant There are some structural parts in an FBR plant which are relatively sensitive to the vertical component of a seismic load. Conceptually, a horizontal base-isolated structure is designed so that there is no, or at least a small, amplification of the vertical earthquake component at the bearing. But in reality, a certain amount of amplification is unavoidable due to a finite axial stiffness of the bearing. Therefore it is valuable to estimate the vertical response of the isolated structure for component design purpose. From this point of view, the method was applied to evaluate a vertical response of a base-isolated plant, in addition to a horizontal response. The restoring force characteristics of the isolation system was supposed to be a bi-linear type and fixed at isolation frequency f = 1.0 Hz, yielding coefficient £ =0.1, and secondary stiffness ratio a = 0.1. A two- directional artificial wave ( N-S : 500 gal, U-D : 350 gal ) was used for a seismic input and shown in Fig. 18. Earthquake response analyses were performed with two kinds of soil properties and three kinds of vertical stiffness of the bearings as parameters. - An analytical model consisted of a multiple-lumped-mass model, an isolation system model, and soil springs as shown in Fig.10. The isolation system was modeled as an equivalent linear model. Values of the soil springs were estimated by Tajimi's method. The time history responses of acceleration and displacement at the support position of the reactor vessel and the floor response spectra in the horizontal

-266 and vertical direction were obtained through these analyses, among which an example is shown in Fig. 19. Table 3 lists the maximum response acceleration and displacement at the support position. From the table, it can be observed that the vertical response can be controled at a low level, when the vertical stiffness of the device is designed sufficiently high. Note here, however, a care should be taken for the relation between the peak of the response spectrum and the component frequency, so that excessive vertical response of the component do not occur. Moreover, from a point of view of component design, the soft rock site has the advantage of response reduction as far as the vertical response is concerned. Since the seismic isolation has its advantage on a hard rock site, we must assess the effect of soil properties more synthetically.

5. Conclusion

From studies on application of a base-isolation system to an FBR plant, a range of appropriate restoring force characteristics of the isolation devices was obtained. An analysis method was developed based on mathematical models for the restoring force characteristics of several types of isolation bearings. The method was verified by a series of shaking table tests. Parametric studies to assess the effects of various factors were done. The base-isolated plant was not so much influenced by soil properties and soil-structure interaction as the non isolated plant. The method was applied to evaluate the vertical response of a base-isolated FBR plant.

ACKNOWLEDGEMENT The shaking table tests and the simulation analysis described in the section 3 of this paper were carried out at Shimizu Corporation under a contract with PNC. The authors would like to thank Mr. Kobatake who was in charge of the tests and analyses.

REFERENCE [1] M.Morishita, M.Kobatake, K.Ohta, Y.Okada; Investigation on Base Isolation for Fast Breeder Reactor Building, lOth-SNiRT, Anaheim, U.S.A., (1989)

-267- Stopper function/ It 8 Huge earthquake /' Ks"

~/\ Second stiffness ; 2 Upper structure en First stiffness • Isolation bearing

Trigger function '< /i K—ttjKi

oT Displacement Fig. 2 A two-lumped-mass model Fig. 1 Generic restoring characteristics of an isolation system

Table 1 Results of Parametric Study

f a. 6 Max.Accel. Max. Disp. Shear force (Hz) (gal) (cm) coefficient

0.5 0.5 0.05 65 8.3 0.067 0.5 0.5 0.1 109 12.1 0.111 1.0 0.1 0.05 66 5.2 0.066 1.0 0.1 0.1 116 6.6 0.117 1.0 0.1 0.2 206 6.9 0.208 1.0 0.2 0.05 85 5.7 0.086 1.0 0.2 0.1 130 6.5 0.132 1.0 0.2 0.2 212 6.8 0.214 2.0 0.05 0.05 77 3.7 0.077 2.0 0.05 0.1 127 3.8 0.125 2.0 0.1 0.05 104 3.7 0.104 2.0 0.1 0.1 148 3.7 0.149

Artificial wave input, Input acceleration = 300 gal Hard rock site (Vs=1500 cm/sec)

-268- Displacement

LLd. •y; / /

(a) Bi-linear model (b) Tri-linear model

Equivalent danping Equivalent stiffness _ ¥77

0 1 6 = a+bF " heq =•

(c) Ramberg-Osgcod model (d) Equivalent linear model

Fig. 3 Models of the restoring force characteristics

2500

Unit: ittn 2 ton

o o

6 ton s

8 ton

STEEL DAMPER \ RUBBER BEARING

Fig. 4 A steel fraTie structure

-269- one-directional input three-directional input 3rd floor 5000 l h=l% * JUooo 1 2nd floor - 3 3000 u 2000 ft 1000 l 1st floor -

1 1 1 0 200 400 600 0.02 0.05 (0.1 0.2 0.5 1.0 2.0 Response acceleration (gal) Period (sec) (a) Maximum response acceleration (b) Floor response spectrum

Fig. 5 Comparison of the responses between one-directional and three-directional input

1200, , , , ,

1000

_ 800 Rocking Horizontal spring spring

600 Upper structure 400 Isolation system 200

Fig..6 A three-lumped-mass model Q.QS 0.1 0.2 0.5 1.0 2.0 5.0 Period (sec) Fig. 7 The Hachinohe Harbor record in Tokachi-oki Earthcruake

-270- Time (sec) Period (sec) 250.0 5000 Bi-linear model h=l% E»i-linear model 0.0 4000 1 -250.0 &3000 J « « 8 (a) Acceleration ,-. s.o E2000 Bi-linear model c.o 1000 a

... • —^-\* a -S.i 10 0.02 0.05 0.1 0.2 0.5 1.0 2.0 (b) Displacement (c) Floor response spectrum Tri-linear model 5000 ' ' 1 ' ' " 1 ' ' Tri-linea r model h=l% 0.0 4000

"•"0 2 4 6 6 10 £? 3000 (d) Acceleration S.O I Tri-linear model 2000 0.0 1000 a ft a -5.0 •v 2 < 6 0;02 0.05 0.1 0.2 0.5 1.0 2.0 (e) Displacement 25C.0 (f) Floor response spectrum R-0 nodel 5000 ... j R-0 h=l% o.o ft ft A-ft. ft A model /v y y v vvv\j\jyy~T\ 4000

-2S0.0 10 JJ3000 (g) Acceleration 5.0 R-0 model E 2000 cm )

CO A r, cL V *" 1000

D -5.0 2 < 6 6 10 0.00° 2 0.05 0.1 0.2 0.5 1.0 2.0 (h) Displacement 250.0 (i) Floor response spectrum 5000

o.o 4000

2 4 6 & 3000 (j) Acceleration _ s.o Equivalent model 2000-

0.0 ... A A 1000-

2 4 6 6 10 O.82 0.05 0.1 0.2 0.5 1.0 2.0 (k) Displacement (1) Floor response spectrum

Fig. 8 Comparisons between the test and analysis results test analysis

-271- Time (sec) ~ 250.0 3rd floor <3 2

J 0. 2nd floor 0 2 4 6 8 10 < -250, (a) Acceleration in the E-W direction

JH 250. a 200 ; 400 600 .2? "i .•/.• A A v. .1.-A /\ Rssponse acceleration (gal) ~ 0. c^ vy V "' V W (d) Maximun response acceleration 5000 <^ -250. 0 2 4 6 8 10 ^4000 (b) Acceleration in the N-S direction

•^sooo £ 250. I : 0. ^2000 o ?>1000

^ -250 t. 0 2 4 6 8 10 (c) Acceleration in the U-D direction 0.02 0.05 0.1 0.2 0.5 1.0 2.0 Period (sec) (e) Floor response spectrum

. . 3.

8 t u -2 '{'Displacement -3 Vf Displacement ? (cm) 1 (cm) E-W direction

-2 (f) Orbit curve (g) Restoring force characteristics

Fig. 9 Coroarisons between the test and analysis results test analvsis

-272 1200 1 1 1 1 ii(i

h=!3%

1000

Equivalent 800 <*> 3 I linear mode] en *\ I! 600 u \ 0) J/I 8 \ 400 J J -—- r \ 200 V

1 1 i i i i i i i t i i 0^02 0.05 0.1 0.2 0.5 1.0 2.0 Period (sec)

Fig. 10 A multiple-lumped-mass model with an isolation Fig. 11 An artificial ground motion system model 30.0 000.0

I f=0.5 Hz, Vs= 700cm/sec 4J 25.0 c o o E=0.5 Hz, Vs= 1500cm/sec U 600.0 A t, f=1.0 Hz, Vs= 700cm/sec u

I 8 0O6 0.0 0.0 100 200 300 400 500 600 0 1 2 3 4 5 The maximum input acceleration (gal) Case No. shown in Table 2 Fig.14 Effect of soil-structure interaction Fig.12 Effect of soil properties

Table I Restoring Force Characteristics

f 0 Damping Isolation Hz Factor % Devices urn 0 Non Isolation 1 0.5 0 .5 0 05 2 LB + IID — 2 0.5 0 .5 0 1 2 • 0.025 IB + HD + TR 3 1.0 0 .1 0 1 2 LB + HD (a) A Sway-Rocking ntxlel (b) A Lattice model (c) An FEM model 4 1.0 0 .1 0 1 0.025 2 LB + HD + TR 5 0.5 10 L3 + VD Fig.13 Analysis models Bt: Trigger yielding coefficient, LB: Laminated Bearing, IID: llysteretic Damper, VD: Viscous Damper, TR: Trigger stiffness ±20% stiffness ±20% yielding disp. ±10%

0 6 0 6 (a) Laminated rubber bearing (b) Steel dampler

46.0

I i (XI

Upper bound 1.00Q Target Lower bound

-31.0 0 100 200 300 0 2 4 6 8 (a) Acceleration(gal) (b) Displacement(cm) (c) Total characteristics Fig. 16 The variability of the response Fig. 15 Upper bound and lower bound of hysteretic characteristics 10 1500

8 2 1000 \ I 6 5 \

600 / \ V Table 3 Responses of anTOR plan t xJ J a! 400 Horizontal vertical No. £v Max.Accel. Max.Disp Max.Accel. Max.Disp 200 (m/s) (Hz) (gal) (an) (gal) (an)

1 1500 IB 146 10.39 656 0.26 0 I I 11 2 1500 12 149 9.C5 681 0.33 0.02 0.05 0.1 0.2 0.5 1.0 2.0 5.0 3 1500 6 140 10.50 979 0.97 Period (sec) 4 700 18 148 10.58 354 0.35 (b) U-D direction 5 700 12 148 10.59 379 0.42 Fig.18 A two-directional artificial wave 6 700 oI* 147 10.69 455 0.70

Vs: Shear wave velocity fv: Vortical isolation frequency -LLZ-

Accel, (gal) Accel, (gal) Disp. (cm) Diso. (cm)

en 13

Ul 3 o

o rn a Acceleration (gal) Acceleration (gal)

£* Ul O O o o <7 00 O O O o o O o o O O o o o oo o o o o o o b —. o — o o KX>C-<>o — 0 o la KXXXK3 — 3 is 0 ui IB n IB 8 : \

•—. tn ts 3 ? o - "D (DO CO (fl • • (0 ". O : / 0) "8 S'«. : t £M I nft • \ o 1 o 2 rr o - to / 2 b o - : • o / Ul y b XA0055392

SEISMIC ANALYSIS FOR THE ALMR

Frederick F. Tajirian Bechtel National, Inc. SO Beak Street San Francisco, California 94119 U.S.A.

ABSTRACT

The Advanced Liquid Metal Reactor (ALMR) design uses seismic isolation as a cost effective approach for simplifying seismic design of the reactor module, and for enhancing margins to handle beyond design basis earthquakes (BDBE). A comprehensive seismic analysis plan has been developed to confirm the adequacy of the design and to support regulatory licensing activities. In this plan state-of-the-art computer programs are used to evaluate the system response of the ALMR. Several factors that affect seismic response will be investigated. These include variability in the input earthquake mechanism, soil- structure interaction effects, and nonlinear response of the isolators.

This paper reviews the type of analyses that are planned, and discuses the approach that will be used for validating the specific features of computer programs that are required in the analysis of isolated structures. To date, different linear and nonlinear seismic analyses have been completed. The results of recently completed linear analyses have been summarized elsewhere. The findings of three-dimensional seismic nonlinear analyses are presented in this paper. These analyses were performed to evaluate the effect of changes of isolator horizontal stiffness with horizontal displacement on overall response, to develop an approach for representing BDBE events with return periods exceeding 10,000 years, and to assess margins in the design for BDBEs.

From the results of these analyses and bearing test data, it can be concluded that a properly designed and constructed seismic isolation system can accommodate displacements several times the design safe shutdown earthquake (SSE) for the ALMR.

1.0 INTRODUCTION

The ALMR program sponsored by the U.S. Department of Energy (DOE) uses seismic isolation to simplify design, to enhance safety margins, and to support the development of a standardized design for the majority of the available U.S. reactor sites. The ALMR [1], employs compact reactor modules sized to enable factory fabrication, and shipment to a wide range of sites. The isolated structure, consists of a stiff rectangular steel-concrete box structure 52 ft by 91 ft. It supports the reactor vessel, the containment dome and the reactor vessel auxiliary cooling system stacks, see Figure 1. The reactor vessel has a 20 ft diameter and a height of 62 ft and is supported from the top. The relatively large height to diameter ratio of the vessel provides sufficient intrinsic resistance in the vertical direction to minimize amplifications in the vertical ground motions making vertical isolation unnecessary.

-278- REACTOR VESSEL CONTAINMENT DOME AUXILIARY COOLING SYSTEM INLET AND EXHAUST STACKS

CONTAINMENT VESSEL ^SEISMIC ISOLATOR BEARING REACTOR VESSEL IN-VESSEL TRANSFER MACHINE INTERMEDIATE HEAT EXCHANGER (2) EM PUMP (4)

REACTOR SILO CORE - METAL FUEL

Figure 1 Isometric View of ALMR Plant

The isolated structure weighs 13,000 kips and is supported on 25 seismic isolation bearings. The isolators consist of steel-laminated elastomeric bearings using a high damping compound. The average load per bearing is 525 kips. The bearings were designed to give a horizontal frequency of 0.75 Hz (at SSE level displacements), and a vertical frequency greater than 20 Hz. Additional details on the design of the bearings are given in a companion paper to be presented at the meeting [15].

2.0 SEISMIC ANALYSIS PLAN

The seismic response of the ALMR in the horizontal directions is mainly controlled by the dynamic stiffness of the bearings at different horizontal displacements. At small displacements, the bearings are stiff and provide inherent wind resistance capacity. As the displacements increase, the stiffness is reduced and remains fairly constant up to bearing displacements greater than those expected for the SSE. At very large displacements, the bearing stiffness increases again, providing a mechanism for limiting displacements during extreme events. This increase in stiffness occurs due to recrystalization of the rubber molecules. Shake table test results performed in the past on different building models have shown that an equivalent linear approach is adequate for predicting the response of

-279- isolated structures, using this type of isolators for earthquakes resulting in displacements that are in the range when the bearing stiffness is relatively constant [7].

To date various equivalent linear time history analyses have been performed to assess the system response of the ALMR. The analyses were performed using the Bechtel structural finite element program BSAP [2]. The results of these analyses have been used for preliminary design of the bearings and the isolated structures and components.

More recently, nonlinear time history analysis was performed to model the nonlinear stiffness of the bearings and to compute the response in the displacement range when the bearing stiffness cannot be assumed to remain constant Several computer programs are available and could be used for performing this analysis. For two-dimensional analysis, DRAIN-2D [6], has been widely used by the structural design community. For three- dimensional analysis, general purpose programs such as ABAQUS, ANSYS, MARC can be used. Specialized programs have also been specifically developed for the nonlinear analysis of isolated structures. These include NPAD [18], SISEC [17], and 3D-BASIS [10]. Among the latter three programs, 3D-BASIS is the only one currently in the public domain. For the ALMR analysis, the computer program SeaStar[ 12] was used. This program was selected because of easy access, and because it is highly efficient on workstation platforms. Further details of this program and the analysis results are given in Section 3.

Future plans to support the development of a final design, include additional linear and nonlinear analyses to evaluate the effects of distant great earthquakes, low magnitude nearby earthquakes, different earthquake time histories, variations in bearing properties due to manufacturing and aging, foundation settlement and other effects. Additionally, three-dimensional soil-structure interaction (SSI) analysis using the computer program SASSI [8,13] will be performed. The objectives of the SSI analysis are to assess the effects of various seismic wave types including body waves and surface waves, embodiment effects, soil layering, SSI effects with emphasis on vertical, rocking and torsional response, and foundation-soil-foundation interaction effects.

Before final design, the computer programs selected for performing the system analysis will be validated by comparing the analysis results with the results of shake table tests to be performed on sub-scale models of the ALMR. Additional comparisons will be made with the actual response of isolated buildings during strong earthquakes, if such data becomes available.

3.0 NONLINEAR SEISMIC ANALYSES

The objective of the nonlinear analyses was to evaluate the effects of the high stiffness of the bearings at small and very large displacements on the response. Modeling these effects may be necessary for calculating the response for small earthquakes as well as very large events exceeding the SSE. Additionally, understanding the effect of the increasing stiffness at large displacements will quantifying the available margin in design beyond the SSE. Finally, findings of this analysis may lead to changes in the bearing design.

-280- 3.1 Seismic Criteria

The ALMR is designed to accommodate the seismic conditions expected at a wide range of sites, including deep soil sites with a minimum shear wave velocity of 1000 fps, to stiff rock sites. The seismic design basis is a safe shutdown earthquake (SSE) with a maximum horizontal and vertical acceleration of 0.3g specified at the surface of the free-field. The design earthquake is consistent with the U.S. NRC Regulatory Guide 1.60 spectra [16]. Figure 2 shows a comparison of the RG 1.60 design spectrum with the tentative Japanese design spectrum developed specifically for seismically isolated FBR [4]. It can be seen that for frequencies that are important to the response of isolated structures (0.5 to 1.0 Hz), the two curves are similar. Since the majority of potential ALMR sites are in the Central or Eastern United States, Figure 2 also shows response spectra more applicable for this region as recommended by the Electric Power Research Institute (EPRI) in [3]. Spectra for two sites are shown, a rock site and a deep soil site. This figure shows that the spectra have stronger high frequency components than the RG 1.60 spectrum which was originally developed from earthquakes recorded in the Western United States. Although this difference in spectral values may influence the response of conventional structures, the response of isolated structures will be smaller for the EPRI inputs. Thus the design spectrum selected for the ALMR is conservative and it can be concluded that the selected design spectrum is appropriate for all regions falling within the selected SSE acceleration. 18 q U.S. RG 1.60 JAPANESE SPECTRUM •*-«-• EPRI SOIL SITE **~*EPRI ROCK SITE o 5 ,-

i ^^^

1 I 10 FREOJENCY (HZ.) Figure 2 Comparison of Various Horizontal Design Response Spectra Scaled to SSE Level (5 Percent Damping)

-281- 3.2 Seismic Input Time Histories

The nonlinear analyses were performed for earthquake levels ranging from one-fourth the SSE (0.075g input) to more than six times the SSE (2.0g input). The time histories used for input levels of 0.3g and less consisted of two horizontal and a vertical synthetic acceleration time histories, whose response spectra enveloped the RG 1.60 spectra. These will be referred to as the RG input Each time history has a duration of 24 seconds and is digitized at 0.005 second intervals. The response spectra for these time histories are compared with the RG 1.60 spectra in Figure 3. These time histories were scaled to three levels; 0.3g, 0.15g, and 0.075g. These time histories were also used for events greater than the SSE, in which case the motions were scaled to 0.75g and 2.0g. For these levels two extra sets of time histories were used. These additional time histories were generated through numerical analysis for a hypothetical site in California. One set was developed assuming an earthquake with a moment magnitude of 8.0 and a distance of 12 km to the source. The resulting peak horizontal acceleration was 0.85g and peak vertical acceleration was l.Og. The total length of the record considered was 60 seconds, with about 25 seconds of strong shaking. The response spectra for these motions are shown in Figure 4 and are compared with the RG input spectra when scaled to 0.85 g. The second set was developed for an extreme event, where the moment magnitude was raised to 8.5. This represents rupture of the entire San Andreas fault This condition is very conservative, but nevertheless represents an upper bound event for California. The resulting time histories had a maximum horizontal acceleration of 2.0g and a maximum vertical acceleration of 1.2g. The duration of the record and of strong shaking was 60 seconds. The horizontal spectrum is shotvn in Figure 5. For comparison purposes the figure also shows the RG spectra scaled to 2.0g. Additional description of the numerical procedure and assumptions used in generating the synthetic time histories are given in [14]. The time histories and the acceleration scaling factors used are summarized in Table 1.

Table 1 Summary of Time Histories Used in the Analysis

NAME ABBREVIATION DURATION MAXIMUM HORIZONTAL MAXIMUM VERTICAL (SEC.) ACCELERATION (G) ACCELERATION (G) RG 1.60 Compatible RG 24 0.075,0.15,0.3.0.75.1.0,2.0 Same as Horizontal ^arge Synthetic Synth 60 0.85.1.7 1.0 Extreme Synthetic Extreme 60 2.0 1.2

33 Description of Computer Program

The computer program SeaStar [12] was used for the nonlinear analysis. This program is a general purpose three-dimensional program for linear and nonlinear structural analysis. It is based on the ANSR-III program the latest version of the ANSR series of computer programs [9,11] developed at the University of California, Berkeley. Input can include both imposed displacements and arbitrary static and dynamic forces. The element library includes 3-D large displacement beam element, 3-D inelastic beam element, 3-D nonlinear truss element, contact or support element, and cable element among others. The program has been extensively verified and is widely used by the offshore industry.

-282- 5* DAMPING 105B DAMPING S

• • • • i i I I I I I I i 10" i FREOJENCY (HZ.) (a) Horizontal Direction

10* DAMPING 5* DAMPING S I

FREQUENCY (HZ.) (b) Vertical Direction Figure 3 Comparison of Synthetic RG 1.60 Compatible Time History Spectra with Design Spectrum

-283- 18 I

RG 1.60 COMPATABLE SYNTHETIC 6 5

1 -

i I I I I I t\ ie* FREQUENCY (HZ.) Figure 4 Comparison of Response Spectra for Synthetic and RG 1.60 Compatible Time Histories (10 Percent Damping)

6 S

BDBE SYNTHETIC 1 RG 1.60 COMPATABLE 1 -

18"

FREOLENCY (HZ.) Figure 5 Comparison of Response Spectra for Extreme and RG 1.60 Compatible Time Histories (10 Percent Damping)

-284- 3.4 Description of Computer Model

The layout of the seismic bearings modeled in this analysis is shown in Figure 6. The center of gravity for the isolated structure is located 16 ft above the top of the bearings. In the lateral direction the load is symmetric around the long axis and is offset by 3 ft around the short axis.

The reactor deck was modeled as an assemblage of rigid beams as shown in Figure 7. This configuration was based on the lateral dimensions of the ALMR platform. Each node (11 to 25) represents the location of a seismic isolation bearing. Additionally node 40 was used to connect the platform to the reactor model and the additional lumped mass required to yield the vertical mass eccentricity defined above. In the lateral direction the lumped masses were distributed to give a minimum eccentricity of 5 percent of the platform lateral dimensions in both directions.

The isolators are modeled as individual springs to properly model rocking and torsional effects. In the axial direction, the springs are linear. In the two lateral directions, the springs are nonlinear, and have three stiffness segments as shown in Figure 8. Damping of the isolators is provided as hysteretic damping in the springs. Additional effective damping of about 1 percent is introduced as Rayleigh damping. A single lumped mass spring model was used to represent the reactor. The mass was representative of the reactor weight (1,992 kips), and the spring in the horizontal direction (3,320 kip/in.) was representative of the reactor vessel stiffness in the horizontal direction giving a horizontal frequency of 4 Hz. The vertical spring stiffness used was 38,200 kip/in., giving a vertical frequency of 13.7 Hz.

Previous linear analyses that included the soil flexibility had shown that the soil had very little effect on the horizontal response of the isolated structure [14]. Consequently for this analysis The foundation was assumed to be rigid (fixed-base analysis).

33 Cases Analyzed

A parametric study was performed to compute the response of the platform. The parameters that were modified include the input time history, the peak acceleration, the ratio of the third stiffness, K3, to the mid stiffness, K2, and the horizontal deflection 6S at which the bearing stiffness increases (see Figure 9). The values for Kj, K2, and 5y were kept constant for all cases and were Kj = 1620 kip/ft, K2 = 360 kip/ft, and 5y = 0.167 ft (2 in.). The parameters that were varied for the different cases are summarized in Table 2.

4.0 ANALYSIS RESULTS

4.1 Summary of Results

The first step of the analysis was to compute the fundamental frequencies of the model under the initial stiffness state of the isolators (Kj). The first two frequencies were 1.51

-285- Hz and 1.56 Hz corresponding to the two translational modes, the third frequency was 2.69 Hz corresponding to the torsional mode. The frequencies were also computed for the softened state, (K2), and the first three frequencies corresponding to the above modes were 0.72 Hz, 0.75 Hz, and 1.23 Hz respectively.

The maximum horizontal displacements across the bearings are summarized in Table 3. In this paper only the results of large earthquakes, larger than 0.75g are reported. Values for a typical inner and comer bearing are given. The vertical displacements were computed to evaluate the effects of rocking, and possible uplifting of the structure from the bearings. These values are summarized in Table 4 and represent the net vertical displacement (upward seismic displacement-vertical compression due to dead load. Negative net values indicate that the bearing is under compression.

The maximum forces in a corner bearing are summarized in Table 5. Additionally, the forces are normalized by the weight supported by each bearing to get an idea of the effective horizontal acceleration of the isolated platform these are also summarized in Table 5.

Figure 6 Layout of Seismic Isolation Bearings

-286- Z X

Figure 7 ALMR Analysis Model

Figure 8 Seismic Isolator Nonlinear Model

-287- 7000.0 y

6000.0 • gj. 5000.0 • g, 4000.0 • O 3000.0 • O U- 2000.0 •

0.0 0.5 1.0 DISPLACEMENT (FT.) Figure 9 Variations in Isolator Spring Properties

Table 2 Summary of Cases Analyzed

CASE No. FILENAME TIME HISTORY HORIZONTAL K3/K2 6s(in.) ACCELERATION (G) 1 REV3 Low Strain (&<&y) Eigen Value Analysis 2 REV4 High StTain (8y

-288- Table 3 Summary of Maximum Horizontal Displacements (Inches)

INNER BEARING CORNERBEA]RING CASE X-DISP. Y-DISP. MAX-DISP. X-DISP. Y-DISP. MAX-DISP. 3 10.3 12.3 12.7 11.5 13.0 16.6 4 13.3 15.1 16.7 15.4 17.5 21.6 5 33.2 34.8 42.3 37.8 37.9 46.0 6 33.2 36.0 43.2 40.2 39.2 48.9 7 30.7 33.9 39.8 33.6 35.5 40.7 8 35.0 38.4 48.2 44.1 46.3 62.5 9 31.7 35.9 40.0 35.8 37.5 43.7 10 27.5 32.7 42.4 31.3 35.4 47.1 11 25.3 30.0 39.1 29.7 32.8 44.2 12 8.2 13.3 13.4 10.6 13.6 14.0 13 25.2 28.6 28.7 29.2 32.7 33.8 14 23.0 29.6 37.3 27.8 32.4 42.4

Table 4 Summary of Maximum Vertical Displacements (Inches)

CASE MAX-VERTICAL DISPLACEMENT 3 -.035 4 -.026 5 .041 6 .035 7 .061 8 .019 9 .099 10 .126 11 .092 12 -.016 13 .069 14 .140

Table 5 Summary of Horizontal Force and Peak Effective Acceleration

CASE FORCE-X FORCE-Y EFFECTIVE EFFECTIVE (KIPS) (KIPS) ACCELERATION-X(g) ACCELERATION-Y(g) 3 389.4 455.4 0.74 0.87 4 491.1 570.9 0.93 1.1 5 1630.5 1862.5 3.1 3.5 6 1557.0 1668.0 3.0 3.2 7 1557.5 2501.3 3.0 4.8 8 1238.2 1397.9 2.4 2.7 9 1975.5 2418.0 3.8 4.6

-289- 4.2 Discussion of Results

The ALMR bearings were designed for an SSE of 0.3g. The shear strain expected in the bearing for this earthquake level is about SO percent Tests performed on the half-scale bearings [IS], indicate that the full-scale bolted bearing are capable of accommodating displacements around 37 in. (S times margin beyond SSE). This displacement corresponds to a shear strain of 2S0 percent and is equal to about 70 percent of the bearing diameter. Based on the test results and results of this analysis, it can be concluded that the current ALMR bearings are capable of accommodating earthquakes up to l.Og acceleration. However, for extreme events, when the input is 2.0g, the displacements exceed the bearing capacity by about 20 percent This is in general a conservative assessment for sites in the Eastern and Central U.S.. As was mentioned in Section 3.1, the response spectra for the time histories used in the analysis are significantly higher in the range of frequencies of interest, than response spectra being used for these regions in other advanced nuclear projects see Figure 2. If time histories compatible with the Eastern U.S. spectra are used the margins in the existing bearing designs would be even larger, and earthquakes with peak accelerations of 2.0g could be accommodated. Inprovements in manufacturing techniques could also result in bearings capable of accommodating large displacements. It is conceivable that bearings with the current ALMR geometry can be manufactured which can displace 45 in.

To examine the effect of the magnitude of bearing stiffness on limiting displacements, the results for four different K3/K2 ratios are plotted in Figure 10. The input for all for cases is the RG compatible scaled to l.Og. It can be seen that with no bearing stiffening, the maximum combined displacement is 62.S in. The displacement decreases as the stiffness ratio increases. Although the increase in stiffness is beneficial in limiting displacements this results in an increase in the forces, see Table 4. Another parameter which was varied was the lateral displacement 8S beyond which the bearing was assumed to stiffen. The K3/K2 was held constant at a value of 4.2. The analysis was performed with the extreme input and three value of 8S, 1.5 ft, 2.0 ft, and 2.5 ft The same analysis was performed with the RG 1.60 input scaled to 2.0g with 8S of 2.0 and 2.5 ft The maximum combined displacements for the various cases can be seen in Figure 11. It can be concluded from this figure that the displacements are reduced if the bearing design is such that they are allowed to stiffen starting from a lower displacement It can also be concluded from this figure that the two different time histories used result in similar responses.

Finally, it can be seen from Table 4, that when a 2.0g input is specified, uplift in some of the bearings may be expected. Tests on bearings have shown that larger tensile strains would be required to damage the bearings. The effect of tension in reducing the shear capacity of the ALMR bearings has not been considered to date. This type of loading condition has been considered in Japan and should be included in future testing of ALMR bearings.

-290- 2.6 4.2 10 RATIO OF K3/K2

Figure 10 Effect of Bearing Stiffening on Maximum Combined Displacement RG 1.60 Input Scaled to 2.0g

1.5 2 2.5 BEGINNING OF HIGH STIFFNESS (FT.)

Figure 11 Effect of Varying Displacement at which Bearing starts to stiffen

The next phase of this program will evaluate wheteher the current ALMR bearing design and seismic gap is adequate. The evaluation will use the results of this analysis, coupled with the findings of the ALMR half-scale bearing tests to evaluate the current seismic risk. Hazard curves for Eastern and Central U.S. will be used to determine the probability of the

-291- seismic gap from closing. If the probabilities are not sufficiently small, the size of the gap and the design of the bearings may have to be modified to accommodate the increased demand.

5.0 REFERENCES

1. Berglund, R. C, Gyorey, G. L., Tippets, F. E., "Progress in Safety and Performance Design of the U.S. Advanced Liquid Metal Reactor (ALMR)," ASME-AIEE Joint Power Generation Conference, Boston, MA, October, 1990. 2. BSAP - Bcchtcl Structural Analysis Program, CE800, Bechtel Power Corporation, 1976. 3. Electric Power Research Institute, "Probabilistic Seismic Hazard Evaluation at Nuclear Power Plant Sites in the Central and Eastern United States: Resolution of the Charleston Earthquake Issue, NP-6395-D, April, 1989. 4. Ishida, K., et al., 'Tentative Design Response Spectrum for Seismically Isolated FBR," Transactions of 10th SMIRT Conference, Vol. K2, Anaheim, California, August 1989. 5. Ishida, K., "Recent Research and Development Activity in Japan," Post-SMiRT 11 Conference on Seismic Isolation of Nuclear and Non-Nuclear Structures, Nara Japan, August, 1991. 6. Kannan, A. M., and Powell, G. H. (1975), "DRAIN-2D: a General Purpose Computer Program for Dynamic Analysis of Inelastic Plane Structures with Users Guide," Report No. UCBIEERC-73122, Earthquake Engineering Research Center, University of California, Berkeley, California, 1975. 7. Kelly, J. M., Griffith, M. C, and Aiken, I. D., "Earthquake Simulator Tests of a Neoprene Base Isolation System for Medium-Rise Structures," Proc. ASME Pressure Vessels and Piping Conference, American Society of Mechanical Engineers, Vol. 147, Pittsburgh, PA, 1988. 8. Lysmer, J., Tabatabaie, M., Tajirian, F., Vahdani, S., and Ostadan,F., "SASSI-A System for Analysis of Soil-Structure Interaction," Report No. UCB/GT/81-02, University of California, Berkeley, CA, April, 1981. 9. Mondkar, D. P. and Powell, G. H., "ANSR-II, Analysis of Nonlinear Structural Response, Users Manual," Report No. UCB/EERC-79/17, Earthquake Engineering Research Center, University of California, Berkeley, CA, July 1979. 10. Nagarajaiah, S., et al., "Nonlinear Dynamic Analysis of Three Dimensional Base Isolated Structures (3D-BASIS)," Report No. NCEER-91-0005, National Center for Earthquake Engineering Research, State University of New York, Buffalo, N.Y., 1991. 11. Oughourlian, C. V. and Powell, G. H., "ANSR-IH: General Computer Program for Nonlinear Structural Analysis," Report No. UCBIEERCS2I21, Earthquake Engineering Research Center, University of California, Berkeley, CA, November 1982. 12. SeaStar - Version P3.00, PMB Engineering, San Francisco, CA, January, 1990. 13. Tajirian, F. F., Longstreth, M., and Appleford, A., "Seismic Analysis of the Modular High Temperature Gas-Cooled Reactor," Proc. of Specialists'Meeting on Seismic Behavior of Gas Cooled Reactor Components, IAEA Report IWGGCR/22, France, November, 1989.

-292- 14. Tajirian, F. F., and Abrahamson, N. A.," Response of Seismic Isolated Structures During Extreme Events," Transactions of 11th SMiRT Conference, Vol. K2, Tokyo Japan, 1991. 15. Tajirian, F. F., Gluekler, E. L, Chen, P., and Kelly, J. M., "Qualification of High Damping Seismic Isolation Bearings for the ALMR," IAEA Specialists'Meeting Seismic Isolation Technology, San Jose, CA, March 1992. 16. U.S. Nuclear Regulatory Commission, "Design Response Spectra for Seismic Design of Nuclear Power Plants," Regulatory Guide 1.60, Rev. 1, Washington, D.C., December 1973. 17. Wang, C. Y. et al., "System Response Analyses of Base-Isolated Structures to Earthquake ground Motions," Transactions of 11th SMiRT Conference, Vol. K, Tokyo, Japan, 1991. 18. Way, D., and Jeng, V., "NPAD - a Computer Program for the Analysis of Base Isolated Structures," Proc. ASME Pressure Vessels and Piping Conference, American Society of Mechanical Engineers, 147,65-69,1988.

-293- IAEA SPECIALISTS1 MEETING ON SEISMIC ISOLATION TECHNOLOGY San Jose, California, USA, March 18-20, 1992

NUMERICAL ACTIVITIES ON SEISMIC ISOLATION IN ITALY F. Bettinali (ENEL-CRIS, Milano, Italy) A. Martelli (ENEA-RIN, Bologna, Italy) G. Bonacina (ISMES S.p.A., Bergamo, Italy) XA0055393 M. Olivieri (ANSALDO-RICERCHE, Genova, Italy)

ABSTRACT The numerical activities which are in progress in Italy in the framework of the seismic isolation studies mainly concern the definition of models for bearings and isolated structures, and their use for test design and the analysis of experimental results. Simple bearing models have been set up, and the development of finite-element (f.e.) three-dimensional (3D) and 2D axisymmetric models is in progress. Simple models have been based on the results of single bearing tests: models formed by a spring in parallel to a viscous damper, where both horizontal stiffness and viscous damping vary with displacements, have been developed by ENEA. Models based on hysteretic damping have also been developed by DISP and ISMES. Detailed bearing models include separate elements for the rubber and steel plates. A 3D model has been implemented by ENEA in the ABAQUS code. Linear elastic calculations have been performed with this model. The implementation of an elastic-plastic model for steel is also being completed, together with that of a hyperelastic model of the rubber, based on tests on specimens. Detailed models will be validated based on measured data. They will be used for bearing design and analysis of the effects of defects: some bearings with artificial defects have been fabricated to this purpose. As to the isolated structures, finite-difference programs were set up for the analysis of such structures in the case that they can be represented by sets of one-degree-of-freedom oscillators. The program ISOLA includes the aforementioned simple bearing model of ENEA, where both stiffness and damping depend on displacement and the effects of viscous creep are accounted for. A similar program has been based on the bearing model developed at ISMES. These models have been successfully used to analyse the experimental results concerning both isolated structure mock-ups and actual isolated buildings, based on the single bearing test data for both horizontal stiffness and damping (see a separate paper). Furthermore, f.e. models with constant viscous damping were implemented by ENEA in the ABAQUS program, and were used for the analysis of the previously tested isolated structures. For buildings, ABAQUS analysis concerns both models where a very large stiffness is assumed for the structure also, and sophisticated 3D f.e. models of the superstructure. The first were found sufficient to describe the overall motion of the SIP isolated buildings at Ancona. This paper describes the main features of the numerical models developed and reports some first comparisons between calculations and measurements.

-294- 1. INTRODUCTION Martelli & Bettinali [1] have explained that considerable efforts are being devoted in Italy to the development of seismic isolation and its application to both civil buildings and industrial constructions. Bonacina et al. [2] have shown that, in this framework, wide- ranging R&D work was undertaken in Italy by the National Agency for New Technologies, Energy and Ambient (ENEA), the National Utility (ENEL), ISMES, ALGA and ANSALDO-RICERCHE in 1989, taking advantage of collaborations established within the activities of the Working Group on Seismic Isolation (or GLIS, see Martelli & Bettinali [1]). The above-mentioned authors have also explained why the Italian R&D activities are focusing at present on the use of the high damping steel-laminated elastomer bearings (HDRBs), in particular those used in the five isolated buildings of the SIP • Administration Center at Ancona; finally, they have cited that this work consists of numerical studies, in addition to experiments on isolators, bearing materials and isolated structures. Numerical activities mainly concern the definition of detailed and simple models of bearings and isolated structures, as well as their use for test design and the analysis of experimental results. This paper describes the main features of such models and reports some first comparisons between calculations and measurements. The latter concern tests described by Bonacina et al. [2]. 2. BEARING MODELS Simple bearing models have been set up, and the development of finite-element (f.e.) three-dimensional (3D) and 2D axisymmetric models is in progress. 2.1 Simple bearing models Simple models of bearings are necessary for the analysis of isolated structures. Our models have been based on the results of the single bearing tests described by Bonacina et al. [2]. These tests have shown a highly non-linear behaviour of bearings. In particular, bearing horizontal stiffness varied considerably with displacement; furthermore, the energy dissipated in each cycle of the sinusoidal tests performed depended on the maximum deformation, but was almost independent of the rate of strain application. This result - which implies a mostly hysteretic damping nature - agrees well with those obtained by other researchers in the USA and Japan for similar types of bearings, although it is valid only in the frequency range which is characteristic of an isolated structure (a certain influence of strain rate was detected for rubber specimens, when wider frequency ranges were investigated). The above-mentioned experimental evidence shows that simple bearing models must first of all account for the variation of the horizontal stiffness with displacement. According to Martelli et al. [3], we obtained the horizontal stiffness at each displacement by evaluating the secant value of the final hysteresis loop of sinusoidal tests (i.e. after at least two conditioning cycles):

" FR(xmin^ /

-295- where FR is the restoring force and x and x. are the maximum and minimum values, respectively, of tne horizohrSI displacement in the hysteresis loop. Furthermore, the dependence of the energy dissipation on displacement should also be accounted for, at least to correctly analyse the measured data. To this aim, two different approaches are possible: (a) the use of equivalent viscous damping coefficients varying with displacement; (b) the use of an hysteretic damping model. Both kind of models have been developed in the framework of this study. 2.1.1 Equivalent viscous damping model. Models consistent with the first approach have been developed by ENEA, where such models were used to analyse the results of all tests performed by Bonacina et al. [2] on the isolated mock-ups and the SIP building (see Sect. 3). In these models, according to Martelli et al. [3], the equivalent viscous damping has been obtained, at each displacement, the equivalent viscous damping was calculated from the results of sinusoidal dynamic tests on single bearings as:

2 15,, = (Area of the hysteresis loop) / [0.5 n k. (xma -xm. ) ]. (2) 2.1.2 Hysteretic damping model. A model based on hysteretic damping, valid for displacements lower than 100% a (i.e. lower than the total rubber height, t ) was developed and used at ISMES by Serino et al. ([4], based again on the previously cited results of single bearing tests. A model based on a similar assumption had been developed by Sand and Di Pasquale [5]. More precisely, the model of Serino et al. [4] was derived as follows. As shown by Bonacina et al. [2], the tangent shear modulus of rubber, G, decreases rapidly with increasing strain r in the small deformation range, from the initial value G , then it tends asympto- tically to a value G^ for r increasing towards a* Thus, the two functions G,(r) and G (r), corresponding to the loading and unloading curves, respectively, can be expressed as:

G1(r) = GW + a exp [-b(r-rmin)] (3)

where G(0) = Gn = G^ + a. (5)

The parameters Gw, a and b of eqs. (3) to (5) must fixed based on the single bearing test data. It is noted that, at large shear strains, say at above 100% a, an increase of the shear modulus has been observed, due to crystallization under deformation of the natural rubber matrix. The formation of crystals is extremely rapid; they disappear as soon as strain reduces. This phenomenon has not been accounted for until now. The integration of eqs. (3) and (4) leads to the shear stress curves in the loading and unloading phases. From these, the corresponding correlations between restoring forces and horizontal displa-

-296- cement can be easily obtained considering that when a laminated elastomer bearing is displaced by the quantity x, an uniform state of pure shear is added in the rubber and the additional strain is equal to r = x/t :

F (X) F + + R1 = R

" «Ml-exP[-b(xinax-x)/tr]}/b (7) (A = cross section area). The above equations neglect the effects of bending deformation and those of the vertical compressive load, but they were applicable in our case, because the bearing shape factor was sufficiently large and the axial force was well below the critical value. When several bearings act in parallel, as it occurs for an isolated structure which moves horizontally without torsion, the total restoring force is obtained by multipying those of eqs. (6) and (7) by the number of isolators. From eqs. (6) and (7) it is finally easy to obtain the secant stiffness of an isolator subjected to hysteresis cycles as: kh(X) = GwA/tr + aA[l-exp(-2bx/tr)]/(2bX), (8) where X=(x__ -x_. )/2, and the equivalent viscous damping ratio as: iucLX mm aA//b {2X[l+exp(-2bX/{[p(/t ))]] - 2t /b[l-exp(-2bx/t/[p(/)]) } fi (X) = 0.5/71 - - - . (9) G AX2/t +0.5 aAX [l-exp(-2bX/t

The values of GR, a and b which provided the best approximation of the test data concerning SIP bearings tested by Bonacina et al. [2] were Gw = 0.631 MPa, a = 1.61 MPa and b = 6.3. Fig. 1 shows the comparison between the hysteresis cycles measured for the 1/4 scale bearings of Bonacina et al. [2] at 50% shear strain and values computed through the above mentioned mathematical model. The first loading analytical curve has been obtained through a similarity transformation of ratio 0.5, thus following Masing's rule: the agreement is excellent. The only appreciable difference is that the model slightly underestimates the bearing tangent stiffness at the very beginning of a loading or unloading curves; consequently, the energy dissipation capacity of the model at small deformations is lower than that observed experimentally. This result is confirmed by the comparison between the analytical and the measured & (X) values: the agreement is good for cycles of strain amplitudes equal or larger than 20%, but B(X) decreases and tends to 0 when X tends to 0, while an almost linear increase of £ is observed experimentally by decreasing shear strain (Bonacina et al. [2]). The problem of having no energy dissipation at zero strain is common to all hysteretic models: it may be attenuated if a constitutive law more complicated than that given by eqs. (3)-(5) (thus, with more than three parameters) is assumed for the rubber. 2.2 Detailed bearing models Detailed models include separate elements for the rubber and steel

-297- Figure 1: Measured and calculated hysteresis dynamic ainutoidal test on « ngle bearing loops in a dynamic sinusoidal test of a sin gle bearing.

-20 -10 0 10 displacement [mm]

Mode n t. 25 Hi. Shea in X dr«ction Uode n 2.25 H?. Shear in Y dbectian Mods n J. 26 ft. Torsion

Mode n 4.70 ft. 2nd Shea in X *de»»;80ft.2nd'foion

Figure 2: 3D model of SIP-type bearings (solid elements only) modes computed for the unloaded bearing in the range [0-100 Hz].

-298- plates. Three element sub-layers have been defined for each rubber layer and one for each steel plate. The 3D model that is being considered at present is formed by solid elements (8 nodes) for the rubber, and shell elements (4 nodes) for steel plates; a previous model consisted of solid elements only, with 15 or 20 nodes. These models were implemented in the ABAQUS computer program. Linear elastic ABAQUS runs were performed, to evaluate stiffnesses and natural frequencies of bearings loaded or unloaded by the superstructure. Fig. 16 of Martelli & Bettinali [1] and Fig. 2 of this paper show the first modes of the unloaded bearing: the first two (25 Hz) correspond to shear in the two horizontal directions; the third (26 Hz) is the first twisting mode; those at 70 Hz and 80 Hz are the second shear and twisting modes, respectively. The implementation of an elastic-plastic model for steel is also being completed (so as to evaluate the effects of steel plate deformations), together with that of a hyperelastic model of the rubber, based on the tests on specimens mentioned by Bonacina et al. [2] (Figs. 3 and 4 show two examples of the excellent agreement between stress-strain curves measured in the tests on specimens with compression and tensile loads and the values obtained by approximation of the polynomial equations used in ABAQUS). Detailed models will be validated based on measured data. They will be used for bearing design and analysis of the effects of defects: two bearings with artificial defects have been fabricated to this purpose also (Bonacina et al. [2]). 3. MODELS OF ISOLATED STRUCTURES Simplified numerical models of isolated structures have been developed at ENEA and ISMES, based on the simple bearing models described in Sect. 2. They are being used to analyse the experimental results described by Bonacina et al. [2], concerning both the isolated rigid structure mock-ups in full- and 1/4 scale, and the isolated buildings that were subjected to in-situ tests. Analysis performed by use of other bearing models (elastic, elasto-plastic, improved version of the Davidenkov-Martin model) have been illustrated in a separate paper by Sand et al. [6]. Furthermore, detailed models of structures have also been developed by ENEA. 3.1 Simplified models of isolated structures 3.1.1 Models using viscous damping for bearings. A finite difference program (ISOLA) was set-up by ENEA for the analysis of an isolated structure in the case that this can be represented by a cluster of one-degree-of-freedom (1-dof) oscillators. This program solves the motion equation of such an oscillator in the case that both stiffness and viscous damping depend on displacement. It also accounts for viscous creep effects. This program is being used by ENEA used for the analysis of the above-mentioned experimental results concerning isolated structures, based on the single bearing test data reported by Bonacina et al. [2] for horizontal stiffness and equivalent viscous damping. Thus, it applies the simple bearing models described in Sect. 2.1.1. Furthermore, f.e. models were also implemented by ENEA in the ABAQUS program for the different isolated structures (by assuming a

-299- B.WJ

-8.88 -8.15 / -8.23

-8.38 -8.58 A-8.38 -8.25 -8.13 8.88

Figure 3: Approximation with ABAQUS (•) of the stress-strain curve measured (—) for rubber specimens with compression load.

O.3O

8.23

8.15- /

8.88

8.88 8.88 8.38 8.75 1.13

Figure 4: Approximation with ABAQUS (•) of the stress-strain curve measured (—) for rubber specimens with tensile load.

-300- very large stiffness for the SIP building also). ABAQUS runs allow for calculations in the case of multidirectional excitation also. However, while the dependence of bearing horizontal stiffness on displacement can be taken into account by the program, only constant viscous damping can be assumed until now. The consequence is that a good agreement between calculations and measurements must be expected - especially for snap-back tests - only by use of damping values that are considerably larger than those obtained in single bearing tests (Bonacina et al. [2]). Indeed, the numerical analysis of free-vibration test data concerning the isolated structures, performed with ABAQUS by use of the above-mentioned model, while showing that horizontal stiffnesses (kh) are consistent with those related to single bearing tests, if the dependence of k. on displacement is correctly taken into account (Figs. 5 and 6), led to a good agreement between calculations and measurements by use of a constant & value that was considerably larger than that measured for the single bearings. (It is noted that, in the calculations, the force-displacement curve has been translated from the origine along the abscissae axis, of a quantity equal to the residual displacement, so as to account for the non- zero center of the motion cycles that is due to creep and other phenomena - see Bonacina et al. [2]). Anyway, the & values used in the calculations agree well with those found experimentally for snap-back tests of the structures, being only slightly larger than the average measured data of Bonacina et al. [2]. The agreement with 13 values measured in the single bearing tests was much better by use of the ISOLA program, i.e. accounting for the dependence of both k, and & on displacement (Figs. 5 to 7). This result was also obtainea for the sinusoidal and seismic tests that had been performed by Bonacina et al. [2] on the 1/4 scale mock-up (Fig. 8). It is noted that such an agreement is good not only for the mock-ups (which were actually formed by rigid masses), but for the building also. This means that single bearing tests are adequate to determine the parameters needed for the analysis of isolated structures, provided that their dependence on displacement is taken into account. For the SIP building, this also means that rigid body motions can be evaluated by use of a very simple model of the superstructure: this model might be applied to evaluate the motion at the superstructure base, thus limiting the use of detailed models of the superstructure itself to fixed base analysis. 3.1.2 Models using hysteretic damping for bearings. A computer program was written by Serino et al. [4] to calculate the non-linear response of a 1-dof system with a hysteretic restoring force given by eqs. (6) and (7), subjected to free or forced excitation. Compu- tation of the response is performed by step-by-step integration of the motion equation. The numerical scheme adopted is the linear acceleration method, where the system total acceleration is assumed to vary linearly, while the tangent stiffness is considered constant, during the time increment. When the system skips from a loading to an unloading curve, or vice-versa, large changes of the tangent stiffness occur, which may result in substantial equilibrium violations if a constant time in-

-301- Figure 5. Comparison between the horizontal displacement measured for the 9,500 kN mock- up in the most severe snap-back test (•) and values computed with ABAQUS (•) and ISOLA

-.1

Figure 6. Comparison between the horizontal displacement measured for the SIP building in the most severe snap- back test (•) and values computed with ABAQUS (•) and ISOLA (—).

Figure 7. Comparison between the horizontal displacement measured for the 394 kN mock-up in the most severe snap-back test (•) and values computed with ISOLA (—).

-.84

-302- crement is used. To avoid this problem, the program automatically reduces the integration time step when the velocity sign changes, so as to achieve the desired level of accuracy in satisfying the dynamic equilibrium equation. Figs. 9 to 11 show some comparisons between the dynamic responses measured by Bonacina et al. [2] for the 1/4 scale mock-up of the SIP buildings and those computed by Serino et al. [4]. Fig. 9 which refers to the snap-back tests at 100% a (36 mm initial displacement) - shows that the model is able to predict the change in stiffness by decreasing displacement, as well as the level of energy dissipation, except in the final portion of the response curve, because of the already mentioned insufficient damping capacity of the model at very low levels of deformation. Furthermore, the amplitude of the first peak is overestimated in the numerical analysis, probably because the residual displacement of bearings should not be taken constant during the entire test duration. (Because of creep of the elastomer, it is reasonable to suppose that the first residual displacement is slightly larger in the very first instants after the mass release: this would generate a restoring force on the mock-up smaller than that computed and thus, it would lead to a first ex- cursion with smaller amplitude). The computed and measured hysteresis loops which represent the total inertia force as a function of displacement during the previous snap-back test are shown by Fig. 10. The calculated cycles may be almost perfectly suporposed to those observed experimentally. Similar results were obtained for the other snap-back tests. Fig. 11 - where the computed and measured acceleration time- histories of the 1/4 scale mock-up are compared for the ID Tolmezzo WE record at 0 db (see Bonacina et al. [2]) - shows the adequacy of the model for forced excitations also. Comparisons of other response parameters demonstrated that the model is able to accurately predict the kinetic energy, energy absorption and total energy input, as well as the displacement relative to the shake table; only the peak displacement values were slightly overestimated. The extension of the work of Serino et al. [4] to implement the aforesaid hysteretic damping model in f.e. codes such as ABAQUS is being considered. Furthermore, new developments of the approach of Sand and Di Pasquale [5] are also in progress: the model defined by them has been made more stable and bearing hardening has been introduced. 3.2 Detailed models of isolated structures In addition to the aforesaid simple models, a sophisticated 3D f.e. model was also developed for the SIP building (Fig. 12). This work was performed by ENEA in collaboration with the designer (Giuliani [7]) - within the cooperative activities of GLIS - to allow for pre- test analysis of the building, as necessary to get the permission for the testing campaign (Bonacina et al. [8]). As required, pre-test calculations were performed by use of the SYSTUS program, which had already been adopted by Giuliani [7] for the design. The building model of Fig. 12 was derived from that of the (somewhat different and less complicated) building analysed in the design. It is characterized by 1820 (mostly isoparametric) elements and 2740 nodes. Pre-test calculations - performed on the VAX 11/751 computer of

-303- Figure 8. Comparison between the horizontal displacement measured for the 394 kN mock-up in the most severe ID seismic test for Calitri earthquake (o) and values computed with ISOLA (—).

40 394 kN ISOLATED MOCK-UP Figure 9: Comparison •nap-back test from 36 mm between the displacement time-history measured for the 394 kN isolated structure mock-up in the 36 mm snap-back test and values computed with the hysteretic damping model of Serino et al. [4]. experimental -10- model

-20 2 time(s]

40 40 394 kl I ISOLATED MOCK-UP 394 kl I ISOLATED MOCK-UP tnap back test from 36 mm 30- snap back test from 36 mm 30-

20 - _ 20

10- "a 10 -

-10 model -20-

-30 -30 -20 0 20 40 -20 0 20 40 displacement [mm] displacement [mm]

Figure 10: Comparison between the hysteresis loops measured for the 394 kN isolated structure mock-up in the 36 mm snap-back test and values computed with the hysteretic damping model of Serino et al. [4].

-304- ENEA - provided results that were later found in a reasonable agreement with measurements, in spite of considerable uncertainties that still affected the data and the use of constant horizontal stiffness and damping, which were necessary to apply methods limiting computer time to still acceptable values (Bonacina et al. [8]). In particular, the experimental response frequencies were reasonably well forecasted for the first deformation (bending and shear) mode of the superstructure (5.5 Hz, against measured values of 5.4 Hz - 5.7 Hz), together with the elevation where the deformation of the building is minimum (fourth floor) and longitudinal building displacements induced by transverse translation. In order to allow for a faster, more accurate, detailed analysis of the data measured for the SIP building, the model of Fig. 12 was recently implemented in ABAQUS, so as to enable calculations on the IBM computer of ENEA. These calculations have already been started by use of direct integration; a refinement of the mesh to make it more homogeneous is in progress, so as to allow for accurate modal analysis also, which is considerably faster (the mesh of Fig. 12, which is very detailed in the stairs section of the building only, is not adequate for ABAQUS calculations of modes where bending is important). 3D analysis of the test results of both houses at Squillace is also being started by ENEA within the collaborative activities of GLIS. 4. CONCLUSIONS This paper has provided an overview on the numerical activities that are in progress in Italy in support of seismic isolation development and application. It has described the main features of detailed and simplified models of bearings and isolated structures. It has shown through comparisons between calculated and measured dynamic responses of isolated structures - that single bearing tests are sufficient to determine the data (stiffness and damping) that are necessary for a correct analysis of such structures. However, the dependence of these data on displacement - at least that of stiffness - shall be taken into account in the analysis, in order to obtain accurate results. Finally, analysis of the isolated SIP building showed that rigid body modes can be well calculated by use of one-degree-fredom models of the superstructure, provided that its first bending fraquency is sufficiently larger than the values corresponding to rigid body modes (as it is for SIP buildings). REFERENCES [1] A. Martelli and F. Bettinali, Status Report on activities on seismic isolation in Italy, Paper presented to this meeting. [2] G. Bonacina, F. Bettinali, A. Martelli and M. Olivieri, Experiments on seismic isolation in Italy, Paper presented to this meeting. [3] A. Martelli, P. Masoni, G. Di Pasquale, V. Lucarelli, T. Sano, G. Bonacina, E.L. Gluekler and F.F. Tajirian, Proposal for guidelines for seismically isolated nuclear power plants - Hori- zontal isolation systems using high damping steel-laminated elastomer bearings, Energia Nucleare, 1 (1990) 67-95.

-305- [4] G.Serino, G. Bonacina and B. Spadoni, Implications of shaking table tests in the analysis and design of base isolated structures. Paper to be presented to the 10WCEE, Madrid, Spain (1992). [5] T. Sano and G. Di Pasquale, A constitutive model for high damping rubber bearing, in: Proceedings of the 9th ASME-PVP Conference, San Diego, CA, USA, Seismic, shock and Vibration Isolation, PVP-Vol. 222 (H.H. Chung, 1991) pp. 37-43. [6] T. Sano, G. Di Pasquale and E. Vocaturo, Linear analysis for base isolated structures, Paper presented to this meeting. [7] G.C. Giuliani, Design experience on seismically isolated buildings, in: Proc. First Int. Post-SMiRT Conference Seminar on Seismic Base Isolation of Nuclear Power Facilities, san Francisco, CA, USA, ANL Report, Argonne National Laboratory, CONB-8908221 (1989) pp. 220-245; Nucl. Engrg. Des. 127 (3) (1991) 349-366. [8] G. Bonacina, G.C. Giuliani, A. Martelli and G. Pucci, Full scale tests on a base isolated building, Proc. IASS Symposium 91, Copenhagen, Denmark, Vol. Ill (1991), pp. 271-282.

0.5 394 khj ISOLATED MOCK-UP Figure 11: Comparison Tolmezzjlj WE seismic test at 0 dB between the displacement time-history measured for the 394 kN isolated structure mock-up in the Tolmezzo seismic test at 0 db and values computed with the hysteretic experimental damping model of Serino model et al. [4].

8 10 time [s]

Figure 12: First vibration mode computed for the superstructure of SIP building subjected to in-situ tests.

-306- XA0055394

Recent Results of Seismic Isolation Study in CRIEPI -Numerical Actfvxtks-

Hiroo SHIOJIRI, Dr.Eng, Katsuhiko ISKtDA Dr.Eng, Shuichi YABANA, Kazuta HIRATA, Abdko Research Laboratory, Central Research Institute of Electric Power Industry.

L Introduction Development of detailed numerical models of a bearing and the related isolation system is necessary for establishing the rational design of the bearing and the system. The developed numerical models should be validated regarding the physical parameters and the basic assumption by comparing the experimental results with the numerical ones. The numerical work being conducted in CRIEPI consists of the following items. (1) Simple modeling of the behavior of the bearings capable of approximating the tests on bearings, and the validation of the model for the bearing by comparing the numerical results adopting the models with the shaking table tests results. (2) Detailed three-dimensional modeling of single bearings with finite-element codes, and the experimental validation of the model (3)Simple and detailed three-dimensional modelings of isolation buildings, and, experimental validations.

The results of the numerical work obtained so far are described here.

2. Simple modeling of bearings (1) Modeling The nonlinear diaracteristics of bearings were determined from an engineering point of view so as to meet the values of both equivalent stiffness(Keq) and equivalent viscous damping ratio(Heq) which were calculated based on the results of the preliminary element testa Fig.l shows how to determine the bilinear characteristics from the restoring loops obtained in the preliminary bearing tests using this engineering method. Bilinear models which give both Keq and Heq as obtained from the preliminary element tests results must meet the following conditions. The primary curve of the bilinear hysteresis loop must pass point C shown in Fig. 2-1, which corresponds to a specified target displacement 5 d, so that this bilinear primary curve gives a target Keq. The points on the primary curve of the bilinear hysteresis loop must correspond to a set of yielding shear displacement 6 y and yielding shear Qy in order for the area

-307- enclosed by the bilinear loop A W to be the same as that obtained from the bearing test results, so that this bilinear hysteresis loop gives a target Heq. In these relationships, numerous of yielding horizontal displacement 8 y and yielding shear force Qy which give the specified value of A W could exist on aline A-B which is parallel to line 0-C as shown in Fig.2-L To select one particular set of 6 yand Qy.a loop fitting method was used in this study. In tills method, the sum of square of differences between the assumed bilinear loop and the loop obtained in the test circular marks in Fig2-3) was naimiwmri at yJJTtfd values of 6 y in a range of 0 to 6 d and the one which gives the least value for the sum of square of differences was selected for 6 y in the bilinear primary curve. In this study ,the displacement s d was determined to be the maximum displacement of the bearing recorded in the shaking table test. Four different types of nonlinear model were used in the simulation analyses (see Fig. 2.2) Bilinear model: the model properties can be determined by the method described above. This model gives the required Heq solely by hysteretic damping, therefore, this model has no damping for the displacement less than 5 y. ) Bilinear + linear model; LRB consists of laminated rubber bearing and lead plug. To consider the internal viscous damping of laminated rubber bearing in the analysis model. This model consists of the linear visco-elastic model with the equivalent Keq and Heq of laminated rubber bearing alone and the bilinear model which represents the nonlinear characteristics of lead plug. The bilinear model were calculated by subtracting linear stiffness and equivalent hysteresis area corresponding to elastmeric rubber bearing from the bilinear model defined in ®. > Overlay model: The bilinear model described in ® gives the target Keq and Heq for a specified displacement amplitude. The Overlay model was obtained by combining several bilinear models so that it gives target A VTs and Keq's for several different displacement amplitudes. As an example of overlay model.Keq and Heq calculated from the bearing test results of LRB, and the nonlinear characteristics of a set of shear springs whose combination produces a overlay model were shown in Table 2-1 and Table2-2, respectively. Furthermore, Fig, 2-3 shows the hysteresis loops calculated by the overlay model of HRB subjected to gradually increasing displacement steps. Nonlinear Elastic + Damper modeLIt is known that Heq of HRB has dependence on displacement amplitude. In this model, the nonlinear elastic characteristics which give Keq for several different displacement amplitudes were combined with the viscous damper which gives Heq for a specified displacement amplitude 6 d.

(2) Validation by Shaking Table Tests

-308- The outline of the analytical model for one directional excitation is shown in Fig. 2.2. The steel frame of superstructure and the isolation layer( bearings) were modeled by shear-flexural beam elements, and a nonlinear shear with a linear rocking spring, respectively. The isolation layer of the test specimen was mnynrnd of a set of nine bearing. The modeling of bearings are discussed above. The isolation layer was modeled by a single nonlinear shear spring and a single linear rocking spring that represent characteristics of 9 bearings. Three types of modeling method were applied for each type erf bearings which was tested on the shaking table The maximum displacements of bearings recorded in the shaking table tests and the target displacements 6 d, as used for modeling were shown in Table 2-3. It should be noted that in the analysis cases of A-3 or C-3 a overlay model was used for different input earthquake levels. Fig. 2-5 shows the time histories recorded in the tests and analytical results for comparison. The analytical results show a good agreement with the test results in both response acceleration and displacement waveforms. However, in the test results, the acceleration responses at RF and IF levels contain some high frequency components. The maximum acceleration distribution and the floor response spectra at 2F level(ccrrespanding to the reactor location in an actual plant) of the model with LRB subjected to SI level earthquake input are shown in Fig.2-6 and Fig.2-7, respectively. The results by the analytical model which gives the specified Keq and Heq at the amplitude around the maximum displacement recorded in the shaking table test show a good agreement with the test results. As for the floor response spectra, the maximum peak values and the overall shapes of peak obtained from the analyses are in a good agreement with the test results in the period range from 0.1 sec to 1.0. sec where the first mode of the isolation system becomes dominant. The maximum acceleration distributions and the floor response spectra for the model with HRB under SI level earthquake input are shown in Fig. 2-8 and Fig. 2-9, respectively. As for the maximum acceleration distribution, the results of the analysis case B-2 show the best agreement with the test results. In the comparison of floor response spectra, the analytical results of all three analysis cases show a good agreement with the test results in the range from O.lsec to LOsec The hysteresis loop of LRB and HRB isolation layers under SI level earthquake input are shown in Fig.2-10 and Fig.2-11, respectively Although there is a tendency that the maximum displacements of analyses are slightly smaller than those of test results, agreements between the analytical results and the test results are regarded to be satisfactory. The maximum acceleration distribution and the floor response spectra of the model with LRB subjected to L5S1 level earthquake input are shown in Fig.2-12 and Fig.2-13, respectively. As input earthquake level increases, differences among the analytical results with different modeling methods become noticeable, in particular, in the floor

-309- response spectra The results of the analytical case C-3 which used an overlay model are in good agreements with the test results for both the maximum displacement distribution and the floor response spectra The hysteresis loops of LRB isolation layer subjected to L5S1 level earthquake input are shown in Pig. 2-14. Tho mff*i>nBn Aiapi**mmt flf the analytical case C-l which used a bilinear model is smaller than the corresponding test value The uni'directional nonlinear models were extended to the two-dimensional horizontal nonlinear models by using Multiple Shear Spring (NSS) method of the test model with LRB for multi-directional inputs of SI level earthquake were performed. The concept of NSS is shown in Kg. 2-15. The nonlinear properties of each direction used in this model is the same as those of the analytical case A-2. Fig. 2-16 shows the relative horizontal displacement orbit of bearings. The analytical results properly simulated the behavior of bearings in the horizontal plane as observed in the shaking table test.

(5) Summary of results The analytical results are in a good agreement with the test results in case of SI level earthquake input for the first mode response, in which the deformation of the isolation layer was dominant Therefore, the adequacy of modeling methods for nonlinear properties was confirmed. An overlay model can adequately simulate the test results of different input earthquake levels, Le., SI and 1.5S1.

height(ea) B EF RF . 653.9 Beaa Eleaent. Result of Element Test _....;££ N L 3F 442.4 3F < Qy

2F 2F 259.4

IF laolator IF Shaking Table layer Atsuae Several «T in Kance WMMSsS// Error of 0<<5r<<5d and Select 6r Fig.2-4 Analytical Modeling with Mioiaua false of SError* Fig.2-1 Loop fitting Method

-l.pl -3.8-I.t -1.0 0.0 1-0 20

« Wlta-r • U-r (O0».rl« M Wta-r ll-tl. Fig.2-2 Types of Honlinear Properties used in Analyses -310- Table.2-1 Eleaent Test Results »ei20HTAL HEIZOXTAL Keq STU1.1 DISPL (ton/ca) (X) 15 O.J75 0.«21 14. SI 50 0.750 0.S4] 1J.13 t.soo 0.4M 11. H ISO i.iii 0. W 10.13

TabLe.2-2 Nonlinear Properties of Overlay Model Component rtlFFSUJ (OK-UKA* U-UK&U Bt-USEAi BKIKtU IH.IXW8 TOTAL CU«IK IUSTIC (1) (2) (J) (4) JTITOBS (tan/ei) (ton/c«> (to»/c«) 11.090 1* HZ 0.471 0.011 15. M0 •1.0 o.osr tjoi 0.000 t t t 4.14) Q.ilS j. m t 0.000 t t 4. MS 0.1H tut 1 t 0.000 f tKi 1.100 1.017 t 1 t 0.000 Fig.2-3 Exaiple of Overlay Model "tit ltt jtiffRtu ch»|iiu point «>s caUulttad br tr»lef«ii loop torlzonca! ttr*U of lit. Table.2-3 Analytical Cases TEST IJOUTOE ftfCT HUftf* AiMimCAt. ntt Of «>lAirSIS T CSEO CASE OIJFL CASE K»a TO 0CTEMIS KOEL urn (TE5T:aO i-\ (a)Bi-lisur ]0.7SO A us si rl.ll t-2 (b)Bl-llnear • linear A-3 (e)Overlar* 0.J7S. 0.7JO. 1.50. I. ?5 (a)BI-liMar rO.« B-1 B us st L0.|7 (d)nonlinear Elastic • 0a*«er B-J (t1Orer!»r 0 4S1. O.Hi. l.«. J. 90 C-1 (a)Bi-tlnear rl.J} 3 I.SO C IRB 1.5S1 C-J (b)BI-l(nur • Linear C-3 ti propertr •odel.

I •[••• A-RC-X

A-3C-X A-3C-X

A-ZC-X

A-1C-X A-1C-X

thlS-X

•W" D-1C-X }u

D-1N-X «•• •§.*

H"" AWALYStS Fig.2-5 faveforas of Test Results and Analytical Results(A-2)

-311- 1000 MOO i-u-i —TKTtt) :::; !!:::: — JH :i 1 :i_i '' :::::: : : : :::::: : ::::::{ : : j ::::!: ' .! • — A-3 "TTT1TIH i"( 'SiiniT t-IC-I 'til

I 0.02 0.10 Pit IK 4-K-J 1000 B -A-3 IHHi £1000 tiltii I~I 1" "TIT ;Ti;~ T~Tflfcijif !"TT 4-It-I- ] Siooo .ivi!ii i...«..«.. lit 111 IK 0 :V?; : : : 1 10 0.02 0.10 0. SO 1.00 t.oo Fig.2-6 MaxiBun Acceleration ••" '•'fcinoAS - s.oo JEIIDO(SEC) Profiles(LRB:Sl) Flg.2-7 Floor Response Spectra(LRB:Sl) 1000, 1000- •—TEST(B) — B-l -- B-2 'TV

0 10 ... O.SC 1.00 $.00 HSItO(SBC)

010 0. 50 J.00 i.OO Rg.2-8 Maxinutj Acceleration reiiDo(sec> rtiiDot Profiles(HRB:Sl) Fig.2-9 Floor Response Spectra(HRB.SU

-312- 5.0

o.e DEPOBHiTIOB(Ca) Fig.2-10 Sestoring Hysteresis Loop&RB ~ 1.0

O B-1 j L o.o w -M i oa -5.0 In -1.5 .. O.O . 1.5 *° -1.5 CO -t.5 0.0 1.1 DEFORHATIOK(CH) BEPOBBATIOI(CH) Fig.2-ll Kestoring Hysteresis Loop(HRB:Sl)

5.00

0 100 200 300 400 500 0 5.00 Fig2-12 Maximum Acceleration 0.02 0.10 0.50 1.00 6.00 0.02 0.10 0.50 1.00 Prof iles(LRB:l. SSI) Fig.2-13 Floor Response Spectra(LRB:1. 5S1)

-:. 5 -1.0 0.0 1.0 2.5 -2.5 -1.0 0.0 1.0 2.5 -I. J-1.0 0.0.1.0 >.5 -2.5 -1.0 0.0 1.0 :. 5 DEFORMATIONS) DEPOEMATION(CM) DEF0RHATI0H(C«) DEFOIMATION(CM) Fig2-14 Restoring Hysteresis Loop(LRB:l.SSl)

1. ST 1.0 Anai) sis

'•i. S 1 Nonlinear Shear Spring ' 0.0 • ii -0.5 wv -1.0 -IK -1.5 -li "7.J 7.5 -1.6-1.0-0.5 O.O 0.5 Fig.2-15 MSS Model Fig.2-16 Orbits of Isolator Layer

-313- 3. Three-Etimensknal Modeling of Single Bearing (1) Numerical Model The numerical method for the bearing must be able to handle geometrical and materia nonlinearities, since maximum strain of rubber can be as much as several hundred percent and elastic modulus are strain dependent. * General purpose finite element programs MARC and ABAQUShave such capabilities and were applied to the analysis of the bearings. For simplicity, rubber is tmimort to be isotropic hyperelastic and nearly incmpreasdble. Element type used in the analysis by MARC and ABAQUS are listed in Table 3-L Constitutive equations used were as follows. Model A (Mooney Rivlin Model)

Where, l.- u-AAA ii-i

W=energy density function, W" ener«y density 'unction' and M i-1 ,2.3)«the principal extension ratios, and Coefficients of Eq. (1) were determined to fit the shear test of rubber bearing. Model B f d. expCei(Ii-3)3 ... (2) Coefficients of Eq. (2) were determined to fit the bi-axial test of rubber material (Fig. 3-1)

(2) Simulation of Laminated Elastomer Bearing 1) Laminated Elastomer bearing and PE mesh. 3 The bearing which support 4.9 x 10 N(500tB)of weight were simulated At first, vertical load was applied, and tiien forced horizontal displacement was given to the upper flange. Considering the symmetry of the geometry of the bearing, loadings, and displacements, only a half of the bearing was modeled. Linear hexahedral elements were used for steel plate and rubber. The number of meshes in circumferential direction and in radial direction were 8 and 2, respectively. Each rubber sheet or steel plate is divided into two meshes in thickness direction. The number of layers were 25. FEM mesh is shown in Fig.3-2.

2) Comparison of the Results of MARC and ABAQUS

-314- Model A was adopted for the constitutive equation of rubber. The results of MARC and ABAQUS were compared, and they show good agreement [Fig. 3-3].

3) Comparison of computed and experimental results The resultB of the ABAQUS code were compared with the experimental onea Rig.3.4 tttmn the computed results using Mooney-Rivlin model, together with experimental am the agreement is good only up to medium sheer strain. In Fig. 3.5, computed mulls using model (B) are compared with experimental ones. Numerical results were «Hnin^ up to 400% of nominal shear strain. (Horizontal displacement divided by total rubber thickness.) The agreement is fairly good even in large strain. The computed deformation are shown in Pig. 3-6.

4) The results of compressible material model The compressible model with constant bulk modulus were also used for the simulation of forced horizontal displacements of elastomer bearings. The bulk aodulus was calculated from the value of Mooney-Rivlin constants assuming poison's ratio Y =0.498. The constitutive model(B) were adopted for shear strain. Shear forces computed using compressible and incompressible models are shown in Fig. 3-7. The difference of the two model is small In Fig. 3-8, the computed vertical displacements using both models are shown together. Tableto Type of elements ABAQUS MARC Steel C3D8R 7 C3D8 Rubber C3D8H 84

Fig.3-1 Relationship between li(-I,) and 3W/31,, 3W/3I,

-311- 300

200

100

Fig.3-2 FEMmesh 0 200 400 600 tOO 1000 •»**• •II»l«

0 100 200 300 Shear Strain(X) Fig.3-4 Comparison rf computed results vising Mooney-Rivlin model and experimental results 80. "Computed Result Expaimentai Result JB 60 Experimental Result I, » 40 w / M 20

0 0 100 200 300 400 500 Shear Strain(X) Pig.3-5 Comparison of oomputed results using model (2) and experimental results

-316- (a) Horizontal Displaceroent(25cnO

(b) Horizontal Displacement(50cni) Fig. 3-6 Computed deformation

80 f 1 taoompriffiive Modei I [T.. .:"] Compressive Hcdd I e60

CO 20

0 100 200 300 400 500 Shear Strain(X) Fig.3-7 Comparison of compressive and incompressive models ( horizontal direction)

JL 5 4-> \ 4 Incompressiye^lodei " Ccimpressive Model \ 3 EjcperimentalResuit(r/ij'

- - •* -60C -40C -200 0 200 400 600 80-V0 lo'oI o Horizontal Displacement(mm) Fig.3-8 Comparison of compressive and incompressive models (vertical direction)

-317- The difference between the computed results are large, and neither results agrees with experimental ones. The vertical displacement obtained using compressible model decrease as horizontal displacement increases. It is qualitatively contrary to experimental results.

(4) Development of Special Purpose 3-D Finite Element Code. \ *L Special Purpose 3-D finite element code is developed for the incorporation of •ore general constitutive equations and interpolation functions, and remeshing of the excessively distorted elements.

1) Functional Whereas rubber can be considered to be nearly incompressible material, uumprwHtbiiity of the rubber must be taken into account to evaluate vertical stiffness of elastomer bearings. The bulk stiffness of rubber is also known to be bulk strain dependent The bulk modulus is due to the changing of interatomic spacing, where shear modulus of rubber is mainly attributable to entropy change It is, therefore, reasonable to assume that the energy density function has the following separated form.

Since the bulk modulus is by far larger than shear modulus, it is still considered to be adequate to adopt mixed formulation with displacements and pressure as independent variables. The functional for mixed formulation is expressed as follows. . • (4) fl=W,U, , i, )+fU)P+g(p) (4) where,

Ii*-Ii/Ii* and 1,*-12/I?"

g(p) ." complementary energy due to bulk strain 2) Constitutive equations Model B is adopted for shear strain, and the following constitutive equation can be assued for bulk strain.

Were, I is ultimate deformation, and C, and N are material constants.

3) Derivation of Finite Element Scheme Applying the variational principle on the above functional, the following weak form

-318- ' equation is obtained (6)

Where, t is 2nd Kola- Kirchhoff stress tensor, y is Green's strain tensor, p is density, F is internal forccand T is surface stress. Introducing adequate discrete interpolation function, the equations to be solved are obtained.

4) Miscellaneous Besides the elements described in table 3-1, it is added the 27 nodes isoparametric element with 3 components of displacement at each node and 4 degree of freedom of pressure for a element. Total Lagrangian formulation is considered to be adequate for hyperelastic large deformation analysis . Remeshing algorithm is being incorporated for the analysis of extremely large deformation

4. Earthquake Response Analysis of Base frolqt^ Building (1) Base Isolated Building The base isolated building, a object of analysis is a four-story reinforced concrete building. The floor area is 1,330 nf and the weight is 2,250ton. The seismic isolation devices consist of laminated rubber bearings and elasto-plastic steel dampers. The rubber bearing is considered to behave as a linear elastic spring up to the displacement of 200mm from the results of bearing tests conducted before installation. The yielding displacement of the steel elasto-pilastic damper is about 30mm. The periods of this building with the isolation devices are L4sec and 2.1sec corresponding to pre-yielding stiffness and post -yielding stiffness,respectively.

RF pi r 1 IMI I nil HJ ••-•• — - 4F 09 8 3F CO* 2F

1 F Y-dir. 6.500 | 7.400 | 6.500

i^i Lj-paj«,itA • 1 n ©: Laminated Rubber Bearing 3.000 6.500 7.400 6.500 L • :£lasto-Plastic Steel taper Fig.4-1 Section and plan of building

-319- 700 loo"

Rubber Sheet t7. OX 14 (=98.0) Steel Plate t3.2X13(=41.6)

Pig. 4-2 Laminated rubber bearing Pig. 4-3 Elasto-plastic steel damper

(2) Earthquake Response Analysis 1) Lumped Mass Model The building above isolation devices was modeled by five lumped mass system. Isolation devices were modeled by a pair of sway and rocking springs. Fig.4-4 shows the lumped mass model Modal analyses were used to evaluate the earthquake response of the building. The response of each mode was integrated by Newmark- Q method. Model dampings were estimated from both the forced vibration test and the earthquake response observation. (Table 4-1) Fig.4-6 shows the mode shapes obtained by the analysis. The comparison of the observed acceleration of first floor with calculated one was shown in Fig4-7 High frequency components corresponding to the second mode was excited in observed record,but excitation of high frequency components was much less in the lumped mqgB model As to the acceleration of the third floor, (See Fig 4-8) the calculated result agreed well witii the observed record. As to the relative displacement between the base mat and the first floor,(See Fig.4.8) the shape of calculated time history is quite mmfiar to that of observed record.

1 Table4.1 Analysis Condition of Luaped Mass Model

Mass or Meaber Veight Stiffness Mode D*«pint Ma (tonf) (tonf/a) fe RatioGO 5 408.9 1719 1 2.5 4 479.2 2718 2 0.3 3 424.0 3534 3 0.3 2 431.3 4515 4 0.3 502.5 71.66 5 0.3 1 1.34X10* 2.05X10" 6 0.3 (tonf-ca*) (tonf *ca) Fig.4-4 lutped Kass Model

-320- Table.4.2 Analysis Condition of FEM Model

Isolation Device Rubber Bearinc Steel Oaaiper A Horizontal Daapinc Vertical Oaapinc Horizontal Oaapinc Stiffness Ratio Stiffness Ratio Stiffness Ratio (tonf/ca) (*) (tonf/ca) (*> (tonf/ca) (f) 1.61 2.5 1304 1.0 2.61 2.S

Reinforced Concrete Modulus of Shearinc Modulus Poisson's Density Daepinc Elasticity of Elasticity Ratio Ratio (tef/.*) (l«f/e.») <«f/ca«) (« 6.3X101 2.7X10' 0.167 2.9 0.3 Fig.4-5 FEM Model

2) Finite Element Model Floors and main walls were modeled by shell elements, and pillars and girders by three-dimensional beam elements. Stiffness and damping of the upper structure was estimated from both design values and the results of the forced vibration tests. Steel dampers are modeled by horizontal shearing springs, and rubber bearings by horizontal and vertical springs. The damping matrix was assumed to be proportional to the stiffness matrix. Fig.4-5 shows the finite element model. Analysis condition was shown in Table 4-2. The results of finite element model is very similar to the observed records of the first floor. (See Fig.4-7)Especially excitation of the second mode is expressed well by the model* The acceleration of the third floor and the relative displacement between the first floor and the base mat calculated by this model are almost the same as those by the lumped-mass model, since these responses are dominated by the first mode.

f,=0.88Hz f,=7.56H2 f,=0.89Hi f, = 12.39Hx

X-direction i ) Lu>ped Mass Kodel Y-direction

{ / i 1 / ! i /i 1 ! 1 1 i i - ; / i I j j ! i i i 1 1 { / t 1 H f,=0.89Hi f,=10.80Hi f ,=0. 88Hz f ,=7.30Hr Y-directlon X-direction ii) FEM Model Fig.4-6 Results of Modal Analysis

-321- «*« 2.5«*IC AT 1.6$ 3.001 1IN-2.36U0 AT 7.05 iito.e 1 | n.t «** ?.l««IO AT •.T4MQ* ; SO.Q : H.O o.e XT 10.0 0-0 J.O 10.0 ll.o to •rmnuci 1) Okiarvad Uc*r4 IMOI MAX i.esxio AT 1.07X10 S.00 HIN-J.77X10 At I.O2XIO H 100.0 § H.I »T t.oo j: w.o

-3.00 0.0 IO.O :o.o 10.0 o.o ••• ••• in TW(SK) mo (SEC) a) CIIO I MAX 1.61X10 AT 7.42 3.00 KIN-J.67X10 it 1.09 HA* 7.26X10 AT «.7»xiO*< 0.00

-3.00 0.0 10.0 20.0 10.0 •0 5.0 MO TOW3 »m Mi »« R&4-7 Comparison of Observed Record with Calculated Results (Acceleration of First Floor)

MAX 1.17X10 AT t.ti 3.001 100.0 HIM-1.I6U0 At 9.1 3 M < t.e^Aio AT i .Til 10"

0.00 k 80.0 6 15.0 -3.00 Q..0 1 O.O 10.0 20.0 30.0 0.0 s.o IO.O it.o ie.o TIME(SEC) FREO(SEC) i) 0bttrv«4 laeerd

I MAX 1.2SH0 AT 8.59 3.00 M1N-1.J0X10 AT S.13 loo.o HAX 6.16*10 AT 6.79X10" 7S.0 • * 1 1 0.00 t«Hm>»n>» .. o! so aI r i "-1 -3.00 i.o, L 0.0 10.0 20.0 to.o 0.0 t.O 10.0 K.0 ».l TIME (SEC) nsxsEci R**i

MAX l.JIXlO AT ».J« 3.001 100. i MIH-I.JOM0 At I.OXIO MA/ 7.0741b AT ».74X10"

0.00 I ! C »o -3.00 0.0 lO.o to.e to.o g o.oe.o t.o to.o u.o n.o nBE(SEC) PFREO(SK) H) Coaparison of Observed Record vith Calculated Results (Acceleration of Third Floor)

-322- MAX 3.Si AT 4.1} SO. AT AT

0.0 S.O 10.0 15 0 20 0

Ucort

S.00 . MAX 3.90 AT 9.14 $ 30.a M1N-4.OI AT a.60 i «*« J.UXIO AT ».79X10" = 20. 10.00 ILLX 10.0

-5.00 0.0, 0.0 10.0 20.0 SO.O 0.0 5.0 10.0 15.0 20.0 TIMClSEC i *• fFEO.(rill fl) Uwfti Rats IM«1

HA< 4.11 AT I.OJXIOS «fi r AT a. 7*110" 1

-S.00 1IV 0.0 0.0 5.0 10.0 15.0 20.0 ffVl.'Hl <

Fig.4-9 Comparison of observed reoord with calculated results (Relative Displacement between Base Nat and First Floor) 3) Summary of results Earthquake response analyses by using both lumped mass model and finite element model were carried out The results were compared with the observed ones, and the fallowings are confirmed 0 lumped mass model is available to estimate the response of the upper structure and the deformation of the isolation devices, as the dynamic behavior of the base isolated building is dominated by the first mode @ Higher mode is simulated more accurately by finite element model, and the use of this model is recommended when the effect of high frequency mode is significant

-323- 5. Conclusions Accurate and objective numerical modeling of behaviour of bearings are possible to simulate the shaking table tests of isolation systems. In detailed three-dimensional analyses of single bearing, the identification of oomtitutive equations of rubber in bi-or tri-axial stress state is mandatary to predict the behavior of the bearing under large strain. Simple lumped mass models are sufficient for the wiim^wtinn of the earthquake response of existing base-isolated buildings. FEN model may be required to predict the floor response in high frequency range.

(Acknowledgements) Simulations of shaking table tests are a part of research project "Verification Test of Seismic Isolation for Fast Breeder Reactor" sponsored by Ministry of International Trade and Industry, and were conducted under the guidance of Prof H. SHTBATA, Prof T. Fujita and other members of the Advisory Committee.

(References) (1) ISHIDA,K,et,al:"Shaking Table Test on Base Isolated FBR Plant Mode^Part 2 Simulation AnalysisUlth SMiRT Vol,K,TOKYO(1991) (2) SHIOjmi,H,et,al: "Numerical Method for Analysis of Laminated Elastomer Bearings'1, 11th SMiRT VoLK TOKYO(1991) (3) MAZDA,T,et,aL"Earthquake Response Analysis of Base Isolated Building", 10th SMiRT Vol,K ANAHEIM(1989)

-324- XA0055395

LINEAR ANALYSIS FOR BASE ISOLATED STRUCTURES T. Sand^1), G. Di Pasquale^1^, E. Vocaturo^2) (1) ENEA DISP Rome (2) ENEA AMB/ING Rome

1. ABSTRACT Different constitutive laws are used to study the dynamic behaviour of a simple base-isolated mass supported by high- damping laminated elastomeric devices and seismically excited. The criteria of selection of the parameters of different laws on the basis of the ciclic shear tests carried on the isolation devices are illustrated. The resuslts of the numerical analyses are compared with those of shake-table tests. The agreement is similar for all the analyses showing that, in the considered deformation field, a simple linear analysis can give satisfactory and conservative results.

-325- LINEAR ANALYSIS FOR BASE ISOLATED STRUCTURES T. Sand^1), G. Di Pasquale^1), E. Vocaturo^2) (1) ENEA DISP Rome (2) ENEA AMB/ING Rome

2. INTRODUCTION The typical behaviour of high-damping laminated rubber bearings (HDLRB) used as isolation devices is clearly nonlinear as can be viewed from the experimental results shown in figure 1. Moreover the loading and unloading curves show different paths each other, so that the combined cycle has a characteristic ' sharp edged ' shape and its area , that is^ the dissipated energy (Ev) , is dependent on the maximum" displacement. The secant stiffness (Ksec), when is computed from cycle edge to edge , presents a similar dependence on deformation and is decreasing until it reaches an asymptotic value in the range of 50 - 120% of deformation (fig. 2). On larger deformations the behaviour of such devices is more complex because other phenomena as geometric nonlinearities and distortion of steel layers that produce an increasing secant stiffness and eventually degradation and ruptures. In this work only the 'working' behaviours of the isolation devices , that is only deformation until 100- 120%, are considered. The aim of the work is to assess if a linear equivalent analysis can reproduce , with good approximation, the non linear behaviour of HDLRB.

-326- Three different kind of analysis are performed: - linear - non linear with elastoplastic constitutive law, - non linear with an improved version of the modified Davidenkov-Martin model,/2/. The theoretical results have been compared with experimental data obtained during shake table tests performed at ISMES laboratories during an extensive R&D program founded by ENEA /I,6/.

3. MODELING OF ISOLATION DEVICES The main features of the material behaviour of isolators are the following: -The slope of force-displacement cycles is generally decreasing with the displacement. -The intersection of the tangent with the force axis is also growing with the displacement increasing. The experimental behaviour can be analytically reproduced by a non linear model derived by the Davidenkov-Martin one that has been used for soils (DMM : Davidenkov-Martin Modified)/3/ . The DMM model has been described in /2/ and others of the same type in /4/. That kind of model is able to match the experimental response in a large range of deformations ( from 20 % to 120% ). On the other hand it shows very low damping for low deformations (< 20%), which is contrary to the eperimental results. A new version of this model, able to reproduce also the hardening effect at large strains has been developed and is described in Appendix A. For better clarity the results obtained with this new version of the model are identified in the following by the 'DRTN'. The elasto-plastic model with kinematic strain hardening, is more easy to use and can satisfactorily reproduce the Ksec variation v.s the displacement. It can not reproduce very well the variation of damping on a wide range of deformations but it can be calibrated to give good approximations in a limited range (say from 30 to 60 % or from 60% to 100%). The linear elastic model has the advantage to allow the use of largely known numerical programs, and of linear analyses, which are more stable and easy to perform but it cannot reproduce neither the sharp-edged shape nor the hardening effec. In figure 3 the experimental and the theoretical cycles obtained with the above mentioned models are compared for three different maximum strains : 51 %, 78 % and 104 % .

-327- The experimental stress-strain curves have been obtained on small bearings 125 mm in diameter and 51 mm in total heigth (fig.l) used also to support the mass on the shake-table. It can be easily seen that in the range of deformation considered, which is the 'working range' for most of the applications of base isolation, all the models give good results but the DMM and the elastoplastic model are able to match better the experimental shapes.

4 EXPERIMENTAL ANALYSES Experimental tests have been held in february 1991 at ISMES /6/ on a 394 KN isolated mass subjected to both snap-back tests and forced excitation on an shake table.(fig.4,5) The mass has been supported by four 1/4 scale bearings 125 mm in diameter (fig.l). The Seismic Test have been based on real accelerograms recorded in Italy : - Tolmezzo (medium soil condition) recorded during the 1976 Friuli earthquake; - Calitri (soft soil) recorded during the 1980 Irpinia earthquake . The original accelerograms have been time-scaled in order to follow the the scale factor of the bearings (1/4) The Tolmezzo earthquake has been also scaled im magnitude from the recorded value of 0.32 g up to lg of maximum acceleration. The Calitri record has been used with its real maximum value amax=0.21g. In fig. 6 the time histories used and the relative response spectra are reported.

5 NUMERICAL ANALYSES 5.1 Calculation models Three analyses,using the above mentioned different" material model have been performed. The mass has been modeled as a plane 2 story frame. It has been isolated by means of devices located directly under the ground floor. The components EW of the same earthquakes used for the shake-table seismic tests have been considered. The parameters used to simulate the behaviour of the isolators are as follows: - Elastic model Shear modulus G = 8 Kg/sqcm , Damping ratio ft = 0.13 - Bilinear model

-328- Initial modulus GO = 23 Kg/sqcm, Hardening modulus Gl = 5.8 Kg/sqcm, yielding shear stress ty = 1.7 Kg/sqcm - DRTN model GO = 35 Kg/sqcm A = 0.8 B=0.32 p0=0.03 pi = 0.005 The meaning of the parameters is explained in appendix A. In fig. 3 the stress-strain cycles obtained with the threee models are compared with the experimental results. It can be seen that the DRTN model gives the better agreement but also the elastoplastic law gives appreciable results. The simple linear model, also if is not able to reproduce well all the features of the experimental tests, gives a simulation that seems acceptable in the range of shear strains analyzed. 5.2 Results 5.2.1) Calitri earthquake The comparison of the computed and experimental displacement time histories are reported in fig. 7;in the table below the peak values are summarized. Experimental Bilinear Elastic DRTN displ. 24.23 22.89 35.10 37.91 (mm) -22.03 -29.07 -26.36 -31.32 All the three models give comparable results: the elastic and DRTN models reproduce better the dominant frequency while the bilinear give a closer estimate of the peak displacement. 5.2.2) Tolmezzo earthquake The displacement and acceleration time histories are reported in fig. 8 and 9. In the tables below the peak values are summarized. Experimental Elastic Bilinear DRTN displacement 19.47 25.43 28.37 27.39 (mm) -19.45 -22.64 -29.71 -28.12 acceleration 0.77 0.80 1.22 0.79 (m\sec2) -0.78 -0.92 -1.29 -0.80

As is possible to see a very good agreement with the experimental acceleration time-histories has been reached

-329- with the elastic and DRTN models. The bilinear approximation gives a satisfactory simulation of the dominant frequency but a considerable overestimate of the peak acceleration and, consequently, of the maximum shear. All the three models give similar results for the peak displacements and a good simulation of the dominant frequency. The linear model give better results in the high displacement zone in wich its cycle dissipate a little more energy than the experimental and the stiffness is very close to the secant experimental value. The DRTN model follows better the variation of the dominant frequency with the maximum amplitude. This fact seems to indicate that same additional source of energy dissipation must be taken into account in real tests, probably due to the vertical displacement cycles.

6. CONCLUSIONS

Three different mathematical models of the HDLRB isolation devices have been compared. Appropriate choice of parameters can lead to satisfactory agreement of results in the working strain range (less than about 120 %). In this range an elastic analysis of the isolated structure can be appropriate giving an estimate of displacement and accelerations (forces) comparable with those furnished by non-linear models and in favour of safety.

7. REFERENCES

/I/ Martelli A. & al, "State of the art on base isolation development in Italy", Proc. International Meeting on Earthquake Protection of Buildings, Ancona (Italy) , June 1991. /2/ Sand T, Di Pasquale G. "Modellazione del legame_ costitutivo di isolatori in gomma armata", Ingegneria" Sismica 2/90. /3/ Martin P.P, Bolton Seed H "MASH, a computer program for the non linear analysis of vertically propagating shear waves in horizontally layered deposits", Report UCB/EERC 78/23 University of California, Berkeley /4/Lazan B.J, "Damping of materials and members in structural mechanics", Pergamon Press, 1968 /5/ Sand T., Di Pasquale G. "Modellazione del legame costitutivo di isolatori in gomma armata e studio delle conseguenze sulla risposta sismica di strutture isolate", Rapporto Tecnico RT DISP 90/02. /6/ Forni M. & al. "Dynamic tests on seismically isolated structure mock-ups and validation of numerical models", Proc. International Meeting on Earthquake Protection of Buildings, Ancona (Italy) , June 1991.

-330- Fig. 1 Cross section of the bearings used for the shake- table tests and experimental stress-strain cycles.

10.0 n E u .-'i 6.0- \ .*•• / 2.0-

0 -2.0-

UCIKSA

1-2 retU In ttxvi. tp. 1.5 ai -•- mox displ 17.63 mm -6.0- ••»•• mox displ 27.45 mm 2-2 plactrt tceUlo ap. 3,8 r» O -•- max displ 36.31 mm 3 • 10 luKMnl icelalo •;. 0.75 «. CD

* - 11 focll In io

3 - Tocllo In fo"u*« >p. 2.S »• -10.0 -120. -80. -40. 0. 40. 80. 120. shear strain (%) Fig. 2 Definition of the secant shear modulus and of the equivalent viscous damping and experimental results.

Secant shear modulus and equivalent damping 20.0 -i C7 C E 15.0-i c o 10.0-1 cr G(ga) m

Tig) d 5.0- Jciclo Secant modulus G* D X) Equivalent viscous damping -G(ga) o 2 E 0.0- 0.0 20.0 40.0 60.0 80.0 100.0 shear strain (%)

-331- Fig. 3 Comparison between the experimental stress-strain cycles and the numerical models: a) elastic model b) bilinear model c)DRTN model

o CM II

UJ O o *—- i§ -3- UJ UJ f f c r f 'o O "to -40 . hea r in -80 . 20 . i 120 . 80 . s—> II d ^—. c \ c d

d O

o I

O CM i J 1

o CN -s

o C 'p to o

d -co I

q q q q q q q CD CN cd to d I d o I I (iuobs/6>i) (tuobs/6>j) (Luobs/6>i)

-332- Fig. 4 View of the isolated 394 kN mass, mounted on the MASTER shake table.

MASS

MASTER

A1Y

Fig. 5 Location of bearings and instrumentation for the isolated 394 k.N mass (S = displacement transducers; A = acce1erometers; F = force transducers; T = thermocouples).

-333- Fig. 6 Acceleration time histories used in the shake-table tests and relative response spectra.

. Tclmezzo lime-scaled rccp. spcclio (kj omox) 10.0-

O <1) 6.0- tn — oco.cUO % cJnmp. \ occel. 15 " clomp. E 2.0- — displ 10 ',', damp. ID

Calilri time-scoled resp. jpsctro (0.2g amox] 5.0-1 2.5-i 2.0- 1.5-

1.0- — Displ. 10% ijnmp. (1J 0.5- Displ. 15 % damp. c Accel 15 % domp o 0.0- — Accel 10% domp. o -0.5- u -1.0-

••1.5- O O -2.0- 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 tempo

-334- 10

Fig. 7 Comparison beween computed and experimental displacement time histories : Calitri time-scaled a) elastic model b) bilinear model c) DRTN model

calitri e—w

-25.0- — sperimentale — ELASTICO -35.0

35.0-1

25.0-1

D % -5.0 H O 8" -15.0-1 -25.0- sperimentale DREP

% -5.0 H o & -15.0-)

-25.0- — sperimentole — dnjm -35.0 5.0 10.0 15.0 20.0 25.0 30.0 tempo

-335- oo

accelerazione(m/sec2) accelerazione(m/sec2) accelerazione(m/sec2) 0) $» O —' n o O 3 (0 (D t3 O> (D ••< 0) H( H- ft JD 01 H-rt O O H-S O 3 t3 tr O (D arts (D H- (0 (-3 (D (D 3 crcro •^ H-o to 3 Cf H-O H- H-(D t« (D a (D CO a O O (D

H1 (D (D N M N H- O O 3 —• (0 (+ O OH-ft W 3 0) t-3 (0 I-1 to 3 O O (U (D (D 00 12

Fig. 9 Comparison beween computed and experimental displacement time histories : Tolmezzo time-scaled a) elastic model b) bilinear model c) DRTN model

35.0 -i

— ISMES — ELASTICO -35.0-

35.0-

25.0- o 1—J 15.0- c CD E 5.0- D i—> (0 -5.0- O oo I-J-' — ISMES -25.0 H — DREP -35.0 35.0 -i

25.0-

2 15.0- c CD 5.0- I ilA,rt/S 1 -5.0- V' O 8" -15.0-1

-25.0-1 — DRTN — ISMES

-35.0 1 1 1 1 1 1 1 i 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 tempo

-337- FILE DRTM91E

APPENDIX A : Evaluation of the optimum values of the Modified Hart in-Davi denkov law to reproduce the experimental behaviour of 1/U scale high-damping laminated elastomer bearings.

1.DHM CONSTITUTIVE LAU

The D H H (Davidenkov-Hartin Modified) constitutive law [1 - 2 ] has been further modified to match the experimental results also in the high strain range (up to 200 X shear strain). Three features have been added to the previous version of the law [2]: - a non-linear elastic behaviour in parallel to the original element to simulate the hardening at high strains, - a variable value of the exponent P , which is now a linear function of the strain, to improve the description of the dissipated energy, - a •memory1 of the maximum and minimum values of the shear strain.

The original parameters are:

G : = 35 Shear modulus at low strain (Kg/sqcm) r 0 := .083 Reference shear strain

A := .8 B := .32

The exponent P is now a function of the shear strain according a linear law described by the two parameters:

P 1 : = .03 Value of the exponent at zero strain

a I := .0 0 5 angular coefficient

so the xpression for P is:

P(X) := P1 - al-(X-rO) where X is the normalized strain V/TO

The hardening effect has been modeled by means of a cubic law of the strain whose amplitude is determined by the parameter

1 .5 a : = and has the simple expression D ( X ) : = o • X G

The analytical expressions of the constitutive law are:

A P(X) -X which describes the variation of the H(X) : 1 - secant shear modulus with the strain: 2 B Gsec(X) = G (1 - H(X) ) 1 + (X)

rmi n TI(D :• rmin + G rm i n ) • + G -0

-338- rmax TU(D := max - G'tTnax - r) • 1 - H • G • D < r for unloading 2 • TO-H which describe the shear stress (r) corresponding to the shear strain (r)

2. EVALUATION OF THE THEORETICAL STRESS - STRA IN CYCLES FOR SEVERAL AMPLITUDES

Let us evaluate now the theoretical stress-strain cycles for three different maximum strains and compare them with the experimental results.

Cycle at 51 X of maximum strain Tmax := .51 rmin := Tmax

Tmax - rmin dr : = 40

T1 := Tmin.rmin + dr ..rmax

rmax rmax := G- 11 - H ( rma x) Tmin := -Tmax

Tmin r s 1 < r1 ) : * rmin + G • ( r 1 - rmin)- 1 - H + G • D(T1 ) 2 TO

rmax - T1 := rmax - G-(Tmax - ri)1 1 - H + G • D c n ) 2 TO

Cycle at 78 X of maximum strain Tmax := .78 rmin := -rmax

Tmax - Tmin dr : = 40

V2 := Tmin,Tmin + dr ..rmax

I Tma x Tmax := G • 1 - H (rmax ) Tmin := -Tmax

T 2 - rmin" TS2CT2) := ruin + G • < r 2 - rmin) 1 - H • G•D CT2 ) I 2 TO rmax - T2 rd2

-339- ' i m d A i in u m s i i

rmax - ruin dr : =

T3 := rmin.rmin + dr ..rmax

rmax rma x :• G 1 - H (rmax) rmin := -rmax ro T3 - rmin"J"J T s 3 ( r 3 ) := ruin + G • ( r 3 - r m i n) • 11 - H • G •D(T3 ) [ 2T0 JJ

rmax_T3 Td3(T3) := Tinax - G-CTmax - T3>- 1 - H • G-D(T3) 2T0

3. EXPERIMENTAL RESULTS

The experimental results have been obtained from a static cyclic test on a 1/4 scale bearing (n.1) 12 5 mm in diameter and 11 x 3 + 2 x 1.5 =36 mm total rubber height:

The reuslts are stored in the following files

50 X max strain : fs1s.DAT Fs1f.DAT np=19 75 X ' ' : fs2s.DAT Fs2f.0AT np=15 100 X « ' : fs3s.DAT Fs3f.DAT np=19

The data are expressed as displacement (mm) and forces (KN) and are to be converted into strains (-) and stresses (Kg/cmq)

* 2 Areaiso := —12.5 cmq hgiso := 35 mm np1 := 19 np2 := 15 np3 := 19 4

1st test j1 := 1 . . np1 us := READ(fs1s ) Fs : = READ(fsif) J1 J1 Fs J1 J1 rsi TSp1 ji hgiso J1 .01 'Areaiso

2nd test j 2 := 1 . .n p 2 us : = READ(fs2s) Fs : = READ(fs2f) j 2 J 2

-340- Fs J2 j2 rs2 T s p2 j 2 hg i so J2 .01 -Areaiso

3rd test j3:=1..np3 us :=READ(fs3s> Fs :=READ(fs3f> j 3 j 3

us Fs J3 J3 rs3 T sp3 j 3 hg i so j 3 . 0 1 • A r e a i s o

4. COMPARISON BETWEEN THEORETICAL AND EXPERIMENTAL STRESS-STRA I N CURVES

In the following figures the experimental data (continuous line with crosses) are compared with the theoretical ones (dotted line) showing a remarkable agreement in the investigated field of strains.

SI X max. strain

TS1 , rdi , rsp1 J1

-5 . 5 5 n.n.rsi .55 ii

-341- 78 X max. strain

Ts2(r2),Td2

- 1 0 .8 T2,T2,Ts2 .8 J2

104 X max. strain

TS3(T3 ) ,Td3(T3),rsp3 J3

- 1 0 -1.1 T3,T3,Ts3 1.1 J3

-342-