ML20128F354

Seismic Safety Research Program Plan

ML20128F354
Person / Time
Issue date:	06/30/1985
From:	NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:
References
NUREG-1147, NUDOCS 8507080215
Download: ML20128F354 (195)

Text

.

NUREG-1147 Seismic Safety Research Program Plan U.S. Nuclear Regulatory Commission Office of Nuclear Regulatory Research g 700 g g 850630 1147 R PDR

NOTICE Availability of Reference Materials Cited in NRC Publications Most documents cited in NRC publications will be available from one of the following sources:

1. The NRC Public Document Room,1717 H Street, N.W.

Washington, DC 20555

2. The Superintendent of Documents, U.S. Government Printing Office, Post Office Box 37082, Washington, DC 20013-7982

3. The National Technical Information Service, Springfield, VA 22161 Although the listing that follows represents the majority of documents cited in NRC publications, it is not intended to be exhaustive.

Referenced documents available for inspection and copying for a fee from the NRC Public Docu-ment Room include N RC correspondence and internal N RC memoranda; NRC Office of Inspection and Enforcement bulletins, circulars, information notices, inspection and investigation notices; Licensee Event Reports; vendor reports and correspondence; Commission papers; and applicant and licensee documents and correspondence.

The following documents in the NUREG series are available for purchase from the NRC/GPO Sales Program: formal NRC staff and contractor reports, NRC-sponsored conference proceedings, and NRC booklets and brochures. Also available are Regulatory Guides, NRC regulations in the Code of Federal Regulations, and Nuclear Regulatory Commission Issuances.

Documents available from the National Technical Information Service include NUREG series reports and technical reports prepared by other federal agencies and reports prepared by the Atomic Energy Commission, forerunner agency to the Nuclear Regulatory Commission.

Documents available from public and special technical libraries include all open literature items, such as books, journal and periodical articles, and transactions. Federal Register notices, federal and state legislation, and congressional reports can usually be obtained from these libraries.

Documents such as theses, dissertations, foreign reports and translations, and non-NRC conference proceedings are available for purchase from the organization sponsoring the publication cited.

Single copies of NRC draft reports are available free, to the extent of supply, upon written request to the Division of Technical Information and Document Control, U.S. Nuclear Regulatory Com-mission, Washington, DC 20555.

Copies of industry codes and standards used in a substantive manner in the NRC regulatory process are maintained at the NRC Library, 7920 Norfolk Avenue, Bethesda, Maryland, and are available there for reference use by the public. Codes and standards are usually copyrighted and may be purchased from the originating organization or, if they are American National Standards, from the American National Standards Institute,1430 Broadway, New York, NY 10018.

NUREG-1147 RA,RD,RM Seismic Safety Research Program Plan Manuscript Completed: June 1985 Data Published: June 1985 Division of Engineering Technology Office of Nuclear Regulatory Research ,

U.S. Nuclear Regulatory Commission W:shington, D.C. 20555 p" ~%,,

Q..M)...

1 1

FOREWORD This document presents a plan for seismic research to be performed by the Mechanical / Structural Engineering Branch, Division of Engineering Technology, and the Earth Sciences Branch, Division of Radiation Programs and Earth Sciences, in the Office of Nuclear Regulatory Research. This plan describes the safety issues, regulatory needs, and the research necessary to address these needs. The plan also discusses the relationship.between current and proposed research within the NRC and research sponsored by other government agencies, universities, industry groups, professional societies, and foreign sources.

We perceive this plan to be a living document and expect to revise it periodically to take into account our experience in implementing the plan and comments received from interested parties within the NRC and among the public.

Comments on this document are welccmed at any tinie and will be considered in the development of subsequent editions of this plan. They need not be restricted to the research activities described herein; comments identifying omissions or recommending additional research are also welcome. Comments should be addressed to:

Roger M. Kenneally Mechanical / Structural Engineering Branch Division of Engineering Technology Mail Stop 1130SS U.S. Nuclear Regulatory Commission Washington, DC 20555 Telephone: (301) 492-7000 i

l l

l B

iii

I-ACKNOWLEDGMENTS The NRC st6ff contributing to the developnent of this plan included:

E. Gunter Arndt, John J. Burns, James F. Costello, Daniel J. Guzy, Roger M. Kerneally, Andrew J. Murphy, John A. O'Brien, Jacob Philip, and i

Thomas J. Schmitt. Frances Miller and Shirley Pons rendered word prccessing services and Lousie Gallagher did the technical editing on the text of the plan.

'l i

i e

l l

I a

1 Y

.- e TABLE OF CONTENTS Section Page Foreword........................................................ iii Acknowledgments................................................ v 1.

Introduction............................................... 1-1

2. Issues 2.1 Seismic Hazard...................... ................... ?-1 2.2 Seismic 2.3 Seismic Risk........................ ................. 2-2 Margins...................... .................

2-2

3. Regulatory Needs, Justifications, and Program Objectives 3.1 Seismic Hazard........................................ 3-1 3.2 Seismic Risk........................... .............. 3-2 3.3 Seismic Margins........................ .............. 3-3

4. Project Description 4.1 Seismotectonic Program................................ 4-1 4.2 Soil Res ponse Projec t. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-3 4.3 Structural Response Project........................... 4-4 4.4 Containment Buckling Project.......................... 4-6 4.5 Load Combinations for Structural Design Project....... 4-7 4.6 Piping Reliability Projects........................... 4-8 4.7 Component Fragility Project........................... 4-10 4.8 SSMRP-BWR Risk Assessment Project..................... 4-11 d.9 Validation of Seismic Calculational Methods Project... 4-12 4.10 NRC Seismic Design Margins Project.................... 4-14

5. Glos sa ry - Acronyms and Defini tiens . . . . . . . . . . . . . . . . . . . . . . . . . Gl-1 6.

References.................................................. R-1

7. Bibliography................................................ Bib-1 Appendices 1 A. Sei smotec ton ic Prog ram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-1 i B. Soil Response Project................................... B-1 i C. Structural Response Project............................. C-1 D. Piping Reliability Projects............................ 0-1 E. Component Fragility Project............................. E-1 F. Validation of Seismic Calculational Methods Project..... F-1 G. NRC Seismic Design Margins Project...................... G-1 vii

1. INTRODUCTION The Seismic Safety Research Program Plan describes NRC research activities to improve estimates of earthquake hazards and the effects of earthquakes on nuclear power plants. This plan addresses short-term, high-priority research to support immediate needs relative to quantifying and reducing uncertainty associated with current regulatory requirements and long-term research to resolve concerns related to seismic design margins. Short-term research improves requirements by removing conservatisms where they are unnecessary and adding conservatisms where weaknesses in the regulations exist, thereby achieving a more balanced design. Long-term research, on the other hand, aims at more strategic questions involving the overall view as to how important earthquakes are in the regulatory process. NRC's seismic research hos been divided into the broad general categories of seismic hazard, seismic risk, and seismic margins. Within the context of this plan, the following definitions apply:

seismic hazard - the natural source of vibratory ground motion and its propagation; seismic risk - the probability of core melt and/or radioactive release due to earthquake effects. The Seismic Safety Research Program addresses the methodology to determine seismic risk; and seismic margins - a measure of how much larger than the design basis earthquake an earthquake must be to compromise plant safety. The Seismic Safety Research Program addresses the methods and procedures to determine seismic margins.

This plan describes the safety issues, regulatory needs, and the research necessary to address these needs. The plan also discusses the relationship between current and proposed research within the NRC and research sponsored by other government agencies, universities, industry groups, professional societies, and foreign sources. The aim of the Seismic Safety Research Program is to support NRC licensing decision-making and to maximize the utilization of research findings from all data sources. The scope covers (in decreasing order of priority) presently operating plants, plants under construction, and future construction that may be proposed. Disciplines covered include the geosciences and geotechnical, structural, and mechanical engineering.

The NRC Office of Nuclear Reactor Regulation (NRR) needs in seismic analysis methodology were described in an April 8,1982 memorandum from H. R. Denton to R. B. Minogue (Ref.1). This memorandum called for a redirection of the Seismic Safety Margins Research Program (SSMRP) to meet short-term needs and suggested an integrated program plan for long-term seismic analysis research. Additional comments on seismic-related 1-1

research are contained in NRR's input for the FY 1986-1990 Long-Range Research Plan (Ref. 2) and ACRS/ staff correspondence on the quantification of seismic design margins (Refs. 3 through 6).

The seismic issues addressed in this plan are described in Section 2.

Section 3 describes the regulatory needs, justifications, and program objectives. The projects to meet the regulatory needs and objectives are described in Section 4 of this plan and are composed of the following:

1. Seismotectonic Program - research to quantify and reduce the uncertainty in seismic hazard assessments and to develop methods of dealing with the related uncertainties; 2.

Soil failureResp (onse Project settlement - research or liquefaction) to essess at nuclear theplant power potential sites for soil subjected to earthquakes at and above the Safe Shutdown Earthquake;

3. Structural Response Project - analytical and experimental research to reduce uncertainties and provide accurate estimates of how noncontainment structures transmit loads from earthquakes larger than the design basis to safety systems and components;

4. Containment Buckling Project - experimental research to validate current methods used to predict buckling loads and locations to ensure that containment shell stability and leak integrity is maintained;

5. Load Combinations for Structural Design Project - analytical research to reduce uncertainties in the methods for combining the loads (earthquakes, temperature, pressure, etc.) used in the design of nuclear power plant structures to ensure structural integrity;

6. Piping Reliability Projects - analytical and experimental research to reduce uncertainties in determining response and failure modes and to develop more realistic design criteria resulting in balanced safety between operation and accident conditions;

7. Component Fragility Project - experimental research to reduce the uncertainties associated with predicting the earthquake level at

! which critical components fail to perform their safety functions; 1

1 8. SSMRP-BWR Risk Assessment Project - analytical research to expand the i

existing simplified seismic risk methodology developed for a PWR to include a BWR;

9. Validation of Seismic Calculational Methods Project - foreign experimental and analytical research to validate uncertainties in

! computer codes used to predict the response, particularly in the nonlinear range, of structures, systems, and components; and

10. NRC Seismic Design Margins Project - research to provide ard demonstrate a procedure for assessing the adequacy of nuclear power plants to withstand earthquake levels above their design basis.

1-2

Figure I shows the project interrelationships and regulatory applications. Project integration is the responsibility of the NRC Office of Nuclear Regulatory Research.

A glossary containing the acronyms and definitions of seismic-related terms used in this plan is provided in Section 5. Section 6 contains references and Section 7 contains a bibliography of project-related reports.

I i

1-3 l

i  ?

FOREIGN AND DOMESTIC l NRC StiSMIC SAFETY RESEARCH PROGRAM  !

RELATED PROGRAMS SEISMIC HAZARD l {

o EPRI and Industry Probabilistic  ; j SEISMIC HAZARD Seismic Hazard Program EARTH SCIENCES g o Charleston Earthquake l

o USGS Ground Motion Studies o Setssetectonic Program l .! o New Brunswick Earthquake o National Ear ke Hazard Re- l o soll Response Project duction Prog. . NSF. FEMA. ,! o Site-Specific Spectre Nas)  :

. l o Corps of Engineers Seismic Soils !

Settlement Program  ; l ETHODS DEVELOPENT AND VALIDATION APPLICATIONS  !

!  ! SEISMIC RISK VALIDATION OF SEISMIC ETH005 ins l o o Seismic Design  !

o Germany - KfK HOR Program  !

Seismic Program Research Safety Mag (SSNtP) ) Margins Project o Licensee PRAs I o Validation of Seismic o Load Combinations  : o Integrated Safety o EPRI - Taiwan SSI Experiments  ! Caleviation Methods Project for Structural l Assessment Prograse o Japan - MITI/MUPEC Seismic A 8 '"  !.

Experiments o Risk Methodology

!  ! Integration &

I Evaluation Program FRAGILITIES AND RESPONSE  : 3Y g L o EPRI Piping Program I FRAGILITIES AND RESPONSE  !

l  ;

o EPRI Equipment Qualification

• '" "E**** **

,! SEISMIC MnaGINS Program o Component Frag 111 ties Project l l o NTOL Seismic Issues o NRC E ipment Quellfication o Piping Reitability Projects o Pipe Damping (R.S. 1.61) o NRC Aging Pmgram o Containment Buckling Project "

l  !

o Seismic Qualification Utility  !  !

o ASME Code-Piping &

Group Data  ! Structures I e l j SEISMIC MARGINS o Equipment Qualification o EPRI Seismic Margins Research l o Structural Damping Program {

l

' l (R.G.1.61) o Industry Probabilistic Risk j j o Independent Pipe Support Analyses "'"

l* Figure 1 Project Interrelationships  :

and Regulatory Applications l _.- _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ - . _ _ . - - . . -

2. ISSUES The seismic issues discussed, below are divided into three categories; seismic hazard, seismic risk, and seismic margins. Whereas these issues are discussed from the NRC perspective, the NRC is not alone in sponsoring research addressing these related areas. Seismic safety research is being performed by other organizations within the United States (e.g., the Electric Power Research Institute and the United States Geological Survey (USGS)) and in other countries (e.g., Japan and the Federal Republic of Germany) with several joint projects and significant interaction and cooperation in project planning.

2.1 Seismic Hazard The fundamental seismic hazard issues are how to quantify and to reduce the uncertainties in seismic hazard assessments and how to develop techniques to deal with the uncertainties in a regulatory environment.

Factors that contribute to the uncertainty in the seismic hazard assessment are (1) the uncertainty in establishing seismic source zones, (2) the uncertainty in the propagation of seismic energy, and (3) the uncertainty in the site-specific ground-motion response, including soil response.

In addition to the fundamental seismic hazard issues, the issue of the recurrence of an 1886 Charleston-size earthquake anywhere on the eastern seaboard has comanded significant NRC attention. In November 1982, the USGS clarified its position with respect to the 1886 Charleston earthquake (Ref. 7). The clarifying statement represents not so much a new understanding but rather a more explicit recognition of existing uncertainties with respect to the causative structure and mechanism of the 1886 Charleston earthquake. Many hypotheses have been proposed as to seismogenic mechanisms and potential location on the eastern seaboard of future Charleston-size earthquakes. Some of these hypotheses would limit such an earthquake in both size and location while others would allow this earthquake to occur over very large areas of the Eastern United States and Canada. Presently, none of these hypotheses is definitive and all contain strong elements of speculation.

Also, a Policy Issue memorandum for the Commissioners (SECY-82-53,  ;

February 5, 1982) on the January 9-11, 1982 New Brunswick, Canada l

earthquake (Ref. 8) stated in part: "Although all information relating to )

the size and location of the event is preliminary, it eventually may be '

concluded that this earthquake could have occurred anywhere within the New England Piedmont Tectonic Province and in accordance with Appendix A to 10 CFR Part 100, would represent the largest historical earthquake in that province...which includes much of New England and southern New York." If the NRC were to act conservatively on the Charleston and New Brunswick earthquake issues, using the procedures of Appendix A to 10 CFR Part 100 to establish revised values for the Safe Shutdown Earthquake (SSE), some Eastern United States nuclear power plants could have their revised ground acceleration significantly increased over their original design value.

The Commission would then be faced with the possibility of having to shut down these plants pending reanalysis of existing margins or requiring extensive structural and equipment modifications to meet the desired safety level.

2-1

2.2 Seismic Risk Probabilistic risk analyses completed to date (Refs. 9 through 15) indicate that accidents initiated by large earthquakes are major contri'utors o to public risk. In calculating the seismic risk for a given

plant, beyondit their is necessary to p(redict design basis the behavior i.e., inelastic of structures or nonlinear and systems response). An understanding of inelastic response is necessary since it is generally recognized that structures and components may be damaged (in the sense that they are permanently distorted) as a result of earthquake excita-tions and yet may continue to perform their safety functions.

Sophisticated methods such as those developed under the Seismic Safety Margins Research Program (SSMRP) are available to predict a "best estimate" behavior of nuclear power plants subjected to large earthquakes (up to roughly six times the SSE). However, these calculational methods are based on linear assumptions with factors applied to account for nonlinear behavior and have not been validated to provide the confidence and creoibility necessary to make sound reaulatory decisions. In addition, much of the response and fragility (failure) data used in these analyses are based on subjective judgment or extrapolation and thus have large uncertainties.

The seismic risk issue addressed in this plan focuses on developing and validating a methodology to determine the probability of core melt and/or radioactive release. The calculational methods used, as noted above, have large uncertainties. In order to improve our estimates of seismic risk, programs have been developed to experimentally determine the fragility levels of key structures, components, and systems and to analytically and experimentally validate, or calibrate to the extent possible, calculational methods, particularly those predicting inelastic behavior. Then, if the seismic risk is determined to be too large, potential changes in regulations can be assessed using a validated technical approach.

2.3 Seismic Margins Frequently, the NRC is faced with decisions related to the seismic design of operating plants. As operating reactor experience (e.g., design and construction deficiencies) or research results (e.g., new information on the seismic hazard, soil-structure interaction, and plant seismic response) becomes available, changes in the design loads on structures, systems, and equipment may be recommended. The seismic margin issue relates to whether these changes can be accommodated within the inherent capacity of the original design or whether plant modifications are warranted.

Recent studies such as the SSMRP indicate that nuclear plants generally have high margins against seismically induced failure. The performance of conventional power plants in past earthquakes confirms the existence of substantial seismic capacity in nuclear power plants.

However, current design practices lead to the excessive use of pipe supports, which may have an adverse effect on safety by making the pipe system too stiff. More flexible piping systems will reduce the number of 2-2

supports providing a better safety balance between operation and accident conditions and will reduce radiation exposure to workers performing inspection and maintenance. A sound, practical seismic margins program using margins-to-failure analysis and seismic probabilistic risk assessment techniques will serve to minimize the need for changing requirements'and licensing actions as estimates of the seismic hazard and system respo'ise change. In addition, seismic margin studies can provide a sound basis for establishing confidence in the seismic capacity of nuclear power plants and serve to indicate, if necessary, places where seismic risk should be reduced.

2-3

3. REGULATORY NEEDS, JUSTIFICATIONS, AND OBJECTIVES 3.1 Seismic Hazards 3.1.1 Regulatory Needs and Justifications

1. Need: Data concernino seismic source zones in the Eastern United States, including Charleston, New Madrid, New England,

-and others, and data concerning seismic energy propogation in the Eastern United States, needed on a day-to-day basis in the licensing process for the analysis of seismotectonic provinces as required by Appendix A to 10 CFR Part 100, to be the basis for amendments to Appendix A to 10 CFR Part 100 and for developing regulatory guidance.

Justification: Except for the New Madrid seismicity, the distribution of seismicity in the East, including New England ,

and the vicinity of Charleston, S.C., is not well defined. No working hypothesis for the cause of the seismicity is generally dCCepted by the geoscience conmunity. The stress fields that drive the major faults are poorly known.

2. Need: An information base for the development of site-specific spectra, to be the basis for amendments to Appendix A to 10 CFR Part 100 and for developing regulatory guidance.

Justification: The recent earthquakes in New Brunswick, New Hampshire, and Arkansas have generated important strong-motion records that for the first time provide a significant opportunity to compare real data with theoretical ground motion and attenuation models for the Eastern United States. Analysis of these records will address important regulatory questions concerning the interpretation of this type of record and its use in licensing decisions.

3. Need: Methods for handling the uncertainties in assessing the potential risk from seismic hazards, including such topics as the applicability of using the historic method of seismic hazard analysis, the impact of the 1982 New Brunswick earthquake ,

sequence, and the verification of Holocene movement on the Meers l fault, to be used to revise current siting regulations.  !

Justification: The current seismic siting regulations do not provide guidance on how to handle the uncertainty associated with licensing decisions and judgments being made at the forefront of a rapidly developing science, seismology.

Decisions must be made based on the best available data that are l steadily being updated. This program is designed to develop l statistical or probabilistic tools to aid the decisionmaking process.

l 3-1 l

_ . .e_m w - . . - . . - - - - .

4._ Need: _ Improved' in situ and analytical procedures are required for predicting soil' liquefaction potential for acceleration

. values at or above the Safe Shutdown Earthquake (SSE).

Justification: The recent high acceleration records from New Brunswick, New Hampshire, and Arkansas and the potential for

'large " anchor-point" accelerations because of the Charleston earthquake issue may reduce the safety margins associated with current soil liquefaction prediction techniques.

5.- Need: Verification of' methods for predicting seismically induced soil settlements.

Justification: The consequences of dynamic soil settlement and liquefaction at earthquake levels near or above the SSE could be a significant contributor to overall risk. Current probabilistic risk assessments (PRAs) do not adequately address

. this problem, and at the present time there are no verified methods for estimating seismically induced soil settlement.

3.1.2 Program Objectives

1. Improve methods for assessing the seismic hazard of the Eastern United States sites.

2. Reduce the uncertainties related to estimating seismic hazards and seismically induced soil liquefaction and settlement.

3.2 Seismic Risk 3.2.1 Regulatory Needs and Justifications 1.. Need: ~ Complete development of seismic PRA methods.

Justification: The detailed and simplified seismic risk methodology developed in the Seismic Safety Margins Research Program (SSMRP) was demonstrated by application to Zion Unit 1, a PWR plant. Thus, all the systems analysis models (initiating events, event trees, and fault trees) and all the structural and piping models were developed for a PWR plant, even though the methodology developed in the SSMRP should be equally applied to both PWRs and BWRs. A simplified seismic risk analysis of a BWR is being performed to verify the applicability of the SSMRP methodology to both PWRs and BWRs.

2. Need: Validation of current seismic PRA methods.

Justification: The SSMRP and industry methods for estimating structure, system, and component responses used in seismic risk analyses are fairly new and have not been subjected to experimental validation. The fragility data rely heavily on expert opinion, and critical system modeling assumptions are made in the risk analyses. Validation will increase our 3-2

confidence in seismic PRA methods and their effectiveness so that they may be confidently used in the regulatory decisionmaking process.

3.2.2 Program Objectives l 1. Develop a simplified seismic risk methodology to assess the l probability of core melt and/or radioactive release.

2. Determine which elements of the probabilistic response calculations used in seismic risk analysis methodology are most in need of validation.

3. Identify tests and apply results that adequately challenge the elements of the probabilistic response calculations to be validated.

4. Assess the adequacy of the assumptions and subjective information used in probabilistic response analysis.

3.3 Seismic Margins 3.3.1 Regulatory Needs and Justifications

1. Need: Evaluate current seismic design criteria and the effects of uncertainties to assist the licensing staff in their assessment of near-term operating licenses and Systematic Evaluation Program (SEP) plants.

Justification: Current regulatory requirements are based on conservative assumptions and expert opinion and lack appropriate test or earthquake experience data to quantify the degree of conservatism. Quantifying these conservatisms will enable the staff to judge the necessity of modifying and requalifying i

' structures and components in older plants or improving design criteria for new plants.

2. Need: Data to predict the nonlinear response and failure modes l and levels of nuclear shear wall structures.

l Justification: Recent seismic PRAs indicate that structural Tailures leading to failure of the equipment housed or supported l therein are dominant risk contributors. Our lack of appropriate l test or earthquake experience data has resulted in large uncer- l tainties associated with structural fragilities. It is also necessary to more clearly understand the nonlinear response of shear wall structures so that the high seismic load input to equipment attached to the structures can be better defined.

3. Need: An improved seismic fragility data base for mechanical and electrical equipment.

Justification: Seismic response quantities in all PRAs are compared against probabilistically defined failure levels called 3-3

1 s fragilities. Since many structures, components, and systems demonstrate several failure modes, there can be several fragilities for each individual structure, component, or system.

Most of the fragility data presently used are based on subjective judgment or military data that, unfortunately, do not characterize earthquake phenomenology. The few comparisons between real data and PRA estimates of fragilities suggest that PRAs are using conservative estimates of component fragilities, which in turn leads to pessimistic evaluations of the seismic threat. Realistic estimates of fragilities for electrical components are particularly needed, but attention must also be focused on mechanical components. Moreover, the role of such factors as anchorage, size, age, and concurrent loads in defining fragility must be established. Without reliable seismic fragilities for electrical and mechanical components, seismic PRAs are of little value in licensing.

4. Need: Estimate and evaluate the margins available in the seismic design of structures, piping, and equipment in nuclear power plants using simplified approaches to determine the risks associated with seismic events beyond the design basis.

Justification: Potential changes in the ability of a nuclear power plant to withstand large earthquakes based on new information on seismicity, soil-structure interaction, plant response, and fragilities must be evaluated and compared with the margin of the original design. The seismic design margins in older piants are of particular interest since these generally had less conservative design criteria for their original design.

3.3.2 Program Objectives

1. Provide research to support improvements and reduce uncertainties in the current seismic design criteria.

2. Reduce the uncertainty in determining the response and failure modes of structures, equipment, and piping subjected to earthquake levels above the SSE.

3. Develop a simplified review method to assess seismic design margins.

4. Provide a basis of technical recommendations, data, and procedures that will serve in the development and implementation of an NRC plan to quantify seismic design margins.

5. Assess the seismic margins adequacy of two trial plants. These assessments will serve as benchmarks for possible future licensing reviews.

3-4

4. PROJECT DESCRIPTION The following is an executive summary of the project descriptions.  ;

More detailed descriptions for most of these projects are contained in the '

appendices to this plan. Each of these project descriptions includes the following sections:  !

1. Scope - A brief description of the project and how it supports the program objectives.

2. Integration and Relationship with Other Research - A description :

of how the projects included in this plan are integrated to address the program objectives. Also' included is a description l of the pertinent research and development being performed by other government agencies, universities, industry groups, professional societies, and foreign sources.

3. Products - Research products and their scheduled completion dates. The program categories they address (i.e., seismic hazard, seismic risk, or seismic margins) are indicated in parentheses.

4.1 Seismotectonic Program (Appendix A)

1. Scope This research will quantify and reduce uncertainty in seismic hazard assessments and will develop means to deal with the related uncertainties.

It will also support seismic risk and seismic margins objectives by assessing the adequacy of the assumption cnd subjective information used to generate the seismic hazard curves.

The uncertainty in the characterization of seismic source zones is the fundamental issue addressed by the regional projects described in Appendix A. This uncertainty arises from several aspects of the issue such as delineation of the seismic source zone, estimation of the maximum credible earthquake, characterization of source parameters, and occurrence relationships.

The two specific issues described in Section 2.1 of this plan, i.e.,

the recurrence of a particular earthquake (Charleston) and the designation of the province of the New Brunswick earthquake, are prime examples of the problem of characterizing seismic source zones. Resolution of the fundamental issue requires improved knowledge of the seismotectonics of the Eastern United States. The earth sciences research program is directed toward determining the how and why of the occurrence of earthquakes in the East.

The seismographic networks are the mainstay of the regional research program. They provide the basic seismological data set on the recent occurrence of earthquakes, their locations, depths, magnitudes, slip orientations, source parameters, etc. They provide important information on the propagation of seismic energy over distance. They guide the siting of supplementary geological / geophysical investigations on seismically active structures.

4-1

l 1

l l

The topical projects in Appendix A address the uncertainty in the seismic hazard issues that are not identified with a specific source zone or that cover several source zones, such as seismic energy propagation, site-specific spectral analysis, and soil settlement or liquefaction. ,

The resolution of this set of issues generally reouires additional data and its analysis and interpretation. The uncertainty in seismic energy propagation is a reasonable example to consider. The seismographic networks are recording the basic data required, which are seismic energy amplitudes at many distances over many travel paths for numerous earthquakes. The basic data need to be correlated and analyzed to develop models to predict the attenuation of ground motion with distance from specific source zcnes. This topic is being investigated by several network operators covering large areas such as the Southeastern United States.

" Handling Uncertainties" is a section in Appendix A associated with seismic hazard assessments and addresses topics that are not amenable to short-term resolution but for which there are pressing regulatory requirements for interim positions. This research generally takes the form of development of probabilistic techniques for the analysis of existing data sets. The Seismic Hszard Characterization Project (SHCP) is an excellent example of this type of research. The issue of the recurrence of a Charleston-sized earthquake on the Atlantic seaboard is a seismic zonation issue that has been under active investigation for about 10-15 years, and a resolution is not in sight. The SHCP was undertaken to provide NRR with a screening tool to determine which eastern nuclear power plants may require reanalysis of their seismic designs.

2. Integration and Relationship with Other Research Results from seismotectonic research are integrated into the Validation of Seismic Calculational Methods Project (Appendix F) by providing basic background data (e.g., earthquake acceleration, intensity, duration) used in soil-structure interaction (SSI) analyses. In addition, data on the geometry of seismic source zones and attenuation curves are integrated into the Soil Response Project (Appendix B) to quantify the risk of seismically induced soil settlement and liquefaction. Also, products from the Seismotectonic Program when coupled with the NRC Seismic Design Margins Project (Appendix G) will provide insights into the adequacy of the current regulatory criteria and the ability of an existing nuclear power plant to sustain an earthquake larger than the design basis.

In addition to the NRC-funded research in seismic hazard assessment, other Federal, State, and industrial organizations are funding and

conducting this type of research. Notably, the Federal Emergency Management Agency (FEMA), the U.S. Geological Survey (USGS), the National l Science Foundation (NSF), and the National Bureau of Standards (NBS) are l

the Federal participants in the National Earthquake Hazard Reduction Program. Because of higher levels of seismic activity, these principal Federal efforts give emphasis to the Western United States. The State geologic surveys make a significant contribution through their programs of regional hazard mapping and assessments. Industry, in general, and the 4-2 i

.1

, 4 public utilities, in particular, have conducted programs in hazard assessment. The Electric Power Research Institute (EPRI) has instituted a major effort on estimating the seismic hazard at nuclear power plant sites.. The Corps of Engineers (COE) is concerned with seismic hazard primarily as it affects their structures. They have produced a seismic-hazard map of the Cent;al United States and are working on one for the eastern seaboard.-

3. Products h  ;

a.. Seismographic network data used on a day-to-day basis by licensing staff, by staff involved in probabilistic risk assessments (PRAs), by the seismic hazard characterization projects, and for rulemaking decisions and engineering research projects (1986-1990). (Supports Seismic Hazard, Risk, and Margins Issues)

b. Geophysical. data for determining crustal structure in areas of suspicious geologic structures to be used on a day-to-day basis by licensing staff, by staff involved in PRAs, by the seismic hazard characterization projects, and for rulemaking decisions and engineering research projects (1986-1990). (Supports Seismic Hazard, Risk, and Margins Issues)

c. Data from the in situ stress measurement program in the Northeastern United States to be used on a day-to-day basis by licensing staff, by staff involved in PRAs, by the seismic hazard characterization projects-, and for rulemaking decisions and engineering research i

' projects (1986). (Supports Seismic Hazard, Risk, and Margins Issues)

d. Preliminary techniques for calculating site-specific response spectra to be used on a day-to-day basis by the licensing staff and for 7, rulemaking decisions (1988'. (Supports Seismic Hazard, Risk, and Margins Issues)

e. .Probabilistic study of the sensitivity of ground-motion estimates at eastern nuclear power plants on the various theories for the causes l

l of seismicity in the Eastern United States to be used on a day-to-day  !

basis by the licensing staff and for rulemaking decisions (1986).

j (Supports Seismic Hazard, Risk, and Margins Issues)

3. 4.2 Soil Response Project (Appendix B) s

1. Scope' To account for the uncertainties in geotechnical engineering parameters, probabilistic methodologies are necessary. These methods could be used 'in quantifying the risk of dynamic settlement and.

-liquefaction. Soil settlement and liquefaction research assesses the

. potential for soil failure at nuclear power plant sites subjected to )

earthquakes'at and above the Safe Shutdown Earthquake (SSE) level. At present, there are no data to assess the adequacy of the assumptions used in the analytical models to predict seismically induced soil settlement

>. j and liquefaction.

r ,.

no 4-3

+

The research program on seismically induced soil settlements, conducted at the U.S. Army Corps of Engineers Waterways Experiment Station, focuses on validating several settlement models. In addition, the COE is developing a state-of-the-art report on seismically induced liquefaction.

2. Integration and Relationship with Other Research The Soil Response Project is coordinated with others in this plan in that it uses data on seismic source zones and earthquake level intensity, duration, etc., obtained from the Seismotectonic Program (Appendix A).

When available, results from the soil response research will be incorporated into future PRAs and other analytical techniques to provide insights into the ability of nuclear facilities to sustain an earthquake larger than the design basis.

The COE maintains arrays of strong-motion accelerographs at many of their dams. This is particularly important in the Eastern United States where not many such instruments are deployed. These instruments are deployed to provide the COE with data specifically applicable to dam safety so some of the data are not directly applicable to more generic safety problems. Research on seismically induced settlements and liquefaction is mainly being conducted by the COE and the USGS. Futurc NRC research will take advantage of the work sponsored by these groups.

Coordination will be maintained through existing agency contacts.

3. Products

a. Validation of seismic settlement model to assess the potential for soil failure of plant sites subjected to earthquakes at or above the SSE(1987). (Supports Seismic Hazard, Risk, and Margins Issues)

b. State-of-the-art evaluation of liquefaction potential at plant sites (1986). (Supports Seismic Hazard, Risk, and Margins Issues) 4.3 Structural Response Project (Appendix C)

1. Scope This research is focused on answering questions related to the issue of whether existing nuclear facilities can continue to operate in light of more demanding earthquake criteria than those considered in the original design. This need is often the result of changing perceptions of the seismic hazard of an area based on the acquisition of new knowledge or the nearby occurrence of earthquakes. There are uncertainties in the calculational methods used to predict how buildings transmit the larger earthquake loads to safety systems and components.

4-4

The project will provide analytical and experimental data assessing how the parameters used in the design of safety-related equipment and noncontainment structures are affecteo by earthquake loads above the design level. These design parameters include, but are not limited to, structural fundamental frequency; damping; and response spectra, accelerations, and displacements at the various floor levels. The sensitivity of these design parameters to changes in internal and external wall configurations, design practices, and magnitude and duration of seismic input motion will be determined as the model configurations are subjected to quasi-static and seismic tests causing elastic and inelastic responses.

These sensitivity tests are necessary because the Auxiliary Buildings, Diesel Generator Buildings, etc., as placed in the various nuclear power plants, are very plant specific. In order to gain an insight into the effects of additional walls, different numbers of floors, or different size buildings, several parameters are varied so the results can be applied to the various configurations that are actually in operating plants.

This research will support seismic margin objectives by providing analytical methods and experimental data to support improvements and reduce uncertainties in the current seismic design criteria. This research will also verify the adequacy of assumptions and subjective information used in probabilistic structural risk analyses by providing failure modes and levels for the majority of configurations tested. In some limited cases, the model may be so large that the experimental facility would not have the capability to fail it.

2. Integration and Relationship with Other Research This project will provide experimentally obtained elastic and inelastic response information to the Validation of Seismic Calculational Methods Project (Appendix F) and the NRC Seismic Design Margins Project (Appendix G). In addition, information on the changes of floor response spectra that are used in the design of systems and com provided to the Piping Reliability Projects (Appendix D)ponents will be and the Component Fragility Project (Appendix E).

Project planning and evaluation of experiment results will be based in part on data available from shaker tests at the Heissdampfreaktor (HDR) facility in the Federal Republic of Germany (FRG). The HOR tests, which are one element in the Validation of Seismic Calculational Methods Project (Appendix F), will provide vibratory information on the behavior of shear walls contained in an actual nuclear facility.

The Construction Technology Laboratories, a Division of the Portland Cement Association, has performed ::tatic tests on isolated shear walls having different aspect ratios, reinforcing amount and arrangement, and different boundary conditions. Data from these tests may be used to

, evaluate concrete shear wall behavior.

).

)

The Japanese have carried out a number of experimental investigations related to nuclear power plant buildings. The main purposes of these investigations appear to be to develop design criteria and to show that 4-5

l structures do not fail at loads up to the design level. An attempt is being made to establish a technical information exchange agreement with Japan and obtain data from these and other tests that may possibly provide information on structural failure.

3. Products

a. The sensitivity of structural behavior (i.e., fundamental frequency; damping; and response spectra, acceleration, and displacements at various floor levels) to changes in internal and external wall configurations, and increased earthquake magnitude beyond the initial design basis (1986-1988). (Supports Seismic Risk and Margins Issues)

b. Failure mode and failure level data for various concrete (Supports Seismic Risk and Margins configurations (1986-1988).

Issues)

c. Experimental validation of the assumptions and analytical methods currently used by applicants / licensees to predict the fundamental natural frequency, accelerations, and displacements of structures (1986-1988). (Supports Seismic Margins Issue)

d. Revisions, as necessary to Regulatory Guides 1.61, " Damping Values for Seismic Design of Nuclear Power Plants"; 1.92, " Combining Modal Responses and Spatial Components in Seismic Response Analysis"; and 1.122, " Development of Floor Design Response Spectra for Seismic Design of Floor-Supported Equipment or Components" (1988). (Supports Seismic Risk and Margins Issues) 4.4 Containment Buckling project

1. Scope In general, operating nuclear power plants with steel containments were designed to withstand the effects of a prescribed SSE without developing severe shell buckles. However, if the containment should be subjected to a seismic event appreciably greater than the SSE, there is concern that the containment could become unstable and large buckles could develop. Such buckles could result in severe local or general structural distortions. Such severe distortions could mean severe dislocations in a of primary safety system attachments, which in turn could result degradation of the ability of the safety systems to perform their mission.

Also, shell buckles could result in the cracking of the containment or the opening of penetration seals thereby inducing leakage paths throughThere the wall that would allow contained gases to escape to the atmosphere.

are large uncertainties in the current methods used to predict euckling loads and locations.

The Containment Buckling Project will assess the adequacy of analytical techniques used to determine the capability of steel containment structures to resist dynamic buckling when subjected to severe seismic loads. The current analytical techniques involvo a freeze-in-time concept to account for the dynamic buckling motions involved. This 4-6

research will also develop physical benchmark experiments that can be used to verify the adequacy of specific computational shell buckling codes.

Also, the extent of postbuckling distortions of containment shells when subjected to time-dependent seismic forcing functions will be investigated.

2. Integration and Relationship with Other Research This dynamic buckling program is a cooperative program between the NRC and the Pressure Vessel Research Committee (PVRC). Arrangements are currently being made between our contractor (Los Alamos National Laboratory (LANL)) and the PVRC to cooperatively fund experiments to be conducted on steel model containment shells in FY 1986. A number of industrial organizations will participate in the PVRC side of the program.

3. Products

a. .Results of research on the localized buckling of containment shells in the region of penetration framing to provide NRR with information to judge the adequacy of the framing used to reinforce cutouts in nuclear power plant steel containments, to serve as benchmarks to evaluate methods of localized shell stability analysis, and to judge the adequacy of engineering design codes related to shell buckling (1985). (Supports Seismic Margins Issue)

b. Results of the dynamic buckling research program to be used by NRR to judge the adequacy of the freeze-in-time technique currently being used to analyze the dynamic buckling of shell structures and the postbuckling test results to provide NRR with an insight on the deformations that may be expected within buckled steel containments durin extreme seismic events (1986). (Supports Seismic Margins Issue 4.5 Load Combinations for Structural Design Project l 1. Scope Structures for nuclear power plant facilities must be designed to withstand combinations of loads (earthquake, temperature, pressure, etc.)

that may be expected to occur during the plant lifetime. There are l

uncertainties integrity. The as to how these loads are combined to ensure structural i traditional methods of structural design attempt to account for the inevitable variability in loads, material strengths, i l

' inservice environments, fabrication process, etc., through the use of l factors that increase the loads and reduce the allowable stresses.

These approaches may result in an unknown and nonuniform reliability because of the subjective manner in which these factors have been deter-mined. In addition, the stochastic nature of the loads and the variations in material properties for nuclear seismic Category I structures are better characterized by a probabilistic approach for a rational assessment of structural safety and performance.

l l

4-7 L

The objective of this research program is to develop a rational approach, based on probabilistic considerations, for the safety evalua-tion of seismic Category I structures subjected to multiple static and dynamic loadings. A load combination methodology for the analysis of seismic Category I structures in the elastic and inelastic response ranges will be established. This improved methodology will be used to evaluate the structural adequacy of operating plants whose design loads have been increased. These plants may have sufficient margins to withstand new loads because conservative load factors were used in their original design.

2. Integration and Relationship with Other Research The methodology used in this project is similar to and integrated with the methodologies used in the NRR-supported studies on Probabilistic Response and Response Spectra for Category I Structures and on Fragility Analysis for Concrete Structures. Structural behavior and failure data from the Structural Response Project (Appendix C) will also be integrated into the methodology being developed.

3. Product Probabilistic-based procedures to establish load combinations and load factors for the analysis of seismic Category I structures and provide a basis for changing the Standard Review Plan and ASME Boiler and Pressure Vessel Code criteria (1987). (Supports Seismic Risk and Margins Issues).

4.6 Piping Reliability Projects (Appendix D)

1. Scope The majority of the seismic needs and objectives adoressed by the piping research described in Appendix D are concerned with procedures and criteria within the design envelope, i.e., at or below the SSE level. The principal objective of the research described in Appendix 0 is to provide the basis for improved balance between operating and accident conditions in terms of overall safety. This is to be done by better defining the response and failure behavior from various design loadings (particularly dynamic loads) and by making integrated assessments on how these loadings and their associated design criteria affect the overall reliability and sa fety of piping systems. Thermal effects must be explicitly considered.

The NRC/EPRI program included in Appendix D will provide data to assess the adequacy of the assumptions used in PRAs related to piping.

Current pipe failure models used in seismic PRAs are necessarily simple and need to be evaluated to determine if there is any significant error.

A systematic bias could affect the probability estimations of both initiating events and the loss of mitigating systems that are used in seismic risk studies. In addition, the piping research described in Appendix D will support the seismic margins objectives by reducing the uncertainty in determining the response and failure modes of piping subjected to earthquake levels above the SSE.

4-8

2. Integration and Relationship with Other Research The Seismic Safety Margins Research Program (SSMRP) and the Structural Response Project (Appendix C) will supply data on structural stiffness, changes in floor response spectra, and camping uncertainties to be used in the seismic spectra studies. The NRC Degraded Piping Program will provide data on the dynamic capacity of cracked piping. These data will be factored into the planning for future capacity testing of piping systems.

The Validation of Seismic Calculational Methods Project (Appendix F) will provide nonlinear piping response data from the PDR and other tests.

Information regarding nonlinear piping response and failure levels and modes will serve as input to the NRC Seismic Design Margins Project (Appendix G) and the Validation of Seismic Calculational Methods Project (Appendix F).

The pipe capacity tests will be done in cooperation with EPRI. EPRI is developing projects regarding snubber elimination and nonlinear piping efforts. prediction that will be coordinated with NRC piping research response

3. Products

a. Failure data from piping capacity tests to be used as a basis for changing ASME Boiler and Pressure Vessel Code criteria and for improving current seismic PRA fr'a gility estimates (1986-1988).

(Supports Seismic Risk and Margins Issues)

b. Recommendations for high-frequency damping values to be used in piping analysis and revising Regulatory Guide 1.61, " Damping Values for Seismic Design of Nuclear Power Plants" (1986). (Supports Seismic Margins Issue)

c. Improved guidance on nozzle boundary conditions to be used as a basis for changing ASME Boiler and Pressure Vessel Code criteria and the NRC Standard Review Plan (1985). (Supports Seismic Margins Issue)

d. Recommendations for improving piping input spectra defined in Regulatory Guide 1.122, " Development of Floor Design Response Spectra for Seismic Design of Floor-Supported Equipment or Components" (1986). (Supports Seismic Margins Issue) e.

t

! Recommendations for improving NRC Standard Review Plan piping response combination methods (1986). (Supports Seismic Margins Issue) f.

Recommendations for simplified methods to account for inelastic piping response to assist the regulatory staff in evaluating the continued operation of plants subjected to loads beyond the design basis (1986). (Supports Seismic Margins Issue) 4-9 L-

g. Intergranular stress-corrosion cracking (IGSCC) study in BWR primary piping using the PRAISE code to assist the regulatory staff in-evaluating the need to replace piping in operating plants (1988).

4.7 Component Fragility Project (Appendix E)

1. Scope This research supports the need for reliable inputs for PRAs and margin studies. Current fragility estimates depend on military data (which do not characterize seismic environments well), extrapolation from design calculations, and subjective judgments. This research seeks to test the hypothesis that electrical and mechanical components have higher failure levels than those presently assumed in seismic PRAs and, as a consequence, that the significance of the earthquake threat may be diminished in licensing decisionmaking. This research will contribute to the development of simplified seismic risk methodologies by eliminating certain branches on event trees and fault trees that do not contribute to risk. These branches can be identified when the actual component fragilities are shown to be extremely large in comparison with predicted seismic responses. Realistic component fragilities are a prerequisite to validation of current seismic PRA methods.

This research will also provide reliable component fragility data to reduce uncertainties in estimating failure modes and levels for equipment.

These data are particularly useful in evaluating older plants since aging effects are explicitly treated in the derivation of mechanical and electrical ccmponent fragilities.

2. Integration and Relationship with Other Research Seismic inputs to mechanical and electrical components are required from other programs. For earthquake levels below the SSE, much of this intrmation already exists. However, this information cannot be projected to " fragilities," which implies that testing of components is conducted to failure levels (many times greater than SSE levels). The reat,on for this is that the design procedure is to increase linearly the amplitude of motion and to assume that frequency characteristics do not change. It is widely recognized that earthquake excitations beyond the design basis will produce nonlinear effects in soils, structures, and systems. This leads to a general decrease in input frequencies that are not accounted for in present fragility testing. In summary, the component fragilities program looks to the Structural Response Project (Appendix C) and the Piping Peliability Projects (Appendix D) for information on appropriate frequency contents to input motions when seismic motions are much greater than the design basis (several times the SSE). This research is also coordinated with the Environmental Qualification of Mechanical and the Dynamic Qualification of Electrical and Mechanical Equipment Program (E0QP).

l I

4-10

The component fragility project provides essential information to the NRC. Seismic Design Margins Project (Appendix G) and the Validation of Seismic Calculational Methods Project (Appendix F). Both these projects make comparisons between predicted and failure response levels. Component fragilities, with respect to electrical and mechanical equipment, identify failure modes and specific failure levels. Because the present data base comes from subjective judgment and military data, it represents a particularly weak link- for both seismic margins and seismic PRA endeavors. .This tends to emphasize the importance of component fragilities and the need to expedite the work.

This .research effort is coordinated with EPRI through a proposed cooperative agreement on data acquisition for component fragility and generic qualification.

3. Products

a. Proceedings of the Workshop on Component Fragility (1985). (Supports Seismic Risk and Margins Issues)

b. fragilities used in current Improve seismic PRAsunderstanding (1985). (of component Supports Seismic Risk and Margins Issues)

c. Results of the Phase I demonstration tests wherein standard procedures for component fragilities testin

.(Supports Seismic Risk and Margins Issues) g are established (1985).

d. Test plans and test results for fragility tests of selected components (1986). (Supports Seismic Risk and Margins Issues)

e. Improved component fragilities of electrical and mechanical components installed in nuclear power plants to be used by the licensing staff to assess seismic PRA submittals and assist in seismic design margins decisions (1986). (Supports Seismic Risk and MarginsIssues) 4.8 SSMRP-BWR Risk Assessment Project

1. _ Scope The major project objective is to develop a simplified BWR seismic risk methodology to assess the probability of core melt and/or radioactive release by adding BWR-specific features to the existing simplified SSMRP PWR methodology. In addition, major differences between the seismic risk of the Zion Unit 1 PWR and the LaSalle County Station Unit 2 BWR will be identified.

The simplified seismic risk methodology developed in the SSMRP was demonstrated by aplication to Zion Unit 1, a PWR plant. Thus, all the systems analysis models (event trees and fault trees) and all the structural and ' pfping trodels were developed for a PWR plant. The 4-11

I methodology developed in the SSMRP is being expanded to include a BWR so that licensee-submitted PRAs (both BWR and PWR) can be assessed against the NRC-sponsored benchmark.

2. Integration and Relationship with Other Research This research is providing the seismic risk methodology to a larger RES program, the Risk Methodology Integration and Evaluation Program (RMIEP), which will assess the risk from internally and externally initiated events. This research will also provide insights on the strengths and weaknesses of PWRs and BWRs and support studies conducted in the NRC Seismic Design Margins Project (Appendix G). The SSMRP and other seismic risk methodologies will be validated by the Validation of Seismic Calculational Methods Project (Appendix F) so that they may be used with confidence and credibility in the regulatory decision process.

3. Product Summary reports evaluating the techniques used to model seismic hazard and seismic response; comparison of these research results with available data; and experience applying these research results to the Zion Unit 1 PWR and the LaSalle County Station Unit 2 BWR (1986). (Supports Seismic Risk and Margins Issues) 4.9 Validation of Seismic Calculational Methods Project (Appendix F)

1. Scope Seismic PRA methods have been applied to clarify safety issues for nuclear power plants since seismic events can simultaneously affect many plant systems and therefore can be a significant or even dominant contributor to overall risk. The randomness of the seismic hazard, the uncertainties and variabilities of fragility and response data, and the approximate nature of the methodology raise questions of credibility with respect to the results of seismic PRAs and subsequent regulatory actions.

While the ultimate answer to these questions depends on the intended use of seismic PRAs, it is nevertheless necessary to validate the methodologies so they may be used with confidence and credibility in the regulatory decision process. The objective of validation research is to obtain information that can be used by NRC to develop acceptance criteria for predicting the behavior of nuclear power plants subjected to large earthquakes and thus improve the regulatory process. The predictive methods to be validated are used in both probabilistic and deterministic predictions.

The fundamental strategy is to engage in cooperative research programs in order to maximize available resources. Three efforts have

been developed

a. Participation in an SSI experiment being performed near Lotung, Taiwan, by EPRI.

i l

4-12 L

b. Participation in the Phase II experiments being performed at the HDR facility in Kahl, FRG, by Kernforschungszentrum Karlsruhe (KfK).

c. Possible participation in tests to be performed on the large shaker table in Tadotsu, Japan, by the Nuclear Power Engineering Test Center (NUPEC).

This research will verify the adequacy of the assumptions and subjective information used in PRAs. It also supports the seismic margins issue by providing experimental data to improve and reduce uncertainties in the current seismic design criteria. The experimental data will reduce uncertainties in determining the response and failure modes of structures, equipment, and piping subjected to earthquake levels above the SSE.

2. Integration and Relationship with Other Research This research will use experimental data generated as part of the Structural Response Project (Appendix C) and the Piping Reliability Projects (Appendix D). Results of the validation effort will be used in the NRC Seismic Design Margins Project (Appendix G).

a. EPRI Cooperation Argonne National Laboratory is the NRC coordinator for this effort.

EPRI is constructing, in a seismically active area in Taiwan, a model about 1/4 the size of a concrete containment.. EPRI will also install instruments in the model and in vertical and horizontal arrays in the vicinity of the model and will record earthquake responses over a 5-year period. NRC has agreed to perform low-level vibratory tests of the

, model to provide baseline data on modal parameters. Future NRC effort will be in providing analytical models to predict SSI effects for the

! recorded earthquakes and in comparing predictions with observations.

b. HDR Cooperation Argonne National Laboratory is also the NRC coordinator for this effort. The HDR facility is a modification of a superheated steam reactor that was decommissioned and modified for research in 1973. A series of experiments (called Phase I tests) was conducted from 1975 to 1983 involving experiments on materials engineering, thermal hydraulics, and mechanical and earthquake engineering. The primary focus of the experiments was comparison of predictions by analytical models with experiments. The Phase II experiments, from 1984 to 1988, are similarly motivated and will examine higher levels of response where damage to structures, systems, and components is expected. NRC participation in the seismic tests will involve providing predictions for the response of structures and piping systems excited by shakers. One series of experiments in which the containment building will be excited by a large shaker is planned for June-July 1986. Another series in which piping systems will be excited into the inelastic range by servo-hydraulic shakers is planned for May-June 1987.

4-13

c. NUPEC Cooperation Brookhaven National Laboratory is the NRC coordinator for this effort. A massive testing effort was started in 1974 under the sponsorship of the Japanese Ministry for International Trade and Industry (MITI). The Nuclear Power Engineering Test Center was established and the largest shake table facility in the world was constructed at the Tadotsu Engineering Laboratory. The table is 15m x 15m with a capacity of 1,000 tons. (For purposes of comparison, the largest shake table in the U.S.,

operated by the Richmond Field Station of the University of California, is 6m x 6m with a capacity of 60 tons.) The test series, from 1982 to 1988, will involve eight specimens representing containment vessels, primary loops, reactor pressure vessels, and reactor internals for both PWRs and BWRs. All specimens will be excited with time histories representative of the Japanese design earthquakes, designated S, and S., and similar in intent to our Operating Basis Earthquake (0BE) and SSE. Responses to those motions will be monitored. The series is known officially as the Seismic Proving Test Programe. The total cost of the program exceeds

$500M, including about $250M for the test facility.

NRC's main interest lies in determining the ability of analytical methods to predict the onset of component damage under very large earthquake motions. To that end, we are negotiating for tests to be performed af ter the " proving tests" have been completed. The tests in which NRC may cooperate will involve increasing the excitation within the limits of the table and possibly modifying the specimens if necessary to induce inelastic response.

3. Products

a. Nonlinear response data from HDR shake tests (1986-1987). (Supports Seismic Risk and Margins Issues)

b. RMIEP's evaluation of the SSMRP's system analysis methods (1986).

(Supports Seismic Risk and Margins Issues)

c. Earthquake response data from EPRI's scale-model containment in Taiwan (1986-1990). (Supports Seismic Risk and Margins Issues) 4.10 NRC Seismic Design Margins Project (AppendixG)

1. Scope The Seismic Design Margins Project (SDMP) discussed in Appendix G provides the technical basis for assessing the significance of design margins in terms of overall plant safety and will identify any potential weaknesses that might have to be addressed. In conjunction with past studies and ongoing validation and fragility efforts, the SDMP should be effective in resolving questions about the ability of plants to withstand large earthquakes.

The SDMP has three phases. The first phase is an intensive effort of about 6 months duration leading to a preliminary assessment of margin 4-14

adequacy and a set of trial guidelines to determine seismic margins in existing plants. In Phase II trial reviews of two plants will be accomplished. Phase III continues with further plant reviews and studies dependent on the results of Phases I and II.

a. Phase I: Methodology Development A panel of consultants who have expert knowledge in seismic design and risk. assessment will:

(1) Assess existing seismic margins information (seismic PRAs, fragility data, etc.), and critique the state of the art of this information.

(2) Make a preliminary judgment as to existing margins in terms of what earthquake levels compromise plant safety and which components, systems, functions, etc., contribute to those compromises at the earthquake levels of importance.

(3) Make a preliminary identification of those generic attributes that are important contributors to plant strength and those that appear to contain important vulnerabilities against earthquakes. This will be done for broad groups of plants, and/or broad groups of functions, and/or broad groups of systems and components. A statement of attributes and vulnerabilities that are clearly nongeneric will be made.

(4) Summarize the results of Tasks (1) through (3) and recommend to NRC a set of " adequacy criteria" whereby NRC might reach a judgment on the adequacy of seismic design margins for specific groups of plants and/or functions and/or systems.

b. Phase II: Conduct Trial Plant Reviews The NRC Working Gioup on Seismic Design Margins will review the expert panel report, interact with the expert panel, and make their own recommendations on the extent of effort and course of action for future generic and plant-specific seismic margins reviews. Upon completion of the NRC Working Group's recommendations, screening guidelines for plant seismic margins review will be developed by the expert panel. Two trial plant reviews will be conducted to demonstrate the use of the screening guidelines and to serve as benchmarks for possible future licensing reviews. Plant selection will be made by the NRC Working Group. The reviews will be conducted by one or two contractors, with the expert panel serving as an advisory group. The expert panel will assess the results of the two trial plant reviews and make as-needed improvements to the plant review guidelines,

c. Phase III: Implementation of Plant-Specific Reviews and Continuing Studies At completion of the first two phases of this research, it is anticipated that further work will be defined, possibly including some 4-15

further plant reviews. Althcugh it may be necessary to do a rough assessnent of all plants, detailed reviews on a representative set should be sufficient. Phase III consists of this additional work, including the implementation o( additional plar.t reviews, as necessary, over a 2-year period. Details will be determined during Phases I and II. This work might include further guideline development, risk assessment, or further review of those existing risk assessments, including some requantification and similar work to that done in Phase I.

2. I n t e g ra t i o n a n d Re l a,t i o,n,s h ip, ,w i,t h, ,0,t,h,e r, ,R,e,s,e a,r,c h The results from the Seismotectonic Program (Appendix A) will redefine, as appropriate, the seismic hazard at plant sites. Of particular importance is the RES Seismic Hazard Characterization Project and the future NRR-planned " production runs" using this methodology.

The program will make use of results and insights from industry and NRC-sponsored seismic PRAs. NRC programs dealing with seismic PRA include the SSMRP, the SSMRP-BWR Risk Assessment Project, and the Validation of Seismic Calculational Methods Project (Appendix F). In addition, research adaressing seismic response and failure levels and

n. odes will serve to better define seismic design margins. Research in this area includes the Structural Response Project (Appendix C), the Piping Reliability Project (Appendix E .) Projects (Appendix D), and the Component Fragility The SDMP will develop a list of the research needs as an outgrowth of its review and use of seismic PRA and earthquake experience data. Areas currently thought to be important include fragility development, relay chatter, operator action (error), and design and construction errors.

These needs will most likely fit into the scopes of either the Validation of Seismic Calculational Methods Project (Appendix F) or research work to be addressed by the NRC RES Division of Risk Analyses & Operation.

EPRI's development of a seismic margins research plan is being coordinated with the NRC effort. In the near future, industry groups will be represented on the seismic design margins expert panel. Coordination withrelatedforeignresearchisbeing(conductedthroughtheValidationof Seismic Calculational Methods Project Appendix F).

Ten to fifteen seismic PRAs will have been performed on plants in the near term. These PRAs are usually ar, adjunct to an internal events PRA and are a likely source of information for the determination of the adequacy of seismic design margins in risk terms. These plants and their PRAs are also a likely source of information for the development of seismic design margins screening guidelines. Two of the seismic PRAs to be reviewed will be the SSMRP Zion and La Salle studies since these are the most comprehensive analyses and were conducted with NRC funds instead of utility funds. Thus, information from these studies is not only the most detailed but also the most readily available. Utility seismic PRAs such as were done on Millstone 3, Seabrook, Limerick, Oconee, and Indian Point 2 and 3 will also give useful insights.

4-16

3. Products

a. The expert panel's assessment of the adequacy of seismic margins (Spring of 1985). The NRC will review this to help determine the need for and extent of future licensing reviews. (Supports Seismic Margins Issue)

b. Initial screening guidelines to be used to assess plant seismic margins with a minimum of required effort (1985). (Supports Seismic Margins Issue)

c. Two trial plant reviews using these guidelines (to be initiated in 1986). These will serve as benchmarks for future licensing reviews.

Insights obtained during the trial application will help define what parts of the guidelines need improvement. (Supports Seismic Margins Issue)

d. guidelines to be used, as needed, in Final future seismic margins NRC licensing screening (1987),

reviews (Supports Seismic Margins Issue) i 4-17

5. GLOSSARY - Acronyms and Definitions t

, ACRS Advisory Committee on Reactor Safeguards ASME American Society of Mechanical Engineers  !

BWR Boiling water reactor Charleston Earthquake An earthquake occurring in 1886 in Charles-ton, South Carolina.

COE~-

-(U.S. Army) Corps of Engineers Degraded Pipe Program A program to develop the basic elastic-plastic fracture mechanics analysis methods to allow the structural integrity "

evaluation of inservice cracked nuclear piping.

EDQP Environmental Qualification of Mechanical-and the and Mecha_ Dynamic Qualification of Electrical nical Equipment Program - a program to develop criteria and methodologies to improve national standards or other regulatory documents used for qualifying electrical and mechanical equipment.

EPRI Electric Power Research Institute Event Tree Describes sequences of system failures which, if they fail in certain combinations, may lead to the release of radioactive material from the reactor to the environment.

Failure The inability to perform the intended

?

function.

Failure Level Magnitude of the input load (e.g., accelera-tion or displacement) causing failure. l

, Failure Mode Phenomenon causing failure, e.g., cracking, j' binding, or collapse.

Fault A tectonic structure along which i differential slippage of the adjacent earth i

materials has occurred parallel to the fracture plane. '

Fault Tree Describes the various ways by which a system j

can fail.

FEMA Federal Emergency Management Agency I. \

! si-1 i

~ Fragility The probability of failure at a given failure level.

FRG Federal Republic of Geraany HOR Heissdampfreaktor facility - a superheated steam reactor located in the Federal Republic of Germany that has been deconnissioned and modified for research.

Holocene Movement The current geological time period that began approximately 18,000 years ago.

IGSCC Intergranular stress-corrosion cracking.

Inelastic Response Plastic flow of material resulting in permanent deformation when the load is removed.

Initiating Event An occurrence that activates the safety systems of a nuclear power plant.

KfK Kernforschungszentrum Karlsruhe LANL Los Alamos National Laboratory Meers Fault A geological fault located in southwest Oklahoma MITI Ministry of International Trade and Industry, Japan.

NBS National Bureau of Standards New Brunswick A January 1982 earthquake occurring in New Brunswick, Canada.

Nonlinear Response Condition where the response of a component or structure is not proportional to load due to inelastic material behavior or geometry.

NRC Nuclear Regulatory Connission NRR (Officeof)NuclearReactorRegulation,NRC NSF National Science Foundation NTOL Near-term operating license NUPEC Nuclear Power Engineering Test Center, Japan OBE 0)erating Basis Earthquake - an earthquake t1at could reasonably be expected but not affect the operation of a nuclear plant during its life.

GI-2

PRA Probabilistic risk assessment PVRC Pressure Vessel Research Committee PWR Pressurized water reactor RES Office of Nuclear Regulatory Research, NRC Response Spectrum A plot of maximum responses (acceleration, velocity, or displacement) of a family of idealized single-degree-of-freedom damped oscillators against natural frequencies (or periods) of the oscillators to a specified vibratory motion input at their supports.

RMIEP Risk Methodology Integration and Evaluation Program SDMP Seismic Design Margins Project Seismic Pertaining to an earthquake SHCP Seismic Hazard Characterization Project -

the development and application of a seismic hazard characterization methodology for the entire region of the United States east of the Rocky Mountains.

Seismicity Relationship of the frequency and distribution of earthquakes.

SEP Systematic Evaluation Program - a program where 11 of the older operating plants were evaluated against the " intent" of the current licensing criteria for selected issues.

Soil Liquefaction Any total or near total loss of shear strength of the soil due to cyclic loading.

i SSE Safe Shutdown Earthquake - the earthquake producing the maximum vibratory ground

! motion for which certain structures, systems, and components are designed to remain functional.

SSI Soil-structure interaction SSMRP Seismic Safety Margins Research Program Stress Field Term used to describe forces acting on a geologic structure. l Gl-3 t

USGS United States Geological Survey 10 CFR Title 10 to the Code of Federal Regulations.

Title 10 -- Energy is composed of four volumes. The first volume, Parts 0-199, contains the regulations of the Nuclear Regulatory Comission.

GI-4

6. REFERENCES

1. Memorandum from H. R. Denton to R. B. Minogue,

Subject:

NRR Research Needs in Seismic Analysis Methodology (RR-NRR-82-2),

oated April 8, 1982,

2. Memorandum from H. R. Denton to R. B. Minogue,

Subject:

NRR Input for the Long-Range Research Plan (LRRP), FY 1986-1990, dated December 6, 1984.

3. Memorandum from J. J. Ray (ACRS Chairman) to N. J. Palladino,

Subject:

Quantification of Seismic Design Margins, dated January 11, 1983.

4. Memorandum from J. C. Ebersole (ACRS Chairman) to N. J.

Palladino,

Subject:

Quantification of Seismic Design Margins, dated January 18, 1984.

5. Memorandum from W. J. Dircks to J. C. Ebersole (ACRS Chairman),

Subject:

Quantification of Seismic Design Margins, dated April 12, 1984.

6. Memorandum from W. J. Dircks to N. J. Palladino,

Subject:

Quantification of Seismic Design Margins, dated April 12, 1984.

7. Letter from J. F. Devine, USGS, to R. E. Jackson, NRC, dated November 18, 1982.

8. SECY-82-53 (Policy Issue Paper),

Subject:

Possible Relocation of Design Controlling Earthquakes in the Eastern U.S., dated February 5, 1982.

9. M. P. Bohn et al., " Application of the SSMRP Methodology to the Seismic Risk at the Zion Nuclear Power Plant," Lawrence Livermore National Laboratory, Livermore, California, NUREG/CR-3428, January 1984.

10. Power Authority of the State of New York, " Indian Point Probability Safety Study," Vols. 1-12, 1982.

11. Commonwealth Edison Company, " Zion Probability Safety Study "

Pickard, Lowe and Garrick, 1983.

12. " Limerick Generating Station Severe Accident Risk Assessment,"

NUC Corporation, 1981. l t

13. " Millstone Unit 3 Probabilistic Safety Study" (Vols.1-12),

Northwest Utilities, August 1983.

14. "0conee PRA: A Probabilistic Risk Assessment of Oconee Unit 3."

Nuclear Safety Analysis Center, EPRI, June 1981.

1 J. Garrick et al., Seabrook Station PRA, PLG-0365, June 1984.

15.

R-1 l

l

l l

7. BIBLIOGRAPHY 7.1 Seismotectonic Program New Madrid Seismotectonic Study, Activities During Fiscal Year 1978, NUREG/CR-0450, October 1978.

Recent Vertical Movement of the Land Surface in the Lake County Uplift and Reelfoot Lake Basin Areas, Tennessee, Missouri and Kentucky, NUREG/CR-0874, June 1979.

Bedrock Geology of the Cape Ann Area, Massachusetts, l NUREG/CR-0881, September 1981.

Analysis of Faults in the Delaware Aqueduct Tunnel, Southeastern New York, NUREG/CR-0882, June 1979.

New England Seismotectonic Study Activities During Fiscal Year 1978, NUREG/CR-0939, September 1979.

An Integrated Geophysical and Geological Study of the Tectonic Framework of the 38th Parallel Lineament in the Vicinity of its Interaction with the Extension of the New Madrid Zone, NUREG/CR-1014, September 1979.

i Nemaha Uplift Seismotectonic Study, Regional Tectonics and Seismicity of Eastern Kansas, NUREG/CR-1144, November 1979.

Seismicity and Tectonic Relationships for Upper Great Lake Pre-cambrian Shield Provence.

NUREG/CR-1569, July 1980.

Geophysical Investigations of the Anna, Ohio Earthquake Zone Annual Progress Report, NUREG/CR-1649, September 1980.

Aeromagnetic Map of the East-Central Midcontinent of the United States, NUREG/CR-1662, October 1980.

Bouguer Gravity Anomaly Map of the East-Central Midcontinent of the United States, NUREG/CR-1663, October 1980.

Investigation of the McGregor-Saratoga-Ballston Lake Fault System East Central New York Final Report,

NUREG/CR-1866, August 1981.

l l Bib-1

Structural Framework of the Mississippi Embayment of Southern Illinois, NUREG/CR-1877, March 1981.

An Integrated Geophysical and Geological Study of the Tectonic Framework of the 38th Parallel Lineament in the Vicinity of its Intersection with the Extension of the New Madrid Fault Zone, NUREG/CR-1878, January 1981.

Investigations into the State of Stress, in the Crust Under Northeastern United States, NUREG/CR-2093, August 1981.

New Madrid Seismotectonic Study Activities During Fiscal Year 1980, NUREG/CR-2129, September 1981.

Brittle Deformation of the Manhattan Prong, NUREG/CR 2138, August 1981.

A Tectonic Study of the Extension of the New Madrid Fault Zone Near its Intersection with the 38th Parallel Lineament, NUREG/CR-2741, June 1982.

Earthquake Hazard Studies in New York State and Adjacent Areas Final Report, NUREG/CR-3079 January 1983.

7.2 Soil Response Project Current Methodologies for Assessing Seismically Induced Settlements in Soil, NUREG/CR-3380, August 1983.

7.3 Structural Response Project Seismic Response of Nonlinear Systems, NUREG/CR-2310, September 1981.

Margins to Failure - Category I Structures Program:

Background and Experimental Program Plan, NUREG/CR-2347, February 1982.

Engineering Characterization of Ground Motion, Vols. 1-5, NUREG/CR-3805 Vol. 1: Effects of Characterization of Free-Field Motion on Structrual Response, May 1984.

Vol. 2: Effects of 6round Motion Characteristics on Structural Response Considering Localized Structural Nonlinearities and Soil-Structure Interaction Effects, March 1985.

Vol. 3: Empirical Data on Spatial Variations of Earthquake Ground Motions.(tobepublished).

Bib-2

Vol. 4: Soil-Structure Interaction Effects on Structural Response, (to be published).

Vol. 5: Summary of Conclusions and Recomendations published). (to be 7.4 Containment Buckling Project An Investigation of Buckling of Steel Cylinders with Circular Cutouts Reinforced in Accordance with ASME Rules, NUREG/CR-2165, June 1981.

Buckling Investigations of Ring-Stiffened Cylinderical Shells Under Unsymetrical Axial Loads.

NUREG/CR-2966, October 1982.

Buckling Investigations of Ring-Stiffened Cylindrical Shells with Reinforced Openings under Unsymetrical Axial Loads, NUREG/CR-3135, February 1983.

Buckling of Steel Containment Shells under Time Dependent Loading, NUREG/CR-3742, May 1984.

7.5 Load Combinations for Structural Design Project Probability Based Load Criteria for Design of Nuclear Structures: A Critical Review of the State of the Art, NUREG/CR-1979, April 1981.

Tornado Damage Risk Assessment.

NUREG/CR-2944, February 1983.

Charact'erization of Earthquake Forces for Probability-Based Design of Nuclear Structures.

l NUREG/CR-2945, February 1983.

l First Excursion Problems for Gaussian Vector Processes, j NUREG/CR-3283, April 1983.

A Consensus Estimation Study of Nuclear Power Plant Structural Loads, l NUREG/CR-3315, May 1983.

Probabilistic Description of Resistance of Safety Related Nuclear Structures.

NUREG/CR-3341, May 1983.

Probabilistic Models for Operational and Accidental Loads on Seismic Category I Structures, NUREG/CR-3342, December 1983.

Probability Based Safety Checking of Nuclear Plant Structures.

NUREG/CR-3628, May 1984.

i Bib-3 U

I Reliability Assessment of Indian Point Unit 3 Containment Structure, NUREG/CR-3641, January 1984.

Probability Based Load Combination Criteria for Design of Concrete Containment Structures, NUREG/CR-3876, (to be published).

7.6 Piping Reliability Projects Probability of Pipe Fracture in the Primary Coolant Loop of a PWR Plant, Vols. 1-9.

NUREG/CR-2189 Vol. 1: Summary, September 1981.

Vol. 2: Primary Coolant Loop Model, September 1981.

Vol. 3: Non-Seismic Stress Analysis, August 1981.

Vol. 4: Seismic Response Analysis, September 1981.

Vol. 5: Probabilistic Fracture Mechanics Analysis, August 1981.

Vol. 6: Failure Mode Analysis, September 1981.

Vol. 7: System Failure Probability Analysis, September 1981.

Vol. 8: Pipe Fracture Indirectly Induced by an Earthquake, September 1981.

Vol. 9: PRAISE Computer Code User's Manual, August 1981.

A Survey of Experimentally Determined Damping Values in Nuclear Power Plants.

NUREG/CR-2406, November 1981.

Parameters That Influence Damping in Nuclear Power Plant Piping Systems, NUREG/CR-3022, November 1982.

Pipe Damping Studies and Nonlinear Pipe Benchmarks from Snapback Tests at the Heissdampfreaktor, NUREG/CR-3180, July 1983.

Comparisons of ASME Code Fatigue Evaluation Methods for Nuclear Class 1 Piping with Class 2 or 3 Piping, huREG/CR-3243, June 1983.

In Situ and Laboratory Benchmarking of Computer Codes Used for Dynamic Response Predictions of Nuclear Reactor Piping, NUREG/CR-3340, May 1983.

81b-4 l

l Impact of Changes ii Damping and Spectrum Peak Broadening on the l Seismic Response of Piping Systems, l

NUREG/CR-3526, March 1984.

Sources of Uncertainty in the Calculations of Loads on Supports of

! Piping Systems, NUREG/CR-3599, June 1984.

l Probability of Pipe Failure in the Reactor Coolant Loops of Westinghouse PWR Plants Vols. 1-4, NUREG/CR-3660

Vol. 1

Summary (to be published).

Vol. 2: Pipe Failure Induced by Crack Growth, August 1984.

Vol. 3: Guillotine Break Indirectly Induced by Earthquake, February 1985.

l Vol. 4: Pipe Failure Induced by Crack Growth, West Coast Plants (to be published).

l Probablity of Pipe Failure in the Reactor Coolant Loops of Combustion Engineering PWR Plants, Vols. 1-3, NUREG/CR-3663 l Vol. 1: Summary, January 1985.

Vol. 2: Pipe Failure Induced by Crack Growth, September 1984.

l Vol. 3: Double-Ended Guillotine Break Indirectly Induced by Earth-quakes, January 1985, i Reliability Analysis of Stiff Versus Flexible Piping - Status Report l NUREG/CR-3718, April 1984. ,

Prediction and Experiment Comparisons for German Standard Problem 4A:

Piping Response to Blowdown, NUREG/CR-3720, March 1984.

Damping Test Results for Straight Sections of 3-inch and 8-inch Unpressurized Pipes, l NUREG/CR-3722, April 1984.

Alternate Procedures for Seismic Analysis of Multiply Supported Piping Systems, NUREG/CR-3811, August 1984.

Preloading of Bolted Connections in Nuclear Reactor Component Supports.

l NUREG/CR-3853, October 1984. l l

l Bib-5

Laboratory Studies: Dynamic Response of Prototypical Piping Systems, NUREG/CR-3893, August 1984.

Tests to Determine How Support Type and Excitation Source Influence Pipe Damping, NUREG/CR-3942, September 1984.

Case Study of the Propagation of a Small Flaw Under PWR Loading Conditions and Comparison with the ASME Code Design Life - Comparison of ASME Code Sections III and XI, NUREG/CR-3982, November 1984.

Response Margins of the Dynamic Analysis of Piping Systems, NUREG/CR-3996, October 1984.

7.7 Component Fragility Project A Research Program for Seismic Qualification of Nuclear Plant Electrical and Mechanical Equipment, Vols. 1-4 NUREG/CR-3892, August 1984.

Vol. 1: Survey of Methods for Equipment and Components Evaluation of Methodology, Qualification Methodology for Line Mounted Equipment.

Vol. 2: Correlation of Pethodologies for Seismic Qualification Tests of Nuclear Plant Equipment.

Vol. 3: Recommendations for Improvement of Equipment Qualification Methodology and Criteria.

Vol. 4: The Use of Fragility in Seismic Design of Nuclear Plant Equipment.

7.8 Seismic Safety Margins Research Program Structural Building Response Review, Vols. I and II.

NUREG/CR-1423, Vol. I February 1980 Vol. II May 1980 Regional Relationships Among Earthquake Magnitude Scales, NUREG/CR-1457, September 1980.

Best Estimate Method vs. Evaluation Method: A Comparison of Two Techniques in Evaluating Seismic Analysis and Design, NUREG/CR-1489, July 1980.

Structural Uncertainty in Seismic Risk Analysis, NUREG/CR-1560, October 1980.

Seismic Hazard Analysis, Vols. I to V, NUREG/CR-1582.

Vol. I: Overview and Executive Summary, April 1983.

Bib-6

e Vol. II: A methodology for the Eastern U.S., August 1980.

Vol. III: Solicitation of Expert Opinion, August 1980.

l t

Vol. IV: Application of Methodology, Results and Sensitivity t

l Studies, October 1981.

l Vol. V: Review Panel, Ground Motion Panel and Feedback Results, October 1981.

The . Effects of Regional Variation of Seismic Wave Attenuation ' on Strong Ground Motion from Earthquake, '

huREG/CR-1655, October 1981.

L

' Compilation, Assessment and Expansion of the Strong Earthquake Motion Data Base, NUREG/CR-1660, September 1980.

l' i

Variability of Dynamic Characteristics of Nuclear Power Plant i Structures, 1 NUREG/CR-1661,' September 1980. '

Subsystem Response Review, i NUREG/CR-1700,. January 1981.

Specifications of Computational Approach,.

NUREG/CR-1701, January 1981.

Specifications of Computational Approach, NUREG/CR-1702, January 1981.

Preliminary Failure Mode Predictions for the SSMRP Reference Plant (Zion 1),

NUREG/CR-1703, January 1981.

Potential Seismic Structural Failure Modes Associated -with the Zion Nuclear Plant.

NUREG/CR-1704, March 1981.

Plant / Site Selection Assessment Report, NUREG/CR-1705, January 1981.

Subsystem Response Review, NUREG/CR-1706, July 1981.

Interim Report on Systematic Errors in Nuclear Power Plants.

NUREG/CR-1722, October 1980.

l ARMA'Models for Earthquake Ground Motions.

NUREG/CR-1751, February 1981.

, Simulating and Analyzing Artifical Non-Stationary Earthquake Ground Motions, NUREG/CR-1752, November 1980.

Bib-7 l

Soil-Structure Interaction: The Status of Current Analysis Methods and Research, NUREG/CR-1780, January 1981.

SSMRP Phase I Final Report, Vols. I to 10, NUREG/CR-2015.

Vol. 1: Overview, April 1981.

Vol. 2: Plant / Site Selection and Data Collection (Project I), July 1981.

Vol. 3: Development of Seismic Input (Project II), January 1983.

Vol. 4: Soil-Structure Interaction (Project III), June 1982.

Vol. 5: Major Structure Response (Project IV), August 1981.

Vol. 6: SubsystemResponse(ProjectV), October 1981.

Vol. 7: Deleted.

Vol. 8: Systems Analysis (Project VII), September 1984.

Vol. 9: SMACS - Seismic Methodology Analysis Chain With Statistics (ProjectVIII), September 1981.

Vol. 10: The Uses of Subjective Input, July 1981.

Uncertainties in Soil-Structure Interaction Analysis Arising from Differences in Analytical Techniques, NUREG/CR-2077, November 1983.

Ranking of Sources of Uncertainty in the SSMRP Seismic Methodology Chain, NUREG/CR-209E, August 1981.

Scaling and Estimation of Earthquake Ground Motion as a Function of the Earthquake Source Parameters and Distance, NUREG/CR-2103, June 1981.

Seismic Structural Fragility Investigation of the Zion Nuclear Power Plant, NUREG/CR-2320, October 1981.

4 Subsystem Fragility, NUREG/CR-2405, February 1982.

Application of the SSMRP Methodology to the Seismic Risk at the Zion Nuclear Power Plant, NUREG/CR-3428, January 1984.

SSI Sensitivity Studies and Model Improvement for the USNRC Seismic Safety Margins Research Program, NUREG/CR-4018, November 1984.

Bib-8 l

g-t\

Seismic Fragility; of ' Reinforced Concrete Structures and Components for Application to Nuclear Facilities, NUREG/CR-4123, March 1985.

7.9 Validation of Seismic Calculational Methods Project Review of Current Analysis Methodology for Reinforced Concrete Structural Evaluations, 4 NUREG/CR-3284, April 1983. '

7.10 NRC Seismic Design Margins Project None to date.

i 1

?

t

/

1 t 9

1 h

s

/

/

o Bib-9

E Appendix A Seismotectonic Program NRC Project Manager: Andrew Murphy

SEISM 0 TECTONIC PROGRAM EXECUTIVE

SUMMARY

Purpose Seismic hazards are significant contributors to the overall hazard at some nuclear power plants. Estimation of these hazards is a very significant factor in siting of nuclear power plants and evaluating the safety of existing power plants. There is considerable uncertainty in estimating the seismic hazard particularly in the Eastern United States. The objectives of the Earth Sciences Branch (ESB) seismotectonic research are to quantify and reduce the uncertainty in seismic hazard assessment and to develop methods of dealing with uncertainties.

The purpose of the Seismotectonic Program Plan (STPP) is to describe: (1) the technical and regulatory issues generated by the uncertainty in seismic hazard assessment; (2) the ESB research program designed to address the issues; and (3) the relationship of the ESB research to other research programs within NRC, within the government agencies, and in the private sector.

Issues The fundamental regulatory issue addressed by STPP is the seismic hazard assess-ments used in estimating seismic risk to nuclear power plants. This issue is specifically addressed by confirmatory research in support of the data and analysis techniques used for lice sing decisions and by research in support of regulatory guidance and possible revision of the current regulation, Appendix A to 10 CFR Part 100.

The recent (November 1982) U.S. Geological Survey (USGS) clarification of its position on the 1886 Charleston earthquake raised a regulatory issue that needs a timely response. The USGS notes that it has not been able to associate the Charleston earthquake with a geologic structure and that there is a probability, albeit very low, that the level of ground motion associated with a Charleston-sized earthquake could occur elsewhere on the eastern seaboard. This presents a problem. Under existing seismic siting criteria, Appendix A, to 10 CFR 100, the ground motions produced by an earthquake not associated with a specific structure are postulated to occur at any location within the tectonic province that it occurred in. If a nuclear power plant is within that tectonic province then the earthquake is postulated to occur at the plant site. If they are in different provinces, the ground motions are postulated to occur on the earth-quake's province boundary at the point closest to the plant.

If the 1886 Charleston earthquake cannot be correlated with a structure then it could be considered to occur anywhere within the tectonic province. Because j the tectonic provinces in the East are large, this could result in the postu-  !

lation of ground motions as large as those of the Charleston earthquake at almost j every reactor site in the eastern seaboard. This could have very troublesome  ;

implications for the licensing process.

The Geosciences Branch, NRR, has developed an interim licensing position and an action plan to address the Charleston issue. The action plan has t'een transmit-ted to DHSWM via a memorandum dated March 22, 1983 from Vollmer to Arsenault with a response dated May 20, 1983. (The Division of Health Siting and Waste Management, has been reorganized into the Division of Radiation Programs and Earth Sciences).

A-1

Program Plan There are tnree issues that contribute to the uncertainty in seismic hazard assessment; the program plan is organized around them. Those issues are: the uncertainty in establishing seismic source zones; the uncertainty in the propa-gation of seismic energy; and the uncertainty in the site specific response of Regional the site including soil failure. The program consists of three parts:

programs that address the uncertainty in seismic source zone configurations and seismic energy propagation; topical programs which deal with developing site specific spectra, strong ground motion models, and soil failure models; and probabilistic programs which deal with techniques to handle the uncertainties.

The regional program consists of seismological and geologic / geophysical programs in the Southeast, the Northeast, the New Madrid-Anna Ohio area, the Nemaha Up-lift area and the Pacific Northwest. The seismic networks define the zones of active seismicity and the basic seismological parameters of t' at activity. The geologic / geophysical investigations attempt to verify the proposed seismic zonations and seismogenic mechanisms. This is necessary because of the short seismic history that we have in the United States.

The topical programs include analysis of strong motion data from the eastern U.S. and analysis of seismic energy propagation. Validation of soil failure models and the feasibility of utilizing data from underground nuclear explosions will be examined. The probabilistic programs include a study of the sensitivity of predicted seismic ground motion to variations in models of seismic source zones. A previous study utilizing expert opinion is now being improved and expanded for use in resolution of the Charleston problem.

i A-2 l

CONTENTS INTRODUCTION 1.0 TECHNICAL ISSUES 1.1 Hazard Uncertainties 1.1.1 Regional Issues 1.1.1.1 Southeastern United States 1.1.1.2 Northeastern United States 1.1.1.3 New Madrid Region 1.1.1.4 Nemaha Region 1.1.1. 5 Pacific Northwest Region 1.1.2 Topical Issues 1.1.2.1 Source Parameters 1.1.2.2 Propagation / Attenuation Characteristics 1.1.2.3 Site Response Characteristics 1.1.2.4 Soil Failure 1.2 Handling Uncertainties 2.0 PROGRAM PLAN 2.1 Estimation and Reduction of Uncertainty 2.1.1 Regional Programs 2.1.1.1 Southeastern United States 2.1.1.2 Northeastern United States 2.1.1.3 New Madrid Region 2.1.1.4 Nemaha Region 2.1.1.5 Pacific Northwest Region 2.1.2 Topical Programs 2.1.2.1 Source Parameters 2.1.2.2 Propagation / Attenuation Characteristics 2.1.2.3 Site Response Characteristics 2.1.2.4 Soil Failure 2.2 Program to Address Uncertainties 2.2.1 Seismic Hazard Characteristics of Eastern United States 2.2.2 Earthquake Recurrence Intervals for Nuclear Power Plants 2.2.3 Selection of Earthquake Resistant Design Criteria for Nuclear Power 2.2.4 Site Specific Response Project A-3

3.0 RELATION OF STPP TO OTHER PROGRAMS 3.1 NRC Seismic Research 3.1.1 RES 3.1.2 NRR 3.1.3 NMSS 3.2 Other Government Agencies 3.2.1 FEMA and the NEHRP 3.2.2 USGS 3.2.3 NSF 3.2.4 Corp of Engineers 3.2.5 Veterans Administration 3.2.6 State Agencies 3.3 Non-Government Research 3.3.1 Electric Power Research Institute Appendices Attachment A National Center For Seismological Studies Attachment B Program Elements Required to More Expeditiously Resolve Seismic Issues I

A-4 L-

INTRODUCTION Seismic hazards are significant contributors to the overall hazard to some nuclear power plants. Research directed at estimating and reducing the uncer-tainty in seismic hazard and research directed at coping with the uncertainty 2

are central themes of the Seismotectonic Program Plan II (STPP). It is an Earth

{ Sciences Branch (ESB) position that the uncertainties in seismic hazard assess-ments are~ higher in the Eastern United States than in the Western United States

,i (the eastern foothills of the-Rocky Mountains constitute the dividing line).

Because the majority of nuclear power plants are in the East, the ESB research

, is concentrated there.

4

) Three of the principal sources of uncertainty in establishing seismic hazards in the East are uncertainties in the seismic source zones, the propagation of seismic energy, and the site response.

The Charleston 1886 earthquake is an example of how these three factors combine in the East to produce high uncertainties. The 1886 earthquake has not been

~ associated with a structure in Charleston; thus, it might be assumed to have an equal probability of occurrence throughout the eastern seaboard. Although the Charleston earthquake was of about the same magnitude as the 1971 San a

Fernando Valley earthquake, the Charleston earthquake had about 10 times the

area of equivalent damage. The site specific response throughout the coastal plain is additionally complicated by the possibility of severe differential soil settlement and soil liquefaction leading to severe differential structural

settlement. Thus all the factors contribute to produce an overall high uncer-tainty in seismic hazard estimate in the eastern seaboard.

The ESB conducts research addressing each of these sources of uncertainty and also conducts research on methods to handle or cope with the uncertainties in a regulatory environment. These projects include development of "probabilistic" 4

techniques for estimating seismic hazard curves and recurrence information for long return periods. The increased use of probabilistic risk assessments (PRA) for analysis of nuclear power plant sites has created a need for this type of study.

Probabilistic assessment of risk requires methodologies to handle the uncer-

-tainties in seismic hazard assessments. The present licensing criteria are '

" deterministic," and are based on procedures in Appendix A to 10 CFR Part 100. '

The probability of ground shaking above the design level is not explicitly l i considered. However, PRAs show that a major contribution to seismic risk is

due to earthquakes several' times the design level. Three approaches are pre-sently being utilized to develop a probabilistic approach to handle the known uncertainties

expert panels, formal sensitivity stuaies, and examination of

, . methods of estimating recurrence rates.

The issues in seismic hazard assessment are addressed in this document in

, three-sections.

j Section I - The technical issues of seismic hazard assessment.

4

Section II' - ESB research program to address those issues.

~Section III - Relationship of other research programs to ESB research programs.

E A-5 i

1 TECHNICAL ISSUES The three fundamental technical issues addressed by ESB seismotectonic research are: how to quantify and reduce the uncertainties in seismic hazard assessments and how to develop techniques to deal with the uncertainties in a regulatory environment. These issues are closely related. The more accurately the princi-pal contributors to the uncertainty are quantified or understood, the more pre-cisely the contributor can be handled in a regulatory environi ent. These issues are discussed below in the sections, 1.1 Hazard Uncertainties, and 1.2 Handling Uncertainties.

1.1 Hazard Uncertainties As noted earlier, three of the principal contributors to the uncertainty in quantifying the seismic hazard at a site are the characteristics of the seismic source zone, the propagation of seismic energy between the source and the site, and the site response including soil response. The relative levels of contribu-tion of these three to uncertainty are region dependent. (For programmatic ease, the Eastern United States has been divided into four regions: Northeast, Southeast, New Madrid / Anna, Ohio, and Nemaha Ridge.) Currently, there is a rea-sonable level of confidence in a working hypothesis for the source of seismicity in the New Madrid area and a moderate understanding of some of the regional pro-pagation characteristics. There are a number of hypotheses for the source of Southeastern seismicity including the Charleston, S.C. area. No generally accepted hypotheses are available for the Northeast or the Nemaha Ridge. There is a low level of knowledge about the propagation and site response character-istics in the East except as noted for the New Madrid region.

Source zone uncertainties are the most visible problems in the eastern seaboard.

This was underscored by the recent United States Geological Survey (USGS) clarifi-cation of position on the Charleston 1886 earthquake. That position established the Charleston seismicity as an eastern seaboard earthquake problem. In the opinion of the USGS, there is an insufficient historical data base to establish seismic zonation on the basis of seismicity alone, and the causative mechanisms of eastern earthquakes are, with few exceptions, not known.

In areas like the Eastern United States, this inability to identify the causa-tive mechanisms is particularly difficult to deal with; knowledge of the causa-tive mechanisms can be used to compensate for that lack of a long historical data base by establishing source zones on the basis of geology. Studies of the Chinese data base have shown the validity of that approach, i.e., determining the causative mechanism. Unfortunately, studies of that same data base have shown that without the causative mechanism the 2000+ year seismic history is too short for seismic zonation if the source zones cannot be constrained on the basis of geology.

The issues are presented in two groups: regional issues, which are mainly issues of seismic source zone uncertainties and topical issues, which are issues of seismic attenuation, site-specific spectral analysis, and soil liquefaction.

A-6

1.1.1 Regional Issues There are four eastern regional programs and a limited regional program in the Pacific Northwest. In the regional areas there is one basic problem: what is the cause of the seismicity. This problem is fundamental to the other problem of seismic source zone uncertainties which are; where can seismicity occur, with what magnitude and with what frequency. In some areas experts have more confidence in the seismic zonation than others. A short discussion of the source zone issues for each of the regions follows.

1.1.1.1 Southeastern United States The Charleston 1886 earthquake dominates the seismic history of the South-eastern United States. The cause of that earthquake is not known. Numerous hypotheses have been proposed, mostly involving reactivation of older faults.

Most of the hypotheses imply the possibility of Charleston like motion else-where in the eastern U.S. Faults that cut Mesozoic and Cenozoic sediments have been found by seismic refraction and reflection investigations in the Charleston region. However, the faults do not cut the surface. The Cooke fault and the Helena Banks fault, both trending NE, are examples of this. The Bel Air fault near Augusta cuts Mesozoic beds and is also NNE trending. None of these faults have been demonstrated to be seismogenic or even active in the Quaternary.

There are other areas of seismicity in South Carolina in addition to the Charleston area.

Two other areas of significant seismicity in the Southeast are the Giles County and Central Virginia seismic zones. In these areas some progress has been made toward the correlation of seismicity with structure.

Preliminary data from the Central Virginia seismic zone show the possibility of correlation of earthquake hypocentral locations and a structure defined by seismic reflection profiling. Good quality hypocentral locations are available because of the station distribution and a good velocity model of the area.

Excellent information on crustal structure was available because of detailed geologic mapping, good magnetic and gravity data, and a state-of-the-art seismic reflection profile. The locations of earthquake hypocenters apparently correlate with an ESE dipping listric fault that was a splay fault to the decollement.

No fault plane solutions are available, and consequently the sense of motion is unknown.

Recent developments on the stress field in the Southeast raised questions about the sense of motion on faults previously considered unambiguous. The Stafford fault zone of northern Virginia is a NNE trending fault zone that shows about a hundred meters of reverse motion. Previously it was considered that the direction'of stress east of the Blue Ridge was NW-SE directed consistent with 1 the reverse motion of the Stafford fault system. New data suggests a N60 E '

direction of the maximum compressive stress. If this direction.is correct, then the Stafford, Bel Aire, Cooke, and Helena Banks faults all could be primarily strike slip faults. The observed vertical components could be secondary.

This new stress information is critical for interpreting the Central Virginia seismic zone. If the stress direction is correct, the listric splay of the Central Virginia seismic zone is reactivated as a primarily strike slip fault and motion along the decollement would be secondary.

l i

A-7 L

The hypocenters in the Giles County seismic zone are below the decollement surface. Consequently, the observable structure at the surface or in the first few kilometers are not expected to give information useful in interpreting the seismicity there. Previous hypocentral locations have delineated a source zone of the seismicity, which is proposed to be a deeply buried fault system. The existence of such a structure is consistent with the tectonics of the area.

The structure should be observable by detailed seismic experiments described in Part II.

1.1.1.2 Northeastern United States The seismic activity of January 1982 dominates the recent seismic history of the Northeast. The magnitude 5.7 and 5.9 earthquakes in New Brunswick and the magnitude 4.7 in New Hampshire supplied critical information. There are now over 15 strong motion records from those events and the aftershock sequences.

Although structures are observable on the surface, no cause and effect relation-ship has been generally accepted. The most significant geologic development in New England since the first version of this seismotectonic plan was written is the apparent clarification of the issue of the stress direction in the New England area. The direction of maximum compressive stress throughout the region is most likely about N50 E to N80*E. This is the same as the direction in the midcontinent. Previously, it was thought to be oriented about NW-SE and it was thought to have the same orientation in the Southeast also. This new direction is supported by four different studies: a utility study of the seismicity of the Ramapo fault zone, deep hydrofracture stress measurement in New York State by Dr. Marc Zoback of the USGS, an NRC-funded USGS investigation along the coast of Maine utilizing shallow stress measurements.

An important implication of this stress direction concerns the Ramapo fault system. There is a coincidence of seismicity with the Ramapo and other fault systems in the Highlands although the cause and effect relationship is not established. The Ramapo fault system forms the western boundary of a Triassic basin. A N50 E stress field would act on those structures to cause a strike slip type motion. Previously, most reactivation mechanisms proposed for pre-existing faults have been gravity backsliding or reverse motion.

Seismicity and other recent crustal movements in the Passmaquoddy Bay area may be related to reactivation of the Triassic basin in the Bay of Fundy. Recent stress measurements indicate that the basin would be reactivated in oblique normal faulting. This area is important for the " Triassic Basin" hypotheses of eastern seaboard seismicity. The Bay of Fundy basin is the basin close in both structure and dimension to the true continental rifts, such as the Rhine Rift, the Reelfoot Rift, or the East African Rift system. The Bay of Fundy basin shows a very strong magnetic signature, magnetic highs paralleling the edges, that is suggestive of mantle involvement in the structure. (This is different from most Triassic basins which are probably restricted to the crust.)

1.1.1.3 New Madrid Region The seismogenic mechanism in the New Madrid region is the best understood in the East. A zone of seismicity follows the trend of a basement rift, the Reel-foot rift system. The zone of the very severe 1811-1812 main shock sequence is centered about an offset in the rift that also is the location of a zone of A-8

uplift. The proposed seismogenic mechanism is the reactivation of the rift as a strike-slip fault zone. This is consistent with the structure, fault plane solutions, and stress measurements.

The issue that is outstanding is the extent of the seismic source zone to the North and South. The trend of the rift has been extended from the New Madrid zone up to the Anna, Ohio, earthquake zone. This trend does follow photo linears and some surface faulting trends. Sesimic reflection profiling also indicates structure along the extension. However, the gravity and magnetic signatures of the rift along the proposed northern continuation are not as strong as in the segment of the Mississippi Embayment. Some short extensions of the rift are apparent to the Northwest, but if the model of rift reactivation is correct, they should not be significant seismic source zones because the stress field will not be properly oriented to reactivate the segments into a strike slip system.

1.1.1.4 Nemaha Region The Nemaha Uplift in the Central U.S. is associated with moderate earthquake activity. Together with the Midcontinent Ge'ophysical Anomoly, the uplift may represent a rift zone which is accompanied by clastic filled basins on the east side, the basins being separated from the rift zone by a series of faults.

The Humboldt fracture zone in Eastern Nebraska and Kansas represents such a fault that is associated with seismicity. The Thurman Redfield and Northern Boundary fault zones in Iowa have been intermittently active since Paleozoic time.

Elsewhere, the relationship between the uplift and the seismicity is not very clear. However, seismic activity is present all along the Nemaha Uplift. The activity seems to occur particularly at locations where NW trending structures or anomalies cross the Nemaha Uplift.

1.1.1.5 Pacific Northwest Region The Pacific Northwest is a region that is a transition trom the strike-slip tectonic to the south in California to the subduction zone tectonics to the north in Alaska. The regional eeismicity is associated with a moribund subduc-tion zone beneath Washington and northern Oregon. The outstanding seismic issue is the characterization of the current tectonics of the region i.e., are the large earthquakes in the region within a subduction zone, is there a seismic gap, and what other features, including the volcanoes, are controlling the seismicity?

1.1.2 Topical Issues

.There are a number of issues that are less amenable to regional characterization and are called topical. These include propagation / attenuation characteristics, soil response studies, and strong ground motion studies. They are basically ground motion issues and will be considered in this section in the direction of propagation, i.e., from source to receiver site, 1.1.2.1 Source Parameters The issue involves the proper characterization of source parameters for Eastern U.S. earthquakes. These parameters include source spectra, dynamic and static stress drop, duration, focal mechanism, corner frequency, moment, and magnitude.

1 A-9

Knowledge of the parameters and how they vary with earthquake location and size is critical to understand eastern earthquakes.

The stress drop, focal mechanism, source spectra provide significant constrains on the source models and the state of stress within active tectonic / geologic features.

Records from the recent (1982) New Brunswick, New Hampshire, and Arkansas earth-quakes appear to have a strong high frequency component. This could have signi-ficant impact on the type of plant components that would be susceptible to dam-age from a small nearby earthquake and could change the importance of documen-ting the " Tau" effect* and its possible frequency dependence.

1.1.2.2 Propagation / Attenuation Characteristics There are several aspects of the propagation of seismic energy that are problematic. They are the variation of the attenuation function with distance (lateral heterogeneity) and the dependence of "Q,"** the attenuation factor, on seismic wave frequency and direction of motion.

Some very promising work has been done on attenuation by St. Louis University and MIT with partial NRC support, but this must be broadened to more fully address the problems mentioned above, particularly the frequency dependence of "Q" . Dr. Singh's work at St. Louis University has provided some rudimentary information on the regional variation of "Q" between 0.5 and 3.5 Hz. Informa-tion about the high frequency component of motion is important for the NRC.

1.1.2.3 Site Response Characteristics Under this issue there are two basic problems, calculation of site specific response spectra and estimation of the site soil response. There are currently no widely accepted techniques available to calculate site specific response spectra. Site specific spectra may be used as substitutes for Reg. Guide 1.60.

There is virtually no guidance available on calculating them, and considerable applicant and NRC staff is spent in development and evaluation.

Also included in site specific response is the issue of the partition of seismic energy between the horizontal and vertical components. The work of Dr. Gupta of Teledyne Geotech indicates that the ratio of the " average" horizontal to vertical acceleration may vary by as much as 200-300%.

1.1.2.4 Soil Failure Many nuclear power plants in the East are sited on soil. Soil settlement and liquefaction can become a significant issue when ground shaking greater than the SSE is considered. There are numerous predictive models for soil settlement and liquefaction; but there has been little work to validate them.

• The " Tau" effect is an observed phenomenon related to the wavelength filtering capacity of large structures such as a power plant foundation. In a favorable orientation of structure to incoming seismic energy, it is postulated that the structure could filter strong ground motion.

• • "Q" is the attenuation factor and is inversely proportional to the decrease in amplitude of a seismic wave as it passes through a medium, i.e., the higher the "Q" the smaller the change in amplitude.

A-10 t

1.2 Handling Uncertainties A fundamental issue that needs to be addressed concurrently with attempts to quantify and reduce the uncertainty in seismic hazard assessment is how to handle the uncertainty in a regulatory environment. Some of the techniques used to handle the seismic hazard uncertainty require further analysis and development.

Two significant factors that affect the implementation of STPP research results into regulatory decisions: (1) how the results will be used in regulatory decisions and (2) the nature of the results to be used. Probabilistic Risk Assessment (PRA) is one technique that takes the results from STPP and builds a tool that can be used in the regulatory process. Thus, part of the responsi-bility of ESB under STPP would be to make certain that the STPP results and their associated uncertainties were in a form compatible with PRA techniques. A number of PRAs that have included external events have identified seismic hazard as one of the dominant contributors to overall risk.

Historically, seismic design parameters are presented in terms of anchorpoint acceleration values and then Reg. Guide spectra. These are not the input specifications that PRA practitioners can most accurately and et.sily handle.

They would prefer input in the form of probability density distributions of ground motion incorporating estimates of sensitivity to spectral distribution, duration, mode and the like. Development of that type of information requires improvement of existing methods and development of new ones to handle the individual uncertainties and their interrelated sensitivities. This is an important step in the development of techniques to compare the saismic risk at two different sites or to compare the relative significance of internal and external contributors to tne overall risk. However, within the time frame of a normal life cycle of a nuclear power plant, there will always be some uncertainty in the assessment of seismic hazard of the eastern seaboard. This uncertainty needs to be factored into the decision making process.

A-11

l 2 PROGRAM PLAN The program plan to resolve the technical issues described in the previous section is presented here in a parallel format. There are two parts:

programs related to quantification and reduction of uncertainty in seismic hazard assessment and programs related to dealing with the uncertainty. The programs described in the first part generally involve basic data collection and interpretation. The programs described in the second part generally involve probabilistic analysis of existing data sets.

2.1 Estimation and Reduction of Uncertainty The programs concerning the estimation and reduction of uncertainty are grouped into regional programs and topical programs. The regional programs consist of operation of seismic networks and geophysical / geologic investigations of key regions. The topical programs consist of studies related to generic issues rather than source zone problems. There is some overlap, the seismic networks also contribute to the topical program.

2.1.1 Regional Programs The uncertainty in the characterization of seismic source zones is the fundamental issue addressed by the regional programs. This uncertainty arises from several aspects of the issue; delineation of the seismic source zone, estimation of the maximum credible earthquake, characterization of source parameters, occurrence relationships and the like.

The programs are subdivided into the previously described four regions. This subdivision is partially because the nature of the problems or issues vary with the regions and the tectonic processes may differ in the different areas.

The role of the seismic networks as the mainstay or backbone of the regional part of the STPP will be described in a separate section. This deviation from the regional presentation of the program plan is being made so that the common element for the networks will not have to be repeated for each region.

Regional Seismographic Networks The four regional networks and one independent local network are the mainstay of the regional STPP (Table 1). Operation of these networks provides data of primary importance in identifying seismic source zones, depth, recurrence statistics, attenuation information, earthquake source parameters, and other basic seismological data.

A-12

r i

Table 1

1. The Northeastern United States Seismographic Network (composed of six local networks, about 130 stations)

2. The Southeastern United States Seismographic Network l- (four local networks, about 65 stations)

3. The New Madrid / Anna Ohio Seismographic Network (three local networks, about 65 stations)

4. The Nemaha Ridge Seismographic Network (two local networks, about 25 stations)

5. The Northern Oregon .ceismographic Network (a single local network, about 10 stations)

As part of the previous STPP, a program was initiated to promote the upgrading of seismographic network instrumentation. The first phase of the transition from analog recording to full digital recording of digital signals, i.e.,

conversion to digital recording of analog signals is nearly complete. The second phase, development of a number of high quality digital test stations for several networks, is progressing.

The issues that this upgrading and testing address are two-fold. The first is mainly economic, and the second is the number of significant seismological problems that are critical to siting nuclear facilities and do not appear to be readily soluble with the current narrow-band limited-dynamic-range-seismographs.

One of these significant problems is hypocentral depth determinations.

Significant progress has been made in specifying the epicentral location of recent seismicity in the East. However, the depth is still difficult to obtain.

It is becoming apparent that, without either more seismographic stations, or new analytical techniques, the depth is going to remain elusive. The proposed solution of increasing the information content of the current recording capabil-ity at a number of test stations is more economical and scientifically produc-tive than adding more stations. By increasing the information content it is possible to reduce the uncertainty in the depth term by making greater use of the earthquake time series. Accurate knowledge of the depth of epicenters is required for correlation of seisinicity with " suspect" geologic structures.

There is an apparently little correlation between seismicity and surface-mapped geologic features. Uncertainty about the depth of seismicity and the subsurface trend of the feature contributes significantly to this lack of correlation.

Accurate depth information is also required as input to fault plan solutions the earthquake modeling codes used to generate synthetic seismograms, spectra, and attenuation information.

Relationship Between Seismographic Networks and Geological / Geophysical l Investigations l However different the regional issues are, the relationship between the

) seismographic networks and the geological / geophysical studies remains the same, j j The networks provide the basic seismicity data, location, depth and magnitude

! which are the fundamental data for establishing the seismic source zones and i 1

]

A-13 I

L

the recurrence relations in them. This information is used to select the areas which are currently seismically active and thus most likely to be productive for geological and geophysical investigations. One of the primary objectives of the STPP is to establish, if possible, the relationship between seismicity and tectonic features.

One of the important contributions of the geological and geophysical investi-gations is their ability to augment a short seismic history. Geological and geophysical investigations, such as stratigraphic studies and geodetic studies can confirm delineation of seismic source zones by " lengthening" the seismic record, in-situ stress measurements can confirm focal mechanism-derived stress directions and seismic profiling can confirm crustal structure models. With the added dimension of geological and geophysical investigations the character-istics of the seismic source zones can be established with more confidence.

2.1.1.1 Southeastern United States A primary issue in the Southeast is determination of a causative mechanism or structure of the 1886 Charleston earthquake. A number of hypotheses have been proposed and the major effort of the STPP in the Southeast is testing these hypotheses for their relevance to eastern seismicity both in the immdiate Charleston area and in other areas in the Southeast. The USGS is the prime contractor to the NRC in the immediate Charleston area; its program there is partially funded by the NRC and the USGS.

The program in the Southeast has 6 geological / geophysical projects.

Geological / Geophysical Program

1. Seismic Reflection Profiling in Giles County, VA The purpose of this project is to investigate a postulated structure that is suggested to be the source structure of the Giles County seismic zone. A deeply buried structure, interpreted as an Iapetan normal fault reactivated by modern stresses, has been proposed as the source structure based on the alignment of seismicity. This project will make use of some state-of-the-art Vibroseis techniques being developed by a Virginia Polytechnic Institute (VPI) consor-tium to penetrate the shallow karst structures in Giles County. This is a cooperative cost-sharing project coordinated by VPI with State, Federal, and oil industry support. The NRC support will be about 10% of the total cost.

2. Seismic Reflection Profiling and Supportive Geology in Central Virginia The primary purpose of the project is to determine the structure of the earth at hypocentral depths in the Central Virginia seismic zone. Preliminary studies show a promising correlation of earthquake hypocenters with structure defined by multichannel seismic studies. The structures are visible at the surface as lithologic boundaries that are fault controlled. The supportive geology in FY 84 will include neotectonic studies of young deposits to determine if there is recent movement.

3. Neotectonics of Southeast The purpose of the investigation is to gain neotectonic information along several proposed seismogenic structures in the Southeast to see if there is A-14

- - . .-. - - - = - - . . .- . .

l evidence that they have been reactivated. Three structures will be investi-

~

gated: a Triassic Basin; a structure related to the decollement, and one

other. The location of these structures will be determined during the l

investigation.

4. In Situ Stress Measurement The direction of in situ stress defines what faults can be active in a region and how they can move. Several measurements will be conducted at critical locations. The first will be conducted in the Central Virginia seismic zone where the structure and the distribution of earthquakes are known. The stress information will give critical information on the type of motion occurring on the fault. The next site chosen will probably be at or near a Triassic Basin.

5. Charleston Clarification Project A project will be initiated in FY 84 to solicit information critical for testing hypotheses on the cause of the Charleston event. The project will  ;

build on results of the USGS program, the neotectonics and the in situ stress projects.

6. U.S. Geological Survey Charleston Projects The USGS has been the major investigator in the Charleston region. This work was cooperatively funded by NRC and USGS. The work includes operation of a seismographic network which is being upgraded to include some three axis instruments. Considerable seismic reflection profiling has been conducted both on shore and off shore. This led to the discovery of the Helena Banks and Cooke faults. Several deep boreholes and many shallow boreholes were drilled which defined the stratigraphy of the region.

Future work in the region may continue the similar types of projects. The

-impact of the USGS clarification of its position on Charleston is being fully factored into the annual formulation of the USGS-NRC program.

2.1.1.2 Northeastern United States In the Northeast there are some suspicious structures, e.g., Ramapo fault system, and some activities that may be related to seismogenic mechanisms, e.g.,

the rapid crustal subsidence along the coast of Maine. However, the level

of understanding is not as well developed as in the Southeast where several models have been proposed and can be tested. In the Northeast the models are being developed.

The Ramapo fault is a critical structure that has been targeted for detailed studies. There is an apparent correlation of earthquake epicenters with the strike of the feature. The geology of the zone is well known, a Triassic j basin parallels older features. In situ stress measurements and multichannel

l. seismic reflection surveys will be conducted under this seismotectonic plan.

With these results, the level of knowledge about the Ramapo seismic zone will be very high, and the relationship between the structure and seismicity will be made clearer.

A-15 l r

l

The program in the Nartheast has 4 geologic / geophysical projects.

1. State of Maine Cooperative Study There is a cooperative agreement with the State of Maine to investigate rapid subsidence along parts of the coast. The rates are up to 9mm per year, and seismicity is spatially associated with the subsidence. This is a cooperative agreement to investigate the subsidence. This is important because the subsid-ence is occurring along the Bay of Fundy rift. The reactivation of rifts has been proposed as the cause of the Charleston event and other events in the Eastern United States. The State of Maine also assists in earthquake investi-gations in the region, such as the 1982 New Brunswick events.

2. Neotectonics of Northeast /New England Critical structures will be investigated for evidence of recent movements or reactivation. These will include at least one Triassic basin. The area of the Gaza, New Hampshire, earthquake and areas near the Long Island epicenters will also be investigated. Verification of the Adirondack uplift is also a possible target. Given the success of paleoliquefaction studies in New Madrid, thase are also likely studies.

3. In Situ Stress Measurements In situ stress measurements will be conducted in the Ramapo seismic zone and the Moodus, Connecticut seismic zone. This is to gain information concerning which faults may be active and what type of motion is likely on them. This informa-tion should help clarify the type of fault motion that is occurring on the Ramapo. Previously, reverse faulting was suspected; however, recent seismic results suggest strike slip faulting. These will be conducted in FY 84. Other areas will be targeted for measurements, including, perhaps the site of the Gaza, New Hampshire earthquake.

4. Seismic Profiling of the Ramapo Fault Zone Three cross sections are planned across the Ramapo fault zone to define the structure at hypocentral depths. The cross sections have been chosen on the basis of the well mapped surface geology and the distribution of earthquakes.

The USGS is supervising the study.

2.1.1. 3 New Madrid / Anna, Ohio The cause of seismicity in the New Madrid region is reasonably well established.

The seismicity is probably occurring along a continental rift zone. The rift is too deeply buried to be detected geologically; however, evidence of recent movement is observable at the surface. The regulatory issue is what is the extension of the rift to the south and north. Geophysical data indicate that the rift spreads out to the north into several different branches. Seismic studies suggest that one of the branches reflection continues to the northeast toward the Anna, Ohio, seismic zone; however, the strong magnetic signal that defines the Maine part of the rift is weaker there.

A-16

.The relationship of the Arkansas 1982-83 swarm to the New Madrid activity is not yet known. The issue is whether the swarm occurred south of the New Madrid zone or not. This suggests the possibility of even further southern extensions.

NRC work is directed primarily in determining the extent of the seismic zone.

This work is proceeding by three processes: seismic monitoring, geologic investigations, and geophysical investigations.

1. Geological Investigations For Evidence of Seismicity There.is geologic evidence of the seismic activities within the documented New Madrid seismic zone. The most striking evidence is the Lake County uplift.

There are also numerous sand blows in the area. Geological investigations outside of the documented seismic zone are directed at determining if such evidence is present along the proposed extension of the Reelfoot Rift. At present, they have not been found. The funding level for this work is being reduced after FYP4.

2. Extension The definitive evidence for the Reelfoot Rift has been geophysical, both by seismic reflection and refraction and potential field studies. Geophysical studies are proposed through FY85 to examine the proposed extension of the Reelfoot Rift to determine if seismogenic structures are present. Preliminary studies show the presence of faulting along some of the extension but, the age is old.

2.1.1.4 Nemaha Region The research activity in the Nemaha Region consists of the operation of a seismic network. Because of the recent discovery of a possibly active fault in Oklahoma, the meers faults, a limited trenching program may be initiated there.

2.1.1.5 Pacific Northwest Region The research activity in the Pacific Northwest consists of the operation of a seismic network. Presently there is no geophysical / geological work planned.

2.1.2 Topical Program The Topical Program is designed to address the topical issues raised in the previous section.

2.1.2.1 Source Parameters A number of projects have been initiated under this topical element. The most iNportant of these currently is the acquisition of a " viable" data base of strong ground motion accelerograms of eastern U.S. earthquake. This is being accomplished by a continuing project to acquire and install an array of l accelerographs in the eastern U.S. to augment the instrumentation programs of l the U.S. Army Corps of Engineers and the Veterans Administration. (The i instrumentation for these two agencies is usually at dam sites or Veterans hospitals.) This project will support the acquisition and installation of A-17 i i

about 50 strong-motion accelerographs. A number of these instruments will be deployed close to a " host" institution so that they may be quickly redeployed in the event of a moderate to large eastern earthquake.

The NRC is also supporting several projects for the analysis and dissemination of some of the existing eastern strong motion accelerograms in particular those from the 1982 earthquakes in New Brunswick, New Hampshire, and Arkansas.

Currently as part of the NRC-USGS Interagency Agreement, the NRC is funding detailed analysis of the New Brunswick and Arkansas data set with some field work.in New Brunswick (NB) with the Canadians. This work is intended to develop more precise data on the crustal structure at the NB site to more fully explain the NB accelerograms. The reason for the high accelerations for small earthquakes remain a puzzle. The U.S. Army Corps of Engineer records from the New Hampshire earthquakes are available as NUREG/CR 3327 with magnetic computer tapes from the NOAA data center in Boulder, Colorado.

An explicit program for work under this subelement has not oeen developed but the above type work will continue to be supported.

2.1.2.2 Propagation / Attenuation Characteristics Much of this work is being carried as part of the seismic data analysis conducted by the regional networks. It is being highlighted as part of the Topical Pro-gram because of its importance. It is also included here because the data base that is currently available is too sparse to provide propagation information on a regional scale and thus should be handled on a broader scale. (This problem of data paucity is further addressed elsewhere in this program plan by improve-ments in network instrumentation.)

The limited amount of data available is being used to the fullest extent possible.

Projects being' planned that use data from special systems that are currently being operated or were operated in the past by 00E, 000, and others for specific limited projects. The data from these sources may significantly augment the network data.

2.1.2.3 Site Response Characteristics A major effort is planned in the area of site specific response development.

It is anticipated that the exact focus of this effort will be developed following a major workshop on the subject which is to be held in July 1983.

This workshop is being jointly supported by the NRC, USGS and other sponsors and is designed to establish the state-of-the-art on site specific response problems. We plan that this new work will explicitly address the problem and representation of ground response in excess of the SSE.

A current project in this area that overlaps with the two previous elements is a project with Structural and Earthquake Engineering Consultants entitled

" Selection of Earthquake Resistant Design Criteria for Nuclear Power Plants; Methodology and Technical Basis." This project is to examine the use of geologic data in the characterization of seismicity, the scaling of strong ground motion, and the scaling of response characteristics for various site and earthquake parameters.

A-18

I 2.1.2.4 Soil Failure Development of a validated model for soil failure is the goal of the soil failure project. A large number of nuclear power plants are on soil founda-tions and given that PRAs are examining the risk of nuclear failure due to ground shaking at 2-3 times the SSE, then the potential for soil failure becomes a contributor to the overall risk. There is presently no validated code for soil failure / settlement.

In addressing this issue we have taken advantage of the experience of the Corps of Engineers who have long been concerned with the effects of soil failure on their structures. They have considerable experience in both labor-atory and field studies. They conducted for the NRC a detailed study of available soil failure codes and concluded that DESRA, a code developed by Prof. Liam Finn of the University of British Columbia, was the best candidate code for validation. There were three reasons: the physical basis of the code was the best; the code utilized standard engineering data, and it had been partially validated by data taken at an artificial island in Japan during an earthquake.

The validation experiments will be conducted at Cambridge University under a subcontract using soil samples tested in a centrifuge. A sand saturated with glycerine will be loaded with model structures and subjected to model seismic loads. Pore pressure and displacements will be monitored during the loading.

As an independent effort, the possibility of a full scale validation of DESRA using using data from the Nevada Test Site will be investigated.

In FY 1984 the NRC will participate in a soil settlement workshop sponsored through The National Academy of Science, the National Science Foundation and the U.S. Geological' Survey.

2.2 Program to Address Uncertainties The objective of this portion of the STPP II is to develop techniques for the analysis or evaluation of existing data sets. Currently these projects all have some probabilistic aspect. There are two projects under this program already in place, one more in the procurement process, and one still being formulated.

2.2.1 Seismic Hazard Characterization of the Eastern United States (SHC)

This is a joint project between the Offices of Nuclear Reactor Regulation (NRR) and Nuclear Regulatory Research (RES) which is being conducted at LLNL.

It is an outgrowth of the probabilistic seismic input for the Systematic Evalua-tion Project (SEP) and the Site Specific Spectra Project. It is designed to provide input as a simplified seismic input for PRA and to be used by NRR for resolution of the issues raised by the USGS clarified position on the Charleston Earthquake.

A principal objective of the SHC project is to improve the basic project computer code developed for SEP and to produce seismic hazard curves with appropriate sensitivity studies for ten test sites. The project uses a solici-tation of expert opinion as the source of basic seismological data. The i

A-19

seismicity panel provides seismic source zone maps, estimates of maximum magnitude earthquakes for each source zone, and estimates of earthquake recur-rence parameters. The attenuation panel provides input on the attenuation and source parameters to be used to propagate the earthquake ground motion from the source to the site.

Results for the ten test sites are scheduled to be ready by mid-February 1984 at which time the codes will be turned over to NRR as a licensing tool. It is anticipated that preliminary hazard curves will be available for all eastern U.S. nuclear power plant sites by about mid-1984.

2.2.2 Earthquake Recurrence Intervals for Nuclear Power Plants This project is being carried out by ERTEC to provide advice to RES concerning the quality of available techniques for calculating earthquake recurrence param-eters. The project consists of five tasks: (1) identify the classes of avail-able recurrence interval estimating techniques and to recommend some for evalua-tion, (2) develop an evaluation scheme, (3) to evaluate the techniques, (4) make a recommendations to RES based on the evaluation of the techniques, and (5) develop a new technique at the option of RES. One of the important aspects of this project is that the contractor is to explicitly address the problems and uncertainties associated with recurrence interval estimates out to very long times, as great as 100,000 and 1,000,000 years.

2.2.3 Selection of Earthquake Resistant Design Criteria for Nuclear Power Plants; Methodology and Technical Bases The proposed topics are grouped into the five separate tasks. The first two, to be carried out during the first year of effort, will provide for formulation and selection of models to describe slip rates at known active faults and to scale strong motion close to these faults. The third and fourth tasks, planned for the second year of the effort, will update the direct-scaling relationships of spectral amplitudes of strong ground : notion and will update and further generalize the Uniform Risk Spectrum method we developed some five years ago (NUREG-0406 Vol. 2). The fifth and the final task, planned for the third year of the effort at the option of NRC, will produce the recommendation on standard (typical) and special case methods and procedures for use in licensing and will combine the deterministic with probabilistic methods of analysis for selecting the balanced seismic design basis.

2.2.4 Site Specific Response Project This project is in the preliminary planning stage. It is to develop tech-niques for calculation of site specific response spectra and will specifically address the problem and the representation of ground response in excess of the SSE. It was also briefly mentioned under Topical Programs.

A-20

3.0 RELATION OF STPP TO OTHER PROGRAMS In this section the STPP II addresses its relationships with the three groups with which it is associated. The first group is composed of users of STPP products within the NRC. This group includes NRR, NMSS, DRA/RES, and DET/RES.

The second group is composed of supporters of earthquake hazard research with whom the ESB has informal or formal cooperative agreements. This group includes Federal organizations such as the USGS and FEMA and State agencies such as geological surveys. The third group is composed of private sector organizations.

3.1 NRC Seismic Research 3.1.1 RES >

Research related to the uncertainties in seismic risk is contracted for by three divisions within the Office of Nuclear Regulatory Research: Division of Radiation Programs and Earth Sciences (DRPES); Division of Engineering Technology (DET); and Division of Risk Analysis (DRA). Briefly the interface between DRPES and the other two RES divisions is briefly described below.

DET is concerned primarily with the mechanical response of nuclear power facil-ities to earthquakes. They need the seismic hazard input to conduct their work.

They supported some research on hazard estimation as part of SSMRP. The Seismic Hazard Characterization for the Eastern U.S. (previously described in the third part of the program plan) is a spin off from that part of SSMRP. That project which is being monitored by DRPES is intended to provide a simplified seismic input for SSMRP use. There is a DET and DRPES interface in soil mechanics also.

DRPES sponsors research concerned with soil settlement, soil failure and lique-faction while DET sponsors research concerned with soil-structure interaction.

DRA is concerned with seismic hazard as a element in the assessment of seismic risk at nuclear power plants. The Full Scope Risk Assessment Program is a major program in DRA which includes external input such as seismic information.

This is the principal interaction point between DRA and DRPES. DRPES provides information to DRA for development of seismic input.

3.1.2 NRR

- RES coordinates its research activities on seismic hazards very closely with i its counterparts in NRR. As mentioned in the third part of the Program Plan, NRR and RES are jointly supporting the development of the Seismic Hazard Characterization project. RES, in effect, is sponsoring the development or improvement of a computer code for hazard curve calcualtions, and NRR is supporting its use at specific sites.

3.1. 3 NMSS NMSS, under the high level waste program conducts considerable technical assistance in the geologic and geotechnical areas and is concerned with )

seismic siting and seismic design.

l A-21

3.2 Other Government Agencies A number of other federal agencies.have research programs associated with seismic hazard evaluation and reduction. A large portion of this other research effort is associated with the National Earthquake Hazards Reduction Program (NEHRP) with FEMA serving as the " Lead Agency." The four major participants in the NEHRP are: FEMA, the U.S. Geological Survey, the National Science Foundation, and the National Bureau of Standards. The USGS is the principal agency in this group supporting research related to earthquake hazards in the U.S. The NSF is responsible for fundamental geological and seismological research and for most earthquake engineering research.

In addition to the NEHRP, other agencies such as the Departments of Energy and Defense have programs that provide basic geological, seismological and engineering data and analysis techniques that are useful to the NEHRP.

3.2.1 FEMA and the NEHRP FEMA as the " Lead Agency" for NEHRP is charged to coordinate a national effort to counter the serious earthquake threat in the United States. As stated in the law, its objectives are:

o The development of technologically and economically feasible design and construction methods and procedures to make new and existing structures, in the areas of seismic risk, earth-quake resistant, giving priority to the development of such methods and procedures for nuclear power generating plants, dams, hospitals, schools, public utilities, public safety structures, high occupancy buildings, and other structures which are especially needed in time of disaster; o The implementation in all areas of high or moderate seismic risk, of a system (including personnel, teshnology, and procedures) for predicting damaging earthquakes and for identifying, evaluating, and accurately characterizing seismic hazards; o The development, publication, and promotion, in conjunction with State and local officials and professional organizations, of model codes and other means to coordinate information about seismic risk with landuse policy decisions and building activity; o The development, in areas of seismic risk, of improved understanding of, and capability with respect to, earthquakerelated issues, including methods of controlling the risks from earthquakes, planning to prevent such risks, disseminating warnings of earthquakes, organizing emergency services, and planning for reconstruction and redevelopment after an earthquake; o The education of the public, including State and local officials, as

, to earthquake phenomena, the identification of locations and structures which are especially susceptible to earthquake damage, ways to reduce the adverse consequences of an earthquake, and related matters; A-22

o The development of research on

a. ways to increase the use of existing scientific and engineering knowledge to mitigate earthquake hazards;

b. The social, economic, legal, and political consequences of earthquake prediction;

c. ways to assure the availability of earthquake insurance or some functional substitute; and

d. the development of basic and applied research leading to a better understanding of the control or alteration of seismic phenomena.

In addition to coordinating the NEHRP activities, FEMA leads the effort to improve the seismic safety of Federal buildings and conducts a program of earthquake preparedness planning for areas of high and moderate seismic risk.

FEMA was one of the co-sponsors of the May 1983 Charleston Earthquake Workshop.

3.2.2 USGS The U.S. Geological Survey is the principal federal agency charged with carry-ing out the geological and seismological aspects of NEHRP. The second objective of the Earthquake Hazard Reduction Act of 1977 (P.L.95-124 [S.1267]) specifies "the implementation (of NEHP) in all areas of high or moderate seismic risk."

Without providing an exact definition of the term "high or moderate seismic risk," the USGS is currently carrying out an earthquake hazards and prediction program under the auspices of NEHRP which places an overwhelming majority of its NEHRP funds in California.

For FY 1984 the USGS is reported to be developing a program plan to address seismicity problems in the Eastern U.S. The thrust and content of this plan is being developed.

In addition to the relation between the USGS and the NRC implied by the above,

! the USGS is the NRC's prime contractor for seismic hazard research in the l Charleston, S.C., area. The USGS also does work for the NRC on the topics of I

strong ground motion and probabilistic sensitivity. Table 1 lists current USGS projects.

3.2.3 NSF The Geophysics Program of NSF sponsors research on earthquake processes and phenomena. The Civil and Environmental Engineering Program sponsors work on l carthquake engineering and socioeconomic implication of earthquakes. The research supported by DHSWM under STPP more closely interfaces with the Geophysics Program at NSF while the DET program interfaces more closely with the Engineering Program.

A principal difference between the DRPES program and the NSF Geophysics Program is that the former is structured to produce a specific product, i.e., earthquake hazard information, and the latter is structured to produce basic scientific cdvancements.

A-23

Coordin-tion between the two programs, NRC and NSF, is very good as evidenced by the numerous jointly funded projects such as the recent Charleston Earthquake Workshop and the Soil Settlement / Liquefaction Workshop being planned. This close coordination and cooperation is fostered by the Naional Academy of Sciences' Interagency Geophysics Discussion Group.

3.2.4 Corps of Engineers The Corps of Engineers (COE) is concerned with seismic hazard primarily as it affects their structure. They have produced a seismic hazard map of the central U.S. and are working on one for the eastern seaboard. They are actively concerned with the problems of soil settlement and liquefaction.

The Corps maintains an array of strong-motion accelerographs at many of their dams. This is particularly important in the eastern U.S. where not many such instruments are deployed. These instruments are deployed to provide COE with data specifically applicable to dam safety so some of the data is not directly applicable to more generic safety problems. In spite of this limitation ORPES and COE have been able to successfully cooperate on a strong motion data analysis project. DRPES and COE are also cooperating on a joint project to validate theoretical models for soil settlement or liquefaction induced by earthquake ground motion.

3.2.5 Veterans Administration The Veterans Administration maintains strong motion seismographs in their facilities. Although this is a minor effort, the data from those is potentially very significant.

3.2.6 State Agencies A number of states conduct research in geology and seismology, primarily through the various state geologic surveys. Generally, their work is more related to economic aspects of geology, such as mineral deposits or water resources. Some states do have hazard related research or research directly related to NRC information needs. The DRPES presently has cooperative programs with several states to share costs on projects of interest to both.

3.3 Non-Government Research 3.3.1 Electric Power Research Institute (EPRI)/NRC Interactions EPRI has developed a draft plan to assess probabilistic seismic hazard in the Eastern United States in response to the USGS clarification of position on the Charleston earthquake and to the NRR plan to address that clarification. The EPRI plan is a parallel but independent effort to the hazard assessment being conducted by NRR. NRR is using the methodology developed by the joint RES/NRR Sesimic Hazard Characterization project mentioned earlier in the STPP. The current EPRI plan is primarily a response to licensing initiative; however the EPRI plan leaves open the possibility for a research effort to clarify some technical issues. If EPRI begins such work, then it will be necessary to coor-dinate efforts. They are kept informed of our activities, and they received a copy of the previous seismotectonic program plan.

A-24

Attachment A NATIONAL CENTER FOR SEISMOLOGICAL STUDIES One of the.truly significant problems that is facing the entire seismological community is the efficient management, both in time and cost, of the data that is being currently collected. A report entitled, " Effective Use of Earthquake Data" written by the Panel on Data Problems in Seismology of the Committee on Seismology of the National Academy of Sciences / National Research Council was published.in August 1983. To quote that report, "the most important recommenda-tion of this panel is that a National Center for Seismological Studies be established that will overcome the key data management problems that we have identified and enhance the availability and effective use of high quality data

' sets by the entire seismological community."

An unofficial working group including several members, past and present, of the Committee on Seismology is attempting to develop a program plan for the implementation of this recommendation. One of the solutions involves the use of the Center for Seismic Studies operated by the Defense Advanced Research Project Agency of D00.in Rosslyn, VA. Discussions involved use of the center itself or as a prototype depending on DARPA's commitments and the availability.

of support. The establishment of a national nonprofit corporation was one of the vehicles suggested for supporting the center with contributions from all

~the organizations making use of the data. The contributions would be expected to reflect in some sense the contributor's need for the information. Thus, the USGS, FEMA and NSF who all have major responsibilities under the Earth-quake Hazards Reduction Act-of 1977 might be expected to be major contributors to the corporation. The NRC because of its continuing concern for the safety of nuclear power plants might be expected to be a contributor. The same very clearly applies to the nuclear utility owners as well. There will be further comment of this issue later in this section.

As currently being discussed, the principal function that the National Center for Seismological Studies would have would be as a data facility. This function requires some explanation. One of the truly onerous and time consuming tasks of seismological research is collection and preparation of a suitable data base.for analysis. The computer systems distributed by the USGS in about 1978-79 were intended to help reduce this burden to allow the seismologist more time for productive research. The DARPA center-is designed to collect, collate, and disseminate its data base in an extremely efficient manner and to provide a master file of data analysis and display codes for research problems. The data management problem is handled once and as close to the data source as practical. This would provide more efficient use of manpower and financial resources.

When such a National Center comes into being it can serve as the appropriate nucleus-for a National Digital Seismographic (NSD) Network. In the most general terms, the NDS network would be composed of three-component broad-band seismographs with high dynamic range digitization of the signals at the i seismometer site with final permanent recording at some central site. This I squipment is in sharp contrats to the low dynamic-range, narrow bandwidth  !

seismographs with analog recording that are currently in use. Very simply l

-put, the principal advantage to such a new system would be the significant '

increase in the information content of the recorded signal.

A-25

This may be well illustrated by the preliminary results of an NRC supported experiment in New Brunswick. A number of portable high dynamic range seismographs were temporarily deployed in the vicinity of the 1982 New Brunswick earthquakes. Preliminary results indicate that seismic signals above 100H were recorded. These are well above the f predicted for this case by cufrent theory. Thus,thesenewrecordingsmay*Naveafundamental impact on theories about earthquake source mechanisms.

The Earth Sciences Branch (ESB) is supporting some experimentation with these broad-band seismographs and the development of some of the necessary analysis codes. (Some of this effort is being conducted in cooperation with the Air Force Office of Scientific Research.) This work is being carried out to explore the advantages and limitations of such a system technically and to provide the NRC with some experience using this type data to resolve regional seismic issues.

One of the recurrent problems concerning RES management is the recurrent costs of operating the regional seismic networks in the eastern U.S. This is particularly worrisome when it is recognized that there is no point in the foreseeable future beyond which the NRC will not need seismic information to carry out its mission. A National Center for Seismological Studies and a National Digital Seismographic Network operated by a non profit corporation which is capable of accepting funding from private as well as governmental organizations may be the mechanism that would resolve this problem.

A vigorous campaign is being carried out to increase to level of outside financial support for the STPP. ESB is committed to developing $1.2 million from external sources to support the STPP in FY 1985 and beyond. The National Center and Network represent two excellent vehicles to carry out this campaign, i.e., some of the financial burden for the eastern seismographic networks could be more evenly divided among the users of the data.

A-26

Attachment B i

PROGRAM ELEMENTS REQUIRED TO MORE EXPEDITIOUSLY RESOLVE SEISMIC ISSUES The availability of financial resources is always one of the important controll-ing factors on how expeditiously a research issue can be addressed. In this cppendix we present a series of program elements that if they are funded now rather than at some time in the future may lead to a more rapid resolution of the seismic issues. All of these elements are important to resolution of the issues and very probably will have to be conducted in the future. They were not selected for inclusion in the STPP because of resource constraints. The projects established in the STPP are those, based primarily on the informed judgment of the ESB staff, which are most needed to resolve the seismic issues i

f in a timely fashion.

In this appendix other key problems in the seismotectonics of the East are defined along with the expected impact of the results. The projects are seismo-logical and geologic / geophysical. The seismological investigation involves more cccurate location;and description of earthquake hypocenters, source parameter, i

propagation characteristics, and site-specific response characteristics. The l

seismological investigation involves more accurate location and description of carthquake hypocenters, source parameter, propagation characteristics, and site-l specific response characteristics. The geologic / geophysical investigation involves determination of the structure, strain and stress of the earth's crust at hypocentral depth. Both the seismologic and geologic / geophysical investiga-tion address the key issue of what is causing the earthquakes.

l Seismological Elements In this appendix the seismological elements will not be separated into regional, topical or probabilistic categories. The main difference between the program presented here and in the main body of STPP is the compression of the milestone schedule.

(1) Acceleration of the upgrading of the regional networks into the National

! i Digital Seismographic Network and the deployment of strong ground motion records in the eastern U.S. are probably the two most important elements for this program. A delicate balance of time must be maintained for the I first element because it cannot be allowed to develop into an NRC show.

,0t.1er participants should be allowed some time to come up to speed. The second element should be advanced as quickly as prudent expenditure of funds will allow.

I I

(2) Work on propagation characteristics using both new and existing data

, sets.

(3) Site specific response characteristics (4) Source parameter characteristics.

These three elements are all important to modeling, and propagating, ground motion to the nuclear power plant site and then estimating the amount of seismic energy that will be important to the plant. It is difficult if not impossible to prioritize these elements.

A-27

(5) Development of techniques to handle uncertainties and estimate the sensitivity of regulatory decisions to the various uncertainties.

Geological / Geophysical Elements The geological / geophysical elements are not as easily handleo by simply compressing the milestone schedule. These eiements are beyond the time scale of the STPP.

The three major seismotectonic issues that are addressable by geologic /

geophysical means are: 1) The structure of the Earth's crust; 2) the stress field active in the East, and 3) the recent strain field in the East. These three factors are important in all theories on the cause of seismicity in the East and are important in differentiating among the theories. The one common aspect to virtually all theories on the cause of eastern seismicity is the reactivation of previously existing structures. Thus, the crustal structure studies defined are the types of structures present; the stress field will define the type of motion that can occur (or if motion can occur) on the structures; and strain measurement detected if motion occrured and if it is still active.

The seismotectonic problems and the proposed geologic / geophysical projects in each of the areas is described below.

Crustal Structure of Florida, Georgia, South Carolina and North Carolina Magnetic surveys and borehole penetration demonstrate that the structure of the crust under the coastal plane in northern Florida, southern Georgia, South Carolina, and southern North Carolina is distinctly different from the crust exposed in the Piedmont of Georgia and the Carolinas and is alss different from the crust beneath the coastal plane north of central North Carolina. Rankin called this the Charleston Terrain. It is interpreted to be an old continental rift terrain such as the East Africa Rift System on the basis of lithology of the borehole that have penetrated it.

The structure of this basement is important. If it is a rift then this could be a cause of the seismicity in the region. Two types of structural studies could be conducted: (1) magnetic and gravity, and (2) deep seismic refraction.

Seismic reflection profiles have been conducted in several areas; however, the technique is not sensitive to strike slip motion along faulis and does not penetrate deep enough to define a rift. Also, there is a basalt layer in Charleston which is difficult to penetrate by seismic reflection and this masks structures beneath it.

Crustal Structure in New Brunswick Earthquake Epicenter One of the most critical areas for study of crustal structure is the area of the New Brunswick Earthquake. That earthquake and its hypocentral area are known well. The process of earthquake generation in that area will be significantly clarified with seismic profiling. Perhaps related to that is the neotectonically active areas in the Pasamaquada Bay area of Maine. Its relation to the New Brunswick Earthquake and the Bay of Fundy area is not clear.

1 A-28

The Pasamaquada activity may be related to reactivation of the Bay of Fundy Rift. Seismic reflection is probably the best method for crustal studies in the New Brunswick area, though seismic refraction may be the best method in the Bay of Fundy region.

Stress Measurements in the Eastern United States The current STPP plans for three stress measurements in FY 83/84. Several more are planned for FY 85/86. However, these measurements are relatively shallow, 0300m. More measurements are necessary to define the stress condition in the East.

Even with the five or so measurements the regional patterns will not be known well. Also, deeper measurements are necessary. The measurements to 300m are reasonable but 1000m or 1500m measurements in earthquake hypocentral areas have been very helpful in defining seismogenic mechanics. Potential sites for movements are: (1) Triassic Basin site; (2) an eastern overthrust splay site; (3) the New Brunswick epicenter site; (4) the Pasamaquada Bay area; (5) the

-Giles County site, and (6) the Charleston area.

Strain Measurements The proposed seismogenic mechanism should be systematically investigated for svidence of strain, both geodetic and geologic (such as terrace movements, and paleoliquefaction features). This would involve examining site for strike slip faulting and also for vertical motions. The sites should include: (1) a Triassic Basin; (2) overthrust zones and splays to them, and (3) Mafic Plutons.

1 A-29

l l

l l

l 8

Appendix B Soil Response Project NRC Project Manager: Jacob Philip

SOIL RESPONSE PROJECT A. INTRODUCTION

1. General The NRC geotechnical engineering staff of the Office of Nuclear Reactor Regulation (NRR) reviews information that is presented by applicants concerning the properties and stability of soil and rock formations which may affect the safety of nuclear power plant facilities under both static and dynamic condi-tions. The dynamic conditions include the vibratory ground motions associated with the Safe Shutdown Earthquake (SSE). The review encompasses several spe-cific areas; the geologic features in the vicinity of the site, the static and dynamic engineering properties of the soil and rock formations underlying the site, the responses of site soils or rocks to dynamic loading, the liquefaction potential of the soil and its consequences including the resulting settlement of structures, the change in groundwater conditions and the piezometric pres-sure in all critical strata as they affect the loading and stability of founda-tion materials.

The Office of Nuclear Regulatory Research, Earth Sciences Branch, has been sponsoring research to develop state-of-the-art information in the area of soil liquefaction and dynamic settlement. The purposes of the research are to:

(a) define the major factors influencing seismically induced liquefaction and settlement of soils; (b) describe and evaluate current methodologies having the potential for predicting seismically induced settlement; (c) validate the models for use in the analyses.

2. Issues The fundamental regulatory issue that needs to be addressed regarding the geotechnical engineering aspects of nuclear power plant facilities in the Seis-mic Safety Research Plan concerns the stabilityoof the facilities when subject-ed to seismic loads equal to or greater than the SSE.

In November 1982, the U.S. Geological Survey (USGS) noted that it has not l been able to associate the Charleston earthquake of 1886 with a known geologi-cal structure. There is a probability, however very low, that the level of ground motion associated with a Charleston size earthquake could occur else- l where in the eastern seaboard. Furthermore, high se.ismic accelerations have recently been recorded at New Brunswick, New Hampshire and Arkansas. Under present siting criteria it becomes possible therefore to postulate ground mo-tions higher than the SSE at many reactor sites in the eastern United States.

Consequently, the question of stability of the nuclear power plant structures, B-1

systems and components resulting from possible liquefaction and dynamic settle-ment becomes a matter of concern, and the development of validated dynamic models to predict the phenomenon becomes essential.

NRR reviews Probabilistic Risk Assessments (PRA) submittals from appli-cants. Geotechnical engineering related.PRA reviews are included in the section on External Events. Geotechnical factors affecting the stability of power plant structures, systems and components vary and should be considered as being uncertain in magnitude. For example, soil properties are highly variable since soil deposits are neither homogeneous or isotropic. In the conventional methods of _ evaluating settlement, liquefaction, bearing capacity, etc., average or conservative values of soil properties are either estimated or taken from laboratory test information and these values are then used along with other parameters to obtain safety factors against' failure. The conventional site specific and deterministic engineering design approach should be broadened to-include probabilistic methodology in order to suitably account for the vari-abilities in geotechnical parameters (e.g., site exploration and investigation programs, variations in material properties, soil or rock stratigraphy, etc.)

and as input for estimating nuclear plant fragility due to seismic motions.

B. BACKGROUND

1. Assessing seismically induced liquefaction and settlements in soils.

The current simplified assumption of earthquake loading in a scil mass is that the earthquake induces oscillating shear stresses and strains from the horizontal components of motion with the primary effect of inertia loading, on the vertical component of motion. Soil response to earthquake loading is com-plex and nonlinear. Depending on soil type and site parameters, soil may ex-perience liquefaction and/or settlements. Methodologies for analysis or prediction of liquefaction and settlements must take into consideration the major variables influencing this behavior. These variables include soil type, relative density, nonlinear pore pressure effects, the magnitude and intensity of the dynamic loads, etc.

Most of the dynamic analysis methods presently in use for the one dimen-sional response of horizontal soil deposits are based on total stress, which at any point on a section through a saturated soil, is the summation of the water

pressure and the effective pressure due to intergranular stress. Deformations are controlled by effective stresses, and these methods of stability analysis are widely used in static problems. Until recently, no dynamic effective stress analysis techniques existed for lack of a model to predict the porewater l

pressures developed by seismic loading. Analysis of total stress does not di-

-rectly yield information on porewater pressures nor the effect of strain sof-tening which results from reduced effective stress which results in decrease of the shear modulus. A nonlinear effective stress model that adequately couples porewater pressure generation, material softening, and porewater pressure dis-sipation can rationally treat those aspects of the problem of seismically in-duced liquefaction and settlement.

The desirability of an effective stress model for the dynamic analysis of nuclear power plant structures takes on added importance with the November 1982 position of the USGS on the 1886 Charleston earthquake; that the level of ground motion associated with the Charleston earthquake could occur elsewhere l

B-2

on the eastern seaboard. Since this could result in the postulation of ground

! motions several times the SSE at many reactor sites in the eastern U.S., many of which are founded on soil, the development of a validated model for studying liquefaction and dynamic settlement is the objective of ongoing research. The analysis techniques should be supplemented with additional research on in situ l methods for assessing liquefaction potential and settlement.

In addressing the issue of soil response to earthquakes, NRC has utilized the experience of the Corps of Engineers (C0E) Waterways Experiment Station, Vicksburg, Mississippi. COE has extensive experience in both analytical and experimental techniques for evaluating .he effects of soil response to seismic loads. Presently they are conducting a detailed study of soil failure models and are conducting verification experiments at Cambridge University using soil samples tested in a centrifuge apparatus. The research will extend through 1986.

The objectives of the experiments are to assess seismically induced lig-uefaction and settlements in soil. COE is also conducting research to develop a state of-the-art report on seismically induced liquefaction. This latter project will be completed in FY 86.

Work on in situ techniques for the study of liquefaction potential of soil is being conducted by the University of Texas, Austin, for the USGS. The tech-niques involve the characterization of the subsoil strata using high and low frequency profile Raleigh of the waves and the development of a continuous shear wave velocity subsurface. The method has been found to be useful to depths of about 25' below the surface and continuing research may extend the application of the technique to depths in. excess of 100'. The experiments are being con-ducted in areas known to have liquefied. The results are used with nonlinear effective stress dynamic analysis models to correlate shear wave velocities with varying levels of acceleration to estimate the potential for liquefaction.

This is similar to the well known Standard Penetration Testing (SPT) versus ground acceleration charts developed by Seed and Idris. The Earth Sciences Branch is monitoring these developments.

2. Probabilistic methodologies in geotechnical engineering The conventional method of evaluating the stability of nuclear plant structures is to use a deterministic site-specific approach in which conserva-tive estimates of soil properties based on available test results are utilized along with the appropriate site parameters to compute safety factors against bearing capacity and liquefaction failures and for estimating settlements. A safety factor equal to unity indicates critical stability while factors greater than unity imply the presence of a margin of safety.

An assessment of the general stability of nuclear power plant structures, systems and components reached based on deterministic factors of safety are only as reliable as the data used. The geotechnical engineering parameters that impact the analysis include adequacy of site exploration and investigation programs, strength properties of the foundation soils, potential for settlement or liquefaction, the existence of collapsible or dispersive soils and aging of the facilities.

The reliability of the geotechnical parameters used for conducting analy-sis pf the stability of structures is vital particularly when the factor of safety approaches unity. For instance, when boreholes are drilled, information B-3

on the soil properties are derived from tests from a limited number of samples, that may be quite variable. Since the cost of extensive drilling of boreholes, sampling, testing, etc., is high, research in the use of probabilistic methodo-logies in geotechnical engineering becomes desirable. These methodologies must incorporate the uncertainty and variability of the basic controlling geotechnical parameters in so far as possible.

The development of probabilistic techniques would provide additional tools to evaluate the geotechnical engineering aspects of an applicant submittal. It would also be particularly useful in evaluating older plants under the System-atic Evaluation (SEP) and Integrated Safety Assessment Program (ISAP).

Presently there are no NRC funded research programs for geotechnical probabi-listic assessments for nuclear plant structures. There is a paucity of re-search on the subject and the scientific community has only very recently begun to examine the use of probabilistic techniques in geotechnical engineering.

The Earth Sciences Branch has had discussions with the COE on probabilistic methodologies for the assessment of soil failure and has solicited a proposal.

C. PROGRAM PLAN The program plan to address the technical issues described in the previous section is presented here. The first deals with the stability of nuclear power plant structures, systems and components when subject to seismic loads at or higher than the SSE which involves research into both analytical and field techniques. The second involves probabilistic methodologies applicable to geotechnical engineering reviews of nuclear power plant structures, systems and components.

1. Performance of nuclear power plant structures, systems and components subjected to large earthquakes motions 1.1 Soil settlement 1.1.1 Research by the Corps of Engineers Acceleration and frequency of input motion have significant effects on soil behavior. Typical results indicate that doubling the ground acceleration can increase the porewater pressure by as much as 70 percent and increase the settlement by more than a factor of three.

A study done for the NRC by the COE describes various currently available models for evaluating seismically induced settlement. The study constituted Phase I of a three phase investigation, and recommended the settlement model formulated by Professor L. Finn of the University of Brit.ish Columbia as the most complete and directly formulated methodology.

Most settlement models including Finn's required input parameters obtained from cyclic simple shear tests. Since data from these tests are not usually available, methodologies for the use of the more readily available cyclic triaxial test data in the various models is being investigated in Phase II.

For Phase III centrifuge tests have been conducted and the data are being analyzed for verification of methodologies for predicting and analyzing l

seismically induced settlements.

I B-4 l I

Phase III includes two primary subtasks:

(a) Phase III-A: Centrifuge tests have been conducted with earthquake simulation for conditions of level ground and of embedded structures simulating nuclear reactor buildings to generate equivalent response data for analysis. Testing to determine the dynamic properties of the sand used in the centrifuge apparatus will be performed.

(b) Phase III-B: In this phase the test results will be reduced and analyzed to evaluate the capabilities of all previously identified (in Phase 1) methodologies to model the prototype response data, with the primary focus on the Finn dynamic analysis technique.

The physical tests (Phase III-A) were cen'trifugal tests with seismic simulation loading. These tests were conducted at Cambridge Univer-sity, in Cambridge, England. Internal measurements were made by the use of accelerometers, porewater pressure transducers and displace-ment transducers. Realistic prototype in situ gravity loading stress fields were obtained both in the free field as well as beneath the model structures. The earthquake excitations were conducted in both increasing and decreasing order to determine shear strain history effects. Both dry and saturated specimens were tested.

In addition to centrifuge tests, engineering properties of the sand used in the centrifuge will be necessary, including such tests as cyclic and conventional triaxial loading, cyclic simple shear, and resonant column or torsional testing. Analyses of these centrifugal test results and sand property determinations will provide data for comparing theoretical calculations with laboratory controlled physi-cal performance tests. The test results will be used to validate settlement prediction models identified in Phase I of this study and to assess their capabilities to predict or analyze seismically in-duced settlements.

1.1.2 Research at the HDR facility, Federal Republic of Germany The staff is following the testing of the integrity of the containment building at the HDR facility located about 50 km SE of Frankfurt, Federal Republic of Germany (FRG). It is planned to subject the containment to vibra-tory motion, simulating earthquakes, in 1986. The testing covers the generation of data concerning the soil response to the containment motions. The advantages of gathering such data are many and include, (1) obtaining the freefield response of soil in the vicinity of an actual full size containment building, (2) compar-ing the response to that of soil around structures modeled in the centrifuge to represent the prototype, and (3) analyzing the data for input soil structure interaction.

Argonne National Laboratory is the NRC coordinator for the effort at the HDR facility. A series of tests were conducted at the facility during 1975-83 involving experiments on materials engineering, thermal hydraulics and mechanical and earthquake engineering. The dynamic properties of the subsurface soils B-5

have been determined and the site has been instrumented with accelerometers, pore pressure devices and ceformation gauges. Preliminary soil response data from the vibratory tests on the containment, planned in mid 1986, will be made available to NRC by the end of calendar year 1986. Analysis of the data and comparison with the centrifuge experimental data conducted under 1.1.1 above can be useful in the validation and prediction of dynamic settlement and liquefaction potential.

1.1.3 Research at the Electric Power Research Institute (EpRI) Sita, Taiwan EPRI is constructing a model about 1/4 the size of a concrete containment in a seismically active area in Taiwan. Argonne National Laboratory is the NRC coordinator for this effort. Instruments installed by EPRI in the model and in vertical and horizontal arrays in its vicinity, will record responses to earth-quake over a five year period. Taiwan Electric Power has the responsibility to monitor the instruments. NRC has committed to perform low-level vibratory tests of the model to provide baseline data on model parameters. Future NRC effort will be in providing analytical models to predict soil-structure interaction effects for the recorded earthquakes.

The staff is following the EPRI research program to explore the feasibility of expanding the study to obtain additional data on soil response. The objec-tives of the latter is similar to that outlined in 1.1.2 above viz., comparison with centrifuge data (Section 1.1.1) and consequently the validation and prediction of dynamic settlement and liquefaction potential.

1.2 Soil liquefaction The Corps of Engineers have in the past, conducted research for the NRC to evaluate the state-of-the-art of earthquake-induced soil liquefaction. A work-ing draft report on the research was submitted to the NRC in November 1984, for comments and suggestions. The report was reviewed by appropriate NRR and RES staff members and their comments transmitted to COE. Final completion of the report and its publication as a NUREG document is expected sometime in 1986.

In a related program the National Science Foundation at the suggestion of the NRC requested that the National Research Council convene and conduct a workshop and seminar on liquefaction. The purpose is to review the state-of-the-art research and submit recommendations on the most promising research approaches to resolve critical technical issues relating to soil liquefaction.

The National Research Council, through the Conmittee on Earthquake Engineering proposed to convene and conduct a workshop and seminar of recognized experts in liquefaction of soils, under earthquake loadings, and prepare reports based on the workshop and seminar discussions. The workshop was conducted at MIT in the spring of 1985. The seminar is scheduled for 19 September 1985 in Washington, DC.

Reports resulting from the workshop and seminar will be prepared in sufficient quantity to ensure distribution to the various federal agencies, including NRC, and to the engineering community, researchers and educators concerned with earthquake induced liquefaction.

B-6

7 I

It-is envisaged that funding by the NRC in the form of a grant to support the workshop would be confined to costs incurred in final editing, printing and distribution of reports.

1.3 In situ methods for prediction of soil liquefaction potential and settlement

-A promising in situ technique for the study of liquefaction potential of soil deposits is being currently followed by the Earth Sciences Branch. The technique uses low and high frequency Raleigh waves and were employed at several liquified soil sites in the Imperial Valley, California by the University of Texas, Austin for the USGS. By utilizing a fast Fourier transform and spectral analysis, Raleigh wave velocity for each frequency is calculated. With inver-sion, shear wave velocity, shear modulus and layering of the media are deter-mined. The shear wave velocity profile of the subsurface strata is then correlated with varying levels of earthquake accelerations from dynamic analy-sis to predict the potential for liquefaction. This method has been found to be an accurate measure of shear wave velocity to depths of up to 25 feet.

NRC participation and/or support of this research project has not yet been defined and discussions are in progress with the USGS and the University of Texas on the program. NRC interest in the project would be confined to that research which would extend the applicability of this method to depths of 100 feet or more.

The duration of the program and costs can only be roughly estimated at this time. Definitive estimates of the NRC contribution to the research have not yet been determined.

2. Probabilistic methodologies for geotechnical engineering license reviews The objective of this study is the development of probabilistic techniques to be applied to geotechnical engineering license reviews for nuclear power plants. The probabilistic approach can be a very useful tool which minimizes the pressures to make decisions based on average conditions while not eliminat-ing engineering judgement and the use of all existing information.

Factors affecting the geotechnical engineering aspects of nuclear power plant structures, systems and components are variable, and should be considered as being uncertain in magnitude. These parameters include the adequacy of the site exploration and investigation programs, geotechnical engineering proper-ties of soils and rock, the potential for subsidence, the presence of collaps-ible or dispersive soils, dynamic settlement and liquefaction, aging of the facilities, etc. The probabilistic model should address and incorporate the uncertainties and variabilities of the factors contributing to the stability.

Sensitivity analysis should be considered for several profiles in order to de-termine the governing variables to be considered in the probabilistic model.

As an example, in the dynamic bearing capacity analysis, if the sensitivity analysis indicates that the cohesion intercept C', the angle of internal fric-tion (', the location of the groundwater table, location of the soil interface lines, and the earthquake accelerations were the governing variables, the anal-ysis can be conducted by varying one variable while holding all the remaining B-7

variables constant. If other variables in the analysis were assumed to be mea-sured accurately then a-sensitivity analysis need not be conducted on these

-variables, as they would not, in this particular case be classified as govern-ing variables.

Probabilistic analysis techniques account for uncertainties in geotechnical engineering parameters and besides other uses could conceivably quantify seis-mically induced settlement and liquefaction risk to estimate plant fragility for use in seismic Probabilistic Risk Assessments. It could provide the licens-ing office with a methodology fo- reviewing nuclear power plants, particularly the older plants where complete deterministic analysis would be both inappro-priate and costly.

Based upon the desirability of probabilistic techniques for the analyses of nuclear plant structures, additions and revisions to licensing regulations, procedures and associated regulatory guides should allow for the use of proba-bility methodologies. A comprehensive check list for evaluating geotechnical aspects of the probabilistic analyses should be developed as well as a user manual for NRC staff use.

Definitive estimates on the level of funding can be provided only after discussions with consultants and contractors after development of a work state-ment and request for proposals.

D. ADDITIONAL RESEARCH NEEDS IN GEOTECHNICAL ENGINEERING The research projects identified in the Appendix to the Seismic Safety Research Plan (SSRP) are those most needed to resolve the fundamental issues that are required to be addressed regarding nuclear power plant structures. In addition we present a listing of research that needs to be performed in the future in order that NRR may more expeditiously evaluate the technical contents of applicant documents leading to licensing decisions about existing or planned nuclear plant structures, systems and components.

Listing of research needs 2.1 Influence of multiple structures on settlement determination 2.2 Engineering procedures and criteria for determining acceptable limits of differential settlement 2.3 Determination of subgrade modules for layered soils 2.4 Effects of varying support and backfill conditions on dynamic response of buried pipes 2.5 Influence of engineering properties of in situ soils on amplification /

attenuation of near field earthquake motions 2.6 Compressibility of residual soils B-8

Appendix C Structural Response Project NRC Project Manager: Roger M. Kenneally

STRUCTURAL RESPONSE PROJECTS A. Introduction

- The issues of safety as related to nuclear power plant Category I buildings have, in the past several years, shifted from the design of new buildings to issues related to operating plants. The principal issues related to operating plants center around aging effects and changing criteria. The Seismic Category I Structures Program was established at the Los Alamos National Laboratory to address the licensing issue: can existing facilities continue to operate in light of more demanding criteria and potential changes in operating modes than those that were considered in the initial design?

The principal users of the research results from the Seismic Category I Structures Program will be the Systematic Evaluation Program Branch, the Equipment Qualification Branch, and the Structural and Geotechnical Engineering Branch. The needs identified by the staffs of these Branches are:

1. understanding the behavior of Category I structures (other than the containment) subjected to earthquake loads beyond design.

The behavior of the structure loaded such that both elastic and inelastic response occurs is of interest;

2. identi fy the changes in fl.oor response spectra at various magnitudes and duration of input loading; and

3. identify the changes in damping associated with the various magnitudes and duration of input loading.

Based on the licensing issue and the user office needs, the (Ebjectives of the Seismic Category I Structures Program are:

1

1. to address the seismic response of Category I structures other than containments;

, 2. to develop experimental data for determining the sensitivity of structural behavior in the elastic and inelastic response range of Category I structures to variations in configuration, design l practices and earthquake loading;

3. to develop experimental data to enable validation of computer i programs used to predict the behavior of Category I structures i

during earthquake loading that causes elastic and inelastic response;

4. to identify floor response spectra changes that' occur during earthquake loading that causes elastic and inelastic structural response; and C-1

5. to develop a methoa for representing damping in the inelastic range, and demonstrate how this damping changes when structural response goes from the elastic to the inelastic ranges.

The Seismic Category I Structures Program is a combined experimental / analytical program with the initial work heavily emphasizing the experimental aspects to develop a good data base. In the latter phases of this program the emphasis will shift toward more analytical development with verification experiments used to validate the sensitivity of certain significant geometry or configurational changes. Existing finite element codes are used to predict structural response and to design tests. In addition, the structural code (INRES-B) developed by Dr.

Frankling Cheng of the University of Missouri-Rolla has been used and will be further developed tu p. edict the dynamic response of model Category I buildings.

The failure mode or load of a Category I structure in itself may not be the most significant product from this research. The reduction in stiffness of a structure as it is loaded into the inelastic range can mean increased equipment or piping response amplitudes because of frequency shifts induced in floor response spectra. Safety factors and calculated margins may not only be inaccurate but can be reduced significantly.

Thus, the information derived from this program will support other ongoing research and licensing activities and resolve issues that will arise in other branches of the USNRC.

B. Background Domestic and foreign research that has been conducted, is currently underway, or planned that may have applicability for determining the seismic behavior of Category I buildings is summarized in this section.

1. USA Seismic investigations to understand structural behavior, and develop design criteria for commercial multi-story frame structures that may or may not include shear walls have been conducted at many different institutions in the USA. The greatest number of investigations have been strictly analytical; however, a significant number of experimental investigations have also been conducted. Because a seismic Category I structure is shear-wall dominated, only a few of the investigations have applicability to the needs addressed by this program. There are a number of experimental investigations that have been and are being performed that have direct applicability to reinforced concrete containment buildings. The Seismic Category I Structure Program excludes consideration of containment buildings, but keeps abreast of the literature for results that can be applicable. A summary of the research performed by the major institutions that have been active in testing of reinforced concrete elements and structures is given below.

C-2

1.1 Construction Technology Laboratories (CTL)

The CTL, located near Skokie, Illinois, performs both analytical and experimental investigations on reinforced concrete elements and buildings.

It has extensive experience in the construction and testing of model reinforced concrete structures. Of particular pertinence to the Seismic Category I Structures Program are the CTL tests on isolated shear walls.

The purpose of these investigations was to develop strength relationships that could be used in the design of shear walls. These shear walls had different aspect ratios, reinforcing amount and arrangement, and different boundary conditions. The shear walls were statically loaded, both monotonically and cyclically. All tests at the CTL use static loading.

1.2 The University of California at Berkeley (UCB)~

Both large scale static tests and dynamic tests are conducted at UCB.

A large shake table is located at UCB and is accessible to the University researchers. The frequency limit on this shaker is about 20 Hz, thus giving the shaker limited utility for structural modeling on this program because our model structures require excitation frequencies up to 200Hz to fulfill the similitude requirements. The UCB has performed considerable testing related to concrete structures; however, the testing has been directed toward determining the behavior of frame buildings and has little applicability to shear wall buildings. The tests at UCB have included the load cycling of beam-column connections, earthquake loading of small masonry buildings, and earthquake loading of a 1/5 scale reinforced concrete frame-shear wall building.

1.3 UniversityofIllinois(UI)

The UI researchers have considerable experience in the construction of reinforced concrete scale models, and have contributed their experience to this program through participation in the technical review group.

The UI has capabilities for both static and dynamic testing. Large load capacity for quasistatic loads exists, but the capacity of their seismic simulator is limited. Many of their static test results on reinforced concrete structural elements found in frame buildings have been incorporated into the ACI Building Code.

Tests that have been performed on their seismic simulator include scale models of frame buildings containing a limited number of shear walls. The correlation of anayltical-experimental comparisons from these tests are used in planning our research to ensure credible results are obtained.

1.4 Cornell University (CU)

The testing of reinforced concrete structural elements has been limited to static loading at CU. Many of their tests have used scale model structures and CU researchers have extensive experience in the design and construction of scale model reinforced concrete structures. We C-3 m

l l

l are not aware of any tests on shear wall structures; however, they have conducted tests that are directly applicable to containment technology.

1.5 University of Texas (UT)

The testing of full-scale reinforced concrete building elements under static loading has been ongoing for many years. The tests were conducted to investigate particular aspects of reinforced concrete behavior--such as connections, reinforcing bar splice lengths, bond strength, etc. Much of their research results have been incorporated into the ACI Building Code.

Very little of UT's research results are directly useful in evaluating the seismic capacity of Category I buildings.

1.6 Classified Reports A data base scan was made of restricted and/or classified reports on reinforced concrete investigations. These reports are classified because of the applications and not because of concrete behavior characteristics.

Many of these investigations used model reinforced concrete structures.

The loadings used in the tests were mostly quasistatic. Dynamic loading, when used, considered blast loads. No new information that could be applied to Category I buildings was discovered in these classified reports; however, these reports contained investigations that could be applicable to reinforced concrete containment buildings.

2. Japan The Japanese have performed a large number of tests on model shear wall structures for the purpose of establishing design criteria and to show that the structures do not fail at loads less than their design loads. These model buildings had either square or circular cross sections. The reinforcement ratio was varied and either a monotonic or cyclic load applied. In addition, a large number of tests were shape, performed on model structures with openings. The opening size, and peripheral reinforcing were varied. The test results were used to establish rules for the design of prototypical buildings. The Japanese have not always tested their structures to failure, and the information reported at conferences is not always useful to support other purposes.

The Japanese have dynamically and statically tested 1/30 scale models of a prestressed concrete containment vessel and components of a reinforced concrete reactor building. The model vessel and reactor building components were both loaded into the inelastic range. The model reactor building components included shear walls. The stated purpose of these tests was to investigate the safety factor for the design load and the lateral load displacement relationship of each component. The components were models of the outer box, the inner box, and the shield wall. The principal data published were the load-deflection relation-ship for cycled loadings (quasistatic).

The containment vessel was tested both statically and dynamically, the dynamic loading being artificial earthquake motions applied by a shake Generally, table. Also published were results of analytical studies.

C-4

where appropriate data can be compared, trends being observed in the Seismic Category I Structures Program are being confirmed by the Japanese studies.

The Japanese are currently doing research that may directly compliment our effort. For example, one recent publication appears to compare test data with elastic analysis and put limits on the " scope of their applicability." Another paper of interest is the pseudadynamic method that is being used to test larger structures. In this method a dynamic analysis of the structure is used to predict the response at various locations to a given base motion. Given this response, computer controlled hydraulic actuators load the structure at various points resulting in a presumably correct relative displacement field within the structure. Such methods may be of use in our program and they are currently under study.

A better understanding of past, present and future research efforts in Japan may be obtained when a technical exchange agreement between the U.S. NRC and the Japan Ministry of International Trade and Industry (MITI) is established.

3. New Zealand A number of experimental studies of structures subjected to seismic loading have been conducted in New Zealand, principally by T. Paulay. Of particular interest are their experimental studies on shear walls. Their shear wall studies did not use real or scaled time earthquake motions, but used monotonic and cyclic quasistatic loading to obtain failure mode, cracking patterns, stiffness and strength degradation, and ductility.

Shear wall aspect ratios, and the arrangement and amount of reinforcing steel were varied in their tests. Their tests were aimed at usual building type of walls, but if one considers their test structures as models, it is possible to relate their results to Category I buildings.

The static tests on the model structures in the New Zealand tests showed the same behavior trends as observed in the Seismic Category I Structures Program.

4. Germany l Shaker tests at the Heissdampfreaktor (HDR) facility in the Federal i Republic of Germany (FRG) may aid us in future program planning and l evaluation of experimental results. The HDR is a decommissioned l superheated steam reactor that has been modified to serve as a test facility. In June-July 1986, the containment building will be excited by a large shaker mounted on the operating floor. Results from this test will provide vibratory information on the shear walls contained in an actual nuclear facility.

C. Program Plan

1. Background and Past Accomplishments In recognition of the increasing need for research to support work on seismic related issues within the USNRC, the Seismic Category I C-5

Structures program began in FY 1980. The first task was an extensive literature survey on research activities applicable to or related to Category I structures. Major A/E's of nuclear power plants were contacted and a number of general drawings on different power plants were obtained.

From this information, an initial program plan was developed. A Technical Review Group (TRG) composed of nationally recognized experts was assembled to aid in the planning and to comment on the progress and direction of the program. No major revisions to the initial program plan were made as a result of the first TRG and NRC staff review and the program plan was published. (Reference 1).

Category I structures are constructed mainly from low aspect ratio shear walls as opposed to structural elements whose predominant behavior is governed by bending. Thus, the first experimental tests at Los Alamos were carried out on isolated model shear wall structures. Static tests were conducted to obtain strengths and load-deflection behavier.

Vibration (sine-sweep) tests were than conducted to obtain natural frequencies and equivalent viscous damping values. The results of these tests were inconsistent, and it was concluded that traditional sine-sweep tests at measured load levels on reinforced concrete structures tend to overfatigue the structure resulting in confusing data and should be avoided. Thereafter, all dynamic tests used either a recorded or a computer generated accelerogram as the base input motion.

Both the static and seismic tests using the isolated shear wall structure (about 1/30 scale) indicated that the initial stiffness was significantly less than the stiffness computed assuming an uncracked concrete cross section. These results generally agreed with results reported in the literature where structures were tested to comparable load levels.

The initial program plan was developed with foreknowledge that scale model testing of reinforced concrete structures is a controversial issue in the civil engineering community, particularly when the structures are loaded into the inelastic range. Two different scale model sizes of isolated walls were planned to demon'.trate scaling to be followed by tests on 3-dimensional box-like structures. The TRG recommended that the scalability demonstration should be carried out using the box-like structures. Thus, using all the information found (stiffness, damping and inelastic behavior) and the testing and modeling technology that was obtained from the isolated shear wall test phase, a revised program plan was developed incorporating TRG guidance as well as addressing new (current) user needs. This test plan was directed primarily at the use of box-like structural models.

The actual seismic testing of the box-shaped test structures commenced in FY 1983. Quasistatic loads, both monotonic and cyclic, were applied parallel to the long (longitudinal) direction and to the short (transverse) direction of the box-shaped test structures. The static tests yielded load-deflection information (stiffness, " yield" point, inelastic behavior, ultimate load carrying capacity and failure mode) on these structures. Failure of the test occurred at the base (foundation)

C-6

i

(

l and wall interface. The concrete developed a penetrating crack around the model building perimeter. The same basic failure mode was observed in the >

isolated shear wall tests.

i Seismic tests were conducted on one story box-shaped model structures

! to obtain their fundamental frequencies and to observe their general i

behavior (crack patterns, freqeuncy shifts, peak responses, damping, i etc.). These seismic tests also showed the reduced initial stiffness over the computed uncracked cross sectional value and large equivalent viscous i damping values that were observed in the isolated shear wall tests.

9 i

To verify that the scaling relationships could be used to translate 1 test results to difference size structures and to obtain general

l. structural behavior, two 1/30 scale (one inch wall) and one 1/10 scale 4 (three inch wall) two-story model Diesel Generator Building structures

were seismically tested. The first 1/30 scale model structures was tested l to aid in the development of the test program for the 1/10 scale structure. After the 1/10 scale model test, the second 1/30 scale model i

was tested in a manner similar to the 1/10 scale model. The results to date indicate that the scaling relationships were adequate to predict the behavior of these modeled structures.

i i To illustrate this point, Fig. I shows the data taken during FY 1983 I from tests on two 1/30 scale model Ofesel Generator Buildings (3D-10-2 and

' 3D-11-2) and one 1/10 scale model (CERL No. 1). When the measured first '

mode frequency is normalized by the frequency scale factor, N,, and the  :

peak acceleration is normalized by the acceleration scale factor, N, the i data can all be plotted on the same curve. (In this notation, the Yscale 1

factor indicates the ratio of the prototype to the model). In addition,

the test conditions had the appropriate added masses and the base motion properly frequency scaled such that the 1/30 scale structure is a 1/3 e

scale model of the 1/10 scale structure while both structures are models

! of the assumed prototype. When the data is illustrated in this manner, the prototype behavior is shown directly, while the individual model data requires knowledge of the scale factors (1/30 scale: Nf = 1/11.8, N'j=1/4.6and1/10 scale: N f=1/6.8,Ny=1/4.6).

Clearly, from Fig.1, the scalability of the two models themselves is demonstrated, but because both models are made of microconcrete with simulated rebar, scalability to the prototype structure is still an issue. ,

The lower than expected initial stiffness or scalability issue is further addressed in Fig. 2. This figure explains how data taken from

, previous experiments (both static and dynamic) can be used to deduce the stiffness of the structure. The figure illustrates the secant stiffness i plotted against the concrete modulus, E . The secant stiffness was taken at 50% of the ultimate load (meas 6 red from experimental results) normalized by the structure's theoretical value calculated from an t uncracked cross-section strength-of-material approach. The concrete modulus was obtained from the equation c E =57000[asrecommendedinACI i 349 for normal weight concrete. With the exception of a single point (a

" wet" test in an aging study) the data consistently show that calculated I

! C-7 1

stiffness are down by a factor of 3 or 4 at this load level. Similar differences have been reported in some parts of the literature. On the other hand, values reduced by 20% or less have been indicated in other part of the literature.

Several questions now arise. How good or credible ~are our previous experimental data because of microconcrete and modeling effects? What is the appropriate value of the stiffness that should be used in design and component response spectra computations in these structure ? Should it be a function of load level? Have the equipment and piping in existing buildings been designed to incorrect response spectra?

Primary program emphasis is to obtain credibility of previous experimental work by beginning to resolve the stiffness difference issue.

The Technical Review Group (TRG) for this program believes that this important issue must be addressed before program objectives can be accomplished.

2. Current and Planned Research To address the alove "stlffness related concerns a series of credibility experiments will be carried out using both large and small scale structures. For the large scale structure, the TRG set priorities on the design. Their recommended " ideal" credibility structure characteristics in order of decreasing priority are as follows:

1. maximum predicted first mode natural frequency - 30 Hz,

2. minimum wall thickness = 4 in,

3. height to depth ratio of shear wall 51,

4. use actual #3 rebar for reinforcing,

5. realistic material for aggegrate,

6. 0.1 to 1% steel (0.3% each face, each direction ideally),

7. water blasted construction joints to assure good aggregate frictional interlock.

They further agreed that the best plan is to build two of these structures as identical as possible. To compare the results from these tests with previously obtained data, the first model should be tested quasistatically and cyclically to failure; the second model should be j tested dynamically.

2.1 Design of the TRG Model i

j Following these recommendations, other TRG suggestions and after

analyzing a number of potential designs, the structure shown in Fig. 3 was proposed to both the TRG and NRC as being a test structure fulfilling the l C-8

e design requirements. Table I gives some of the details of this structure.

After resolving a number of questions relating to the details and the potential response (out of plane bending of walls, torsion, etc.) of the structure, the decision has been made to construct and test this particular configuration and its models.

TABLE I COMPUTED CHARACTERISTICS OF THE TRG MODEL STRUCTURE I

uncracked transformed section = 2.06 x 10 6 in.4 Aeffective shear = 379 in.2 Area total = 1288 in.4 7

Total uncracked bending stiffness = 3.5 x 10 lb/in.

6 Shear stiffness = 5.3 x 10 lb/in.

Total stiffness 6

= 4.4 x 10 lb/in.

Max dead weight normal stress = 43 psi Max shear stress in flange at 5 g due to assumed 5% torsion (approx.) = 35 psi Total concrete = 6 yards Total added weight = 37,600 lb Total weight = 60,800 lb 2.2 Design and Testing of a Scaled TRG Model Two quarter scale models of the large TRG structure having 1 inch walls, a single row of hail screen reinforcing and appropriate added mass will be constructed and tested prior to the construction of the large model. These tests will be used primarily to confirm conclusions from design analyses for the large structure regarding torsional effects and bending of individual end and shear walls, as well as to further demonstrate the scaling effects for these structures. Both low level static and dynamic testing will be performed. The low level tests will be used to derive virgin elastic stiffness information. One model will then be tested statically into the inelastic range and the other model tested seismically into the inelastic range.

If the models behavior is as expected, the large TRG structure construction will proceed as designed. If unexpected or unexplainable effects occur, modifications will be made and additional small models will be tested. Such modifications could include thicker walls to accommodate steel on both faces of the structure or a complete redesign. The TRG and the MC will be consulted regarding proposed modifications.

C-9

2.3 Testing of the TRG Models Because a large part of the purpose of these tests is to resolve the

" stiffness difference" issue, a great deal of low level as-built static and dynamic testing will be carried out on all of these structures and their models. The low level results will be used to detennine the following:

a. the as-built static stiffness at as low of a load as we can measure it,

b. the as-built first mode natural frequency at as low of excitation level as we can measure it.

Comparison of these two quantities from a single model will confirm and demonstrate the accuracy of our methods of stiffness determination.

Comparison of these two quantities between models will be used to characterize the stiffness variation that occurs from structure to structure even when efforts to construct identical models are attended to.

Following the low level testing, the models will be moved to a suitable test site for higher load testing. Prior to large load testing, however, the low level tests (static and dynamic) will be repeated to determine any changes that have occurred during shipping.

The static model will then be tested to failure using monotonically increasing loads with increments in load closely spaced enough to accurately determine stiffness as a function of load. The dynamic model will be tested from a low seismic input up to the capacity of the table.

Broad band low level tests between seismic tests will be used to assess response and permanent stiffness changes. Complete test plans for both large models will be available prior to testing. .

2.4 Specific Set of Contingency Plans Four specific scenarios have been identified as likely possible outcomes of the TRG experiments and they have been addressed accordingly.

(1) Results from the 1 inch and 4 inch TRG models agree with one another and are consistent with previous testing results. In this case, the program will focus on three parallel efforts. One of these efforts will be to estimate the magnitude and seriousness of the structural stiffness problem as it affects existing equipment and piping. These estimates can be made using relatively simple lumped mass structural models with the appropriately adjusted stiffness. Another effort will involve developing the theories that will describe the correct stiffness and damping and incorporating them into analytical models. A third effort will be to inform the nuclear civil structure industry with workshops, presentations, and publications. Two workshops are suggested. The first will inform industry about the problem and its impact as well as get industry feedback and opinion. The second workshop will deal with the computational models that are expected to come out of the results of this C-10

~

program. This specific scenario is the expected one. To expand on this thought, there is some evidence that the differences in stiffnesses may be due to the inherent nonlinear elastic behavior of concrete. If this is the case, the use of linear analysis procedures must be evaluated in terms the error made, i.e., is the error so large that they can not be usedL Also, in this case, all reported stiffnesses can probably be shown to be consistent, in that at low load levels, the agreement with theory will be within 20% or less whereas at high loads, the values may be low by 60-80%.

(2) Results from the small TRG models agree with prior experimental results concerning stiffnesses at high loads, but results from the large TRG models agree with theoretical uncracked cross section analysis. In this case, the conclusion is that microconcrete models can give trends, but must be corrected for material distortion effects to be used to predict the prototype structural behavior. This scenario will have several impacts on the direction of this program. First, a scheme for correcting our previous data will be sought. Second, future scale model testing will be limited to structures made from real construction materials. Because of seismic table capacity limitations these tests will have to be static and pseudo-dynamic in nature for any investigation of significant inelastic effects (margin studies). To expand our data base a 1/5 scale model of a Diesel Generator Building with 6 inch walls and No. 3 rebar will be tested seismically, but complete failure is not feasible. A combined test (dynamic damage followed by quaisistatic to failure) at the CERL facility of this model will be carried out. Because large model tests will be very expensive, the analytical effort will be intensified.

Since the type of behavior that the analytical model predicts should be scale independent if enough material properties are known and the correct physical behavior is captured by the equations, then a model developed that describes the true concrete structural behavior may still be verified using the previous data by inputting microconcrete properties.

(3) Results from the small TRG models agree with theoretical predictions but results from the large TRG models do not. This scenario is probably the worst possible outcome in that with what we know now, it is unexplainable. In this case, the entire program will have to be reexamined with regard to assumptions, scaling laws and effects, testing procedures, and analysis methods. Experts will be paid to " audit" the program in detail to identify those areas for which erroneous assumptions have been made or for which erroneous testing has been carried out. It will also be necessary to determine which data is useful or can be corrected. In view of the consistency of our past data, this scenario is deemed unlikely.

! (4) For both scales of TRG models, the low load level static data l agrees with theoretical predictions; high load level static data agrees l

with previous experimental results; low excitation dynamic data agrees with theory; high excitation dynamic stiffnesses for the large models are even lower than previous results. This scenario is possible, and means that a " spring" has been " missed" in the large test. The test data can be corrected analytically, if, the spring and its value is identified.

Another large test with more and different placement of instrumentation might be required. But in general, this case is a subset of scenario No.

I and program direction would proceed as described there.

C-11

2.5 Sensitivity Studies and Other Experiments Following the resolution of the stiffness difference issue, a limited number of experimental tests will be carried out to meet program objectives and aid in benchmarking the analytical model development. If settlement of the scalability and stiffness difference issues allow, these tests will be carried out on one-inch thick wall concrete models. A statistican, knowledgeable in experimental design will be used to comment on the test configurations reconmended and assure that the controlled variables (i.e., number of floors, wall arrangement, etc.) and uncontrolled variables (i.e., ccncrete strength) are incorporated into a cost-effective test matrix to meet program objectives.

Two models of a one-story building with a partial interior wall will be tested to transverse and longitudinal monotonically increasing loads to determine their load-deformation relationships. A single transversely-loaded one-story model building with one end wall thicker than the other will also be tested statically to obtain load-deflection behavior where torsional stiffness is important. Finally, a one-story model structure with an end wall opening will be tested. All four of these tests are also expected to yield information on ductility of the prototype buildings.

Seven seismic tests on one-story model structures with one-inch thick walls will form the effort to investigate sensitivty of structural behavior to earthquake input. Sensitivity of building response to a major wall opening (e.g., for a railway car), to a partial partition, to the presence of a nonsymmetrical closed section plan, and to different accelerograph inputs will be investigated.

Two model structures will be seismically tested both longitudinally and transversely to determine response sensitivity to buildings with wall openings and load induced torsional effects. Two model structures will be tested both longitudinally and transversely to determine the response Two basic model sensitivity structures (to buildingsplan symmetric withsection) a partialwill interior wall.

be reismically tested using completely different recorded earthquakes as input. Since the Technical Review Group pointed out that the 1940 El Centro accelerogram currently used in the program may not be the best to use in testing, these tests will determine the sensitivity of responses if different accelerogram inputs are used. Finally, one model structure will be tested to determine the response sensitivity of a building with a nonsynenetrical closed

). section, i Test data from the small scale structural models with interior l partitions and open end walls will be used to design four intermediate scale models that will be used for computer code benchmarking and floor

response spectra investigation. The models will be two story and will i possess the following first floor design modfiications:

! 1. one model with a complete partition

2. one model with a partial partition l

i I

j C-12 l

L

3. one model with an open end wall

4. one model with a partial open end wall.

As in previous years seismic tests will be conducted on the models with increasing earthquake amplitude. Floor response spectra will be determined and the ultimate capacity of the structures will be measured.

Pre- and post-test predictions of the seismic response of the structures will be carried out using the INRES-B computer code and material models (including the representation for damping) will be refined so as to give the best predictions of time histories and floor response spectra compared with the experimental observations in both the elastic and inelastic range of model behavior.

A substantial effort will be required in the areas of data reduction and analytical modeling with the INRES-B computer code. These efforts will be increased over those needed to support the experimental program.

In addition, the Los Alamos National Laboratory will begin investigations (using the Technical Review Group, NRC staff and other national laboratories) into how the existing program should be modified to establish a data base or analysis procedures useful for probabilistic risk assessment of nuclear power facilities.

One further effort will be investigated and possibly initiated. The program management has noted instances in the shock and vibration literature of measuring and reporting natural frequencies and mode shapes of very large reinforced concrete structures. This developing technology will be investigated and the possibility of performing such a test on a full scale Category I shear wall structure evaluated. If a suitable structure can be found, such a test combined with analytical modeling could be used as a final confirmation of the as-built stiffness of these structures.

D. Application of Results The types of information that will be available to potential users, such as the Equipment Qualification Branch, the Systematic Evaluation Program Branch, and the Structural and Geotechnical Engineering Branch, from this research includes:

1. Floor response spectra from seismic loads producing elastic and inelastic structural response, and floor response spectra changes as the response progresses from the elastic to the inelastic ranges.

2. Bounds on earthquake magnitudes at which the structural response l changes from elastic to inelastic.

j

3. The initial stiffness of structures (as related to the stiffness {

computed using the principles of mechanics). The initial stiffness will be investigated as a function of scale of the test structures.

(

C-13

\

4. The change in stiffness as the structural response becomes inelastic and the degradation of stiffness and load carrying capacity as the structure is subjected to load cycling.

5. The deformation that can be used as a measure of the yield point of the. structure and the ductility ratio of the structure.

6. Changes in equivalent viscous damping as the structural response proceeds from the elastic to the inelastic ranges.

The test results from the Seismic Category I Structures Program provide information on the overall or gross behavior of simple buildings and on variations in these buildings. These various structural behavior trends can be reflected in the licensing process by inclusion in Standard Review Plans and Regulatory Guides. By incorporating the behavior trends into design codes, the results can be applied during the design of new plants.

The data from this program can be used to evaluate computer models and programs that are applied to more complex structures. For example, initial tests show that the fundamental frequency decreases to about one-half of its original valve before significant cracking occurs. By using the ratio 1/2 and applying it to an existing analysis, the frequency of the actual building just prior to major cracking can be estimated. In addition, the analytical methodt used to obtain floor level responses and response spectra in the elastic range can be compared to experimentally obtained results. If necessary, new models can be formulated to more accurately predict structural responses. Once acceptable correlation between predicted and experimental elastic response is achieved, inelastic responses can be estimated by comparing inelastic experimental results to predicted inelastic results. By considering increasing earthquake magnitudes, comparing elastic predicted results to experiment trends, and adjusting parameter values, the behavior of a Category I building can be calculated in a rational manner.

E. References

1. E. Endebrock, R. Dove, and C. Anderson, " Margins to Failure--Category-I Structures Program: Background and Experi-mental Program Plan," Los Alamos National Laboratory report LA-9030-MS, NUREG 2347 (September 1981).

2. Sozen, Mete A., "A Note on Nonlinear Seismic Response of Reinforced Concrete Structures With Low Initial Periods,"

Section 4 of the informal proceedings of the EPRI/NRC workshop on Nuclear Power Plant Re-evaluation For Earthquakes Larger Than SSE, October 15-17, 1984, San Francisco, California.

3. Umemura, H. , Aoyama, H. , Ito, M. , and Josokawa, Yo, "Aseismic Characteristics on RC Box and Cylinder Walls," unknown publication source supplied by Cheng, F. Y., Program Consultant.

The data cited is taken from the figures.

C-14

I I l l A 3D-10-2,1/ 30 SCALE 3 9 3D-11-2,1/ 30 SCALE 7 -

B C E R L No.1,1/10 SCALE -

W

[6

~

g _

C E R L No.1 MODEL g E FAILED IN NEXT s g TEST FOR WHICH

" 5 -

Vpg /N9 > 3.5 g -

$ NOTES:

$4 y

__ 1. FOR 1/30 SCALE MODEL i

u- N, = 1/11.8 1

$ Ny = 1/4.6

~

k FOR 1/10 SCALE g MODEL E N, = 1/6.8 2

_ Ny = 1/4.6 _

FOR PROTOTYPE

• ^'

N, = Ny = 1 EXAMPLE: AT POINT 'A' _

CE R L No.1 TEST f, = 2.6 x 6.8 = 17.7 Hz

@YPK = 3.5 x 4.6 = 16 g o i I I I o 1 2 3 4 5 PEAK ACCELERATlON 9pg / Np ( g )

Fig. 1. Data taken from three FY 83 tests illustrating the first mode frequency. shift as the model structures were progressively damaged by increasing peak seismic accelerations.

C-15

l NORMRLIZED STIFFNESS VS CONCRETE MODULUS l o e

d-d-

s b ' ,

3 m d-m S. W. 1-30 Scale Isolated bd- Structures Shear Walls n l p

• 9 O

L m

i

=

U,

- g d- " 0 D o ,

o I h, a a, ,

l z d- _n

" a O

N Sozen (Ref. 2 )

!  ;- / y Umemura 5' 1 = 0.25 (Ref. 3 )

• g. K Theory 4 45.0 N.o S.o 10.0 15.0 S.o S.o N.o S.o 45.0 5 0.0 5'.o =10' CONCRCTC N00UWS (PSil h-Figure 2 Normalized Stiffnesses versus Concrete Modulus From This Prognn and Others.

NU x

l -

. l

' l

• j TWO STEEL PLATES l x APPROX.17,800 lb EACH

/

i 8'1 12 0 s N

4 N '

\l -

i 106  % -

p O

\ ,n ' '

90 ' "

,- ll N

N .- ,,, 4 l' '

', k .k I ALL FOUR INCH

\ s WALLS HAVE No. 3 f (Q i s A W l N 8 REBARS ON 4.9 INCH

\ s i

's

's  :',

i p

s'

/

90 ,-,' I20 DIMENSIONS IN INCHES Fig. 3. TRG structural test model.

l

l i

l l

Appendix D Piping Reliability Projects NRC Project Manager: Dan Guzy i

PIPING RELIABILITY PROJECTS I. INTRODUCTION This plan will detail the piping system research to be sponsored and monitored by the Mechanical Structural Engineering Branch (MSEB/DET/RES) in roughly the Spring of 1985 to the Fall of 1987 period. This research is primarily directed at those research needs identified by the NRC Piping Review Committee (PRC) in Volumes 2 and 4 of NUREG 1061. Other research included in this plan deals with needs related to PRC recommendations but not specifically stated (e.g., quantification of the combined effect of regulatory changes),

, and needs identified in the seismic margins and PRA areas.

Recently completed piping research will also be discussed briefly, if related to the efforts above.

I l

l D-1

II. BACKGROU_ND A. Research Tasks Recomended by_theh PR,C The USNRC Piping Review Comittee.(PRC) was formed in 1983 at the reouest of the USNRC Executive Director for Operations.

Its charter was to review NRC piping criteria, to recommend changes to this criteria (based oncurrentdata),andto identify areas that would benefit from future research. The PRC recomendations are published in the following volumes of NUREG 1061:

Volume 1 - Investigation and Evaluation of Stress Corrosion Cracking in Piping of Boiling Water Reactor Plants Volume 2 - Evaluation of Seismic Designs - A Review of Seismic Design Requirements for Nuclear Power Plant Piping Volume 3 - Evaluation of Potential for Pipe Breaks Volume 4 - Evaluation of Other Dynamic Loads and Lead Combinations Volume 5 - Sumary - Piping Review Comittee Conclusions and Recomendations Volumes 1 through 4 are the reports of the four PRC task groups. Recomended research tasks are identified in these along with detailed discussions of the needs and objectives for these tasks. Volume 5 is the PRC Sumary report which sets priorities for all research deemed necessary to resolve outstanding piping issues. The research recommended in the five volumes of NUREG 1061 will be addressed by both ongoing and new programs conducted by the NRC/RES or industry. The division of projects within RES is such that the Materials Engineering Branch (MEBR) will primarily address the tasks dealing with pipe cracking and pipe break that are outlined in Vols. I and 3. (MEBR has its own Piping Plan, a major component of which is the Degraded Piping Program.) The scope of the piping research to be conducted by MSEB and discussed in the. plan that follows is directed primarily towards those research needs identified in Vols. 2 and 4. These deal with criteria for seismic and other dynamic loadings, and for load combinations.

NRC projects directly recomended by NUREG IU61, Vols 2 and 4, are sumarized below along with their PRC priority ranking:

PRC Priority Task A-3 Support EPRI's piping structural capacity tests A-4 Complete LLNL studies of GE and B&W RCL pipe break

• 1 B-2 Complete ORNL development of nozzle design guidance B-3 Complete INEL damping studies

• B-3 Additional studies for nonseismic damping C-1 Assess uncertainties in seismic spectra C-1 Develop basis for flattened spectra D-2 L

C-2 Additional studies on phase correlation and modal combinations C-7 Develop simple estimation methods for inelastic response C-9 Complete PNL study on snubber performance (sponsored by EEICB/RES)*

Those tasks followed by an asterisk (*) will essentially be completed under ongoing programs and will not be discussed here in detail. All the other tasks form an integral part of this plan.

Additionally, there will be one research task recommended in NUREG 1061, Vol. 1, that will be sponsored by MSEB. This is categorized as C-6 in Vol. 5 and deals with the future modification and use of the PRAISE Computer Code to investigate intergranular stress corro-sion cracking (IGSCC). (MSED asored the development and use of the PRAISE Code in LLNL's Load Combination Program and the Stiff vs Flexible Piping Study.)

)

8. Other Needs Related to the PRC Recommendations The previous section discussed research tasks now explicitly recom-mended by the PRC. There will be future needs that will evolve from the implementation of the PRC recommendations for regulatory changes (See Chapter 10 of NUREG 1061, Vol. 5) and from the results of the presently recommended research tasks.

The PRC made its evaluations and recommendations largely on an item-by-item basis. With the exception of LLNL's study of the im-pact of changes in both damping and spectra broadening, most of the supporting research data did not consider the combined effects of criteria changes. Justifications were largely based on seismic margins eva,1uations and reliability studies of currently designed piping systems. As the various proposed criteria changes become implemented, the nature of piping designs will. change (in the direction of lower natural frequencies and perhaps lower seismic structural capacity). It is important to investigate combined effects and reassess previous conclusions as regulatory changes are made.

A specific item to be investigated is the use of the new damping criteria in conj action with the currently recommended independent support motior method.

Some possible areas that could warrant future study are:

1. The need to include seismic anchor motions in SSE evaluations, as OBE design loadings decrease (through use of the new damping values, decoupling of OBE and SSE, etc.)

2. The impact of a possible future reclassification of seismic inertial loads on 11 other piping criteria

3. The possible decrease in damping as supports are removed (thus invalidating damping values derived from stiff systems)

D-3 J

4. The increase in nozzle loadings as systems become more flexible

5. The effect of support strength and reliability on pipe system reliability, when systems become more flexible

6. The impact of potential design errors on piping systems designed to recommended new criterie

7. The effect of inservice degradation (corrosion, embrittlement, etc.) on piping system reliability.

C. Seismic Margins and pRA Validation Needs The current NRC efforts in the quantification of seismic margins and the validation of seismic PRA methods present needs to better quan-tify piping system seismic fragilities. Although earthquake experi-ence data at non-nuclear heavy industrial facilities gives us confi-dence that the lower parts of piping fragility curves are reasonable (i.e., failure is not likely at earthquake levels up to roughly .3g ground motion) there is little or no physical data in the median fragility range. Piping fragility models are based chiefly on enaly-sis and some of the assumptions used in their development have been criticized (see below), eroding confidence in seismic PRA con-closions about the importance (or non-importance) of piping failure.

The extensive use of piping in nuclear plant safety systems gives definition of the median fragility range a greater importance than for lesser used equipment. If a systematic bias has created fragil-ity estimates higher than actual, then current risk studies may have masked out the effect of piping failure, now generally considered a minor risk contributor. On the other hand, if it can be shown that piping fragility curves are biased significantly low, then it may be possible to take piping system review and analysis out of the PRA process. This could apply to defining initiating events. (e.g.,

small LOCAs) as well as to the failure probability of mitigating systems.

Current pipe failure models used in seismic PRAs are necessarily simple and need to be evaluated to determine if there are any sig-nificant errors. Some areas where these piping fragility models (and the associated response analyses) are criticized include:

1. Collapse is assumed to be the failure mode (many feel that fa-tigue ratcheting is the more realistic mode).

2. Linear methods are used to predict non-linear behavior and may not capture important phenomena such as the "detuning" of systems.

3. Assumptions regarding support response / failure and the effect of these on piping response / failure are very crude.

4. Whole piping systems are represented by single-point moment values without attempting to integrate total system capacity.

5. Assumed flaw sizes may be conservative. l

6. The approaches used are very generalized and may miss impor-tant, line-specific weaknesses (e.g. lugs, bolted flanges, etc.)

7. Scaling from ASME Code allowables may be inappropriate since these are based on static tests.

0-4 L ..

Such items need to be addressed to either add confidence to current estimations, or to improve them. This can best be done through the acquisition of physical failure data for piping systems. Earth-quakes which cause physical failure are rare, so testing is the only real option.

D. Ongoing a_nd Recently Completed MSEB Piping Research While this plan is to be mainly forward looking (starting roughly in the Spring of 1985), recognition should be made of piping research projects that have been completed recently, or will be in the near future. MSEB-sponsored studies served as part of the information base from which the PRC made its recent reconnendations. The MSEB projects most influential to the PRC include:

o Load Combination Program (FIN A0133)

This LLNL program has demonstrated the low probability of oc-currence of double-ended guillotine breaks in Westinghouse and CE designed reactor coolant loops. The program will end in FY 85 with the completion of studies of GE and B&W reactor coolant 1 cop break probabilities. This program is providing the justi-fication for proposed modifications to General Design Criteria 4 to allow the use of " leak before break" as a justification for eliminating pipe whip restraints and jet impingement shields, o Damping Studies (FIN A6316)

Tests at INEL will be completed that will better define the parameters that affect seismic damping. INEL has also devel-oped the world data bank on pipe damping that was used by the PVRC in developing the new damping curves.

o Piping Benchmark Program (FIN A3225)

The BNL Piping Benchmark Program has evaluated the adequacy of varic1s dynamic analysis methods using both physical and ana-lytic benchmarks. Recently, this program has made recommenda-tions on the use of multiple-response spectra methods that was approved (with slight modifications) by the PRC.

o Stiff Vs Flexible Piping (FIN A0383)

This LLNL program has assessed the relative reliability of stiff and flexible piping systens considering failures due to both seismic and thermal loads. It has also quantified the impact of using the new damping and spectra shifting criteria on the reduction of seismic supports.

D-5

III. PLAN ELEMENTS The following MSEB-sponsored piping research projects will be conducted in the FY 1985 to FY 1988 timeframe.

A. EPRI/NRC DYNAMIC LOAD PIPING TESTS The NRC and EPRI will cooperate jointly in the " Piping and Fitting Reliability Research Program." The cbjectives of this program are to clearly determine the likely dynamic failure modes of uncracked piping systems, to identify procedures to predict failure (and mar-gins), and to provide a basis for changing ASME Code rules, if ap-propriate. This program consists of dynamic capacity testing of piping at the system, component, and specimen levels, plus analyses needed in support of test planning and the development of recommen-dations for criteria changes. The NRC Piping Review Committee has ranked this research effort in its highest priority category (identified as item A-3).

Currently, ASME Code requirements for the seismic design of piping assume that plastic collapse is the dominant failure mode. If it can be demonstrated by this program that fatigue /ratcheting is the principal failure mode (as many now believe), then significant changes can be made with regards to how the ASME Code sets limits on seismic inertial stresses. The resulting criteria changes could significantly reduce the number of seismic supports and snubbers used in piping design.

This program, and related follow-on efforts, will also serve to meet needs in the Seismic Margins and PRA validation areas. Current pipe failure models used in seismic PRAs are necessarily simple and need to be evaluated to determine if there are any significant errors.

The EPRI/NRC " Piping and Fitting Reliability Research Program" will begin in mid-FY 85 and will last 2 1/2 to 3 years. EPRI developed the initial planning for this program and will fund the larger portion of it. In exchange for its contribution, the NRC will receive full and immediate access to all data generated and will participate in project reviews. EPRI will retain ownership of the data, but will provide for the NRC and its contractors' use of the data as needed.

1. ETEC Demonstration Test As a part of the NRC's contribution in FY 85, a pipe system capacity test will be conducted at ETEC. The objectives of this test are to demonstrate the . feasibility of failing a representative piping system under a high earthquake-like load, and to provide information and insights needed in the test planning of the main EPRI/NRC program.

The ETEC test will use a modification of the 6" pipe system previously tested at ANC0 in an earlier EPRI/NRC effort. The earlier tests were unable to produce pipe failures, (except due D-6

to over-fatigue). ETEC will use much higher load capacity seis-mic tables that are intended to produce pipe rupture in a limited number of cycles.

2. Piping and Fitting Reliability Research _ Prog, ram The tasks of the EPRI/NRC " Piping and Fitting Reliability Research Program" are outlined below. The NRC's contribution to this program will be through sponsorship of parts of Tasks 2, 3, and S.

Task 1: Program Plan Development General Electric will perform the overall management, coordination and integration of the program. G.E. will work with consultants and the testing sub-contractors to develop test matrices, configurations, and data acquisition procedures.

EPRI and the NRC will review and approve the plans as they develop.

Task 2: Pipe Comp,onent Testing Dynamic capacity testing will be conducted at the component level for loadings simulating severe seismic, hydrodynamic loads, and water hammer events. The objectives will be to characterize the mode of component failure and to measure the response level and number of cycles to failure.

Elbows, tees, reducers, support connections, nozzles and lugs will be tested. The test specimens will consist of both carbon and stainless steel 6" NPS (with various schedules) piping components under different internal pressures. The seismic and hydrodynamic tests will consist of the components attached to shakers at one end. The other ends will have weights attached to induce inertial loadings in the components. The water hamer tests will have a different loading technique (to be specified at a later time).

Task 3: P_ipc,,Sy,s, tem,,Te,s,t_i n2 Three different 6" piping systems will be tested to failure under simulated earthquake, hydrodynamic, and water hammer loadings. These pressurized systems will use components compatible with those in Task 2 and will also have piping supports at the 4 or 5 load input points. The tests will be repeated at least once to give two (or more) tests per load-type.

The objective of these tests is to show realistic piping system failure under the combined effects of pressure and gravity loading plus large dynamic loadings. Response levels and cycles to failure will be recorded, as will the nonlinear behavior of the systems.

D-7

Task 4: Specimen Fa_tigue_ Ratcheting_T,e,s,ts.

The failure mechanism for most of the tests in Tasks 2 & 3 is expected to be low cycle fatigue ratchet. The basic phenomena of fatigue ratcheting will be studied in Task 4 using laboratory speci-mens. Since at the present time there is no standard laboratory test specimen which addresses this failure behavicr, such specimens need to be developed. Both uniaxial and bending loads will be ap-plied. The influence of different material and temperatures will be studied. The outcome of this task will be the basis for evaluating ratchet effects under dynamic fatigue for ASME Code pipe materials.

Task 5: Analysis of Tests and Design Rules The three basic tests described in tasks 2, 3, and 4 will require analysis support so that (1) the tests can be properly designed, instrumented, and interpreted, (2) the underlying failure behavior can be generically characterized for all piping design situations and (3) so that a significant data base can be generated to support changes to the ASME Code pipe design rules.

All post-test failure interpretation of tasks 2, 3, and 4 will be included in this task. A technical area of interest which will be explored is the equivalent linear damping which results when the task 3 tests are driven into the plastic regime. Other methods for predicting nonlinear response will also be evaluated.

Another goal from this task is the development of specific failure criteria which generically apply to combined static and dynamic loading of nuclear piping components. These criteria will be de-sign-oriented.

Task 6: Identification and Development of Alternative _ Design Rules-and Regulations This task involves the identification of alternative design rules which are practical to implement into the ASME Code. It will also identify any NRC regulations which supplement ASME piping code rules and which need to be revised if the benefits of the project are to be realized. This task involves liaison with the PVRC, NRC and the ASME Section III Code body which ultimately makes the rule changes.

If new design procedures are developed, the evaluation under task 7 will be implemented after feedback is given from the ASME Section III Working Group on Piping and other appropriate standards groups.

Task 7: Evaluation of Alternative Design Rules 4 This task has the objective of evaluating the alternative design l rules identified in task 6. The evaluation criteria to be used will

! be developed by General Electric and then reviewed and approved by EPRI and the NRC as well as by the ASME Section III Working Group on Piping. The objective of this task is to determine the i D-8

acceptability, ranking, and if necessary, the required fine tuning of the alternative design rules developed in task 6.

The following criteria will be used in evaluating alternative design rules:

Analytical justification and consistency with test data Economic impact Ease of application and conformity with present code rules Effect of new design rules on safety margins for cracked piping Task 8: Project Final Report _s EPRI will publish the following reports:

Vol. 1 Pipe Component Test Results Vol. 2 Specimen Fatigue-Ratcheting Test Results Vol. 3 Pipe System Test Results Vol. 4 Piping Design Rule Studies Vol. 5 Project Summary Report

3. Separate and, F,ollow,,on,, Tasks As part of the NRC's review of Tasks 6 and 7 above, separate NRC-sponsored evaluations may be performed. The NRC would coordinate these efforts with EPRI and GE and attempt to resolve any disagreements if they occur.

The EPRI/NRC program outlined above is directed primarily at evaluating and improving piping system design rules of Section III of the ASME Code. Additional work may be needed by the NRC to use the failure data in validating, or improving, the piping fragility models used in seismic risk and margins studies.

The piping design design rules of Section III of the ASME Code do not explicitly address pipe cracking. The EPRI/NRC capacity tests will not include flawed piping; however, the NRC Degraded Piping Program will perform seismic tests on relatively short pipe specimens with known crack sizes. The results of these two programs need to be evaluated in parallel. Data about the capacity of flawed piping needs to be evaluated along with data concerning the nonlinear response behavior of piping systems at high input levels (Task 3 above) to give an integrated assessment of failure.

The objectives of the Degraded Piping Program are deterministic and '

may lead to the introduction of conservations. Additional work may be needed to provide a probabilistic best-estimate model of the capacity of cracked piping for use in seismic risk and margins studies.

D-9

B. Nozzle Design Guidance The need for improving design guidance on nozzle flexibility and allowable loads was recognized by the NRC Piping Review Committee (identified as item B-2). Because.of this, the following task was initiated as part of ORNL's ongoing technical -support program (FIN B0474). The guidance to be. developed is intended to reduce the need for seismic supports under current design criteria, and to provide a more realistic. basis for evaluating the higher nozzle loads that will result from the more flexible piping systems design that are being considered.

ORNL will prepare improved design guidance for pressure vessel and equipment nozzles and piping system connections in typical Class 1, 2, and 3 nuclear power plant system. Particular attention will be given to the interaction effects at the. vessel-piping (and piping-piping) juncture ' including the effects of nozzle flexibility on calculated bending and thrust loads under both static and dynamic conditions with a view toward reducing the required number dnd rigidity of dynamic. supports (including snubbers) and anchors in the piping systems. Design methods used to prevent plastic collapse and fatigue failure will be addressed through an assessment of current primary and secondary load stress indices and stress limits based on recently developed theoretical data (shell theory and finite element solutions) and existing experimental results. Of particular importance are the recent reconsendations from NRC Bulletin 297 " Local Stresses in Cylindrical Shells Due to External Loadings on Nozzles - Supplement to WRC Bulletin No. 107."

ORNL will develop recommendations for improvements in ASME Code design rules for vessel _riozzles and piping. These will be' reviewed by the ASME and NRC.This project is. currently underway and should be ,

completed in 1985.

C. InvestigationofHigh,Fre,quencylamp,ing The NRC Piping Review Coteittee has recommended (item B-3) that ad-ditional studies be made to determine whether the recently approved pipe damping values for seismic design can be extended to higher frequency vibratory loadings. The PRC has recommended that the new damping curves be applied only for seismic analysis of piping.

Neither RG 1.61 nor the ASME Boiler and Pressure Vessel Code-explicitly addresses damping due to. hydrodynamic loads. To.date, only a small amount of data has been generated in the frequency range of 33 to 100 Hz. Guidelines for damping in this range would eliminate an important gap in piping system design guidance.

As part of the overall NRC/INEL pipe damping study program (FIN A6316), INEL will add a new task of evaluating high frequency damping. The task will be accomplished by reviewing the high frequency data available in the open literature, reviewing the additional data from Nutech Engineers which was contributed to the damping data bank, conducting high frequency /high stress tests on D-10

two piping systems previously used for seismic damping studies, reducing and evaluating the data, and preparing and publishing a NUREG report that includes the data and recomendations for allowable damping at higher frequencies. The testing will be done on already fabricated piping systems using available supports and existing data acquisition and reduction equipment.

The proposed test matrix for the high frequency pipe damping testing is outlined below. Each test series would begin with modal characterizations (using instrumented hamers) to identify the high frequency mode shapes. Eachtest(exceptmodalcharacterizations) would include at least one high stress level and one low stress level test for each configuration. Random input motion using one or more hydraulic shakers will be used. The number of frequency ranges required to excite the pipes adequately would be determined during initial shaker characterization runs. It is anticipated that several frequency ranges will be required to ensure sufficient input energy is available in each range.

5-in. 3-in.

3-D System 1-0 System Modal characterizations X ~ ~- X End supports X X Hydraulic snubber X Mechanical snubber X X Rigid strut (s) X X Combination strut and snubber X X These studies will begin in mid F85 and end in mid F86.

D. Piping Spectra Project The NRC Piping Review Comittee has recomended (item C-1) that fur-ther studies be conducted to more completely assess the uncertainties in the seismic response of building structures and piping systems. The results of these studies would then be used along with sensitivity studies to evaluate the potential benefits of using design spectra that are lower and broader than those now developed by Regulatory Guide 1.122. The feasibility of using equivalent static analyses for evaluating inertial loading will also be assessed.

Regulatory Guide 1.122 now accounts for uncertainties in structural response frequencies by requiring that peaks in raw spectra be broadened by 115%. The effect of uncertainties in piping system response frequencies are not considered at all. Recent studies (e.g., the SSMRP) indicate that structural response uncertainties are underestimated by the regulatory guide's specification. Further broadening, however, may unrealistically increase load levels in 0-11

the analyses. A probabilistic approach.that considers the uncertainties in high level building and piping response will lead

.to a more rational specification of in-plant spectra. Although the following-project is directed chiefly towards piping design and analysis, those who are involved with equipment qualification and structures will also be interested in its results.

The recomenced spectra studies will be performed in a two-phased

-project at LLNL, " Assessment and Im

~ Procedures.Used in Piping Design" FINA0453). (provement of Spectrum-Broad Phase I will consist 4 -of the gathering and synthesis of information on response frequency uncertainties (both for nuclear structures and piping). A 4 _" state-of-the-art" sumary will be made concerning what has been

.done to date in this area. All related NRC-sponsored work be considered along with cutside research. -(NRC programs to be inves-tigated include the SSMRP, BNL's Structural Load Combination and i Piping Benchmark Prcgrams, LANL's Category I Structures Program, USI A-40 and the LL/00R Seismic Conservatism Program.) Little, or no, new analysis will be performed in this phase.

Phase 1 will be completely funded in FY85. An interim report will be issued at the end of Phase 1 that will include (1) the

" state-of-the-art" sumary discussed above, (2) an assessment of the ,

adequacy of the 15%Ipeak broadening range specified by Regulatory Guide-1.122, and (3) an outline of the effort proposed for Phase 2.

The approval to continue with Phase 2 will be based to some extent on the success of Phase 1. The objective of Phase 2 is to develop and justify a probabilistically-based procedure for " simple" in-structure spectra that realistically account for the maximum energy a design earthquake can transmit. A goal would be to establish the basis for " flattened and broadened" spectra that could replace those now specified by Regulatory Guide 1.122. The extreme of this would be a completely level spectram that would facilitate the use of static analyses for piping inertial loads,'and thus greatly reduce the complexity of current design practice.

! Upon eventual completion of Phase 2, a final report should be issued that will sumarize the work in both phases of_ the project. Phase 2 is to be completed in FY 86.

E. Combinational Procedures for Piping _R,espon_s,e_

o Analyses In Vol. 4 of NUREG 1061, the NRC Piping Review Comittee has made recomendations to modify present piping design criteria regarding response combination methods and the use of multiple response

, spectra methods. In developing these recomendations, the PRC identified (item C-2) the following needs for further analytical study:

1. Clarify the impact of phase correlations between support groups on the recomendations for the independent support motion i method, 0-12

2. Evaluate methods for calculating the effect of closely spaced modes, and

3. Establish the transition frequency between high and low fre-quency when implen.enting the algebraic sumation rule for high-frequency modal combinations.

The investigations above could provide the justifications for removing some conservatism from the current PRC recommended criteria for piping response combinations.

Through the Mechanical Piping Benchmark Project (FIN A3225), BNL conducted a study of the Independent Support Motion (ISM) method.

This study used an extensive set of time history response data (much of this from the SSMRP) to serve as a statistical, best-estimate, benchmark for comparisons with response spectrum analyses employing various combinational methods. These comparisons were instrumental in justifying response spectrum procedures that were recommended by BNL in NUREG/CR-3811, and then adopted (with modification) by the PRC. This study did not address the impact of using the PVRC (Code Case N-411) damping values in conjunction with the ISM method. The combined effect of using both these new criteria needs to be evaluated.

BNL will extend their previous work in new studies to answer the needs identified above. The tasks to be performed as part of the "Combinational Procedures for Piping Response Spectra Analyses Project" (BNL FIN A3287) are:

Task 1: Investigation of New Damp,ing Values with ISM Method The approach to this task will be similar to that reported in NUREG/CR-3811. The same piping models will be used as before, but Code Case N-411 damping will be bsed for both the time history and spectra analyses that will be compared. (SSMRP-SMACS Code output will be provided under BNL FIN A3267; this project will also assess the effect of Code Case N-422 damping on time history analysis results.)

Task 2: Modal Combination Methods The response spectrum method is most commonly used to estimate the dynamic response of systems. In this method a response quantity is computed for each mode and each direction of excitation. The total response is then formed by summing over all modes and over all directions of excitation. Current criteria require that the square root of the sum of the squares (SRSS) rule, modified to account for closely spaced modes, be used to combine the modal responses. This rule is based on the assumption that well spaced modes are randomly phased (uncorrelated) while closely spaced modes are essentially in-phase. There are two issues associated with this criteria, applying to both single (" uniform") and multiple response methods:

0-13

Closely Spaced Modes: For closely spaced modes, Reg. Guide 1.92

. requires-the moTa Tcontributions to be combined using the absolute sum (ABS) rule, using a worst. case phasing assumption for these modes. Algebraic sumation between these modal contributions may be nere appropriate for certain specific conditions. Methods theoret-

-ically based in random vibration theory (e.g., the DSC and CQC methods) will be investigated.

High Fregue_ncL M odes: ' Current research indicates that some degree of correlation may exist between modes above 2 Hz and the high frequency modes (above 33 Hz). .To account for this correlation modified model combination rules have been proposed by Gupta, Hadjian and by Lindley and Yow.

BNL will use the time history data base previously developed to evaluate the appropriateness of new response combination methods that should better address closely-spaced modes and high frequency modes. Uniform response spectrum analyses using the new combination methods will be compared against the tine history results.

It is intended that this task be initiated in FY85.

Task 3: Structural Phase Correlation in ISM Method The recomendations on the Independent Support Motion Method con-tained in NVREG/CR-3811 were all adopted by the PRC, with one exception. This dealt with the combination between group contributions where, in the absence of. phasing information, the BNL recomendation of SRSS combination was not accepted. The NRC chose

- to recomend the more conservative absolute suming of group responses.

In FY85, as'part of the ongoing effort, BNL will determine the de-gree of phase correlation between the structural inputs used in the original study and assess the impact of this information on the BNL recomendations. The results of this will impact future planning for Task 2, which can thus not begin until FY86.

A proposed group combination scheme to be evaluated would. continue to use absolute suming if group phase correlation is undetermined, but use other suming methods (such as algebraic or SRSS) depending on what is known about group phase correlation. ISM results using these group combination assumptions will be compared against time history results where the input phase correlation has been determined.. If new sets of time history response data are needed to meet specific correlation requirements, then the SMACS code may be used in new time history analyses. The results of these comparisons

.will be used to make recommendations to modify the current PRC position on the ISM group combination method.

D-14

i F. IGSCC Study using Praise Code The NRC Piping Review Committee has recommended in NUREG 1061 Vols.

I and 5 (item C-6) that the PRAISE computer code be used to (1) investigate the impact of intergranular stress corrosion cracking IGSCC in primary piping according to pipe size, (2) study the failure frequencies of components due to IGSCC, fatigue, etc.,

and (3) calculate the conditional probability of multiple failures associated with IGSCC to check the point value and to test the assumption of a log normal distribution.

The PRAISE computer code was developed by LLNL as part of the MSEB-sponsored Loads Combination Program. This code estimates the probabilities of pipe leak and rupture due to fatigue and stress corrosion cracking using a Monte Carlo simulation technique. In the Loads Combination Program, PRAISE was used to analyze leak and rupture probabilities for the main coolant piping of all four NSSS vendors. PRAISE was also used to analyze other piping systems in the Stiff vs. Flexible Piping Project.

It is proposed that the study recommended above by the PRC be con-ducted as an extension of the BWR Ripe Rupture Investigations that were a part of the LLNL Loads Combination Program. The BWR work completed in FY85 dealt only with calculating leak and rupture probabilities of Mark I recirculation, main steam and main feed water piping. Not covered were the LPCI, LPCS, HPCS, RCIC and RHR piping. This piping, which is also part of the reactor coolant boundary, is of generally smaller diameter than the piping previously analyzed.

It is proposed in this extension of prior efforts to evaluate the relative importance of fatigue and IGSSC in the smaller diameter piping of GE Mark I designs, and to also include piping from Mark II and III configurations. The work would be performed by Lawrence Livermore National Laboratory using the already validated IGSSC modeling algorithms developed and applied to the large diameter piping. A key aspect of the new work would be an evaluation of the assumption that multiple cracks may initiate and " link-up" during the operation of the plant and an assessment of how this phenomenon affects assumptions of probability distributions.

It is proposed that this task be conducted in FY87-88.

G. Inelastic Resp _onse of Pip,ing , Systems The NRC Piping Review Committee has recommended (item C-7) that studies be made to develop pseudo-linear-elastic estimation methods, and design procedures to account for inelastic piping response. It is hoped that the end product of this would a methodology simple l enough to be used in the routine design evaluation process.  !

Existing techniques that are based on modifying standard response i spectra analyses will be evaluated and adopted if justified. These include the dynamic-to-static margin ratio technique (by Campbell) and nonlinear response spectra methods. Otherwise, new methods will i be developed.

D-15

. .- - -= --. . .- -__. .

The PRC's recomendation is addressed mainly towards improving the design of piping for loads within the design envelope. In doing

'this, the margins-tc-failure that exist in both the current criteria and in future criteria shculd be assessed and compared. Thus, linear / nonlinear response factors at failure levels need be considered as well as the response factors at the design level (i.e., Service Level D).

For design purposes, these assessments can introduce some conservatisms to account for uncertainties. However, for seismic PRA validation and seismic margins evaluations the goal would be to establish best-estimate factors and also try to realistically quantify the associated uncertainties in these factors.

The " Nonlinear Piping Response Prediction" project at HEDL (FIN D1611) will consider nonlinear response prediction techniques with differing degrees of accuracy (and complexity) and will compare the various analytical results both against each other and against physical benchmarks. Because of their experience in nonlinear piping analysis, HEDL was selected to conduct this project. HEDL is currently involved with a related DOE task to predict nonlinear piping response.

It is currently planned that pretest response and failure predictions of the ETEC Demonstration Tests will be made by HEDL using ABACUS and NONPIPE (a less sophisticated nonlinear piping

code), along with candidate simpler techniques such as inelastic spectra, a " dynamic margins" approach, and current seismic PRA fragility estimation methods. Post test assessments will be made of the various predictions with the hope that a simple design-oriented method could be endorsed. This project will be conducted mainly in FY 85.

H. Other Projects As discussed in the " Background" section of this report, there will be needs to assess the combined effects of the criteria changes currently recomended by the PPC, and future changes that evolve from the research discussed previously in this plan. The most significant of the criteria changes now endorsed by the PRC (in terms of removing seismic margins) is the new damping values. The implementation of this would change piping design and possibly

invalidate assessment of other criteria changes that were made considering only current design criteria and practices. The margins t impact of other criteria changes, such as adoption of the In-dependent Support Motion method, needs to be re wsntified in light of criteria changes that precede them. It is e M sioned that such studies would be done on an as-needed basis as t? standards are changed item-by-item. ,

e D-16

1 Appendix E Component Fragility Project l

NRC Project Manager: John O'Brien l

E

COMPONENT FRAGILITY PROJECT A. Introduction The " fragility" of a component is generally defined as the probability of its failure as a function of some forcing or response function. Seismic fragility is usually expressed as a function of local floor response or is tied to free field peak ground acceleration. The testing and data collection activities outlined in this program concentrates on seismic fragility; however, it is important to keep in mind that, in principle, fragility can be defined for any type of forcing function, as for example, hydrodynamic loads.

Current seismic probabilistic risk assessment (PRA) methods for nuclear power plants utilize component fragilities which are for the most part based on a limited data base and engineering judgement. The seismic design of components is based on ASME Code and NRC requirements that do not reflect the actual capacity of a component to resist failure. In order to improve the present component fragility data base and establish realistic component seismic design margins, a projected three-year program was initiated to compile the existing fragilities data base, and at the same time independently perform fragilities tests on components determined to be important to safety. Besides determining actual seismic design margins and fragilities for selected components, this program will generally enhance understanding of component failure modes and of the various issues affecting fragility.

The specific objectives of the Component Fragility Project are to:

o systematically identify and categorize electrical and mechanical components important to safety, taking into account system and subsystem functional descriptions, operating and maintenance experience, expert opinion, past PRA results, regulatory concerns, and existing test data.

o develop procedures for component fragilities and seismic margins testing, and demonstrate the effectiveness of these procedures through actual component tests. Obtain useful fragilities and seismic margins data for the components tested.

o based on the test procedures and component grouping, develop a plan for comprehensive fragilities and seismic margins testing to be performed in FY86 and FY87.

o initiate cooperation with domestic and foreign institutions to obtain already existing component fragility data.

o assemble, analyze and interpret existing component fragility data and compare with information currently being used.

o improve seismic PRA's and obtain better estimates of seismic margins by using more realistic test based component fragilities.

E-1

o learn which components are most likely to fail through a comparison of their fragilities with the expected environments to which they will be subjected. This could potentially lead to simplifiod PRAs by the elimination of certain branches on event trees and fault trees.

o learn whether the integrity of a component is more challenged by seismic loads or by other events, such as water hammer.

This program supports the need for reliable inputs for probabilistic risk assessments and margins studies. It is now believed that pessimistic estimates of the seismic threat to nuclear power plants have developed because of undue conservatism in the input component fragility data. This has forced the NRC to react in an overly cautious manner to difficulties in seismic design such as at Diablo Canyon. Moreover, for older operating reactors without any equipment qualification, the staff is now beginning to depend on experienced data to allow continued operation. This is a more or less qualitative approach requiring a better foundation which can only come from controlled laboratory testing such as proposed in this effort.

B. Background There have been no prior efforts directly involving this project.

Some fragility data were developed as part of the Seismic Safety Margins Research Program (SSMRP), but this information is based primarily on expert opinion and test data from non-seismic situations. In addition, the nuclear industry, as a result of work sponsored by the Seismic Quali-fication Utility Group (SQUG), has developed a data base which used past earthquake experience data of non-nuclear facilities to demonstrate the seismic adequacy of equipment in operating facilities.

The NRC has begun a cooperative agreement with EPRI to collect component fragility data. Terms of this agreement are as follows:

1. NRC and EPRI will use a common data format for collection of information from various sources. It is understood that owing to the specific nature of the individual program, NRC and EPRI may gather additional information about a test to meet their needs for compilation of the data.

2. NRC and EPRI will coordinate their data acquisition effort, e.g.,

visiting a source organization, collecting a test report, etc., in order to avoid duplication as far as possible. It is agreed that EPRI will primarily collect data from the utilities and west coast testing laboratories, and NRC will primarily approach vendors and east coast testing laboratories.

3. NRC and EPRI will exchange data sheets and pertinent information. It is recognized that in many instances, information in the form of a data sheet may not be adequate for either program; a review of the test report may be required.

E-2 1

- - - - - _ - _ . = - . _ _ _ _ - _ - . _

! i i

l The Component Fragility Project will be conducted in twa Phase I (FY85) activities will consist of parallel efforts to (1) phases. develop and demonstrate procedures for performing component tests to obtain new ,

fragilities data, (2) identify, through systematic grouping, components important to safety and therefore candidate for extensive testing, and (3) acquire and evaluate existing component fragilities data from sources

}

world-wide. Near the end of Phase I, the results of these activities will be convolved into a detailed plan for comprehen:ive component testing to j be performed in Phase II of the program (FY86 and FY87).

a Phase II (FY 86 and FY87) of the Component Fragility Project will be devoted to ~ the planning, performance and evaluation of comprehensive

! component fragility tests and supporting analyses, i

The completion of Phase I will provida assurance that any testing j performed during Phase II will not duplicate any already existing and usable data. Demonstration tests also will have been conducted that will define the procedures and clarify the approaches for arriving at usable fragility data. In addition, the prioritization scheme will permit the appropriate selection of components for fragility testing during Phase II.

Those components essential to the safe shutdown of. the reactor in the event of an earthquake and those components essential to the safety of the

' public in the event of a severe accident will be given the highest priority for further testing.

The plan of approach as well as the objectives and scope have been approved by NRR (Equipment Qualification Branch) and by the ACRS subcommittee on Extreme External Phenomea. We are coordinating our work with major efforts ongoing in RES's Electrical Engineering Branch and the Mechanical / Structural Engineering Branch.

r C. Program Plan In the purest statistical sense, empirically developing a meaningful seismic fragility for a given component would require that a large population ' of identical components (e.g., several hundred or several thousand) be subjected to successively higher levels of acceleration and the distribution of failures as a function of acceleration level recorded.

Within practical constraints on time and resources, this is hardly feasible for a single component under well-defined load conditions, let alone for the effectively infinite combinations and permutations of 4

component type, mounting, loading conditions, etc., that could be identified for actual nuclear power plants. Therefore, an alternate

approach is necessary to experimentally gain insight into fragility.

. The approach adopted in this program takes advantage of the fact that, for PRA application, a limited fragility description may be ade-quate. This is because in a probabilistic analysis, failure occurs only

, when the probability distributions of response and fragility overlap. In those cases where only the tails of these distributions overlap (as would j be the case, for example, for a component with a high safety margin) only

the lower portion of the fragility curve would be of interest from a PRA

standpoint. Therefore, the number of tests on such a component could be E-3 1

limited substantially, although a relatively large number of identical components might still have to be tested to assure statistically meaningful results.

For a component having a high margin against seismic failure, the degree of overlap of the response and fragility distributions could conceivably be so small as to imply that the probability of a seismic failure is effectively zero. In such a case, it would have little meaning to empirically describe fragility for PRA purposes; however, testing would be useful for quantifying experimentally the level of seismic margin.

This case has the further attraction that if the level of seismic margin is very large (i.e., failure does not occur until acceleration reaches levels many times that associated with a safe shutdown earthquake),

testing of a small number of components could deliver adequate results.

Phase I of our work is already underway at Brookhaven National Laboratory (BNL) and Lawrence Livermore National Laboratory (LLNL).

For Phase I of the Program the Scope of Work is as follows:

The BNL portion of the Phase I effort involves five tasks:

a. 1. Establish lines of communication with vendors, owners, and testing laboratories to determine availability of already existing component fragility data;

2. Negotiate, to the extent possible, the transfer of existing component fragility data to BNL;

3. Host an international workshop on component fragility to identify sources of information and issues and concerns.

4. Utilize all sources of information to assemble, analyze and interpret available fragility data for selected components. The selected components will take into account the component prioritization being undertaken by LLNL and the recommendations of the consultants which will be retained by LLNL. In turn, the LLNL component prioritization will be based in part on the information obtained by BNL.

5. Compare results with component fragilities used in current PRA and seismic margin studies and recommend improvements where possible.

The LLNL portion of the Phase I effort invovles five tasks:

b. 1. Develop Component Group and Prioritization

2. Develop Phase I Test Procedures

3. Perform Phase I Tests

4. Review and Document Test Results

5. Develop Phase II Program Plan E-4

In order to reduce to a managable level the number of components included in a comprehensive test program, it will be necessary to prioritize components important to safety. Past efforts to develop generic equipment lists through the use of probabilitic risk assessment models, such as by LLNL during the Seismic Safety Margins Research Program (SSMRP) as well as by BNL, have met with limited success. The results of these efforts suggested that the apparent importance of components was strongly influenced by event / fault trees, which are characteristic of a specific plant, and by the fragility data used in the PRA study, which was based primarily on expert judgment.

In view of these limitation, an alternate approach to systematically identify, categorize, and prioritize electrical and mechanical components important to safety will be taken. This approach will take into account plant system and sub-system functional descriptions, expert judgment, past PRA results, regulatory concerns, existing test data and plant operating and maintenance experience. Components will be grouped by function.

Sub-groups will be established according to the technical issues affecting the fragility of a given component; the sub-groups will in turn provide a basis for defining detailed test specifications. The intent of this approach is to reduce to a minimum the number of component tests required while at the same time maximizing the applicability of test results.

The actual component grouping and prioritization will be performed by LLNL and its consultants. Component groups and important components within each group, and eventually candidate components for comprehensive testing, will be systematically identified as follows:

1. The NRC will be solicited for specific regulatory concerns that should underly component prioritization.

2. Detailed descriptions of power plant systems will be provided by selected architect-engineers, who will also identify which

~ components, in their judgment, are most important within each system and subsystem as based on the functional descriptions.

3. The relative contribution of each system to overall plant safety (or risk) will be determined through consultation with PRA experts. No PRA analyses will be performed.

4. Operating and maintenance histories, obtained from utilities and component vendors, will be examined to determine the importance and degree of aging effects on components.

5. Once specific components or component types have been identified as important, the existing fragilities data base compiled by BNL will be examined to determine to what extent data is already available on these cocponents.

i

6. The component grouping and prioritization will be subjected to review by a select group of consultants. It is anticipated that development of a final grouping and prioritization will require one or more iterations.

E-5

4

7. The NRC will be asked to . review the grouping and prioritization, and to provide additional input, if any. ,

'8. LLNL and its consultants will then generate a list of components for comprehensive testing, divided into three groups.

Group 1: components judged to be of highest priority, which will be tested in FY86.

Group 2: components judged to be of lower priority, which will be tested in FY87.

Group 3: lowest priority components for which the necessity of >

testing will be subject to NRC decision.

Parallel to the grouping and prioritization of power plant components, LLNL and its consultants will develop fragilities test procedures which will be applied in Phase I demonstration tests on selected components. The actual tests will be performed by Wyle i

Laboratories (Norco California) under subcontract to LLNL. LLNL and Wyle

! will jointly develop the Phase I tests procedures. The components to be j tested will be selected by LLNL and Wyle, based on judgment of their

-importance and the component characteristics that can be used to demonstrate the overall testing philosophy and procedure. This makes it possible to begin actual testing early in Phase I before component i prioritization is completed.

The demonstration tests will be performed on two types of . motor control centers, one with a stiff frame and one with a flexible frame.

Anchorage of the motor control centers will be one of the variables in the testing.

! As the final Phase I task, LLNL and its consultants will develop a program plan for comprehensive Phase II testing, including a recommended list of components to be tested and test specifications. The individual L

Phase I efforts will contribute to the Phase II test plan as follows:

!' o Component prioritization will provide insight on the relative j safety significance of various components, thereby having the most direct-bearing on the number and types of components to be tested in Phase II.

o Component grou sing will identify the functional relationship among components and thereby the extent to which test results for a particular component might be a;olicable to others that are similar. Component sub-grouping will influence specified test procedures.

o The fragilities data survey will identify available data for components selected as candidates for testing, and therefore influence the extent of testing. For certain components limited testing may suffice to fill in " holes" in the available data base.

E-6

The proposed approch for defining a recommended Phase II test plan is intended to systematically identify components for testing and optimize the applicability of data obtained from individual component tests, in other words maximize the cost effectiveness of component testing. We recognize, however, that certain regulatory considerations (for example, limiting testing to specific components in reactor coolant systems) may in the end outweigh logistical factors. The final Phase II test plan will therefore be developed in close consultation with the NRC.

For Phase II of the Program the scope of work is summarized below.

To further reduce the scope of Phase II testing, the components will have been grouped during Phase I by function and technical issues. Only representative components from each group will be selected for fragility testing. To arrive at an appropriate component selection, statistical methods will be utilized during Phase II to design the test plan for each component group. Furthermore, analytical methods will be developed during Phase II to predict, to the extent practical, the expected fragility level of each component to be tested. Methods will also be developed to extend the test results to all of the components within each group.

For each component group, the following questions will be addressed.

a. number and size of specimens to be tested;

b. loading environments to be considered;

c. support conditions to be simulated;

d. parameters to be monitored so as to identify all significant failure modes, including partial and intermittent failures.

Utilizing the information obtained in Phase I, analytical methods will be developed to predict, to the extent practical, the fragility level of each component to be tested. It is recognized that at present it is not possible to analytically predict fragility levels of complex electrical equipment. However, some failure modes such as anchorage can be predicted. Such information, when combined with empirical data can be very useful in arriving at an analytical prediction of the fragility of a given component. The ability to analytically predict a fragility level is very important since it is not practical to perform extensive fragility testing. By pre-predicting the test results it will be possible to test the validity of the analytical approach. Furthermore, the results will permit the refinement of the analytical procedures.

The contractor will obtain the test specimens and subcontract to one or more testing laboratories to conduct the tests. The contractor will review the testing laboratory procedures to ensure compliance with the test plan, monitor the test program, witness tests and review testing laboratory reports to ensure compliance with approved testing procedures.

Upon completion of the testing associated with each component group, the contractor will:

E-7

a. analyze the test results and compare with fragilities used in current PRA's and seismic margin studies;

b. identify ali signifie nt failure modes, including partial and intermittent component failures;

c. define the limitations of the present fragility data base for the tested components and recommend improved fragility curves; J

d. assess potential for failures under seismic conditions as opposed to

other dynamic load conditions, such as SRV discharge loads.

As part of this task, the contractor will also develop methods to extend the test results to all of the components within each group.

D. Application of Results These efforts could lead to the establishment of a Component Fragility Data Bank at one of the National Laboratories. No SRP sections or Regulatory Guides will be influenced by this work.

Results from the Component Fragility Project will primarily be used in seismic PRAs and margin studies. It is expected that more realistic component fragilities will lead to simplification of the seismic PRA methodology through the elimination of insignificant branches on event trees and fault trees. Additionally, uncertainties in the present data base, which in turn cause uncertainties in risk estimates, should be reduced as a consequence of these endeavors.

1 i

E-8 l

M

'l 1

i Appendix F Validation of Seismic Calculational Methods Project NRC Project Manager: James F. Costello E

s 0

d s

VALIDATION OF SEISMIC CALCULATIONAL METHODS PROJECT

(

A. INTRODUCTION There is a high-priority need for reliable assessment of the possible consequences to the public if a nuclear power plant were subjected to an earthquake significantly larger than considered in its design basis. That need is documented in the following correspondence:

o Memorandum from J. J. Ray (ACRS Chairman) to N. J. Palladino (Commission Chairman), dated 1/11/83 on "Quantification of Seismic Design Margins."

o Memorandum from J. C. Ebersole (ACRS Chairman) to N. J. Palladino (Commission Chairman), dated 1/18/84 on "Quantification of Seismic Design Margins."

o Memorandum from W. J. Dircks (ED0) to J. C. Ebersole, dated 4/12/84

[ on "Quantification of Seismic Design Margins."

o Memorandum from W. J. Dircks (ED0) to Chairman Palladino, dated 4/12/84 on "Quantification of Seismic Design Margins."

Seismic Probabilistic Risk Assessment (PRA) methods have been applied to identify safety issues for nuclear power plants. A common outcome from these assessments' is the perception that seismic events can affect many plant systems simultaneously, and therefore, can be a significant or even dominant contributor to overall risk. The randomness of the seismic hazard, the uncertainties and variabilities of the needed data, and the approximate nature of the methodology used raise questions of credibility with respect to the results of seismic PRAs. This, in turn, leads to questions about the conclusions drawn regarding safety implications and regulatory actions. Some possible uses, such as comparisons of alternative courses of action, may be less sensitive to residual uncertainties than others, such as quantitative estimates of public consequences. But, while the accuracy needed depends on the intended end use of seismic PRAs, it is nevertheless necessary to validate the methodologies so they may be used with confidence and credibility in the regulatory decision process.

The objective of validation research is to obtain information that can be used by NRC to improve predictions of the behavior of nuclear power plants subjected to large earthquakes and thus improve the regulatory process. The predictive methods to be validated are used in both probabilistic and deterministic calculations.

-The fundamental strategy is to engage in cooperative research programs in order to best stretch available resources. Three efforts have been developed:

n

1. Participation- in a soil-structure interaction experiment being

. performed near Lotung, Taiwan by the Electric Power Research

. Institute (EPRI). r

2. Participation in the Phase II experiments being performed at the Heissdampfreaktor (HDR) facility in Kahl, W. Germany, by Kernforschungszentrum Karlsruhe (KfK).

3. Possible participation in tests of a 1/2.5 scale model of a PWR piping loop to be performed on the large shaker. table in Tadotsu, Japan, by the Nuclear Power Engineering Test Center (NUPEC).

The scopes of these efforts are outlined in Section C.

B. BACKGROUND y

There have been attempts, mostly since the 1950's, to assess the adequacy of mathematical models of. the behavior of large civil engineering structures during earthquake ground motions. For the most part, excitation levels .have been low and the responses measured may well not reflect behavior during very strong ground shaking. However, a necessary step toward understanding behavior in the non-linear range is an understanding of how well linear models do in the range where they should work well.

4 For most civil engineering structures in seismic regions, the largest

dynamic design load comes from postulated natural earthquakes. Most of the dynamic? tests were performed mainly for verifying the design model.

But since in- most tests, the excitation levels were very low compared to seismic excitation levels the design model could not be said to be truly verified for earthquake loading. If a structure, which was dynamically tested, also was instrumented and monitored during natural earthquakes, a good means of evaluating the test results is available. Only a few L structures were subjected to both dymamic tests and natura'. earthquakes, with the response to the latter serving .the same purpose as the test records.

Tanaka et al. [1] compared the period and damping values of fundamental translational modes of 17 multistory buildings in Tokyo obtained from pre-earthquake ambient testing with those determined from response - to a 1968 earthquake of magnitude 6.1. The maximum re amplitudes of buildings ranged from 3 to 25 microns (1 micron = 10 gponsem) in the ambient tests, and from 0.3 to 8 cm during the earthquake. Thus the earthquake-response amplitudes were about three orders of magnitude higher L than the ambient-response amplitudes. The structural periods during the -

l earthquakes increased on the average by about 20% compared to the corresponding values obtained from ambient tests. However, as the authors note, no such simple trend is observed for damping ratios. Though there is a -tendency for the damping to increase with amplitude for most buildings, an opposite tendency was observed in a few others. Since the method' of determination of damping values from ambient or earthquake records is not considered 'very reliable, the value of the comparison of damping. values is somewhat limited, as the authors themselves point out.

F-2

[

Udwadia and Trifunac [2] compared ambient, forced and earthquake-excited vibrations of the Millikan Library (nine-story reinforced concrete) building and Buf1 ding 180 of JPL (nine-story steel frame building). In the Millikan Library building, the forced-vibration tests and three sets of ambient tests were pre-earthquake tests. These tests

.were followed by the Lytle Creek earthquake of 1970 and the San Fernando earthquake of 1971. Post-earthquake vibration tests included ambient tests soon after the 1971 earthquake and vibration tests many months later l in 1972. - It was noted that dynamic parameters showed marked variations during the earthquakes.

As noted by McVerry et al. [3], the level of excitation in the San Fernando earthquake was 10' times that of the Lytle Creek earthquake. The Lytle Creek response was six orders of magnitude greater than typical ambient-test response. Yet for the low level of excitation in the Lytle Creek earthquake, it is noted [3,4] that the ipentified natural periods i were close to those measured in pre-earthc;uake tests. On the other hand, for the San Fernando earthquake, a marked reduction from the  !

l pre-earthquake test values of apparent natural frequency is noted.

According to figures given by Udwadia and Trifunac [2], the reductions are -

about 20% in the N-S fundamental frequency and about 33% in the E-W fundamental frequency. McVerry and co-workers [3,4] also state, on the basis of their analysis of the San Fernando earthquake records, that the N-S and E-W fundamental periods increased from vibration-test values of 0.52 and 0.69 seconds to 0.62 and 1.00 seconds, respectively, during the strongest portion of the earthquake. The ambient tests that followed the ,

San Fernando earthquake showed that, compared to the pre-earthquake i values, the natural frequencies still showed a reduction ranging from 4.8 '

to 14%. However, the vibration tests of 1972 showed that the natural frequencies continued to recover (f.e., increase) and approached almost their pre-earthquake values [2].

In the JPL Building 180, the forced vibration tests of 1963, three earthquakes during 1968-71, and the post-earthquake ambient tests of 1971 form the -basis of comparisons. Here again, for the purpose of these comparisons the entire record of each earthquake was used by the authors to compute transfer functions. The authors demonstrate through these transfer - functions that the apparent E-W and N-S translational natural frequencies during the San Fernando earthquake decrease from the values of the forced-vibrations tests by about 15% and 20%, respectively. The other '

two earthquakes of 1968 and 1970 generated response amplitudes of only about one-tenth of the San Fernando earthquake, and during these earthquakes the apparent natural frequencies decreased somewhat, but recovered' to the pre-earthquake, test-determined values. The ambient tests (performed about six months after the San Fernando earthquake) showed that the natural frequencies recovered (i.e.,-increased from earthquake values) to within 6% of the forced vibration test values.

Udwadia and Trifunac (2) performed a moving-window Fourier analysis of the 1970 and 1971 earthquake records to study the variation of natural frequencies during the different phases af an earthquake. Based on a F-3 m

4 f comparison of different natural frequency values (of a given mode) during an earthquake with those obtained from pre-earthquake tests, the authors

, state that 'the fundamental frequency during moderate earthquake excitations may decrease by as much as 50% of the values determined by pre-earthquake dynamic tests. This decrease (in the two cases reported) was not accompanied by any observable damage, and in both cases the post-earthquake tests showed that the fundamental frequency recovered with time and approached pre-earthquake test values without any repairs. While

being unable to ascribe the nonlinear behavior shown by the comparisons to

' any particular cause, the authors speculate that non-structural damage

might be responsible for such behavior and that the apparent frequency changes may be characterized by time-dependent nonlinear hysteretic behavic.. Foutch and Housner [5] have considered the various possible i

contributing factors to the nonlinear behavior of this building.

4' .The changes observed in the fundamental frequencies of the Millikan l Library building during 'the course of many dynamic tests and the San L Fernando earthquake also gave rise to many studies [6,7,8,9]. In one of

these studies, Luco et al. [6] give tables showing the values of ,

fundamental frequencies determined from different dynamic tests (both before and after the earthquakes) and the Lytle Creek and San Fernando earthquake records. The values of apparent modal damping of the first 2-W mode from some of these tests and earthquakes are also given. The frequency values generally confirm the trends noted by Udwadia and Trifunac [2] and McVerry et al. [3,4]. The apparent damping ratios during the San Fernando earthquake are noted to have increased to about 5.5% from

the pre-earthquake test values of about 1.5% and subsequently to have

decreased to post-earthquake test values of 1.3-1.8%.

An 11-story steel frame building in Japan was subjected to l forced-vibration tests subsequent to which it was subjected to many earth-quakes. Murakami et al. [10] have reported on the analysis of the test

. response to two of the earthquakes. Since the test amplitudes are given in terms'of displacement and the earthquake amplitudes are given in terms of acceleration, the levels of excitation of the test cannot be compared with those of the earthquake. The authors ' note that the natural periods

determined from earthquake response are longer than those obtained from forced-vibration tests. For example, the fundamental N-S period increased from 0.48 to 0.52-0.56 s and the E-W period from 0.31 to 0.38-0.42 s.

. Thus it is seen the increase in period ~ ranges from-8-30%. The authors do I not seem to have investigated the variation of damping characteristics.

One other point noted by the authors is that the higher-mode vibrations of

j. the slabs observed in the dynamic tests were suppressed in the largest-amplitude earthquake vibrations. This is interesting, because if the j earthquake records had been used as input for parameter estimation, the

slab modes would net have been identified. A nine-story reinforced-concrete frame shc?r-wall building in Japan was subjected to forced-vibration tests in 1969, 1970, and 1971. Earthquake response of this j building was observed from 1969 onwards, and Shiga et al. [11] report that, during 1969-1970, 17 earthquakes were recorded. The authors have j selected three 1970 earthquakes for comparison with the three sets of i forced-vibration tests. In this case too, it is not possible to compare l the levels of excitation of the tests on the one hand with those of the i

F-4

i l

earthquakes on the other due to insufficient data. The fundamental 4 natural periods for the earthquakes are seen to have increased by various

, percentages from test values. The vibration-test results in themselves seem to indicate that the natural periods tend to increase with time. But 4

this may be misleading, as the intervening earthquakes may have been at i least partially if not fully responsible for this apparent period lengthening.

Ohta et al. [12] compared the natural periods of 11 tall buildings in Japan obtained from forced-vibration tests with those obtained from the

earthquake response records. This comparison is somewhat questionable 3 because the vibration tests were performed before i.he buildings were completed while the earthquake responses were measurea after the buildings .

were completed. The authors note that the acceleration amplitudes of 21 '

. earthquakes that occurred from 1968 to 1979 ranged from 10 to 50 times larger than the acceleration amplitudes of the vibration tests. According

. to the authors, the fundamental periods determined from earthquake

l. response were on the average 30% longer than those determined from '

vibration tests.

Ogawa and Abe [13] report on their investigation of stiffness degradation of 205 buildings due to earthquakes, by performing ambient i tests before and after two strong-motion earthquakes. Strictly speaking, this work is not a comparison of test and earthquake behavior. Yet, since the post-earthquake tests were performed immediately after the earthquake, possible recovery would have been small and it is not unreasonable to assume that the post-earthquake t sts reflect the same dynamic characteristics of the buildings that the final, low-excitation portion of the earthquake records would have indicated. The buildings, located at the Sendai area of Japan, ranged in height from 1 to 18 stores and included various types of construction. The maximum recorded ground accelerations during the two 1978 earthquakes was in the range of 0.25-0.4

g. The buildings under investigation showed various extents of damage, ranging from almost no damage to severe shear failure in columns. The authors also relate the degree of damage to the change in natural period.

They state that the average value of the ratio of the natural period measured after the quakes to that measured before is 1.31 and that there

, is a strong correlation between the extent of damage and this ratio.

1 J

Test- and earthquake-determined characteristics were compared for the 42-story steel frame Union Bank building in Los Angeles [3,14]. Ambient L tests were performed on this building both before and after the 1971 San i Fernando earthquake [15]. The San Fernando earthquake response was

) analyzed by Beck and Jennings [14] to determine the modal parameters of the building. Comparing the pre-earthquake ambient-test values and the

~

values determined from earthquake response, it is seen from [14] that all the four natural periods determined increased by about 50% during the earthquake. The fundamental-mode damping ratio also showed an increase from the pre-earthquake ambient-test values of 1.7% to a value of 4.2%

< during the earthquake. The higher-mode damping ratios showed an even l larger increase. The effective participation factors determined from ambient-test response were also found to be different from those J

F-5

l' J

determined .from the earthquake response. The effective participation factors represent the mode shapes in a sense. Beck and Jennings [14] note that this=last mentioned result was somewhat unexpected. They conjecture

.that the fundamental-mode mode shape during the earthquake might have been

different.from that during the ambient test.

A comparison of natural periods determined from the earthquake response with those obtained from the post-earthquake ambient tests shows that all the values decrease and tend toward their pre-earthquake test values. The recovery was not complete. The fundamental period, which .

increased from a pre-earthquake value of 3.1 s to an earthquake value of  !

j. 4.61 s, decreased to only 3.8 s at the time of the post-earthquake ambient test. Since no further series of post-earthquake tests is reported to

-have been performed, it is not known if the recovery continued with time.

1 The final example of a building subjected to both testing and earthquake excitation concerns the Bank of California building in Los 4

Angeles i3]. This 12-story . reinforced concrete building suffered l substantial structural damage during the San Fernando earthquake. Ambient tests were performed after the earthquake. McVerry et al. [3] note that this . building showed a great period lengthening during the earthquake.

! They state that the fundamental period identified from a late segment of

, the earthquake response record was 2.37 s, whereas the post-earthquake l' ambient tests indicated a value of only 1.62 s.

Another large civil engineering structures subjected to both vibration tests and earthquakes was the Santa Felicia earth dam. The response of this modern rolled-fill embankment to two earthquakes (of f magnitudes 6.3 and 4.7) was analyzed by Abdel-Ghaffar and Scott [16].

c After the two earthquakes, ambient and forced-vibration tests were i performed on the dam [17,18]. Abdel-Ghaffar and Scott' also made a comparative to the tests [19 study].of Thethe response following of the dam observations to the earthquakes are summarized from thatwith that study.

The maximum accelerations at the crest were'0.2 g and 0.05 g for the i earthquakes and 0.000037 g in the forced-vibration tests. The , period of

, the first upstream-downstream shear mode was longer during the earthquake by 13% compared to ' the forced-vibration tests. The authors note that I

though the water level in the reservoir was not the same .during the

, earthquakes and the tests, the differences in levels were not large enough to affect the period values significantly. During the earthquakes, the response was primarily in the first mode, and very few higher modes were exicted .in the upstream-downstream direction. In contrast, in all the dynamic tests many higher modes were clearly excited. The authors also discuss the difference between the earthquakes and the different tests in

! exciting various other modes. The damping ratios are given to have L increased from a range of 3-4% for the tests to a range of 5-15% for the stronger of the two earthquakes.

The previous example completes the review of comparisons of dynamic test and earthquake response of structures. A different but closely

- related comparison will be briefly reviewed now before concluding this section. Blume [20,21] compares the dynamic behavior of buildings F-6

subjected ~ to vibrations from underground nuclear explosions (UNE) with that observed during natural earthquakes. Many high-rise buildings in Las Vegas, Nevada, were monitored to measure their response to many UNE. ;The-response of the same buildings to natural earthquakes was also measured, j with the same instruments. The strongest of these earthquakes is the San ,

Fernando earthquake of 1971. Unlike the comparisons reviewed until' now, scme of the UNE events produced response amplitudes greater than those i from the San Fernando earthquake. The response amplitudes were of the

- order of a few hundredths of a "g." Comparing the variation of the period at which the maximum response is noted to occur in the -UNE and the i-earthquakes, .for six typical high-rise buildings, Blume notes that the periods are not constant and that they vary with both amplitude and time..

4 He notes that though some correlation has been obtained between period and 1

' amplitude and between period and time, some of the variations must be considered random. Noting that natural periods of complex buildings are

- not constant, and that the natural separation of random bonding between

nonstructural elements has resulted in noticeable period changes, he remarks that period cha'nges do not necessarily indicate damage. He states

• that, in general, period and damping tend to increase with ' time, with loading, with amplitude, and with prior history of loading. He points out that fundamental-period increases of up to 40%, and damping increases of

' - up to 100% have been found in the Las Vegas buildings in the course of the many UNE and earthquakes.

Finally, in the context of this section, reference to a recent paper by Yao and Schiff [22] is appropriate. In this paper, the authors have

- considered the application of system-identification techniques ' to three situations: the analysis of test data, the analysis of strong motion response data, and the analysis of post-earthqua'ke test data. They note that in low-level testing the ' distribution of transducers is becoming more i extensive than before. They state that a logical extension of this would be to use system identification ' methods for determining the optimal i locations of the transducers. Regarding the analysis of earthquake-response- data, they point out the need for characterizing nonlinear

elements of the system and for assessing structual' damage. They consider ambient tests to be a useful means for determining whether a given structure can be occupied prior to detailed inspection and possible repair after- a severe earthquake. They also observe that usually there is a large. variability in the results of data analysis and that there is a need for including an estimation of variability of the results.

The comparisons reviewed above demonstrate both the value and the

. limitation of dynamic tests as a means for predicting the response to strong-motion earthquakes. The major finding that emerges is that if

, large civil engineering structures are modeled as linear systems then it j should be recognized that their modal parameters could show large j variations not only - between low-level tests and earthquakes, but also i during various phases of the earthquake.

The periods generally increase from low-amplitude, pre-earthquake test. values by as much as 20-50% during earthquakes that hardly caused any i

-structural damage. Housner and Jennings [23] note that the increase in

~

the pre-earthquake, test measured period for buildings can be much larger (by as much as a factor of two or three) during strong earthquake motion i

! F-7 t

-m._- ~ , . - , _ . - , , - . , - __ . , , ,-,,e , -w.,_-c,c.,,we,_,- ,

that causes some structural members to yield. (In the extreme case of a structure that collapses, its fundamental period, of course, goes to infinity. ) Damping also generally tends to increase from tests to earthquakes, though this trend has not always been uniform. The substantial changes in the modal parameters from tests to earthquakes indicate that such parameters determined from low-amplitude tests may not be applicable for calculating response to strong-motion earthquakes. Yet one has to note that but for the performance of dynamic tests at various amplitude levels, coupled with earthquake-response measurements, the nonlinear trends would remain even more obscure than they are now.

Another point to be noted is that, despite the knowledge that modal parameters vary with time and amplitude, the models identified from both tests and earthquakes are almost always time-invariant or time-varying linear models. This is mainly due to the lack of availability of widely applicable nonlinear models for the soil-structure system and the appealing simplicity of the linear models.

Regarding the actual methods used in dynamic-testing procedures and earthquake-response analyses, it is seen that the instrumentation has not generally been as extensive in response measurements during earthquakes as during tests. While the reason for this is obvious, this lack of more than one or two earthquake records from a complex structure makes it difficult to identify more than one or two lower modes from the earthquake-response data. It has been noted, further, that any particular earthquake may not excite all the important modes, i.e., even some of the lower modes. The methods used for parameter estimation from earthquake response may not have given reliable results, as many of them are based on statistical assumptions not generally satisfied by earthquake response records. Damping estimates from earthquake-response are generally not considered reliable. For these reasons, some of the results of the comparisons reviewed should only be considered as tentative. More data, collected from low-level tests, more extensive data on response to many earthquakes, and better parameter-estimation methods are all needed before these tentative conclusions are confirmed.

C. PROGRAM PLAN

1. HDR COOPERATION Argonne National Laboratory is the NRC coordinator for this effort.

The HDR facility is a modification of a superheated steam reactor that was decommissioned and modified for research in 1973. A series of experiments (called Phase I tests) was conducted during 1975-83, involving experiments on materials engineering, thermal hydraulics, and mechanical and earthquake engineering. The primary focus of the experiments was comparison of predictions by analytical models with experiments. The Phase II experiments, during 1984-88, are similarly motivated and will examine higher levels of response where damage to structures, systems, and components is expected. NRC participation in the seismic tests will involve providing predictions for the response of structures and piping systems excited by shakers. One series of experiments, in which the con-tainment building will be excited by a large shaker, is planned for June-July 1986. Another series in which piping systems will be excited F-8

into the inelastic range by servo-hydraulic shakers is planned for May-June 1987.

HDR System The Federal Ministry for Research and Technology (BMFT) decided to use the HDR Superheated Steam Reactor for safety research in 1973. It is a 100 MWTH experimental reactor built to demonstrate the feasibility of direct nuclear superheating. The plant, designed in 1967/68, suffered extensive fuel element damage after less than 2000 hours0.0231 days <br />0.556 hours <br />0.00331 weeks <br />7.61e-4 months <br /> of operation.

In order to carry out safety experiments the plant was decommissioned and a new safety loop built which enabled the plant to simulate BWR and PWR operating conditions. Decommissioning work included dismantling and removal of all activated components and equipment no longer needed. Plant modifications included the addition of systems to introduce heat to bring the whole test loop to pressure and temperature. For this purpose, a 4 MW electric boiler is connected into the loop. Water circulation takes place by means of the former primary circuit partial-load pump.

Extensive recabling was carried out to provide instrumentation and electrical power lines protected from blowdown conditions. A Central Data Acquisition System (ZMA) and amplifiers were installed in the control room.

The HDR facility lies on the River Main approximately 50 km east of Frankfurt. The first German nuclear power station (VAK) and a conven-tional power station are immediately adjoining. Figure 1 is a cut away view of the HDR facility showing various features of the plant and attached structures. Plant dimensions, design pressures and volumes are tabulated in Table 1. Table 2 contains other technical data for test loop operations.

HDR Program l The HDR Safety Program has been under way since 1975 and covers all components important to the safety of Light Water Reactors (LWR) such as:

containment containment structures reactor pressure vessel internal fittings to the pressure vessel piping systems valves The principal objectives of the investigations concerning plant structures, systems, and components are:

Improvement of the understanding of the properties and structural behavior of these systems and components in specific incidents relative to safety and quantification of the available safety margins.

Demonstration of the reliability of design methods and in-service inspections of these components.

Optimization and improvement of analytical techniques. ,

F-9

Phase I of the HDR Safety Program was concluded in 1983. This phase involved testing beginning at low levels with tests increasing in severity but not so high that an early failure would put later tests in question.

Phase II comprises tests which will deliberately weaken or damage some of the HDR facilities.

Phase I Earthquake Investigations (EV4000) comprised three different types of low-level excitations of the HDR containment building and additional medium- and high-level excitations of certain piping systems.

In different series of tests the containment building and certain subsystems were excited with (i) a pair of rotating eccentric mass vibrators placed on the operating floor and producing a steady state sinusoidal force in the desired direction, or (ii) impulse forces (100 to 400 kN) produced with reaction rockets mounted normal to the hemispherical dome of the building or (iii) the detonation of small (5 to 10 kg) explosive charges buried at distances of about 30 m from the center of the building. Additional tests on the piping system involved shaker excitation directly applied to the piping system and snapback tests.

Phase II will include high level testing in serveral areas. These are: Pressurized Thermal Shock, Piping Failure Tests /Pipewhip, Piping Stress under Blowdown, Containment and RPV Blowdown and Earthquake High Level Tests. The tests useful for seismic PRA validation are the Earthquake High Level Tests. The main objective of the Phase II earthquake tests is to provide high level excitation to the building and mechanical equipment in the HOR facility. The excitation simulating the earthquake motion to be provided by a large shaker at the operating floor level of the reactor building and additional shakers as needed for excitation of the piping, polar crane and other components. A breakdown of the EV 4000 Phase II test groups are shown in Tables 3 through 6.

7 These tests are expected to move towards severe loading (up to 10 Newton at 1-8 Hz) and high response levels (perhaps as high as 1 "g" acceleration) for structures and components. The minimum expected peak accelerations at different locations during the large shaker tests are as follows:

EXPECTED PEAK ACCELERATIONS 2

Structure / Component Elevation, m Peak Acceleration, m/s Reactor Building, outer shell 40.5 5 Reactor Building, Inner Structure 30.85 4.5 Reactor Building, Foundation Mat -11.0 0.6 Reactor Vessel 13.85 2.5 Role of HDR Tests in Seismic PRA Validation The contribution to the testing needs for seismic PRA validation that can possibly be made by the test data from the HDR Safety Program is discussed here. While the major emphasis is placed on the possible contribution of the high level tests of Phase II, the use of Phase I data is also addressed where appropriate.

F-10

The role of HDR to assist in validation of seismic PRA response predictions accuracies can be a useful one. The HDR offers an integral structure-piping facility of significant complexity. This facility can potentially provide information on structure and piping responses of significant magnitude for comparison with seismic PRA predictions. The use of HDR can be expected to provide useful information on the degree to which the seismic PRA linear elastic analysis methodologies can predict load transmission and structure and piping response in a realistic situation where nonlinear effects and non-proportional damping is to be expected. Specifically concerning piping response, testing in the higher frequency range (above 20 Hz) is planned for Phase II HDR experiments.

This information could be valuable to validate response predictions in a frequency range which appears to be an important component in Eastern U.S.

Earthquakes.

It should also be mentioned that significant information and insight concerning response analysis accuracies was obtained during Phase I HDR experiments. Thus, it was found that structural modeling details are not always necessary to capture the response, as long as the salient feature are modeled. In fact, well thought-out simple models often predicted with greater precision than more complex models. This aspect is very important for seismic PRA analysis simplification and can be further pursued during the Phase II HDR experiments. Finally, HDR test data should permit the evaluation of conservatisms and limits of applicability for the typical linear elastic analyses employed in seismic PRAs.

The HDR system was, during Phase I of the program, subjected to a l variety of loadings in the low and intermediate range of excitations. The '

large shaker tests of Phase II will provide even higher multiple loadings.

The system also is well instrumented. Thus, the application of the SMACS or similar methodologies to analyze a selected group of past and future HDR tests should provide good insight into the expected variability in response and physical properties of a complex reactor system. While it is realized that none of the HDR excitations are representatives of seismic excitation, such an exercise should, nevertheless, give good indications of how structural and component frequencies and damping vary with loading magnitude, type and duration. Similarly, the HDR system could be used to test the effects of modeling variability by performing a number of I

analyses, each by a different analyst, for a single preselected test.

During Phase I of the HDR testing significant insight into the '

variability of structural response parameters, such as damping, for both the HDR building and piping was obtained. This included both variations with load level and type as well as the statistical variability of these parameters under a given loading. This information will now be augmented in the Phase II tests with higher excitation level data. This information should be a valuable contribution to estimation of response variability.

The HDR facility also contains many equipment items which belong to single generic categories, e.g., valves. The possibility thus exist of obtaining quantitative information of the variability, or dispersion of expected response in a generic group of equipment. 5 F-11

F The HDR offers an integral structure-equipment system of significant complexity. It_ is thus likely that response correlations during high level- loadings will occur. This holds particularly for correlations -

arising from- the structural configurations. The use of HDR data for-establishing the expected correlations will therefore be explored.

The proposed HDR testing program appears to present favorable conditions under which boundary interaction effects may be discovered.

These discoveries _ may be more - qualitative than , quantitative, but will serve to identify areas which require attention and candidates for future ,

- testing.

The shield building of HDR is a reinforced concrete structure and will be subjected to dynamic shaker tests as proposed in the testing program. .A significant effort will be made to instrument the structure at locations of expected maximum stress and deformation. However, due to the requirement that containment not be breached it is highly doubtful that gross _ structural failures will be observed during the tests. The most that can be expected is some concrete cracking and separations in shear walls on the interior close to the shaker level. While this is insufficient to yield failure data the information may be useful in deter-mining.-the effect of such nonlinear behavior on the global structural response, i.e., the softening effect. These data can be used in the effort described, in Appendix C, on structural response.

Safety considerations for tae HDR containment integrity preclude the possibility of liquefaction or significant soil failure during testing in the direct vicinity of the building. Thus no significant validation of seismic PRA methods in this area is possible. On the other hand, the HDR site is 'relatively soft, has a high water table and its properties are well documented. This provides an opportunity to evaluate both the ade-quacy of the viscoelastic soil modeling and the construction of impedance functions used as part of typical SSI analyses. It should also be noted that the _ HDR site contains a recompacted soil region in the vicinity of the HDR building. Preliminary calculations by German investigators indicate that, at the boundary of this region and the undisturbed soil,-

sufficient straining may occur during the -large shaker experiments to approach liquefaction ccnditions. The usefulness of this information for validating soil response predictions in seismic PRA analyses should ' be explored.

i

2. NUPEC COOPERATION Brookhaven National Laboratories is the NRC coordinator for this effort. A massive testing effort was started in 1974, under sponsorship i- of the Japanese Ministry for International Trade and Industry (MITI). The l Nuclear Power Engineering Test Center was established and the largest i shake table facility in the world was constructed at the Tadotsu

! Engineering Laboratory. The table is 15m x 15m with a capacity of 1,000 l tons. Characteristics of the table are shown in Table 7[24]. (For

purposes of comparison, the largest shake table in the U.S., operated by

the Richmond Field Station of the University of California, is 6m x 6m w'ith a capacity of 60 tons.) The test series, during 1982-88, will F-12 l

l e

involve 8 specimens representing containment vessels, primary loops, reactor pressure vessels and reactor internals for both PWRs and BWRs.

The test specimens are depicted in Table 8[25]. All specimens will be excited with time histories representative of the Japanese design earthquakes, designated S and S and similar in intent to our OBE and SSE. Responses to those kotions b,ill be monitored. The series is known officially as the " Seismic Proving Test Programme".

NRC's main interest is in the ability of analytical methods to predict-the onset of component damage under very large earthquake motions.

To that end, we are negotiating for tests to be performed after " proving tests" have been completed on one specimen, a PWR tiping loop model. We, and the Japanese, anticipate that the specimen will respond elastically to the proving tests and will be in an undamaged state. The tests, in which NRC will cooperate, would involve increasing the excitation (within the limits of the table) and modifying the specimen to induce inelastic response. Possible modifications under discussion include both adding mass and removing one or more supports. The PWR piping loop test will be performed in 1987. The 1/2.5 scale test model with its support structures is depicted in Figures 2 and 3. As can be seen, the model consists of a steam generator, a coolant pump, and three sections of the primary piping system, i.e., the hot leg, the crossover leg and the cold leg. The reactor vessel is not included because the piping-reactor connection is assumed to be fixed or rigid. Dimensions of the piping section are given in Table 9 and Figure 4.

The steam generator is supported at three locations (see Figures 2 and 3). The top and intermediate supports provide translational restraints in the two lateral directions (i.e., x and y), whereas the bottom supports provide translational as well as rotational restraints in all three directions (x, y and z). The coolant pump, on the other hand, has one set of lateral supports with translational restraints near the top, and another set of support at the bottom with six directional restraints similar to that of the steam generator.

3. EPRI COOPERATION Argonne National Laboratory is the NRC coordinator for this effort.

EPRI is constructing, in a seismically active area in Taiwan, a model about 1/4 the size of a concrete containment. EPRI will also install instruments in the model and in vertical and horizontal arrays in the vicinity of the model and will record earthquake responses over a five-year period. NRC has committed to perform low-level vibratory tests of the model to provide baseline data on modal parameters. Future NRC effort will be in providing analytical models to predict soil-structure interaction effects for the recorded earthquakes and in comparing predictions with observations.

The 1/4-scale model is located within the SMART-I instrumentation network, near Lotung in Taiwan. Recorded data from the SMART-I array (PGA of 0.24 g horizontal and 0.09 g vertical) and an artificial " broad-band" time history were used with the FLUSH code for the design analysis of the scale model. These analyses, based on soil-data from preliminary site  !

F-13

1 tests, indicate significant SSI with rocking being a major portion of the response. The foundation motion from the SSI analyses were used in a 3-D analysis of the containment model structure for estimates of peak response.

Following the completion of construction and the NRC-sponsored vibration tests, the structure will be instrumented with strong motion accelero-graphs. In addition, both surface and downhole response measurement instrumentation will be installed in the field around the structure. Kinematrics instrumentation has been selected by EPRI. The structural response and soil response at the surface will be measured by the Kinemetrics triaxial FBA-13 (force balance acceirometer). FBA-13DH Triaxial Downhole Accelerometer Assemblies will be used to measure the variations in site response with depth. Finally pressure transducers will be used to monitor the contact of the structure with the surrounding soil.

The gages will be located both at the side and the bottom of the basemat.

The free-field and the structural response instruments would be triggered at the 0.01 g level. The Institute of Earth Sciences of the Sinica Academica of Taiwan will maintain the instrumentatic'. It is expected that sufficient data would be collected in 3 to 5 years from the time measurements begin.

Based on the nature of strong motion measurement transducer placements and the availability of dynamic test data, we believe that the following specific investigations will be possible:

o verification of deconvolution methods using surface and downhole measurements o verification of analytical procedures for constructing soil impedance functions based on soil property data. (The test generated impedance functions will be used as basis of comparison).

o verification of SSI methodologies through predictive analyses based on different methods, viz. soil-springs, substructuring, finite element. (This will be based on seismic data).

o verification of structural response analyses.

- based on dynamic tests

- based on seismic response data In addition to the above, if the seismic excitation proves to be strong enough to induce inelastic deformations, verification of methods based on the use of ductility-modified spectra may also be considered.

Studies on the softening and recovery behavior as indicated by frequency shifts during and after earthquakes would also be feasible.

D. APPLICATI0_NS Seismic PRA methods have been used in recent years to clarify safety issues for nuclear power plants. The reason for this is the realization F-14

that seismic events can affect many plant systems simultaneously, and therefore, can be a significant or even dominant contributor to overall risk. The randomness of the seismic hazard, the uncertainties and variabilities of the needed data, and the approximate nature of the methodology raise questions of credibility of the results of seismic PRAs.

At this stage in the development of seismic PRA methodology, we are not sure how " good" the results have to be to support a given regulatory action. But, one thing is certain: the use of probabilistic l.

methods in licensing activities will continue to increase. - For example.

the Comission's 1985 Policy and Planning Guidance (NUREG 0885, Issue 4) includes the following planning guidance in its section on " Assuring The Safe Operation of Licensed Facilities":

i

12. Attention should be given to refining the use of probabilistic assessment techniques to implement Comission policy on safety '

goals, as directed by the Comission, and in other regulatory i applications especially amenable to risk assessment, e.g., in dealing with generic safety issues, formrlating new regulatory requriements, assessing and revalidating or eliminating existing regulatory requirements, evaluating new designs, and setting reactor research and inspection priorities.

13. Whenever probabilistic risk assessment is used in the 4

decision-making process, there must be clear statements of the models used in the analysis with a clear identification of the most significant assumptions and a systematic evaluation of the -

most important uncertainties.

The most likely applications of the results of the research effort will be in:

a. ' ; Reassessment of the risk to the public associated with the

' ' possibility that earthquakes, larger than considered at the j design stage, may occur at operating plants.

b. ' Assurance that the next generation of standardized plants, which are likely to utilize different seismic design approaches than i the current generation, will meet the Comission's safety goals. i t

4 F-15 i

. - - . _ . . . _ . . _ , _ - - _ ~ _ _ , . _ _ _ _ - - - - - - . , _ - _ _ _ , _ . . - _ , _ - . _ _ _

E. REFERENCES

1. Tanaka, T. et al., " Period and Damping of Vibration in Actual Buildings During Earthquakes," Bull. Earthquake Research Institute 47, 1073-1092, 1969.

2. Udwadia, F. E. and Trifunac, M. D., " Time and Amplitude Dependent Response of Structures," Earthquake Eng. and Struct. Dyn. 2, pp.

359-378, 1974.

3. McVerry, G. H. et al., " Identification of Linear Structural Models From Earthquake Records," Proc. 2nd U.S. National Conference on Earthquake Engineering, EERI, pp. 515-524, Aug 1979.

4. McVerry, G. H. , " Structural Identification in the Frequency Domain from Earthquake Records," Earthquake Eng. Struct. Dyn. 8, 161-180, 1980.

5. Foutch, D. A. and Housner, G. W., " Observed Changes in the Natural Periods of Vibration of a Nine Story Steel Frame Building," Proc.

Sixth World Conference on Earthquake Engineering, New Delhi, Vol.

III, pp. 2698-2704, 1977.

6. Luco, J. E. et al., " Soil-Structure Interaction Effects on Forced Vibration Tests," Report No. CE 80, University of Southern California, Department of Civil Engineering, Los Angeles, Oct 1980.

7. Iemura, H. and Jennings, P. C., "Hysteretic Response of a Nine-Story Reinforced Concrete Building," Earthquake Eng. Struct. Dyn. 3, 183-201, 1974.

8. Luco, J. E., " Soil-Structure Interaction and Identification of Structural Models," Second ASCE Conference on Civil Engineering and Nuclear Power, Vol. II: Geotechnical Topics, Paper No. 10-1, Knoxville, Sept 1980.

9. Foutch, D. A. and Jennings, P. C., "A Study of the Apparent Change in the Foundation Response of a Nine-Stor ~ Reinforced Concrete Building," Bull. Seismol. Soc. America 68 (1)y219-229, Feb 1978.

10. Murakami, M. et al., " Earthquake Resistance of a Steel Frame Apartment House with Precast Concrete Panels," Proc. 5th World Conference on Earthquake Engineering, Rome, Vol. 2, pp. 2688-2697, 1974.

11. Shiga, T. et al., " Dynamic Properties and Earthquake Response of a 9-Story Reinforced Concrete Building," Proc. 5th World Conference on Earthquake Engineering, Rome, Vol. 2, pp. 2680-2683, 1974.

12. Ohta, T. et al., "Results of Vibration Tests on Tall Buildings and Their Earthquake Response," Proc. Sixth World Conference on Earth-quake Engineering, New Delhi, Vol. III, pp. 2717-2722, 1977.

F-16

13. Ogawa, J. and Abe, Y., " Structural Damage and Stiffness Degradation of Buildings Caused by Severe Earthquakes," Proc. 7th World Conference on Earthquake Engineering, Istanbul, Vol. 7, pp. 527-534, 1980.

14. Beck, J. L. and Jennings, P. C., " Structural Identification Using Linear Models and Earthquake Records," Earthquake Eng. Struct. Dyn.

8, 145-160, 1980.

15. Udwadia, F. E. and Trifunac, M. D., " Ambient Vibration Tests of Full-Scale Structures," Proc. 5th World Conference on Earthquake Engineering, Rome, Vol. 2, pp. 1430-1439, 1974.

16. Abdel-Ghaffar, A. M. and Scott, R. F., " Analysis of Earth Dam Response to Earthquakes," J. Geotech. Eng. Division, Proc. ASCE 105 (GT12), 1379-1404, Dec. 1979.

17. Abdel-Ghaffar, A. M. and Scott, R. F., " Experimental Investigation of the Dynamic Response Characteristics of an Earth Dam," Proc. 2nd U.S.

National Conference on Earthquake Engineering, EERI, pp. 1026-1035, 1979.

18. Abdel-Ghaffar, A. M. and Scott, R. F., " Vibration Tests of Full-Scale Earth Dam," J. Geotech. Eng. Division, Proc. ASCE, 107 (GT3),

241-269, 1981,

19. Abdel-Ghaffar, A. M. and Scott, R. F., " Comparative Study of Dynamic Response of Earth Dam," J. Geotech, Eng. Division, Proc. of ASCE, 107 (GT3), 271-285, March 1981.

20. Blume, J. A., " Response of Highrise Buildings to Ground Motion from Underground Nuclear Detonations," Bull. Seismol . Soc. Am. -59 (6),

2343-2370, Dec. 1969.

21. Blume, J. A., "Highrise Building Characteristics and Responses Determined from Nuclear Seismology," Bull. Seismol. Soc. Am 62 (2),

'-~

519-540, April 1972.

22. Yao, J. T. P. and Schiff, A. J., " System Identification in Earthquake Engineering," Proc. 2nd Speciality Conference on Dynamic Response of Structures: Experimentation, Observation, Prediction and Control, Ed. Gary Hart, ASCE, pp. 649-657, 1981.

23. Housner, G. W. and Jennings, P. C., Earthquake Design Criteria, Monograph, Earthquake Engineering Research Institute, Berekley, CA 1982.

24. Omori, T., "Present Status of Tadotsu Engineering Laboratory", Pro- ,

ceedings of the MITI-NRC Seismic Information Exchange Meeting, Vol.

I, pp. 103-111, Palo Alto, CA, July 1984. J

25. Watabe, M., " Overview of Research Activities on Earthquake Resistant Design for Nuclear Power Plants", Proceedings of the MITI-NRC Seismic  !

Information Exchange Meeting, Vol. I, pp. 71-93, Palo Alto, CA, July 1984.

F-17

i Reaktorgeb3ude 051,76 Sicherheitsbeh81ter Reactor building V__

containment d.00 I

d Materia 1schleuse

{$p Materials lock

=== m o h,_

8 Y 'I 4 i in

..[.

4 _

[

l i Elementbecken

' Fuel element i +30.85 / / \ -

o v /

storage pool 3 ,

',.g' g:%(3 - \

~~

l

.\

j l

Reaktordruck-  ; i  !.

h m,1

"~

(,j gef30 N k " - ,

i Reactor pressure 5

< Warte vessel s

[',\;l - -

'pj

[' g :2 ,

Control room A K r:- -

:,,- . . /s '

t

+18.00 -

yyl1- -

% - - --  ? .

n v -

y't

'- ,~y &,  ;

i i , -

. _w l

,3.4 -

\ '  ; ,

~ 'Pe,~MW

\ ,. ] '

l 1 .

~ *

, \

l

- a ,

I [~ ~]

b

- wg n 4 i gg m

~q .

[, W ,i .,

n E '

E

,%  %[% ,- s. .:

U r

_0 --- --- --- .

=

' r  :, 1  ; ..

ffh M W/N$;

r-- 5 ],X

'N W N y l t-l  ;

g p p_ x >

m , ,.

EE y ,,

m .J.e.e m .m n _ .,

d .. =

=

g,,.,,

Umw31zpumpe 9

i  ;: . . m __ ,

, = Recirculation pump

's '

.x

.i;

=4, .:p 1

n., ' -

i

~  :.

l G ... .. , /

h$$bb5$ ,'?-(kN.,.,bb'?ki?b!,'E A Q S 10c

/

E-Kessel und VKL-Pumpen i

Electric boiler and test loop pumps FIGURE 1 f

.=

=

Anlage III 1.

! ReaktordruckgebXude LWngsschnitte Appendix Ill 1.

Reactor building vertical sections i

F-18 f

LEGEND:

1 1 - steam g nerator 2 - reactor coolant pump gl ] 3 - hot leg pipe

'i

, 4 - cross over leg pipe 5 - cold leg pipe

.b 6 - lateral supports 7 - support frame r i, r WJ '

3 W

Fig. 2 Proving Test Model of PWR Coolant System with Support Structures -

Front View .

4 1

l l

1 g D F ., IB 4L_ A 4- >

l .$.~

, s 2  %

1 m:,

f g 2 ,

3 7-

. w ._

y

. ... M i

Fig. 3 Proving Test Model of PWR Coolant System with Support Structures -

Side View with Coolant Pump F-19

. - _ _ _ _ - _ - . _ _ _ _ . _ _ _ . . _ _ . _ _ _ _ . ~ . - _ , _ _ _ _ _ _ - . _ _ _ - _ _ .

,~. < ' s. ,

+ - . , >

.,$ ,< N A

+ ::' .~

O,. STEAM 1

, , ,, r :>. ; o, ~ < ,f,iGENERATOR A,, ,

s ,i

,., .K g ,r,< <d- *,% ^ ,/*i v'%v.:,,-

2-y,,

~ ,

n , < , ,.:, v: ,,.s a..,, -.z

, ::,1,,,.,v,".'

,-r,

^ .,

P, > , >.

y<, , ,,n

,,,f",.~, , ,,

,>,;,e.',;, y n:

,. - , >><sy;g...

>-^ +' ' 1.e: ,"

y

< 5', 5 v, y4,. . t, v, f .>j^s,, g /,'

~' igg:W1 ~-

REACTOR Qy . , . ,

. ,,; >:,x ,

, , ,e*

Q;' COOLANT

,s.,,~_>^.,. PUMP

), ^,'<<,',< <5 s, ;3

• d,'y4 .

h8 z$

m?f?'?l;<,bS?' aO w%iw':?<;

r,+, ~ < +, ,-,.+se , ',< sN>

b

> < f.h. , , ,

. <:::,;' 'y;~,"'I

~~ w*

COLO LEG

~~ -

g>,,,

c,,c. ,z ~, s , . - ##f4-

, e f, ' 9,<u~ ;s ' , , ,~,,,,'

~~ ._,,>>.,,9, v% p

' '6 W, ! ,,'R Y j',> ; '< < s. v , , g,,

. , + , > , <-, ,, ^^$,,>~~,'

h>Q f

'A

<l't: n%+

, ,v d, <' + , , , .,

• , ' e'/

b, c.,112,

~-

< t y,%l >

,I#*

^~,w 9 g <<

,, p 14 4

,s '<,, , "'

^

/k V

, /e.

1r

, s l

Fig. 4 Isometric Schematic of PWR Coolant System i

l 1

1 F-20 1

l L

1l Design data Containment RPV Pressurizer Subcooler Dimension

pressure vessel (HOU) f Height 60 12.70 14 11 4.365 m

Inside diameter 20 2.96 1 765 07 m J Wall thickness 17/30 112 45 28 mm i Design pressure

-- overpressure 5.60 110 110 110 ate.g.

j -- underpressure 350 - - - um wat. col, f i Test pressure 7 12 - - - '

3 ate.g.*C

{ Design temperature 155 360 550 320

Material Fine-grained steel 23 NiMoCr 36 BHW 38 BH 395 -

1 FB 50 S -

l Free volume 13349 - - -

m' i  ? Flow: primary - - -

170 t/h j 5 secondary - - -

130 t/h h*ter ir ts Heater circuit Cooling circuit c ,]lation r rs ith j Design pressure 120 120 112 180s wat. col. /m wat.ca

Design temperature 324 324 320 ate.g.,C Material No.1.4550

15Mo3 No.1.4550: 15Mo3 No. 1.4550 - -

Nominal bore 150-250 150-250 350-450 80 mm i Capacity 40-90 90-190 1400-1600 100 t/h 1

)

TABLE 1 Principal design data for HDR Plant Appendix III 1.1 I

I l

Units BWR operation PWR operation No, . . _ _

Soiler power MW 3.5 const. 3.5 const.

1 th 2 Pressure in upper RPV region atm. abs. 40.56 - 90 110 O 250 - 310 3 Temperature downstream of C 250 - 302 electric boiler 4- Enthalpy downstream of kcal/kg 259.2 - 323.6 259.2. - 334.3 electric boiler 5' Capacity in of primary circuit t/h 40 - 90 40 - 90 <

k half-load pump (boiler throughput) 6 Mixed temperature t, in lower *C 180.5 - 276.1 180.5 - 284.2 RPV region 7 kcal/kg 183.9 - 290.1 183.9 - 300 Mixed enthalpy g in lower RPV region t/h 90 - 190 90 - 190 8 Capacity (, of primary circuit full-loadpump(heatsink)

O C 148 - 264.1 148 - 271,3 9 Cold water temperature t,, -

downstream of both subcoolers (heatsink) 10 Cold water enthalpy h,, kcal/kg 150.5 - 274.3 150.5 - 284.1 downstream of both subcoolers (heatsink)

I l

TABLE 2 Appendix 111 1 Technical data for HDR test loop f

1 F-22

= _ - - ..--- _. .

Desired Result of No. h t Groups / Investigations / Tests Data / Calculation Renarks Test Parameters Experimental Objectives Comparison

__ . ..- ~ ... .. ....- _ _

Shaker Testa e R,"GE b) m ed Analyses (stress

"'7 HAGg analysis and structural behavior):

1 - structural non-linearities verification of High excitation level

- load propagation in building analytic techniques for reactor kilding, - transfer behavior of concrete > for load propagation, cuie s.cnts u pnsit in*, soil-

- ccamponent loads structure interection

- soil-structure interaction

- damping of structure, soil, ocsgonents 2 hk r es/sa.desh$.equipar. - reaction and stresses of pipes verification of non- Relationship khhe to EV 2000 4 and vessels under extrene dynanic linear analytic

[ loads (plastificatifon) techniques, staplified test groups 6 Pips and vessel ugadons - damping, as a function of load, techniques "ROBL" and in the inelastic regime.

frequ n y,anterial, preexisting (RENA with ductility) " ROV" (see

• g . Test loop (VIG.)

damage

- hanger influences Appendix A)

. Superheated Stama Generator - coupling /hrling *

(11D0)

. Vessols on resonance level (BES

. Finrvhter storage tanks (BE) 2 and 3 e Pipe investigations in I'*9- - behavior of pipes under air- Verification (also mncy regM20 Hz, and craft innact loads sisplified) analytic -

ally d ined

- damping, as a function of load techniques for ocagonent with high forces on soin and frequency design against aircraft in one m steam line or VIG. - evaluation of support clearance impact, verification of in upper frequency regime structural dynmaics models for bl W calculations M

'6 TABLE 3 PlinR Phase II Test Groups, Sub-Project EV 4000 rig, III,1,j

Test Groups / Investigations / Tests Desired Besult of No. hst Parameters Experlantal Objectives Data / N1ru1

• inn m '

Cogarison

- --- -~- - . - . . . . . . - . . ._

I 4 Inpulse Excitation of Outer Simm21ation of Aircraft Impact Verification of ana- -

investi-

! Containment Shell ("S10 ") lytic techniques for gations l t b) Ioad analysis: aircraft impacts of local

- !=mlaa excitation with behavior s m asively higher inada - safety margin in Japact (100 ms,160 kN-sec) - structure design a

- operating conditions " cold" - global effects of - Intal

- ccaponent design and " hot" (70 bar, 285 CC) ig safety analyses t

- 1rr-al loading of outer .. follow load propagation against ,

7 containment shell in building / auxiliary local l

g building / free field failure

. w= wit loading 7

- damping structure, soil, ocuponents  :

l 0

TABLE 14 A PilDR' Phase II Test Groups, Sub-Project EV 4000 pf ..

Irpulse Excitation of Containment Outer Shell, "SIO " i

. . - . . . - - - - - - = . - . - -

I l

tb.

' inst Groups / Im'estigations / Tests $ 1 red M t of Test Parameters pimental Objectives Data / Calculation Omparison

! 5 Crane Slip Tests ("RJK") b) Ioad analysis (structural Demonstrate safety dynamics reaction) of containment building ,

(and crane) for external  !

Test method: horizontal N E - slip displace ==nts of loads shaker, vertical shaker #""

i noJnted on crane; - impact forces of the sisultaneous shaking slipping crane

- general vibrational j response

,, 1 b

I l

t

! TABLE 5 i P10R Phase II Test Groups, Sub-Project EV 4000 j Crane Slip Tests, "RJK" '9' **

Test Groups / Investigations /Tosts Desired Result of No. Test Parameters Experimental Objectives Data / Calculation Remarks Coiparison 6 S_eismcneter Measurements under Environmental Noise Excitation

(" M -

- Evaluation of actual vibration - Development of a systematic

- HDR reactor building with characteristics to verify the evaluation technique in-known vibrational behavior building model used to develop cluding statistical and floor responses (excitation probabilistic methodolo-values for consonent response) gles

- Development of a scheme for optimized design of maamwo-monts over the estimated 7 - IIDR ocuponents (e.g., BPV) - Evaluation of actual vibration g eo with known vibration characteristics for verifica-response tion of analytic models

- IOR structures and ocuponents - Verification of the paammunent- - Pree=Milation for special with unknown vibration be- and evaluation methods design of man =wesents havior - Evaluation of results and operating condition " cold", camparison with results with and w/o water from existing calculations and maa=wements W -tenn measurements Ca. 1 hr TABLE 6 PIDR Phase II Test Groups, Sub-Project EV 400o Fig. III.4.1 Scisnometer Measurements Under Enviromental Noise -

r .m t y ,3- i ,- n , apyl*

Table 7. Main Performance of the Vibration Table Iten Performances Basarks (1) Maximum Loading Capacity 1,000 ton Size 15m x 15m (2) Table Weight 600 con Including movable parts of exciting me hanism (3) Excitation Directions X*Z Axis Is Horizontal simultaneously Z: Vertical (4) Maximus Stroke Horizontal 1200mm 3

Vertical 1100mm (5) Maximum velocity Horizontal 75ca/s Vertical 37.5ca/s Horizontal 2,670 Gal 500t Inertia load (6) Maximus Acceleration , 1,800 Gal 1,000t Inertia load Vertical 1,335 Gal 500t Inertia load 900 Gal 1,000t Inertia load (7) Excitation Horizontal 3,000 tonf Vertical 3,300 tonf 6,500 tonf-m To satisfy at the time of maxista vertical

(*) Permissible Overturning Momene excitation 12,000 tonf-a No vertical excitation applied (9) Permissible Yaving Homent 3,000 ton-a (10) Duration of Excitation 20 sec. Simultaneous two-axis excitation (at maximum speed sine-wave)

(11) Continuous Excitation 5% of maximus Simultaneous two-axis speed excitation (12) Frequency Range 0-30 Ha l

i F-27

w 5

... ... g i '

$< 0 3 GE E E

N-

= W o e a s

, dg m e  : E E B E = = Q

, a el g = r

= _

Ca3 w Q e w ,

k

• d S $- $ 5 W

a g

th b h e QE z

O e61 aEl 86 59L gg b -

w N l d O

5 j l ," k. W &

th 4

m CO g a I # d- m o e a o o m g0 o 2 3 4 A gl  % 2 co 4 -

e.

}g Q 3

e.

h h h dF "o e' o E

o E

E E Dl d "

QE ,

il -

B

! 55 o sg s

g = e- ud

a E88 mB E.! w!

1 8 F-28

Table 9. Specification of Piping System.

Hot Le.g Cold leg Cross-over Leg Unit Straight Bend Straight Bend Straight Bend Outside Diameter cm 35.3 38.1 33.5 35.0 37.7 39.4 Thickness em 2.9 3.4 2.8 3.2 3.1 3.5 i

F-29

1 l

l Appendix G NRC Seismic Design Margins Project NRC Project Manager: Dan Guzy i

l TA8LE OF CONTENTS l

l A. INTRODUCTION G-1 General G-1 i Regulatory Needs G-2 Strategy and Assumptions i Ge

, B. OBJECTIVE G-5 '

C. SCOPE OF WORK G-6  !

D. TASK DESCRIPTIONS G-9

Part I

Assess Margins - Develop Guidelines - Review Plants G-9 l

Phase I Task I.1 Assessment of Existing Information G-9 Task I.2 Estimation of Existing Margins G-ll Task I.3 Identification of Generic Attributes G-12 Task I.4 Assessment of Margin Adequacy G-13 Task I.5 Development of Screening Guidelines G-15 Phase II Task I.6 Conduct Trial Plant Reviews G-18 Phase III Implementation of Plant Reviews G-19

& Continuing Studies Part II: Identification of Infonnation Needs G-21 Task II.1 Assessment and Development of Failure Data G-21 Task II.2 Assessment of Margins Infonnation G-22 Task II.3 Assessment of the Effect on Margins of G-23 Relay and Circuit Breaker Performance During Strong Motion Task II.4 Assessment of the Behavior of Operators G-23 During and Immediately After Strong Motion l Task 11.5 Assessment of the Effect of Design and G-24 Construction Errors on SDM Issues Task II.6 Assessment of Inherent Calculational G-25 Design Margin Task II.7 Assessment of System Changes, Such as Added G-26 Redundancy and Enhanced Operational Modes, On SDM Task II.8 Assessment of the Effect of Uncertainty in G-27 Non-linear Structural Behavior on SDM Issues E. SCHEDULE AND DELIVERABLES G-29

ACKNOWLEDGEMENTS This Program Plan was developed by the Expert Panel on Quantification of Seismic Design Margins

• in collaboration with Lawrence Livermore National Laboratory.

Other contributors were Paul Smith, who formulated an earlier version of the Plan and Abel Garcia and Don Bernreuter who provided valuable assistance in reviewing the Plan.

I 1

• The-Panel is made up of the following members: R.J. Budnitz (Chairman),

P.J. Anico, C.A. Cornell, W.J. Hall, R.P. Kennedy, and J.W. Reed.

)

NRC SEISMIC DESIGN MARGINS PROJECT l l

l G.E. Cummings, J.J. Johnson, R.J. Budnitz i l

l A. INTRODUCTION General Recent studies such as the Seismic Safety Margins Research Program (SSMRP) estimate that seismically induced core melt frequencies come from earthquakes in the peak ground acceleration range from 2 to 4 times the safe shutdown earthquake (SSE) acceleration. Other studies indicate that Seismic Category I structures and PWR primary coolant piping have similar high margin against seismically induced failure. The performance of conventional power plants in past earthquakes confirm the existence of substantial seismic capacity in nuclear power plants. However, from a licensing perspective, there is a continuing need for consideration of the inherent quantitative seismic margins because of, among other things, the changing perceptions of the seismic hazard. A sound, practical seismic margins program, utilizing margins to failure analysis and seismic probabilistic risk assessment techniques would serve to minimize the need for changing regulatory requirements and licensing actions as estimates of the seismic hazards change. In addition, it can provide a sound basis for confidence in the seismic capacity of nuclear power plants and serve to indicate, if necessary, places where seismic risk should be reduced.

The Seismic Design Margins Program (SDMP) discussed in this Plan will provide the technical basis for assessing the significance of design margins in terms of overall plant safety and will identify potential weaknesses that might have to be addressed. This, in conjunction with past studies and ongoing validation and fragility efforts, should be effective in resolving the quantification of seismic design margins issues.

A general definition of seismic design margin (SDM) is expressed in terms of how much larger than the design basis earthquake an earthquake must be to compromise 4

I G-1

th2 safety of a plant._ Margin is dLfintd at tha plant level and at the level of functien/ system, structures, equipment and compcntnts.

Regulatory Needs At the June 11, 1984 joint meeting of the NRC Staff Working Group on SDMs and the Expert Panel on SDMs* there was an extensive discussion of regulatory needs. The regulatory needs were distilled to the following:

1. There is a need to understand how much SDM exists. Margin in this context is expressed in terms of how much larger than the SSE an earthquake must be to compromise the safety of the plant.

2. There is a need to create a seismic margin framework that can filter, and to some extent absorb, the effects of changing knowledge and hypotheses in geology and seismology. This framework is needed to provide an engineering perspective and to avoid, when possible, over-reaction to these changes.

3. There is a need to understand the influence of design and t construction errors, systems interactions and effects of operator behavior on the seismic response of plants.

4. There is a need for research to understand the behavior of plants under loads induced by low-aagnitude earthquakes characterized by high frequencies, short duration, and highly localized ground

. ~ motion. It is recognized that the response of the plants may be qualitatively different for these earthquakes than for those for which the plants are designed. There is also a need to put into

, perspective the significance of increased high frequencies (above 10 Hz) for lanjer earthquakes.

5. There is a need to provide additional assurance concerning the validity of the models and input data now used in seismic probabilistic risk assessments (PRAs) so as to increase confidence in

, the validity of PRA results.

The hierarchy is that the first need is the most important and the remaining four are secondary.

• The Panel is made up of the following members: R.J. Budnitz (Chairman),

P.J. Amico, C.A. Cornell, W.J. Hall, R.P. Kennedy, and J.W. Reed.

i G-2

Th] first Regulatory Need is too broad to sirva th] purp;se of defining scope and sh:uld be furth;r specified. Review of this plan sh:uld lead to a core focused d;scription and highlight other potential needs. This iteration process is n:cessary to assure a successful, balanced, cost-effective and responsive SDMP.

Desirable inputs would include a current list (last two years and in the near future) of regulatory concerns and issues relating to seismic margins. This list should be constructed by the NRC staff and updates to it made as the SDMP progresses. Typical current issues might be:

o Margin implications of recently recorded high frequency features of earthquakes such as measured in the New Brunswick earthquake.

1 o Margin implications of differences found in earthquakes recorded at '

anomalous sites such as shallow soil sites, o Margin implications of recently recorded earthquake spectra on the design assumption that the peak vertical acceleration is 2/3 the peak horizontal acceleration.

l The central regulatory issue is that the safe shutdown earthquake (SSE) used for the design of plants can be exceeded with finite probability. This exceedance is i due to a variety of reasons: 1) the SSE has a finite return period and thus I larger earthquakes are expected but with longer return periods and 2) the shape of design spectra can be exceeded. The basis for the adequacy of the seismic design of plants thus cannot rest on the size of the SSE alone and must also rest on there being adequate SDM.

The criteria used for plant design are known to embody SDM which in most cases is b211eved to be large. However, this SDM primarily arises from prescriptive procedures rather than performance requirements that specify the various margins quantitatively. This means that the existence and sources of SDMs are generally known but their quantitative values are generally unknown. Since qucatitative SDMs are unknown, a natural regulatory question is: "What minimum level of earthquake will compromise plant safety and where are the weakest links?"

SDMP is intended to take the next step. It will quantify the earthquake levels that could compromise plant safety as part of the process of assessing SDMs. To the extent possible this will be done by quantitative studies that will be planned to develop results with generic implications. To the extent that this generic work may fall short, SDM screening guidelines will be developed. These guidelines G-3 m

\

will be us:d to assess thu adequacy cf SDMs through varicus types of plant-specific reviews. Although some quantitative results on SDMs do exist, they '

l are not based on sufficiently broad and varied studies to meet NRC needs adequately.

Strategy and Assumptions The overall strategy for the SDMP relies upon a preliminary set of conclusions that seem to be a consensus in the connunity of knowledgeable experts familiar with seismic design margin issues. Perhaps the most important consensus is a confidence that reactors designed, built and operated according to the NRC's current regulations in most cases possess significant margin above the SSE levels for which they have been designed -- stated another way, there is a high degree of confidence that earthquakes must be significantly larger than the SSE before they will compromise the safety of the plants.

This confidence, which the SD@ hopes to confirm through specific studies (but is prepared to fail to confirm depending on how the studies turn out), has been taken into account in the development of the Plan. There is a conviction not only that such margin exists, but that it should be possible to demonstrate the existence of this margin quantitatively. Moreover, the strategy of the Plan is based on the t

assumption that it is possible to group the ensemble of plants into a manageable number of sub-groups, characterized by similar properties, such that statements about SDM will be feasible for each sub-group separately.

l If these consensus opinions and convictions are borne out by the studies contemplated, then the result will be statements about how much SDM exists for each sub-group of plants. It is recognized that the statements about SDM I resulting from this work cannot be comprehensive -- that is, they cannot cover all i issues involved in the plants' seismic responses. In particular, there are a few l

! issues (discussed separately below) for which new research is needed before their effect on margin can be stated confidently. Nevertheless, there is the expectation that there will be groups of plants, and groups of plant attributes j (such as groups of structures or groups of equipment types), where statements about SDM can be made confidently.

l The Plan also rests on several assumptions made as a starting point for the Plan's development. These assumptions may change as the work of the Plan evolves, and as the input of other knowledgeable parties is factored into the Plan.

G-4

1. We assume that both deterministic and probabilistic techniques will b2 used to analyze h a much SDM exists, and as tools in the SDM screening guidelines to be developed as part of the Plan. We also assume that plant risk, using the traditional risk end-points of PRAs figure -of-merit (core-melt forfrequency, offsite determining risk, etc.)

that plant safetywill is be used as thein compromised the SDM Analyses. _

2 We assume that plants, systems and components can be grouped usefully for the purpose of studying SDM.

3. We assume that guidelines will be required to conduct plant reviews in the event SDM adequacy cannot be resolved in a generic manner.

4. We assume that requiring plant-specific seismic PRAs as the principal vehicle for analyzing SDM at various plants is not a desirable solution to the task of finding a screening method for SDM. We assume that an approach can and will be found involving less extensive analysis, although it is possible that seismic PRAs may be needed for some plants to provide a piece of the required technical information.

5. We assume that those accident sequences that are principal contributors to the seismic part of plant risk can be identified in a generic way insofar as there is any pattern identified among the plants.

6. We assume that during the execution of the SDMP the validity of l seismic PRAs will be established sufficiently to permit confidence in the conclusions based on their use.

B. OBJECTIVE The objective of the Seismic Design Margin Program (SDW) is to develop the technical basis to resolve SDM issues. This will be accomplished through specific studies using both deterministic and probabilistic techniques. The SDW has several goals:

o To define hierarchical relationships of margin at the plant, function, system, structure, equipment and component level.

o To assess the amount of SDM.

o To identify generic attributes related to SDM.

o To determine the adequacy of SDM's.

o To develop SDM screening guidelines.

This SDW Plan is to provide a comprehensive approach (set of tasks) to address the SDMP objective and goals.

G-5

i C. SCOPE OF WORK The scope of the SDMP has been developed in two parts. Part I encompasses the-body of the Program within six tasks. Part II tentatively identifies eight tasks thich will provide infomation to help resolve SDMP issues. Part I is split into three phases. The first phase is an intensive effort of about six months duration

< leading to a preliminary assessment of margin adequacy and a set of trial guidelines. In Phase II trial reviews of two plants will be accomplished. Phase III continues with further plant reviews and studies dependent on the results of Phases I and II.

Part I: Assess Margin - Develop Guidelines - Trial Review of Plants Phase I Task I.1 Assess Existing Information Task I.2 Estimate Existing Margins Task I.3 Identify Generic Attributes Task I.4 Assess Margin Adequacy Task I.5 Develop Screening Guidelines and Methods for their Application Phase II Task I.6 Conduct Trial P1 ant Reviews Phase III Implementation of Plant Reviews and Continuing Studies Part II: Identification of Information Needs Task II.1 Assess Failure Data Task II.2 Assess Margins Information Task II.3 Relate Capacity and Performance of Relays and Breakers During Strong Motion to Margin Issues Task 11.4 Relate the Behavior of Operators During and Immediately Af ter Strong Motion to Margin Issues Task II.5 Assess the Contribution of Design and Construction Errors to the Compromise of Safety Task II.6 Assess Inherent Calculational Design Margin (best estimate vs. design code)

Task II.7 Assess the Impact of System Changes on SDM Task II.8 Assess the Impact of Non-linear Structural Behavior on Margin Issues The relationship between these various tasks is illustrated in the SDMP Flow Chart shown in Figure 1 and the SDMP Schedule shown in Section E. The timing is aimed at resolving NRC concerns about seismic design margins in three years. To do this

! effectively, two trial plant reviews will be done during the first year with further work to be defined following these trial reviews. Following completion of G-6

^

L

Tasks I.1 - I.4 the NAC staff must establish, eith:r firmly or tentatively, criteria whereby adequacy is to be judged, building on the work of Task I.4, before screening guidelines can be developed in Task I.S. Also, NRC needs to comment'on the screening guidelines developed in Task I.5 before the trial plant reviews can be conducted in Task I.6. The infomation needs identified in Part II

- must be available in a timely manner to allow completion of the Part I tasks.

e J

1 l

G-7

SDMP FLOW CHART PART I I.1 A Assess Existing Infonnation

[\ Part II

!.2 I.3 Estimate Identify Ex isting Generic Identification of Information Needs Margins Attributes II.1 Assess Failure Data

% [ Phase I II.2 Assess Margin Info.

II.3 Relay / Breaker Performance I.4

& Assess Margin II.4 Operator Behavior Adequacy 11.5 Design & Constr. Errors II.6 Calculational Margin II.7 Systems Margin II 8 Nonlinear Struc. Behavior NRC Input V

I.5 Develop Screening Guidelines V

=h A l I.6 l Conduct Trial Plant Phase II Reviews g ah A l Implementation of Plant Reviews Phase III and Continuing Studies g l

l G-8

)

D. TASK DESCRIPTIONS Part I: Assess Margin - Develop Guidelines - Review Plants Phase I Task I.1: Assessment of Existing Information.

Background. Significant infomation is currently available relating to SDM issues although it is known that gaps in knowledge do exist. To make sure that relevant existing information is utilized it is necessary to identify a task to assess this information. The information exists in a variety of j sources although it is believed that the information most relevant to SDMP will be found in seismic PRAs. Nevertheless, all sources need to be reviewed.

Results from existing programs may give qualitative if not quantitative insights into the margins issues. Decisions and supporting studies from the Systematic Evaluation Program (SEP) may help in this regard. Also, existing fragility and equipment qualification data needs to be assessed. This data may give insights related to existing margin on equipment and components if not for system and plant margin. Existing earthquake experience data must be reviewed to give insights about margins in structures as well as equipment.

Ten to fif teen seismic PRAs will have been performed on plants in the near term. These PRAs are usually an adjunct to an internal events PRA and are a likely source of information for the determination of the adequacy of SDMs in risk tems. These plants and their PRAs are also a likely source of t information for the development of SDM screening guidelines. ,

l 1

The SDMP approach requires a consistent set of quantified seismic PRA results. At a minimum we anticipate the use of NRC, utility and EPRI developed seismic hazard functions which will require re-quantification of the existing seismic PRA results. The sensitivity of PRA results to this hazard f unction will have to Fa tested. We also anticipate the need to examine closely the development of the plant logic models and fragility and uncertainty descriptions and the possible need to modify some of these relative to their characterization in the existing seismic PRAs. Any ,

modifications will lead to a need for some requantification. Also, the development of SDM screening guidelines will require a close interrogation of the existing seismic PRAs in the form of sensitivity studies, uncertainty G-9

analyses and tha evaluation of various alternate configurations. One of tha seismic PRAs to be reviewed will be the SSMRP Zion Study since this is the most comprehensive analysis and was conducted with NRC instead of utility funds. Thus, information from this study is not only the most detailed but also the most readily available. Utility seismic PRAs such as were done on Millstone 3, Seabrook, Limerick, Oconee and Indian Point 2 and 3 will also give useful insights.

Objective. This task will evaluate existing information to extract that information useful to SDM issues and has two main objectives:

1. To provide information to be used to identify generic attributes, determine margin adequacy, and develop screening guidelines (Tasks I.3, I.4 and 1.5).

2. To provide infonnation to help in estimating existing margins (Task I.2). Results from Task I.2 will be used to establish what we know and don't know about existing margins.

Approach. The approach will be to review as much information as possible during the first phase of this Program. Further review including

requantification of the existing seismic PRA's may be required later depending on Phase I findings. For Phase I, results of all studies including PRAs will

be taken as stated with limited interpretation and no requantification. With i this in mind, the steps in the review process will be as follows.

1. Review existing studies relating to SDM issues, e.g. SQUG.

( 2. Review existing fragility and equipment qualification data.

3. Review existing earthquake experience data.

4. Review existing information on hazards analysis, e.g. LLNL and EPRI studies.

5. Review existing seismic PRAs (approximately 6). In Phase I, quantities of interest will be tabulated and compared on a plant, system / function, and component basis. Such quantities would include core-melt frequency, important accident sequences, systems / functions, components and structures, and earthquake level at the median failure point. This assessment will be based on the PRA results as presented with any necessary requantification-determined after Phase I.

6. For each of the reviews the applicability of the results to each of the Part I tasks needs to be assessed and the findings documented in a way useful to each task.

l G-10

)

t l

Justification of Task Based en Regulat::ry N_ ecd. Th2 results of this task will be used by the other Part I tasks to address regulatory needs as stated for those tasks.

Relationship to Other SDMP Tasks. The results of this task provide input to the other Part I tasks. ]

Milestones. See Schedule in Section E. Details to be developed later, i

Task I.2: Estimation of Existing Margins l

Background. There is a need to establish to the extent possible what the SDM is in existing plants and particularly to establish where the uncertainties appear too great or where gaps in knowledge occur. Based on the infomation in Task I.1 an early determination of what these margins are will be made by this task to establish a base of knowledge from which the remainder of SDMP can be conducted.

Objective. The objective is to estimate the margins present in existing plants by estimating what size earthquake is necessary to compromise plant safety. It is desired to estimate SDMs for the plant as a whole, as well as at the system / function level, the structure level, and the component level.

Approach.

1. Identify candidate plants based on Task I.1 results for which sufficient infomation is thought to exist to enable an estimate of SDMs.

2. Review the analysis for each candidate plant, and develop additional specific information as needed.

3. Determining how SDM will be defined in an operational sense for the purposes of this Task, in tems of specific ground motion i characteristics or other physical parameters. The definition of SDM given in the introduction to this SDM Plan is quite general: "SDM is expresseo in terms of how much larger than the design basis earthquake an earthquake must be to compromise the safety of a plant".

4 Determine the SDMs for each candidate plant, including an estimate of I the uncertainties in the determination. The SDMs are to be estimated at the plant level and/or the level of functions / systems, structures and components as feasible.

G-11 [

1 l

l

5. Determina, thrcugh existing sensitivity studies or n::w sensitivity studies to be carried cut where nuded, the extent to which thm SDMs calculated above depend on various assumptions, models, and generic data. Particular attention must be paid to the sensitivity of the results to the shape of the hazard curve or response or fragility function curves.

6. Document the results.

Results of the Task. The results of this task will be estimates of the '

SDMs existing for each of a small group of candidate plants chosen because sufficient infonnation is available upon which to base such a set of estimates. The SDMs are to be expressed where feasible at the plant, system / function, structure, and component levels. The uncertainties in the SDM estimates are to be presented and discussed, along with insights through sensitivity studies as to where the SDM estimates most depend on various assumptions, models, and generic data.

Justification of Task Based on Regulatory Need. Results of this task will partially satisfy Regulatory Needs 1 and 2.

Relationship to Other SDMP Tasks. This task will take input from Task I.1 and supply input to Task I.4.

Milestones. See schedule in Section E. Details to be developed later.

Task I.3: Identification of Generic Attributes Background. To make the assessment of SDM as efficient as possible, generic attributes need to be identified. This will help group plants or plant systems so that further review can be better focused.

Objective. The objective of this task is to identify those generic attributes of the plants studied that seem to be important contributors to plant strength, and those that appear to contain important vulnerabilities to earthquakes. Focus should also be on identifying systems, structures and components which can be eliminated from further investigation.

This is to be accomplished, insofar as feasible, for broad groups of plants,and/or broad groups of functions, and/or broad groups of systems and components. The specific groupings are to be one result of this effort.

Approach.

1. Identify candidate generic attributes that emerge from the studies in Task I.1 and I.2.

2. For each identified candidate generic attribute, determine the i extent to which the attribute is present in each of the plants  ;

studied. The focus of this effort is to achieve a rough grouping of I plants and/or attributes.

G-12

3. Grsup th2 plants and/or attributes, taking into account the extent to which each plant group or attribute group possesses a high, medium, or low degree of correlation and consistency. Determine the extent to which the groupings seem ' natural' or ' forced' -- that is, whether the groupings seem to arise from some generic property of the plants or attributes that might be present-in other similar plants not studied, or whether the groupings seem not to arise from any identified generic property.

4. Estimate the extent to which the groups of plants or groups of attributes might be extendable to include other plants or attributes not specifically present within the analytical infonnation set studied.

5.

Estimate the extent to which the conclusions in (3) and (4) above

' depend on. (are sensitive to) differences in assumptions, models, or data used. Particular attention must be paid to whether the insights are sensitive to the hazard curves used or shapes of the fragility or response function curves.

6. Estimate the overall confidence in the groupings arrived at (high, medium, or low confidence, for example), based on the analyses in (4) and (5), and on the confidence thought to be present in the overall groupings.

Results of the Task. The results of this task will be identified generic attributes of plants studied that seem to be important contributors to plant strength, and that seem to be important aspects of plant vulnerability to earthquakes. For each identified generic attribute, the task will identify and discuss the sensitivity of the conclusions to assumptions, models, and data used. An overall confidence rating will be given to the groupings specified.

Justification of Task Based on Regulatory Need. Results of this task are mainly to support the remaining Part I tasks.

Relationship to Other 50MP Tasks. This task will take input from Task I.1 and supply input to Tasks I.4 and I.5.

Milestones. See schedule in Section E. Details to be developed later.

Task I.4: Assessment of Margin Adequacy.

Background. Based-on the results of the preceding tasks, an assessment of margin adequacy needs to be made to determine the necessity for proceeding on to final guideline development or the plant review stage. It may be that margins will be found adequate at this stage or found adequate for some G-13

classes of plants, systems or compon:nts. Such a finding w:uld eliminate or minimize the need for further effort. Close coupling with NRC will be necessary in this task since the final judgement as to adequacy of SDM must be made by the NRC. Any required additional studies are included in this task rather than in the preceding tasks.

Objective. The objectives of this task are:

1. To assist NRC by providing results for their decisions on the adequacy of SDMs.

2. To provide inputs to the ' development of SDM screening guidelines.

. (Whether or not this objective is needed will dependent on NRC decisions on whether SDMs are adequate.)

Approach.

1. Interact with NRC to finalize an approach to determine adequacy .

based on the experience gained in the preceding tasks.

2. Review and sumarize the reports from the preceding tasks.

1

3. Revise this Sumary or perform special studies as a result of the review in (2), as required by NRC.

4. Interact with NRC on their decision on adequacy of SDMs, as required by NRC.

Justification of Task Based on Regulatory Needs. Regulatory Needs 1 and 2 would be addressed by this task since existing margins will have been assessed and statements about adequacy made. Also, Need 3 will partially be addressed since in reviewing the seismic risk assessments some feeling concerning the effect of design and construction errors, systems interaction and operator behavior will be made. Also, in going through tne process of making statemants of margin adequacy, a better feeling will be forthcoming about the validity of the models and input data (Regulatory Need 5).

Relationship to Other SDMP Tasks. This task will use input from Tasks I.1, I.2 l

and I.3 and provide input to Task I.S. i l

1 Milestones. See schedule *a Section E. Details are to be developed later.

G-14 a- - - - - - - _ - - . - - _-

~ . . . - - - - . .-

Task I.5: Development of Scretning Guidelines.

Background. The purpose of screening guidelines is to help detennine how to proceed with plant specific reviews and to help assess the adequacy of SDMs in these reviews. The need to develop screening guidelines necessarily rests on NRC conclusions on t'ne technical results developed in Task I.4.

It may be necessary to revise some of the efforts that were perfonned in Task I.4 to reficct NRC insights that resulted from their decision process on SDM adequacy or to answer various NRC questions that arise. The primary goal of SDMP at this point is to identify screening guidelines that can be used to support an NRC finding that SDMs are adequate or inadequate at a plant for which no seismic PRA exists. It is anticipated that plants which do pass the guidelines will be judged to have adequate SDM, but those plants that fail may or may not have inadequate SDM. Additional effort will be required to determine if the SDM is not adequate, e.g., a risk assessment.

Assuming some areas are found of questionable adequacy, the key NRC decision will be the priority that will be assigned to further plant or topic reviews.

Possible decision areas are plants located at sites expected to have local site amplification, plants with current or earlier seismic criteria, geographical location of plants, Westinghouse versus other PWRs, PWR versus BWR, magnitude of the risk estimates in the existing PRAs, and so forth.

There are two major f actors that impact the development of SDM screening

-guidelines:

i o NRC insights as a result of Tasks I.1, I.2, I.3 and I.4 and their decision process on the adequacy of SDMs.

o A close examination of the differences between the plants or parts of plants that were and were not found to have adequate SDMs.

l- Close coordination with NRC will be required in this task since they play the dominant role in one of these two factors. For example, it is possible to develop guidelines relating to the adequacy of SDMs as follows:

"The seismic loads used in the design of structures shall be shown to be a factor of 2 or more times the median loads that are expected to occur assuming the occurrence of realistic earthquakes with a peak acceleration equal to the l SSE."

l G-15 i

and/or:

"The median capacity of a structure (including consideration of inelastic energy absorption) shall be shown to be 5 or more times the best estimate loads that are expected to occur assuming the occurrence of a realistic earthquake with a peak ground acceleration corresponding to the 5 x 10-4 per year probability level."

It is important to specify areas and forms of guidelines as completely as possible before the SDMP technical efforts in this task begin. This is because:

o The area (structural response or capacity in the above example) of applicability of the guideline must be viewed as necessary and acceptable to NRC.

o The form of the guideline (factor of 2 or 5, etc., in the above exaniiTeT must also be acceptable to NRC as an appropriate one for determining adequacy of SOMs in the specific area (structural response in the example).

Note that in the above examples an additional complication arises.

Specifically, most seismic design and PRA information available does not include the best estimate seismic loads for the structures. These loads would need to be calculated if NRC desired a guideline of this type.

A key issue to consider in developing guidelines is current versus earlier seismic design criteria. This is because the SDMs for plants designed to current criteria are thought to be larger than for plants designed to earlier criteria. Global guidelines are appealing as they would simplify the plant-specific SOM reviews significantly, independent of the original criteria used.

i Although desirable, we anticipate that it may not be possible to develop acceptable global guidelines. One of the reasons for this is there are many site- and plant-specific features that have a significant impact on seismic risk even when the plants are designed to the same criteria. This is a widely recognized consideration in PRAs for internal-initiated events and seems likely to be true for seismic PRAs also. If this consideration is a significant factor in seismic PRAs then it means that design practice is as or i

more important than design criteria. Data on the performance of non-nuclear i

G-16 L

facilitics in past carthquakes tends to confirm the importance of design practice in seismic vulnerability. However, global guidelines may help in grouping plants for consideration and/or for determining which systems in plants of a certain vintage or vendor should be considered.

We thus anticipate that it may be necessary to develop SDM screening quidelines that are less global and more technical. One problem with such guidelines is that they may require significant efforts by the assessor as part of the plant-specific review. Recall that compliance with our example guideline would require the assessor to perform best estimate response analyses. Also, the uncertainty and variability within and between plants needs to be taken into account. If, for instance, the structures do not satisfy the guidelines but the piping does, then additional guidelines or a risk analysis may be required.

To reduce the ultimate burden on the assessor, the guidelines should be developed in such a way that they offer a spectrum of options and/or levels.

For example, for structures, one sequence of such guidelines might be the following:

1. Structural response margins (as in the example).

2. Margin against structural yielding.

3. Margin against structural failure.

NRC staff and utility efforts in (1) could be used in (?) if the guidelines in (1) were not satisfied and those efforts in (1) and (2) could be used in (3) if the guidelines in (1) and (2) were not satisfied. All of this technical effort would be of use in a seismic PRA if the guidelines in (3) were not satisfied.

Objective. The objective of this task is to develop SDM screening guidelines.

The purpose of these guidelines is to assess the adequacy of SOMs in plant-specific reviews and to help structure the type and priority of such reviews.

Approach.

1. Develop a trial set of screening guidelines to be used in subsequent plant reviews and evaluations. The area and form should be specified in the guidelines as appropriate.

G-17

2. Devaltp a procedure for applying thase scresning guidelines fcr postulated plant reviews. It may be that certain classes of plants need no or minor review (for instance, post-1973 plants), it may be that only certain parts of plants need reviews (reactor coolant loop piping may have adequate margin), or it may be that plants can be grouped into broad classes (by vendor, A/E firm, age, etc.).

3., Elicit NRC guidance on the proposed guidelines and method of implementation. This may involve further work in Tasks I.2, I.3, and steps (1) and (2) of this task.

4. Finalize guidelines.

5. Document guidelines and review methods.

Justification of Task Based on Regulatory Needs. This task is preparatory to conducting plant reviews.

Relationship to Other 50MP Tasks. This task receives information from Task I.4 and provides information to Task I.6.

Milestones. See schedule in Section E. Details are to be developed later.

Phase II Task I.6: Conduct Trial Plant Reviews Background. To meet the Regulatory Needs concerning SOM issues, it may be necessary to review plants individually or on some sort of selective group basis. As an example of this, selective review of reactor coolant loops in a few plants within vendor categories helped in resolution of pipe restraint issues in the LLNL Load Combinations Research Program. The SDM reviews will be necessary if uncertainty exists concerning SOM adequacy after the completion of Task I.4. It is now believed such uncertainty will exist because current seismic risk assessments and SEP studies show that unique plant features frequently dominate risk and therefore would need to be looked at to assure margin adequacy. To test this review concept trial plant reviews will be conducted with further reviews implemented in Phase III if found necessary by NRC.

The screening guidelines and reconnendations for their use from Task I.5 will be used to conduct these reviews. Detailed interaction with the NRC staff will be required not only in establishing the guidelines but in implementing the G-18

l l

raviews. The reviews could be dono in ecnjunction with the utilities as was i

+

done in the Seismic Qualification of Auxiliary Feedwater Systems Program.  :

l l

Objective. The objective of this task is to conduct trial plant reviews and i report the results so that the adequacy of margin is established.  !

Approach.

l

1. Use data from Task I.3, I.4 and I.5 to establish a plant review '

process (walkdowns, PRAs, systems / components to be included, plant  !

groupings,etc.).

2. Catalog the important components and characteristics of each plant l to be reviewed with annotation. '

3. Do a rough assessment of each plant to establish priorities.

Factors to be taken into consideration include generic attributes ,

and work related to other NRC efforts (ISAP, licensing issues, etc.). '

4. Do more detailed reviews if required or until NRC feels the margins issue is resolved.

5. Document findings.

Justification of Task Based on Regulatory Needs. This task is expected to satisfy Regulatory Need 1 relating to the assessment of existing margin and to help satisfy Regulatory Need 2 relating to stability of the seismic review process. Some information relating to the other three Regulatory Needs should also come out of this task. Of course, if SDMP were to be carried to the ultimate extreme of doing complete reviews of all plants including individual j risk assessments, all Regulatory Needs would be met. At this time it is felt that reviews of all plants should not be necessary to satisfy all Regulatory Needs.

Relationship to Other SDMP Tasks. This task receives guidelines information from Task I.S.

Milestones. See schedule in Section E. Details are to be developed later.

Phase III: Implementation of Plant Specific Reviews and Continuing Studies At completion of the first two phases of SDMP it is anticipated that further work will be defined possibly including some further plant reviews. Although it may be necessary to do a rough assessment of all plants, detail reviews on a repre-G-19

s ntativa set should ba suffici:nt. Phase III cf SDMP is thm ccnduct cf furth:r work including the implementation of additional plant reviews, as necessary, over This work a two year period. Details will be determined during Phase I and II.

might include further guideline development, risk assessment or further review of those existing risk assessments including some requantification and similar work to that done in Phase I.

1 l

G-20

Part-II: Identification of Infcmation Needs The following tasks are identified as being important to establishing actual seismic design margin including performance of the plant and operator immediately af ter the occurrence of an earthquake. These tasks are executed concurrently with those of Part I and provide input to the end product, i.e. the ability to make statements about seismic design margin for a specific plant or for groups of plants. Their execution may fall under a aifferent program either within the NRC cr the industry. However, they are presented here to emphasize their importance to the end objective.

t Task II.1: Assessment and Development of Failure Data.

Description. Considerable seismic qualification testing has been perfomed using shake tables. Some informal descriptions exist of weaknesses or f ailures that were observed during testing. This is contrasted with the data on the observed performance of equipment in past earthquakes where few if any failures have been reported. Obtaining these qualification test results may l

be difficult since they may be considered proprietary. In addition, other NRC and industry programs are gathering and generating useful data concerning fragility, e.g., Component Fragility Program, Structural Fragility Program.

The objective of this task is to obtain general information on failure modes and failure levels of equipment and structures as an input to NRC decisions on the adequacy of SDMs.

i The approach is to engage testing laboratories to develop reports sumarizing general infomation on weaknesses or failures of components observed during testing. Typical information sought would be year of test, general 4 description of component, description of failure or loss of function,

excitation description, and mounting conditions. The infomation from all testing laboratories would be assembled for like components and added to the existing fragility data base. Candidate components for additional testing would be identified. Note that significant effort would be expended to assure confidentiality of the data and its source.

G-21

l Justification of Task Based on Regulatory Nends. Regulatory Need 1 is l

partially addressed in the sense that this task provides limited infomation on the fragility of equipment. Accurate fragility information is the most important factor in the detemination of the acceleration range that is the dominant contributor to seismic risk. This range is assumed to be an important f actor in NRC's decisions on the adequacy of SOMs. Regulatory N[

5 is partially addressed by this task in the sense of validation of assumptions on equipment fragility.

/

Relationship to Other SDMP Tasks. This task provides inputs to Tasks I.2, I.3, I.4 and I.5 11.2: Assessment of Margins Information.

Description. The state of knowledge in the fields of seismic risk analysis, component and structural fragility, and systems behavior is evolving at a rapid pace. Many research programs and additional seismic PRAs are underway, or planned, with results expected in the next two to three years. Examples include the NRC Component Fragility Program, NRC Equipment Qualification Program, NRC Category I Structures Program, EPRI Hazard Program, EPRI Piping and Fitting Dynamic Reliability Program, foreign research programs (HDR, These Japanese, French, etc.), ISAP, RMIEP, Diablo Canyon Seismic PRA, etc.

programs will provide valuable input to the seismic . design margin issue and need to be explicitly recognized and included in the'5')MP. The timing of their results is not compatible with the schedule for 'lask I.1, hence, an additional task is identified. L

\

The objective of this task is to assess newly developed information related to seismic design margin on an ongoing basis and provide input to Tasks I.4 and I.5 of the 50MP. To do so requires identifying and monitoring research programs and other studies which are likely to provide infomation pertinent to the assessment of seismic design margins.

(

Justification of Task Based on Regulatory Needs. Regulatory Need 1 is  ! ,

directly addressed by this task.

Regulatory Needs 2 - 5 are addressed in the same manner as the subject research programs address these issues.

G-22

Relationship to Other SDMP Task.s. The results of Task 11.1 provide input to Tasks I.4 and I.S.

Task II.3: Assessment of the Effect on Margins of Relay and Circuit Breaker Perf ormance Daring Strong Motion Description. Recent studies have shown that seismically-induced circuit breaker faifures may inhibit the proper operation of safety systems during and after strong motior.. .These f ailures may be caused by relay chatter in the circuit breaker electrical systems. For example, inadvertent operation of the anti-pumping relays may lock out the circuit breaker. Failure of manual and test switches in these circuits may also be a problem. Such failures have been ignored in most seismic risk assessments to date, yet relay chatter can be caused by a relatively low intensity of ground motion. Therefore, the effect of relay chatter during strong motion may have a pronounced effect on system and plant SDM.

Since some circuit breakers, relays and circuits may be affected by this type of failure, the number of these and the generic implications need to be assessed.

l The objective of this task is to estimate the influence on plant and system

, SDM of circuit breaker misoperation caused by relay chatter or switch malfunctions. An examination of nuclear power plant systems will be made to detennint how prevalent are circuit breaker system designs subject to strung motion failure. If large populations of systems are susceptible to these

( problems, effort needs to be expended on developing their fragility functions and the consequences of their failures on plant safety.

Justification of Task Based on Regulatory Needs. This task will contribute to the resolution of Regulatory Need 1.

Relationship to Other SDMP Tasks. The results of this task will provide information for Tasks I.2 and I.3.

II.4: Assessment of the Behavior of Operators During and Immediately Af ter Strong Motion Description. Concern exists that reactor operators or the displays they monitor may be so affected by tr.e grcund motion that they will ce unable to perform their required functions. Recent experimental data from Japanese G-23

tests suggests the optratcr cay b2 provented from reading and reacting to his displays at ground motions above 0.2 to 0.4 g. These tests indicate the actual level is influenced by chair design, chairs with casters being the better perfomers. Other efforts to assess the effect of earthquakes on operator performance are being undertaken by the NRC Office of Nuclear Regulatory Research as part of their human factors research program.

Information from these efforts as well as results from the revisits of seismic PRAs need to be assessed. From this assessment, a feeling for the relative effect of seismic induced operator error compared to various hardware and structural failures needs to be made. The overall impact of operator behavior during and af ter earthqitakes to SDM issues can then be understood.

Justification of Task Based on Regulatory Needs. Regulatory Need 1 is directly addressed by this task.

Relationship to Other SDMP Tasks. Results from this task will provide input for Tasks I.2 and I.4.

II.5: Assessment of the Effect of Design and Construction Errors on SDM Issues Description. The amount of seismic design margin at a plant is dependent not only on the design as envisioned but also as constructed in the field. Design and/or construction errors can play a significant role in this issue. Several reports in the literature have demonstrated the large effect that design errors can have on seismic risk and consequently on SDM. To date, however, no generally accepted, practical means of including the effects of design and construction errors in a risk calculation is available. Further, no methodology or procedure exists to identify design and construction errors outside of the current QA programs. Because design or construction errors many times lead to surprises in the performance of facilities when subjected to an earthquake, this issue needs to be addressed in the format of SDMs.

Justification of Task Based on Regulatory Needs. Regulatory Needs 1 and 5 are addressed by this task. Regulatory Needs 2 - 4 are not addressed.

Relationship to Other SDMP Tasks. This task provides input to Tasks I.2 and 1.4 G-24

Task II.6: Asstssment of Inhirent Calculational Design Margin.

Description. Over the past 15 years, a significant evolution has occurred in seismic analysis and design procedures. In most cases, the evolution has been to increased design requirements introduced by the methods of seismic analysis and specification of the hazard and system parameters. Examples include:

o Seismic design ground motion defined by US NRC Regulatory Guide (RG) 1.60 -- three components of motion and broad-band response spectra.

o Damping values defined by US NRC RG 1.61. .

o Control point definition at foundation level -- US NRC Standard Review Plan (SRP) 3.7.2.

o Broadened in-structure response spectra for equipment and piping system qualification defined by US NRC RG 1.122.

o Envelope procedure for analyzing multi-supported systems such as piping systems -- US NRC SRP 3.9.3.

o Modal combination rules for closely-spaced modes -- US NRC RG 1.92.

These and other requirements were introduced due to legitimate concern regarding uncertainties in analysis methods and parameter values. However, they were introduced with little consideration of their ramifications on subsequent elements in the seismic analysis chain. It is well-recognized that conservatisms compound as one moves from the seismic input - to soil-structure interaction (SSI) - to structure response - to-equipment and piping response. This compounding was not explicitly considered in developing new requirements. It is a major source of seismic design margin. Quantifying these margins contributes to our ability to make definitive statements concerning seismic design margin on a plant-by-plant basis or for groups of plants. In addition, it is items such as these that may be potential screening guidelines.

An effective approach to this task is first to assemble existing information on quantification of calculational margin. Also, one must identify candidate seismic design criteria (methodologies, parameter values, etc.) which introduce substantial conservatism in calculated values of response. To quantify these conservatisms, one must perform comparative calculations with L

best estimate technology, the result being the margin introduced by the specified calculational procedure.

G 25

Justification cf Task Based an Regulatory Nerds. Regulatory Need 1 is partially addressed by this task, i.e. the results of this task lead to quantification of seismic design margin due to response prediction techniques. Regulatory Need 2 is addressed by demonstrating the large calculational margin which exists due to seismic analysis methodology for specific situations. Small perturbations in the definition of the ground motion should not exceed this margin.

Regulatory Needs 3 - 5 are not addressed.

Relationship to Other SDMP Tasks. Task II.4 obtains input from Task I.1 and provides results to Tasks I.2 and I.3.

Task II.7: Assessment of System Changes, Such as Added Redundancy and Enhanced Operational Modes, on SDM Description. Over the past 15 years, systems design has evolved as have seismic design criteria. Thoughts have evolved concerning redundancy of components, redundancy of safety systems, isolation of components, manual operation of portions of systems, power trains and isolation, etc. Changing NRC requirements have led to this evolution such as the implementation of fire

-protection regulations. Many of these changes may have overall plant safety consequences when considering the seismic hazard even though they were not implemented to enhance seismic safety as in Task II.6. Consequently, those plants with f avorable systems aspects may be more reliable under the seismic  ;

hazard and this may lead to screening criteria for the SDMP.

The objective of this task is to identify systems aspects of nuclear power plants which lead to significant SDM and consequently constitute a screening criterion by themselves.

Justification of Task Based on Regulatory Needs. Regulatory Need 1 is partially addressed by this task. Regulatory Needs 2 - 5 are not addrgsstd.

Relationship i.o Other SDMP Tasks. Task II.7 obtains input from Task I.1 and provides results to Tasks I.2 and I.3.

G-26

~ ._ - _ _ . . _ _ - _- ._J

Task II.8: Assessment of the Effect cf Uncertainty in Non-linear Structural Benavior en SUM Issuts De scription. In assessing seismic design margin, the realistic behavior of structures under earthquake loadings must be taken into account -- in particular, the non-linear behavior of structures. The performance of structures in past earthquakes has demonstrated the significant reserve capacity of ductile structures subjected to earthquake loadings and the poor estimates of behavior made by linear elastic predictive techniques.

Currently, when non-linear behavior is taken into account, it is treated by very approximate techniques. The ductility modified response spectrum technique, originally developed by Newmark, is the most extensively used approach to date. It is based on numerous studies of single-degree-of-freedom systems but lacks correlation with physically realistic structural configurations.

The problem is clearly two-fold: data acquisition on the behavior of structures and analytical modeling of the behavior. Data acquisition is partially addressed by the NRC Category I Structures Program although for limited structure types and scale models. Additional data for full-scale structures and of differing construction are needed. Analytical techniques, adequately benchmarked, need to be developed to permit analysis of structures with significant non-linear behavior.

Seismic PRAs consider the range of possible earthquakes at the site and, hence, consider earthquakes substantially higher than the design level event.

Seismic PRAs quantify structural failure predictions which become an important element in seismic risk analysis and in the assessment of seismic design margi n. Non-linear structure behavior dominates these predictions and requires validation.

In addition to the effect on structure forces, non-linear structure behavior has a significant impact on the input environment to subsystems (piping l systems and equipment). This needs to be taken into account when estimating their capacity for seismic PRA and seismic design margin analyses purposes.

The objective of this task is to identify and quantify the margin introduced by linear or approximate calculations for structures that respond in a non-linear way to strong seismic excitations. The approach to executing G-27

this task is a combinaticn of data acquisition (existing and n:w) at the structure and structure element levels and analytical development and verification of non-linear analysis techniques. Benchmarking these techniques with existing data and application to physical structures completes the effort.

Justification of Task Based on Regulatory Needs. Regulatory Need 1 is directly addressed by this task. The non-linear behavior of structures is an important source of margin -- one which requires consideration and quantification to permit quantitative margin statements to be made.

Regulatory Needs 2 and 4 are indirectly addressed by this task since the damage potential of earthquakes defined therein is assessed.

G-28

E. 50MP SCHEDULE AND DELIVERABLES Task FY 85 FY 86 & FY 87 PART I Phase I @

I.1 Assess Existing Information V I.2 Estimate Existing Margins V-I.3 Identify Generic Attributes 7 I.4 Assess Margin Adequacy V T' I.5 Develop Screening Guidelines Phase II @

1.6 Conduct Trial Plant Reviews V T Phase III "

Implementation of Plant Reviews and Continuing Studies Reports as tasks are completed PART II @

Identification of Information Needs V Work to be done separately following identification of information needs.

Deliverables:

@ LLNL input to Expert Panel on Capacity & response f actor data. (11/84)

@ Input to the Expert Panel. (2/85)

@ Expert Panel Report on Tasks I.1 - 1.4 & Identification of Information Need s. (4/85)

@ Expert Panel Report of proposed guidelines and trial review procedure.

@ Report on results of the 2 trial planc reviews & recommendation for Phase'III implementation.

G-29 i

81RC POfteI 25 U.S. NUCLE AR LE3 ult. TORY COMastOS40N 1. *e.EPORY NUMeEA (AssvaeW 6P TiOC. amp ved Man, ,f ears (2 Sol

","o*"323'- BIBLIOGRAPHIC DATA SHEET NUREG-1147 SEE INSTRUCTIONS ON TME REVERSE.

2. TITLE AND SUSTITLE 3 LE AVE SLANyf Saismic Safety Research Program Plan r y A

A 4 DATE REPORT COMPLETED t, MONTM YEAR t AuT ORise ,

,dJune 1985

. rp 6. DATE REPORT ISSUED f' MONTH VEAR 5., ,/

June 1985

,. PE_,ORMno ORoAN,2Ar,0,. NAME ANo MA,uma AooREs ,, ,.i Co., , ,RO4cmA ,,,OR u,,,T Nu,,,ER Division .of Engineering Techilology /,

Office 'of Nuclear Regulatory Research [ "'""6"^"'*****"

U.S. Nuclear Regulatory Comission / '

Washington, DC 20555 i /

. ,1

10. SPONSORINa ORGAN 42ATION NAME AND MAILINo ADDRES$ fince,eele Coel 11a. TYPE OF REPORT j

Same as 7 above i J

% i

• h [ a. PERICO COyERED flacsusewe essa;

't M 4 </

i i 12 SU8PLEMENT ARY NOTES 4 Research Program Plan is AssTRAcT isco.oren er re i

\f

.g y't l This' plan describes the safety issues, regyla' tory needs, and the research necessary to address these needs. The plan also discusses the relationship between current and proposed research within the NRC and fesear'ch sponsored' by other government ~

agencies, universities,industrygroups,[professionalsocieties,andforeignsources. a h, Y t

C] i;{

y a

ff

$$ S sy 1 a t

'r? )

i f

e Q 14 DOCUMENT ANALYSIS -e KEYWORDS /DESCRsPTOR$ ,< gg gygggggggyy ResearchProgramPlan, Seismic /Research,SelsmicPlan,SeismicSafety,

TATEME,.T Seismic Hazard, Seismic Risk,. Seismic Margins, Seismo, tectonic frogram!

Soil Response, Structural Resp ~onse, Piping Reliability, Component Fragilit /, Unlimited Validation of Seismic Calculational Methods *

,, ,E cur,,, cL,,,,,,c,,,og (TA,s espel m OnTi,iERs,0,1 ENDEOTE,,Ms unclassified e rne ->

unclassified

17. NLMSER OF PAGE$

18 price

UNITED STATE 3 cou==cenes an

• • * "c'8 usma PA'o NUCLEAR RET _ULATORY COMMISSION WASHINGTON, D.C. 20655 wasw. o c.

"" h 8#

OFFICIAL BUSINESS PENALTY FOR PRNATE USE, $300

)

I 1

' ~ , ,