Sie sind auf Seite 1von 208

Seismic Fragility Application Guide

SED
R I
A L

LICE

M AT E

WARNING:
Please read the Export Control
Agreement on the back cover.

Technical Report

Seismic Fragility Application Guide


1002988

Final Report, December 2002

EPRI Project Manager


R. Kassawara

EPRI 3412 Hillview Avenue, Palo Alto, California 94304 PO Box 10412, Palo Alto, California 94303 USA
800.313.3774 650.855.2121 askepri@epri.com www.epri.com

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES


THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN
ACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCH
INSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THE
ORGANIZATION(S) BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM:
(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I)
WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR
SIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS
FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON OR
INTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTY'S INTELLECTUAL
PROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USER'S
CIRCUMSTANCE; OR
(B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER
(INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVE
HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOUR
SELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD,
PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT.
ORGANIZATION(S) THAT PREPARED THIS DOCUMENT
ABSG Consulting Inc.

NOTICE: THIS REPORT CONTAINS PROPRIETARY INFORMATION THAT IS THE INTELLECTUAL


PROPERTY OF EPRI, ACCORDINGLY, IT IS AVAILABLE ONLY UNDER LICENSE FROM
EPRI AND MAY NOT BE REPRODUCED OR DISCLOSED, WHOLLY OR IN PART, BY ANY
LICENSEE TO ANY OTHER PERSON OR ORGANIZATION.

ORDERING INFORMATION
Requests for copies of this report should be directed to EPRI Orders and Conferences, 1355 Willow
Way, Suite 278, Concord, CA 94520, (800) 313-3774, press 2 or internally x5379, (925) 609-9169,
(925) 609-1310 (fax).
Electric Power Research Institute and EPRI are registered service marks of the Electric Power
Research Institute, Inc. EPRI. ELECTRIFY THE WORLD is a service mark of the Electric Power
Research Institute, Inc.
Copyright 2002 Electric Power Research Institute, Inc. All rights reserved.

CITATIONS
This report was prepared by
ABSG Consulting Inc.
300 Commerce Drive, Suite 200
Irvine, CA 92602
Principal Investigators
R. Campbell
G. Hardy
K. Merz
This report describes research sponsored by EPRI
The report is a corporate document that should be cited in the literature in the following manner:
Seismic Fragility Application Guide, EPRI, Palo Alto, CA: 2002. 1002988.

iii

REPORT SUMMARY

The Seismic Fragility Application Guide provides utilities with in-depth guidelines for
performing fragility analysis as part of a seismic probabilistic risk assessment (SPRA). These
cost-effective and practical procedures and the resulting SPRA can support riskinformed/performance-based (RI/PB) applications.
Background
The American Nuclear Society (ANS) has developed draft standard ANS 58.21, External Event
PRA Methodology Standard, for addressing the risk to nuclear power plants from earthquakes
and other external events. The standard provides requirements for addressing external events
ranging from simplified screening to sophisticated levels of probabilistic risk assessment. The
primary focus is on seismic PRA. The standards approach for SPRA is intended to be identical
to the American Society of Mechanical Engineers (ASME) standard (ASME, 2001) for internal
event PRA. The standard uses a graded approach and considers the studys scope and level of
detail, plant specificity, and degree of realism. Requirements for three graded SPRA levels are
provided. The graded levels are labeled Capability Categories 1, 2, and 3. This document focuses
on developing seismic fragilities for structures, systems, and components (SSCs) in an overall
seismic PRA. Existing methodologies for developing fragilities, ranging from simplified
methods to detailed analyses, were used in the individual plant examination for external events
(IPEEE) program in varying degrees of detail. The U.S. Nuclear Regulatory Commission
(USNRC) review of IPEEE (USNRC, 2000a) identified some shortcomings in methodology and
practice that require improvements for future work. This document addresses USNRC
comments, correlates existing methodologies with requirements for the three Capability
Categories in the ANS standard, and provides implementation guidelines and example problems.
Objective
To provide utilities with seismic fragility analysis methods to support regulatory and nonregulatory applications.
Approach
The project team reviewed the ANS External Events PRA Standard requirements for the three
levels of performing seismic PRA and summarized the steps necessary to satisfy the standard.
Team members reviewed existing fragility methodologies (as applied to IPEEE and other
commercial and research seismic PRAs) and correlated them to the standards requirements.
Comments by USNRC on the IPEEE submittals were reviewed and are addressed in this report.
Additionally, more recent methodological suggestions for development of fragilities (Kennedy,
1999) and supplemental test data that could enhance the database for developing fragilities
(Ueki, et.al., 1999) also were reviewed and addressed.

Supplemental methodology and procedures were developed in this study for cases where existing
methodology and procedures in the technical literature were undefined. The project team
developed example problems to supplement existing sample problems in the literature and to
address technical methods not adequately addressed.
Results
This report provides an implementation guide for deriving seismic fragilities together with
representative example fragility calculations. The basic fragility methodology has been
documented in selected technical papers and industry publications (for example, EPRI report
TR-103959). This report updates that fragility methodology to reflect recent methodological
changes in the literature and correlates the appropriate steps to the requirements of the ANS
External Events Standard. Examples of fragility development that complement examples in TR103959 are provided to enhance information available to personnel who perform fragility
analysis. This document and TR-103959 provide methodology, procedures, and an array of
example problems that encompass most situations that fragility analysts will encounter. The
following have also been included in this report: a summary of USNRC comments on fragility
methods from the IPEEE program; recommendations to address appropriate NRC comments; a
review of recent experience data (test and earthquake data), which can be used for the fragility
development process; a review of recent advances in the technical literature relative to seismic
fragility methodology; and, a review of recent fragility requirements within the ANS External
Events Standard and recommendations for meeting these requirements.
EPRI Perspective
The industry and the regulators are moving toward RI/PB methods; therefore, it is important that
utility engineers understand, become familiar with, and use seismic PRA methods. The basic
parts of a seismic PRA are identifying the hazard, analyzing the system, and evaluating structural
fragility. Of these, calculating fragilities is the closest to the structural engineering discipline.
Accordingly, EPRI considers developing tools such as this report to be critical for structural
engineers who use SPRA methods.
Keywords
Earthquakes
Seismic Risk
Fragilities
Probabilistic safety assessment
Individual plant examination for external events

vi

EPRI Proprietary Licensed Material

ABSTRACT

The American Nuclear Society has developed draft standard ANS 58.21, External Event PRA
Methodology Standard, for addressing the risk to Nuclear Power Plants from earthquakes and
other external events. The Standard provides requirements for addressing external events from a
risk-informed perspective. The requirements range from simplified screening to progressively
more detailed levels of Probabilistic Risk Assessment. For seismic events the standard provides
requirements for conducting Seismic Margin Assessment (SMA) and Seismic PRA. Although
some examples are provided in the Standard for risk informed applications of SMA, the primary
focus of the Standard is on Seismic PRA. The primary steps in conducting seismic PRA are the
development of the seismic hazard, the development of a fault tree/event tree model of the plant
response to earthquakes and the development of fragilities for basic events included in the plant
model. The ANS Standard provides high level requirements for the seismic hazard, the plant
system modeling and the development of fragilities for three progressively more detailed levels
of Seismic PRA. The three levels of detail are denoted as Capability Category 1, 2 and 3. This
document focuses on the development of seismic fragilities of structures, systems and
components for use in Seismic PRA for any of the three catagories. The importance of the
interface between the fragility analysts and the hazard and plant system modeling analysts is
emphasized.
Existing methodology for development of fragilities ranges from simplified methods to detailed
analysis. Existing methodology that would meet Capability Categories 1 and 2 and in many
cases, Capability Category 3, is portrayed in EPRI TR-103959 and was used in the IPEEE
program in varying degrees of detail. The USNRC review of IPEEE submittals (USNRC,
2000a) identified shortcomings in methodology and practice that require improvements for
future work. This document addresses the USNRC comments and correlates existing
methodologies with requirements for the three Capability Categories in the ANS Standard.
Supplemental methodology and example problems are provided to enhance the existing
methodology applied in IPEEE.
First the basic methodology for conducting Seismic PRA and development of fragilities is
summarized. Then a detailed discussion of the NRC comments on IPEEE, as they relate to the
ANS Standard and the existing methodology, is provided. The most important USNRC
comments on methodology are related to the use of a uniform hazard spectrum, use of surrogate
elements and scaling of soil-structure interaction analysis. The use of uniform hazard spectra
and scaling of soil-structure interaction analysis are addressed in the ANS Standard and this
document. The use of surrogate elements is not addressed in the ANS Standard. The Standard
requires that screened out components have a low contribution to risk implying that they do not
need to be included by representation as a surrogate element. The NRCs principal issue in
IPEEE was that the screening level was too low and representation of screened out elements by
vii

EPRI Proprietary Licensed Material

use of a surrogate element or elements resulted in a significant contribution to risk. Screening


criteria and the failure rate target for screening are addressed in this report and an example
screening level fragility is presented. Other NRC comments on IPEEE are related to expertise of
the analysts and reliance of non-seismically designed structures controlled by organizations
outside of the plant boundary. These are not methodological issues but are nevertheless
important and are discussed in this report.
A step-by-step discussion of the development of fragilities to meet the three Capability
Categories in the ANS Standard is presented. Table 4-1 is a comprehensive comparison of the
fragility parameters that are to be addressed, the existing methodology, lack of methodology or
procedures and the requirements of the ANS Standard. Where existing methodology and
examples are determined to be adequate, the methodology is referenced and is not repeated.
Where existing methodology or procedures are not complete, supplemental criteria and examples
are provided.
The example problems in the Appendices are primarily related to issues raised by the NRC
review of IPEEE or methodology not addressed in existing Seismic PRA and fragility
methodology guidelines. They address scaling of spectra, developing fragilities from earthquake
experience data, developing screening levels and applying screening criteria. Other examples
address the derivation of fragility from design analysis, liquefaction related fragility and the
fragility of un-reinforced masonry walls.
A Capability Category 2 analysis would, in general, be required for future risk informed
applications although, in some cases, Capability Category 1 should be acceptable. The minimum
requirements of NUREG-1407 for Seismic IPEEE would correspond to Capability Category 1
and most IPEEE submittals did not go beyond the NUREG-1407 requirements. NUREG-1407
required that only a point estimate of CDF be calculated using a mean hazard curve and a single
composite fragility curve, thus, most submittals did not contain an uncertainty analysis as
required for Capability Category 2. The EPRI and LLNL hazard studies are considered to
comply with Capability Category 2. In most cases, if new structural analysis was conducted,
fragilities developed for IPEEE would either comply with Capability Category 2 or could easily
be updated to Capability Category 2. For many cases where in-structure spectra were scaled
from the design basis spectra, the fragilities would likely only be applicable to Capability
Category 1 based on requirements within the ANS Standard.
The information in this document, in conjunction with EPRI TR-103959, is intended to envelop
most cases for development of seismic fragilities for Capability Categories 1 through 3. Some
very unique cases (such as for dams) as noted in the USNRC review of IPEEE submittals,
require specialized expertise that is not documented here.

viii

EPRI Proprietary Licensed Material

ACRONYMS
AFRS
ANS
BWR
CDF
CDFM
CUS
DBE
EUS
FOAKE
FPS
GERS
GIP
GMI
HCLPF
HFD
IPEEE
IRS
LERF
LLNL
NEP
NPP
PGA
PSA
PSHA
PSV
PWR
RAI
RE
RLE
RRS
SA
SD
SDOF
SMA
SPRA
SPSA
SQRT
SQURTS
SSCs

Amplified Floor Response Spectra


American Nuclear Society
Boiling Water Reactor
Core Damage Frequency
Conservative Deterministic Failure Margin
Central United States
Design Basis Earthquake
Eastern United States
First of a Kind Reactor Engineering
Feet per Second
Generic Equipment Ruggedness Spectra
Generic Implementation Procedure
Ground Motion Incoherence
High Confidence of Low Probability of Failure.
High Frequency Ductility
Individual Plant Examination of External Events
In-Structure Response Spectra
Large Early Release Frequency
Lawrence Livermore National Laboratories
Non-Exceedance Probability
Nuclear Power Plant
Peak Ground Acceleration
Pseudo Spectral Acceleration
Probabilistic Seismic Hazard Analysis
Pseudo Spectral Velocity
Pressurized Water Reactor
Request for Additional Information
Reference Earthquake Spectrum from Probabilistic Hazard Study
Review Level Earthquake
Required Response Spectrum
Spectral Acceleration
Spectral Displacement
Single Degree of Freedom
Seismic Margin Assessment
Seismic Probabilistic Risk Assessment
Seismic Probabilistic Safety Assessment
Seismic Qualification Review Team
Seismic Qualification Reporting and Testing Standardization
Structures, Systems and Components
ix

EPRI Proprietary Licensed Material

SSE
SSI
SSMRP
Standard
TER
TRS
UHS
ZPA
ZPGA

Safe Shutdown Earthquake


Soil-Structure Interaction
Seismic Safety Margin Research Program
ANS 58.21 External Events PRA Methodology Standard
Technical Evaluation Report
Test Response Spectrum
Uniform Hazard Spectra
Zero Period Acceleration
Zero Period Ground Acceleration

EPRI Proprietary Licensed Material

CONTENTS

1 INTRODUCTION ....................................................................................................................1-1
1.1

Objective of Applications Guide ...................................................................................1-1

1.2

Scope of the Applications Guide..................................................................................1-2

2 STATE OF THE ART AND PRACTICE OF SEISMIC PRA IN THE U.S. AND OTHER
COUNTRIES..............................................................................................................................2-1
2.1

Methodology ................................................................................................................2-2

2.1.1 Key Elements of Seismic PRA ................................................................................2-3


2.1.2 Output of Seismic PRA ...........................................................................................2-6
2.1.3 Discussion of Seismic PRA Tasks ..........................................................................2-6
2.1.4 Acceptable Seismic PRA Methodology...................................................................2-9
2.1.4.1 SSMRP Method ............................................................................................2-10
2.1.4.2 Zion Method ..................................................................................................2-10
2.2

Seismic Fragility Analysis Methodology.....................................................................2-10

2.2.1 Generalized Fragility Descriptions ........................................................................2-11


2.2.2 Detailed Fragility Model.........................................................................................2-13
2.2.2 Failure Modes .......................................................................................................2-16
2.2.3 Estimation of Fragility Parameters ........................................................................2-17
2.2.3.1 Fragility of Structures ....................................................................................2-18
2.2.3.2 Fragility of Equipment and Other Components.............................................2-20
2.2.4 Information Sources ..............................................................................................2-23
2.2.5 Other Fragility Models ...........................................................................................2-24
2.2.6 Hybrid Method.......................................................................................................2-25
2.3

Plant Level Fragility ...................................................................................................2-26

3 FRAGILITY METHODOLOGY ISSUES AND ENHANCEMENTS .........................................3-1


3.1

Methodological Issues from USNRC Review of IPEEE Submittals .............................3-1

3.1.1 Use of Uniform Hazard Spectrum ...........................................................................3-2

xi

EPRI Proprietary Licensed Material

3.1.2 Use of Surrogate Elements in SPRAs.....................................................................3-3


3.1.3 The Use of New Soil Structure Interaction Analysis Versus the Use of
Scaling..............................................................................................................................3-5
3.1.4 Reliance on Structures for Which the Original Design Documentation is no
Longer Available...............................................................................................................3-6
3.1.5 Importance of Analysts Expertise in Component Fragility/HCLPF
Assessments ....................................................................................................................3-7
3.2

Comments and Suggestions from Industry on Methodology for SPRA .......................3-8

3.3

New Test and Earthquake Experience Data..............................................................3-13

4 DEVELOPMENT OF FRAGILITIES IN ACCORDANCE WITH ANS 58.21 ...........................4-1


4.1

Understanding the Seismic Hazard .............................................................................4-2

4.2 Understanding the Development of the Risk Model and Equipment List and the
Significance of Screening Thresholds ...................................................................................4-6
4.3

Determine the Seismic Response of Structures ..........................................................4-8

4.3.1 Scaling of Existing Design Analysis ........................................................................4-9


4.3.2 Conducting New Analysis .....................................................................................4-10
4.4

Plant Walkdown .........................................................................................................4-11

4.5

Structural Capacity ....................................................................................................4-12

4.6

Determine Ductility Beyond the Limit State Capacity.................................................4-13

4.7

Structural Response Factor .......................................................................................4-14

4.7.1 Spectral Shape Factor ..........................................................................................4-15


4.7.2 Damping................................................................................................................4-15
4.7.3 Modeling................................................................................................................4-16
4.7.4 Mode Combination ................................................................................................4-16
4.7.5 Earthquake Component Combination ...................................................................4-17
4.7.6 Foundation-Structure Interaction...........................................................................4-17
4.7.7 High Frequency Effect...........................................................................................4-18
4.8

Probabilistic Response: .............................................................................................4-18

4.9

Equipment Response and Capacity...........................................................................4-20

4.9.1 Initial Prescreening Using Licensing Criteria.........................................................4-21


4.9.2 Prescreening Using Earthquake Experience Data................................................4-22
4.9.3 Prescreening using the EPRI SMA Screening Tables ..........................................4-23
4.9.4 Development of Fragilities Using Plant Specific Data ...........................................4-24
5 INDEX TO EXAMPLE FRAGILITY CALCULATIONS ...........................................................5-1

xii

EPRI Proprietary Licensed Material

6 REFERENCES .......................................................................................................................6-1
A BENCHMARK STUDIES TO VERIFY AN APPROXIMATE METHOD FOR SPECTRA
SCALING.................................................................................................................................. A-1
A.1. Background.................................................................................................................. A-1
A.2. Verification of Original Spectra to be Scaled ............................................................... A-1
A.3. Development of Scaled Spectra by Rigorous and by Simplified Means ...................... A-2
A.4. References................................................................................................................... A-4
B DEVELOPMENT OF IN-STRUCTURE RESPONSE SPECTRA FOR SEISMIC
MARGIN OR SEISMIC PRA EVALUATION BY SCALING ..................................................... B-1
B.1

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

B.2

Incoherence Reduced Ground Motion ........................................................................ B-1

B.3 Estimation of Floor Spectra Compatible with Incoherence Reduced Ground


Motion .................................................................................................................................. B-6
B.3.1 Scaling of Floor Spectra......................................................................................... B-6
B.3.2 Spectral Estimation Method ................................................................................... B-6
B.3.3 Incoherence Reduction of Selected Reference Locations ................................... B-10
B.3.4 Total Spectra Incoherence Reduction for All Locations ....................................... B-12
B.4

High Frequency Reduction of Floor Spectra Due to Ductility Effects........................ B-13

B.5 Estimation of Floor Spectra Compatible with High Frequency Ductility Reduced
Pseudo Ground Motion ...................................................................................................... B-16
B.5.1 Damage Consistent Scaling of Floor Spectra ...................................................... B-16
B.5.2 Damage Consistent Reduction of Selected Reference Locations ....................... B-16
B.6

References ............................................................................................................... B-19

C ESTIMATION OF EQUIPMENT CAPACITY BASED ON EARTHQUAKE


EXPERIENCE DATA................................................................................................................ C-1
References ........................................................................................................................... C-5
D EXAMPLE FRAGILITY FOR INSTRUMENT CABINET DERIVED FROM
EXPERIENCE DATA................................................................................................................ D-1
D.1

Demand ...................................................................................................................... D-1

D.2

Capacity ...................................................................................................................... D-2

D.3

Capacity Factor........................................................................................................... D-4

D.4

Structural Response Factor ........................................................................................ D-4

D.4.1 Spectral Shape (SS) .............................................................................................. D-5


D.4.2 Damping (D)........................................................................................................... D-6

xiii

EPRI Proprietary Licensed Material

D.4.3 Modeling (M) .......................................................................................................... D-6


D.4.4 Mode Combination (MC) ........................................................................................ D-6
D.4.5 Ground Motion Incoherence (GMI) ........................................................................ D-7
D.4.6 High Frequency Ductility Reduction (HFD) ........................................................... D-7
D.4.7 Scaling Using Random Vibration Theory (RV)....................................................... D-7
D.4.8 Structural Response Factor (FRS) ........................................................................... D-7
D.5

Fragility ....................................................................................................................... D-8

D.6

References ................................................................................................................. D-9

E DEVELOPMENT OF GENERIC FRAGILITY DESCRIPTIONS FOR PURPOSES OF


SCREENING BASED UPON DESIGN CRITERIA ................................................................... E-1
E.1

Establishment of Screening Level............................................................................... E-1

E.1.1 Seismic Hazard ...................................................................................................... E-1


E.1.2 Uncertainty in the Median Fragility......................................................................... E-2
E.1.3 Target Failure Rate ................................................................................................ E-2
E.2

Development of Demand on Components.................................................................. E-3

E.3

Screening Evaluation of Equipment and Distributive Systems ................................... E-3

E.4 Screening of Flexible Equipment and Distributive Systems Designed by


Analysis ................................................................................................................................ E-5
E.4.1 Strength Factor ...................................................................................................... E-6
E.4.2 Equipment Response Factor.................................................................................. E-6
E.4.3 Structural Response Factor ................................................................................... E-8
E.4.4 Fragility Description for Flexible Components Designed by Analysis .................... E-8
E.5

Components Qualified by Test.................................................................................... E-9

E.6

References ............................................................................................................... E-12

F EXAMPLE PROBLEM FOR SERVICE WATER PUMP ........................................................F-1


F.1

Description of Equipment.............................................................................................F-1

F.2

Strength Factor ............................................................................................................F-3

F.3

Equipment Response Factor .......................................................................................F-4

F.3.1 Qualification Method ...............................................................................................F-4


F.3.2 Damping..................................................................................................................F-5
F.3.3 Modeling..................................................................................................................F-6
F.3.4 Mode Combination ..................................................................................................F-6
F.3.5 Earthquake Component Combination .....................................................................F-6
F.3.6 Equipment Response Factor...................................................................................F-7

xiv

EPRI Proprietary Licensed Material

F.4

Structural Response Factor .........................................................................................F-7

F.4.1 Spectral Shape........................................................................................................F-7


F.4.2 Damping..................................................................................................................F-8
F.4.3 Modeling..................................................................................................................F-8
F.4.4 Mode Combination ..................................................................................................F-9
F.4.5 Ground Motion Incoherence....................................................................................F-9
F.4.6 Structural Response Factor ....................................................................................F-9
F.5

Fragility for Service Water Pumps ...............................................................................F-9

F.6

References ................................................................................................................F-10

G GENERAL METHODOLOGY FOR LIQUEFACTION SEISMIC FRAGILITY


ASSESSMENT AND EXAMPLE ANALYSIS ........................................................................... G-1
G.1

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

G.2

Background................................................................................................................. G-1

G.3

Basis of Approach....................................................................................................... G-2

G.4

Example Case Study .................................................................................................. G-3

G.4.1 Overview of Approach............................................................................................ G-3


G.4.2 Initial Liquefaction Analysis and Results ................................................................ G-4
G.4.3 Detailed Liquefaction and Settlement Analysis and Results .................................. G-5
G.5

References ................................................................................................................. G-9

xv

EPRI Proprietary Licensed Material

LIST OF FIGURES
Figure 2-1 Seismic Risk Assessment Methodology ...................................................................2-4
Figure 2-2 Seismic PRA Flowchart ............................................................................................2-7
Figure 2-3 Fragility Curves.......................................................................................................2-12
Figure 2-4 Mean, Median, 5% Non-Exceedance, and 95% Non-Exceedance Fragility
Curves for a Component ..................................................................................................2-15
Figure 2-5 Discrete Family of Fragility Curves for a Component .............................................2-16
Figure 4-1 Annual Probability of Exceedance of Peak Ground Acceleration .............................4-3
Figure 4-2 Uniform Hazard Spectra for the 10-4 Annual Probability of Exceedance.
th
th
th
Spectra shown for three percentiles: 15 , 50 , and 85 ....................................................4-4
Figure A-1 Lumped Mass Model of Reactor Building .............................................................. A-5
Figure A-2 Reactor Building EQ Floor Spectra, Node 11 ........................................................ A-6
Figure A-3 Reactor Building NS Floor Spectra, Node 11......................................................... A-7
Figure A-4 Reactor Building Vertical Floor Spectra, Node 11.................................................. A-8
Figure A-5 RG 1.60 Spectrum Compatible Time Histories ...................................................... A-9
Figure A-6 Reactor Building E-W Floor Spectra Reconstructed Model, Node 11.................. A-10
Figure A-7 Reactor Building N-S Floor Spectra Reconstructed Model, Node 11................... A-11
Figure A-8 Reactor Building Vertical Floor Spectra Reconstructed Model, Node 11............. A-12
Figure A-9 EW Floor Response Spectrum Developed From Eigensolution of DBE
Analysis, Using RG 1.60 Time Histories ......................................................................... A-13
Figure A-10 Comparison of DBE with UHS ........................................................................... A-14
Figure A-11 RB Estimated SDOF Oscillator Response Node 11 .................................... A-15
Figure A-12 RB Estimated SDOF Oscillator Response Node 11 .................................... A-16
Figure A-13 RB UHS Scale Factors Node 11 .................................................................. A-17
Figure A-14 Scaled DBE Spectra Node 11......................................................................... A-18
Figure B-1 Reduction Function for Incoherence Across a 43.3 M (142-Foot) Foundation....... B-3
Figure B-2 Uniform Hazard Horizontal Response Spectra ...................................................... B-4
Figure B-3 Incoherence Reduced Horizontal Ground Motion for Building E............................ B-5
Figure B-4 Incoherence Reduced Vertical Ground Motion for Building E ................................ B-5
Figure B-5 Response Spectra Relationships ........................................................................... B-7
Figure B-6 Incoherence Reduced Spectra for Building E Node 162610................................ B-11
Figure B-7 Incoherence Reduction Functions for Selected Nodes of Building E ................... B-12
Figure B-8 Overall Incoherence Reduction Factors for Building E ........................................ B-13

xvii

EPRI Proprietary Licensed Material

Figure B-9 Reduced Horizontal Ground Motion Spectra for Building E ................................. B-15
Figure B-10 Reduced Vertical Ground Motion Spectra for Building E ................................... B-15
Figure B-11 Overall Reduced Spectra for Building E Node 162610 ...................................... B-18
Figure D-1 Equipment Class 20 Control and Instrumentation Panels and Cabinets ............... D-3
Figure E-1 Comparison of DBE Vs Probabilistic Response Spectra Reactor Building El.
547 ................................................................................................................................... E-5
Figure E-2 Typical Overtest at High Frequency..................................................................... E-12
Figure F-1 Model of the Service Water Pump...........................................................................F-2
Figure F-2 Simplified Motor Stand Model .................................................................................F-3
Figure F-3 Demand Response Spectrum, 5% Damping...........................................................F-5
Figure G-1 Weighted Fragility Curves, Accounting for Random Variability and
Composite Variability, for End-States of: (i) Incipient Liquefaction, and (ii) Gross
Liquefaction....................................................................................................................... G-7
Figure G-2 Weighted Fragility Curves, Accounting for Random Variability and
Composite Variability, for End-States of Component Failure Due to Settlements
Caused by Level-Ground Liquefaction, for Cases Where the Component Median
Capacity Against Failure Equals (iii) 5 cm, (iv) 10 cm and (v) 20 cm ................................ G-8

xviii

EPRI Proprietary Licensed Material

LIST OF TABLES
Table 4-1 Correlation Of Fragility Development Elements And Requirements Of Ans
Standard For External Events ..........................................................................................4-25
Table 5-1 Index To Example Fragility Calculations....................................................................5-2
Table B-1 Reduction Factors for 150-Foot Foundation............................................................ B-2

xix

EPRI Proprietary Licensed Material

1
INTRODUCTION

1.1

Objective of Applications Guide

Seismic Probabilistic Risk Assessment (SPRA) studies have been conducted for many of the US
Nuclear Power Plants over the last 20 years. Initially they were conducted to answer safety
concerns in heavily populated areas. The most recent wide spread application of SPRA was to
satisfy the USNRC request for information regarding severe accident vulnerabilities in Generic
Letter 88-20, Supplement 4, (USNRC, 1991a). The USNRC is encouraging the use of PRA for
making risk informed decisions and has developed a Risk-Informed Regulation Implementation
Plan (USNRC, 2000b) and associated regulatory guides. The Licensees in turn are moving
toward using PRA for Changes to Licensing Basis, Changes to Technical Specifications, Graded
Quality Assurance, etc.
Seismic issues continue to arise in operating NPPs to address the risk from installations that were
not designed and constructed in accordance with current standards or in looking at potential
safety issues associated with life extension. There is a desire and a need to utilize seismic PRAs
to address these issues on a risk informed basis rather than applying the conventional
deterministic licensing basis approach to all seismic issues.
A recent Draft ANS Standard 58.21 (ANS, 2002), External Events PRA Methodology
Standard, hereafter referred to as the Standard, sets multi-level requirements for conducting
SPRAs. The Standard sets requirements for three levels of PRA, Capability Category 1, 2 and 3.
Most of the initial SPRAs conducted in the US in the 1980s, contained an uncertainty analysis
that examined the uncertainty spread in the computed Core Damage Frequency (CDF) arising
from the uncertainty in the seismic hazard and the uncertainty in the fragilities of structure,
systems and components (SSCs). These studies corresponded to the fundamental requirements
of Capability Category 2 in the Standard, although, the hazard studies at that time would not
meet current requirements. In IPEEE, the Licensees who chose to do SPRA were only required
to compute a point estimate of CDF. This would correspond to Capability Category 1, mainly
because uncertainty analyses were not conducted. Capability Category 3 is more along the lines
of what was done in the USNRC sponsored Seismic Safety Margins Research Program (USNRC,
1981). Capability Category 3 requires extensive effort to compute probabilistic response of
structures and severely limits screening unless screened out SSCs can be shown to have very low
seismic failure rate and are uncorrelated. For purposes of risk informed regulation it is intended
in this Applications Guide that a Capability Category 2 SPRA will generally be the approach
taken by licensees although, depending upon the issue, an application meeting Capability
Category 1 may be adequate. This decision is up to the Licensees to demonstrate that full
sensitivity analyses are not required to demonstrate the merits of a risk informed decision and of
course requires agreement by the regulators, who must develop Safety Evaluation Reports
(SERs) on the issue.
1-1

EPRI Proprietary Licensed Material


Introduction

1.2

Scope of the Applications Guide

Significant information in the literature exists on how to generate seismic fragilities and how to
conduct a SPRA. The EPRI Methodology for Developing Seismic Fragilities, (EPRI, 1994)
and EPRI Methodology for Assessment of Nuclear Power Plant Seismic Margin, (EPRI,
1991b), contain most of the background and guidance needed for an analyst to develop seismic
fragilities of SSCs. This Applications Guide will not repeat the guidance in those documents.
Instead it will focus on the applicability of the methodology to the requirements in the Standard
(including how to use the methods with respect to Capability Category 1, 2 and 3 of the
Standard) and provide additions and enhancements to the existing methodology where
applicable. EPRI (1994) contains methodologies for rigorous and simplified methods for
developing seismic fragilities, but since the Standard was written subsequent to EPRI (1994),
some direction to the analyst is appropriate on which of the methods in EPRI (1994) are
applicable to the three capability categories of the Standard.
Chapter 2 presents an overall summary of the SPRA methodology and fragility methodology.
This Applications Guide focuses on the detailed development of fragilities, which is one of the
important technical steps in conducting a SPRA. It is important, however, for the fragility
analyst to understand the background of the seismic hazard development and the systems
modeling in order to assure a clear interface with the hazard and systems analysts. Chapter 3
discusses the USNRC comments on seismic IPEEE submittals and provides guidance on how to
address these comments in accordance with the framework of the Standard. Chapter 3 also
addresses some recent industry-suggested alternate approaches to current practice in SPRAs, and
their compliance with the Standard.
Chapter 4 goes through a step by step description of the important elements of developing
seismic fragilities and the correlation of each step to requirements in the Standard. Detailed
fragility development procedures in EPRI (1994) are not repeated. Rather, the focus is on
guiding the fragility analyst through the process of interfacing with the hazard and systems
analysts and developing fragilities that are compatible with the overall SPRA process. The
appendices contain a complementary set of sample problems which focus on different areas of
fragility analysis than those published in EPRI (1994) in order to enhance the database of
examples available to the user.

1-2

EPRI Proprietary Licensed Material

2
STATE OF THE ART AND PRACTICE OF SEISMIC PRA
IN THE U.S. AND OTHER COUNTRIES

The use of seismic probabilistic risk assessment (SPRA) methods to supplement the
deterministic processes of licensing and design of nuclear power plant facilities started in the mid
1970s. Prior to this time, deterministic procedures were primarily used. In 1975 the U.S.
Nuclear Regulatory Commission (USNRC) published WASH-1400, a reactor safety study of
U.S. commercial nuclear power plants that employed probabilistic risk assessment procedures to
assess accident risks (USNRC, 1975). In that study the annual frequency of seismically-induced
-7
core damage for an average site was reported to be 5x10 . It was concluded that seismic events
were not major contributions to risk. This study considered seismic events in only a rudimentary
manner.
An SPRA was conducted in the late 1970s for the Oyster Creek Unit 1 Nuclear Generating
Station. This study became the foundation for SPRA as currently practiced and characterized
SPRA in terms of the integration of a site hazard curve with a plant level fragility curve to
compute core damage frequency. The plant level fragility curve was formulated from individual
structures, systems and components (SSCs) fragilities using fault tree/event tree logic models of
the plant systems. A lognormal fragility model was used to define the fragilities. Lognormal
models are still used in SPRAs conducted for nuclear plants today. A detailed fragility model
was developed that addressed the randomness and uncertainty in the various underlying response
and capacity variables that contribute to the success or failure of SSCs. In 1981 the Zion SPRA
was submitted to the NRC (Zion, 1981). This was the first complete SPRA study of a
commercial NPP. The first technical paper published that described in some detail what is
referred to as the Zion method Kennedy, et.al., 1980. The method was patterned after the
Oyster Creek and Zion SPRAs. About the same time, the NRC sponsored the Seismic Safety
Margin Research Program (SSMRP) at the Lawrence Livermore National Laboratory (LLNL)
(USNRC, 1981). The SSMRP method for performing SPRA involved detailed response analyses
using the Latin hypercube simulation procedure. The Latin Hypercube procedure ensures that
the full range of uncertainties of important variables are utilized but requires considerably fewer
simulations than the classic Monte Carlo simulation procedure. Monte Carlo analysis usually
requires thousands of simulations to assure that the full range of uncertainties of variables are
incorporated. In the SSMRP method, fragilities were referenced to local accelerations rather
than acceleration at the ground level. The SSMRP approach was resource intensive and is
generally not used today. However, simplifications of the SSMRP approach have been used in
subsequent research studies; for instance, NUREG-1150 (USRNC, 1990).
Several other studies were conducted in the early 1980s using the Zion Method methodology
(Indian Point, Limerick, Susquehanna, Seabrook, Milestone 3, Oconee, Browns Ferry).

2-1

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

In August 1985 a Severe Accident Policy statement was issued by the USNRC commissioners
(USNRC, 1985). It required limited scope PRA evaluations of all commercial nuclear power
plants in the United States for severe accident events. The USNRC staff was also given the
responsibility for establishing the methodology and development of an alternate seismic
margin approach to SPRA which used fragility and SPRA concepts in conjunction with some
simplifying deterministic screening evaluation procedures. Trial guidelines for performing
seismic margin reviews of nuclear power plants were developed and recommended to the
USNRC (Prassinos et al., 1986). A trial review using these guidelines was performed for the
Maine Yankee Atomic Power Station (Prassinos et al., 1987; Moore et al., 1987; Ravindra et al.,
1987). As an alternative to the USNRC Seismic Margin Approach, EPRI developed a
deterministic Seismic Margin Assessment methodology (EPRI, 1988). A trial plant applications
was conducted for the Catawba PWR (EPRI, 1989b). A later trial plant application was
conducted for the Hatch BWR (EPRI, 1991d).
In 1988, Pacific Gas and Electric Co. (PG&E) submitted the results of the detailed SPRA
conducted for the Diablo Canyon Nuclear Power Plant to the NRC (PG&E, 1988). This was part
of the PG&E Long Term Seismic Program that was a licensing condition required for plant
operation. This was the most detailed SPRA performed to date. Several studies conducted
during this program confirmed the validity of the methodology originally developed for the
Oyster Creek and Zion studies and enhanced this methodology in some areas to reduce
uncertainty.
In 1988 the USNRC issued Generic Letter 88-20 (USNRC, 1988) to nuclear power plant utilities
and operators, requesting that an individual plant examination (IPE) for internally initiated
events be performed. This letter was written as part of the Severe Accident Policy. In 1991 the
USNRC issued Supplement 4 to Generic Letter 88-20 (USNRC, 1991a) requesting an Individual
Plant Examination of External Events (IPEEE) for plant-specific external-event-initiated severe
accident vulnerabilities. The USNRC also issued a procedural and submittal guidance document
(USNRC, 1991b) for IPEEE programs. Probabilistic risk assessment procedures, seismic margin
methodology, deterministic screening methods, and success path processes were recommended
as the preferred method to resolve significant external events, primarily earthquakes.
Since 1980, seismic PRAs or seismic PSAs have been conducted for over 50 nuclear power
plants worldwide. The methodology has been well established and the necessary data on the
parameters of the PRA model have been generally collected. Detailed descriptions of the
procedures used in SPRA are given in several published reports - PRA Procedures Guide
(USNRC, 1983), PSA Procedures Guide (USRNC, 1985), (EPRI, 1994) and Budnitz, (1998).

2.1

Methodology

Seismic PRA is different from an internal event PRA in several important ways:

Earthquakes could cause initiating events different from those considered in the internal
event PRA.

All possible levels of earthquakes along with their frequencies of occurrence and
consequential damage to plant systems and components should be considered.

2-2

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

Earthquakes could simultaneously damage multiple redundant components. This major


common cause effect should be properly accounted for in the risk quantification.

The objectives of a seismic SPRA are to:

Understand the most likely accident sequences induced by earthquakes (useful for accident
management),

Develop an appreciation of accident behavior (i.e., consequences and role of operator),

Gain an understanding of the overall likelihood of core damage induced by earthquakes,

Identify the dominant seismic risk contributors,

Identify the range of peak ground acceleration that contributes significantly to the plant risk
(this is helpful in making judgements on seismic margins), and

Compare seismic risk with risks from other events and establish priorities for plant backfit.

2.1.1 Key Elements of Seismic PRA


The key elements of a SPRA can be identified as:

Seismic Hazard Analysis: to develop frequencies of occurrence of different levels of


earthquake ground motion (e.g., peak ground acceleration) at the site.

Seismic Fragility Evaluation: to estimate the conditional probability of failure of important


structures and equipment whose failure may lead to unacceptable damage to the plant (e.g.,
core damage). Plant walkdown is an important activity in conducting this task.

Systems/Accident Sequence Analysis: to model the combinations of structural and


equipment failures that could initiate and propagate a seismic core damage sequence.

Risk Quantification: to Assemble the results of the seismic hazard, seismic fragility, and
systems analyses to estimate the frequencies of core damage and plant damage states.
Assessment of the impact of seismic events on the containment and consequence analyses,
and integration of these results with the core damage analysis to obtain estimates of seismic
risk in terms of effects on public health (e.g., early deaths and latent cancer fatalities).

Figure 2-1 shows the key elements of seismic risk assessment methodology along with the
typical databases and software.

2-3

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

1A

Region Specific Seismicit y

EQHAZARD
SEISRISK III
HAZARD

Seismic Motion
Parameter

Plant Specific Unavailabilit y

2B

SOFTWARE
Event Trees
Fault Trees
Containment Analysis
Systems Analysis

Software

SMACS
SAP
SHAKE
SASSI
ANSYS
ETC.

Conditional Probability
of Failure

3B

Figure 2-1
Seismic Risk Assessment Methodology

2-4

Atmospheric
dispersion

Population
Evacuation

Health effects

Release Frequency

Databases

Component-Fragility
Evaluation

Weather data

Release
Category

Frequency

Seismic Motion
Parameter

SEISIM
SRACOR
EQESRA
SEISMIC
SEIS

4
Probability Density

DATABASE

SETS
RISKMAN
NUPRA
CAFTA
RISK SPECTRUM

1B

SOFTWARE

2A

Software

Frequency of
Exceedance

Frequency of
Exceedance

4A

DATABASE

1 Seismic Hazard Analysis

Earthquake Experience
GERS
Fragility Tests
Analyses

Property damage

3A

Consequence Analysis

Damage
Risk

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

Box 1 shows the result of a seismic hazard analysis, i.e., a family of seismic hazard curves
relating the frequency of exceedance to different levels of ground motion. Box 1A describes the
databases needed to perform this seismic hazard analysis; note that region specific seismicity
data is required for this analysis. Some of the available software for performing the seismic
hazard analysis are indicated in Box 1B.
Box 2 is a pictorial representation of the systems analysis; it consists of event trees, fault trees
and containment analysis. The database (shown in Box 2A) needed to perform the system
analysis is the unavailabilities (i.e., random failures and operator failures) modified to reflect the
severe stress induced by earthquakes. The software available for performing the systems
analysis are indicated in Box 2B; they are typically used in the internal event analysis.
Box 3 shows the result of component fragility evaluation, i.e., a family of seismic fragility
curves. These are developed using plant design information and realistic response analysis.
There are many response analysis software programs available (Box 3B). The databases used for
fragility analysis include earthquake experience data, generic equipment ruggedness spectra and
fragility test results as indicated in Box 3A.
Box 4 shows the probability density functions of core damage frequency and release frequencies.
These are obtained using the sequences of component failures, fragilities of components and
seismic hazard curves. This is accomplished using a quantification software. Note that the
quantification procedure is different from the internal event analysis in that the entire spectrum
of earthquakes is considered and at each earthquake level, the component failure probabilities are
different and dependencies between different component failures are to be explicitly included in
the analysis. Some of the software developed for the seismic quantification are SEISIM,
SEISMIC, SEIS, EQESRA, SRACOR and SECOM-2 as indicated in Box 4A. Most SPRAs
conducted have been a Level 1 PRA that stops at the computation of core damage frequency
(CDF). IPEEE required that the Level 1 analysis be extended to the evaluation of contaminant
integrity but did not require a full Level 2 evaluation of Release Frequency.
Box 5 refers to the dispersion analysis using weather data that estimates the consequences of a
core damage accident resulting in a radiological release to the atmosphere. Population
distribution around the site and emergency evacuation procedures in place are considered in
assessing the consequences in terms of health effects and property damage. Usually, the
software and databases employed for internal events are adequate to estimate the consequences
of seismic induced accidents. Sometimes, the analysts assume a reduced evacuation rate for
seismic events.
Box 6 shows the risk curves. For each level of damage (e.g., number of deaths, cancer fatalities
and property damage), the risk curve gives the annual frequency of exceedance of damage. The
uncertainties in the risk assessment are displayed by means of a family of risk curves. Therefore,
the annual frequency of exceeding a given level of damage is distributed and one could state this
frequency with different levels of confidence.

2-5

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

2.1.2 Output of Seismic PRA


The output from a SPRA consists of:

Seismic fragilities of components and seismic margins.

Seismic fragilities of accident sequences and seismic margins.

Seismic accident sequence frequencies and uncertainty distributions.

Impact of nonseismic unavailabilities on seismic risk.

Identification of dominant risk contributors: components, systems, sequences and procedures.

Distribution on range of accelerations contributing to seismic risk.

Risk reduction as a function of seismic upgrading to aid in backfit decisions.

2.1.3 Discussion of Seismic PRA Tasks


Figure 2-2 shows a flow chart of the SPRA. In the following we describe the different tasks
1. Review Plant Safety Systems and Modify Internal Event PRA Event and Fault Trees: The
systems analyst will review the plant safety systems from the viewpoint of seismic safety,
identify any seismic-specific initiating events and modify event and fault trees in the internal
event PRA. Redundancy of multitrain safety systems is usually not credited due to
correlations of response and capacity of similar or identical components.
2. Develop PRA Components List: Based on Task 1 and past seismic PRAs of similar plants,
the systems analyst and fragility analyst develop a preliminary components list. The list
includes the equipment and systems required to provide protection for all seismically induced
initiating events, including those needed to address seismic induced fires and floods and to
prevent early containment failure in an earthquake. Non safety systems are also included to
take credit for non failures of normal shutdown systems.
3. Conduct Soil Failures Evaluation: The potential for soil liquefaction, slope failures and
damage to buried pipelines is assessed in this task. Procedures for assessing these effects are
described in EPRI (1991b). For most plants, a review based on design and construction
records is considered adequate to screen these types of failures out. A detailed analysis is
needed only if soil failure is deemed significant. This task is usually carried out by specialist
geotechnical engineers.
4. Perform Structural Response Analysis: This task involves the derivation of the best estimate
(or median-centered) seismic responses and their variability in the form of structural loads or
floor response spectra that define the demand for which structures, systems and components
are evaluated. These best estimates and variabilities are obtained by simulation probabilistic
response analysis, by new deterministic analysis with estimated variability or by scaling of
the safe shutdown earthquake (SSE) responses and assigning variability. The ground
response spectrum usually used as input for this analysis is the median spectral shape for a
10,000-year return period along with variability estimates.
2-6

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries
Review Plant Safety
Systems and Modify
Internal Event Analysis
Event and Fault Trees

Perform Response
Analysis

Develop PRA
Components List
Including
Containment Systems

Soil
Failures
Evaluation

Develop Floor Spectra


and Structural
Response
Select Peer Review
Team

Relay Chatter
Evaluation

Perform Plant
Walkdown Using
EPRI Margin Procedures

Screen Out
Components From
PRA List

Develop Seismic
Fragilities of
Screened-In Components

Modify Fault Trees,


Develop Sequence
Equations

Seismic Risk
Quantification

Seismic
Hazard
Curves

Develop Seismic
PRA Outputs

Prepare PRA
Report

Peer Review

Input to Utility
Management

Seismic PRA
Completed

Figure 2-2
Seismic PRA Flowchart

2-7

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

5. Perform Peer Review: In order to assure the technical quality of the seismic PRA and to
provide validation of the judgments made by the analysts, a peer review of the entire seismic
PRA is often performed with the peer review team participating in the project at various
critical times (e.g., review of response analysis, systems modeling, walkdown, fragility
analysis and final documentation). For example, the requirements of a peer review are
described in NUREG-1407 Chapter 7 (USNRC, 1991b). At this stage of the PRA, the task is
to set up the peer review team and identify the scope and schedule for its participation.
6. Perform Plant Walkdown: The plant walkdown of essential components is particularly
emphasized in modern seismic PRAs. In order for the walkdown to be efficiently performed,
review of the design basis, preparation of procedures, collection of design/qualification data
and training of the walkdown team is essential. All items on the components list must be
physically examined for seismic vulnerabilities if possible using the procedures given in
EPRI (1991b). The emphasis is on compliance to screening caveats, anchorage and
attachment of subassemblies and parts, and seismic spatial systems interactions. The
walkdown is conducted by a team of systems engineers and seismic fragility analysts.
7. Perform Screening of Components: Certain high capacity components may be screened out
of the components list based on the review of seismic qualification criteria and qualification
documents and walkdown screening. The decision to screen components should be based on
the seismic hazard and the associated unconditional failure rate of a component with a
fragility corresponding to the screening level. Deterministic screening targets are typically
set based upon the lower tail of the component fragility. The reference point for screening is
an acceleration level where there is 95% confidence of less than 5% probability of failure,
commonly referred to as a HCLPF (high-confidence-of-low-probability-of-failure). For
example, some PRA analysts screened out components with HCLPF capacities larger than
0.3 g peak ground acceleration (pga) in the IPEEE program. Based on previous seismic
PRAs, the CDF contribution of components screened out at 0.3g pga HCLPF capacity was
judged to be very low. However, as discussed in Chapter 3 the screening level was often too
low and masked the CDF results. Screening for more seismically active regions (e.g.,
western US and higher seismic regions in the central and eastern US) should only be done at
a higher earthquake level. Screening is primarily done by seismic fragility analysts using
earthquake experience and plant specific qualifications criteria.
8. Perform Relay Chatter Evaluation: Relays whose chatter during an earthquake could result
in adverse effects on plant safety must be identified and evaluated. This evaluation may be
done probabilistically or by deterministic methods. The identification of relays and
evaluation of the consequence of chatter on the electrical circuits are done by the systems
analysts and electrical engineers; the seismic ruggedness of relays including the amplification
of response through the cabinet into the relays is evaluated by the seismic fragility analysts.
Often, rather than modeling the response of the systems to relay chatter, a deterministic
screening is conducted to identify relays with high and low capacity and to determine if relay
chatter is detrimental. Low ruggedness relays that can cause adverse effects are usually
replaced. Some relays with intermediate capacities may be modeled. In this case specific or
generic data on relay capacity is used to develop fragilities.
9. Develop Seismic Fragilities: Estimation of the conditional probabilities of structural or
equipment failure for a given level of seismic ground motion for the screened-in components.
2-8

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

Curves are developed using the fragility model whose parameters are the median acceleration
capacity (Am), and logarithmic standard deviations reflecting randomness in capacity ( R )
and uncertainty in the median capacity ( U ) . This task is performed by the seismic fragility
analysts.
10. Develop Accident Sequence Equations: Perform the event tree and fault tree analyses for the
seismic initiating events to obtain accident sequence Boolean equations or cutsets. This task
is performed primarily by systems analysts with assistance from fragility analysts.
11. Input Seismic Hazard Curves: The seismic hazard curves developed for the site are input
into the seismic quantification code.
12. Conduct Seismic Risk Quantification: This task involves assembling the results of the
seismic hazard, fragility and systems analyses to estimate the frequencies of core damage and
plant damage states. For some applications (e.g., Individual Plant Examination of External
Events (IPEEE)), it was sufficient to obtain a mean point estimate of the core damage
frequency using a single mean hazard curve and a single mean fragility curve; however,
NUREG-1407 (USNRC, 1991b) encourages the analyst to make a careful study of the
uncertainties. The approach followed in recent seismic PRAs is to identify the dominant
sequences by point estimation and to perform uncertainty analysis of only these dominant
sequences. The risk quantification is done by considering both seismic failures and nonseismic unavailabilities and operator actions.
13. Develop Seismic PRA Outputs: Since the focus of the seismic PRA is not on bottom line
numbers but on the insights of the examination, a number of intermediate outputs are
required as described in Section 2.1.2 above. This task is shared between the systems
analysts and fragility analysts.
14. Prepare PRA Report: This task involves documenting the methodology and results of the
study. Specific reporting requirements are given for example in NUREG-1407
(USNRC, 1991b).
15. Perform Final Peer Review: After the draft report is prepared, a peer review of the
procedures, numerical results and insights obtained from the PRA is conducted. This is a
culmination of the review process that has been implemented throughout the above tasks.
The peer review is expected to produce a short report endorsing the PRA study.
16. Provide Input to Utility Management: This task involves developing a summary of the
Seismic PRA, (a Tier I report) the findings, and risk informed upgrading recommendations.
The objective of the task is to ensure that all responsible persons within the utility are
informed of the Seismic PRA results. Each utility may have its own procedures for this task.
2.1.4 Acceptable Seismic PRA Methodology
The methodology described above has been accepted by the USNRC for the IPEEE Program and
has been the most commonly used method for SPRA of nuclear power plants around the world.
This is also known as the Zion Method wherein the seismic fragilities of components are
2-9

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

referenced to the ground acceleration (either peak or spectral acceleration). In the early stages of
development of SPRA methodology, there was a major research program at the Lawrence
Livermore National Laboratory funded by the USNRC called Seismic Safety Margins Research
Program (SSMRP) (USNRC, 1981). It developed a theoretical approach to estimating the
seismic risk of nuclear power plants. The major differences between the SSMRP method and the
Zion method are outlined in the next subsection.
2.1.4.1 SSMRP Method

The structural and component fragilities are expressed in terms of local response parameters
such as stress, moment and spectral acceleration. Therefore, given an earthquake, the
conditional failure probability of a structure or component is obtained by convolution of the
probability distribution of the local response for the given ground acceleration and the
probability distribution of the seismic resistance (capacity) of the structure or equipment.

A major emphasis of the SSMRP method lies in the computation of structural and equipment
responses using a Latin Hypercube simulation technique. The joint probability distribution
of the responses of different components (i.e., elements in the building, equipment and
piping) characterized by mean values and a covariance matrix is developed.

The quantification of accident sequences is done cutset by cutset. Each cutset probability is
obtained by integrating the joint probability distribution of the seismic response and the
seismic capacity over the failure range. The cutset probabilities are added according to
Hunters bound (Hunter, 1976) approach to obtain the accident sequence probability.

Because of the complexity and required resources, the SSMRP method and the softwares (i.e.,
SMACS and SEISIM) (USNRC, 1991) have not been used in seismic PRAs in the last 15 years.
However, SMACS, which is a probabilistic response code, has been used to develop probabilistic
floor response spectra for some IPEEE programs. The SSMRP method corresponds to
Capability Category 3 of the Standard.
2.1.4.2 Zion Method
When the Zion method is used, some approximations are made by the analysts to account for
correlations between component failures. It is also judged that the probabilistic response
analysis, to capture the correlation and the quantification methodology to conduct multiple
integration over the joint probability distribution, is not essential for commercial applications.
Instead, some thumb rules were established to approximately account for the correlation.
The Zion method corresponds to Capability Categories 1 and 2 in the Standard.

2.2

Seismic Fragility Analysis Methodology

The seismic fragility of a structure or equipment is defined as the conditional probability of its
failure at a given value of acceleration (i.e., peak ground acceleration or peak spectral
acceleration at different frequencies). The methodology for evaluating seismic fragilities of
structures and equipment is documented in the PRA Procedures Guide (USNRC, 1983) and is
2-10

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

more specifically described for application to NPPs in the EPRI Methodology for Developing
Seismic Fragilities (EPRI, 1994). This general methodology has been applied in over 50 Seismic
Probabilistic Risk Assessments of nuclear power plants.
The fragility methodology described herein is in accordance with the Zion Method although the
capacity part of fragility development (strength of SSCs) is applicable to the SSMRP method as
well.
The objective of a fragility evaluation is to estimate the capacity of a given component relative to
a ground acceleration parameter such as peak ground acceleration or spectral acceleration.
Typically, the seismic hazard for a plant site is defined by peak ground acceleration (pga) or
spectral accelerations (Sa) at different structural frequencies; hence all fragility estimates are
referenced to ground acceleration (peak ground or spectral acceleration). Although spectral
acceleration is the preferred ground motion parameter, most existing hazard studies focused
primarily on peak ground acceleration, and most SPRAs have been based on pga. Peak ground
acceleration is used herein as an example indicator only. If the seismic hazard curves are
available in terms of spectral accelerations at different frequencies they could be used as long as
consistency in the hazard and fragility definitions is maintained. In spite of its shortcomings as a
damage measure, peak ground acceleration is a familiar term for all analysts involved in SPRA
(i.e., systems analysts, hazard analysts and fragility analysts). In the Diablo Canyon SPRA,
sensitivity studies indicated only a minor change to the core damage frequency calculated using
fragilities defined in terms of peak ground acceleration compared to those defined using average
spectral acceleration over a specified frequency range covering the fundamental frequencies of
major structures. The important conclusion is that proper interface between the analysts (i.e.,
hazard, fragility and systems) should take place and it does not matter so much what parameter
the fragility is referenced to as long as the failure mode is properly defined and the seismic
response and capacity values are consistently calculated.
2.2.1 Generalized Fragility Descriptions
The ground acceleration capacities of the components are usually estimated using information on
the plant design basis and responses calculated at the design-analysis stage. The ground
acceleration capacity is a random variable that can be described completely by its probability
distribution. However, there is uncertainty in the estimation of the parameters of this
distribution, the exact shape of this distribution, and in the appropriate failure model for the
component. For any postulated failure mode and set of parameter values describing the ground
acceleration capacity and shape of the probability distribution, a fragility curve depicting the
conditional probability of failure as a function of peak ground acceleration can be obtained.
Hence, for different models and parameter assumptions, one could obtain different fragility
curves. A satisfactory way to consider these uncertainties is to represent the component fragility
by means of a family of fragility curves. A subjective probability value is assigned to each curve
to reflect the analysts degree of belief in the model that yielded the particular fragility curve.
When represented in this fashion, the fragility curves need not appear to be smooth S-shaped
curves, approximately parallel to each other. They could theoretically intersect each other and
they may not even be non-decreasing functions of peak ground acceleration. The only
requirement is that fragility, being a probability, should be between 0 and 1 (see Figure 2-3).
Since each curve represents a different model, the fragility curves could intersect. Sometimes, a
2-11

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

fragility curve for a cutset containing a failure event and a success event is plotted. This could
show a decrease in the fragility (i.e., conditional probability of failure) at increased acceleration
values.

Figure 2-3
Fragility Curves

At any acceleration value, the component fragility (i.e., conditional probability of failure) varies
from 0 to 1; this variation is represented by a subjective probability distribution. On this
distribution we can find a fragility value (say, 0.01) that corresponds to the cumulative subjective
probability of 5%. We have 5% cumulative subjective probability (confidence) that the fragility
is less than 0.01. Similarly, we can find a fragility value for which we have a confidence of
95%. Note that these statements can be made without reference to any probability model. Using
this procedure, the median high (95%) and low (5%) confidence fragility curves can be drawn.
On the high confidence curve, we can locate the fragility value of 5%; the acceleration
corresponding to this fragility on the high confidence curve is the so-called high-confidence-oflow-probability-of-failure (HCLPF) capacity of the component. By characterizing the
component fragility through a family of fragility curves, the analyst has expressed all his
knowledge about the seismic capacity of the component along with the uncertainties. Given the
same information, two analysts with similar experience and expertise would produce
approximately the same fragility curves. Development of the family of fragility curves using
different failure models and parameters for a large number of components in a seismic PRA is
impractical if it is done as described above. Hence, a simple model for the fragility was
proposed as described in the above-cited references. In the following section this fragility model
is described.

2-12

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

2.2.2 Detailed Fragility Model


The entire family of fragility curves for an element corresponding to a particular failure mode
can be expressed in terms of the best estimate of the median ground acceleration capacity, A m ,
and two random variables. Thus, the ground acceleration capacity, A, is given by:
A = Am eR eU,

Equation 2-1

in which eR and eU are random variables with median values of 1.0, representing, respectively,
the inherent randomness about the median and the uncertainty in the median value. In this
model, we assume that both eR and eU are lognormally distributed with logarithmic standard
deviations, R and U, respectively. The formulation for fragility given by Eq. 2-1 and the
assumption of a lognormal distribution allow easy development of the family of fragility curves
that appropriately represent fragility uncertainty. For the quantification of fault trees in the plant
system and accident sequence analyses, the uncertainty in fragility needs to be expressed in a
range of conditional failure probabilities for a given ground acceleration. This is achieved as
explained below.
With perfect knowledge of the failure mode and parameters describing the ground acceleration
capacity (i.e., only accounting for the random variability, R), the conditional probability of
failure, fo , for a given peak ground acceleration level, a, is given by:

a

ln
A
fo = m
R

Equation 2-2

where [.] is the standard Gaussian cumulative distribution of the term in brackets. The
relationship between fo and a is the median fragility curve plotted in Figure 2-4 for a component
with a median ground acceleration capacity A m = 0.87g and R = 0.25. For the median
conditional probability of failure range of 5% to 95%, (- and + 1.65 log standard deviations from
the mean) the ground acceleration capacity would range from A m exp
(-1.65 R ) to A m exp (1.65 R ), i.e., 0.58g to 1.31g as shown in Figure 2-4.
When the modeling uncertainty U is included, the fragility becomes a random variable
(uncertain). At each acceleration value, the fragility f can be represented by a subjective
probability density function. The subjective probability, Q (also known as confidence) of not
exceeding a fragility f is related to f by:

2-13

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

+U 1(Q)
In
A

f = m

Equation 2-3

where:
Q

= P[f < f | a]; i.e., the subjective probability (confidence) that the conditional
probability of failure, f, is less than f for a peak ground acceleration a.

-1[.] = the inverse of the standard Gaussian cumulative distribution of the term in

brackets.
For example, the conditional probability of failure f at a peak ground acceleration of 0.6g that
has a 95% nonexceedance subjective probability (confidence) is obtained from Eq. 2-3 as 0.79 as
shown in Figure 2-4 on the 95% confidence curve. The 5% to 95% probability (confidence)
interval on the failure at 0.6g is 0 to 0.79. Computations in the seismic PRA are usually made by
discretizing the random variable probability of failure f into different intervals and deriving
probability qi for each interval (Figure 2-5). Note that the sum of qi over all the intervals is
unity. The process develops a family of fragility curves, each with an associated probability qi
(Figure 2-5).

2-14

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

Am = 0.87 g
R = 0.25
U = 0.35

0.8
95%
Confidence

0.79

Median

0.95

Mean

0.6

0.5

0.4
5%
Confidence
0.2
0.58g

0.05
0

0.2

HCLPF
0.32g

0.4

0.6

0.8 0.87

1.2

1.31 1.4

PEAK GROUND ACCELERATION (g)

Figure 2-4
Mean, Median, 5% Non-Exceedance, and 95% Non-Exceedance Fragility Curves for a
Component

A mean fragility curve is also plotted in Figure 2-4. This is obtained using Eq. 2-2 but replacing
2
2 1/2
R with the composite variability C = (R + U ) . In the IPEEE program, only a point estimate
(mean value) of CDF was required, thus single mean fragility curves and the mean seismic
hazard curve were convolved to calculate the unconditional probability of failure of SSCs.
The median ground acceleration capacity Am, and its variability estimates R and U are evaluated
by taking into account the safety margins inherent in capacity predictions, response analysis, and
equipment qualification, as explained below.

2-15

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries
1.0

0.8

q1

0.6

q2

q3

q4

q5

0.4

0.2

0
0

0.4

0.8

1.2

1.6

2.0

SSE

Peak Ground Acceleration, g


Figure 2-5
Discrete Family of Fragility Curves for a Component

2.2.2 Failure Modes


The first step in generating fragility curves such as those in Figure 2-4 is to develop a clear
definition of what constitutes failure for each of the critical elements in the plant. This definition
of failure must be agreeable to both the structural analyst generating the fragility curves and the
systems analyst who must judge the consequences of component failure. Several modes of
failure (each with a different consequence) may have to be considered and fragility curves may
have to be generated for each of these modes. For example, a motor-actuated valve may fail in
any of the following ways:

Failure of power or controls to the valve (typically related to the seismic capacity of such
items as cable trays, control panels, and emergency power). Since these failure modes are
not related to the specific item of equipment (i.e., motor actuated valve) and are common to
all active equipment, such failure modes are most easily handled as failures of separate
systems linked in a series to the equipment.

Failure of the motor.

Binding of the valve stem due to distortion and, thus, failure to operate.

Failure of the pressure boundary due to overstress of the flange joint.

2-16

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

It is usually possible to identify the failure mode most likely to be caused by the seismic event by
observations during the walkdown or by reviewing the equipment design. If there is clearly a
dominant weak link, then only the one failure mode for the weak link is considered. If two or
more failure modes have approximately equal capacity, and the failure modes are uncorrelated,
then fragility curves are developed for each failure mode based on the premise that the
component could fail in any one of the approximately equal capacity potential failure modes.
Identification of the credible modes of failure is largely based on the analysts experience and
judgment. Review of plant design criteria, calculated stress levels in relation to the allowable
limits, qualification test results, seismic fragility evaluation studies done on other plants, and
reported failures (in past earthquakes, in licensee event reports and fragility tests) are useful in
this task.
Structures are considered to have failed functionally when they cannot perform their designated
functions. In general, structures are considered to have failed functionally when inelastic
deformations under seismic load are estimated to be sufficient to potentially interfere with the
operability of safety-related equipment attached to the structure, or fractured sufficiently so that
equipment attachments fail. These failure modes represent a conservative lower bound of
seismic capacity since a larger margin of safety against total collapse exists for nuclear
structures. Also, a structural failure is generally assumed to result in a common cause failure of
multiple safety systems, if these safety systems are housed in the same structure. For example,
the service water pumps may be assumed to fail when the enclosure pump house roof collapses.
Structures that are susceptible to sliding are considered to have failed when sufficient sliding
deformation has occurred to fail buried or interconnecting piping or electrical duct banks.
For piping, failure of the support system or low cycle fatigue failure of the pressure boundary are
credible failure modes. Failure modes of equipment examined may include structural failure
modes (e.g., bending, buckling of supports, anchor bolt pullout, etc.), functional failures (binding
of valve, excessive deflection in rotating equipment), and breaker trip or relay chatter.
Consideration should also be given to the potential for soil failure modes (e.g., liquefaction, toe
bearing pressure failure, base slab uplift, and slope failures). For buried equipment
(i.e., piping and tanks), failure due to lateral soil pressures may be an important mode.
Seismically induced failures of structures or equipment under impact of another structure or
equipment may also be a consideration. Seismically induced failures of dams, if present,
resulting in either flooding or loss-of-cooling-source, should also be investigated.
2.2.3 Estimation of Fragility Parameters
In estimating fragility parameters, it is convenient to work in terms of an intermediate random
variable called the factor of safety. The factor of safety, F, on ground acceleration capacity
above a reference level earthquake specified for design; e.g., the safe shutdown earthquake level
specified for design, ASSE, is defined as follows:
A = FA SSE

where A is the actual ground motion acceleration capacity

2-17

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries
F=

Actual seismic capacity of element


Actual response due to SSE

This relationship is typically expanded to identify the conservatism or factor of safety in both the
strength and the response.
F=

Design response due to SSE


Actual capacity
X
Design response due to SSE Acctual response due to RE

F = F F
C SR

Equation 2-4

where FC is the capacity factor, FSR is the structural response factor and RE is the reference
earthquake spectrum derived from the probabilistic hazard study, anchored to the same pga as
the SSE.
Note that F can also be defined with reference to a different earthquake such as the operating
basis earthquake (OBE) level. However, the fragility analyst must be sure that the realistic
failure mode at high acceleration is the same as would be identified by comparing OBE response
to OBE allowable stresses.
The median factor of safety, Fm, can be directly related to the median ground acceleration
capacity, Am, as:
Fm =

Am
A

Equation 2-5

SSE

The logarithmic standard deviations of F, representing inherent randomness and uncertainty, are
identical to those for the ground acceleration capacity A.

2.2.3.1 Fragility of Structures


For structures, the factor of safety is typically modeled as the product of three random variables:
F = FS FFSR

Equation 2-6

The strength factor, FS , represents the ratio of ultimate strength (or strength at loss-of-function)
to the stress calculated for A SSE . In calculating the value of FS , the nonseismic portion of the
total load acting on the structure is subtracted from the strength as follows:
S - PN
F =
S P -P
T N

Equation 2-7

where S is the strength of the structural element for the specific failure mode, PN is the normal
operating load (i.e., dead load, operating temperature load, etc.) and PT is the total load on the
2-18

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

structure (i.e., sum of the seismic load for A SSE and the normal operating load). For higher
earthquake levels, other transients (e.g., SRV discharge in BWRs) may have a high probability of
occurring simultaneously with the earthquake. The definition of PN in such cases should be
extended to include the loads from these transients.
The inelastic energy absorption factor (ductility factor), F , accounts for the fact that an
earthquake represents a limited energy source and many structures or equipment items are
capable of absorbing substantial amounts of energy beyond yield without loss-of-function. A
suggested method to determine the deamplification effect resulting from inelastic energy
dissipation involves the use of ductility modified response spectra (Newmark, 1977). The
deamplification factor is primarily a function of the ductility ratio defined as the ratio of
maximum displacement to displacement at yield. Riddell and Newmark (1979) have shown the
deamplification factor to also be a function of system damping and the slope of the force
deflections curve. EPRI (1994) expands the derivation of a ductility factor, considering the
shape of the hysterisis loop and duration of the earthquake. Chapter 3 of EPRI (1994) provides
two methods for developing inelastic energy absorption factors. One method is an effective
frequency/effective damping method and the other method is a modification to the RiddellNewmark method. The effective frequency/effective damping method takes into account the
shape of the force-deflection hysterisis loop and the shift in frequency of the structure as it is
stressed beyond the elastic limit. The effective Riddell-Newmark method is a modification to
the original formulations for the durations of the earthquake ground motion. It is recommended
that the average of the two methods be used as a median estimate of the inelastic energy
absorption factor. For an example, refer to Section 6 in EPRI (1994) for the derivation of a
fragility for a low-rise concrete shear walls (typical of auxiliary building walls). A system
ductility, = 4.0, corresponds to a median ductility factor F, of 2.45 at 7% damping. The
variabilities in the ductility factor, F , are both estimated for this case as R = 0.21 and

U = 0.21, taking into account the uncertainty in the predicted relationship between F , ,
system damping, and the sensitivity of the inelastic deformation to the actual earthquake time
history.
The structure response factor, FSR is based on recognition that in the design analyses, structural
response was computed using specific (often conservative) deterministic response parameters for
the structure. Because many of these parameters are random (often with wide variability) the
actual response may differ substantially from the calculated response for a given peak ground
acceleration.
The structure response factor, FSR , is modeled as a product of factors influencing the response
variability:
FSR = FSA FGMI F FM FMC FEC FSSI

Equation 2-8

where:

2-19

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries
FSA

= spectral shape factor representing the difference and variability in response due to
the difference between the SSE spectrum and the RE spectrum defined by the
hazard analyst.

FGMI

= ground motion incoherence factor that accounts for the fact that a traveling seismic
wave does not excite a large foundation uniformly.

= damping factor representing variability in response due to difference between


actual damping and design damping.

FM

= modeling factor accounting for any bias and uncertainty in response due to
modeling assumptions.

FMC = mode combination factor accounting for any bias and variability in response due to
the method used in combining dynamic modes of response.
FEC = earthquake component combination factor accounting for any bias and variability
in response due to the method used in combining earthquake components.
FSSI = factor to account for effect of soil-structure interaction including the reduction of
input motion with depth below the surface.
The median and logarithmic standard deviations of F are expressed as:

Fm = FSm Fm FSAm FGMIm Fm FMm FMCm FECm FSSIm

Equation 2-9

and

F = s + + SA + GMI + + SSI
2

2 1/ 2

Equation 2-10

The logarithmic standard deviation F is further divided into random variability, R , and
uncertainty, U . To obtain the median ground acceleration capacity A m the median factor of
safety, Fm , is multiplied by the safe shutdown earthquake peak ground acceleration.

2.2.3.2 Fragility of Equipment and Other Components


For equipment and other components, the factor of safety is composed of a capacity factor, FC ; a
structure response factor, FSR ; and an equipment response (relative to the structure) factor, FRE
Thus,

FE = FCFREFRS

2-20

Equation 2-11

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

The capacity factor FC for the equipment is the ratio of the acceleration level at which the
equipment ceases to perform its intended function to the seismic design level. This acceleration
level could correspond to a breaker tripping in a switchgear, excessive deflection of the control
rod drive tubes, or failure of an equipment support. The capacity factor for the equipment may
be calculated as the product of FS and F . The strength factor, FS , is calculated using Eq. (2-7).
The strength, S, of equipment is a function of the failure mode.
Equipment failures can be classified into three categories:

Elastic functional failures.

Brittle failures.

Ductile failures.

Elastic functional failures involve the loss of intended function while the component is stressed
below the yield point of its structural elements. Examples of this type of failure include the
following:

Elastic buckling in tank walls and component supports.

Excessive blade deflection in fans.

The load level at which functional failure occurs is considered the strength of the component.
Brittle failure modes are those that have little or no system inelastic energy absorption capability.
Examples include the following:

Anchor bolt failures.

Component support weld failures.

Shear pin failures.

Each of these failure modes has the ability to absorb some inelastic energy on the component
level, but the plastic zone is very localized and the system ductility for an anchor bolt or a
support weld is very small. The strength of the component failing in a brittle mode is therefore
calculated using the ultimate strength of the material.
Ductile failure modes are those in which the structural system can absorb a significant amount of
energy through inelastic deformation. Examples include the following:

Pressure boundary failure of piping or vessel nozzles.

Structural failure of cable trays and ducting.

Failure of component support members (plastic bending, plastic buckling).

The strength of the component failing in a ductile mode is calculated using the effective yield
strength of the material for tensile loading. For flexural loading, the strength is defined as the
limit load or load to develop a hinge mechanism.

2-21

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

The inelastic energy absorption factor, F , for a piece of equipment is a function of the ductility
ratio, . The median value of F is considered to be 1.0 for brittle and functional failure modes.
For ductile failure modes of equipment that respond in the amplified acceleration region of the
design spectrum, the inelastic energy absorption factor may be calculated in a manner similar to
structures (refer to Section 2.2.3.1 for a description of the derivations of inelastic energy
absorption factors).
The equipment response factor FRE , is the ratio of equipment response calculated in the design to
the realistic equipment response; both responses being calculated for design floor spectra. FRE is
the factor of safety inherent in the computation of equipment response. It depends upon the
response characteristics of the equipment and is influenced by some of the variables listed under
Eq. (2-8). These variables differ according to the seismic qualification procedure. For
equipment qualified by analysis, the important variables that influence response and variability
are as follows:

Qualification method (QM) dynamic analysis vs static coefficient used, etc.

Spectral shape (SA) - including the effects of peak broadening and smoothing, and artificial
time history generation.

Modeling (affects of mode shape and frequency results) (M).

Damping (affects of design damping vs. median damping) ().

Combination of modal responses (for response spectrum method) (MC).

Combination of earthquake components (ECC).

For rigid equipment qualified by static analysis, the variables, except the qualification method,
and combination of earthquake components are not significant. The equipment response factor is
the ratio of the specified static coefficient divided by the zero period acceleration of the floor
level where the equipment is mounted. If the equipment is flexible and was designed via the
static coefficient method, the dynamic characteristics of the equipment must be considered. This
requires estimating the fundamental frequency and damping, if the equipment responds
predominantly in one mode. The equipment qualification method factor is the ratio of the static
coefficient to the best estimate spectral acceleration at the equipment fundamental frequency.
Where testing is conducted for seismic qualification, the response and capacity may be
determined from specific criteria contained in Chapter 3 of EPRI (1994), pp. 3-57 to 3-71.
The overall equipment response factor is the product of these factors of safety corresponding to
each of the variables identified above. The median and logarithmic standard deviations for
randomness and uncertainty are estimated following Eqs. (2-9) and (2-10).
The structural response factor, FSR , is based on the response characteristics of the structure at the
location of the component (equipment) support. The variables pertinent to the structural
response analyses used to generate floor spectra for equipment design are the only variables of
interest to equipment fragility. Time-history analyses using the same structural models used to

2-22

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

conduct structural response analysis for structural design are typically used to generate floor
spectra. The applicable variables are as follows:

Spectral shape.

Ground motion incoherence.

Damping.

Modeling.

Mode combination (if mode superposition time history is used).

Soil-structure interaction including reduction with depth of seismic input.

For equipment with a seismic capacity level that has been reached while the structure is still
within the elastic range, the structural response factors should be calculated using damping
values corresponding to less than yield conditions (e.g., about 5% median damping for reinforced
concrete). The combination of earthquake components is not included in the structural response
since the variable is to be addressed for specific equipment orientation in the treatment of
equipment response. Median Fm and variability R and U estimates are made for each of the
parameters affecting capacity and response factors of safety. These median and variability
estimates are then combined using the properties of lognormal distribution in accordance with
Equations (2-9) and (2-10) to obtain the overall median factor of safety Fm and variability R
and U estimates required to define the fragility curves for the structure or equipment. For each
variable affecting the factor of safety, the random ( R ) and uncertainty ( U ) variabilities must be
separately estimated. The differentiation is somewhat judgmental, but it can be based on general
guidelines. Essentially, R represents variability due to the randomness of the earthquake
characteristics for the same peak acceleration and to the structural response parameters that relate
to these characteristics. The dispersion represented by U is due to factors such as the following:

Our lack of understanding of structural material properties such as strength, inelastic energy
absorption, and damping.

Errors in calculated response due to use of approximate modeling of the structure and
inaccuracies in mass and stiffness representations.

Usage of engineering judgment in lieu of complete plant-specific design data for equipment
code capacities, and responses.

2.2.4 Information Sources


Fragility evaluation utilizes data from various sources plant specific and generic. Plant specific
information would be design analysis and qualification test data. Generic information consists of
earthquake experience data (EPRI 1998), (SEQUAL, 2001), and test data for relays (EPRI
1991a, EPRI 1996) and for non-relays devices (EPRI, 1991c, 1999). Fragility parameter values
derived for several components in the past seismic probabilistic risk assessments have been
compiled in Campbell et al. (1988).

2-23

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

Several sources of information are utilized in developing plant-specific and generic fragilities for
equipment. These sources include the following:

Seismic design calculations.

Seismic evaluation results from the USI A-46 resolution or IPEEE program.

Plant safety analysis reports.

Plant specific seismic qualification test data.

Generic equipment ruggedness data (GERS), (EPRI, 1991a, 1991c, 1995, 1998).

Past earthquake experience (EPRI 1998, SEQUAL 2001).

In seismic margin studies such as those conducted for many US NPPs in IPEEE, an index of
seismic margin is the HCLPF capacity of the component. This quantity considers both the
uncertainty and randomness variabilities and is the acceleration value for which the analyst has
95% confidence that the failure probability is less than 5%. For example, Figure 2-4 shows a
HCLPF of 0.32g for a fragility description of A m = 0.87g, R = 0.25, U = 0.35 . That is, it is an
acceleration value for the component for which we are highly confident there is only a small
chance of failure given this ground acceleration level:
HCLPF Capacity = A m exp {- 1.65 (R + U )}

Equation 2-12

For some applications, a point estimate (mean value) of core damage is considered adequate.
Uncertainty analysis is not performed. In developing a point estimate of CDF, a composite
fragility curve is commonly used that incorporates randomness and uncertainty into a single
curve. The composite fragility curve is defined by two parameters A m and C where A m is the
median capacity as previously described and C is the composite variability. Refer to Figure 2-4
as an example.
The HCLPF capacity is then approximately defined as: a 1% conditional probability of failure
(-2.33 log standard deviation below the mean).
HCLPF Capacity = A m exp (2.33C )

Equation 2-13

2.2.5 Other Fragility Models


The lognormal model for fragility has been used in most seismic PSAs conducted to date.
Material strength data tends to follow a lognormal distribution whereas data on the pullout
capacity of expansion anchors tends to be normally distributed. The lognormal model is
mathematically easy to use and can be partly justified by the Central Limit Theorem which states
that the distribution of a product of several variables tends to be log-normal regardless of the
distribution of the individual variables. Ellingwood (1994) and Reed and Kennedy (EPRI, 1994)
have also arrived at similar conclusions. However, there have been some attempts to use other
probability models to describe the fragility of components. As a sensitivity study, Ravindra et.
al., (1984) explored the use of Weibull distribution for seismic fragility. Note that the Weibull
2-24

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

distribution has two parameters that can be derived from the mean and standard deviation of the
variable. It was concluded from this study that a Weibull model gives unrealistically high
fragilities in the lower tail. The basic information needed is still the mean and standard deviation
(equivalently, median, and R and U). There is not much empirical and analytical data available
to justify use of probability models requiring more than two parameters.
2.2.6 Hybrid Method
The fragility methodology of estimating the median and R and U described in this report is
universally applicable. It does however require the median factors of safety for different
variables affecting the response and capacity be estimated as well as their logarithmic standard
deviations. In the U.S. nuclear industry, seismic margin assessments have been done for a
number of nuclear power plants. Seismic margin is defined as the HCLPF capacity of the plant
safe shutdown systems relative to the design basis or safe shutdown earthquake (DBE or SSE).
The HCLPF capacity of the weakest link component in the safe shutdown path is considered the
plant level HCLPF capacity. The HCLPF capacities of components are calculated using a
deterministic procedure called Conservative Deterministic Failure Margin (CDFM) method.
EPRI (1991) describes the CDFM method and provides several examples. In order to simplify
the seismic PRA, a hybrid method is suggested in EPRI (1994) and Kennedy (1999). The main
feature of this method is the development of seismic fragility using the HCLPF capacity. First,
the HCLPF capacity of the component is determined using the CDFM method. Next, the
logarithmic standard deviation, C , is estimated using judgement and the following guidance,
(Kennedy, 1999). For structures and major passive mechanical components mounted on ground
or at low elevations within structures, C typically ranges from 0.3 to 0.5. For active
components mounted at high elevations in structures the typical C range is 0.4 to 0.6. When in
doubt, use of C of 0.4 is recommended as a conservative estimate (higher C results in a larger
ratio of median to HCLPF). The median capacity is calculated using Eq. 2-13 and an
approximate fragility curve for the component is thereby obtained. EPRI (1994) further
recommends that this approximate fragility method initially be used for each component in the
systems analysis to identify the dominant contributors to the seismic risk (e.g., core damage
frequency). For the few components that dominate the seismic risk, more accurate fragility
parameter values should be developed and a new quantification done to obtain a more accurate
mean core damage frequency and to confirm that the dominant contributors have not changed.
The CDFM method, though in concept is universally applicable, has been derived following the
seismic design and qualification practices of the U.S. nuclear industry. The parameters and
implied safety factors in the CDFM procedures should be appropriately modified for use in other
countries reflecting the differences in design practices. The same caveat would apply to the use
of generic C values. These generic values have been arrived at using the results and insights of
a number of seismic PSAs involving thousands of fragility calculations. Judgment should
therefore be exercised in their use for new applications in countries outside the U.S.

2-25

EPRI Proprietary Licensed Material


State of the Art and Practice of Seismic PRA in the U.S. and Other Countries

2.3

Plant Level Fragility

It is sometimes useful to develop the plant level fragility curves. They depict the conditional
probability of seismic core damage (or other damage indicators such as large early release) for
different levels of ground motion input. The plant level fragility curves can be generated by
quantifying the accident sequences consisting of component successes and failures. By entering
the plant level fragility curves corresponding to 95% confidence at 5% conditional probability of
failure, the plant HCLPF capacity can be obtained.
In this case the plant HCLPF capacity is determined from the detailed modeling of the plant
systems responses for an earthquake. In seismic margins assessments, the plant HCLPF capacity
is assumed to be the HCLPF capacity of the weakest component in the highest capacity safe
shutdown path. This is a weakest link in a chain methodology rather than a logic tree evaluation
of the plant systems and components response to a seismic event. If the plant safe shutdown
system capacity is dominated by a single weak link, the plant HCLPF capacities will be very
similar in both cases. However, if multiple component failures all contribute to failure of the
safe shutdown path(s), the results of the two methods of defining plant HCLPF capacity may
differ by a significant amount.

2-26

EPRI Proprietary Licensed Material

3
FRAGILITY METHODOLOGY ISSUES AND
ENHANCEMENTS

Over the last decade, a number of issues have been raised and enhancements have taken place
with regard to seismic fragility methodology. These issues and enhancements eminate
principally from 3 areas:
1. The NRC comments/recommendations based on their review of SPRA submittals as part of
the IPEEE program. These comments and recommendations were primarily focused on
methodological areas that needed to be addressed or upgraded for SPRAs beyond IPEEE
where licensees will be submitting Risk Informed Performance Based (RI/PB) applications.
2. Methodological Improvements to the fragility analysis process.
3. Publication of new data that is useful to the calculation of seismic capacity level in support of
the fragility analysis.
Each of these 3 enhancement areas is discussed in detail in this section.

3.1

Methodological Issues from USNRC Review of IPEEE Submittals

The United States Nuclear Regulatory Commission published a comprehensive report on their
review of IPEEE submittals, (USNRC, 2000). The review primarily focused on the extent to
which the licensees IPEEE submittal had achieved the intent of Generic Letter 88-20,
Supplement 4 (USNRC, 1991a), had satisfied the IPEEE objectives and had followed the
guidance of NUREG-1407 (USNRC 1991b). The reviews focused on verifying that the critical
elements of acceptable IPEEE analyses were performed in accordance with the guidance in
NUREG-1407. The reviews were not intended to validate or verify the licensees IPEEE
analyses or results. Rather, methods, approaches, assumptions and results were reviewed for
reasonableness. If inconsistencies were found they were reported in the plant-specific Technical
Evaluation Reports, TERs. During the review process there were Requests for Additional
Information, RAIs, to clarify or enhance portions of the submittals.
Licensees were provided the option of conducting a Seismic PRA or a Seismic Margins
Assessment. There were a total of 75 licensee submittals covering 108 operating units. Seismic
PRA was conducted in 28 of the submittals covering 41 units. Methodological issues cited by
NRC that are addressed in this report are focused on the issues related to Seismic PRA.
The NRC staff concluded that in general, the licensees had achieved the goals of IPEEE but
noted that varying degrees of detail and methodology had been used and that some of these
3-1

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

methodology issues appeared to have a significant influence on the results obtained. The five
major issues identified in the NRC review of seismic submittals are not all specific to the
development of fragilities but are interrelated so are addressed in this report as they relate to
development of fragilities.
3.1.1 Use of Uniform Hazard Spectrum
USNRC Comment: The uniform hazards spectra (UHS) shapes, as employed by the licensees
of some plants in the eastern U.S. (EUS), for component fragility calculations appear
uncharacteristic of conventional spectrum shapes derived from observed earthquakes. These
UHS appear to have substantially reduced energy content, compared to the respective design
basis SSE spectra, in the frequency range that is typically considered to have the greatest impact
on the SSC responses to seismic motions. As a result, the seismic analyses using the UHS as
input resulted in significant reductions (50% to 70%) in seismic demand, compared to the
corresponding design-basis calculations. Furthermore, since there was no consistent guidance
provided for anchoring the UHS to the zero peak ground acceleration (ZPGA), the licensees
applied their engineering judgment for the anchorage of the UHS.
Discussion of Comment: While the fragility analyst is provided the seismic hazard and
associated spectral shape, he must be aware of the characteristics of the site hazard and may have
some technical input to the hazard analyst regarding the frequencies at which spectral ordinates
are desired and whether the fragilities will be referenced to spectral acceleration in a stated
frequency range or be referenced to ZPGA (pga). Reference to spectral acceleration in a
frequency range of the plant structures is generally preferred in order to minimize the uncertainty
between pga and spectral acceleration.
ANS Standards 2.29 and 2.27 (ANS, 1997 and ANS, 2000), currently in draft form, are to be the
governing documents on the development of the site-specific probabilistic seismic hazard. The
ANS 58.21 External Events PRA Methodology Standard provides requirements for development
of UHS for Capability Categories 1, 2 and 3. For Capability Category 2, fractile and mean
ZPGA and fractile and mean UHS are required. For Capability Category 3, magnitude and
distance deaggregation and seismic source deaggregation are required. The deaggregation
process is described in RG 1.165 (USNRC, 1997). Existing hazard studies such as the LLNL
and EPRI hazard studies used in the IPEEE evaluations would correspond to Capability
Category 2. When using the LLNL and/or EPRI hazard studies, or another study done to a
comparable technical level, the intent of this requirement is not to repeat the entire hazard
exercise or calculations, unless compelling new site specific information and interpretations have
been established in the technical literature that affect the usefulness of the seismic PRA for the
intended application.
ANS 58.21 high level requirement HA-G1 requires that the UHS reflects or bounds the sitespecific considerations. For Capability Category 3, the response spectrum shape used in the
SPRA should be based on site-specific evaluations performed for the Probabilistic Seismic
Hazard Analysis (PSHA), and reflect or bound the characteristics of spectral shapes associated
with the mean magnitude and distance pairs determined in the PSHA for the important ground
motion levels.

3-2

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

The UHS, as used in IPEEE, are a broad banded shape that is not characteristic of a single
earthquake. The ZPGA hazard may be dominated by a close in short duration high frequency
earthquake whereas the low frequency range, typical of design type earthquakes, may be
dominated by a larger magnitude distant earthquake characterized by a lower ZPGA.
Consequently, most of the analyses conducted where the structural response was dominated by
low frequency input motion, should be conservative relative to the ZPGA associated with the
UHS. If deaggregation is conducted using guidelines in RG 1.165, two separate spectral shapes
might be developed; one that is based on sources, magnitudes and distances that produce peak
amplification in the 1-2 Hz range and one where peak amplification occurs in the 5-10 Hz
range. These two spectral shapes defined for a specific return period will have separate ZPGAs
associated with the same return period. In essence, there are two different earthquakes to
consider. Capability Category 3 requires that probabilistic response be conducted, which would
have to take into account the deaggregation. Further guidance is provided in Chapter 4 on
probabilistic response.
The NRC comment on lack of energy in the UHS low frequency range will inherently be
addressed if the hazard is defined as required in ANS 58.21. For Capability Category 1 and 2,
the UHS will be conservative if the fragility is anchored to the ZPGA. If the hazard is carried
out to low enough frequencies for each of the points on the UHS, the fragility can be anchored to
a spectral acceleration in the frequency range of the host structure. If the fragilities are anchored
to spectral acceleration, the risk analysis will be much more accurate and will reflect the correct
spectral shape for the dominant frequency, range, since the frequency of occurrence of that range
will be dominated by earthquakes that produce spectral accelerations in that range.
3.1.2 Use of Surrogate Elements in SPRAs
USNRC Comment: The basis and approach for surrogate element modeling are discussed in
EPRI TR-103959 (EPRI, 1994). The overall concept of a surrogate element is to account
quantitatively (albeit approximately) for the risk contribution of components that are screened
out during the walkdown and screening phase of an SPRA. Hence, the failure of a single
surrogate element represents the potential failures of several components that might normally be
excluded from the SPRA model. Use of the surrogate element helps to assure that the SPRA
does not overlook a potentially significant portion of the seismic CDF. Use of a surrogate
element represents acceptable SPRA practice when (1) screening is performed at a sufficiently
high threshold, (2) the capacity of the surrogate element is appropriately assessed to be
consistent with the screening threshold and (3) the surrogate element is appropriately included in
the seismic plant logic model. Otherwise, the usefulness and validity of SPRA findings may be
compromised.
If the surrogate element is used to represent a low screening threshold, resulting in relatively few
components having lower capacity than the surrogate element, dominant risk contributors can be
masked, and the ranking of dominant sequences may be misleading. If the surrogate element is
applied to represent a reasonably high screening threshold, but is not used appropriately in the
seismic plant logic model, then a fraction of the seismic CDF may be missed in quantification.
Such undesirable cases are more related to pitfalls associated with proper application of the
surrogate element, rather than flaws in the conceptual basis for the surrogate element itself.

3-3

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

Therefore, the use and implications of the findings and perspectives derived from some of the
SPRA IPEEEs that have applied the surrogate element approach need to be treated carefully.
To date, an adequately detailed investigation of the implications of using the surrogate elements
in IPEEEs has not been undertaken. No regulatory guidelines have been developed concerning
its use (particularly with respect to sensitivities in plant logic modeling). However, in most
circumstances, if failure of the surrogate element is modeled as leading to core damage and if the
surrogate element is found to be only a minor contributor to seismic CDF, then its use is
probably reasonable.
Discussion of Comment: In IPEEE, surrogate elements have been inconsistently applied or not
applied at all for screened out components. In several cases, the screening level was too low and
the surrogate element or elements comprised a significant amount of the computed CDF. The
Standard (ANS 2002) does not use the terminology surrogate element. It refers to grouping of
elements into a super element. In internal event PRA, the analyst selects a screening level or
-8
-5
cutset frequency (e.g., 10 per year). After calculating the CDF (e.g., 10 per year), he will
confirm that the screening did not have any significant impact by recalculating the CDF with the
-9
cutset frequency screening set at 10 per year. When he reports the CDF, he may not add a
fraction to represent the unaccounted contribution from the screened cutsets. The methods
available to screen components for seismic capacity generally cannot assure failure rates as low
-8
as 10 or so per year, thus, the concept of using surrogate elements to account for risk from
components meeting achievable screening guidelines was conceived. The goal is to screen at
high enough levels such that the risk contribution from screened out elements is low.
There should be an initial interface between the systems analyst and the fragility analyst to assess
the potential impact of various screening levels. There are various methodologies to screen, as
discussed below that arrive at different surrogate fragilities capacity levels, thus different failure
rates. A single surrogate fragility should not be used unless it represents the lowest capacity of
screened out components and is shown to have very low contribution to the computed CDF.
Depending upon the level of documentation available, several levels of screening can be
conducted and associated generic or surrogate fragilities developed. Examples are:
1. Screen based on walkdown screening levels In EPRI NP-6041 (EPRI, 1991). This is the
most common screening level but the lower level of screen at 0.8g spectral acceleration is
often too low of a level to result in insignificant contribution to CDF when the surrogate
elements are added to the plant model.
2. Conduct generic fragility calculations based on the inherent margin in the design basis.
Often, because of excess conservatism in the design criteria used to qualify components, a
fairly high screening level can be developed, subject of course to a walkdown confirmation
of proper installation and absence of systems intersections. This method is applicable to both
qualification by analysis and qualification by test.
3. Conduct reviews of design calculations or qualification test levels to determine a minimum
design margin or qualification test margin that will correspond to a screening fragility level
that is selected to result in minimal contribution to CDF. Many plants have Seismic
Qualification Review Team (SQRT) forms available for all safety class equipment. These
3-4

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

SQRT forms summarize the seismic qualification level and it is efficient to examine these
summaries and screen out many components with a predetermined margin above the
minimum qualification level. For A-46 plants, the results of anchorage calculations and
outlier evaluations are readily available and can be quickly reviewed to determine which
components can be screened at some target threshold and assigned a surrogate fragility
represented by the screening level.
The above considerations for developing surrogate fragilities could result in several generic
fragility levels that represent different classes of equipment or even the same class in a different
building or floor elevation. A single surrogate fragility that is the lower bound of all components
screened out should not be used unless it is predetermined that a fragility based on the lowest
level of screening, when convolved with the seismic hazard, will result in a very low
unconditional probability of failure.
The modeling of screened out components into the event tree-fault tree logic models is beyond
the scope of this Fragility Applications Guide but the fragility analyst and systems analyst should
mutually have an understanding of how the surrogate fragilities are to be used. Use of a single
surrogate element or even one in each fault tree whose top event terminates at a branch on an
event tree, to represent hundreds of screened out components that would normally be modeled as
basic events in each fault tree, is conceptually non-conservative unless the capacity of the
surrogate element is sufficiently high to result in very low unconditional probability of failure.
3.1.3 The Use of New Soil Structure Interaction Analysis Versus the Use of
Scaling
NRC Comment: Two approaches were used in the IPEEE for developing the RLE in-structure
response spectra (IRS). The first approach is associated with scaling the existing design-basis
SSE IRS to the RLE IRS, following procedures outlined in EPRI NP-6041. While this approach
was applied mostly by the plants founded on rock, which is appropriate under the guidelines
given in EPRI NP-6041, it should be noted that a few plants founded on soil also performed this
scaling. The application of the scaling method for structures founded on soil was often justified
by the licensees by stating that: (1) the shapes of the RLE and SSE design ground spectra are
relatively similar and the SSE IRS used for the scaling are broadened spectra. Therefore, if there
is any shift in frequency due to the soil effect, it should be small and its effect on the scaling
should be negligible, and (2) the damping applied in the scaling is much smaller than the
damping associated with the radiation damping if the soil effects were considered. Therefore
scaling was felt to still produce conservative RLE IRS.
The second approach used in IPEEE requires the performance of a new seismic analysis,
including the soil effect, or the soil-structure interaction (SSI) effect and the detailed structural
modeling. This approach was utilized by the majority of plants founded on soil for the RLE IRS
development.
Regardless of the methods applied, it was noted that the new seismic analyses achieved
substantial, across-the-board reductions in the RLE IRS compared to the SSE IRS, when the SSI
effects were introduced in the seismic analysis. It was also noted that in many cases, especially
for deep soil sites, the deconvolution analysis often produced much higher reduction of the free
3-5

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

field motion from the ground surface to the basemat than would be permissible in the design
process.
When the scaling was used, the fragility/HCLPF computations inherited the conservatisms that
existed in the design-basis analyses, whereas new SSI analyses usually remove as much
conservatism as allowed, sometimes going well beyond what would be permissible in design
practice. It was observed that for the plants using the scaling, the scaled RLE IRS is generally
higher than the corresponding SSE IRS, whereas for those plants that performed new SSI
analyses, the seismic analyses often resulted in much lower RLE IRS demand than the designbasis IRS. Therefore, comparison of the component seismic fragility/HCLPF values for two
plants using the two different approaches could be misleading. The different approaches to
estimating building and component seismic responses (scaling vs new SSI calculations) can
significantly affect the magnitude of the reported fragility (or CDF) or HCLPF values. Hence,
comparisons of the seismic capacities should be made mainly among plants which were analyzed
using similar methods.
Discussion of Comment: The Standard, HLR FR-C1-C6, requires realistic seismic response.
For Capability Categories 1 and 2, justification of scaling is required. HLR-FR-C4 would
require new analysis in most cases of soil sites or sites with UHS spectral shapes significantly
different than design spectra. For Capability Category 3, scaling is not acceptable and new
probabilistic response is required.
The requirements of the Standard and the guidance in EPRI TR-103959 and EPRI NP-6041
appear to be adequate to address this issue. Note that if spectra for a soil site are scaled up based
on a damping value less than that of the soil-structure system, the scaled spectra will be
evaluated conservative. The opposite is true of course if the scaling is down such as could be the
case in the low frequency region of a typical EUS UHS. If scaling is used though, it could be
done in a more accurate manner by scaling on a mode by mode basis rather than on a single
dominant mode basis as is suggested in EPRI NP-6041. In Chapter 4 and Appendices A and B a
more accurate procedure for mode by mode scaling of spectra is demonstrated.
The scaling or reanalysis to develop in-structure spectra should be as realistic as possible.
Intentionally conservatism should not be introduced in a SPRA. One should be cautious in the
case of major structure founded on rock but with smaller structures, such as diesel generator
buildings, founded on overburden. Scaling of the rock founded structure may be justified but
scaling of the surface founded structure may not be very accurate. In this case, there could be
inconsistencies in the development of fragilities for components in the two different structures
that could possibly lead to erroneous risk ranking of components.
3.1.4 Reliance on Structures for Which the Original Design Documentation is no
Longer Available
NRC Comment: Some plants identified dams designed by other agencies (Corps of Engineers,
Bureau of Reclamation, etc.) as critical SSCs for providing emergency cooling water or AC
power. In one case, although the dam is not under the quality assurance program and control of
the licensee, it has been reviewed by the licensee and its consultants and found to be able to

3-6

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

successfully withstand the RLE without catastrophic failure. The availability of lake water to
support the plants service water system is thus not jeopardized.
In another plant, a dam failure was identified as leading to the depletion of the water supply to
the Containment Cooling Service Water, which provides cooling water to the low pressure
coolant injection for decay heat removal. This limited the plants capability to perform an
orderly safe shutdown following a seismic event. However, since the dam was designed by other
agencies, in this particular case, no documentation of the original design was available for
review. Therefore, the seismic capacity of the dam could not be determined. This type of weak
link, resulting from an SSC outside the plant boundary and not under plant control, could not be
resolved and documented without coordination and consultation with the relevant agencies
involved.
Discussion of Comment: This is not a common issue but has appeared in some US plants and
elsewhere, such as in SPRAs in Switzerland. There is some guidance in EPRI NP-6041 (EPRI
1991) on how to calculate the deformation and slope stability of earthen dam structures. The
Standard, does not address dams specifically. The general requirements are there to develop
representative fragilities for all SSCs to be modeled. If dams, dykes, canals, or similar structures
can have an effect on the calculated CDF, they must be included in the model and a
representative fragility must be developed, or the screening guidance in EPRI NP-6041 must be
used to demonstrate the stability at a screening level that has little impact on the calculated CDF.
Guidance on the development of fragilities of earthen dams is not provided in EPRI (1994) or
other fragility methodology documents. However, given a deterministic analysis of a dam on
experienced fragility analyst with basic soil mechanics background should be able to derive a
reasonable estimate of fragility. In many cases, an outside expert may be required to develop a
fragility. Due to the very specialized nature of this issue, no detailed procedures or examples are
presented in this report for dams or similar structures.
3.1.5 Importance of Analysts Expertise in Component Fragility/HCLPF
Assessments
NRC Comment: While a complete detailed examination of a licensees component
fragility/HCLPF assessments was beyond the scope of the IPEEE review, selected fragility
calculations were requested from licensees for certain components that were reported in the
IPEEE submittals as having unusually higher capacity than expected from past SPRA
experiences.
Because the major portion of input to fragility analysis is highly subjective, its quality relies
heavily on past SPRA experience and the analysts expertise in these areas. A limited review of
selected fragility calculations suggests that the analysts expertise in component fragility/HCLPF
assessments and his (her) experience with past SPRAs could have a large impact on the quality
of the component fragility calculations. Of the calculations reviewed, some were good quality
fragility/HCLPF assessments, which followed very closely past SPRA practice in the nuclear
industry and the individual analysts expertise was also reflected in the fragility estimates.
There were other calculations, however, which appeared poorly prepared (i.e., lacking material
documentation) and using unrealistic estimates of uncertainties.
3-7

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

Because of the correlation between the analysts expertise and the quality of the fragility/HCLPF
assessments, guidelines or criteria may be needed so that only analysts with sufficient
qualifications will perform the fragility/HCLPF assessments in future SPRAs.
Discussion of Comment: All of the NRC methodology comments and observations relate to the
expertise and experience of the personnel performing the systems and fragility work. EPRI and
the nuclear industry has strived to develop methodology and applications documents so that the
utility engineer, with some additional specific training, as was the case in the SQUG GIP
Training and the Add On IPEEE Training, can perform the bulk of the necessary calculations.
The SQUG GIP and EPRI NP-6041 have requirements for the personnel performing the
walkdowns and the outlier and HCLPF calculations. EPRI TR-103959 does not address
personnel requirements for performing fragility calculations. Unfortunately, SPRA is practiced
by only a few organizations, thus few have been exposed to the methodologies of SPRA. In
NUREG CR-5270 (Kennedy, et al. 1989) it was determined that there was more variability in
analysts then in the fragility and CDFM methodologies. This of course is also true in standard
design calculations performed to industry codes and standards. While design analysts are
generally conservative by nature, errors are made that in some cases reduce the margin implied
by codes and standards. Appendix F presents an example where the analyst had what appeared
as obvious errors in the model and had focused on a failure node that is not representative of
actual failure.
The Standard does not provide any high level requirements for personnel performing the fragility
work. However, a thorough peer review is required which should uncover the deficiencies noted
in the NRC comments. In general, the personnel qualifications are similar to those in (SQUG,
1991) and (EPRI, 1991b). This document and the information in (EPRI, 1991b) and (EPRI,
1994) should be sufficient for experienced engineers meeting the requirements in (SQUG, 1991)
and (EPRI, 1991b) to develop fragilities in a reasonably consistent manner.
The purpose of this document is to enhance the existing methodology and provide more specific
applications guidance on how to develop seismic fragilities that comply with the three capability
categories of the Standard.

3.2

Comments and Suggestions from Industry on Methodology for SPRA

EPRI TR-103959 (EPRI 1994) provides a good overview of the Seismic PRA process and is the
most complete source of information on the development of fragilities. It focuses on the classic
method of developing fragilities, defined by a double lognormal distribution, using the separation
of variables approach. Deviations to the use of a double lognormal distribution are discussed as
is Monte Carlo and Latin Hypercube simulation as opposed to the separation of variables
approach to define the distribution on response.
Budnitz (1998) describes the state of the art of current methodologies for Seismic PSA (PRA).
In this report, he observes that the six sub-methodologies of Seismic PSA are not equally mature
and therefore the many different types of engineering insights from PSA are not all equally
reliable. However, he goes on to state that, whereas the methodology for seismic PSA suffers
from certain problems, these problems are not necessarily more severe than the problems with
the internal-initiators PSA methodology. The six sub-methodologies discussed are:
3-8

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

1. Seismic Hazard
2. Seismic Local Ground Motion and Building Motion
3. Walkdown Methodology
4. Failure Mode and Fragility Methodology
5. Seismic PSA Systems Analysis Methodology
6. Seismic PSA Methodology for Analyzing Plant response, Offsite Release and Consequences.
It is correctly noted and known in the industry, that the largest distribution in computed seismic
CDF arises from the large uncertainty in the seismic hazard. Even though the methodology is
considered to be mature, reasonable experts differ in their assessments and this difference must
be taken as genuine uncertainty. The methodology for sub methodologies 2, 3 and 4, as
described in EPRI (1994) is considered to range from mature to very mature, it is pointed out that
uncertainties remain for many items of equipment. This means that different analysts will
produce different capacities and fragility curves. In Kennedy (1989) this was demonstrated
when four experts varied by about a factor of 1.5 for many of the identical problems they were
provided for purposes of calculating a fragility and for calculating a HCLPF by deterministic
methods. When the same problems were given to experienced industry designers without
previous background in development of fragility and HCLPFs, the variation was even larger.
This relates to the NRC comment on expertise of the analysts in their review of IPEEE
submittals (USNRC, 2000).
It is noted in Budnitz (1998) that the use of lognormal mathematics is known to be an erroneous
approach in the tails of the lognormal distributions, even when the lognormal shape adequately
represents the data in the main parts of the distribution, because the data do not fit a lognormal
distribution in the tails beyond a couple of log-standard deviations. Despite these limitations, the
lognormal model is commonly used principally for calculational convenience. In a study by
Ravindra et al. (1984) it was shown that a Weibull distribution produced unrealistic results in the
lower tail of the fragility curve. As pointed out in Kennedy (1999) common practice is to cut off
the tails of the lognormal distribution at the HCLPF value (-2.33 log standard deviations if the
HCLPF is defined by a single lognormal curve. The risk results are more sensitive to the HCLPF
value than the median value of the governing fragilities, so it is important to focus on the
accuracy of the lower portion of the fragility curve.
Systems modeling is not within the scope of this guide but it is pointed out in Budnitz (1998) the
importance of the interface between the systems analyst and the seismic fragility analyst to help
each other to focus on the issues deemed important. These interfaces also include the hazard
analyst. Of particular importance is the correlation between component failure modes. The
fragility analysts must advise the systems analysts of correlation between component responses
and component capacities and the systems analysts must not only incorporate these correlations
into the systems model as best as they can, they must conduct sensitive studies for different
modeling interpretations of the correlations. It is pointed out that in some cases, where several
components must fail together to result in core damage, and the correlations among them are not
well understood, the differences between assuming full correlation and zero correlation can
3-9

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

amount to an order of magnitude difference in calculated CDF. This is compared to plus or


minus one order of magnitude or more in calculated CDF that can result from the uncertainty in
the seismic hazard.
Budnitz (1998) does not suggest any specific changes to the current methodology. The purpose
of his report was to describe the current methodology and review the methodology to the extent
to which current methodology produces reliable and useful results and insights. The conclusion
is that, despite the numerically large uncertainties, these uncertainties should generally not
invalidate the key engineering insights concerning potential earthquake related vulnerabilities.
At the OECD/NEA Workshop on Seismic Risk held August 10-12, 1999, in Tokyo, Japan, a
number of papers were given on seismic PSA methods and seismic ruggedness testing. The
papers that contribute most to the current issues of fragility methodology are by Kennedy (1999),
Ravindra (1999), Fleming (1999) and Watanabe (1999).
In Kennedy (1999), suggestions are made for simplifying the seismic fragility development
method and the systems analysis methodology with the objective of obtaining a point estimate of
CDF. These recommendations arise from the conjecture that the uncertainty in the seismic
hazard can be about two orders of magnitude wide in the 15% to 85% NEP range and that even
the most sophisticated SPRA cannot predict a point estimate of the CDF to any better accuracy
than a factor of five. Many of these suggested simplifications are outlined in EPRI (1994) and
were used in the IPEEE program to satisfy the objective of identifying vulnerabilities rather than
an objective of computing numerical values of CDF, and conducting uncertainty analyses, for
comparison to internal event CDF. The main suggestion in Kennedy (1999) is to use a hybrid
method, or a simplified hybrid method, for estimating CDF. The simplified hybrid method in
conjunction with simplified risk modeling would not comply with the requirements of ANS
(2002) thus is not discussed further.
The hybrid method would correspond to Capability Category 1 in the Standard. The hybrid
method follows the normal steps of a SPRA but the fragility curves are developed in a more
simplified manner. It is recommended to estimate the HCLPF capacity of the components by the
CDFM method used for Seismic Margin Assessments, EPRI (1991). The next step is to
approximate the composite logarithmic standard deviation C and calculate the median capacity
using the CDFM HCLPF and the estimated C . For structures or major passive components
mounted on the ground or at low elevations, C usually ranges from 0.3 to 0.5. For active
components mounted high in the structures the typical C ranges from 0.4 to 0.6. The estimate of
C depends upon the uncertainty in the spectral shape relative to the peak ground acceleration,
typical uncertainties for demand and capacity calculations and the degree of inelastic energy
absorption inherent in the structure or component. When in doubt, the lower values are the more
conservative and should be used.
The above recommendation is based on the fact that the final CDF is more sensitive to the
HCLPF than to the median capacity, thus the accuracy of is not so important. Note, however,
that in the discussion of calculating HCLPF, Kennedy (1999) states that for critical components
that he calculates the HCLPF using the CDFM method and the fragility method and if there is a
significant difference, he favors the fragility method. It is not mentioned in the paper that the
3-10

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

CDFM HCLPF is relative to an 84th percentile demand and must be scaled to a HCLPF relative
th
to a 50 percentile demand. In EPRI (1994), pages 5-5 and 5-6, the authors elaborate more on
the development of a median capacity from a CDFM HCLPF. The calculated CDFM HCLPF is
relative to an 84th percentile demand and must first be adjusted to a HCLPF50 by dividing the
CDFM HCLPF by exprs where rs is the combined logarithmic standard deviation for the
horizontal component response spectrum shape basic variable and is the SRSS combination of
the randomness r and uncertainty u for this variable. The random and uncertainty s are
provided in Table 3-2 of EPRI (1994) and depend on the frequency range of interest and whether
the fragility is anchored to spectral acceleration or peak ground acceleration. rs is typically
taken as about 0.3 for the simplified approach to developing fragilities from CDFM calculations.
In some cases, at certain frequencies, the UHS of EPRI (1989) and USNRC (1994) have a very
large difference in spectral amplification between the 50th and 84th percentile UHS and u alone
is greater than 0.3. Thus, in developing fragility curves from a CDFM HCLPF, the analyst
th
th
should examine the difference between the 50 and 84 percentile amplification of the site
specific UHS and either determine the applicability of a standard rs of 0.3 or develop a specific
rs to correspond to the site specific UHS.
Note that in the IPEEE program, the plants that chose to do seismic margins assessments were
required to utilize a NUREG/CR-0098 spectral shape rather than a UHS from EPRI (1989) or
USNRC (1994) in developing a CDFM HCLPF. In these cases, the difference in spectral shape
between the NUREG/CR-0098 and UHS ground motion response spectra must also be accounted
for. In the case of soil sites, scaling from one spectral shape to another may not be reasonable
and a new structural analysis may be required if HCLPFs from an IPEEE SMA are to converted
to fragilities for use in a subsequent SPRA.
Chapters 2, 3 and 5 in EPRI (1994) describe the procedure for converting a CDFM HCLPF to a
fragility in detail and further elaboration is not necessary in this guide.
The use of surrogate elements to represent components that have been screened out is discussed
in Kennedy (1999), Ravindra (1999) and Chokshi (1999) as well as in the NRC review of the
IPEEE submittals, USNRC (2000). It is pointed out that the unconditional failure rates, thus
risk, implied by the screened out components should be small. This was not always the case for
the IPEEE submittals. Typically, components were screened out on the basis of screening tables
in EPRI (1991) or by use of SQUG GIP bounding or reference spectra (SQUG, 1991) and
meeting associated screening caveats. The development of a surrogate fragility to represent
components screened out is described in Kennedy (1999) but as is the case for development of
fragilities from CDFM HCLPFs described above, the Kennedy (1999) description does not
address the ground rule that the screening level is assumed to be a CDFM HCLPF84 and must be
adjusted to a HCLPF50 value. This is described in detail in Chapter 5 of EPRI (1991). The
screening spectral acceleration is adjusted to a HCLPF50 spectral acceleration by dividing by the
factor exprs.
HCLPF50 = Screening Level/exprs

where rs is as discussed above for development of a fragility from CDFM HCLPF calculations.

3-11

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

In Kennedy (1999) and EPRI (1991) it is recommended that the median capacity conservatively
be twice the HCLPF50 value, implying a C of 0.3. This is a conservative assumption since the
rs used to represent the randomness and uncertainty in the seismic hazard is typically 0.3 and
the assumption of a C of 0.3 to represent all randomness and uncertainty in demand and
capacity implies that there is no variability in the demand other than in the earthquake itself and
that the capacity is always equal to the demand. It would seem more rational to assume a C in
the range discussed above for development of a fragility from a CDFM HCLCF. Although, as
correctly pointed out in Kennedy (1999) the end risk result is not very sensitive to the assumed
C , assuming of course we are talking about a reasonable range of values, so assuming a higher
C , and calculating a higher median capacity for the surrogate should not make that much
difference in the final risk results as long as the screening level was set sufficiently high to begin
with. The problem is, that for sites with medium seismic hazard (0.2 0.25g pga), the screening
tables of EPRI (1999) may not be high enough to result in a surrogate element with a failure rate
low enough to effectively not have an influence on the final results. Refer to Appendix E for an
example.
The influence of the surrogate element or elements developed for different types of components
that are screened at different levels depends also on the modeling assumptions and the assumed
correlation between screened out elements. Note that in ANS (2002), High Level Requirement
FR-B1, components cannot be screened out for Capability Category 3 unless their failures can be
shown to be independent of the remaining components. In any case, HLR SAB-3 states to
PERFORM an analysis of seismic-caused dependencies and correlations, in a way so that any
screening of SSCs appropriately ACCOUNTS FOR those dependencies and correlations.
Fleming (1999) and Watanabe (1999) discuss the importance of properly accounting for
correlation and provide examples of case studies. The issue of correlation is primarily a systems
modeling issue and not in the direct scope of the development of fragilities. Nevertheless, it is
important for the fragility analysts and the systems analysts to interface on this issue so that the
screening levels, the effect of the screening levels on CDF and the sensitivity of screening levels
to correlation are well understood by both parties.
The systems analysts should set a target for the unconditional failure rate of screened out
components to be represented by surrogate fragilities. Kennedy (1999) suggests a simple method
to establish the screening level, given the target unconditional failure rate, PFS .
Given PFS, enter the seismic hazard curve (presumed to be the mean hazard curve) at an
exceedance frequency HS(a) given by:

HS(a) = 2PFS
And determine the corresponding ground motion level, as .
Set the screening level, SL at:
SL = 0.8 as

3-12

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

The systems analysts should establish the value of PFS based on their knowledge of the plant
design and operation and their judgment on the effects of correlations between screened out
components and between screened out components and components remaining in the model.
The screening level may be established as a spectral acceleration value or as a peak ground
acceleration value, depending upon how the systems analysts will compute the seismic failure
rate from the seismic hazard definition. It is then the job of the fragility analyst to determine the
appropriate screening value. For instance, if the systems analyst sets a screening target in terms
of peak ground acceleration, the fragility analyst must relate this to spectral acceleration of the
ground motion spectrum for comparison to screening tables in EPRI (1991) that are defined in
terms of spectral acceleration.
Preferably, the systems analyst will convolve the hazard curve with candidate surrogate
fragilities and determine the required surrogate fragility or HCLPF50 value for the surrogate and
provide this information to the fragility analyst. Given the target HCLPF50 for the surrogate, the
fragility analyst can then propagate this to a HCLPF84 value for comparison to the screening
tables in EPRI (1991), taking into account the amplification of peak ground acceleration to
obtain spectral acceleration. If the target screening level is set too high, the fragility analyst then
knows that he must use other sources of information besides EPRI (1991) screening tables for
screening. Other sources of capacity information such as summaries of qualification margins,
generic fragility calculations based upon required qualification procedures and practices or other
sources of ruggedness data such as Generic Equipment Ruggedness Spectra (GERS) may be
used to screen at different levels.

3.3

New Test and Earthquake Experience Data

There is an ongoing effort through SQUG and EPRI to enhance and develop new GERS based
upon the results of qualification testing. The Seismic Qualification Reporting and Testing
Standardization (SQRTS) organization is an EPRI sponsored group of utilities which has sought
to achieve both qualification test standardization and economy by conducting collective
(multiple item) shake table tests of replacement equipment for NPPs. The qualification and/or
fragility testing results are summarized in a standard test report format and made available
through a seismic qualification data library service to all SQURTS utilities. Most of the testing
is at the device level such as relays, switches, molded case circuit breakers, transmitters, etc.
These data are then analyzed to either enhance existing GERS as originally reported for non
relays and for relays (EPRI, 1991a, c) or develop new GERS for addition to the EPRI GERS
reports. New GERS and updates of GERS were developed in EPRI TR-105988-V1 (EPRI,
1996) and EPRI TR-105988-V2 (EPRI, 1999) and incorporated as addenda to the EPRI GERS
reports for non relays and relays (EPRI, 1991a, c).
A pilot program was initiated to gather test data from other countries to enhance GERS. Data
was obtained from the United Kingdom for Control and Instrument Panels (EQE, 2001), referred
to as Instrumentation and Control Panels and Cabinets (I&C) in the SQUG GIP. Candidate
GERS were developed for cabinets, including devices within the cabinets, using the rules for
developing GERS for low diversity class equipment. A joint SQUG/USNRC review panel has
previously ruled that I&C panels and cabinets are a high diversity class, thus there are as yet no
official GERS for this class. The data from the UK is very useful in a general sense though for
3-13

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

demonstrating typical damping values and fundamental frequencies. There were two groups of
panels. The first group was circa 1980-1985 tested in a biaxial or triaxial mode. The second
group was circa 1989 to 1997, all tested triaxially. Fundamental frequencies ranged from 7.5 to
21.4 Hz front to back and 7.8 to 33.7 Hz side to side. Panels in the second group tended to be a
bit stiffer but there were individual panels in the second group with frequencies near the lower
level cited. Damping ratios ranged from 3% to 13%. The lower damping ratios were generally
associated with very stiff cabinets. The average of the 16 cabinets tested was above the 5%
damping considered as median damping in the development of fragilities (EPRI, 1994).
At the OECD/NEA Workshop held in Tokyo, 1999, several Japanese papers were presented that
summarized high level shake table tests of components conducted on the NUPEC Tadotsu
Engineering Laboratory shake table. Many tests are for passive large scale model structures or
equipment that are specific designs. One test of a diesel generator determined, through
extrapolation of response measurements, that the crankshaft locating bearing housing would be
the governing failure mode when it yielded. In general, these tests are for specific designs and
cannot be effectively used to develop generic ruggedness levels for typical US equipment. They
do, however, reinforce the ruggedness presumed in the EPRI (1991) screening tables.
NUPEC tests of more typical generic electrical equipment are summarized in Ueki, et. al. (1999).
From the quoted fundamental frequencies, the equipment tested is considerably stiffer than
typical SQUG database equipment, thus typically more rugged. The tests were initially
conducted as proving tests and then the shaking level was increased to high levels with no
reported malfunction or damage. The high level tests ranged from 3.4g zpa for metal clad
switchgear (minimum side to side fundamental frequency of 22 Hz) to 8.0 g zpa for an upright
panel (minimum side to side fundamental frequency of 27 Hz). These frequencies correspond to
the zpa so the low frequency peak of the test response spectra at about 2 Hz has no effect on
response.
Due to the differences in design, the very stiff construction and the differences in manufacturers
of internal devices, the data are not directly applicable for enhancement of GERS developed for
US equipment. The tests do, however, further reinforce the ruggedness of stiff cabinets and
devices within stiff cabinets.
The development of fragilities from GERS is clearly described in EPRI (1994), Chapter 3 and
needs no further explanation here.
There is an ongoing effort by the SEQUAL Owners Group (SEQUAL, 2001) to utilize
earthquake experience for seismic qualification of equipment in non-A46 plants. In this effort,
extensive study has been made of earthquake records and the associated representation of
equipment classes with the records. In the SQUG GIP, a Reference Spectrum was defined as the
average of 4 ground motion records that subjected 20 generic classes of equipment to high level
seismic input motion. This Reference Spectrum represented the seismic ruggedness of the 20
generic classes of equipment, subject to meeting the associated demand to capacity evaluation
and specific caveats for each of the classes. The SQUG SEQUAL program has enhanced the
database that formed the basis for the SQUG GIP and has taken a more statistical approach to
demonstrating that the seismic experience data can be used to qualify equipment. In SEQUAL
(2001), applicable earthquake records representing a minimum population of equipment in each
of the 20 classes have been averaged using a weighted average method and compared to the
3-14

EPRI Proprietary Licensed Material


Fragility Methodology Issues and Enhancements

SQUG Reference Spectrum in the frequency range of 2.5 to 7.5 Hz. This is the broad-banded
amplified acceleration portion of the SQUG Reference Spectrum. This updated compilation of
representative spectra shows small differences for each class of equipment. For some classes,
the SEQUAL capacity spectra exceed the Reference Spectrum and for other classes, the
SEQUAL capacity spectra are a little bit lower than the Reference Spectrum but in no case are
they less than 1.1g spectral acceleration in the frequency range of 2.5 to 7.5 Hz, which is 90% of
the 1.2g Reference Spectrum. These individual capacity spectra and the number of components
represented by each equipment class capacity spectrum are contained in Appendix D of
SEQUAL, 2001 and can be used in the development of fragilities using experience data for
equipment included in the SQUG 20 generic classes of equipment. A procedure for development
of fragilities from experience data using survival analysis is presented in Appendix C and an
example is presented in Appendix D.
It is noted that the NRC approved ground motion spectra for piping contained in USNRC (1999)
would result in average spectral acceleration over the 2.5 Hz to 7.5 Hz range that is less than the
SQUG Reference Spectrum. It is however noted that in many cases, the spectral accelerations
are high at very low frequency and these data can be used to develop displacement capacities for
very low frequency systems such as rod hung piping. The SQUG GIP reference spectrum cuts
off at 2 Hz. Low frequency systems are displacement and velocity sensitive and in many
instances of developing fragilities for low frequency systems such as rod hung piping or for
sliding or rocking of rigid bodies, experience spectra are needed at 1 Hz and below.
EPRI (1994) does not address the development of fragility from seismic experience data.
SEQUAL (2001) has a generic example in Appendix B. Salmon and Kennedy (Salmon, 1994)
also have an example. Both examples use the SQUG Reference spectrum as a capacity spectrum
for comparison to in-structure response spectra. In these cases, the assumed C varies slightly,
thus the ratio of HCLPF to median capacity is slightly different. The median capacity in each
case is statistically derived based upon an assumed number of components represented in the
database. There are slight differences in the methodology but the differences would not result in
any significant difference in the end item fragility, given identical demand on the component.
The example developed in Appendix D of this report demonstrates a procedure for development
of fragility from an earthquake experience capacity spectrum as defined in SEQUAL (2001).
The example is based on the survival analysis methodology in Appendix C, as used in the
SEQUAL generic example (SEQUAL, 2001).
Other earthquake records may be added to enhance the SEQUAL capacity spectra. A SQUG
method for gathering and validating earthquake experience data (SQUG, 2001) has been
approved by the regulators (USNRC, 2001). There have been several more recent severe
earthquakes world wide that have been investigated to some degree (Turkey, Taiwan, Kobe,
Hualien, Lotung) but specific data gathering and evaluation in accordance with current approved
criteria has not been carried out for them. Therefore, the data collected cannot effectively be
used in enhancing the SEQUAL capacity spectra. In general, the observations from these events
have concurred with observed damage or lack of damage from SEQUAL and SQUG database
facilities.

3-15

EPRI Proprietary Licensed Material

4
DEVELOPMENT OF FRAGILITIES IN ACCORDANCE
WITH ANS 58.21

The new ANS Standard 58.21 for External Events PRA, ANS (2002), sets down high level
requirements for conducting a SPRA. There are three different capability categories in the
Standard that require varying degrees of sophistication. The Standard is not intended to provide
methodology for meeting the requirements but does reference some applicable documents. This
chapter will focus on the detailed development of fragilities and the necessary interface with the
hazards analysts and systems analysts in order to make the fragilities compatible with the
Standard and the intended use of the fragilities. EPRI TR-103959 (EPRI, 1994) is the most
complete methodology guide available on the development of fragilities. In most cases, the
methodology discussed would satisfy all three capability categories of the Standard. In some
cases, the approximate methods suggested would not comply with the requirements for all
capability categories in the Standard. In a few cases, the guidance should be enhanced for
compliance with all capability categories in the Standard.
The approach taken in this chapter is to list the step by step progression of developing fragilities
and provide discussion on what is necessary to meet each of the ANS Standard capability
categories. If the methodology in EPRI TR-103959 is adequate, for one or all of the capability
categories it will be cited and not repeated. If the methodology needs to be expanded or is not
contained in EPRI TR-103959, then guidance will be provided herein. In all cases, the guidance
will be with reference to the specific capability categories in the Standard for which it is
applicable.
EPRI TR-103959 provides excellent examples of the development of fragilities for common
structures and equipment. In cases where further examples are beneficial to cover areas not
addressed in detail in EPRI TR-103959, they are developed in the Appendices.
Table 4-1 provides a summary of the steps in the development of seismic fragilities and the
associated high level requirements in the Standard. It also includes a column to reference the
appropriate coverage in EPRI TR-103959 or other documents, if applicable. If an addition to, or
expansion of, EPRI TR-103959 is required, it is noted. These additions and expansions are then
discussed in more detail in the following text and references are made to example calculations if
they are provided in the Appendices.
The following major steps are required in a Seismic PRA.
1. Develop the seismic hazard
2. Develop a systems model that represents the plant response to earthquakes.
4-1

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

3. Calculate the response of the structures housing systems and components to the earthquake
input motion (demand on structures, systems and components).
4. Review the design basis and conduct prescreening.
5. Conduct a Walkdown of the plant to identify potential seismic vulnerabilities and to screen
out rugged components.
6. Develop seismic fragilities for structures, systems and components that are not screened out.
a) Determine the response of the systems and components
b) Determine the capacity of the structures, systems and components to withstand the
seismic input motion.
7. Calculate the response of the plant due to the seismic input motion and the progressive
failures of structures, systems and components.
Some of these tasks may be conducted concurrently but each requires an understanding of the
compatible tasks and continuous interface with the different parties conducting the tasks. The
focus of this document is on steps 3, 4, 5 and 6. The interface with steps 1 and 2 is important in
the implementation of steps 3, 4, 5 and 6 and this interface is also elaborated.

4.1

Understanding the Seismic Hazard

The Standard cites ANS (1997) and ANS (2000) for the development of the hazard and provides
some high level requirements for three Capability Categories. The seismic hazard is typically
defined as the annual probability of exceedance of a ground motion parameter or parameters,
such as peak ground acceleration (pga) or spectral acceleration, Sa, occurring at the site. As was
typically done in EPRI (1989) and USNRC (1994), a peak ground acceleration was developed
for a range of 1E-1 to 1E-7 or lower annual probability of exceedance. Uniform Hazard Spectra
(UHS) were developed for a few annual probability of exceedance values such as 1E-3, 1E-4 and
1E-5. Examples of a typical seismic hazard definition are shown in Figures 5-1 and 5-2 for peak
ground acceleration and for a uniform hazard spectrum associated with 1E-4 annual probability
of exceedance.
Since there is uncertainty in the many parameters used to develop the seismic hazard curves and
UHS, the distribution about the mean or median value of the pga hazard or UHS is also defined.
In the calculation of seismic induced failure rates of SSCs, the fragility curves are convolved
with the seismic hazard to compute the unconditional probability of failure. In order to conduct
an accurate computation, the hazard must be defined out to a very low probability of annual
exceedance, typically 1E-8 or 1E-9. In NUREG-1407, USNRC (1991b) It was required to
extend the mean seismic hazard to 1.5g peak ground acceleration unless it could be shown that a
cut off at a lower acceleration did not result in a significant difference in the results. As shown
in Figure 4-1, this would have required an extrapolation of the mean seismic hazard to less than
1E-7 annual probability of exceedance. Note that in IPEEE, only a point estimate of CDF was
required, thus only the mean seismic hazard curve was used. This would correspond to
4-2

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Capability Category 1 in the Standard. Capability Category 2 and 3 require uncertainty analysis
and the full uncertainty in the seismic hazard must be considered.
Given, the information as shown in Figures 4-1 and 4-2, regarding the seismic hazard, the
fragility analysts and systems analysts are somewhat forced to reference the fragilities to the
peak ground acceleration as opposed to spectral acceleration in a frequency range of interest.
The Standard and EPRI (1994) both recommend that the fragilities be developed relative to
spectral acceleration, as this reduces uncertainty in the response of the SSCs. For instance, if a
structure is responding at 5 Hz, defining the hazard in terms of spectral acceleration at 5 Hz will
result in less uncertainty in the structural response than if the hazard is defined as peak ground
acceleration and the uncertainty between peak ground acceleration and spectral acceleration must
be considered as well.

Figure 4-1
Annual Probability of Exceedance of Peak Ground Acceleration

4-3

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Figure 4-2
Uniform Hazard Spectra for the 10-4 Annual Probability of Exceedance. Spectra shown for
three percentiles: 15th, 50th, and 85th

EPRI (1994) in Chapter 3, discusses three important features of the hazard definition that the
fragility analyst must consider in developing fragilities. They are; 1) the relationship between
the peak ground acceleration and spectral acceleration, thus spectral shape; 2) the relationship
between the two orthogonal horizontal components of earthquake and 3) the relationship
between the horizontal and vertical ground motion. The hazard curves developed in EPRI
(1989) and USNRC (1994) were assumed to represent the average of the two horizontal
components of earthquake. A specific seismic hazard for the vertical direction was not
developed and the vertical hazard was typically assumed to be 2/3 of the horizontal hazard. A
vibration frequency was not associated with the peak ground acceleration which, as noted in the
NRC review of IPEEE submittals, USNRC (2000), resulted in one case of a misinterpretation
that resulted in an unconservative CDF computation.
In IPEEE there was also some confusion and misinterpretation of the control point for the
hazard. The hazard was stated to be in the free field. In developing seismic hazard the
earthquake is propagated from the source to the site through underlying rock. If the site is
overlain by soil, the hazard analyst must then propagate the hazard to the surface to define the
hazard in the free field. For shallow soil sites, this may not always be done, therefore the
fragility analyst must take into account any amplification of the soil column to the foundation of
structures that are not founded on the rock control point where the earthquake hazard is defined.
The fragility analyst must have a clear understanding of where the control point is located for the
defined hazard.
EPRI (1994) in Table 3-2, provides estimates of uncertainty to be included in the fragility
computations for cases where the fragility is reference to peak ground acceleration vs spectral
4-4

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

acceleration and for cases where the vertical acceleration is assumed to be two thirds of the
horizontal acceleration. These uncertainties should be taken into account if the fragilities are
referenced to peak ground acceleration. The random variability is also tabulated in the table.
The values provided are ranges of values from studies of earthquake records. For a specific site,
the random and uncertainty variables may be more or less than the ranges shown. For instance,
th
th
in some of the IPEEE sites, at certain frequencies, the difference between 84 and 50 percentile
UHS spectral acceleration is quite large when they are both anchored to the same pga. Besides
the peak to peak variation random variability given in Table 3-2 of EPRI (1994), uncertainty
variability may be larger than provided in EPRI (1994) and should be considered in these cases
where there is a pronounced difference in the amplification of pga. Of course, if the fragility is
anchored to spectral acceleration, the uncertainty variability of the spectral acceleration
amplification above the pga is captured in the hazard curves during the convolution of the hazard
and fragility and only the peak to peak random variability should be included in the fragility
derivation.
Other information that the fragility analyst may need is the peak spectral displacement, and peak
spectral velocity of the earthquake ground motion. For very low frequency systems, the failure
is often governed by large displacement or for cases of rigid body rocking or sliding, peak
velocity is often desired. The UHS have typically been defined down to 1 Hz. Spectral velocity,
Sv, and spectral displacement, Sd, can be derived from spectral acceleration, Sa, by the
relationships:
Sv = Sa/
Sd = As/2

where is the circular frequency.

The peak spectral displacement usually occurs below 1 Hz, thus the hazard analyst should
provide these values also or else provide hazard parameters down to say 0.25 Hz.
The existing EPRI (1989) and USNRC (1994) hazard studies are considered to be applicable to
Capability Category 1 and 2 in the Standard. Capability Category 3 would require a more
detailed hazard analysis. In this case, de-aggregation is required and if the high frequency range
and low frequency range of the UHS are governed by two or more different sources, then in
effect, two or more different earthquake hazards are present with different spectral shapes and
different annual frequency of occurrence. In this case, two or more separate fragility
descriptions are required for each SSC and two or more separate calculations of CDF are
required in order to arrive at the final CDF.
The development of the seismic hazard is to be conducted in accordance with draft ANS
standards 2.29 and 2.27 under preparation, ANS (1997 and ANS 2000). These standards will
specify the scope of the hazard studies and the output that conforms to each of the Standards
Capacity Categories. The fragility analyst must understand the source and meaning of the
important hazard parameters.

4-5

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

4.2 Understanding the Development of the Risk Model and Equipment


List and the Significance of Screening Thresholds
The systems analyst will develop an event tree/fault tree risk model that will be used to calculate
the CDF due to seismic events, coupled with random failures and human error probabilities. The
systems model will typically contain non-safety components as well as safety components in
order to capture the success (non-failure) of normal operating systems as well as the failure of
safety systems. The model will also contain the non-seismic failure rate or unavailability of
systems or components. A typical example is the failure of diesel generators to start or continue
running, given the loss of off site power. These random failure rates are usually contained in the
internal event risk model. The fragility analyst is provided a list of equipment that is to be
modeled. This list may consist of 1000 or more components plus an even larger list of relays. It
is not practical to develop detailed fragility descriptions for each component even if the systems
analyst models each component. Typically a screening threshold is established by the systems
analyst whereby the components can be screened out and either not modeled in detail, or else
surrogate elements can replace groups of elements that are screened at a high capacity level.
In EPRI (1994) the terminology surrogate element is used to denote an element introduced
into the risk model that represents several components that are screened at a stated level. In
practice there may be several screening levels, representing several generic types of components.
The surrogate element or elements then require a fragility description that is to be developed by
the fragility analyst. However, the systems analyst should set a target for the surrogate element
or elements so that they have a very small contribution to risk. In practice, this use of a surrogate
element has been over simplified and in some cases, as noted in USNRC (2000), some IPEEE
submittals utilized a single surrogate element to represent all screened out components. This
over simplification would appear to be unconservative for cases where uncorrelated components
in series (OR Gates in the risk model) have similar capacities close to the screening level. Their
individual failure rates are counted only once as a single failure instead of being added. This
report will not provide criteria for the modeling of screened out components but it is pointed out
that there needs to be close interface between the systems analyst and fragility analyst in order to
assure that screening levels are either high enough to completely screen out the components or
surrogate fragilities are included in the model in a manner that captures the risk contribution
from the components represented by surrogate fragilities. Note that in the Standard the focus is
on setting a screening level so that the screened out components do not contribute significantly to
CDF or LERF. This may not always be practical and some groups of components may
necessarily be represented by a fragility that does result more than an insignificant contribution
to CDF or LERF.
A good place to start is for the fragility analyst to develop some preliminary fragility descriptions
for screening levels that may be achieved. For instance, there are two screening levels in EPRI
(1991) that may be used to screen most of the components. Fortunately, the equipment list is
dominated by valves and most can be screened at the higher (1.2g spectral acceleration)
screening level. In many cases, certain groups of components or subsystems like piping, HVAC
and electrical raceways can be preliminarily screened at some level or levels based on their
conservative design criteria. Not all components or subsystems have equal margins built into
their design criteria so different levels of generic fragility can be assigned to different classes of
equipment based upon their seismic qualification criteria. The systems analyst can then advise if
4-6

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

the proposed or achievable screening level will have a significant effect on final risk results. If it
is determined that a certain easily achievable screening level is too low, then the fragility analyst
will understand that the screening level needs to be increased by some means. More effort will
be required to further subdivide a large group into smaller groups, with associated generic
fragilities, that may be based on different methods of generic fragility development, or it may be
determined that plant specific fragilities are required for certain groups of equipment. Appendix
E shows an example where the EPRI screening tables are not sufficient for screening but,
because of conservatism in the design process, flexible subsystems and components could be
screened. In this case, the use of surrogate elements for the achievable screening level did result
in a significant, though not dominant, contributions to CDF. Hindsight indicates that the
screening level should have been higher but if so, the level of effort required to develop
individual or generic subclasses of fragilities would have increased significantly.
The fragility analyst should also provide some guidance to the systems analyst on the correlation
of components within each group of components falling into a screening level and between
groups of components for different screening levels. For instance, there may be a fairly low
screening level that applies to non-safety components in the power conversion system. This
screening level fragility, along with a fragility for loss of off site power based on switchyard
failures, may determine the failure or success of normal shutdown systems, given an earthquake.
In this case, the failure of mechanical components in the power conversion system would be
independent of the loss of off site power. Likewise, the failure of flexible piping in the power
conversion system would be uncorrelated to the failure of rigid equipment. At the first screening
level in EPRI (1991) several dissimilar types of components may have the same screening level.
In this case, the fragility analyst would group the components that are considered to have
correlated failure modes. The different groups of components screened at this level could then
be assumed to be independent (uncorrelated). Other screening levels based on different methods
for preliminary screening would be treated in a similar manner. If this information is provided to
the systems analyst early in the study, the systems analyst will then have a head start on how to
model screened out groups of components, accounting for expected correlation between groups.
They can also feed back some guidance to the fragility analyst as to what screening levels could
be tolerated without requiring plant specific fragilities.
It is convenient for the fragility analyst to screen components using the screening guidelines in
EPRI (1991). However, in several IPEEE submittals, it was evident that the screening level was
too low and the surrogate fragility ended up to be a significant contributor to CDF, if not the
most significant contributor. As discussed in Chapter 3, the selection of a screening threshold,
thus the associated surrogate fragility, requires some careful thought by the systems analyst as to
the correlation of components represented by the surrogate element or elements.
For Capability Category 3 in the Standard, high capacity components can be screened only if
they can be considered as fully independent of the remaining components. In the case of
Capability Category 3, the treatment of correlation is also much more complicated. All
components are to some extent partially correlated since they experience the same earthquake.
This raises the issue of partial correlation. Fleming (1999) suggests that for partial correlation
that fragilities be developed, where instead of the traditional C being broken down into R and
U , he suggests decomposing the randomness and uncertainty into IND and DEP . The systems
analyst and fragility analyst would have to caucus on how to handle partial correlation. This
4-7

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

report does not provide guidance on modeling of partial correlations. It does, however,
emphasize the importance of a close interface with the systems analysts on the issue.

4.3

Determine the Seismic Response of Structures

Seismic response of structures to the earthquake hazard specified for the site is an initial part of
the development of fragilities and screening levels of SSCs. The structural response is calculated
to develop loads on structural members and to develop amplified floor response spectra (AFRS)
to define the demand on systems and components in the structures. The Standard requires that
realistic seismic response be preformed. In general, if the ground motion spectra are
significantly different or if the structures are founded on soil, a new analysis is required.
Reference is made to ASCE (1998) for guidance on soil-structure interaction analysis and to
EPRI (1994) for determining median response and variability. Scaling of existing design
response must be justified based on demonstrating adequacy of the existing models, foundation
characteristics and similarity of input response spectra. Reference is made to EPRI (1991) for
scaling methodology. For Capability Category 3, scaling is not permitted and probabilistic
response analysis is required. Probabilistic response analysis may also be conducted for
Capability Categories 1 and 2 but is not a requirement. If probabilistic response analysis is
conducted, the Standard requires that that the number of simulations done using Latin Hypercube
or Monte Carlo techniques is large enough to obtain stable median and 85% non-exceedance
responses.
The minimum requirements for Capability Category 1 and 2 are to scale existing design response
analysis if justified, or to conduct a new deterministic analysis for the median ground motion
spectrum to be used in the PRA. For purposes of this report, UHS will be used to represent the
median ground motion spectrum determined appropriate for the PRA. For Capability Category 3,
a probabilistic analysis is always required and if there are two or more dominant seismic sources
as determined by the deaggregation of the probabilistically developed UHS, the probabilistic
response analysis may be required for two or more ground motion spectral shapes. A situation
such as this arose in a PRA of a plant in Eastern Europe where a probabilistic UHS was
developed for large magnitude earthquakes and used in a probabilistic response analysis in the
base PRA analysis. Another high frequency spectral shape from an actual close in lower
magnitude earthquake was then used in a subsequent analysis for comparison of responses. In
this case the second earthquake response was only used in a deterministic comparison of
response since a complete hazard study for local sources was not carried out. The close in
earthquake that had actually occurred resulted in higher structural response than the SSE in a few
locations but for most locations, the SSE response was greater as was the response to the 1E-4
median UHS.
Most often in past seismic PRAs, either a new deterministic analysis has been conducted or
existing design analyses have been scaled. In most cases of scaling in earlier PRAs, the assumed
median spectral shape was similar in broad band frequency content to design spectra and scaling
was approximately conducted on a mode by mode basis. All plants in the US were designed
before the regulatory requirements for soil structure interaction in USNRC (1989) were changed
to be more realistic. Some of the newer plants on soil sites used soil-structure interaction
computer codes in the response analysis but the previous restrictions on control point for the
input motion and limits on radiation damping resulted in very conservative response. Earlier
4-8

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

NPPs used simple soil-spring models that were not capable of capturing the complex behavior of
structural response on layered soil sites. For older NPPs on soil sites it is especially important
that the structural response be recomputed. Usually the existing structural model can be used but
the soil-structure interaction method and modeling must be updated to more modern methods.
EPRI (1991) and ASCE (1998) provide detailed discussions and up to date methodology for
deterministic soil-structure interaction analysis that meets all requirements of the Standard for
Capability Category 1 and 2 and no further guidance is necessary here.
4.3.1 Scaling of Existing Design Analysis
Scaling of existing analysis to develop realistic loads for a different ground motion input is much
easier than scaling existing in-structure spectra. The scaling of loads is discussed first. EPRI
(1991) provides equation 4-1 for mode by mode scaling of loads. If the mode by mode structural
member loads are available the scaled load for a new UHS ground motion spectrum can be
derived from:

Pi, j UHS = Pi, j SSE [Sa j UHS / Sa j SSE ]

Equation 4-1

Where Pi, j is the seismic load in element i for mode j, Sa j UHS is the spectral acceleration from
the UHS for mode j at UHS modal damping and Sa j SSE is the spectral acceleration for the SSE
for mode j at mode j modal damping. The element load PUHS is then the SRSS of PijUHS .
For simple structures with dominant response in a single mode, the scaling of total load in a
structural member can be approximately done using the applicable spectral accelerations and
damping for the dominant mode.
For simple lumped mass models if member loads are not provided on a mode by mode basis but
the nodal masses, mode shapes and participation factors are provided, the seismic force from
each node for each mode and be computed as:

Fi j = Mi j j Saj
Where FI, j = force at node i for mode j
Mi = mass of node i

j = participation factor for mode j


j = mode shape for mode j
Sa j = Spectral acceleration of mode j at frequency fj

4-9

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

The base shear, V, can be calculated as:


2
V = (j j Mi Sa j )

1/ 2

Scaling of spectra is more difficult. The only justification for scaling of spectra using a single
mode is if the spectral shape of the UHS is similar to that of the SSE, which is rarely the case. If
the complete eigensolution parameters are available, existing spectra can be scaled on a mode by
mode basis with reasonable accuracy. However, for soil sites, especially sites with older, more
simplified SSI analyses, scaling has little merit and new analysis is recommended.
Also, in some cases it has been found that, sample models of steel frame structures that respond
at very low frequencies do not have a design basis analysis that can be used for mode by mode
scaling. In these cases, 3D models may have been used but the design basis response analysis
was only carried out to 40 or so modes and the eigensolution was cut off at fairly low frequencies
like 6-8 Hz. Thus scaling of spectra using existing eigensolution parameters could not propagate
the high frequency portion of a typical EUS UHS.
Appendix A shows an example of scaling of DBE spectra developed from a simple lumped mass
reactor building model subjected to a RG 1.60 spectral shape anchored to 0.05g to develop
spectra for a high frequency UHS anchored to 0.1g. In this case, the original model was
recreated and the original spectra were verified before the scaling procedure was attempted.
Scaling was done by using original eigensolution results and random vibration theory as
described in Appendix B and by simplified methods using only the participation factors and
scaling at only a few frequencies. The simplified method might be applicable to Capability
Category 1 of the Standard. The more rigorous method using random vibration theory and the
original eigensolution is considered to be applicable to Capability Category 2 of the Standard.
The example conducted was done on Excel spread sheets. Since the model was simple, it was
not tedious to set up the spread sheets. For complex models it would only be practical to do this
if the eigensolution could be obtained in electronic form. Otherwise, the model should be rerun
using a time history that matches the UHS or alternatively using a direct generation code and a
new eigensolution from the existing model. The random vibration theory method in Appendix B
could, though, be applied to a reduced number of modes without sacrifice of much accuracy.
In Appendix B, a scaling example is presented for a more complex structure that accounts for
reduction in the ground motion spectra due to the ground motion incoherence associated with the
large structure. The high frequency portion of the ground motion UHS is also reduced in
accordance with EPRI (1997) to account for limited ductility of components in the structure.
This second reduction is applicable for mechanical components but wouldnt be applicable for
evaluation of relays.
4.3.2 Conducting New Analysis
It is often beneficial, and in many instances mandatory, to conduct new response analysis.
Existing results for rock sites may be scaled with reasonable accuracy as described in Section
4.3.1. In most cases, new analysis should be conducted for soil sites. Many of the design
response analyses for soil sites were conducted using very conservative criteria specified by the
regulators at the time. Often soil spring models were used with limits on damping. The control
4-10

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

point input motion was defined as the motion at the surface and deconvolution was not allowed.
In such cases, any form of scaling, especially if the UHS and DBE/SSE spectral shapes were
significantly different is very inaccurate.
The new analysis may be deterministic using best estimate properties for soil and damping and
taking credit for deconvolutions of the free field input motion with depth of embedment.
Guidance for new deterministic analysis, including variations in soil properties, is provided in
ANS (2002), EPRI (1991b) and ASCE (1998).
Probabilistic spectra for several seismic PRAs performed for IPEEE and for NPPs in other
countries with high seismic hazard have been developed using Latin Hypercube probabilistic
methods as described in USNRC (1981) for computing structural response. Although EPRI
(1994) mentions conducting probabilistic response analysis by Latin Hypercube or Monte Carlo
Methods, no specific references are provided for methodology or software for conducting such
analyses. In EPRI (1991) limited criteria are provided for scaling of loads and for scaling of
spectra. The emphasis in EPRI (1991) and EPRI (1994) is to conduct new deterministic analysis,
especially for soil sites, and to determine the variability in results using the classic separation of
variables approach.
Regardless of the method used, the objective in calculating structural response is to determine
the median response in terms of structural loads and amplified floor response spectra, and the
distribution about the median value.

4.4

Plant Walkdown

The walkdown methodology is well documented in EPRI (1991). This methodology was
referenced for IPEEE in USNRC (1991 b). The general sequence of performing a plant
walkdown contains the following steps:

1. Review the equipment list prepared by the systems analysts: The walkdown team should
review the list of components to understand what it to be included in the model. Often,
questions will arise as to the detail of the list. The internal events systems model may have
itemized items that can be grouped together by the rule of the box criteria. On the other
hand, items may have been too coarsely grouped and some should be separated. For
example, the jacket water heat exchanges and lube oil coolers for emergency diesel
generators may or may not be engine mounted. If they are on the engine, or engine skid, they
can be considered part of the engine assembly under the rule of the box criteria. If however,
they are in a separate substructure mounted on the floor or a wall, they should be listed as
separate components. Often, this decision cannot be made until some preliminary walkdown
is performed by the systems engineers in combination with the fragility engineers. The
interface between the systems engineers and fragility engineers is important at this stage to
assure an accurate list of equipment to be considered in the model.
2. Determine applicable screening levels: As discussed in Section 3.1.2, there may be several
screening levels that are applicable to different types of equipment. The most common
screening criteria applied during the walkdown are the screening tables in EPRI (1991b).
There are two levels of screen that apply to different types of components and structures.
4-11

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

These screening criteria are subject to verification of the inclusion rules, including elevation
above grade level, the adequacy of anchorage and the absence of systems interactions. Other
levels of screening may be developed based upon the seismic design or qualification criteria.
These levels of screen may be greater or less than the experience based screening levels in
EPRI (1991b). For instance, it may be determined that the damping value for equipment
qualified by analysis was very conservative, thus flexible components within given classes
would have large margin due to over prediction of response. This over design based on over
prediction of response would affect the seismic demand on the component itself and its
anchorage. If this screening level based on design criteria exceeds one or both of the
screening levels in EPRI (1991b) then it would be the basis for a higher capacity generic
fragility description to represent those components screened.
This type of prescreening in IPEEE studies has ranged from situations where practically all
components were prescreened based upon design conservatism to cases where most
components could only be screened during the walkdown at the lower screening level in
EPRI (1991b). Whatever is done for screening, a representative fragility must be convolved
with the hazard curve in order to assure that the calculated failure rate is lower than a target
set by the systems analysts.

3. Prepare walkdown sequence: This step involves coordination with plant operations to lay out
a schedule and route that minimize any effect on plant operations or maintenance.
Occasionally, it is necessary to plan non-routine inspections that require bus outages for
access. Often, more that one visit into the plant is necessary to access all equipment required
to be examined.
4. Conduct the walkdown: During the walkdown, SSCs are examined to apply the EPRI
screening, if applicable, verify that prescreening by other methods is valid, identify
vulnerabilities such as spatial systems interactions, fire sources, potential for actuation of fire
suppression systems, vulnerabilities to fire protection systems, verify anchorage, take field
dimensions if necessary, etc.
5. Documentation of walkdown. The walkdown and screening process and screening criteria
should be well documented. The walkdown documentation in effect is equivalent to a
calculation that documents the verification of capacities of equipment based on various
methods of field determination or predetermination of these capacities. EPRI (1991b)
provides excellent guidelines on the content of the documentation and provides screening
forms to document the screening level based on the EPRI screening criteria. The Standard
refers to this document for guidance on documentation. Additional documentation of
prescreening criteria and supporting calculations will also be necessary.

4.5

Structural Capacity

The Standard requires the identification of all relevant failure modes and the evaluation of
fragilities for critical failure modes. Depending upon the design documentation available, the
structural analyst may extrapolate an ultimate or limit state capacity from the design analysis. In
other cases, the analyst may need to conduct a new linear analysis or conduct some degree of
non-linear analysis. Usually, this task is achieved using results of linear analysis to determine
4-12

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

the demand level at the ultimate capacity for concrete or limit state for steel structures. EPRI
TR-103959 provides an example for determining ultimate capacity of a typical shear wall
structure. The Standard makes reference to EPRI NP-6041 and ASCE-4 for determining
ultimate capacity or limit state capacity. The key issue in determining the capacity is to focus on
the global instability, rather than the local instability of a single element. In typical design
calculations, the focus is on element capacity. To this extent, the design calculations must often
be extended to determine the redundant load paths to the point of global instability. This is often
done by using simplified non-linear models. Chapter 3 of EPRI TR-103959 provides a
simplified non-linear analysis method based on the use of effective frequency and effective
damping. Appendix G of EPRI NP-6041 provides guidance on the use and limits of equivalent
viscous damping in non-linear dynamic analysis.
On rare occasion, a more detailed non-linear dynamic analysis may be necessary to determine
the redundant load paths and inelastic deformation in a critical structure. The guidance afforded
in EPRI TR-103959 and EPRI NP-6041 is considered adequate for almost all non-linear analyses
that may be required to develop the capacity of structures.

4.6

Determine Ductility Beyond the Limit State Capacity

The requirements in the Standard are general on this topic. The capability of a structure to
cyclically deform beyond its limit state without failure depends upon the ductility of the
construction. Ductility refers to the amount of total deformation relative to the elastic
deformation. In fragility analysis, a ductility factor is derived such that the elastically calculated
demand from a dynamic analysis is divided by the ductility factor to determine an equivalent
spectral acceleration loading for purposes of determining stability. However, in developing
fragilities for structures, inelastic deformation less than that of the stability point is considered a
failure threshold due to the fact that equipment may be damaged before the structure totally
collapses.
EPRI NP-6041 provides conservative guidance on the ductility that can be considered when
determining a HCLPF. The elastically calculated demand is multiplied by a factor of 0.8 (equal
to a ductility factor of 1.25). This is considered conservative as long as the structure is ductile.
DOE Standard 1020 (DOE, 1994) provides ductility factors for common types of concrete and
steel structures. The recommended ductilities are associated with performance goals and are
generally considered to be a HCLPF value, about 95% non-exceedance values.
The median ductility factor is more judgmental since the level of distress less than collapse,
associated with affecting equipment within the structure, is not a precise value. Typically, a limit
on the deformation is defined by story drift. Story drift is utilized to determine the system
ductility. Recommended median story drift values and their random and uncertainty variability
for structures with and without safety related equipment attached are provided in Table 3-5 of
EPRI TR-103959. Median system ductility of multi-story buildings can be calculated from
equation 3-23 in EPRI TR-103959.
Newmark (1997) original proposed a formula for calculating the ductility factor for a simple
elastic-plastic model. Riddell and Newmark, refined the relationship between system ductility
and ductility factor as a function of structural damping and accounting for a bilinear stress strain
4-13

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

relationship. In USNRC (1994) this relationship was further refined to account for the shape of
the hysteresis loop and duration of the earthquake and utilizing an effective frequency and
effective damping in a simple non-linear model. EPRI TR-103959 summarizes two procedures
derived from these studies for determining a median value for the ductility factor. One
procedure incorporates the simplified non-linear model using effective frequency and effective
damping from USNRC (1994). The other procedure utilizes a modified Riddell-Newmark
ductility method. It is recommended that the average of the two methods be used to define the
median value of the ductility factor.
Additional randomness and uncertainty is to be added to that stated in EPRI TR-103959 for story
drift values for purposes of calculating the total R and U for ductility.
As a sanity check, the ductility factor HCLPF values calculated from the lognormal fragility
model can be compared to recommended ductility values in EPRI NP-6041 and DOE Standard
1020.

4.7

Structural Response Factor

For Capability Categories 1 and 2, the Standard states Estimate Seismic Response on a realistic
basis using site specific earthquake response spectra. Probabilistic Response is required for
Capability Category 3. The separation of variables procedure described in EPRI TR-103959 for
development of structural response factors is generally applied for Capability Category 1 and 2
applications although a probabilistic response analysis may be done. In the separation of
variables approach a Structural Response Factor, FSR, is developed that defines the conservatism
or unconservatism present in the derivation of seismic demand on structures or in the
development of floor response spectra.
The Structural Response Factor is applicable to both the structural fragility and the equipment
fragility. EPRI TR-103959 provides a detailed description of the development of the Structural
Response Factor. This factor may be different for the structures and for the equipment in the
structure due to the different analytical methods used to develop structural loads and floor
response spectra. For rock sites it is common practice to develop structural loads for design by
the response spectrum method whereas the floor response spectra are usually developed from a
mode superposition time history analysis. In some cases, floor response spectra are developed
by the direct generation method that is based on random vibration theory. In all cases, the
models should be the same. The differences result primarily from the spectral shape factor and
damping.
For soil sites, the derivation of a structural response factor from existing analysis involves
scaling of response as previously discussed. In general a new median centered analysis should
be conducted for soil sites, especially if the existing analysis was not based on state of the art soil
structure interaction (SSI) methodology. Typically the floor response spectra are developed
from a soil-structure interaction code such as SASSI or CLASSI. Often times, the structural
loads are developed from a more simplified soil structure interaction model. In these cases, soil
impedance functions are developed from an SSI code such as SASSI and are used to develop
equivalent soil spring and damper values. This approach may be a best estimate case considered
to be median centered but introduces additional uncertainty.
4-14

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

A brief discussion is provided for each of the important variables considered in the separation of
variables approach to developing a structural response factor. More detail on the numerical
values associated with each variable are provided in EPRI TR-103959, Chapter 3.
4.7.1 Spectral Shape Factor
The spectral shape factor is derived from the ground motion variables discussed in EPRI TR103959. They include the earthquake response spectrum shape used in design vs the UHS shape,
the relationship of the horizontal direction peak response to the average response and the
relationship of the vertical response to the horizontal response. If existing analysis is scaled, the
median values of the three variables are compared to the design values to obtain an overall factor
of conservatism or unconservatism. If new analysis is conducted, it should be median centered
using the median values of the three variables. In this case, the spectral shape factor should be
unity.
EPRI TR-103959 discusses time history simulation in Chapter 3. If time history analysis is
conducted, any difference between the average response spectrum resulting from the time history
and the target ground motion spectrum is usually incorporated into the spectral shape factor.
There is also uncertainty associated with the spectrum resulting from the time history vs the
target spectrum.
Note that if the seismic hazard can be defined in terms of spectral acceleration instead of the
traditional peak ground acceleration, the uncertainty in the spectral shape can be significantly
reduced. Refer to the discussion in Section 4.1 and in EPRI TR-103959, Chapter 3, for detail on
the uncertainty associated with defining the hazard as pga as opposed to spectral acceleration.
This uncertainty in spectral shape is incorporated into the overall variability of the spectral shape
factor.
In developing the spectral shape factor for equipment that is analyzed using the floor response
spectra as the demand, additional variability exists due to the use of broadened and smoothed
spectra. If raw spectra are not available and the strength factor for equipment is evaluated using
the broadened and smoothed floor response spectra, then additional conservatism should be
accounted for in the spectral shape factor applicable to equipment. Note that for soil sites, the
floor response spectra are usually a broadened and smoothed envelope of the response of the
structure for three soil stiffness cases. Thus, the fragility analyst should not double count the
uncertainty associated with the soil stiffness variation in developing the spectral shape factor and
the factor for Soil Structure Interaction.
4.7.2 Damping
The damping factor is the ratio of response of the structure for design damping vs the response
for median damping. If a new analysis is conducted, median damping should be utilized, in
which case the factor should be unity. Note that the median value of damping for development
of floor response spectra for equipment may be different than for defining structural failure.
Modern reinforced concrete structures are quite robust and if their capacity is greater than the
capacity of some critical equipment that ultimately governs the seismic risk, the median damping
4-15

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

value associated with large inelastic deformation of the structure may be too high to represent
median elastic response of the structure if some critical equipment failures occur while the
structure is elastic. The fragility analyst needs to make final adjustments to the damping factor
to reflect the appropriate value associated with the failure level of the equipment under
consideration. In theory, damping appropriate for median structural response would vary with
the capacity of the equipment. In practice, a damping value associated with the yield level of the
structure is usually chosen as a constant value. In some plants where the concrete structures are
much stronger than the equipment, a damping value associated with about of yield is more
appropriate.
4.7.3 Modeling
Structural modeling is usually considered to be a best estimate. But in some earlier analyses, the
models were very simple and torsional coupling effects were not adequately modeled. The
fragility analyst must make a judgment as to the adequacy of the model. If the model is in
question, it should be redone or modified to an extent that the fragility analyst considers it to be a
best estimate representation. The current trend is to make very detailed finite element models as
opposed to the more simplified stick models used in earlier design analyses. With modern
computers, the more detailed models can be utilized, however, using an overly complex model
does not necessarily guarantee better results.
The UHS provided in EPRI (1989) and USNRC (1994) for IPEEE for central and eastern US
sites were characterized by ground motion spectra with their peaks at high frequency. Some of
the earlier models of US NPP structures were simplified stick models and in cases of lower
frequency steel structures, the models would not properly capture the response at high
frequencies. In these cases, the models may be acceptable for developing structural loads,
governed by the low frequency portion of the UHS, but would not be appropriate for developing
floor response spectra to define demand for higher frequency equipment with sensitive devices
such a relays. The fragility analyst always has to deal with making best estimate judgments and
the uncertainty in these judgments. However, if a Capability Category 2 analysis is to be
conducted, it is likely to require some improvements in structural models of older NPPs.
Uncertainty in the response of structures due to modeling results from uncertainty in frequency,
uncertainty in mode shapes and for some models, torsional coupling. EPRI TR-103959 provides
guidance on the uncertainty resulting from these three variables that constitute the variability
resulting from modeling.
4.7.4 Mode Combination
This variable is applicable to all response spectra and mode superposition time history analyses.
Typically, modal responses have been combined by the SRSS rule with modifications for closely
spaced modes. This is considered to be median centered. If the results of an existing analysis
are being used and modes were combined by other methods, then the fragility analyst needs to
make an assessment of any conservatism that may result from such a combination. EPRI TR103959 provides guidance on the range of variability associated with simple two mode models to
multi-mode models.
4-16

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

4.7.5 Earthquake Component Combination


Current design criteria in RG 1.92 (USNRC, 1978) require that the responses to three orthogonal
directions of input be combined by the SRSS rule. This is interpreted to be the combination of
the end item of interest by SRSS such as stress, bolt force, etc. Often in design analyses, the
analyst has combined three orthogonal forces or accelerations by SRSS to result in a vector,
which implies that the three components are in phase. This is conservative and is not consistent
with the intent of RG 1.92. SRSS combination of the end item of interest is considered to be
median centered. Alternatively, Newmark (USNRC, 1977) recommends a simpler 100%, 40%,
40% rule as being median centered. In this combination, 100% of the dominant direction load or
stress is combined arithmetically with 40% of the load or response in the other two orthogonal
directions. The two criteria result in close to the same answer. The advantage of the 100-40-40
rule is that accelerations, loads or resulting stresses can be combined using the rule without
altering the final end item of interest. Many fragility analysts use the 100-40-40 rule in rederiving median centered responses.
Earlier NPPs often combined the worst direction horizontal response by absolute sum with the
vertical response. Depending upon the shape of the structure, this could be conservative or nonconservative relative to SRSS or the 100-40-40 rule. In these cases, use of existing analysis
results requires the fragility analyst to make an assessment of the degree of conservatism or nonconservatism to determine the earthquake component combination factor.
Having all three components in phase is considered to be an extremely low probability event and
is typically considered to be a 3 log standard deviation condition. Consequently, the random
variability associated with combination of earthquake components is usually small.
As noted in EPRI TR-103959, the earthquake component combination factor and variability
should not be double counted when deriving fragility of equipment. It is usually examined in
developing the equipment response factors and is deleted from the structural response factor as it
is applied to developing equipment fragility.
4.7.6 Foundation-Structure Interaction
This general variable, like the modeling variable, consists of several parts; soil structure
interaction (SSI) modeling, vertical spatial variation of ground motion (deconvolution with
depth) and ground motion incoherence (GMI). GMI reflects the fact that high frequency seismic
waves cancel each other across the foundation/soil interface, thus do not excite the foundation
uniformly along its entire surface. GMI only effects the high frequency response of structures,
thus is only applicable to stiff structures on rock foundations. The vertical spatial variation of
ground motion applicable for soil sites is a measure of the reduction of the effective input motion
defined at the surface to a structure foundation below the ground surface.
The variability of response due to foundation-structure interaction is really part of the overall
structural modeling but is typically evaluated separately in SPRA fragility development. Using
the separation of variables approach is perfectly valid, however, the fragility analyst must be sure
that the uncertainties associated with frequency and mode shape in the modeling factor are not
dominated by SSI effects and then counted again in developing uncertainty in SSI. The
4-17

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Standard, EPRI TR-103959 and EPRI NP-6041 all strongly recommend that a new SSI analysis
be conducted if the UHS is significantly different in shape than the design spectrum or if the SSI
modeling is not state of the art. The details of deconvolution of surface motion to the structure
foundation and the modeling of the soil and foundation interface is adequately covered on ASCE
4 and EPRI NP-6041.
In the Section 3.1.3 discussion of the NRC comments on IPEEE results, it was noted by the NRC
that the deconvolution analysis often produced higher reduction in free field motion from the
ground surface to the base mat than would be permissible in the design process. The details of
this comment are unknown or whether it applied to SPRA analysis or SMA analysis. Design
analyses prior to 1989, which is the case for all US NPPs, were severely restricted on
deconvolution. The Standard Review Plan (USNRC, 1989) now allows a more realistic
deconvolution, and restricts the reduction of input motion at the foundation to 60% of the free
surface motion. EPRI TR-103959 suggests that if the deconvolution analysis predicts a larger
reduction, it should not be restricted when evaluating median response for an SPRA. The
guidance in ASCE-4 and EPRI NP-6041 should be followed for determining the degree of
reduction of free surface motion at the foundation. Restrictions on the degree of reduction
should not be applied for SPRA median response analysis.
4.7.7 High Frequency Effect
In EPRI TR 102470 (EPRI, 1993), It is shown that even the smallest amount of ductility in
equipment will effectively reduce the response to high frequency input motion. Rather than
develop ductility factors for equipment that are variable with frequency, it was determined that
the high frequency portion of the ground motion spectrum can be reduced before generating floor
response spectra to account for this frequency dependent ductility factor. This is applicable to
most equipment but not necessarily for relays, consequently, it was not a common practice for
IPEEE. An example of a variation of this method has been conducted for a stiff structure
founded on rock. Appendix B shows how existing floor spectra were scaled downward in the
high frequency regime using the eigensolution of the structure and random vibration theory. In
this example, the existing floor response spectra were first scaled to reflect a reduction in the
high frequency portion of ground motion input due to Ground Motion Incoherence. In the next
step, the GMI reduced floor response spectra were again scaled to account for reduction in
response due to a minimum amount of ductility in equipment. In this case, the relay evaluations
were based on the first scaling for GMI and other equipment was evaluated based upon the
double scaling. If the eigensolution is available for the structural model used to develop floor
response spectra, this process can be carried out efficiently in a spread sheet format. Otherwise
new analyses using existing models may be an easier solution.

4.8

Probabilistic Response

Probabilistic response is required for Capability Category 3 SPRA. It may also be conducted for
Capability Categories 1 and 2 and was conducted for several IPEEE studies. When probabilistic
response analysis is conducted for the development of response spectra or structural loads, all
important variables that affect the structural response are included. Typically, probabilistic
analysis has been conducted using the SMACS computer code (USNRC, 1981). SMACS is
4-18

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

based on a Latin Hypercube stratified sampling simulation process that requires significantly
fewer simulations (about 30) than a Monte Carlo process. In this approach, the variables that
affect response are assumed to be lognormally distributed and the probability distribution of the
variables is broken up into equal parts, equal to the number of simulations. Combinations of
each variable are randomly selected for inclusion in an analysis. Once a value of a variable is
selected, it is not used again. In this manner it is assured that the 30 or so simulations include the
total distribution defined for each variable. The time history results for each simulation are used
to develop response spectra. Statistics are then applied to the resulting response spectra in order
th
to define median and 84 percentile response spectra.
The variation in spectral shape is simulated by utilizing 30 scaled natural and synthetic time
histories with median and 84th percentile response spectra ordinates that match the median and
84th percentile UHS ground motion spectra. In developing the time histories, the UHS may be
first modified to incorporate GMI effects and high frequency spectral reduction to account for
limited ductility of components. However, if relays are to be included in the evaluation, this
second reduction is best done for non relay evaluation on the final probabilistic floor response
spectra using methods in Appendix B. Otherwise, the 30 simulation analyses would have to be
conducted for two cases, relays and non-relays.
Other variables included in the probabilistic analysis are structural stiffness, structural damping,
soil stiffness and soil damping. USNRC (1981) provides the background for the SMACS
computer program and the typical ranges of the variables affecting structural response.
The original studies in the SSMRP program (USNRC, 1981) comparing probabilistic response
analysis using SMACS to design analysis conducted in accordance with regulatory requirements,
demonstrated very large margins in typical design analyses, particularly for soil sites. With
subsequent changes made to the Standard Review Plan (USNRC, 1989), much of the
conservatism was removed from the structural response analysis requirements. In FOAKE
(1993) the degree of conservatism remaining in the current USNRC design analysis practice was
examined. In that study, three typical NPP models were evaluated for varying foundation-soil
conditions. The models used in the study were a PWR Reactor Building, a BWR Reactor
Building and a PWR Auxiliary Building. The three models were evaluated for an assumed fixed
base rock site, a medium soil site (Vs = 1000 fps) and a soft soil site (Vs = 500 fps). The
structures were embedded for the soil cases plus a surface founded case was included for the
PWR Reactor Building. Spectra developed for design analysis in accordance with current
regulatory criteria were compared to spectra developed probabilistically. In this study, the 30
time histories used for the simulations had a distribution on the free field response spectra such
th
that the 84 percentile spectrum matched the RG 1.60 spectrum.
Resulting 84th percentile response spectra were compared to the design response spectra. The
following conclusions regarding the conservatism in design requirements were made.
1. The response spectra produced from design analysis were always conservative relative to 84th
percentile probabilistic spectra.
2. The minimum degree of conservatism was for the cases of structures founded on rock. The
factor of conservatism was 1.1 or greater. The greatest degree of conservatism was for the
soft soil cases with factors as high as 2.0.
4-19

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

3. The taller reactor building structures resulted in more conservatism than the shorter, more
squat, auxiliary building model.
4. The degree of conservatism increased with building elevation.
The conclusion from this study is that there may be a lot of conservatism in response of
structures founded on soft soil that is not captured in SPRA when so called median centered
deterministic analysis is conducted. In EPRI TR-103959 it is recognized that probabilistic
results typically produce lower response than so called median centered deterministic results. A
Demand Reduction Factor, DR, is suggested to have a median value of 0.92. This would
correspond closely to the conclusions from the study described above for rock sites but is likely
very conservative for soil sites.
In Reed et. al. (1994) a simplified approach to developing probabilistic spectra is proposed that
required only two independent structural response analyses with two different damping levels
instead of multiple simulations. Resulting mean in-structure spectra from the simplified
approach are shown to be very close to results from a Latin Hypercube simulation. The model
used in the study was a simplified six mass fixed base model and most of the response is in a
single fundamental frequency mode. The simplified probabilistic approach results in a
significant reduction of the peak of the in-structure spectra as compare to a single deterministic
median centered analysis. This approach appears to develop more realistic in-structure spectra
for SPRA than a deterministic best estimate approach. It is not clear how the results would
compare for a soil site since the reduction in response is more complex than in a fixed base
model with increased damping. As noted In the FOAKE study above, for soft soil sites, the
difference between probabilistic spectra and spectra from a so-called median centered
deterministic analysis is quite large and can be as much as a factor of 2 at high elevations in tall
structures.
The conclusion at this point is that conducting probabilistic response for soil sites can result in a
significant reduction in response. If there is a mix of structures, the reductions can vary
significantly. Consequently, if components that are significant contributors to the seismic risk
are located in these different structures, the difference in demand between probabilistic spectra
and deterministic spectra could alter the conclusions as to the more significant contributors to
risk. The computed CDF would also be quite different. Rock sites would not be nearly as
sensitive and the standard method of scaling design results or conducting a simplified
probabilistic response as described in Reed et al. (1994) should not pose the question of whether
structural response analytical methods can alter the ranking of dominant contributors.

4.9

Equipment Response and Capacity

For Capability Category 1, the standard allows generic data to be used if it is demonstrated to be
conservative. Capability Category 2 required that fragilities be based upon plant specific data
and allows generic data to be used for screening if conservative. Capability Category 3 requires
that fragilities be developed from plant specific data and to ASSURE that they are realistic. It
also required that fragilities be developed as a function of local response and to DERIVE the
joint probability distribution of the seismic capacities of different components.

4-20

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

EPRI (1994) provides the methodology for developing the capacity and response of equipment
designed by analysis or qualified by test. Methodology is also provided for deriving fragilities
from Generic Equipment Ruggedness Spectra (GERS). The methodology provided complies
with the requirements of the Standard for Capability Categories 1 and 2 and is generally
applicable to Capability Category 3 as long as the seismic hazard is defined at the
equipment/structure interface rather than at the ground.
Development of equipment and subsystem fragilities from different sources of information can
take on many forms. It is not practical to develop specific fragilities for all components to be
addressed in the risk model. Therefore, a sequence of screening out high ruggedness
components, derivation of generic fragilities for different classes or types of components and use
of plant specific design documents will optimize the process and assure that screening and
generic fragilities will have minimal impact on the risk results.
As discussed in Section 3.1.2 and 3.2, the systems analyst should set targets for fragilities of
elements that can be screened out or that will be represented by surrogate elements. This is an
important initial step so that the fragility analyst does not assume that screening at some
moderate level defined by the plant design requirements, EPRI screening tables (EPRI, 1991) or
the SQUG Reference Spectrum (SQUG, 1991) will be acceptable in the final risk assessment.
Often these levels of screening are adequate. However, before the fragility analyst begins the
sequential process of screening and development of plant specific fragilities, some targets should
be set by the systems analyst.
4.9.1 Initial Prescreening Using Licensing Criteria
The general approach to meeting the requirements for all three levels is to make use of the design
and qualification information as much a possible. For non A-46 plants, an initial screening can
be done based upon the licensing criteria used for qualification of equipment and subsystems.
The industry standards and regulatory requirements for seismic qualification have variable safety
margin (Campbell, 1993). Generic fragilities can often be developed for different equipment and
subsystem types based upon the licensing basis design and qualification criteria. These generic
fragilities may demonstrate capacity above a predetermined screening target. If this initial
generic approach using design and qualification criteria does not produce fragilities that are
above a predetermined screening target, then the fragility analyst knows that some individual
fragilities for governing cases are required. The fragility analyst should first determine what
margin between design demand and design capacity is necessary to result in a fragility that meets
predetermined screening levels. Then a quick review of qualification information can target the
components that fall below the margin target and focus on these components for detailed fragility
development.
Typical components and subsystems that fit into this category are valves, cable raceways, piping,
HVAC ducting and many rigid components such as horizontal pumps and chillers. Though
these fragilities are termed generic, they are based on plant specific design data and are
considered to meet the intent of the Standard for development of fragilities from plant specific
data. Components that meet the screening target can be represented by super elements or
surrogate elements in the risk model. This initial screening/generic fragility approach is also
applicable to Capability Category 3 as long as the screening level is based upon local response
4-21

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

parameters, the components failure modes are considered to be independent and the screening
level is at a high enough level that the risk contribution from a generic fragility representing the
screened class is very small.
4.9.2 Prescreening Using Earthquake Experience Data
For A-46 plants, the documented capacity is usually for anchorage and the capacity of the
equipment itself can only be represented by seismic experience data. In these cases, there are
two possible capacities. One capacity is based on the anchorage capacity if it is less than the
capacity of the component as determined by the SQUG GIP Reference Spectrum. The other
capacity is based upon the comparison of in-structure spectra to the SQUG GIP Reference
Spectrum or the more updated SEQUAL spectra (SEQUAL, 2001) for individual equipment
classes. Because of the limited amount of high acceleration seismic experience data, the
experience spectra such as the SQUG GIP Reference Spectrum or SEQUAL spectra are limited.
Consequently, the screening threshold, thus the generic fragility that represents the screening
threshold, is limited. This limit, if represented by surrogate elements in a SPRA, may be too low
and the resulting risk contribution too high. One of the initial steps in a SPRA is to determine if
various screening thresholds, such as the SQUG GIP Reference Spectrum will result in a low
enough failure rate that this form of screening is acceptable. For most central and eastern US
NPPs, the SQUG GIP screening should be sufficient to assure that seismic experience based
fragilities, when included in the risk model, will not dominate seismic risk.
EPRI TR-103959 does not address the development of fragility from seismic experience data. In
previous SPRAs, the SQUG Reference Spectrum has been assumed to represent a HCLPF
capacity, considered to be a 95% confidence level of capacity. Salmon and Kennedy (1994)
postulated a probabilistic methodology based on the SQUG Reference Spectrum representing a
minimum resistance level analogous to a testing level. The underlying assumptions in the
derivation are:
1. There are 50 independent samples in the database for each equipment class.
2. There are no incidences of failure in the database for components that meet the caveats of
SQUG (1991).
3. The database equipment was subjected to a wide range of ground motions centered about the
SQUG Reference Spectrum. The demand distribution about the Reference Spectrum is
assumed to be lognormal with a D of 0.3.
4. A total C of 0.45 is a reasonable value of the total demand and capacity uncertainty.
5. The target confidence level of the median equipment capacity was assumed to be 95%.
A Latin Hypercube simulation was conducted using 50 simulations and a capacity factor of 2.35
was derived. Thus, the median capacity of a component that meets the SQUG GIP caveats was
estimated to be about 2.35 times the SQUG Reference Spectrum.

4-22

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

In SEQAL (2001) a similar approach is taken for deriving a fragility from seismic experience
data. In this approach, the SQUG Reference Spectrum is considered to be a median demand
spectrum with a lognormal distribution. Using survival statistics, where a sample size, n, has no
failures and a sample size of n+1 has one failure, the 95% confidence capacity and median
capacity are derived as a function of the number of independent samples in the database. A
capacity factor of 2.2 is shown for a sample size of n = 30 as opposed to 2.35 derived by Salmon
and Kennedy (1994) by simulation for a sample size of n = 50. For comparison, the SEQUAL
survival statistics derivation method would yield a capacity factor of 2.42 for 50 samples. Both
methods are based on assumptions regarding the distribution of the demand about the Reference
Spectrum and the distribution of capacity about the median value. There are some slight
differences in the assumptions regarding the logarithmic standard deviation of the capacity
(0.45 for the simulation and 0.4 for the survival statistics analysis) which accounts for some of
the small difference in results for the two methods.
Appendix C contains the survival statistics derivation of the capacity factor to define the median
capacity above the SQUG Reference Spectrum and knockdown factor for determining a HCLPF
(95% confidence) capacity for several sample sizes ranging from n equal 15 to n equal 60. It is
show that in order to consider that the SQUG Reference Spectrum is a 95% confidence capacity
(HCLPF capacity), at least 30 samples are necessary. Knockdown factors are tabulated as a
function of sample size. In Appendix D, an example problem is presented where an equipment
class Reference Spectrum in SEQUAL for an Instrument Cabinet is used in derivation of a
fragility in an eastern US plant subjected to the EPRI UHS (EPRI, 1989). In this example
problem, the sample size used in determining the weighted Reference Spectrum is 46
components, resulting in a HCLPF capacity greater than the Equipment Class Reference
Spectrum. SEQUAL recommends that the SQUG Reference Spectrum be used for all equipment
classes which has been a typical assumption in prior SPRAs. However, in the example problem,
the actual weighted spectrum and actual number of samples are used in conjunction with
Appendix C survival statistics to demonstrate the process of developing a fragility from seismic
experience data.
SPRAs include many non-safety components that are not qualified for seismic events. The
experience based screening can also be applied to non qualified components, subject to
performing the walkdown and verifying that the non-safety components meet the GIP caveats
and the anchorage capacity envelops the seismic experience based capacity.
4.9.3 Prescreening using the EPRI SMA Screening Tables
In EPRI (1991b), screening criteria for establishing HCLPF are provided. There are two levels
of screening based on ground motion spectral acceleration, 0.8g and 1.2g. A specific ground
motion response spectrum is not provided but it is implied to be a NUREG/CR-0098 median
spectral shape. It is also generally assumed that since the screening level spectral acceleration is
more damaging at lower frequencies, that it is applicable to higher frequency range as well.
Typical central and eastern US UHS in EPRI (1991) and USNRC (1994) peak at about 25 Hz
and may exhibit higher spectral acceleration at 25 Hz than the NUREG/CR-0098 median
spectrum or the SQUG Bounding Spectrum anchored to 0.8g spectral acceleration. The SQUG
bounding spectrum is the SQUG Reference Spectrum reduced by a factor of 1.5 to be used in
comparing ground motion spectra rather than in-structure spectra. In general the interpretation
4-23

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

that the screening spectral acceleration is applicable to equipment mounted in structures


regardless of the frequency of the structure is reasonable as long as the screening is not applied
to cabinets and panels containing devices that are mounted in very stiff structures.
EPRI (1994) provides a simple example of the derivation of a fragility for equipment screened
using the EPRI screening tables. In this example, it is assumed that the equipment median
capacity is twice the screening level. This is generally more conservative than the above
example of screening using the SQUG Reference Spectrum where it is shown that as few as 30
samples results in a median capacity of 2.2 times the Reference Spectrum. However, since the
EPRI screening criteria are generally more liberal than the SQUG GIP, the assumption of a
median capacity factor of 2.0 as opposed to a higher factor that could be derived, as in
Appendix C from the actual number of samples, is a reasonable comprise. Fragilities derived
from the EPRI screening tables would be applicable to Capability Category 1 of the Standard and
could be used in Capability Category 2 studies provided that it is shown that the resulting
unconditional failure rate of the screened component is above some target set to assure very
small contribution to the overall risk results. The Screening Tables would not be applicable to
Capability Category 3.
4.9.4 Development of Fragilities Using Plant Specific Data
The overall methodology and examples in EPRI (1994) are considered to be applicable to
Capability Categories 1 and 2 of the Standard and if plant specific data is used in conjunction
with probabilistic spectra, the methodology is applicable to Capability Category 3. There are
always, of course, unique cases not specifically addressed.
One such case is the development of fragility for components that fail in a functional mode
where the qualification is by analysis. Review of design reports for equipment such as fans has
shown that often, the designer does not consider a minimum safety factor for deflection of the
fan relative to the housing. Thus, calculating that there is still some minimal amount of
clearance during the SSE does not provide the same safety margin as say a component support
that is designed not to yield. This same situation has been observed in design reports where it is
shown that a systems interaction due to impact will not occur but there is no consideration of a
safety factor on the impact. In general, if impact is considered to be failure, analogous to
reaching the ultimate strength in a component support, then there should be an analogous margin
built into the design process, such as a factor of 1.4 or 1.5. This is not always the case. In these
situations, the fragility analyst must first determine the consequences of impact or the real
margin beyond impact. This will determine the capacity factor relative to the demand. The rest
of the fragility calculation would follow the general guidance in EPRI TR-103959. The real
capacity is, in most cases, a very subjective judgments but must be addressed. If the impact is
considered to be failure and there is very little margin between the capacity and demand, the
resulting fragility may show up as a dominant contributor to seismic risk.

4-24

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21
Table 4-1
Correlation Of Fragility Development Elements And Requirements Of Ans Standard For External Events

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Understand the
development of the seismic
hazard

Defines frequency of
occurrence of hazard
parameter, location of hazard
(bed rock/rock
outcrop/surface), ground
motion spectral shape and
uncertainties.

Methodology is well defined in


RG 1.165 and draft ANS
Standard , SSHAC report.
Uniform hazard spectra should
be defined below 1 Hz to
determine maximum amplified
ground motion displacement.
Very flexible systems require
evaluation for displacement.
Hazard provided for IPEEE
was cut off below 1 Hz.
Guidance is needed on what
frequency to assign to pga.
This affects the shape of the
UHS beyond 25 Hz.
Relationship between
horizontal and vertical spectra
needed. Clarification on
whether hazard is average of
two horizontal components or
maximum of two horizontal
components is needed.

ANS standard correlates 4


SSHAC report categories to 3
capability categories in
standard. ANS standards 2.27
and 2.29 under development
for seismic hazard.

ANS standard correlates 4


SSHAC report categories to 3
capability categories in
standard. ANS standards 2.27
and 2.29 under development
for seismic hazard.

ANS standard correlates 4


SSHAC report categories to 3
capability categories in
standard. ANS standards 2.27
and 2.29 under development
for seismic hazard.

Determine hazard parameter


to reference fragility to
(peak ground acceleration,
spectral acceleration)

Spectral acceleration is
preferred but requires extra
work for hazard analyst.
Usually pga is provided which
results in greater uncertainty in
the development of the
structural response portion of
the structure or component
fragility.

Methodology is well defined in


NUREG 1.165, draft ANS
standards and SSHAC report
for developing hazard in terms
of spectral acceleration or pga.
Discussion on interface with
hazard analyst is provided.

Capability Categories 1, 2 and


3 recommend the use of
spectral acceleration but
states that the use of pga is
acceptable.

Capability Categories 1, 2 and


3 recommend the use of
spectral acceleration but
states that the use of pga is
acceptable.

Capability Categories 1, 2 and


3 recommend the use of
spectral acceleration but
states that the use of pga is
acceptable.

Determine Seismic
Response of Structures

Estimating structural response


is necessary to determine
loads in structural members
and to develop floor response
spectra as input to equipment.

HLR-FR-C1-C6 requires
realistic seismic response

HLR-FR-C1-C6 requires
realistic seismic response

HLR-FR-C1-C6 requires
realistic seismic response

4-25

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element
Develop Structural Loads

Loads for PRA hazard need to


be developed for evaluation of
structures.

Scale Existing Loads

Loads may be approximately


scaled by comparison of
design basis spectra and UHS
spectra.

Single Mode Scaling

Multiple Mode Scaling

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

HLR-FR-C1-C6 requires
realistic seismic response

HLR-FR-C1-C6 requires
realistic seismic response

HLR-FR-C1-C6 requires
realistic seismic response

Some guidance is provided in


EPRI NP-6041 for single mode
scaling. Additional discussion
on applicability of scaling for
loads and for scaling of
spectra would be useful. More
detail on scaling methods is
provided in Appendices A and
B.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

Scaling is not acceptable.


HLR-FR-C2 requires
probabilistic analysis for
Capacity Category 3.

Scaling is done at a single


dominant frequency of the
structure. Only applicable for
similarly shaped spectra and
for rock sites.

Guidance exists in EPRI NP


6041 for single mode scaling
and restrictions on single
mode scaling. Guidance and
restrictions are adequate for
single mode scaling

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

Scaling is not acceptable.


HLR-FR-C2 requires
probabilistic analysis for
Capacity Category 3.

A more accurate method of


scaling by utilizing mode
shapes and participation
factors from design analysis
and scaling on a mode by
mode basis. May not be
practical for complex models,
depending on availability of
eigen solution parameters.
Not addressed in EPRI NP6041 but commonly done in
prior PRAs.

No guidance provided.
Guidance is provided in
Appendices A and B.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

Scaling is not acceptable.


HLR-FR-C2 requires
probabilistic analysis for
Capacity Category 3.

General guidance provided in


EPRI NP-6041, EPRI TR103959 and reference to
ASCE 4. General guidance
considered adequate. Limit on
deconvolution are in SRP.

HLR-FR-C1-C6 requires
realistic seismic response.
General requirements are
outlined in HLR-FR-C1 to C6.

HLR-FR-C1-C6 requires
realistic seismic response.
General requirements are
outlined in HLR-FR-C1 to C6.

Deterministic analysis is not


allowed for Capability
Category III. HLR- FR-C2 has
requirements for probabilistic
analysis

New Deterministic Analysis for Recommended in EPRI NPUHS


6041 for soil sites and for
cases where spectra are not
similar in shape.

4-26

Applicable
Methodology

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology
Monte Carlo and Latin
hypercube methods discussed
in Chapter 4 of EPRI
TR103959 but no specific
reference to SMACS
developed in SSMRP
program. Additional guidance
is provided. References to
LLNL SSMRP methods using
SMACS and Reed/Kennedy
et. al. paper on Approximate
Probabilistic response
methods are discussed.

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

HLR- FR-C2 has requirements


for probabilistic analysis.
Applications Guide needs to
provide more detail than what
is in EPRI TR 103959.

HLR- FR-C2 has requirements


for probabilistic analysis.
Applications Guide needs to
provide more detail than what
is in EPRI TR 103959.

HLR- FR-C2 has requirements


for probabilistic analysis.
Applications Guide needs to
provide more detail than what
is in EPRI TR 103959.

HLR-FR-C1-C6 requires
realistic seismic response.
General requirements are
outlined in HLR-FR-C1 to C6.

HLR-FR-C1-C6 requires
realistic seismic response.
General requirements are
outlined in HLR-FR-C1 to C6.

Deterministic analysis is not


allowed for Capability
Category III. HLR- FR-C2 has
requirements for probabilistic
analysis

New Probabilistic Analysis for


UHS

Not commonly done but


results in much better
definition of median response
and uncertainty in structural
response.

Develop Floor Response


Spectra

Required for development of


fragilities for equipment.

Scale Existing Spectra

Difficult if design and UHS


spectral shapes are not
similar. For soils sites, scaling
is usually inaccurate and in all
cases has more uncertainty.

Some guidance is provided in


EPRI NP-6041 for single mode
scaling. Additional discussion
on applicability of scaling for
loads and for scaling of
spectra is provided in
Appendices A and B.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

Scaling not acceptable. HLRFR-C2 requires probabilistic


analysis for Capacity Category
3.

Single mode scaling

Only reasonable if structure


responds primarily in a single
mode.

Guidance in EPRI Np-6041.


Limited application. Current
guidance and application limits
are considered adequate.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

Scaling not acceptable. ANS


requirement FR-C2 requires
probabilistic analysis for
Capacity Category 3.

4-27

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Correlation of ANS
Standard Capability
Category 3

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Guidance is provided in
Appendices A and B.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

HLR-FR-C3 requires
JUSTIFICATION of scaling.
HLR-FR-C4 would require new
analysis in most cases of soil
sites or sites with UHS
spectral shape significantly
different than design spectra.

Scaling not acceptable. ANS


requirement FR-C2 requires
probabilistic analysis for
Capacity Category 3.

New Deterministic Analysis for Recommended in EPRI NPUHS


6041 for soil sites and for
cases where spectra are not
similar in shape.

General guidance provided in


EPRI NP-6041, EPRI TR103959 and reference to
ASCE 4. General guidance
considered adequate. Limits
on deconvolution are in SRP.

HLR-FR-C1-C6 requires
realistic seismic response.
General requirements are
outlined in HLR-FR-C1 to C6.

HLR-FR-C1-C6 requires
realistic seismic response.
General requirements are
outlined in HLR-FR-C1 to C6.

Deterministic analysis is not


acceptable. HLR- FR-C2
requires probabilistic analysis
for Capacity Category 3.

New Probabilistic Analysis for


UHS

Monte Carlo and Latin


HLR-FR-C2 has requirements
hypercube methods discussed for probabilistic analysis.
in Chapter 4 of EPRI
TR103959 but no specific
reference to SMACS
developed in SSMRP
program. Additional guidance
is provided. References to
LLNL SSMRP methods using
SMACS and Reed/Kennedy
et. al. paper on Approximate
Probabilistic response method,
methods are discussed.

HLR-FR-C2 has requirements


for probabilistic analysis.

HLR-FR-C2 has requirements


for probabilistic analysis.

Fragility Development Description of Fragility


Elements
Development Element
Multiple Mode Scaling

4-28

Can be done using mode


shapes and participation
factors. Difficult for complex
structures and without
electronic files on mode
shapes and participation
factors. More approximate
methods at soil sites have
shown poor correlation to new
analyses.

Not commonly done but


results in much better
definition of median response
and uncertainty in structural
response.

Applicable
Methodology

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Determine screening level to


define minimum surrogate
element (Convolve
candidate surrogate
fragilities and hazard to
determine unconditional
probability of failure that is
an acceptable screening
level)

Surrogate fragilities are


substituted into the fault trees
to represent basic events that
are screened out on the basis
of meeting some minimum
capacity. A single surrogate
fragility is often used to
represent several basic events
of equal or greater capacity.
In IPEEE, the use of surrogate
fragilities often resulted in a
large portion of the computed
seismic risk being driven by
the low capacity fragility of the
surrogate element. Some
licensees used only one
surrogate element to represent
all screened out SSCs. This
assumes that all failures of the
screened out SSCs are
correlated. This can be
unconservative for cases
where several components
appear in series (OR Gates)
and have similar capacities.

Clear guidance from the


systems analyst to the fragility
analyst is needed to determine
the screening level required to
result in a surrogate fragility
that has very low contribution
to seismic risk. Per EPRI TR103959, a surrogate element
should be introduced in each
fault tree to represent basic
events in the fault tree logic
that have been screened out.
In some cases, more than one
surrogate in a fault tree is
appropriate, depending upon
correlation. Additional
guidance is provided in
Applications Guide.

The terminology "surrogate


element" is not specifically
addressed. Lumping of
groups of components into
supercomponents is
addressed in HLR-SA-A-3. In
HLR-SA-E-3, it is stated that
"a detailed set of criteria must
be developed and used to
assure that this screening
does not eliminate elements of
the model that should have
been retained. This implicitly
addresses the use of
surrogates but the emphasis is
on elimination of risk
contributing elements rather
than on using superelements
that have too low of a
capacity. HLR- FR-B1 states
that criteria in NP-6041 and
NUREG/CR-4334 may be
used to screen out high
capacity components but
cautions to CHOOSE the
screening level high enough
that the contribution to CDF
and LERF from the screened
out components is not
significant.

The terminology "surrogate


element" is not specifically
addressed. Lumping of
groups of components into
supercomponents is
addressed in HLR-SA-A-3. In
HLR-SA-E-3, it is stated that
"a detailed set of criteria must
be developed and used to
assure that this screening
does not eliminate elements of
the model that should have
been retained. This implicitly
addresses the use of
surrogates but the emphasis is
on elimination of risk
contributing elements rather
than on using superelements
that have too low of a
capacity. HLR- FR-B1 states
that criteria in NP-6041 and
NUREG/CR-4334 may be
used to screen out high
capacity components but
cautions to CHOOSE the
screening level high enough
that the contribution to CDF
and LERF from the screened
out components is not
significant.

HLR-FR-B1 states "SCREEN


high seismic capacity
components ONLY if the
components' failures can be
considered as fully
independent of the remaining
components." Current
guidance in EPRI documents
would not meet this
requirement.

Walkdown

Important for ruggedness


screening and identifying
vulnerabilities due to design
and construction errors and
systems interactions.

Requirements in ANS
Standard are met by guidance
in EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

4-29

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Review Equipment List

Necessary to prepare for


walkdown and set up interface
with systems engineers. List
may be revised after review to
incorporate rule of box or to
breakdown to subcomponents
as is appropriate for fragility
development.

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Determine applicable
screening levels. Determine if
generic fragilities based on
design requirements or EPRI
NP-6041 screening is
adequate to assure capacity of
surrogate element

Develop preliminary fragility


for screening levels and
convolve with hazard to
calculate unconditional failure
rate.

Addressed in EPRI TR103959, Chapter 5, but 1st


level screen in EPRI NP-6041
often results in a surrogate
fragility that is too low.
Additional guidance is
provided on applicability of
screening tables. Appendix E
provides example where EPRI
screening is not adequate and
an alternate screening
approach is taken.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Prepare Plant Walkdown


Sequence

Coordinate with plant


Adequately covered in EPRI
personnel to optimize time and NP-6041 which is standard
minimize exposure.
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Conduct walkdown screening

Determine components that


can be screened at levels
equal or exceeding surrogate
target. Identify vulnerabilities.

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Apply EPRI NP-6041


Screening Where Applicable

EPRI NP-6041 has two


screening levels. Screening
can only be done if level
meets surrogate target.

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Component Vulnerabilities

Identify vulnerable
components.

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

4-30

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Anchorage

Obtain information on
anchorage and screen if
anchorage is generically
rugged.

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Systems Interactions

Identify potential systems


interactions (falling, proximity,
SAM, spray, flood, systematic
interactions).

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Seismic/Fire Issues

Seismic induced fire,


inadvertent actuation of fire
suppression systems,
unavailability of fire
suppression.

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Documentation of Walkdown

Identification of personnel,
documentation of walkdown
screening criteria and
walkdown findings.

Adequately covered in EPRI


NP-6041 which is standard
referenced in EPRI TR103959 and NUREG-1407.
Additional documentation
guidance in NUREG-1407.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

General requirements in HLRFR-E1 through E-5.


Reference to guidance in
EPRI NP-6041.

Determine Structural
Capacity (Ultimate Load
Analysis)

Determine the point that the


calculated loads would result
in the structure being
unstable. Ductile behavior
may exist beyond this point.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

Scale Existing Linear Analysis

Easiest for screening. May


General guidance on scaling
not be sufficient for detailed
in EPRI NP-6041 and EPRI
fragility development for failure TR-103959.
state.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

4-31

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

New Linear Analysis

May be hand or computer


analysis to address
redistribution of loads of failed
members to point of instability.

Adequate guidance by
reference in EPRI NP-6041,
EPRI TR- 103959 and
ASCE 4.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

New Non-Linear Analysis

Conducted only for critical


cases to more accurately
determine failure point and
reduce uncertainty.

Some guidance in Chapter 3


of EPRI TR- 103959 using
effective frequency and
effective damping in simple
models. Further guidance in
Appendix G of EPRI NP-6041
on use of equivalent viscous
damping in non-linear
analysis. Guidance
considered adequate for
general case.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

ANS Standard HLR-FR-D1


and HLR-FR-D2 require the
identification of all relevant
failure modes and the
evaluation of fragilities for
critical failure modes.

Determine Ductility Beyond


Ultimate Load Capacity

General guidance given for


structures in EPRI and DOE
documents. Use of factors
requires judgment to preclude
any brittle weak link failures.
Generic ductility factors given
in performance goal based
DOE Standard 1020.

Guidance in Chapter 3 of
EPRI TR-103959 on effective
frequency and effective
damping, simplified non-linear
models and on Newmark
ductility method. Both are
recommended and average of
results is recommended as
median value of ductility,
Additional recommendations
on performance goal based
structural ductilities are
provided in DOE Standard
1020. The DOE material is
referenced in the Applications
Guide.

ANS Standard requirements


are general regarding
capacity and include only
small reference to ductility
such as drift. Current
guidance is adequate to meet
ANS requirements.

ANS Standard requirements


are general regarding
capacity and include only
small reference to ductility
such as drift. Current
guidance is adequate to meet
ANS requirements.

ANS Standard requirements


are general regarding
capacity and include only
small reference to ductility
such as drift. Current
guidance is adequate to meet
ANS requirements.

Final Structural Capacity,


Capacity Factor and
Variability

Separation of variables
Separation of variables
ANS standard has no specific
approach is used. This is
adequately addressed in EPRI requirements. Existing EPRI
TR 103959 is adequate.
more practical than a complete TR103959.
simulation process.

ANS standard has no specific


requirements. Existing EPRI
TR 103959 is adequate.

ANS standard has no specific


requirements. Existing EPRI
TR 103959 is adequate.

4-32

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element
Structural Response Factor
and Variability

Defines conservatism or
unconservatism in structural
response analysis and
confidence bounds on
analysis.

Spectral Shape (UHS vs.


Design)

Required if scaling of spectra


and loads is conducted.
Factor is unity if new median
centered analysis is done.

Damping (Median vs. Design)

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

HLR-FR-C1 states to
"Estimate seismic response on
a realistic basis using site
specific earthquake response
spectra."

HLR-FR-C1 states to
Probabilistic response
"Estimate seismic response on required.
a realistic basis using site
specific earthquake response
spectra."

Adequately covered in EPRI


TR-103959. The use of a
UHS is a given parameter to
the fragility analyst. Refer to
the discussion on seismic
hazard that addresses the
ANS Standard comment on
making sure that the spectral
shape is sufficiently rich in low
frequency.

Note to HLR-FR-C1 states that


"Any UHS SHOULD be used
cautiously making sure that
the spectral shape is
sufficiently rich in low
frequency. Refers to NOTE
HA-G1 for further discussion
on this topic. "

Note to HLR-FR-C1 states that


"Any UHS SHOULD be used
cautiously making sure that
the spectral shape is
sufficiently rich in low
frequency. Refers to NOTE
HA-G1 for further discussion
on this topic. "

Note to HLR-FR-C1 states that


"Any UHS SHOULD be used
cautiously making sure that
the spectral shape is
sufficiently rich in low
frequency. Refers to NOTE
HA-G1 for further discussion
on this topic. "

Required if scaling of design


response is done. New
analysis should be based on
median damping. The
resulting damping factor is
then unity.

Adequately covered in EPRI


TR-103959.

No specific requirements

No specific requirements

Probabilistic response
required. Variables are
included in simulation.

Modeling

Modeling uncertainty must be


estimated regardless if
existing design analysis is
utilized or new deterministic
analysis is done.

Adequately covered in EPRI


TR-103959.

No specific requirements

No specific requirements

Probabilistic response
required. Variables are
included in simulation.

Mode Shape

Uncertainty in deformed shape Adequately covered in EPRI


of modes.
TR-103959.

No specific requirements

No specific requirements

Probabilistic response
required. Variables are
included in simulation.

Frequency

Uncertainty in stiffness results Adequately covered in EPRI


in uncertainty in frequency and TR-103959.
response.

No specific requirements

No specific requirements

Probabilistic response
required. Variables are
included in simulation.

Method of Analysis

Uncertainty associated with


Usually incorporated into
analytical methods, particularly spectral shape factor.
in treatment of soil-structure
interaction.

No specific requirements.
No specific requirements.
No specific requirements.
Refers to ASCE 4 for guidance Refers to ASCE 4 for guidance Refers to ASCE 4 for guidance

4-33

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Response Spectrum (capacity


only)

Response spectrum analysis


commonly used for structural
loads. Results are
conservative since maximum
components of force are
reported instead of
coincidental forces. SSI is
simplified.

Analysis methods adequately


described in ASCE 4. EPRI
TR-103959 adequately
addresses uncertainties in
methods.

No specific requirements.
No specific requirements.
Probabilistic response
Refers to ASCE 4 for guidance Refers to ASCE 4 for guidance required. Variables are
included in simulation.

Time History (development of


response spectra or loads for
capacity)

Used primarily for


development of in-structure
spectra. Can be beneficial in
reducing uncertainty and
conservatism in calculation of
structural loads.

Analysis methods adequately


described in ASCE 4. EPRI
TR-103959 adequately
addresses uncertainties in
methods.

No specific requirements.
No specific requirements.
Probabilistic response
Refers to ASCE 4 for guidance Refers to ASCE 4 for guidance required. Variables are
included in simulation.

Mode Combination

Applies to response spectrum


analysis and mode
superposition time history
analysis.

Adequately addressed in EPRI No specific requirements.


TR-103959.

No specific requirements.

Probabilistic response
required. Variables are
included in simulation.

Earthquake Component
Combination

Industry standard of SRSS is


considered median centered.
Random variability depends
on structural geometry
(rectangular vs. circular) and
sensitivity to vertical
acceleration.

Adequately addressed in EPRI No specific requirements.


TR-103959.

No specific requirements.

Probabilistic response
required. Variables are
included in simulation.

4-34

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Foundation Structure
Interaction

Includes soil-structure
interaction vertical spatial
variation of ground motion and
ground motion incoherence.
Soil-structure interaction
parameters must be
addressed as part of modeling
(soil stiffness, soil damping,
deconvolution. New
deterministic analysis should
vary soil properties.
Requirements are in ANS
standard and described in
EPRI NP-6041 and in EPRI
TR-103959. New probabilistic
analysis should have realistic
ranges of variables.

Detailed deterministic
guidance provided in EPRI
NP-6041 and methods to
estimate uncertainty in
deterministic analyses are
provided in EPRI TR-103959.
Soil-structure interaction is
typically treated as a separate
variable but is really a part of
modeling. Emphasis in EPRI
documents is on new analyses
using modern methods. Most
existing NPPs have outdated
SSI models and scaling of
results is not recommended.
Ground motion incoherence is
usually only applicable to rock
foundations. It is considered
separately in this document
with an example spectra
reduction in Appendix B.

HLR-FR-C6 requires
variations in soil stiffness and
provides guidance on using
median soil properties defined
at strain levels corresponding
to input seismic motions that
dominate core damage.
Refers to ASCE 4 for guidance

HLR-FR-C6 requires
Probabilistic response
variations in soil stiffness and required. Variables are
included in simulation.
provides guidance on using
median soil properties defined
at strain levels corresponding
to input seismic motions that
dominate core damage.
Refers to ASCE 4 for guidance

Ground Motion incoherence

Can result in reduction of


effective input motion on large
stiff structures.

Addressed in EPRI TR103959.

No specific requirements.

No specific requirements.

Probabilistic response
required. Variables are
included in simulation.

Adequately addressed in EPRI No specific requirements.


TR-102470. Can be useful for
some sites with stiff structures
and high frequency input
motion. Example spectra
reduction given in Appendix B.

No specific requirements.

Probabilistic response is
required. Reduction in
ground motion spectrum iat
high frequency should be
incorporated into probabilistic
response analysis.

High Frequency Effects (Effect Can result in reduction in high


on Response Spectra)
frequency portion of instructure response spectra for
high frequency input motion.

4-35

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element
Determine Capacity of
Components

Develop the ultimate capacity


of component for function or
structural failure.

Develop Surrogate fragilities


based on screening levels.

Surrogate fragilities are based


on highest screening level
from walkdown or review of
equipment qualification.

4-36

Applicable
Methodology

Guidance in Chapter 5 of
EPRI TR-103959 on
development of surrogate
element from screening.
Generally more than one level
of surrogate element should
be developed depending on
what levels the components
can be screened, considering
component type, walkdown
screening and reviews of
equipment qualification
documentation screening is
discussed in Applications
Guide and examples are given
in Appendices D and E for
development of generic
fragilities from seismic
experience data and from
design criteria.

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

HLR-FR-F1 and F2 require


fragilities to be based on plant
specific data supplemented as
appropriate by earthquake
experience data, fragility test
data and generic qualification
test data. For all SSCs that
appear in the dominant
accident cutsets, ASSURE
that they have site-specific
fragility parameters which are
derived based on plantspecific information.
Exception: JUSTIFY the use
of generic fragility for any SSC
as being appropriate for the
plant.

HLR-FR-F1 and F2 require


fragilities to be based on plant
specific data supplemented as
appropriate by earthquake
experience data, fragility test
data and generic qualification
test data. For all SSCs that
appear in the dominant
accident cutsets, ASSURE
that they have site-specific
fragility parameters which are
derived based on plantspecific information.
Exception: JUSTIFY the use
of generic fragility for any SSC
as being appropriate for the
plant.

HLR-FR-F1 and F2 require


fragilities to be based on plant
specific data supplemented
as appropriate by earthquake
experience data, fragility test
data and generic qualification
test data. For all SSCs that
appear in the dominant
accident cutsets, ASSURE
that they have site-specific
fragility parameters which are
derived based on plantspecific information.
Exception: JUSTIFY the use
of generic fragility for any SSC
as being appropriate for the
plant.

HLR SA-A3 refers to grouping


of elements into a super
element. Surrogate fragility
can be a superelement. HLRSA-E3 cautions on elimination
of SSCs from the model on the
basis of screening. This
implies representation by a
surrogate fragility. HLR-FRB1 cautions against screening
at too low of a level that would
result in a significant
contribution to CDF or LERF
from the screened out
components.

HLR SA-A3 refers to grouping


of elements into a super
element. Surrogate fragility
can be a superelement. HLRSA-E3 cautions on elimination
of SSCs from the model on the
basis of screening. This
implies representation by a
surrogate fragility. HLR-FRB1 cautions against screening
at too low of a level that would
result in a significant
contribution to CDF or LERF
from the screened out
components.

HLR-FR-B1 states to
"SCREEN high-seismiccapacity components ONLY if
the components' failures can
be considered as fully
independent of the remaining
components.

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Walkdown Screening

Two screening levels in EPRI


NP-6041 can define surrogate
fragilities.

Guidance in Chapter 5 of
EPRI TR-103959 on
development of surrogate
element from screening.
Generally more than one level
of surrogate element should
be developed depending on
what levels the components
can be screened, considering
component type, walkdown
screening and reviews of
equipment qualification
documentation screening is
discussed in Applications
Guide and examples are given
in Appendices D and E for
development of generic
fragilities from seismic
experience data and from
design criteria

HLR SA-A3 refers to grouping


of elements into a super
element. Surrogate fragility
can be a superelement. HLRSA-E3 cautions on elimination
of SSCs from the model on the
basis of screening. This
implies representation by a
surrogate fragility. HLR-FRB1 cautions against screening
at too low of a level that would
result in a significant
contribution to CDF or LERF
from the screened out
components.

HLR SA-A3 refers to grouping


of elements into a super
element. Surrogate fragility
can be a superelement. HLRSA-E3 cautions on elimination
of SSCs from the model on the
basis of screening. This
implies representation by a
surrogate fragility. HLR-FRB1 cautions against screening
at too low of a level that would
result in a significant
contribution to CDF or LERF
from the screened out
components.

HLR-FR-B1 states to
"SCREEN high-seismiccapacity components ONLY if
the components' failures can
be considered as fully
independent of the remaining
components.

Screening from Review of


Qualification Reports

Review of equipment
qualification can result in
screening at levels above or
below the target surrogate
fragility.

Guidance in Chapter 5 of
EPRI TR-103959 on
development of surrogate
element from screening.
Generally more than one level
of surrogate element should
be developed depending on
what levels the components
can be screened, considering
component type, walkdown
screening and reviews of
equipment qualification
documentation screening is
discussed in Applications
Guide and examples are given
in Appendices D and E for
development of generic
fragilities from seismic
experience data and from
design criteria

HLR SA-A3 refers to grouping


of elements into a super
element. Surrogate fragility
can be a superelement. HLRSA-E3 cautions on elimination
of SSCs from the model on the
basis of screening. This
implies representation by a
surrogate fragility. HLR-FRB1 cautions against screening
at too low of a level that would
result in a significant
contribution to CDF or LERF
from the screened out
component.

HLR SA-A3 refers to grouping


of elements into a super
element. Surrogate fragility
can be a superelement. HLRSA-E3 cautions on elimination
of SSCs from the model on the
basis of screening. This
implies representation by a
surrogate fragility. HLR-FRB1 cautions against screening
at too low of a level that would
result in a significant
contribution to CDF or LERF
from the screened out
component.

HLR-FR-B1 states to
"SCREEN high-seismiccapacity components ONLY if
the components' failures can
be considered as fully
independent of the remaining
components.

4-37

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Develop Limit State Capacities Capacity is based on the


by Analysis
lowest capacity failure mode.

General guidance in EPRI TR103959 for common modes of


failure such as anchor bolts
and welds. Further guidance
would be useful but due to the
many failure modes and
different types of equipment, it
is not practical. General
guidance in EPRI TR-103959
is considered adequate for
most fragilities.

HLR-FR-A2 allows generic


data for development of
fragilities but must be
demonstrated to be
conservative.

HLR-FA-A2 requires that


fragilities be based on plant
specific data. Generic data
may be used for screening if
conservative.

HLR -FA-A2 requires that


fragilities be developed from
plant specific data and to
"ASSURE that they are
realistic." HLR-FR-F1 also
requires that fragilities be
developed as a function of
local response and to
"DERIVE the joint probability
distribution of the seismic
capacities of different
components"

Scale from Qualification


Analysis

Most common for plants with


well documented seismic
qualification. A-46 plants
generally only addressed
anchorage and the equipment
capacity is defined by the
SQUG screening criteria
(SQUG Bounding spectrum or
1.5 times the SQUG bounding
spectrum).

General guidance in EPRI TR103959 for common modes of


failure such as anchor bolts
and welds. Further guidance
would be useful but due to the
many failure modes and
different types of equipment, it
is not practical. General
guidance in EPRI TR-103959
is considered adequate for
most fragilities.

HLR-FR-A2 allows generic


data for development of
fragilities but must be
demonstrated to be
conservative.

HLR-FA-A2 requires that


fragilities be based on plant
specific data. Generic data
may be used for screening if
conservative.

HLR -FA-A2 requires that


fragilities be developed from
plant specific data and to
"ASSURE that they are
realistic." HLR-FR-F1 also
requires that fragilities be
developed as a function of
local response and to
"DERIVE the joint probability
distribution of the seismic
capacities of different
components"

Conduct New Analysis

Sometimes required for


equipment with incomplete or
no analysis. The amount of
analysis depends upon the
relationship of the target
surrogate fragility and the
screening levels achieved.

General guidance in EPRI TR103959 for common modes of


failure such as anchor bolts
and welds. Further guidance
would be useful but due to the
many failure modes and
different types of equipment, it
is not practical. General
guidance in EPRI TR-103959
is considered adequate for
most fragilities.

HLR-FR-A2 allows generic


data for development of
fragilities but must be
demonstrated to be
conservative.

HLR-FA-A2 requires that


fragilities be based on plant
specific data. Generic data
may be used for screening if
conservative.

HLR -FA-A2 requires that


fragilities be developed from
plant specific data and to
"ASSURE that they are
realistic." HLR-FR-F1 also
requires that fragilities be
developed as a function of
local response and to
"DERIVE the joint probability
distribution of the seismic
capacities of different
components"

4-38

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Develop Ductility if Applicable

Generally applicable to flexible


equipment with structural
failure as the governing failure
mode.

Limited guidance provided in


EPRI TR-103959. EPRI NP6041 provides limited
guidance on ductility
associated with a HCLPF.
Guidance on effective
frequency and effective
damping for non-linear
response in EPRI TR-103959
and NUREG/CR-3805.
Further guidance beyond
general guidance given is
difficult due to the numerous
configurations of equipment.

No requirement on equipment
ductility.

No requirement on equipment
ductility.

No requirement on equipment
ductility.

Develop Fragilities from Test


Data

Qualification is commonly
done by shake table testing to
required response spectra.

Guidance provided in EPRI


TR-103959. Guidance is
result of subjective opinion of
testing community. Guidance
considered adequate.

No specific requirements on
developing fragility from test
data.

No specific requirements on
developing fragility from test
data.

No specific requirements on
developing fragility from test
data.

Specific Qualification Tests

Single test usually has limited


intended safety margin, and
one test is statistically
insignificant, therefore HCLPF
is estimated to be less than
test level.

Guidance provided in EPRI


TR-103959. Guidance is
result of subjective opinion of
testing community. Guidance
considered adequate.

No specific requirements on
developing fragility from test
data.

No specific requirements on
developing fragility from test
data.

No specific requirements on
developing fragility from test
data.

Generic Equipment
GERS are available for vendor Guidance provided in EPRI
Ruggedness Spectra (GERS) and model specific equipment. TR-103959. Guidance is
Limited application for fragility. result of subjective opinion of
testing community. Guidance
considered adequate.

No specific requirements on
developing fragility from
GERS.

No specific requirements on
developing fragility from
GERS.

No specific requirements on
developing fragility from
GERS.

Guidance is not provided in


HlR-FA-A2 allows fragilities
EPRI documents. Example
from earthquake experience
provided in Appendices C & D. for Capability Category I

HLR-FA-A2 restricts seismic


experience to screening.

Fragility must be based on


plant specific data.

Develop Fragilities From


Earthquake Experience

Similar argument to
qualification testing.
Earthquake experience is
equivalent to multiple tests of
generic equipment.

4-39

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

No specific requirements on
development of fragilities from
similarity.

No specific requirements on
development of fragilities from
similarity. Plant specific data
is required.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

Correlation to ANS
Standard Capability
Category 1

Develop Fragilities From


Similarity to Other
Components

This is a generic derivation


commonly used for groups of
similar components and
modeled as supercomponents
or surrogate elements in the
fault trees.

Develop Equipment
Response Factors for
Fragilities Developed by
Analysis

Defines the conservatism or


unconservatism in the
response used to develop
equipment capacity factors.

Qualification Method (Static


Analysis, Response Spectrum
Analysis)

Qualification analysis methods


often incorporate conservative
approximations of equipment
response that must be
quantified.

Effect of inelastic response of


structure on in-structure
spectra.

Generally not addressed due Addressed in chapter 3 of


to fact that weakest equipment EPRI TR-103959 by
capacity is less than yield level references to literature.
in structure. Important in
some cases to address.

No requirements in standard.

No requirements in standard.

No requirements in standard.

Spectral Shape (Design


Spectra Vs UHS Spectra and
Peak Broadening and
Smoothing)

Difference in calculated
response between design
spectra and UHS spectra must
be quantified.

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

Damping (Median Damping Vs Difference in response


Design Damping)
between design damping and
median damping must be
quantified.

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

Modeling

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

4-40

Modeling is usually considered


to be median centered but
uncertainty results from
calculated mode shapes and
frequencies.

Not discussed in specific detail No specific requirements on


but implied in EPRI
development of fragilities from
documents. This is commonly similarity.
done by SPRA practitioners.

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Mode Shape

Uncertainty results from model Guidance is provided in EPRI


simplifications and boundary
TR-103959. Further guidance
condition approximations.
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

Frequency

Uncertainty results from model


and boundary condition
approximations and weight
estimations.

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

Mode Combination

Usually by SRSS. Statistical


Guidance is provided in EPRI
uncertainty must be quantified. TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

Earthquake Component
Combination

Usually by SRSS. SRSS


considered to be median.
Statistical uncertainty must be
quantified. If ECC
combination is not by SRSS,
difference between
combination method and
median response must be
quantified.

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

Equipment Response Factor Developed from individual


factors for each response
and Variability
variable and randomness and
uncertainty for variables.
Separation of variables
approach commonly used.

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

General requirements in HlRFA-C1 through C-6 apply


primarily to structural
response. Equipment
response could be implied.

4-41

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element
Develop Structural
Response Factor for
Equipment

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Must be derived from


parameters determined for
structural response in
development of in-structure
response spectra. (Spectral
shape, damping, modeling
including SSI modeling, mode
combination, ground motion
incoherence, high frequency
effects). Defines conservatism
or unconservatism in
development of response
spectra used to develop
equipment capacity and
response factors. Separation
of variables approach is
usually used to develop final
factor from individual factors
for each important variable.
Probabilistic response should
have factor of unity and
variability is defined by statics
of response (spectral
acceleration vs frequency).

Guidance is provided in EPRI


TR-103959. Further guidance
in EPRI NP-6041. Existing
guidance is adequate.

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

Current practice is to treat dual


train components as
correlated and components
within a train or in other
systems as uncorrelated. All
components are partially
correlated due to the fact that
they experience the same
earthquake. Many also have
correlated failure modes such
as pull out of expansion
anchors.

No guidance in EPRI
documents. Guidance is
added to the extent that it
affects the fragility analyst.
This is primarily a systems
modeling issue.

ANS Guide requirement SAB3 says to PERFORM an


analysis of seismic-caused
dependencies and correlation
in a way so that any screening
of SSCs appropriately
ACCOUNTS FOR those
dependencies and
correlations. This requires
guidance to the fragility
analyst to define correlation
parameter or coefficients in
the fragility description.

ANS Guide requirement SAB3 says to PERFORM an


analysis of seismic-caused
dependencies and correlation
in a way so that any screening
of SSCs appropriately
ACCOUNTS FOR those
dependencies and
correlations. This requires
guidance to the fragility
analyst to define correlation
parameter or coefficients in
the fragility description.

ANS Guide requirement SAB3 says to PERFORM an


analysis of seismic-caused
dependencies and correlation
in a way so that any screening
of SSCs appropriately
ACCOUNTS FOR those
dependencies and
correlations. USE plantspecific dependencies and
correlation values throughout.
This requires guidance to the
fragility analyst to define
correlation parameter or
coefficients in the fragility
description.

Special Topics
Correlation of Basic Events

4-42

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Develop Fragility for Buried


Tanks

Generally screened out for low


seismicity sites. Tank design
must be examined and analyst
must confirm that screening
level meets target surrogate
fragility.

Develop Fragility for Buried


Pipe Lines

Generally screened out for low Criteria for computing strains


seismicity sites. Piping design in buried pipe is provided in
must be examined and analyst Appendix C of EPRI NP-6041
must confirm that screening
level meets target surrogate
fragility.

Develop Fragilities for Items


Affected by Soil Settlement
Due to Ground Shaking or
Liquefaction

Not commonly done. Ground


settlement can occur without
liquefaction. Ground
settlement can affect buried
pipes for service water and
diesel fuel.

No guidance in EPRI
documents. Special case
requiring case specific criteria
and analysis.

Develop Fragilities for


Retaining Walls

Correlation to ANS
Standard Capability
Category 1

Criteria for pressure on walls


No specific requirements in
is developed in Appendix C of Standard
EPRI NP-6041. Can be used
to define approximate external
pressure load on buried tanks.

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

Not a common issue but must


be addressed if failure of wall
can result in failure of
essential equipment.

Criteria in Appendix C of EPRI No specific requirements in


Standard
NP-6041 for pressure on
retaining walls.

No specific requirements in
Standard

No specific requirements in
Standard

Develop Fragilities for Dams

Not a common issue but must


be addressed if failure of
upstream or down stream can
result in flooding or loss of
ultimate heat sink.

No guidance in EPRI
documents. Some guidance
in Appendix C of EPRI NP6041 on slope stability.
Usually requires expert in
dams to evaluate.

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

Develop Fragilities for HVAC

Usually not much attention


paid to HVAC ducting in
seismic PRA. Focus is on
supports.

No guidance in EPRI TR103969. Same guidance in


EPRI NP-6041. Failure mode
is structural and general
guidance in EPRI documents
is applicable.

No specific requirements in
Standard

No specific requirements in
Standard

No specific requirements in
Standard

4-43

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Peer Review

Independent review of
methodology, modeling and
results is required to assure
assumptions and correct
application of methods.

Requirements in NUREG1407. NUREG 1407 is


referenced.

Section 5 of Standard refers to


applicable Peer Review
Requirements in ASME PRA
Standard plus adds additional
requirements for Seismic PRA,
Seismic Margins and PRA of
Other External Events.

Section 5 of Standard refers to


applicable Peer Review
Requirements in ASME PRA
Standard plus adds additional
requirements for Seismic PRA,
Seismic Margins and PRA of
Other External Events.

Section 5 of Standard refers to


applicable Peer Review
Requirements in ASME PRA
Standard plus adds additional
requirements for Seismic PRA,
Seismic Margins and PRA of
Other External Events.

Reporting

Proper documentation is
necessary for reviews and
retention of records for future
risk informed applications.

Subject is addressed in TR 103959. Reporting is not


addressed. Guidance on
reporting is contained in
NUREG- 1407 and EPRI NP6041. Guidance is adequate.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Parameter that fragility is


indexed to.

Clear explanation is required


of the parameter that the
fragility is indexed to. Index
can be pga or Sa over a
frequency range of interest
defined at bedrock, soil
surface or rock outcrop.
Fragility must reflect the
definition of hazard to be use
in quantification of
unconditional probability of
failure including whether the
hazard is the average of two
horizontal components the
peak horizontal component.

Subject is addressed in TR 103959. Reporting is not


addressed. Guidance on
reporting is contained in
NUREG- 1407 and EPRI NP6041. Guidance is adequate.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

4-44

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Failure Mode (Functional,


Recoverable or
Unrecoverable, Pressure
Boundary Failure, Leak,
Break, etc.

Clear guidance must be


provided to the systems
engineers as to the failure
mode. Different failure modes
may have different
consequences in the risk
model.

Subject is addressed in TR 103959. Reporting is not


addressed. Guidance on
reporting is contained in
NUREG- 1407 and EPRI NP6041. Guidance is adequate.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Methodologies used for


development of fragilities

Different methods within


generally accepted practice
may be used and must be
clearly documented.

Subject is addressed in TR 103959. Reporting is not


addressed. Guidance on
reporting is contained in
NUREG- 1407 and EPRI NP6041. Guidance is adequate.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Walkdown team, walkdown


process and results

The walkdown team, process,


screening criteria and
screening results must be
documented.

Subject is addressed in TR 103959. Reporting is not


addressed. Guidance on
reporting is contained in
NUREG- 1407 and EPRI NP6041. Guidance is adequate.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

4-45

EPRI Proprietary Licensed Material


Development of Fragilities in Accordance with ANS 58.21

Fragility Development Description of Fragility


Elements
Development Element

Applicable
Methodology

Correlation to ANS
Standard Capability
Category 1

Correlation of ANS
Standard Capability
Category 2

Correlation of ANS
Standard Capability
Category 3

Surrogate elements

The description of the criteria


for establishing surrogate
element fragility targets and
the development of surrogate
fragilities from walkdown
screening and qualification
report review must be clearly
documented.

Subject is addressed in TR 103959. Reporting is not


addressed. Guidance on
reporting is contained in
NUREG- 1407 and EPRI NP6041. Guidance is adequate.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Consequence of Interactions
(falling, impact, spray,
flooding, fire, explosion, etc.).

The affect of interactions with


essential equipment must be
identified and quantified.

Subject is addressed in TR 103959. Reporting is not


addressed. Guidance on
reporting is contained in
NUREG- 1407 and EPRI NP6041. Guidance is adequate.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

Section 7 provides general


and a few specific
requirements for
documentation. More specific
guidance is provided in
NUREG-1407 but for a
specific end item of developing
risk insights. The ANS
standard focuses on analytical
models and sufficient
documentation for a living
methodology for risk informed
applications.

4-46

EPRI Proprietary Licensed Material

5
INDEX TO EXAMPLE FRAGILITY CALCULATIONS

The methodology presented herein, in EPRI TR-103959, Methodology for Developing Seismic
Fragilities, (EPRI, 1994) and in other references, describes the general procedures for
developing seismic fragilities. The focus is on classic cases for brittle and ductile structural
failures. However, the methodology is unfamiliar to most design engineers who work primarily
with deterministic design codes and standards and, as for any technological application, having
specific examples is essential to assure that the generalized methodology is understood and
applied correctly. EPRI TR-103959 presents several excellent examples that address common
situations that the fragility analyst will confront. The appendices to this document provide
additional examples to supplement the database of examples in EPRI TR-103959. In addition,
NUREG/CR-5270 (Kennedy et.al, 1989) provides a comparison of fragility calculations
performed by different individuals that demonstrates some of the different considerations that a
fragility analyst may make regarding the median capacity of a SSC vs the code type capacity.
Table 5-1 provides an index of sample fragility calculations existing in the public literature.

5-1

EPRI Proprietary Licensed Material


Index to Example Fragility Calculations
Table 5-1
Index To Example Fragility Calculations

Type of SSC

5-2

Reference to Example Fragility


Calculation

Reinforced Concrete Shear Wall

EPRI TR-103959, Chapter 6

Vertical Atmospherical Storage Tank

EPRI TR-103959, Chapter 7

Vertical Atmospheric Storage Tank

NUREG/CR-5270

Horizontal Heat Exchanger

EPRI-TR-103959, Chapter 8

Horizontal Heat Exchanger

NUREG/CR-5270

Expansion Anchor

EPRI TR-103959, Chapter 9

Equipment Qualified by Testing-Motor Control Center


Using GERS

NUREG/CR-5270

Vertical Air Tank With Support Skirt

NUREG/CR-5270

Cantilevered Reinforced Masonry Wall

NUREG/CR-5270

Fragility of Instrument Cabinet Derived from Earthquake


Experience

Appendix D

Generic Fragility Derived from Design Criteria

Appendix E

Vertical Long Column Service Water Pump

Appendix F

Ground Settlement Due to Liquefaction

Appendix H

Simplified Scaling of Spectra

EPRI NP6041SR, Chapter 4

Simplified Scaling of Spectra

Appendix A

Detailed Scaling of Spectra

Appendix B

Derivation of Fragilities from Earthquake Experience


Using Survival Analysis

Appendix C

EPRI Proprietary Licensed Material

6
REFERENCES

1. ANS (1997), ANS 2.29, Standard for Probabilistic Analysis of Natural Phenomena Hazards
for Nuclear Facilities, Draft, December.
2. ANS (2000), ANS 2.27, Standard Covering Guidelines for Investigations of Nuclear
Materials Facilities for Seismic Hazard Assessments, Draft, September.
3. ANS (2002), ANS 58.21, External Events PRA Methodology Standard, Draft, April.
4. ASCE (1998), Seismic Analysis of Safety-Related Nuclear Structures and Commentary on
Standard for Seismic Analysis of Safety Related Nuclear Structures, ASCE Standard,
American Society of Civil Engineers, 1998.
5. ASME (2001), Standard for Probabilistic Risk Assessment for Nuclear Power Plant
Application, American Society of Mechanical Engineers, 2001.
6. Budnitz, R.J. et. al, An Approach to the Quantification of Seismic Margins in Nuclear
Power Plants, NUREG/CR-4334, August 1985.
7. Budnitz, R.J., State of the Art Report on the Current Status of Methodologies for Seismic
PSA, Committee on the Safety of Nuclear Installations, OECD Nuclear Energy Agency,
March 1998.
8. Campbell, R.D., M.K. Ravindra and R.C. Murry (1988), Compilation of Fragility
Information from Available Probabilistic Risk Assessments, UCID-20511 Rev. 1, Lawrence
Livermore National Laboratory, 1988.
9. Campbell (1993), Insights from Probabilistic Risk Assessments and Seismic Margin
Assessments Regarding The Variance of Design Margins for Seismic Events, ASME
Technology for the 90s, American Society of Mechanical Engineers, 1993
10. Chokski, N, et. al., (1999), Lessons Learned from IPEEE Studies Regulatory Applications
Using Risk-Insights, Proceedings of OECD/NEA Workshop on Seismic Risk, Aug 10-12,
1999, Tokyo, Japan.
11. Ellingwood (1994), Validation of Seismic Probabilistic Risk Assessments of Nuclear Power
Plants, NUREG/CR-0008, Jan 1994.
12. EPRI (1988), A Methodology for Assessment of Nuclear Power Plant Seismic Margin,
EPRI NP-6041, EPRI, Palo Alto, California, October 1988.

6-1

EPRI Proprietary Licensed Material


References

13. EPRI (1989a) Probabilistic Hazard Evaluation at Nuclear Plant Sites in the Central and
Eastern United States, Resolution of the Charleston Issue, EPRI NP-6395-D, EPRI, Palo
Alto, California, April, 1989.
14. EPRI (1989b), Seismic Margin Assessment of the Catawba Nuclear Station, EPRI NP6359, EPRI, Palo Alto, California, April 1989.
15. EPRI (1991a) Seismic Ruggedness of Relays, EPRI NP-7147, EPRI, Palo Alto, California,
February 1991.
16. EPRI (1991b) A Methodology for Assessment of Nuclear Power Plant Seismic Margin,
EPRI NP-6041SL, Revision 1, EPRI, Palo Alto, California, June 1991.
17. EPRI (1991c) Generic Seismic Ruggedness of Power Plant Equipment in Nuclear Power
Plants, EPRI NP-5223, Revision 1, EPRI, Palo Alto, California, February, 1991.
18. EPRI (1991d), Seismic Margin Assessment of the Edwin I Hatch Nuclear Plant Unit 1,
EPRI NP-7217, EPRI, Palo Alto, California, March 1991.
19. EPRI (1993) Analysis of High-Frequency Seismic Effects, EPRI TR-102470, EPRI, Palo
Alto, California, October 1993.
20. EPRI (1994) Methodology for Developing Seismic Fragilities, EPRI TR-103959, EPRI,
Palo Alto, California, June 1994.
21. EPRI (1996), GERS Formulated using Data from the SQURTS Program, EPRI TR105988, Volume 1, EPRI, Palo Alto, California, April, 1996.
22. EPRI (1998), SQUG Electronic Earthquakes Experience Database Users Guide, EPRI TR110781, EPRI, Palo Alto, California, May 1998.
23. EPRI (1999) GERS Formulated Using Data From the SQURTS Program, EPRI TR105988-V2, EPRI, Palo Alto, California, April 1999.
24. EPRI (2000a) Individual Plant Examination for External Events (IPEEE) Seismic Insights,
EPRI TR-112932, Revision, EPRI, Palo Alto, California, December 2000.
25. EPRI (2000b) Planning for Risk-Informed Seismic Regulations, EPRI 1000896, EPRI,
Palo Alto, California, December 2000.
26. Fleming, K.N. and T.J. Mikschl, (1999) Technical Issues in the Treatment of Dependence in
Seismic Risk Analysis, Proceedings of OECD/NEA Workshop on Seismic Risk, Aug. 10-12,
1999, Tokyo, Japan.
27. FOAKE (1993), Technical Core Group Report on FOAKE Task E-1, ASME Piping,
Appendix M, Quantification of the Margin of Conservatism in Seismic Design In-Structure
Response Spectra, January, 1993.

6-2

EPRI Proprietary Licensed Material


References

28. FOAKE (1996), Advanced Light Water Reactor (ALWR) First-of-a-Kind Engineering
Project on Equipment Seismic Qualification, Prepared by MPR Associates and EQE
International for Advanced Reactor Corporation, Feb. 1996.
29. Hunter (1976), An Upper Bound on the Probability of a Union, Journal of Applied
Probability 13, pgs. 597-603, 1976.
30. Johnson, J.J, et. al., SMACS Seismic Methodology Analysis Chain with Statistics,
NUREG/CR-2015, Volume 9, 1981.
31. Kennedy, R.P., C.A. Cornelle, R.D. Campbell, S. Kaplan and H.F. Perla (1980),
Probabilistic Seismic Safety Study of an Existing Nuclear Power Plant, Nuclear
Engineering and Design, Vol. 59, No. 2, pp. 305-338, 1980.
32. Kennedy, R.P. et. al, (1989), Assessment of Seismic Margin Calculation Methods,
NUREG/CR-5270. 1989.
33. Kennedy, R.P. (1999), Overview of Methods for Seismic PRA and Margins Analysis
Including Recent Innovations, Proceedings of OECD/NEA Workshop on Seismic Risk,
Aug. 10-12, 1999, Tokyo, Japan.
34. Moore, D.L. et al., (1987), Seismic Margin Review of the Maine Yankee Atomic Power
Station, NUREG/CR-4826, Vol. 2, 1987.
35. Newmark, N.M. and W.J. Hall (1978), Development of Criteria for Seismic Review of
Selected Nuclear Power Plants, NUREG/CR-0098, May 1978.
36. PG&E (1988), Final Report of the Diablo Canyon Long Term Seismic Program, Pacific
Gas and Electric Company, Docket 50-275 and 50-323, July 1988.
37. Prassinos, P.G., M.K. Ravindra and J.D. Savy (1986), Recommendations to the Nuclear
Regulatory Commission on Trial Guidelines for Seismic Margin Reviews of Nuclear Power
Plants, Lawrence Livermore National Laboratory, NUREG/CR-4482, 1986.
38. Prassinos, P.G., R.C. Murry, G.E. Cummings (1987), Seismic Margin Review of the Maine
Yankee Atomic Power Station, Summary Report, NUREG/CR-4826, Vol. 1, 1987.
39. Ravindra, et.al. (1984), Sensitivity Studies of Seismic Risk Models, EPRI NP-3562, June
1984.
40. Ravindra, M.K., G.S. Hardy, P.S. Hashimoto, J.J. Griffin (1987), Seismic Margin Review of
the Maine Yankee Atomic Power Station, NUREG/CR-4826, Vol. 3, 1987.
41. Ravindra, M.K. (1999), Seismic PSAs Issues, Resolutions and Insights, Proceedings of
OECD/NEA Workshop on Seismic Risk, Aug. 10 -12, 1999, Tokyo, Japan.
42. Reed, J.R. et. al.(1994), In-Structure Response for Calculating Equipment Capacities in
SMA and SPRA Reviews, Symposium on Current Issues Related to Nuclear Power Plant
Structures, Equipment and Piping, North Carolina State University, December 1994.
6-3

EPRI Proprietary Licensed Material


References

43. Riddell, R and N.M. Newmark (1979), Statistical Analysis of the Response of Nonlinear
Systems Subjected to Earthquakes, Univ. of Illinois Civil Engineering Dept. Report UILU
79-2016, August 1979.
44. Salmon, M.W. and R.P. Kennedy, Meeting Performance Goals by the Use of Experience
Data, Lawrence Livermore National Laboratory, UCRL-CR-120813, December 1, 1994.
45. SEQUAL (2001), Basis for Adoption of the Experience-Based Seismic Equipment
Qualification (EBSEQ) Methodology by Non-A46 Nuclear Power Plants, Topical Report,
April 2001.
46. SQUG (1991), Generic Implementation Procedure (GIP) for Seismic Verification of
Nuclear Plant Equipment, Revision 2, Corrected, Seismic Qualification utility Group, June
1991.
47. SQUG (2001), Procedure for Gathering and Validating Earthquake Experience Data,
Revision 3, Seismic Qualification Utility Group, May 2001.
48. Ueki, T, T. Sueki, Y. Kitada, H. Niwa and Ml Fujii, (1999), High Acceleration Level
Vibration Tests for Electric Components, Proceedings of OECD/NEA Workshop on
Seismic Risk, Aug. 10-12, Tokyo, Japan.
49. USNRC (1975), Reactor Safety Study, WASH 1400, NUREG-73/041, 1975.
50. USNRC (1978), Combining Modal Responses and Spatial Components in Seismic
Response Analysis, Rev. 1, Feb. 1978.
51. USNRC (1981) Seismic Safety Margins Research Program, Phase I Final Report, SMACS
Seismic Methodology Analysis Chain with Statistics (Project VIII), NUREG/CR-2015, Vol.
1-9, 1981.
52. USNRC (1983), PRA Procedures Guide, Vol. 2, NUREG/CR-2200, 1983.
53. USNRC (1984), Engineering Characterization of Ground Motion, Task 1: Effects of
Characteristics of Free-Field Motion on Structural Response, NUREG/CR-3805, May 1984
54. USNRC (1985), Policy Statement on Severe Accidents, Federal Register, Vol. 50, 32138,
August 8, 1985.
55. USNRC (1985), Probabilistic Safety Analysis Procedures Guide, NUREG/CR-2815, 1985.
56. USNRC (1988), Individual Plant Examination for Severe Accident Vulnerabilities
10CFR50.54f, USNRC Generic Letter 88-20, November 23, 1988.
57. USNRC (1989), Standard Review Plan, NUREG-0800, Revision 2, 1989.
58. USNRC (1990), Severe Accident Risk, An Assessment of Five US Nuclear Power Plants,
NUREG-1150, Volumes 1 and 2, December 1990.

6-4

EPRI Proprietary Licensed Material


References

59. USNRC (1991a) Individual Plant Examination of External Events (IPEEE) for Severe
Accident Vulnerabilities 10 CFR 50.54 (f) (Generic Letter No . 88-20, Supplement 4),
June 1991.
60. USNRC (1991b), Procedural and Submittal Guidance for the Individual plant Examination
of External Events (IPEEE) for Severe Accident Vulnerabilities, NUREG-1407, June 1991.
61. USNRC (1994), Revised Livermore Seismic Hazard Estimates of 69 Nuclear Plant Sites
East of the Rocky Mountains, NUREG/CR-1488, Lawrence Livermore National
Laboratories, April 1994.
62. USNRC (1997) Regulatory guide 1.165 Identification and Characterization of Seismic
Sources and Determination of Safe Shutdown Earth Ground Motion, March 1997.
63. USNRC (1999), Safety Evaluation of GE Topical Report, NEDE-31858-P, Revision 2,
BWROG Report for Increasing MSIV Leakage Limits and Elimination of Leakage Control
Systems, September, 1993, USNRC, March 3, 1999.
64. USNRC (2000a), Perspectives Gained from the Individual Plant Examination of External
Events (IPEEE) Program, NUREG-1742, April 2000.
65. USNRC (2000b), Risk-Informed Regulation Implementation Plan, October 2000.
66. USNRC (2001) Review of the Seismic Qualification Utility Group Procedure for Gathering
and Validating Earthquake Experience Data, Revision 2 (TAC NO. MA9464), April 27,
2001.
67. Watanabe, Y., T Oikawa and K. Muramatsu (1999), Reappraisal of the Effect of Correlation
of Component Failures on Core Damage Frequency in Seismic PSA Using DQFM Method,
Proceedings OECD/NEA Workshop on Seismic Risk, Aug. 10-12, 1999, Tokyo, Japan.
68. Zion (1981), Zion Probabilistic Safety Study, Prepared by Pickard, Lowe and Garrick, Inc.
for Commonwealth Edison Company, 1981.

6-5

EPRI Proprietary Licensed Material

A
BENCHMARK STUDIES TO VERIFY AN APPROXIMATE
METHOD FOR SPECTRA SCALING

A.1. Background
Often in conducting a SPRA, the in-structure response spectra developed for design are desired
to be scaled to represent in-structure response spectra for a different earthquake with a different
ground motion spectral shape.
EPRI NP-6041 [A1] for Seismic Margin Assessment (SMA) cautions against scaling spectra
using a simple single mode procedure if the ground motion spectral shape for the newly defined
review level earthquake (RLE) is significantly different than the ground motion spectral shape
used for developing the in-structure spectra to be scaled. More sophisticated scaling procedures
can be applied providing that the eigensolutions for the original models are available. These
scaling procedures can utilize random vibration theory, direct generation computer programs,
also based on random vibration theory, or time history solutions. In some cases, the
eigensolution outputs in the analysis reports are only partially complete. In the case of the
reactor building model in this example, the participation factors () and eigenvectors () were
available. In some cases, participation factors are provided but mode shapes are not. In this
example, spectra are scaled rigorously using random vibration theory and by more simplified
procedures using only frequencies and participation factors.
The reactor building in-structure spectra in this example exhibit evidence of more than one
significant mode contributing to the spectral shape. It is therefore a good selection for a
benchmark study whereby in-structure response spectra are developed using a more sophisticated
scaling method for comparison to spectra developed using a simple scaling method that could be
used on other structures that do not have complete eigensolutions available.

A.2. Verification of Original Spectra to be Scaled


The first step in the study was to recreate the model of the concrete reactor building using the
input data contained in the dynamic analysis report. The model was constructed on SAP 2000
[A2]. Modal frequencies from the reconstructed model agreed quite well with those in the
analysis report. Figure A-1 shows the lumped mass model for the concrete reactor building.
Note that the reactor building is not typical of those in US PWRs. Nodes 11 and 12 represent the
top of the vault structure where the steam generators (boilers) are supported.
The spectral shape of the Design Basis Earthquake (DBE) was a RG 1.60 spectrum anchored to
0.05g pga. The DBE spectra were broadened and smoothed to result in conservative in-structure
A-1

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

spectra to be used by the designers to qualify piping and equipment. Spectra for node 11 at the
upper steam generator support are shown as Figures A-2 through A-4. Spectra for other nodes
are similar in shape. From the shape of the two horizontal spectra, we would conclude that a
mode or modes with frequency beyond that of the peak of the spectra have some contribution to
spectral shape. We would therefore anticipate that scaling must be done at more than one
frequency.
Using the recreated model and artificial time histories that envelope the RG 1.60 spectral shapes,
new DBE spectra were created for the reactor building at selected nodes. Figure A-5 shows the
match of the spectra resulting from the artificial time history to the RG 1.60 spectral shape. A
similar comparison of the original artificial time history induced spectra for the DBE was not
available in the analysis report. Typically, there was conservatism in the time histories used in
older analyses. Using node 11 as a representative case for comparison we find that the original
DBE spectra appear to be conservative relative to the recreated spectra at the peak of the spectra
for the horizontal case. For the vertical case and for higher frequencies in the two horizontal
directions, the recreated spectra are slightly higher than the DBE spectra. This is likely the result
of the integration scheme employed in earlier computer codes vs. that in SAP 2000. Earlier
codes used integration schemes that tended to suppress higher frequency response. Figures A-6,
A-7 and A-8 show the broadened spectra for node 11 for the recreated model. Four percent
damped DBE spectra are partially superimposed for comparison. Since equipment for SPRA is
primarily evaluated at 5% damping, the 4% damping cases developed for designs are the most
meaningful for comparison. This part of the benchmark study leads to the conclusion that the
original DBE spectra are comparable to spectra that would be developed by modern computer
codes using artificial time histories that produced spectra that were a very close match to the
specified ground motion response spectra. The conservatism at the peak of the original DBE
spectra was likely due to conservatism in the original time histories. The lower spectral
ordinates observed at higher frequencies is not a significant deviation from the DBE spectra plus
this higher frequency response is not nearly as damaging to equipment as low frequency
response.
Another check conducted on the DBE spectra was to use the eigensolution values for and
from the analysis report and conduct a response analysis using the same artificial time histories
used in the SAP 2000 model. Results are shown in Figure A-9 for 4% damping, node 11,
direction X. The same observations are made in the comparison of DBE spectra to a time history
solution using the eigenvalues from the DBE model. In this case, the DBE spectra appear even
more conservative at the peak and are slightly unconservative at about 18 Hz where a second
mode is present. This further confirms the validity of the original spectra to be scaled and
demonstrates some conservative bias in the DBE spectral peaks.

A.3. Development of Scaled Spectra by Rigorous and by Simplified Means


Given the participation factors and the modal displacements, an approximate but accurate
method for developing in structure spectra by scaling can be derived from random vibration
theory. This is the basis for so called direct generation computer codes. However, inherent in
the random vibration analysis, is a scale factor that is applied to the RMS response to obtain peak
response. This scale factor often results in very conservative in-structure response if the absolute
value is taken. A scheme to scale in-structure spectra that has been used successfully is to
A-2

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

compute in-structure spectra using random vibration theory for the original DBE ground motion
spectra, and for the new UHS ground motion spectra. The ratio of these two in-structure spectra
at different frequencies can then be used to scale the DBE in-structure response spectra to
represent in-structure spectra for the UHS. The mathematical description of this process is
contained in Appendix B. The process is most easily carried out using a direct generation
computer code, but may also be done on a spread sheet using eigenvalues for modes of the most
interest.
For this study we used a typical UHS for eastern US sites. The UHS has a peak ground
acceleration of 0.1g compared to the 0.05g DBE but the peak occurred at 20 Hz as opposed to
the 2.5 to 9 Hz amplified acceleration range for the RG 1.60 DBE spectrum. Figure A-10
compares these two spectra. Above 5 Hz the UHS spectral amplitudes are higher. Thus we
would expect that high frequency modes would be greatly amplified by the UHS whereas there
was not significant amplification for the DBE ground motion spectrum.
Figures A-11 and A-12 show the node 11 responses to the DBE and the UHS using the random
vibration procedure described in Appendix B. The higher mode at about 18 Hz is highly
amplified by the UHS. The ratios between these two spectra are shown in Figure A-13. These
ratios, as a function of frequency, can then be used to scale the DBE in-structure response
spectra. Figure A14 shows the scaled DBE spectrum to represent response to the UHS.
This method can be used if the participation factors and eigenvectors are available. In some
cases, only frequencies and participation factors are tabulated in design reports. However, from
the shape of the in-structure spectra and the participation factors, one can deduce the modes that
are contributing significantly to response and to the shape of the in-structure spectra. If scale
factors are developed at these significant modes and used to scale the DBE spectra, the resulting
in-structure spectra for the UHS should be a reasonable approximation of UHS in-structure
response spectra. This sample scaling might be considered applicable to capability Category 1 of
the Standard. The next step was then to do this simple scaling for node 11 in the X direction and
compare the results to the more rigorous scaling results.
If we examine the DBE spectrum for node 11, direction X, Figure A-2, we observe that the mode
driving the peak of the spectrum occurs at about 6.6 Hz. This is mode 1. It then appears that
there is another mode or modes less than 20 Hz that contributes to response. Examining the
participation factors for the X direction we find that mode 6 at 18.5 Hz has a high participation.
Only modes 1 and 6 have significant participation. We would then select these two frequencies
to scale the DBE spectrum. By comparing the UHS and DBE spectra in Figure A-10 at these
two frequencies the resulting scale factors are:
Frequency Hz

UHS Sa (g)

DBE Sa (g)

Scale Factor

6.6

0.16

0.136

1.18

18.5

0.24

0.077

3.12

The broadened DBE spectra from 5.8 to 7.2 Hz in Figure A-2 are then scaled up by the 1.18
scale factor determined at 6.6 Hz. At frequencies significantly below the broadened peak
frequency the UHS exhibits lower spectral acceleration than the DBE and the amplified peak can
A-3

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

conservatively be fared into the DBE spectra. From about 12 to 20 Hz the DBE spectra in Figure
A-2 are flat. This is a result of broadening the peaks and smoothing. The scale factor of 3.12
should be applied to this part of the spectra. Between the broadened peak and 12 Hz the spectra
should be fared using the same general shape as the DBE. The DBE zero period acceleration at
33 Hz should be scaled up by the ratio for the 6.6 Hz fundamental mode and from 20 Hz to 33
Hz the spectra should be fared.
This simple scaling is also plotted on Figure A-14 for comparison to the more rigorously scaled
spectrum. As can be seen, the simple scaling results in conservative spectra through the
frequency range of interest. The zpa for simple scaling is about 15% lower than for the more
rigorous scaling. Since the seismic issues are primarily with flexible equipment, the slightly
lower zpa is of no consequence. Note that in Figures A-6 and A-7, the spectra from the recreated
model were higher at about 18 Hz than the DBE spectra. This is also shown in Figure A-9
comparing the time history results using the eigensolutions to the DBE design spectrum.
However, when the simple scaling is done as shown in Figure A-14, the large scale factor at 18
Hz overcompensates for this and the resulting scaled spectra are conservative except in the rigid
range.
This example of simple scaling for a case where more than one mode has significant response
may be extended to address other cases of a dominant single mode or multiple modes of
response.
The scaling process using the original eigensolutions and random vibration theory is considered
to be acceptable for capability Category 1 and 2 of the Standard, when limited to fixed base
models. The simplified Scaling based on identifying important modes by participation factors
and spectral peaks is considered to be reasonable for capability Category 1 for fixed base models.
However, the analyst must be reasonably confident that the original analysis and resulting
spectra to be scaled are realistic. In this case, they were shown to be conservative at the peak
and slightly unconservative at higher frequency.

A.4. References
A1. EPRI NP-6041 SL, A Methodology for Assessment of Nuclear Power Plant Seismic
Margin (Revision 1), EPRI, Palo Alto, California 1991.
A2. SAP 2000, Three Dimensional Static and Dynamic Finite Element Analysis and Design of
Structures, Version 7.4, August 2000, Computers and Structures Inc., Berkeley, CA.
A3. Regulatory Guide 1.60, Design Response Spectra for Seismic Design of Nuclear Power
Plants, US Nuclear Regulatory Commission, 1973.

A-4

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-1
Lumped Mass Model of Reactor Building

A-5

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-2
Reactor Building EQ Floor Spectra, Node 11

A-6

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-3
Reactor Building NS Floor Spectra, Node 11

A-7

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-4
Reactor Building Vertical Floor Spectra, Node 11

A-8

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-5
RG 1.60 Spectrum Compatible Time Histories

A-9

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-6
Reactor Building E-W Floor Spectra Reconstructed Model, Node 11

A-10

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-7
Reactor Building N-S Floor Spectra Reconstructed Model, Node 11

A-11

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-8
Reactor Building Vertical Floor Spectra Reconstructed Model, Node 11

A-12

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Design 11X
11X Modal Response @7%

Figure A-9
EW Floor Response Spectrum Developed From Eigensolution of DBE Analysis, Using RG 1.60 Time Histories

A-13

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

UHS

DBE

Figure A-10
Comparison of DBE with UHS

A-14

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-11
RB Estimated SDOF Oscillator Response Node 11

A-15

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-12
RB Estimated SDOF Oscillator Response Node 11

A-16

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Figure A-13
RB UHS Scale Factors Node 11

A-17

EPRI Proprietary Licensed Material


Benchmark Studies to Verify an Approximate Method for Spectra Scaling

Rigorous Scaling
Simplified Scaling

Figure A-14
Scaled DBE Spectra Node 11

A-18

EPRI Proprietary Licensed Material

B
DEVELOPMENT OF IN-STRUCTURE RESPONSE
SPECTRA FOR SEISMIC MARGIN OR SEISMIC PRA
EVALUATION BY SCALING

B.1

Introduction

In-structure spectra developed for design often do not take advantage of methods to reduce high
frequency spectral peaks that have been developed to address this issue in the US Individual
Plant Examination of External Events (IPEEE). In-structure spectra are typically obtained from
floor time-histories generated from a modal time-history analysis of the structures. The analyses
utilize time-history base input acceleration records whos response spectra envelope the specified
ground motion uniform hazard spectra (UHS) for SPRA or the Review Level Earthquake (RLE)
spectrum for seismic margin assessment. According to the guidelines for SPRA and SMA
conduct (see References B8 and B1), the estimation of seismic demand is to be median-centered
except that the ground motion for SPRA is to be specified as the median UHS whereas RLE is
to be established at the approximate 84% non-exceedance level. This appendix discusses the
various steps that can be used to affect the removal of conservative bias in the response analysis
with the goal of obtaining an estimate of median-centered response conditional on the occurrence
of the SPRA UHS or an RLE.
For the examples shown herein, the primary power of the UHS motion is contained in the
10-20 Hz range (by contrast, the power of the RG 1.60 Spectrum used for design is contained in
the 2-9 Hz range). Therefore, significant conservatism is introduced if the effects of foundation
incoherence and equipment high frequency ductility are not considered. Both of these effects
result in a reduction of the high frequency content of in-structure spectra.
In this Appendix, the following subjects are addressed: (1) Reduction of ground motion for
incoherence, (2) scaling of design floor spectra compatible with reduced ground motion spectra,
and (3) reduction of floor spectra for high frequency ductility effects.

B.2

Incoherence Reduced Ground Motion

Ground motion definition spectra, such as typical UHS spectra, represent the elastic response of
an oscillator due to the motion expected at a single point on the horizontal free surface of a
half-space media. Actual ground motion, however, has horizontal spatial variation due to wave
scattering effects and statistical incoherence of motion in the frequency domain. Considering the
motion of two points on the surface with a given horizontal separation distance, measurements
have indicated that the two motions can be separated into frequency-dependent in-phase
B-1

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

components, or coherent motion, and frequency-dependent components with random phasing, or


incoherent motion. A large rigid foundation mat or a rigid multi-cell box-type concrete structure
can only achieve coherent motion over the characteristic dimension of its base. The random, or
incoherent motion has a net zero average over the characteristic dimension of the foundation.
Further, observational data indicates that the incoherent portion of the surface ground motion
increases in the higher frequency range resulting in reduction in input at higher frequencies for
structures with large plan dimension.
Based on a conservative interpretation of data for motions recorded over short separation
distances, References B1 and B8 provide a procedure for spectral reduction to account for
statistical incoherence of the foundation input motion. Using a characteristic foundation
dimension de = 150 feet as a reference value, the following spectrum reductions are
recommended:
Table B-1
Reduction Factors for 150-Foot Foundation
Frequency, Hz

Reduction factor R150

0.2

1.0

1.0

1.0

10

0.9

20*

0.8232

25

0.8

*Reduction factor determined by linear log-log interpolation

For foundations with different characteristic plan dimensions, d e , the reduction value
(1- R s ) may be extrapolated proportional to the characteristic foundation plan size of 150 feet

and reduction values (1-R150 ), determined from the R150 given in Table B-1, or
R s = 1 (de / 150)(1 R150 )

Equation B-1

For frequencies between the reference values given in Table B-1, the values of reduction
function, Rs , may be linearly interpolated as a log R s vs. log f function.
The characteristic foundation dimension for a building may be taken as the square root of the
building foundation plan area. In the following, an example building, designated as Building E,
is used to demonstrate the spectrum scaling methods. The portion of Building E with the
foundation perimeter directly supported by rock was taken as the effective foundation plan to be
used for determination of the Building E incoherence reduction:

B-2

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

E Building Foundation Plan:


46.65

40.20m

de =

A = (49 .65 X40 .20 = 43 .31m = 142 feet

Using Equation (B-1), the incoherence reduction function for Building E was obtained as shown
in Figure B-1.
Inco herenc e R educ tion - Bu ilding E

0.9

0.8

Redu ction Factor, Rs

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0

10

15

20

25

30

35

40

45

50

F requ ency, hz

Figure B-1
Reduction Function for Incoherence Across a 43.3 M (142-Foot) Foundation

Figure B-2 is a tripartite presentation (log-log) to define an example UHS. This UHS is similar
to the shape of a typical EUS UHS. On such plots, only one dependent variable, usually Spectral
Displacement (SD), can be selected for each frequency. The other variables, denoted as the
Pseudo Spectral Velocity, PSV = 2 f SD, and the Pseudo Spectral Acceleration,
2
PSA = (2 f ) SD, are derived quantities. For the current study, the often used approximation for
Spectral Acceleration, SA g PSA, was used for the UHS. Each UHS was constructed by linear
segments of log SA g vs. log f between frequency break points of 0.2, 1.0, 5, 10, 20, and 50 Hz
B-3

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

defined for damping ratios, , with values 0.5, 1.0, 2.0, 5.0 and 10 per cent of critical damping.
The common ZPA value was at 50 Hz and the common displacement was at 0.2 Hz. At the
break points, the spectral values were linearly interpolated as a log SA g vs. log function. For
frequencies between the break point values, the spectral values were linearly interpolated as a log
SA g vs. log f function.

Figure B-2
Uniform Hazard Horizontal Response Spectra

The incoherence reduced ground motion spectra for a building is then obtained as the product of
the UHS spectral acceleration and building reduction function, SA INg (f, ) = R s x SA g (f, ) , for
the value of modal damping used in the corresponding Building analyses. Example incoherence
reduced horizontal and vertical ground motion spectra for Building E, for the damping used for
the analysis of Building E, are shown in Figures B-3 and B-4.

B-4

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

Building E - Horizontal Ground Motion Spectra


0.45
4.5% Damping
0.4

0.35

SAh

UHS Input Motion

SAh * Rs

Incoherency Reduced Motion

0.3

SA 0.25
,
Hz
0.2

0.15

0.1

0.05

0
0

10

15

20

25

30

35

40

45

50

Frequency, Hz

Figure B-3
Incoherence Reduced Horizontal Ground Motion for Building E

B uilding E - Vertic al G ro und Mo tion Sp ectra

0.4
4.5% Dam p ing

0 .35

E-5 Inp ut Motio n

SAv
SAv * Rs

Incoherency Reduc ed Motion

0.3

SA, g

0 .25

0.2

0 .15

0.1

0 .05

0
0

10

15

20

25

30

35

40

45

50

F requ ency, Hz

Figure B-4
Incoherence Reduced Vertical Ground Motion for Building E

B-5

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

B.3 Estimation of Floor Spectra Compatible with Incoherence Reduced


Ground Motion
B.3.1 Scaling of Floor Spectra
The most direct procedure for obtaining estimates of median-centered response from analyses
that have a conservative bias is to simply scale the computed results by a reduction function
which approximately filters the response to remove the bias. The first step in the scaling
procedure is to determine the reduction of both horizontal and vertical ground spectra, as
discussed above, which accounts for the effect of ground motion incoherence for each building
foundation size. The next step is to provide estimates of the reduced floor spectra that would
result from the revised (reduced) median-centered ground motion spectra. Thus, each
time-history generated building floor spectrum SA ni , at node n for response in (translational)
direction i, is modified in accordance with the scaling relationship:
SA eIN
ni
SAINni = SAni x

SA eEni

Equation B-2

where SA INni is the scaled spectra that accounts for foundation incoherence. In the above
equation, SA eEni and SA eINni are the estimated floor spectra that would result from the elastic, or
unreduced UHS ground spectrum and the incoherence reduced ground spectrum, respectively.
The ratio within the brackets of Equation (B-2), may be identified as the reduction function
associated with the incoherence reduction.
B.3.2 Spectral Estimation Method
Instead of redoing the complete structural response analysis for the incoherence reduced UHS, a
simple method of estimating the modal components of floor oscillator response, given a modal
representation of the structure (, ) and the ground motion spectrum, SA g {f, } , for the
damping value, , used in the building analysis may be done. The first use of a simplified
method of direct (i.e., avoiding time-history analysis) floor spectrum generation was proposed by
Biggs (Reference B2). This method, based on empirical amplification factors, used only the
structure mode shape factors, nij , and the participation factors, jr , defined for mode j, node n,
response direction i, and input direction r. Vanmarke (Reference B3) showed that Biggs
empirical amplification factors are similar to those as obtained from random vibration theory.
The amplification factors presented in these studies compared the response of an uncoupled
Single-Degree-of-Freedom (SDOF) oscillator mounted on the structure to the same oscillator
mounted on the ground. Subsequent work (Igusa and Der Kiureghian, Reference B4) has
emphasized the random vibration formulation in the development of direct structure response
generation methods, including the effect of equipment mass coupling. Traditional generation of
floor spectra, using time-history methods, assume that the floor motion is uncoupled from the
equipment response. This assumption also can result in response conservatism, however, the
consideration of this effect is somewhat application specific (knowledge of the equipment mass
B-6

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

and local floor mass is required), therefore, in this study the scaling procedure did not include
mass coupling effects.
It should be noted that the accuracy of the spectral values obtained using an estimation method is
not that important since only the ratio of the spectral estimates is used for scaling of the original
spectra. The method used in this study for the estimation of floor spectra is based on the random
vibration results (Crandall and Mark, Reference B5) for a cascaded set of SDOF systems
(i.e., uncoupled oscillators) with white noise base motion input. Direct generation computer
codes could also be used and are the preferred choice. For purposes of illustration, the Crandal
and Mack random vibrations model using existing eigensolution values are utilized. This
procedure can be carried out on a spreadsheet using the modes of interest. Modes that contribute
little to response can be eliminated with very little loss in accuracy.
Basically, the floor response spectrum ordinate at each frequency, fk , is the sum of the
contribution of each structure mode, j, at that frequency. Each mode contribution was considered
as the response of two cascaded SDOF systems for which the first stage output was the structure

modal acceleration response component Ynj = jnj SA gj for base motion input where SA gj is the
ground response spectrum ordinate at fj . The output of the first stage is then used as input to the
second stage which is the response spectrum oscillator (on the structure) tuned to frequency fk .
The output of the second stage is the modal component of the floor response spectrum ordinate.
These relationships are illustrated in Figure B-5. In the following development, it will be
understood that the modal response of the oscillator is associated with input direction r.
SA

fj

SAnk = j(SAnkj)

Floor Response Spectrum

Znkj

Zzpa = |Yn|max

k = 2fk

Ynj

Xnj

SAgj

j = 2fj

Xg

fk

SA

Ground Response Spectrum


j

fj

ZPA = |Xg|max
f

Figure B-5
Response Spectra Relationships

The ground response spectrum is a plot of the spectral acceleration for a set of SDOF oscillators
attached to the ground, with damping as a function of oscillator frequency f, for a given ground

motion characterized by an acceleration time-history, Xg . If we denote X j as the absolute


B-7

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

acceleration response of an ground mounted oscillator with frequency fj and damping j , then
the spectral acceleration is given by:
SA gj = max X j

Equation B-3

We assume that sufficient number of modes are included in the modal representation of the
structure such that:

m - j (jrnj ) =

Equation B-4

where || < 0.1 and where m = 1, if the response at node n is in input direction r, and m = 0, if
the response direction is cross-axis to input direction r.

Then the absolute acceleration response for structure node n, Yn , may be obtained using modal
analysis as:

Yn = j Y jn j jnj X j

Equation B-5

The floor response spectrum is a plot of the spectral acceleration for a set of SDOF oscillators
attached to the structure at node n, with damping k , as a function of oscillator frequency, f, for

a given floor motion characterized by an acceleration time-history, Yn . If we denote the

absolute acceleration response of the floor mounted oscillator with frequency fk as Znk , then

the spectral acceleration is given by


SAnk = max Znk

Equation B-6

If the floor input motion is expressed as Yn = j jnj X j , then each mode component may be

considered as an independent input to the floor oscillator, and the contribution of each mode
component to the floor oscillator response can be considered as the response of two cascaded
..

SDOF systems. Using the notation of Crandall and Mark (Reference B5), X 1 is the absolute
acceleration response of the first stage with frequency fj and damping j and

..

X2

is the absolute

acceleration response of the second stage frequency fk and damping k . Now, given that the
base input motion for the first stage is characterized as White Noise (WN), the
Root-Mean-Square (RMS) response of the first and second stage may be obtained from the WN
results presented in Crandall & Mark for a two-SDOF cascade. Denoting the first stage response

B-8

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling
..

..

as X 1RM S , and the second stage response as X 2RM S , the functional relations presented by
Crandall and Mark may be utilized to obtain an amplification factor which compares the
uncoupled response of the floor oscillator to the structure response at the point of contact. We
denote this amplification function as

..

X 2RMS fk
AF = ..
, ,
f j k
j
X1RMS

Equation B-7

Using the notation,


..

Z nkj = Pnk x 2 RMS

Equation B-8

Ynj = jnjSA gj = Pgj x 1RMS


..

Equation B-9

where Pnk and Pgj are Peak Factors introduced by Vanmarke (Reference B3). Using
Equations (B-7), (B-8), and (B-9), the modal floor response component may be expressed in
terms of the modal structure response as:

Znkj = AF jnj (Pnk / Pgj ) SA gj

Equation B-10

Vanmarke showed that the peak factors, P, corresponding to a given exceedance level (such as
84%) may be considered, in general, as approximately constant for a damped oscillator over the
frequency range 5-50 Hz. Vanmarke also showed that the ratio Pk / Pg was approximately 0.8.
We note that the order of this approximation is not relevant to the scaling procedure, since we are
considering ratios of estimated Znk for the original and reduced ground motion spectra.
Thus, considering the input motion to be in direction, r, and the response at node, n, in direction,
i, we use the notation

Znirj {fk } = AF{fk / fj , j , k }rjnij 0.8 SA gr {fj }

Equation B-11

for the response component of the floor oscillator with frequency, fk , due to the structure mode
with frequency, fj , and modal response factor, rjnij .
Then, the total response may be estimated using the SRSS modal summation,

B-9

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

Znir {fk } = j Znirj


1/ 2

Equation B-12

where the summation is over all of the modes used in the modal analysis of the structure, and the
SRSS spatial summation gives the final spectral acceleration estimate,

Zni {fk } = r Znir


1/ 2

Equation B-13

where the summation is over the three translational input directions.


B.3.3 Incoherence Reduction of Selected Reference Locations
For the example presented, the modal frequencies, fj , associated participation factors, rj , and
translational mode shape values, nij , for the nodes with spectra to be scaled are utilized. This
modal information, along with the scaling procedure and modal estimation method developed
above, was then used to reduce the high frequency regions of the unscaled floor spectra for the
selected reference locations. The computation of the ratios of the estimated response quantities,
using the modal response weighting factors and the amplification factors developed above, is the
bulk of the scaling effort. The method of response estimation detailed in Section B.3.2 can be
incorporated into several spreadsheets, which are linked by an application specific Visual Basic
encoded procedure. This procedure utilized the reduced ground motion spectra, the set of modal
response weighting factors, rjnij , and the associated mode frequencies as given quantities. The
spectral ordinates for each oscillator frequency, consistent with the set of frequencies for each
building floor spectrum, is then computed using Equations (B-7), (B-10), and (B-12). The ratios
of the estimated spectral response quantities are then computed as the appropriate set of
reduction functions to be applied to the spectra obtained from the analysis for the reference
building locations selected.
In this example, a 1.0E-5 UHS and compatible time-histories, were originally used in
development of floor response spectra. Later studies indicated that 85% of the UHS was more
appropriate as the site specific RLE. Therefore an additional reduction of the spectra by the 0.85
factor, as indicated in Equation (B-13) was utilized.
RLEINni = 0.85 x SAINni = 0.85 x SAni x (SAEINni / SA eEni )

Equation B-14

An example of the plotted results is shown in Figure B-6, which provides an unscaled RLE
spectra, the computed incoherence reduction functions, and the resulting incoherence reduced
RLE spectra for Node 162610 in the Building E model. Note that the seismic experience based
SQUG Reference Spectrum peak S A screening capacity of 1.2g from Reference B6 is
superimposed on the spectra for comparison. The 1.2g SA peak of the Reference Spectrum is

B-10

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

extended on the basis that higher frequency acceleration is less damaging than low frequency
acceleration of the same value.
Building E- Unscaled Spectra - Node 162610
3

5% Damping

0.85 x E-5 Input Motion

2.5

Comp 1 x 0.85
Comp 2 x 0.85
Comp 3 x 0.85
Ref Cap (= 1.2 g)

2
SA, g
1.5
1
0.5
0
0

10

20

30

40

50

Frequency, Hz

Building E - Foundation Reduction


Incoherence Reduction
1
0.9
0.8
0.7
Reduction 0.6
Factor
0.5
0.4
0.3
0.2
0.1
0

RX
RY
RZ

10

20

30

40

50

Frequency, Hz

Building E - Reduced Spectra - Node 162610


Incoherence Reduction

142 m Foundation
2.5

5% Damping

0.85 x E-5 Input Motion

Comp 1 x rX x 0.85
Comp 2 x rY x 0.85
Comp 3 x rZ x 0.85
Ref Cap (= 1.2 g)

SA, g 1.5
1
0.5
0
0

10

20

30

40

50

Frequency, Hz

Figure B-6
Incoherence Reduced Spectra for Building E Node 162610

B-11

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

B.3.4 Total Spectra Incoherence Reduction for All Locations


Comparison of the reduction functions computed for each subset of reference nodes in a given
response direction indicates that, in general, the variation of reduction factor amplitude with
frequency is similar for the selected representative sample of nodes within a building. An
example of a set of incoherence reduction functions obtained for Building E in one horizontal
response direction is shown in Figure B-7. In general, for all buildings, if the set of reduction
functions in a given direction is viewed as a statistical sample, the maximum coefficient of
variation of the sample is less than 0.1. For SMA, the sample mean plus one standard deviation
level can be utilized as a generic reduction function for the directional response of the building.
If all of the response spectra for a given building direction were reduced for the total effects of
foundation incoherence using such a generalized set of reduction factors, the resulting set of
floor response spectra would have only a 16% chance of being exceeded, which would insure
that equipment demand is, at least, taken at the 84% nonexceedance level. For SPRA, the mean
value of the reduction function can be used and the COV can be incorporated into the structural
response factor uncertainty.
An example set of mean plus one standard deviation total reduction functions, based on the
representative sample selected for Building E is given in Figure B-8.
Build ing E Inco herence R eduction - Component X
0.9

0.89

0.88

R s , Reduction Function

0.87

0.86

RX 113121

0.85

RX 162610
RX 202610

0.84

RX 2197
RX 233607

0.83

RX 3290
RX 4506

0.82

RX m+1s

0.81

0.8
0

10

15

20

25

30

35

40

45

Frequency, Hz

Figure B-7
Incoherence Reduction Functions for Selected Nodes of Building E

B-12

50

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

B uild ing E R eduction Factors


In cohe rence Redu ction
1

0.9

0.8

0.7

Redu ction Factors

RX m+1s
RY m+1s

0.6

RZ m+1s

0.5

0.4

0.3

0.2

0.1

0
0

10

15

20

25

30

35

40

45

50

Frequency, Hz

Figure B-8
Overall Incoherence Reduction Factors for Building E

B.4

High Frequency Reduction of Floor Spectra Due to Ductility Effects

High frequency equipment components, mounted within structures that have floor spectra with
significant high-frequency input energy, can have response levels that are significantly less then
would be predicted by conventional elastic or equivalent static analysis procedures. The
presence of even limited ductile energy absorbing capacity in component attachments, such as a
weld, can lead to substantial reductions in response due to non-linear behavior. As an example,
consider a small fillet weld with a yield displacement on the order of 0.001 inch (0.0254 mm)
and the displacement at weld failure being 0.01 inch (0.254 mm). A 0.4g response at 25 Hz will
require a displacement of 0.0063 inch (0.159 mm) while a 0.4g response at 5 Hz will involve a
displacement of 0.157 inch (3.99 mm). The 20 Hz response level is within the ductile range of
the weld while the 5 Hz response level has failed the weld. This example illustrates the so-called
brittle behavior of welded anchorage associated with low frequency response while the high
frequency response of the same anchorage is achieving ductile behavior. In Reference B7 the
effects of ductility on high frequency response has been explored in great detail.
In order to develop a procedure for a reducing a floor spectrum for high frequency ductility
effects, Reference B7 considered a squat item of equipment that is controlled by weld anchorage
capacity subjected to a pure base shear as the limiting case. It was concluded in Reference B7
that it is conservative to use only a single-degree-of-freedom model to obtain response spectrum
reduction factors. In terms of physical description, this conservative model corresponds to an
electrical cabinet that is anchored at its base by a minimum 3/16-inch (4.8 mm) fillet weld loaded
in the transverse direction. The equivalent length of the weld is taken to be 1 inch (25.4 mm)
with a yield displacement of 0.001 inch (0.0254 mm) and ultimate displacement of 0.01 inch
(0.254 mm) based on published test data. For this limiting example it was also conservatively
B-13

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

assumed that there is no nonlinear response in the cabinet structure, in the connections mounting
the electrical devices to the cabinet structure, or in the devices themselves. This case was
denoted as the simplified sliding model in Reference B7 for which a seven-step iterative
procedure was developed to obtain a reduced spectral ordinate at a given frequency. The
procedure developed in Reference B7, reduces the ground spectrum rather than directly reduce
an individual floor spectrum for ductility effects. This reduced level of input motion does not
represent an actual reduced ground motion but is rather a pseudo ground motion or Damage
Consistent Motion which yields the ductility reduced floor motion when the reduced input is
applied to the elastic building analysis model. The amplification of the building is included in
the sliding model procedure by requiring a larger input scale factor, FSM , to be applied to the
pseudo ground motion input level. The reason for utilizing this computational artifice is to allow
the estimation of ductility reduced floor spectra using the elastic structure analysis model and
direct spectra estimation methods. As a result, the scaling procedure developed for incoherence
reduction of ground motion may also be used for estimation of high frequency ductility reduced
floor spectra.
The sliding model procedure of Reference B7 was applied with the following basic guidance:
(a) the reduction is performed for frequencies 10 Hz and above; (b) the reduced pseudo ground
response spectrum should be connected to the elastic spectrum at 8 Hz; (c) a spectral ordinate
should not be reduced below a value equal to the peak spectral value at 10% damping divided by
1.6 or below the response level associated with just obtaining the yield displacement; and (d) for
SMA evaluations the factor of safety for equipment mounted within a building should be taken
as FSM = 3.0. The focus is on a spectrum reduction associated with a HCLPF capacity level
rather than a median capacity level. The initial input ground motion applied to the base of the
simplified sliding model is the incoherence reduced ground motion, since effects of foundation
incoherence result in a reduction in actual motion input both to the structure and the equipment
mounted therein.
The reduction procedure detailed in Reference B7 was incorporated into an interactive
spreadsheet that implemented the iterative procedure for the building. The combined
incoherence and high-frequency ductility-reduced pseudo-ground-motion was then obtained as a
product of the UHS acceleration and building reduction function for the value of modal damping
used in the corresponding building analysis.
SAHFG (f , ) = (

Rs

( )
F ) SAg f ,

Equation B-15

Figures B-9 and B-10 provide a comparison of the results of both the horizontal and vertical high
frequency ductility pseudo ground motion or damage consistent reduction with the unreduced
ground spectra and the incoherence reduced ground spectra for Building E.

B-14

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

Building E - Horizontal Ground Motion Spectra


0.4
4.5% Damping
0.35

0.3

SAh

E-5 Input Motion

SAh * Rs

Incoherency Reduced Motion

SAh*Rs/Fu

Damage Consistent Motion

SA, Hz

0.25

0.2

0.15

0.1

0.05

0
0

10

15

20

25

30

35

40

45

50

Frequency, Hz

Figure B-9
Reduced Horizontal Ground Motion Spectra for Building E

Building E - Vertical Ground Motion Spectra


0.35
4.5% Damping
E-5 Input Motion

SAv

0.3

Incoherency Reduced Motion

SAv * Rs
SAv*Rs/Fu

0.25

Damage Consistent Motion

SA, g

0.2

0.15

0.1

0.05

0
0

10

15

20

25

30

35

40

45

50

Frequency, Hz

Figure B-10
Reduced Vertical Ground Motion Spectra for Building E

B-15

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

B.5 Estimation of Floor Spectra Compatible with High Frequency Ductility


Reduced Pseudo Ground Motion
B.5.1 Damage Consistent Scaling of Floor Spectra
The basic scaling procedure developed for incoherence reduction can also be used to reduce each
floor spectrum SA ni , at node n in (translational) direction i, for high frequency ductility effects.
The damage consistent, or high frequency ductility, scaled floor spectra is obtained in two
scaling steps:
SA eHFni
SA HFni = SA ni x
SA eIN
ni

SA eINni
x
SA eE
ni

Equation B-16

where SA HFni are the scaled floor spectra which account for both foundation Incoherence and
High Frequency ductility. In the above equation, SA eEni , SA eINni , and SA eHFni are the estimated
floor spectra that would result from the Elastic UHS ground spectrum, Incoherence reduced
ground spectrum, and the High Frequency ductility reduced pseudo ground spectrum,
respectively. It should be noted that SA HFni accounts for both effects while the spectra obtained
using Equation (B-2), SAINni ,account for incoherence only. The reason for utilizing the two step
approach is that the high frequency ductility reduced pseudo ground spectrum is developed from
the incoherence reduced ground spectrum. The ratios within the brackets of Equation B-15 may
be identified as reduction functions associated with either incoherence reduction, high frequency
reduction or both reductions combined.
B.5.2 Damage Consistent Reduction of Selected Reference Locations
The same spectral estimation procedure outlined in Section B.3.2 can be used to obtain the
estimated floor spectra, SA eHFni , associated with the pseudo ground motion spectra developed for
each building to account for high frequency ductility effects. The ratios of the estimated spectral
response quantities are then computed as the appropriate set of reduction functions to be applied
to the spectra obtained from the analysis for the reference building locations selected in
Section B.3.4. As indicated in Section B.3.4, the basic scaling procedure is to factor the floor
spectrum, by the ratio of the estimated floor spectra. In this example a site specific RLE, was
taken as 0.85 x UHS and is applied directly in the scaling equation
SA eHFni
RLEHFni = 0.85 SAHFni = 0.85 SA ni
SA eIN
ni

SA eINni

SA eE
ni

Equation B-17

The quantity, SA eINni / SA eEni , is the incoherence reduction factor which is the ratio of the
estimated floor spectra that would result from the INcoherence reduced ground spectrum and the
estimated floor spectra that would result from the Elastic ground spectrum. The quantity,
B-16

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

SA eHFni / SA eINni , is the additional reduction factor for high frequency ductility which is the ratio
of the estimated floor spectrum that would result from the High Frequency ductility reduced
ground spectrum and the estimated floor spectra that would result from the INcoherence reduced
ground spectrum. As noted above, SA HFni accounts for both effects and SAINni accounts for
incoherence only. The spectra reduced for foundation incoherence may be computed as an
intermediate step since SAINni values are used for relay evaluations. SA HFni is used for screening
against the extended SQUG Reference Spectrum, Reference B6, or the SMA screening tables in
Reference B8 and for equipment anchorage calculations.
An example of the plotted results of this example is shown in Figure B-11, which provides an
unscaled RLE spectra, the combined incoherence and high frequency ductility reduction
functions, and the resulting overall reduced RLE spectra for Node 162610 in the Building E
model.

B-17

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

Building E- Unscaled Spectra - Node 162610


3

5% Damping

0.85 x E-5 Input Motion

2.5

Comp 1 x 0.85
Comp 2 x 0.85
Comp 3 x 0.85
Ref Cap (= 1.2 g)

2
SA, g

1.5
1
0.5
0
0

10

20

30

40

50

Frequency, Hz

Building E - Foundation and Damage Consistent Reduction


Incoherence and High Frequency Ductility
R d ti

1
0.9
0.8
0.7
Reduction 0.6
Factor
0.5
0.4
0.3
0.2
0.1
0

RX
RY
RZ

10

20

30

40

50

Frequency, Hz

Building E - Reduced Spectra - Node 162610


142 m Foundation
1.6

Incoherence and High Frequency Ductility


R d ti

5% Damping

1.4

0.85 x E-5 Input Motion

1.2
SA, g

Comp 1 x RX x 0.85
Comp 2 x RY x 0.85
Comp 3 x RZ x 0.85
Ref Cap (= 1.2 g)

1
0.8
0.6
0.4
0.2
0
0

10

20

30

Frequency, Hz

Figure B-11
Overall Reduced Spectra for Building E Node 162610

B-18

40

50

EPRI Proprietary Licensed Material


Development of In-Structure Response Spectra for Seismic Margin or Seismic PRA Evaluation by Scaling

B.6

References

B1. A Methodology for Assessment of Nuclear Power Plant Seismic Margin, (Revision 1),
EPRI NP-6041-SL, EPRI, Palo Alto, California, August 1991.
B2. Biggs, J.M., Seismic Response Spectra for Equipment Design in Nuclear Power Plants,
Proceedings 1st International Conference on Structural Mechanics in Reactor Technology,
Paper K4/7, pp. 329-343, 1972.
B3. Vanmarcke, E.H., Structural Response to Earthquakes, Chapter 8 of Seismic Risk and
Engineering Decisions, C. Lomnitz and E. Rosenbleuth, Editors, Elsevier, 1976.
B4. Igusa, T. and Der Kiureghian, Dynamic Characterization of Two-Degree-of-Freedom
Equipment-Structure Systems, Journal of the Engineering Mechanics Division, ASCE,
Vol. 111, no. 1, pp1-19, January 1985.
B5. Crandall, S.H. and Mark, W.D., Random Vibration in Mechanical Systems, Academic
Press, New York, New York, 1963.
B6. Seismic Qualification Utility Group (SQUG), Generic Implementation Procedure (GIP)
for Seismic Verification of Nuclear Power Plant Equipment, Revision 2A, March 1993.
B7. Analysis of High-Frequency Seismic Effects, EPRI TR-102470, EPRI, Palo Alto,
California, October 1993.
B8. Methodology for Developing Seismic Fragilities, EPRI TR-103959, EPRI, Palo Alto,
California, June 1994.

B-19

EPRI Proprietary Licensed Material

C
ESTIMATION OF EQUIPMENT CAPACITY BASED ON
EARTHQUAKE EXPERIENCE DATA

Earthquake experience data was used to develop the SQUG Reference Spectrum (Reference C1)
which is implied to be a conservative capacity spectrum for verification of seismic adequacy.
This appendix utilizes survival analysis to determine the distribution on capacity for use in
development of seismic fragilities.
Let the average spectral capacity of a given equipment class, defined as a 5% damped spectral
acceleration value averaged over the 3- 8 Hz frequency range, be represented by the random
variable C. The distribution of C is taken as log-normal with a known (assumed) log-normal
standard deviation, c , but an unknown log-normal mean, ln(C), where C represents the median
capacity.
Let the average spectral demand that the equipment class has been subjected to, defined as a 5%
damped free-field spectral acceleration value averaged over the 2.5- 8 Hz frequency range, be
represented by the random variable D. The distribution of D is taken as log-normal with a known
(assumed) log-normal standard deviation, D , but an estimated log-normal mean, ln(D), where D
represents the median demand.
Next we consider n independent equipment items from the equipment class, with known freefield spectral demand {D1 , --, Di , --, Dn} resulting in an average Reference Spectrum value,
Dave = RS . Each of the n items has survived the respective input motion represented by Di
without damage. Here, caveats (installation specifications) are used to define the equipment
class which exclude items with damage due to non-engineered attributes such as lack of
anchorage or inadequate restraint. For this evaluation we assume that the equipment has been
directly subjected to a level of mounting point motion equivalent to the free-field ground motion
without building amplification (or deamplification) effects. Since we are interested in obtaining
the highest estimate of capacity, we also assume that D is representative of the strong ground
motion which occurs within the epicentral region of a major earthquake.
If now we consider the ratio of capacity to demand for each of the n items, Ci / Di , we conclude
that all n ratios are greater than unity or,
Ci / Di > 1,

since no damage has been observed in any of the n equipment items belonging to the equipment
class. We also note that the ratio of spectral capacity to spectral demand , X = C/D, is a logC-1

EPRI Proprietary Licensed Material


Estimation of Equipment Capacity Based on Earthquake Experience Data

normal variable with mean, ln(X) = ln(C / D) , and log-normal standard deviation,
X =

{ (

)2 + (C )2 }

1/ 2

. The probability of failure for an item of equipment is given by

PF = P(X < 1) = F(X = 1) ,

where F is the cumulative distribution function (CDF) of X. If a reduced variate is defined as

u = ln(X) / x , u0 = ln(X) / x ,
then, given z = u u0 , we may write F(X) = (z) , where is the normal CDF. Thus

PF = F(X = 1) = P(u < 0) = P(z < u0 ) = (u0 )


The probability of survival for an equipment item is then,

Ps = 1 PF
Now, given n pairs of independent Di , Ci with known Di and average RS but unknown Ci , we
apply the constraint, Xi = Ci / Di > 1, since no failure has been observed in the n equipment
items. If the Xi are ordered such that, X1 < Xi < Xn , then the minimum probability of survival
is given by
P(Xi > 1) = i {1 F(Xi )}xi =1 = (1 PF )

Since C is unknown, it can only be specified by the assignment of a confidence coefficient. The
lower confidence limit on PF is found by considering the probability of an assumed failure for an
n+1 item of equipment. This probability of failure is taken as the confidence-level , , such that
the observed result of n cases of no failure is the best that could have occurred. Thus,

= 1 (1 PF )

n+1

is the probability of failure for at least one item given the survival of n items. Now we can
estimate the population mean, ln(X), which assures that, for a given level of confidence, , the
lowest capacity demand ratio of n equipment items will be greater than unity by requiring
PF = 1 (1 )

1/(n+1)

= ( u0 ) ,

or

u0 = 1 1 (1 )1 /(n + 1)

C-2

EPRI Proprietary Licensed Material


Estimation of Equipment Capacity Based on Earthquake Experience Data

Since u0 = ln(X) / X , we may write

X = C / D = euox
If the median demand, D, is estimated as D = Dave = RS , then the capacity associated with 95%
confidence is given by

C95 = RS euox
The High Confidence Low Probability of Failure (HCLPF), or 95% confidence of less than a 5%
failure, value is given by the 5% capacity level, or

CHCLPF = RS euox 1.645c = RS Fk


where we identify the factor, FK = euoX 1.645C , as the reduction or knockdown factor applied to
the Reference Spectrum to achieve a HCLPF capacity value.
Given D = 0.3 and C = 0.4 as representative log-normal standard deviations for spectral
demand and capacity, then X = 0.5 , and we obtain the following tabulation of capacity/demand
ratio for a confidence coefficient, = 0.95, or a 95% confidence-level, for equipment survival for
class group sizes ranging from 60 to 15.
n

PF

(-uo)

X = C/D

FK

60

0.047924

-1.66533

2.299

1.191

50

0.057048

-1.58005

2.203

1.141

40

0.070461

-1.47237

2.088

1.081

35

0.079847

-1.40611

2.020

1.046

30

0.092114

-1.32785

1.942

1.006

25

0.108830

-1.23277

1.852

0.959

20

0.132946

-1.11257

1.744

0.903

15

0.170750

-0.95121

1.609

0.833

From this table, we note that a class group size of 30 is the minimum number of items necessary
to demonstrate that the Reference Spectrum level represents a HCLPF capacity level for a unity
or greater knockdown factor.
The development outlined above provides an estimate of the population mean, ln(C), which, for
high levels of confidence, will be conservative (i.e., low) compared to the true population mean.
Our problem, as a set of n observations of no damage for the demand level recorded or estimated
for each observation, may be interpreted as a sample taken from a large population of equipment
meeting the attribute limits or caveats of the equipment class. We wish to infer an estimate of
the sample mean capacity, or ln(C), for which the conservatism is removed. This would then
C-3

EPRI Proprietary Licensed Material


Estimation of Equipment Capacity Based on Earthquake Experience Data

provide an estimate of the true median capacity of the equipment to be used in risk informed
seismic evaluations of equipment.
One method of achieving this capacity estimate is to consider the HCLPF values computed
above, RS FK , as one-sided lower tolerance limits based on the sample size and sample mean
value. This may be represented by

ln(Cnp ) = ln(C) knp


where Cnp is the lower tolerance limit , such that the probability is p that at least a proportion
lies below Cnp (or a proportion 1 lies above Cnp ) and where knp is the tolerance factor
based on p, , and sample size, n. In general, for the case of a known (or assumed) standard
deviation (Reference C2)
k np = 1() + 1(p) /(n)1 / 2

If p = 0.95 and = 0.05, and we identify, Cnp = CHCLPF = RS FK , then

(C / RS)tol = FK eknp
and the following tabulation is obtained using the prior results for FK :
n
60
50
40
35
30
25
20
15

FK
1.191
1.141
1.081
1.046
1.006
0.959
0.903
0.833

eknp
2.102
2.119
2.142
2.158
2.177
2.202
2.237
2.288

(C/RS)tol
2.503
2.418
2.317
2.258
2.190
2.113
2.021
1.907

The (C / RS)TOT and FK values obtained are based on an assumed c of 0.40. Changes in the
assumed c affect both the FK and (C / RS)TOT values. An increase in the assumed c , lowers
the HCLPF and increases the median capacity. Conversely, a decrease in assumed c increases
the HCLPF and decreases the median value. Thus, within a reasonable range of c , we would
not expect the unconditional failure rate of a derived fragility to vary by a significant amount.

C-4

EPRI Proprietary Licensed Material


Estimation of Equipment Capacity Based on Earthquake Experience Data

For comparison of the effect of changes to the assumed value of C the sensitivity on the results
is checked for n=30:
D
0.3
0.3
0.3

n
30
30
30

C
0.450
0.400
0.335

C
0.450
0.400
0.335

FK
0.978
1.006
1.047

X
0.54
0.50
0.45

(C/RS)tol
2.347
2.190
2.009

References
C1.

Generic Implementation Procedures (GIP) for Seismic Verification of Nuclear Plant


Equipment, Revision 2, Seismic Qualification Utility Group (SQUG), June 1991.

C2.

Hald, A., Statistical Theory with Engineering Applications, John Wiley & Sons, 1952.

C-5

EPRI Proprietary Licensed Material

D
EXAMPLE FRAGILITY FOR INSTRUMENT CABINET
DERIVED FROM EXPERIENCE DATA

In this example a fragility will be developed for an instrument cabinet located in Building E as
described in Appendix B. The instrument cabinet is welded to embeds and the anchorage and
base stiffness are deemed to be rugged relative to the demand. GERS are not available for
instrument cabinets, therefore seismic experience data is selected for derivation of a fragility.
The instrument cabinet has no electro-mechanical relays, computers, programmable recorders or
strip chart recorders and meets the caveats of the GIP, Reference D1.

D.1

Demand

For this example we will use the in-structure spectrum for node 162610 shown in Figure B-11 in
Appendix B. This in-structure response spectrum reflects three reductions in the original
Uniform Hazard ground motion spectrum. The original ground motion UHS for the area was
anchored to 0.12 g and was used to develop amplified in-structure response spectra. Subsequent
site-specific studies determined that 85% of the area spectrum was more appropriate as a median
ground motion input. The resulting pga was 0.102g as shown in the UHS spectrum, Figure B-3
of Appendix B.
The ground motion spectrum peaks at 20 Hz and further reduction can be taken for Ground
Motion Incoherence (GMI) and for High Frequency Ductility (HFD) effects. Since the
instrument cabinet does not contain any electro-mechanical relays, the high frequency ductility
reduction is appropriate to reflect the effective reduction in damage at high frequency
acceleration due to small amounts of ductility that may be experienced in the cabinet base. The
resulting spectra shown in Figure B-11 are the result of the three reductions and are considered to
be median centered response spectra relative to the ground motion input. Since the site was a
low seismic hazard site and the structures and equipment were expected to have HCLPF values
above the 1E-5 median UHS pga, the 1E-5 UHS spectrum was considered the best estimate
spectral shape to define the input motion.
The UHS were only defined out to1E-5/year frequency of occurrence. Peak ground acceleration
was defined beyond 1E-5/year so the fragilities are anchored to pga in lieu of spectral
acceleration.
In accordance with the procedure in Reference D4, the narrow banded demand spectrum can be
clipped to represent a more realistic demand on the instrument cabinet for comparison to the
broad banded seismic experience spectrum. Referring to Figure B-11, and the clipping
procedure in Reference D4, the following parameters and clipping factor are derived.
D-1

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

The parameter f08 is the frequency band width for the spectrum taken at 80% of the peak and is
about 2 Hz. The factor fc is the 14.5 Hz central frequency of the demand spectrum.

B = f0.8 / fc = 0.138
For B less than 0.2, the clipping factor, CC is:

CC = 0.3 + 0.86 B = 0.42


The peak spectral acceleration of the demand spectrum at 14.5 Hz is 1.5g. The clipped peak is
then:

SaC = 0.42(1.5) = 0.63g .


Uncertainty, U is defined in Reference D4 as:

U = 0.37 0.5 B = 0.30

D.2

Capacity

The instrument cabinet has not been tested nor is there any available data on similar models.
Therefore the capacity is determined from earthquake experience. Reference D1 provides a
Reference Spectrum that was determined to be applicable to generic classes of equipment and
subsystems. In support of using seismic experience for verification of new and replacement
equipment in newer non A-46 plants, the SEQUAL Owners Group has done an extensive
reevaluation of the seismic experience data and has presented the results in a topical report,
Reference D2. A weighting of database site spectra was conducted, accounting for the number
of representative examples at each database site. The resulting average of weighted database
spectra very closely matches the GIP Reference Spectrum in the 2.5 to 7.5 Hz range. The
comparison of the SEQUAL experience spectrum to the SQUG Reference Spectrum is shown in
Figure D-1 for instrument cabinets and panels. It closely matches the SQUG reference spectrum.
For purposes of defining capacity, the SEQUAL spectrum will be used.
There are 46 documented independent examples of instrument cabinets and panels in the
SEQUAL database. Using the survival analysis methodology described in Appendix C, the
median and 95% confidence capacity spectra can be derived. For 46 examples, the multiplier,
FK, is interpolated from Appendix C to be 1.117. Thus the 95% confidence spectrum is 1.117
times the SEQUAL spectrum. The median capacity multiplier, (C / RS)TOT , is 2.378.
The actual fundamental frequency of the instrument cabinet is not known. Based on the
guidance in Reference D3, the side to side and front to back fundamental modes would be
greater than 8 Hz for a cabinet with welded anchorage and stiff base. Internal panels could also
be expected to be above 8 Hz. The demand spectrum in Figure B-11 of Appendix B peaks at
14.5 Hz, thus some portions of the instrument cabinet could be in resonance with the demand.
D-2

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

Figure D-1
Equipment Class 20 Control and Instrumentation Panels and Cabinets

For structural type failures we would argue that the spectral acceleration capacity for high
frequency input motion is at least as great as for low frequency input motion due to the fact that
the displacements associated with high frequency are very much smaller than for low frequency.
However, since the instrument cabinet contains numerous devices that could be subjected to
cascading response for high frequency input and could fail in a brittle mode, we will consider the
spectrum, as derived by SEQUAL to be representative, throughout the frequency range, just as a
test response spectrum. At the 14.5 Hz peak of the demand spectrum, the SEQUAL 5% damped
spectral acceleration is about 0.82g. The median capacity is then:

Cm = (C / RS)TOT (0.82g) = 2.378(0.82g) = 1.95g


The 95% confidence capacity is:

C95% = FK (0.82g) = 1.117 (0.82g) = 0.92g


The uncertainty in the capacity is calculated from the ratio of the median capacity to the 95%
confidence capacity.

UC = (1/ 1.65) ln (1.95 / 0.92) = 0.46

D-3

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

D.3

Capacity Factor

The median value of the capacity factor FC , is the ratio of the median capacity to the median
demand.

FC = 1.95/0.63 = 3.10
The uncertainty in the capacity factor is derived from the uncertainty in the capacity and the
uncertainty in the clipped demand.
UFC = (0.30 2 + 0.46 2 )12 = 0.55

D.4

Structural Response Factor

There are several variables that contribute to randomness and uncertainty in development of the
in-structure response spectra that define the demand. The original development of in-structure
response spectra was conducted using a state of the art three dimensional model of the reinforced
concrete structure. Since the model was three dimensional, torsional coupling effects are
automatically included.
The original in-structure spectra were developed by mode superposition time history analysis
using three simultaneous independent time histories whos response spectra closely matched the
ground motion UHS. At the 14.5 Hz fundamental frequency that results in the narrow banded
peak of the in-structure spectrum, the time history spectrum was about 10% higher than the
target UHS spectrum.
The structure is robust relative to the demand and is stressed to less than half of the yield
capacity, therefore, a best estimate of structural damping of 4 % was used in the analysis.
The resulting in-structure spectra were then scaled to account for ground motion incoherence,
(GMI) and for high frequency ductility (HFD) effects using the random vibration methodology
described in Appendix B.
In most cases, the parameters used are considered to be median centered thus the response
factors are unity unless otherwise noted. The variables to be addressed in deriving the structural
response factor and it variability are:

Spectral Shape (SS)

Damping (D)

Modeling (M)
Frequency
Mode Shape

Mode Combination (MC)

D-4

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

In addition, there is some uncertainty associated with the scaling of in-structure spectra to
account for:

Ground Motion Incoherence (GMI)

High Frequency Ductility Effects (HFD)

Scaling using random vibration theory (RV)

D.4.1 Spectral Shape (SS)


At the frequency of interest, the response spectrum resulting from the time history input motion
was about 10% greater than the target UHS. In addition, as discussed in section 5.8, probabilistic
response invariably results in a lower peak response than a so called median centered
deterministic response. In Reference D4, a demand reduction factor of 0.92 is suggested to
account for this. The FOAKE study described in Section 5.8 showed a demand reduction of at
least 10% for the three models used on rock sites. For soil sites the demand reduction was much
larger. For the rock site in this problem, the demand reduction is considered to be 0.9, resulting
in a demand reduction factor of:
FDR = 1/0.9 = 1.11

If no reduction is considered to be about a 1% probability case (2.33 R case),

RDR =

1
ln 1.11 = 0.04
2.33

From Reference D4, Table 3-2, for a 14.5 Hz frequency of concern, the random variability, R ,
would be about 0.18. This represents the peak to peak variation in the actual earthquake
response spectrum as opposed to the smooth UHS. For fragilities anchored to pga, the
uncertainty on the spectral amplification at 14.5 Hz would be about 0.14.
In addition, the input motion is defined at the average of the two orthogonal components of
earthquake. One direction could have a peak greater than the other. The random variability, R ,
in the ratio of horizontal components is about 0.13. The resulting spectral shape factor and
variability are:
FSS = 1.1(1.11) = 1.22

RSS = (0.042 + 0.182 + 0.132)1/2 = 0.23


USS = 0.14

D-5

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

D.4.2 Damping (D)


Structural damping used in deriving the in-structure spectra was 4 %. For structures at less
than yield, Regulatory Guide 1.60 would recommend 4% for reinforced concrete. Reference
D4 recommends 5% as a median value for reinforced concrete with considerable cracking.
Considerable cracking at less than yield was not expected for the building E structure and a
compromise median damping value of 4 % was used as a best estimate.
FD = 1.0

Reference D4 suggests that 3% is a 1U value for damping. In the amplified portion of the
UHS, we can estimate the uncertainty is spectral acceleration due the uncertainty in damping as
the logarithm of the square root of the damping ratio.

UD = ln (4.5/3)1/2 = 0.2
D.4.3 Modeling (M)
There are two variables to consider under modeling, structural frequency and mode shape. The
model is considered to be a best estimate of the structural load path from roof to foundation. Per
Reference D4, when code properties are used for concrete, the calculated and actual stiffness are
similar, thus we consider the calculated fundamental frequency to be median centered. For a
detailed model such as the one being considered, Reference D4 suggests that the uncertainty, f ,
on frequency is about 0.15. The -1 frequency is then:
f1 = e0.15 (14.5) = 12.48 Hz

The UHS ground motion response spectrum shown in Figure B-3 is fairly flat in this region and
the change in spectral acceleration is only about 3%.

Uf = ln (1.03) = 0.03
For complex structures, Reference D4 recommends that the uncertainty on mode shape,
UMS = 0.15.
The resulting uncertainty in the modeling is then:
UM = (0.03 2 + 0.15 2 )1 / 2 = 0.15

D.4.4 Mode Combination (MC)


Mode superposition time history analysis was conducted. The combination of modal responses
by SRSS was considered to be median centered. From the response spectra in Figure B-11, the
D-6

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

dominant direction X spectrum appears to be dominated by a single mode. Some contribution


from higher modes would be present but not very influential for this case. The mode
combination randomness is small and is estimated to be:

RMC = 0.05
D.4.5 Ground Motion Incoherence (GMI)
From Appendix B, the reduction factor for ground motion spectral acceleration at 14.5 Hz is
about 0.87. If we consider the no reduction case to be an upper limit on input motion of about
3, the uncertainty for GMI is:

UGMI = (1/3) ln (1/0.87) = 0.05


D.4.6 High Frequency Ductility Reduction (HFD)
From Appendix B, the high frequency reduction factor for the ground motion spectrum at 14.5
Hz is about 0.7. With no reduction being an extreme 3 limit,

UHFD = (1/3) ln (1/0.7) = 0.12


D.4.7 Scaling Using Random Vibration Theory (RV)
The scale factors developed using random vibration theory are considered to be quite accurate
since the scale factors were developed from a ratio of response between the original and reduced
ground motion spectra rather than comparing a single response analysis for the reduced ground
motion spectrum to the original response analysis. Some uncertainty is of course present in any
analysis and this uncertainty in the method of scaling is estimated to be:

URV = 0.10
D.4.8 Structural Response Factor (FRS)
All variables for structural response were unity except for spectral shape ( FSS = 1.21).

FRS = 1.21
= SRSS of individual s
= SRSS (SS , D , M, MC , GMI, HFD , RV )

D-7

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

RRS = (0.23 2 + 0 + 0 + 0.052 + 0 + 0 + 0)1 / 2 = 0.24


URS = (0.14 2 + 0.20 2 + 0.152 + 0 + 0.052 + 0.12 2 + 0.10 2 )1 / 2 = 0.33

D.5

Fragility

The median capacity is the product of the capacity factor, structural response factor and the pga
of the ground motion spectrum.

A m = FC FRS (zpa)
A m = 3.10(1.21)(0.102g) = 0.38g

R = (0 + 0.24 2 )1 / 2 = 0.24
U = (0.55 2 + 0.33 2 )1 / 2 = 0.64
HCLPF = 0.38g(e)1.65( 0.24 + 0.64) = 0.09g
Note in this case that the median and HCLPF capacity derived from seismic experience data are
quite low. The reasons for such a low median capacity and HCLPF are several fold. First of all,
the ground motion spectrum that was the basis for the structural response analysis has unusually
high amplification of spectral acceleration relative to the zero period acceleration. Examining
Figure B-3, the unreduced horizontal UHS input spectrum has a peak amplification of about 4 at
4 percent damping. A Regulatory Guide 1.60 spectrum has a peak spectral acceleration
amplification of about 3.13 at 5% damping and a NUREG/CR-0098 median spectral shape has
an amplification of 2.12 at 5% damping. Typical amplifications for central and eastern US
uniform hazard spectra are less than 2.5. The derivation of the ground motion spectrum was not
done in accordance with detailed logic tree modeling techniques as was done in the EPRI and
LLNL studies. A review of the derivation of the hazard implies that the spectrum was about an
th
80 percentile spectrum and is referred to as a Uniform Risk Spectrum. From the LLNL and
EPRI studies we dont see much difference in the spectral shapes between the 50th and 84th
percentile UHS, so for purposes of an illustrative problem, the given shape was assumed to be a
median shape even though it appears to have unreasonable spectral amplification of zero period
acceleration. In reality, the spectral shape factor derived should be at least 60% higher but,
without redoing the hazard study, using current methodology, there is no analytical basis to alter
the given shape. Second, the structure is very stiff, is founded on rock, and has low damping, so
the amplification of high frequency input is large and the peak of the in-structure response
spectrum exceeds the peak of the experience based spectrum. Third, there are only 46 official
examples in the database that keeps the statistical capacity lower than if there were more well
documented samples. Fourth, the uncertainty is very large due to the large uncertainty in the
capacity itself and the uncertainty associated with peak clipping and scaling of response. Thus
the ratio between the HCLPF and median capacity is greater than 4.

D-8

EPRI Proprietary Licensed Material


Example Fragility for Instrument Cabinet Derived from Experience Data

In this particular case, the seismic hazard in terms of pga is very low and if convolved with the
fragility, the resulting unconditional failure rate would be reasonably low and a fragility derived
in this manner would be acceptable. For rock site cases at locations high in the structure, where
high in-structure spectra define the demand, experience based fragility derivations may not be
useful for defining seismic fragility.

D.6

References

D1. SQUG, Generic Implementation Procedure (GIP) For Seismic Verification of Nuclear
Power Plant Equipment, Revision 2, Corrected, Seismic Qualification Utility Group, June
28, 1991.
D2. SEQUAL, Topical Report, Basis for Adoption of the Experience-Based Seismic
Equipment Qualification (EBSEQ) Methodology by Non-A46 Nuclear Power Plants,
SEQUAL Owners Group, April 2001.
D3. EPRI TR-102180, Guidelines for Estimation or Verification of Equipment Natural
Frequency, EPRI, Palo Alto, California, March 1993.
D4. EPRI TR-103959, Methodology for Developing Seismic Fragilities, EPRI, Palo Alto,
California, June 1994.

D-9

EPRI Proprietary Licensed Material

E
DEVELOPMENT OF GENERIC FRAGILITY
DESCRIPTIONS FOR PURPOSES OF SCREENING
BASED UPON DESIGN CRITERIA

E.1

Establishment of Screening Level

It is not practical to calculate fragilities for all components that are included in the risk modeling.
Most components and distributive systems are inherently rugged and can be screened out on the
basis that their seismic induced failure rate is low in comparison to the items that will ultimately
dominate seismic risk. It is desirable to establish a fragility target where components exceeding
this target may be screened out.
In developing the target, three variables must be considered; seismic hazard, uncertainty in the
median fragility and frequency of failure (potential core damage) relative to that for other events.
A fourth variable, consequence of failure, is important, but for purposes of establishing a
fragility cut off it is assumed that all failures have equal consequence. Parametric studies were
conducted using the hazard and candidate fragility curves as input variables and examining the
resulting failure frequencies. The example fragility descriptions were convolved with the
seismic hazard to compute mean seismic failure rates. The example cases were then studied to
determine an acceptable cutoff fragility level.
E.1.1 Seismic Hazard
Uniform hazard spectra were defined which describe the spectral accelerations at different
frequencies for different return periods. The peak ground acceleration vs frequency of
occurrence was provided up to 1.0g. NUREG-1407 states that the seismic hazard must be
carried out to 1.5g unless sensitivity studies can show that a lower cutoff is justified. In the study
to determine a target screening level fragility described in Section 1.3, the pga hazard was
extrapolated to 1.5g and cases were run for 1.0g and 1.5g cutoff. For this particular site, the
hazard was reasonably high having a mean 10,000 year return period pga of close to 0.3g. For
components with HCLPFs in the 0.3g range, the extension of the hazard did not make a
significant difference since the range of HCLPF to median capacity was enveloped by the hazard
up to 1.0g. At the fragility level that was ultimately determined to be an acceptable screening
level, there was enough difference between the 1.0 and 1.5g cutoff results that the screening level
decision was based on a 1.5g seismic hazard cutoff.

E-1

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

E.1.2 Uncertainty in the Median Fragility


The uncertainty range for fragilities varies with the failure mode. For ductile modes of failure,
such as for structures or piping, the margin to failure relative to code allowable is much larger
than for brittle or functional failure modes but the uncertainty is also larger so that a dual criteria
must be implemented to establish a minimum value of the median capacity and of the HCLPF.
HCLPF is defined mathematically as 95% confidence of less than 5% probability of failure. For
the double logarithmic fragility curve typically used, the median capacity is denoted as Am, the
randomness as R , and uncertainty as U , where the s are logarithmic standard deviations. The
HCLPF may be computed from
HCLPF = Am exp (-1.65) (R + U )

For ductile failure modes of flexible systems, such as for structures piping, cable raceways, etc.
the ratio of median to HCLPF is typically three or greater. For brittle failure modes of rigid
equipment or functional failure modes such as relay chatter, the ratio of median to HCLPF tends
to be less than three but greater than two. Thus, for the same seismic failure rate, the flexible,
ductile items must have a higher median but may have a lower HCLPF than for a non-ductile
failure mode.
The cases conducted to determine the fragility level for screening revealed that the seismic
failure rate is more sensitive to HCLPF than median. Often it is more convenient to estimate or
compute a deterministic HCLPF for making decisions on screening, hence the final screening
value for fragility was targeted to a HCLPF value, wherein the median value is implied,
depending upon the failure mode. Establishment of a HCLPF above the screening target was the
approach used exclusively for screening of structures and most flexible equipment.
E.1.3 Target Failure Rate
Internal event core damage frequencies typically are on the order of 1.0E-5/yr. Seismic induced
core damage frequencies for higher seismic zones such as the one considered here may have
similar CDF. It is not practical to set the screening level too high (very low failure rates) or else
nothing can be screened. On the other hand, we would not want to screen out seismic failures
that could contribute more than 10% of the expected CDF so the target for screening was set at
1.0E-6/year or less for seismic failure rate for a seismic hazard extended to 1.5g pga. The 1E-6
per year screening target is comparable to screening targets for other external events set at 1E-6
per year CDF in USNRC (1991b). Also in Regulatory Guide 1.174 addressing Risk-informed
changes to the licensing basis, a change in licensing basis that increases CDF by less than 1E-6
per year would be considered. The concern in this case though is if surrogate fragilities for
several screened out components appear in and gates in the SPRA model, the cumulative
contribution to CDF from screened out components could be significantly higher than 1E-6 per
year.

E-2

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

Applying the screening target above and the results of the several analyses conducted, the
following approximate fragilities were determined to be the threshold for screening out
components:
Characteristics of Item

Am, g

HCLPF, g

Med/HCLPF

Flexible, ductile structures or equipment

1.5

0.5

3.0

Brittle or functional modes of failure

1.22

0.57

2.14

These two cases result in seismic failure rates of approximately 1.0E-6/yr for a 1.5g hazard
cutoff. These values were recommended as surrogate fragilities for inclusion in fault trees whose
top event terminates at a branch in an event tree.
In the actual study, the use of these values for surrogate elements resulted in a significant but not
dominant contribution to CDF from the surrogates. The target for screening should theoretically
have been set at a higher level. This would have resulted in significantly more effort to review
each seismic qualification report. Engineering studies are always a comprise between cost and
refinement in the results. In this case, the weakest links in the plant were identified, which was
the objective of IPEEE, but most of the plant structures, systems and components were
enveloped by the surrogate fragility elements.

E.2

Development of Demand on Components

The site under consideration was a soil site. The original design bases DBE response spectrum
resembled a NUREG CR-0098 median spectrum anchored to 0.25g. The UHS developed for the
site was narrower banded, peaked at about 5 Hz, had a higher pga and higher spectral
amplifications at 10,000 year return period than the DBE spectrum. It was determined that a new
median centered SSI analysis would be necessary and that a probabilistic analysis would provide
the most realistic and favorable results.
Probabilistic response spectra were developed for several levels of ground motions, changing the
soil properties as the acceleration levels increased. The in-structure response spectra used for
screening were developed for a 0.5g pga UHS. Comparison of the probabilistic in-structure
response spectra for 0.5g pga UHS to the 0.25g DBE spectra revealed significant conservatism in
the DBE spectra.
Further conservatism in the design analysis process had to be quantified in order to reach the
screening target.

E.3

Screening Evaluation of Equipment and Distributive Systems

Most commercial equipment and distributive systems are inherently rugged as long as they are
adequately supported or anchored. When this same equipment is qualified for seismic loading in
an NPP environment, the qualification requirements and conservatisms in the specification of
E-3

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

seismic demand and calculation of equipment response compound to result in very large
margins. In Reference E1, screening guidelines are provided that, subject to a walkdown
verification of the equipment to search for vulnerabilities, would allow most equipment and
distributive systems to be screened for earthquakes up to 1.2g peak spectra acceleration. This
th
value is to be compared to the 84 percentile ground motion spectrum defined at 5% damping.
In Section E.1 screening thresholds were developed for equipment in terms of median and
HCLPF capacities. The first step in screening is to compare the 84th percentile UHS to the 1.2g
th
screening level of Reference E1. In this case, the 84 percentile input for a 10,000 year return
period was significantly greater than 1.2g so the screening tables in Reference E1 would not
imply a HCLPF of 0.5g or greater required for screening based on the risk contribution criterion
described in Section E.1.
Therefore, an approach was taken wherein quantification of the conservatism in the design
process was used in conjunction with the walkdown in order to screen out generic classes of
components and distributive systems or individual components. Comparisons of the 0.5g pga
median probabilistic floor response spectra to the original design floor response spectra for the
0.25g DBE reveals that they were comparable in the reactor building. This comparison
demonstrates that there is about a factor of two conservatism in the original seismic demand
specified for equipment. This factor is less for rigid equipment but, since the spectral
acceleration in the high frequency (rigid) regime of the floor spectra is much lower than for the
amplified region, the loads for rigid equipment are also low. A typical comparisons of the 0.25g
DBE and 0.5g probabilistic spectra is shown in Figure E-1. The spectra comparisons alone do
not demonstrate sufficient margin to screen out the components but they do make a significant
contribution to the process.

E-4

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

Figure E-1
Comparison of DBE Vs Probabilistic Response Spectra Reactor Building El. 547

E.4 Screening of Flexible Equipment and Distributive Systems Designed


by Analysis
Flexible equipment and distributive systems that are designed by analysis usually have ductile
failure modes but for purposes of screening it is assumed that the failure mode is non-ductile
such as for failure of anchor bolts or welds or buckling. Fillet welds are shown in Reference E1
to have as much larger margin than implied by the design codes so fillet welds should not be the
basis for the screening computation. The following assumptions are made for the screening
calculation:

Probable frequency range is 3-10 Hz (flexible equipment).

2% damping was used for the design whereas 5% is considered median with 2% defined as a
-2 U case.

The screening is only applicable to locations outside of the primary containment where the
effects of hydrodynamic loads are minimal.

Code margins are those inherent in the ASME code where the allowable stress may be as
high as 70% of the specified ultimate strength, resulting in a nominal safety factor of 1.43.

E-5

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

The safety factors of expansion anchors and welds are greater so the ASME criteria governs
for the screening calculation.
The methodology used follows that described in Reference E2.
E.4.1 Strength Factor
The Strength Factor, Fs, is:
Fs =

FU N
DBE

FU is the ultimate strength (stress/load), N is the normal load or stress and DBE is the seismic
load or stress. Since the 0.5g UHS in-structure spectra are essentially equal to the 0.25g DBE instructure spectra in the 3-10 Hz frequency range, the basis for DBE stress is considered to be
anchored to 0.5g pga. Conservatively, consider that the only normal load effect is from weight
which usually is a low contribution to the ultimate load capacity. The median normal load is
assumed to be 10% of the ultimate capacity with a + 1 value equal to 20% of the ultimate
capacity. The median ultimate strength is about 1.1 times the code specified value. The code
specified value is set at the 95% confidence level, which is a -1.65 U value. It is assumed that
the average demand is 70% of the code allowable with 100% assumed as a 95% probability
value (+ 1.65 U ). The code allowable can be as high as 70% of the code specified ultimate
strength, therefore the median load/stress is (.7)(.7) = 0.49 of the ultimate capacity based on code
specified strength. The DBE load/stress is then (0.49-0.1) = 0.39 times the ultimate capacity.
The strength factor is then:

Fs =

1.1 0.1
= 2.56
0.39

Using the approximate second moment method from Reference E2 for calculating , the U is
computed to be 0.30.
E.4.2 Equipment Response Factor
The equipment response factor consists of the product of the individual factors for the variables
of Qualification Method, Damping, Modeling, Mode Combination and Earthquake Component
Combination.
Qualification Method
It is assumed that a dynamic response spectrum analysis was conducted as opposed to a more
conservative static coefficient method. There is no particular bias in the response spectrum
method but considerable conservatism can accumulate from several variables. In the case of
piping, often the envelope response spectra were used which is very conservative. However, if
we consider a single flexible component or a subsystem supported from the same elevation, this
E-6

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

conservatism cannot be counted. The practice of peak broadening and smoothing introduces
conservatism. A prior study of single degree of freedom systems revealed that the degree of
conservatism for spectra soothing in the frequency range of interest is about a factor of 1.2 with a
U of 0.09.

FQM = 1.2
U = 0.09
Damping
The factor of conservatism that results from using 2% damping in design vs 5% median is
quantified by:

FD =

S a2%
S a5%

The design spectra were used to find an average value of 1.17 for the spectral acceleration ratios
between 2% and 5% damping in the 3-10 Hz frequency range. If 2% damping is a -2 U value,
the U for damping is 0.08.
FD = 1.17

UD = 0.08
Modeling
Modeling error can arise from frequency error and mode shape difference between the model and
the actual response. The model would normally be median centered so the modeling factor
would be unity. The spectra peak at very low frequency, are fairly flat between 5 and 10 Hz and
have a factor of 2 difference from 3-5 Hz (Figure E-1). For the steepest slope between 3 and 4
Hz, a 1 f frequency shift results in a 25% difference in response. Thus:

Uf = ln(1.25) = 0.22
The response U due to mode shape error is estimated to be about 0.15. Combining s by SRSS,
the U for modeling is:
UM = (0.22 2 + 0.15 2 ) = 0.27

E-7

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

Mode Combination
Mode combination was by SRSS which is median centered. The response variability due to
mode combination is estimated as a R of 0.15 for multimode response of distribution systems.
Earthquake Component Combination
Earthquake components were combined by SRSS. Response is usually dominated by the two
horizontal directions. Considering the two horizontal components to be in phase as a 3 R case:

R =

1
ln 2 = 0.12
3

Equipment Response Factor Results


Combining the response factors as the product of the individual factors and the s by the SRSS
rule, the equipment response factor and its variability are represented by:
FRE = 1.4
R = 0.19

U = 0.29
E.4.3 Structural Response Factor
Spectra were developed by probabilistic methods using a Latin Hypercube simulation process in
which all important variables associated with structural response are included. The median
results were used to derive the strength and response factors so the structural response factor is
unity. The difference between the 50th and 84th percentile spectral accelerations in the 3-10 Hz
frequency range defines the composite variability, C . This ratio averages about 1.25 for the
reactor building so the C is 0.22. There is approximately equal variability from random and
uncertainty variables and the corresponding R and U are 0.22/ 2 = 0.16 each.
E.4.4 Fragility Description for Flexible Components Designed by Analysis
The median peak ground acceleration capacity is the product of the strength, equipment response
and structural response factors times the reference 0.5g peak ground acceleration for the UHS.
As previously shown the 0.5g in-structure response spectra for the UHS are equivalent to the
0.25g DBE spectra, hence 0.5g is used as the reference earthquake pga to represent the DBE
demand.
Am = 2.56 (1.4)(1.0)(0.5g) = 1.79g

E-8

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

The random and uncertainty variability are the SRSS of the Rs and Us for the three variables
and are computed to be:
R = .25

U = 0.45
The HCLPF is computed as:
HCLPF = 1.79g exp(-1.65) (R + U ) = 0.56g

This exceeds the 0.5g HCLPF and 1.50g median target set as the screening threshold in
Section E.1 and, considering the conservative assumptions used in the derivation, this class of
component and distribution system can be comfortably screened out subject to a walkdown
verification that there are no vulnerable looking details. This calculation was used to screen out
piping, cable trays and valves as well as the passive parts of instrument racks and electrical
distribution cabinets.
Valves are rigid but the piping systems in which they are mounted are flexible and the piping
response dictates the demand for the valves. Valve qualification data were reviewed and the
valves usually had a large design margin above the specified demand so the above derivation is
also considered applicable to valves with the exception of those that may be identified during the
walkdowns that appeared to be outside of the seismic experience database.

E.5

Components Qualified by Test

Reference E2 provides a methodology for developing fragilities for components qualified by test.
The median capacity is expressed as:
Am =

TRSC
FD FRS PGA
RRSC

where:
TRSC

= Clipped test response spectrum

RRSC

= Clipped required response spectrum

FD = Device capacity factor

FRS is the structural response factor and


PGA is the peak ground acceleration for development of the RRS

E-9

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria
TRSC = TRS (CT )(CI )

Where:
CT is set at unity for broad banded multiaxis testing
CI is a capacity increase factor and is recommended to be 1.1 with a U of 0.05
(Reference E2)
RRSC = RRS (CC )(DR )

Where:
RRS is the required response spectrum

CC is the clipping factor. Clipping is not applicable for this case since the spectral peaks
are at very low frequency relative to the equipment fundamental frequencies.
DR is a demand reduction factor if the RRS is calculated by deterministic means.
Probabilistic methods were used for the development of floor response spectra so DR is
unity.
FD is recommended in Reference E2 to be a value of 1.4 to demonstrate function during
the earthquake with R of 0.09 and U of 0.22. FD for function after the earthquake is
recommended to be 1.9 with R of 0.09 and U of 0.28.

FRS is the structural response factor which is unity for probabilistic spectra. The C is
computed as the logarithm of the ratio of the 84th percentile response to the 50th percentile
response at the frequency of the equipment. R and U are assumed to be equal and are
C / 2 . The s vary with the location of the equipment. For purposes of the generic
screening, an average ratio of 84th percentile to 50th percentile response of 1.25 was used
and the C was computed to be 0.22 with R and U equal to 0.16 each.

PGA is the peak ground acceleration of the earthquake record used to develop the RRS.
In this case the RRS was defined as the in-structure floor response spectra for 0.5g
median structural response.
Using the above methodology, the fragility may be expressed as:
Am = (1.4)(1.1)(0.5g)

E-10

TRS
TRS
= 0.77
g
RRS
RRS

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

R = 0.18

U = 0.28
HCLPF = 0.36

TRS
g
RRS

For the example shown the RRSs peak at quite low frequency (2 to 3 Hz) whereas the tested
electrical panels are greater than 6 Hz, thus the comparison of TRS and RRS must be made at the
panel frequency, not at the spectral peaks.
If the TRS and RRS are equal at the fundamental frequency of the equipment, the median
capacity for function during the earthquake is 0.77g and the HCLPF is 0.36g. Referring to the
target screening fragility for functional modes of failure, in order to screen out components with
relays that must be functional during the earthquake, the TRS needs to be a factor of 1.6 greater
than the RRS at the equipment fundamental frequency. This provides a HCLPF of 0.5g and a
median of 1.22g, which is the screening target. This can easily be demonstrated in many cases
since the tendency is to overshoot the RRS for frequencies beyond that of the peak of the RRS if
the RRS peak is at low frequency. Figure E-2 shows a typical overtest condition at frequencies
beyond the peak of the RRS.
Using the larger FD and larger uncertainty for function after the earthquake, the ratio of the TRS
to RRS was computed to be about 1.27 for screening in order to get the HCLPF up to the 0.57g
target screening level.

E-11

EPRI Proprietary Licensed Material


Development of Generic Fragility Descriptions for Purposes of Screening Based Upon Design Criteria

Figure E-2
Typical Overtest at High Frequency

E.6

References

E1. EPRI-NP6041SL, A Methodology for Assessment of Nuclear Power Plant Seismic


Margin, Revision 1, EPRI, Palo Alto, California, August 1991.
E2. EPRI-TR-103959, Methodology for Developing Seismic Fragilities, EPRI, Palo Alto,
California, June 1994.

E-12

EPRI Proprietary Licensed Material

F
EXAMPLE PROBLEM FOR SERVICE WATER PUMP

In Section 2.1.5, the importance of the fragility analysts expertise in component


fragility/HCLPF assessment is discussed. It is pointed out by the USNRC IPEEE reviewers that
some IPEEE calculations appeared to be poorly prepared and used unrealistic estimates of
uncertainties. In the discussion of this observation it is pointed out that this is also a problem
with design calculations. In this example of a service water pump, the existing analysis of the
pump was used to develop a fragility. It was determined that the failure mode focused on by the
design analyst was not a realistic mode of failure, whereas another area of the pump assembly
was improperly modeled and the higher response that would result from a corrected model would
govern the pump failure.
While it is emphasized in this Application Guide to utilize design data as much as possible for
screening or for scaling results to develop fragilities, it must be emphasized that the fragility
analyst must assess the validity of the design analysis. This example demonstrates an
approximate correction to existing design calculations in order to derive a seismic fragility.

F.1

Description of Equipment

The Service Water Pumps are long column vertical pumps that are anchored to the concrete roof
of the pump structure by expansion bolts. The pump columns are supported laterally about 10
feet below the pump motor stand. Since the unsupported column length was less than 20 feet,
one might screen the pump out based on criteria in the SQUG GIP (Reference F1) as having a
HCLPF exceeding 0.3g pga, whereas the DBE was 0.15g pga which corresponded approximately
to the 10,000 year mean pga defined by a Uniform Hazard Spectrum for the site. However,
expansion bolts are not normally used to anchor rotating equipment, thus further investigation
was conducted. A finite element analysis had been conducted in support of replacement of the
pumps and was available for review. In the report it was concluded that the expansion bolts for
the pump anchorage were the critical element. Although the expansion bolt safety factor was
less than normal manufacturers recommendations, it was close to the GIP allowable margin and
it was determined in the analysis that the capacity was sufficient to assure operability in the event
of a safe shutdown earthquake.
The pump model is shown in Figure F-1. The pump column, motor stand and motor were
modeled as beams. The base plate was modeled using plate elements. The nodes representing
the location of the expansion anchors were modeled as fixed in three translational directions.

F-1

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

Figure F-1
Model of the Service Water Pump

A more detailed review of the analysis indicated that the critical loading on the expansion
anchors resulted from a prying action in the mounting plate. With small displacement theory and
the bolt locations modeled as fixed in 3 directions of translation, a small amount of bending in
the base plate produces large calculated bolt tension loads. This is not a realistic failure mode for
the base plate anchorage. With a small amount of slip in the expansion anchors, the tension load
due to prying is relieved and only the shear load contributes to ultimate failure. Taking this into
consideration, other failure modes were examined. It was observed that the second mode of
vibration was about 3.9 Hz and corresponded to translation of the pump motor. This is a much
lower frequency than would be expected, thus the focus shifted to the motor stand and the
validity of the model. The motor stand is a cylindrical shape with two large windows cut out,
resulting in two 120-degree arcs that act as guided cantilever beams in bending when lateral
acceleration is applied to the motor. Refer to Figure F-2 for a simple representation of the motor
stand deformation.
It was noted that the analysis was based on different dimensions of the motor stand than the field
dimensions taken by the IPEEE team. The analysis dimensions were smaller, resulting in a
significantly more flexible motor stand. Thus, in calculating the strength factor and the
equipment response factor, the dimensions taken by the IPEEE team were used to adjust the
results from the computer model.

F-2

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

Motor

20.75

9342#
Motor stand
Base plate

Figure F-2
Simplified Motor Stand Model

F.2

Strength Factor

Using the calculated loads from the design verification analysis, the bending moment and shear
calculated at the motor to stand interface were applied to the corrected dimensions and the
resulting bending plus axial stress was computed to be about 7,000 psi. The material was A-53
Grade B, with 35 ksi specified yield strength. The median yield strength for low carbon steels is
about 1.20 times the specified yield, resulting in a median yield strength of 42 ksi. The median
factor of safety relative to yield is then 42/7 = 6.0.
It was assumed that any appreciable plasticity in the motor stand would result in misalignment
between the motor and pump shafts and damage the coupling between the motor and pump
assembly thus, the function would be lost. The plastic hinge shape factor in this case is greater
than 1.5 but a full plastic hinge in a support assembly that assure alignment of rotating members
is likely beyond a median capacity. Failure was assumed to occur at a factor of 1.5 beyond yield.
The onset of yielding was considered to be about a 95% confidence value. The resulting
capacity factor is:

FC = 1.50(6) = 9.0
There are two sources of uncertainty in the capacity factorthe material yield strength and the
capacity beyond yield. The specified yield is a 95% confidence value, so the uncertainty on
yield strength is:

UY = (1/ 1.65) ln 1.25 = 0.14


The uncertainty in the failure threshold is estimated based on the assumptions that the yield point
is a 95% confidence capacity. The resulting U is:

UF = (1/ 1.65) ln1.5 = 0.245


F-3

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

Combining uy and uf by SRSS results in a uc on capacity of:


UC = (0.14 2 + 0.245 2 )1/ 2 = 0.28

F.3

Equipment Response Factor

Equipment response variables are Qualification Method, Damping, Modeling, Mode


Combination and Earthquake Component Combination.
F.3.1 Qualification Method
The analysis was a response spectrum finite element analysis which is considered to be unbiased.
However, the modal response of the motor stand was computed at 3.9 Hz. The motor stand was
modeled as being more flexible than the actual geometry. The moment of inertia calculated from
the field dimensions taken by the IPEEE team showed that the stiffness in the weak axis was
actually 6.5 times the stiffness calculated in the model. So the actual frequency of the motor on
its stand would be greater by the square root of 6.5, or about 10 Hz vs 3.9 Hz from the computer
model. This frequency is at the peak of the broadened spectrum shown in Figure F-3. The
spectral acceleration defined at 2% damping used in the design analysis was 0.42g. 5% damping
is considered to be a median damping value. As shown in Figure F-3, if the 5% damped design
spectral peak is unbroadened, and the best estimate of frequency is 10 Hz, the qualification
method factor including damping is the ratio of 2% damped spectral acceleration at 3.9 Hz vs 5%
damped spectral acceleration at 10 Hz.

RQM =

F-4

0.43
= 0.86
0.50

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

Assumed
Unbroadened
Spectrum

Figure F-3
Demand Response Spectrum, 5% Damping

The uncertainty associated with the unbroadening of the design basis response spectrum can be
estimated from the ratio of the broadened to unbroadened spectral acceleration at 10 Hz. Since
the peak of the broadened spectrum is an upper bound (at least a 99% non-exceedance
probability value, i.e., 2.33 U value) the qualifications method uncertainty is
UQM =

1
0.55
ln
= 0.04
2.33 0.50

F.3.2 Damping
The calculated dominant frequency of the motor stand was about 3.9 Hz and the associated two
percent damped spectral acceleration used in the analysis was 0.42 g. Five percent damping is
considered to be median centered. However, when computing the damping factor and its
uncertainty on response, the frequency should first be corrected to reflect the actual motor stand
geometry vs. the softer geometry used in the model. For this case the damping factor was
incorporated in the qualification method factor. The response at 10 Hz is near the peak of the
spectrum shown in Figure F-3. In the amplified response regime, the ratio of spectral
accelerations at different damping levels can be approximated as the square root of the damping
rates. If 2% damping is considered a -2 U value, the uncertainty in damping is:
UD =

1
5
ln
= 0.23
2
2

F-5

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

F.3.3 Modeling
At the estimated 10 Hz elastic frequency of the motor stand, and considering the unbroadened
spectrum in Figure F-3, the spectrum is very steep, so a small error in frequency results in a large
change in spectral acceleration. Note that we have already accounted for the conservatism in
broadening in the qualification method factor.
For fragility calculations, it should be considered that as the response goes beyond the linear
region, the effective frequency reduces and in the case for the steep slope of the demand
spectrum and the corrected pump motor stand frequency, the response could significantly reduce.
It is also a common to slightly over estimate stiffness by assuming idealized inflexible bounding
conditions. It was judged that a reasonable estimate of the effective spectral acceleration would
be about half way between the 5% damped peak of 0.55g and the valley of 0.31g, or about 0.43g.
The elastic spectral acceleration used in developing the qualification method factor was 0.5g.
The modeling factor can then be computed as: The modeling factor, including damping, is then:
FM =

0.50
= 1.16
0.43

The range of response from the peak to the valley is considered to be plus or minus 2.33 U and
the U for modeling is computed as:
1
UM =
ln (.55/.31) = 0.12
2(2.33)

F.3.4 Mode Combination


The analysis was multimode dynamic analysis. The first mode was predominantly the pump
column and second mode was the pump motor and stand. The next modes are above 13 Hz and
are not particularly influential to the pump stand response. There is no bias in the mode
combination method so the mode combination factor is unity:

FMC = 1.0
The randomness on mode combination when the response is predominantly in one mode is
estimated to be:

RMC = 0.05
F.3.5 Earthquake Component Combination
The analysis was in accordance with the licensing criteria wherein the worst horizontal
directional response was combined with the vertical response by absolute sum. Median centered
response is considered to be a combination of all three directional responses by the 100, 40, 40
F-6

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

rule. The critical response is all horizontal and then 100, 40, 40 horizontal vector is 1.08 times
the single direction response. The earthquake component combination factor is then:
FECC = 1/1.08 = 0.93

Assuming that both horizontal components being in phase is a 3 extreme, the R for earthquake
component combination is:
= 0.09
RECC = ln 2

1.08

F.3.6 Equipment Response Factor


The equipment response factor is then:
FRE = FQM (FD)(FM)(FMC)(FECC)
FRE = 0.86(1.0)(1.16)(1.0)(0.93) = 0.93

s are combined by SRSS to yield:


R = (0 + 0 + 0 + 0.052 + 0.092)1/2 = 0.10
U = (0.042 + 0.232 + 0.122 + 0 + 0)1/2 = 0.26

F.4

Structural Response Factor

The pump structure is founded on rock and has a fundamental frequency of about 10 Hz as
indicated by the top of roof response spectrum in Figure F-3. Design damping was 4% as
indicated by Regulatory Guide 1.61 for structures at less than half yield. The pump structure is
simple and torsional effects are minimal. Spectra were developed from a fixed base 2
dimensional model.
Variables considered are spectral shape, damping, modeling, mode combination and ground
motion incoherence.
F.4.1 Spectral Shape
The UHS spectrum for the site, when anchored to the DBE pga, exceed the DBE spectrum at the
10 Hz fundamental frequency of the structure by 45%. The resulting spectral shape factor is:
FSS =

1
= 0.69
1.45

From Reference F2, the randomness is about:


F-7

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

RSS = 0.20
For fragilities anchored to pga, the uncertainty at 10 Hz is about:

USS = 0.16
In addition, the earthquake ground motion is specified as the average of two horizontal
components. The peak response in one direction will govern the structural response of the 2D
model. From Tables 3-2 and 3-3 of Reference F2,

Fpeak =

1
= 0.92
1.09

Rpeak = 0.13

Combining factors and s .

FSS = 0.69(0.92) = 0.63


RSS = 0.20
USS = (0.16 2 + 0.13 2 )1/ 2 = 0.21

F.4.2 Damping
The 4% design damping utilized in development of response spectra is considered median for the
low stress level in the structure.
FD = 1.0

UHS spectra are only provided at 5% damping. At 10 Hz the UHS spectra are rising as opposed
to the decayed peak of the DBE spectrum. In this case, the uncertainty in response due to
damping is estimated as:

UD = 0.15
F.4.3 Modeling
The model is simple, is fixed base and dominant response is primarily in a single mode as
indicated by the spectrum in Figure F-3. Code properties were used for concrete thus, according
to the guidance in Reference F2, the calculated frequency is considered to be median centered.
FM = 1.0

F-8

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump

The UHS spectra slope is not very steep at 10 Hz and the uncertainty due to variability in
structural frequency is estimated as:

Uf = 0.10
The mode shape for the short stiff fixed base structure is considered to not vary significantly
from variations in modeling parameters.

UMS = 0.05
UM = (0.12 + 0.05 2 )1/ 2 = 0.11

F.4.4 Mode Combination


The response is primarily in a single mode.
FMC = 1.0

RMC = 0.05
F.4.5 Ground Motion Incoherence
The structure is 58 x 124 in plan dimension. At 10 Hz, the GMI factor was calculated to be:
FGMI = 1.07

U = 0.03

F.4.6 Structural Response Factor


The resulting structural response factor is:

FRS = FSS FD FM FMC FGMI = (0.63)(1)(1)(1)(1.07) = 0.67


R = (0.20 2 + 0 + 0 + 0.05 2 + 0)1/ 2 = 0.21
U = (0.212 + 0.15 2 + 0.112 + 0 + 0.03 2 )1/ 2 = 0.28

F.5

Fragility for Service Water Pumps

The median peak ground acceleration capacity is the product of the capacity factor, the
equipment response factor, the structural response factor and the SSE peak ground acceleration.
F-9

EPRI Proprietary Licensed Material


Example Problem for Service Water Pump
Am = 9(0.93)(0.67)(0.15) = 0.84g

R = (0 + 0.10 2 + 0.212 )1/2 = 0.23


U = (0.282 + 0.262 + 0.282 )1/ 2 = 0.47
The HCLPF is computed from:
HCLPF = 0.84 exp (-1.65)(0.23 + 0.47) = 0.26g

In this case, the original target for screening was 0.3g pga. Suspicions regarding the design
analysis led to the more detailed evaluations and it is shown that the target was not met.

F.6

References

F1. SQUG, Generic Implementation Procedure (GIP) for Seismic Verification of Nuclear
Power Equipment, Revision 2, Corrected June 1991, Seismic Qualification Utility
Group.
F2. EPRI TR-103959, Methodology for Developing Seismic Fragilities, EPRI, Palo Alto,
California, June 1994.

F-10

EPRI Proprietary Licensed Material

G
GENERAL METHODOLOGY FOR LIQUEFACTION
SEISMIC FRAGILITY ASSESSMENT AND EXAMPLE
ANALYSIS

G.1

Introduction

Determinations of liquefaction seismic fragility is uncommon, and hence, the specific methods
for liquefaction fragility assessment are not as well established and commonly accepted/applied
as those for failures of structures and mechanical/electrical equipment.
As for any fragility assessment (FA), however, the development of liquefaction seismic fragility
follows the general FA framework, that requires probabilistic characterization of seismic
resistance parameters and related seismic load effects. For soil liquefaction, the probabilistic
characterization of resistance requires knowledge of the methods and results of in-situ tests and
collection of other geotechnical data, whereas probabilistic characterization of load effects on
resistance (when the seismic load is characterized by a scalar parameter, such as peak ground
acceleration [PGA] or spectral acceleration [Sa]) requires knowledge of the variability in
magnitude, duration, shear-stress time histories, etc., and their relationship to liquefaction failure.
Below are presented a brief summary of background on methods (both deterministic and
probabilistic) for seismic liquefaction assessment, and discussion of the basis and overview of
approach for seismic FA of soil failure due to liquefaction.
As is the case for evaluation of structural/mechanical/electrical components, a fragility analysis
should be undertaken only if the possibility of failure cannot be eliminated on a conservative
basis in a screening evaluation. EPRI NP-6041-SL (1991) describes a set of procedures for
performing a screening evaluation of seismic liquefaction potential. These procedures are
compatible with methods for seismic liquefaction FA, and hence, any liquefaction FA should
first be preceded by the appropriate screening evaluation.

G.2

Background

Assessment of soil liquefaction is a complex issue that has been studied extensively since the
early 1960s. Since the 1980s, significant advancements have been made in identifying the
parameters and relationships important to predicting the occurrence and severity of liquefaction.
Kramer (1996) provides a valuable overview of seismic liquefaction assessment, including
background on the development of modern methods.

G-1

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

Liquefaction failure is manifested, principally, according to one of two failure modes: (1) flow
failure, where the shear strength of the soil drops below the level needed to maintain stability
under static conditions, and very large movements can be produced under the effect of static
gravitational forces; and (2) cyclic mobility (including lateral spreading) failure, where
incremental damage/displacements are accumulated as shear-stress/load pulses momentarily
exceed the shear strength/resistance of the soil during dynamic response. Either mode of failure
can lead to excessive strains, and hence, displacements that could, in turn, lead to unacceptable
performance of supported structures, buried piping, tanks, etc., at a nuclear power plant.
Although detailed approaches for seismic liquefaction assessment can be quite theoretical and
involved, suitably accurate simplified deterministic approaches, based on empirical methods,
have been developed. Perhaps the most recent and widely accepted of such approaches is that
published in 1997 by the U.S. National Center for Earthquake Engineering Research (Youd and
Idriss, 1997), that embodies knowledge acquired from numerous studies by several experts.
Various probabilistic approaches of seismic liquefaction assessment have also been developed,
and these generally are implemented in accordance with one of the following basic strategies: (a)
characterizing the key parameters and relationships in a simplified deterministic model as
random, and developing a derived expression or formulation/procedure for determining
probability of liquefaction for a given soil deposit and seismic load amplitude; or (b) direct
statistical classification and regression analysis of empirical observations where key in-situ
resistance parameters and load information have been obtained. Using the latter approach, Liao
et al. (1998) developed the following expression for probability of liquefaction:

PL =

1
1+ e

[ 0 + 1 ln(CSR) + 2 (N1 )60 ]

where 0 , 1, 2 are parameters determined from regression analysis for specific soil types
(e.g., clean sands, silty sands, etc.), CSR is the cyclic stress ratio (load parameter), and (N1)60 is
the standardized/corrected blow count. The reference list identifies several documents that
describe probabilistic approaches for seismic liquefaction assessment. An approach that would
capture the uncertainty in the methodologies would be a simulation that makes use of multiple
methods.

G.3

Basis of Approach

The key parameters for liquefaction assessment are relevant measures of cyclic (shear) stress
ratio. Thus, the seismic demand is expressed as a loading cyclic stress ratio, CSR, and the
seismic capacity is expressed as a cyclic stress resistance ratio of the soil, CRR. The factor of
safety against liquefaction, FSL , is given simply as:
FSL =

cyclic shear stress required to cause liquefacti on


CRR
=
equivalent cyclic shear stress induced by earthquake CRR

To determine CSR, it is necessary to determine the shear stresses at depth in the soil caused by
the earthquake. This can be accomplished using site response analysis using a simplified
G-2

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

formula, based on a one-dimensional equivalent-linear SHAKE model or using more


sophisticated software such as DE SRA-2C that takes into account the degradation of the soils
due to progressive pore water pressure buildup during an earthquake. At each depth location of
interest, an equivalent cyclic shear stress, cyc , is obtained (in terms of the maximum stress) and
an equivalent number of cycles at this cyclic stress level is estimated based on magnitude. The
cyclic shear stress is normalized by the effective vertical stress, and corrected for field
conditions, in order to obtain the equivalent cyclic stress ratio CSR.
To determine CRR, existing empirical relationships are available that depend on the measured
in-situ resistance [i.e., value of (N1)60, CPT cone resistance, shear wave velocity, etc.], percent
fines, effective confining stress, and magnitude.
For probabilistic analysis based on strategy (a) described above, it is desirable to define and
estimate variabilities R , U associated with the parameters and relationships involved in
determining CRR and the CSR (given that the ground motion parameter is known). In some
instances, it is possible to determine these values from existing results (e.g., researchers have
provided statistics for many of the relationships governing the empirical determination of CRR
and CSR), whereas for some components of the analysis (e.g., the determination of random
variability in maximum shear stress at depth) it is beneficial to estimate variabilities from results
of response simulations.
Because failure of nuclear plant components will generally depend, not just on the occurrence of
liquefaction, but rather on the magnitude of accumulated shear deformations or soil settlement,
the fragility curves should be referenced to a state (or states) of liquefaction capable of
developing shear deformations, or loss of soil strength, causing distress to plant components.

G.4

Example Case Study

G.4.1 Overview of Approach


Using the above general approach, a liquefaction fragility analysis was performed for a site that
would not pass the screening guidelines in EPRI (1991). A family of fragility curves (accounting
for both random [aleatory] and modeling [epistemic] uncertainties) was developed for each of
the following five end-states of interest:

i. Incipient liquefaction
ii. Gross liquefaction
iii. Level-ground liquefaction-induced settlement of about 5 cm (2 inches).
iv. Level-ground liquefaction-induced settlements of about 10 cm (4 inches).
v. Level-ground liquefaction-induced settlements of about 20 cm (8 inches).
Probabilistic fragility calculations were performed using available information on relevant site
conditions, deterministic liquefaction potential, and site-specific probabilistic seismic hazard.
G-3

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

The procedure of NCEER (Youd and Idriss, 1997) was followed in conducting individual
liquefaction analyses, and the procedure of Ishihara and Yoshimine was used in computing
settlements associated with level-ground liquefaction. The soil profile was subdivided into
uniform layers having thickness of about 30.5 cm (one foot), over a range of depths from zero to
about 25 meters (98 feet).
The most important basic random variables and uncertain variables were identified and
characterized by means of appropriate probability distributions. The Latin-Hypercube
Simulation (LHS) approach was implemented to (a) obtain several simulated samples of each
random and uncertain variable, and (b) randomly combine these samples to obtain several
simulated realizations of conditions that govern liquefaction-related behavior. Each simulated
combination/realization defined a set of inputs for performing computational analyses of
liquefaction safety factor, post-liquefaction settlements, and settlement-related failures, for a
range of ground motions.
In conducting each fragility analysis (for each given end-state of interest), two sets of
simulations were performed. The first set accounted for only the random components of
uncertainty, whereas the second set accounted for composite uncertainty associated with both
random and epistemic elements. This approach enabled the total variability of the derived
fragility functions to be separated into its random and epistemic contributors.
G.4.2 Initial Liquefaction Analysis and Results
In an initial fragility analysis pertaining to the state of gross liquefaction, two sets of 50
simulations were generated, and liquefaction calculations were completed manually, using
computational spreadsheets. In performing each of these manual calculations, plots of
liquefaction safety factor versus depth were examined, and the surface peak-ground acceleration
(PGA) was iteratively adjusted until a state of gross liquefaction (corresponding to the somewhat
imprecise condition that liquefaction safety factor was observed to be less than unity over a
significant total depth) was realized. This initial analysis resulted in fragility curves that could
be approximated by a conventional double-lognormal distribution having the following
equivalent fragility parameters defined at the top of the soil surface.
Preliminary Fragility Parameters for Gross Liquefaction:
Motion keyed to:

Ground-surface PGA

Median

0.41

0.24

0.33

A significant observation resulting from this initial analysis was that, in most simulations, only a
relatively small increment in ground-surface PGA was generally required to go from a state of
incipient liquefaction (i.e., safety factor of 1.0 in a single isolated soil layer) to gross liquefaction
(i.e., safety factor below 1.0 in several soil layers). This observation suggested that the fragility
parameters for gross liquefaction would not change significantly if a more precise definition of
gross liquefaction were implemented. In addition, the observation implied that, for a given
simulation, liquefaction-induced settlement (i.e., cumulative post-liquefaction settlement at the
ground surface) would increase dramatically, from a nominal value to a limiting maximum, over
G-4

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

a comparatively narrow increment in ground-surface PGA. Hence, the initial fragility results
could serve as a meaningful approximation to the fragility functions describing liquefactioninduced settlement failures.
G.4.3 Detailed Liquefaction and Settlement Analysis and Results
Although the initial analysis was considered to be adequate in meaningfully conveying overall
liquefaction fragility additional work was performed to refine the fragility assessment and to
provide a confirmatory internal quality check on the validity of the initial analysis. The
additional work consisted of explicitly computing liquefaction-induced settlements, and defining
failure for affected components in terms of a criterion based on settlement. For this more
detailed assessment, two sets of 500 simulations were generated, liquefaction and settlement
analyses were performed for each simulation using a special-purpose computer code, and
fragility curves were developed for surface ground motions ranging from 0.01 g to 2.00 g.
Two significant and distinct observations to be made in regard to the more detailed fragility
assessment are (1) that limiting values of post-liquefaction volumetric strain limit the maximum
amount of cumulative settlement that can be achieved in a given simulation, and (2) a small
fraction of simulations result in a soil profile that is substantially invulnerable to gross
liquefaction. The first observation means that, at some point, for all simulations, the settlement
associated with level-ground liquefaction ceases to increase with ground acceleration. The
second observation means that gross liquefaction will never occur, regardless of ground-motion
level, for some of the simulated soil profiles (although incipient liquefaction and some small
settlements can be produced for these profiles). These two implications are considered to be
physically realistic for the case of level-ground liquefaction, as opposed to undesirable artifacts
of the simulation process. An important consequence of these two points is that fragility curves
associated with both gross liquefaction and settlement-related failures can, in some cases, have a
limiting maximum value that is less than unity.
The more detailed fragility analysis resulted in fragility curves for incipient liquefaction and
gross liquefaction that could be well approximated by the following equivalent fragility
parameters of a double-lognormal distribution.
Detailed-Analysis Fragility Parameters for Incipient Liquefaction:
Motion keyed to:

Ground-surface PGA

Median

0.27

0.24

0.36

Detailed-Analysis Fragility Parameters for Gross Liquefaction:


Motion keyed to:

Ground-surface PGA

Median

0.40

0.23

0.36

In this analysis, incipient and gross liquefaction was defined explicitly and consistently, as
follows:
G-5

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

Incipient liquefaction corresponded to the state where the most vulnerable soil layer just
reached a liquefaction safety factor of unity.

Gross liquefaction corresponded to the state where liquefaction safety factors less than unity
occurred over a cumulative soil depth of about 3 meters (10 feet, or 10 soil layers).

In the set of simulations that accounted only for random variability, none of the simulated soil
profiles was invulnerable to gross or incipient liquefaction. For the set of simulations that
accounted for composite variability, 13 out of 500 (or 2.6 percent) of the simulated soil profiles
were invulnerable to gross liquefaction, but none were invulnerable to incipient liquefaction.
The small fraction of soil profiles that were invulnerable to gross liquefaction limited the
maximum probability of gross liquefaction (at high ground motion), in the resulting weighted
composite fragility curve, to 0.974. Since this value was only a small deviation from unity, and
the shape of the fragility curve was otherwise significantly lognormal in character, it was decided
that the double-lognormal fragility representation was reasonable. Figure G-1 show plots of the
weighted fragility curves resulting from each set of simulations, for end-states of incipient
liquefaction and gross liquefaction.
The definition of incipient and gross liquefaction do not, however, lead to a decision as to what
would fail or what could fail. Therefore, further analyses were conducted to determine the
settlement of the soil as a result of liquefaction and the consquence of the settlement to buried
pipes and concrete electrical ducts.
For end-states associated with component failure due to excessive settlements, more dramatic
constraints on maximum failure probability were realized with increases in the components
median resistance to settlement. As indicated previously, three related cases were examined, for
components having median failure settlements of 5 cm (2 inches), 10 cm (4 inches), and 20 cm
(8 inches). A double-lognormal distribution was used for characterizing a components ability to
resist settlement, and in each of the preceding cases, parameters R =0.30 and U =0.60 were
used to describe, respectively, random variability and modeling uncertainty in settlements
causing failure. The value of R approximately accounts both for randomness in material
properties affecting resistances and randomness associated with spatial non-uniformity in
settlements. The high value for U reflects the situation that specific analysis of buried piping
connecting to buildings was not available to determine settlement failure thresholds.
Due to the constraints on failure probabilities, settlement-related fragility curves could not, in
general, be properly conveyed by equivalent parameters of a double-lognormal probability
distribution. For this reason, only the fragility curves resulting from the simulations are
presented. Figure G-2 show plots of the weighted fragility curve resulting from each set of
simulations, for the three cases of median failure of buried piping for settlement of 5 cm, 10 cm,
and 20 cm. The fragility plots are relative to the surface pga.
Examination of these figures shows that, as component median capacity against settlement
increases, the asymptotic (maximum) level of failure probability decreases markedly. Again, the
reason for the reduced asymptote (i.e., less than unity) of the fragility curve is that postliquefaction volumetric strains in the soil reach limiting values (that depend upon the initial
relative density of the soil). These limiting strains imply maximum settlements that do not
G-6

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

increase with ground motion. The asymptotic failure probability of the fragility curve is thus
controlled by the limiting maximum settlements and the components probability distribution of
settlement capacity.

Figure G-1
Weighted Fragility Curves, Accounting for Random Variability and Composite Variability,
for End-States of: (i) Incipient Liquefaction, and (ii) Gross Liquefaction.

G-7

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

(Median Failure Settlement = 5 cm

(Median Failure Settlement = 10 cm

(Median Failure Settlement = 20 cm

Figure G-2
Weighted Fragility Curves, Accounting for Random Variability and Composite Variability,
for End-States of Component Failure Due to Settlements Caused by Level-Ground
Liquefaction, for Cases Where the Component Median Capacity Against Failure Equals (iii)
5 cm, (iv) 10 cm and (v) 20 cm

G-8

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

G.5

References

1.

Blake, T.F. (1998). LIQUEFY2 Version 1.50 Update, LIQUEFY2 Users Manual,
Thomas F. Blake Computer Services & Software, Newbury Park, CA.

2.

Chameau, J.L., and G.W. Clough (1983). Probabilistic Pore Pressure Analysis for Seismic
Loading, Journal of Geotechnical Engineering, ASCE, Vol. 109, No. 4, pp. 507-524.

3.

Christian, J.T., and W.F. Swiger (1973). Statistic of Liquefaction and SPT Results,
Journal of the Geotechnical Engineering Division, ASCE, Vol. 101, No. GT-11,
pp. 1135-1150.

4.

EPRI (1991). A Methodology for Assessment of Nuclear Power Plant Seismic Margin
(Revision 1), EPRI NP-6041-SL, EPRI, Palo Alto, CA.

5.

Fardis, M.N., and D. Veneziano (1982). Probabilistic Analysis of Deposit Liquefaction,


Journal of the Geotechnical Engineering Division, ASCE, Vol. 108, No. GT-3, pp. 395-417.

6.

Haldar, A., and W.H. Tang (1981). Statistical Study of Uniform Cycles in Earthquakes,
Journal of the Geotechnical Engineering Division, ASCE, Vol. 107, No. GT-5, pp. 577-589.

7.

Haldar, A., and W.H. Tang (1979). Probabilistic Evaluation of Liquefaction Potential,
Journal of the Geotechnical Engineering Division, ASCE, Vol. 105, No. GT-2, pp. 145-163.

8.

Ishihara, K. and Yoshimine, M., Evaluation of Settlements in Sand Deposits Following


Liquefaction During Earthquakes, Soils and Foundations, Vol. 32, No. 1, pp. 173-188.

9.

Kramer, S.L. (1996). Geotechnical Earthquake Engineering, Prehntice-Hall Inc., Upper


Saddle River, New Jersey, 653 p.

10. Liao, S.S.C., D. Veneziano, and R.V. Whitman (1988). Regression Models for Evaluating
Liquefaction Probability, Journal of Geotechnical Engineering, ASCE, Vol. 114, No. 4,
pp. 389-411.
11. Veneziano, D., and S.S.C. Liao (1984). Statistical Analysis of Liquefaction Data,
Proceedings of the 4th Specialty Conference on Probabilistic Mechanics and Structural
Reliability, ASCE, pp. 206-209.
12. Whitman, R.V. (1971). Resistance of Soil to Liquefaction and Settlement, Soils and
Foundations, Vol. 11, No. 4, pp. 59-68.
13. Yegian, M.K., and R.V. Whitman (1978). Risk Analysis for Ground Failure by
Liquefaction, Journal of the Geotechnical Engineering Division, ASCE, Vol. 104, No.
GT-7, pp. 921-938.
14. Youd, T.L. and I.M. Idriss (1997). Summary Report, Proceedings of the NCEER
Workshop on Evaluation of Liquefaction Resistance of Soils (T.L. Youd and I.M. Idriss,
G-9

EPRI Proprietary Licensed Material


General Methodology for Liquefaction Seismic Fragility Assessment and Example Analysis

editors), held at Salt Lake City, Utah (January 5-6, 1996), Technical Report No.
NCEER-97-0022.
15. Youd, T.L. and S.K. Noble (1997). Magnitude Scaling Factors, Proceedings of the
NCEER Workshop on Evaluation of Liquefaction Resistance of Soils (T.L. Youd and I.M.
Idriss, editors), held at Salt Lake City, Utah (January 5-6, 1996), Technical Report No.
NCEER-97-0022.
16. Youd, T.L., K.H. Stokoe, P.K. Robertson, W.D.L. Finn, and P.M. Byrne (1996).
Proceedings of a Symposium on Recent Developments in Seismic Liquefaction
Assessment, Vancouver, B.C., Sponsored by ConeTec, April 12.

G-10

WARNING: This Document contains


information classified under U.S. Export
Control regulations as restricted from
export outside the United States. You
are under an obligation to ensure that you have a
legal right to obtain access to this information
and to ensure that you obtain an export license
prior to any re-export of this information. Special
restrictions apply to access by anyone that is not
a United States citizen or a Permanent United
States resident. For further information
regarding your obligations, please see the
information contained below in the section titled
Export Control Restrictions.

Export Control Restrictions


Access to and use of EPRI Intellectual Property is granted
with the specific understanding and requirement that
responsibility for ensuring full compliance with all applicable
U.S. and foreign export laws and regulations is being
undertaken by you and your company. This includes an
obligation to ensure that any individual receiving access
hereunder who is not a U.S. citizen or permanent U.S.
resident is permitted access under applicable U.S. and foreign
export laws and regulations. In the event you are uncertain
whether you or your company may lawfully obtain access to
this EPRI Intellectual Property, you acknowledge that it is
your obligation to consult with your companys legal counsel
to determine whether this access is lawful. Although EPRI
may make available on a case by case basis an informal
assessment of the applicable U.S. export classification for
specific EPRI Intellectual Property, you and your company
acknowledge that this assessment is solely for informational
purposes and not for reliance purposes. You and your
company acknowledge that it is still the obligation of you and
your company to make your own assessment of the
applicable U.S. export classification and ensure compliance
accordingly. You and your company understand and
acknowledge your obligations to make a prompt report to
EPRI and the appropriate authorities regarding any access to
or use of EPRI Intellectual Property hereunder that may be
in violation of applicable U.S. or foreign export laws or
regulations.

About EPRI
EPRI creates science and technology solutions for
the global energy and energy services industry.
U.S. electric utilities established the Electric Power
Research Institute in 1973 as a nonprofit research
consortium for the benefit of utility members, their
customers, and society. Now known simply as EPRI,
the company provides a wide range of innovative
products and services to more than 1000 energyrelated organizations in 40 countries. EPRIs
multidisciplinary team of scientists and engineers
draws on a worldwide network of technical and
business expertise to help solve todays toughest
energy and environmental problems.
EPRI. Electrify the World

Program:

1002988

Nuclear Power

2002 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power Research
Institute and EPRI are registered service marks of the Electric Power Research Institute, Inc.
EPRI. ELECTRIFY THE WORLD is a service mark of the Electric Power Research Institute, Inc.
Printed on recycled paper in the United States of America

EPRI 3412 Hillview Avenue, Palo Alto, California 94304 PO Box 10412, Palo Alto, California 94303 USA
800.313.3774 650.855.2121 askepri@epri.com www.epri.com

Das könnte Ihnen auch gefallen