ML20086P180

Review of IPEEE Binning for Plant Farley

ML20086P180
Person / Time
Site:	Farley
Issue date:	12/31/1991
From:	Boissonade A, Mccann M, Jeffrey Reed JACK R. BENJAMIN & ASSOCIATES, INC.
To:
Shared Package
ML20086P182	List: ML20086P180
References
NUDOCS 9112260237
Download: ML20086P180 (59)

Text

1 l

REVIEW OF TIIE IPEEE BINNING FOR PLANT FARLEY Prepared for Southern Company Services, Inc.

Birmingham, Alabama Prepared by h!r.rtin W. McCann, Jr.

Auguste C. Boissonade John W, Reed Jack R. Benjamin and Associates, Inc.

Mountain View, California 1

December 1991 l

Ih 4

iit 4:3 P

PDR

i EXECUTIVE

SUMMARY

As part of the implementation of the U.S. Nuclear Regulato.ty Commission (USNRC) policy on severe accidents, each licensee is requested to perform an individual plant evaluation for external events (IPEEE). A plant evaluation for seismic events can be satis 6ed by performing a probabilistic risk assessment (PRA) or a seismic margins analysis.

For purposes of performing a margins assessment, the USNRC staff has grouped the plaats in the castern U.S. (EUS) into evaluation bins. The process of assigning a plant to a bin was based on the site hazard as described in NUREG-1407. This report examines the assignment of the Joseph M. Farley nuclear power plant (Plant Farley) into the 0.3g Focused-Scope (FS) bin. Based on a review of the IPEEE binning and the LLNL estimate of the seismic hazard at Plant Farley, it is concluded that Plant Parley should be re-assigned to the Reduced Scope (RS) bin. The basis for this reclassincation is summarized as follows:

Plant Farley is located in the Gulf Coastal region where many (seven of ten) of the RS plants are located.

LLNL 5GX seismic hazard assessment produces extreme estimates of the likelihood of ground motions that are not supported by the historic experience at Plant Farley and thus do not provide a realistic measure of the site hazard.

based on the EPRI and LLNL 4GX hazard studies, the composite probability (CP) for Plant Farley is closer to the midpoint CP of the RS plants than the FS plants.

An examination of the LLNL 5GX seismic hazard for Plant Farley demonstrates the following:

the LLNL SGX seismic hazard results which strongly contribute to the mean hazard and the 85th fractile are extreme and inconsistent with the historical record, the ground motion models selected by expert 5 are not supported by e

available strong-motion data, and the process used to derive these models is demonstrated to be flawed, ii

the ongoing LLNL seismic hazard program for the Department of Energy at the Savannah River Site, which includes expert 5, does not j

use these ground motion models.

This document summarizes the technical arguments to support the re-assignment of Plant Farley to the RS bin. By excluding the LLNL 5GX hazard results from the binning l

process, this report clearly demonstrates that the seismic hazard at Plant Farley is more closely aligned with the hazard at plants in the RS bin. A reassessment cf the LLNL seismic hazard for Plant Farley that eliminates extreme estimates of the ste hazard reinforces this conclusion.

y LIST OF ACRONYMS Cumulative Probability Distribution Function CDF_

CP Composite Probability EPRI Electric Power Research Institute Eastern United States EUS Focused-Scope FS IPEEE Individual Plant Evaluation for External Events Lawrence Livermore National Laboratory LLNL NPR New Production Reactor Peak Ground Acceleration PGA PRA' Probabilistic Risk Assessment Reduced-Scope RS Seismic-Activity Rate SAR U.S Nuclear Regulatory Commission USNRC iv

CONTENTS HEC EXECUTIVE

SUMMARY

ii LIST OF ACRONYMS iv INTRODUCTION 1

BASIS FOR PLANT FARLEY BINNING REEVALUATION 2

USNCR IPEEE PLANT BINNING PROCESS 3

LLNL SEISMIC HAZARD FOR PLANT FARLEY 4

Review of Individual Seismic Ilazard Estimates 5

Seismic-Activity Rate Assessment 8

Ground Motion Expert 5 Attenuation Models 10 Comparison With Strong Motion Data 11 Model Estimation 13 Seismic liazard Assessment Based on Expert 5 Models 14 LLNL NPR Seismic Ilazard Assessment 15 Summary of LLNL Seismic liazard 15 PLANT FARLEY IPEEE BINNING REEXAMINATION 16 EPRI Seismic Ilazard 17 LLNL 4GX Scismic Hazard 17 CONCLUSIONS 18 REFERENCES 31 APPENDIX A Rank Order Listing of the LLNL PGA Seismic Ilazard for Plant Farley APPENDIX B Rank Order Listing of the LLNL Magnitude Frequency Estimates -

APPENDIX C Reassessment of the LLNL Seismic Hazard For Plant Farley l

l V

l L

INTRODUCTION As part of the implementation of the U.S. Nuclear Regulatory Commission (USNRC) policy statement on severe accidents, Supplement 4 to Generic Letter 88-20 requests that each licensee perform an individual plant examination for external events (IPEEE). The IPEEE program specifically requires that seismic events be included in the plant evaluation.

The seismic evaluation can entall either a seismic probabilistic risk assessment (PRA) or a seismic margins evaluation. For purposes of performing a seismic margins assessment, the USNRC staff has recommended one of two different review-level earthquakes (0.3g or reduced-scope) for plants in the eastern U.S. (EUS). The plants were assigned to bins or groups based on a consideration of the site hazard. Associated with etch bin is a review-level earthquake and the evaluation th9 must be performed. NUREG-1407 (1) describes the binning process and the required evaluations. Based on the binning procedure, the Joseph M. Farley plant (Plant Farley) was assigned to the 0.3g Focused-Scope (FS) b'.

m This document presents the findings of a review of the LLNL seismic hazard assessment for Plant Farley and the IPEEE assignment of Plant Farley to the FS bin. The results of a review of the IPEEE binning indicated there was a high degree of similarity in terms of location and level of seismicity between Plant Farley and the plants in the Reduced-Scope (RS) bin. Furthermore, the characterization of Plant Parley as a FS plant seems to depend primarily on the LLNL 5GX seismic hazard results. With these insights a review of the LLNL seismic hazard assessment for Plant Farley was conducted.

The findings of this review suggest that the placement of Plant Farley into the FS bin was determined largely on the LLNL seismic hazard results based on the input from the five

- ground motion experts (the LLNL 5GX case). From an examination of the LLNL 5GX L

ha7ard results, it is concluded that this assessment does not provide a rational measure of the likelihood of grour.d motions of engineering interest. Two factors contributed to this conclusion: extreme estimates of the rate of earthquake occurrences in the region near Plant Farley and the ground motion models selected by LLNL expert 5. This document 1

1

summarizes the technical arguments to support this position. Excluding the LLNL 5GX hazard results from the binning process clearly demonstrates that the seismic hazard at Plant Farley is more closely aligned with the hazard at plants in the RS bin. A reassessment of the LLNL seismic hazard for Plant Farley reinforces this conclusion. Based on these Gndings it is concluded that Plant Farley should be assigned to the RS bin.

The next section outlines the technical position to support the reclassincation of Plant Parley to the RS bin. This is followed by a brief overview of the IPEEE binning process.

In the next section the results of a review of the LLNL seismic hazard for Plant Farley are presented. The detailed insights from this evaluation provides a clear understanding of the primary factors that contribute to the apparent inconsistency of the results. Next, the IPEEE binning of Plant Farley is reexamined based on the EPRI and LLNL 4GX hazard results.

The final section presents the conclusion of this study.

HASIS FOR PLANT FARLEY IPEEE IIINNING RE-EVALUATION This report presents the technical basis for the position that Plant Farley should be reassigned to the RS bin. This position is based on the results of a close examination of the LLNL seismic hazard for the Plant Farley site. The Gndings of this review identify two factors that lead to the conciusion that the LLNL study does not provide a rational measure of the seismic hazard at Plant Farley. These factors are the assessment of scismicity in the regional vicinity of Plant Farley and the ground motion models selected by LLNL expert 5.

As demonstrated in this report, individually these factors have a major impact on the assessment of seismic hazard. Jointly, they produce extreme estimates of the likelihood of ground motion that are inconsistent with the seismicity of the region.

The review findings have broad implications concerning the LLNL seismic hazard L

assessment at all plant sites in the EUS. Ideally, the LLNL seismic hazard should bc recomputed for each plant. Such a re-evaluation is not within the practical scope of this study. However, as a demonstration, Appendix C to this report presents the results of a re-2 l

i evaluation of the LLNL seismic hazard for Plant Parley. Alternatively, the LLNL seismic hazard results based on the ground motion modelt provided by 4 of the ground motion experts (AGX case) are considered to be an adequate substitute to demonstrate that Plant Farley should be reassigned to the RS bin.

The position that Plant Farley should be classified to tne RS bin is developed on the basis of the following:

review of the IPEEE binning and the relationship of Plant Farley to o

the RS and FS plants, o

results of a close examination of the LLNL seismic hazard for the Farley site that coacludes the LLNL study does not provide a rational m.ture of the site hazard, and use of the LLNL 4GX hazard results (as a surrogate measure of a set of revised a

LLNL seismic hazard estimates for all sites in the EUS) to reassess the appropriate bin for Plant Farley, USNRC IPEEE PLANT BINNING PROCESS In providing guidance to licensees in the FUS for performing plant examinations, the USNRC staff d,eveloped two bins for purposes of performing seismic margin evaluations. A 0.3g peak ground acceleratu NiA) review-lev.:1 earthquake has been set for most plants in the EUS. For plants who e tac.

aic har.ard is low, the staff determined that a reduced-i scope margin evaluation shoulo oc performed. Plants assigned to the RS bin are located in the area near the Gulf and Florida coasts and in the upper Midwest. Figure I shows a map that identifies the locmtion of the RS and FS plants and Plant Farley Plant Farley is located approximately 75 miles north of the Gulf, well within the perimeter of the RS plants located in the Gulf Coastal Plain. Furthermore, Plant Farley is located over 400 km from the Charleston seismic zone which is the dominant seismogenic feature in the southeast. In spite of the favorable geograpPic location of Plant Farley relative to the plants in the RS bin, it has been assigned to the 0.3g FS bin.

3 l

For purposes of evaluating the seismic hazard at plant sites in the EUS, the USNRC binning process used t'ne results of the probabilistic seismic hazard assessments performed by Lawrence Livermore National Laboratory (LLNL) ar:1 the Electric Power Research Institute (EPRI) (2,3). The binning process involved a ranking procedure that represented the seismic hazard at each site in terms of a composite probability (CP) of exceeding the NUREG/CR-0098 (4) response spectrum anchored to 0.3g PGA. The procedure, which is described in NUREG-1407, considered the seismic hazard for PGA and selected spectral response accelerations in the 2.5 to 10 Hz frequency range. The ranking of sites was performed using the median, mean and 85* percentile hazard results from the two studies.

~

LLNL SEISMIC HAZARD FOR PLANT FARLEY In the first step in the IPEEE binning process, plants were ranked using the EPRI and u-

' LNL seismic hazard assessments (2,3). However, the staff noted that the LLNL mean

hazard results at many sites was dominated by the ground motion model selected by es gmund motion expert (LLNL expert 5), in response to concerns that this model may not ce co. 'istent with available data, the LLNL seismic hazard was recomputed using tne input L: c, four ground mo' ion experts, excluding the input provided by expert 5 (4GX case).,The two sets cf LLNL hazaro assessments, the 4GX and SGX cases, were treated as two separate studies in the li)EEE binning process.

An-unination of the IPEEE binning process suggests that the assignment of Plant Farlev FS bin is attributable to the ranking associated with the LLNL 5GX seismic hazar. a

. Given that Plant Farley is located in a low-seisraic region and the significant impac. diat the ground motion model selected by LLNL expert 5 can have at low-seismicity sites (2), a detailed review of the LLNL SGX hazard results was conducted.

Recent experience indicates that in order to develop ar. understanding of the seismic hazard at a site, an assessment must be performed that provides insight into the factors that contribute to the results. Reviewing the hazard results for a site is complicated by the fact 4

there are a number of parameters involved in the analysis (e.g., many seismic sources, seismicity parameters, etc.) and multiple parameter estimates that are develogxl as part of the uncertainty analysis. The assessment of uncertainty is an important part of the hazard analysis and the key factor in determining the mean hazard for a site.

The results of a seismic hazard assessment are generally Iresented in a standard, summary-type format that includes the 0.15,0.50 and 0.85* pewentile and mean seismic hazard curves. For example, Figure 2 shows the EPRI seismic hazard curves for Plant Farley for PGA. The percentile hazard curves correspond to selected probability levels, at each ground moiion level, of a discrete cumulative probability distribution function that quantifies tne uncertainty in the assessment of the estimate of the annual probability of exceedance (hazard value). The mean hazard curve is the expected value or arithmetic average of the annual probability of exceedance. The discrete cumulative probability distribution function (there is one at eaci' ground motion level) is comprised of thousands of estimates of the probability of exceedance. Associated with each hazard value is a probability weight. Each hazard curve corresponds to a particular combination of input parameters that were provided by the seismicity and ground motion experts. To develop an understanding of the contribution of individual hazard estimates (and the parameters that comprise it) to the mean probability of exceedance, a detailed analysis must be made.

In this study, the LLNL seismic hazard results for Plant Farley were examined from two perspectives. The first looked at the probability distribution on seismic hazard and identified the individual hazard estimates that contributed to the mean hazard and the 85*

percentile level. The second part of the review examined the estimated rate of earthquake occurrences in the vicinity of the plant.

Review of Individual Seismic IIazard Estimntes In order to develop detailed insights into the LLNL seismic hazard estimate for Plant Farley, the hazard was recalculated. The LLNL seismic hazard software was modified to 5

provide detailed output that included a summary of the parameters associated with each seismic hazard curve that was calculated in each simulation, including selected intermediate results. The intermediate results that were retrieved included the seismic hazard curves for each seismic source (before they were combined with other sources in an expert map') and the estimated earthquake occurrence rates for each seismic source and simulation. Each intermediate result is identified in terms of the expert that provided the input, the simulation number and the value of each input parameter.

Table 1 presents a summary of the LLNL probability distribution on the probability of exceedance for Plant Farley for a PGA level of 0.13g for the 5GX case. For purposes of this review, a PGA level of 0.13g was selected since it corresponds approximately to the SSE value of 0.lg. The following information is summarized in the table:

rank order number for each hazard value (1 = highest hazard estimate),

ground motion and seismicity expert numbers that provided the input e

to the hazard calculation, discrete probability weight associated with the hazard estimate, e

the, cumulative probability level, e

the contribution (in percent) of the hazard value to the mean e

probability of exceedance, the cumulative contribution (in percent) of the top hazard values to the mean F.obability of exceedance, the hazard value, and the

+

cumulative mean hazard.

y

'To quantify the uncertainty on the assessment of srNmic sources that may be active in a region, alternative maps (or combinations of seismic sources) are defined by each LLNL seismicity expert.

6

1

~

The results in Table 1 provide an insight into the extreme limits of the LLNL hazard results and their contribution to the mean hazard level. Appendix A to this report provides a listing of the top 100 hazard estimates at 0.13g PGA. The top 100 hazard values collectively determine 76 percent of the mean hazard. (Note, the complete probability distribution is defined by 2750 values at each ground motion level.)

Based on a review of the information in Table 1 and Appendix A, the following observations are made:

fifty percent of the mean hazard is produced by the top 21 hazard 3

values whose cumulative probability weight is 7.4x10 the annual probability of exceedance values for the top 21 hazard estimates range from 6.37x102 to 4.13x104 These hazard estimates correspond to average-recurrence intervals for PGA 2. 0.13g that range from 2.4 to 15.7 years.

of the top 100 estimates of the probability of exceedance,95 are e

derived from the ground motion model selected by LLNL expert 5.

An alternative perspective of the extreme PGA hazard results cited above (e.g., the top 21 values) is provided by estimating the range on the expected number of times 0.13g or greater would occur during the period of record. For a period of record of 200 years,0.13g or greater would occur 13 to 83 times at the Plant Parley site. Based on the fact there is no evidence that ground motions of engineering interest have occurred at Plant Farley, the validity of hazaid estimates of this magnitude are ictuitively ojicstioned.

These observations demonstrate two major points. The first is the fact that incorporated in the LLNL seismic hazard assessment are extreme estimates of the likelihood of ground motion. Many of the estimates are so high that their validity as a reasonable measure of seismic hazard is questioned since they are inconsistent with the relatively short historic record. For example, there are a number of LLNL hazard estimates that predict average-recurrence intervals of strong-ground motion (PGA 2,0.13g) of 50 years or less.

7

These estimates of the seismic hazard at Plant Farley are inconsistent with the fact that historically, no carthquake ground motions of engineering interest have occurred at the site.

Thus, there is no historical basis for such short recurrence intervals (e.g.,50 years or less).

The second major observation from the information provided in Table I and Appendix A is the fact that a relatively small number of extreme estimates of the seismic hazard can have a major impact on the mean hazard, despite the fact that they have low-relative weight.

The data in Table 1 and past experience suggests two factors contribute to the LLNL extreme estimates of seismie hazard; extreme estimates of the rate of earthquake occurrences and the ground motion models that were selected by expert 52 These topics are addressed in the following subsections.

Selsmic-Activity Rate Assessment As part of a seismic hazard assessment, seismicity experts must estimate the frequency of occurrence of earthquakes in the vicinity of the site. This evaluation, which includes an assessment of the seismic sources near the site and the carthquake recurrence relationship for each seismic source (2,3), determines the spatial and temporal rate of earthquake occurrences. A key parameter in the assessment is the seismic-activity rate (SAR). The SAR defines the temporal rate of occurrence of earthquakes (of any size) per year, above the lower-bound magnitude, which in this case is m,5.0. Thus, the SAR establishes the level of a hazard curve by defining how frequent carthquakes of any size are to occur near a site, if the SAR is incteased/ decreased by a factor of 10, the entire hazard curve is scaled proportionally.

To investigate the seismic activity assumed in the LLNL study, a square region with Plant Farley at the center was defined. The sides of the square were approximately 300 km LLNL GME 5 selected one attenuation model for each (of six) ground motion parameters 2

(e.g., PGA, pseudo-velocity response spectra) that was evaluated in the hazard analysis.

8

from the plant site. When the LLNL hazard calculations were recomputed, the expert assessments of the SAR in this region were determined. The evaluation considered all the sources of uncertainty in defining the regional seismicity, including the variation among the seismicity experts, their uncertainty in seismic-source activity and in the assessment of the seismicity parameters (i.e., a-and b-values). From these results, the probability distribution on the LLNL estimate of the SAR in the vicinity of Plant Farley was determined. Figure 3 shows the LLNL cumulative probability distribution function (CDP) for the SAR. Table 2 summarizes the basic parameters of the SAR CDF.

In a format similar to Table 1, Table 3 lists the top 35 estimates of the SAR. The table in Appendix B lists the top 100 SAR values. Prior to reviewing Table 3, it is useful to note that only 1 event of magnitude 5 or greater was identified in the EPRI carthquake catalog that has occurred in the region near Plant Farley.

As observed for PGA. there are a number of extreme estimates of the SAR. For example, the highest estimate is a rate of 13.9 events per year of m,15.0. Other observations from the data in Table 3 and Appendix B are:

the mean SAR is 1.02x10' events per year, which translates into an o

average recurrence interval for earthquakes with m,,15.0 of approximately ten years, the mean SAR corresponds approximately to the 87* percentile, the top 52 estimates of the SAR, with a cumulative probability weight of approximately 0.02, contributes to 50 percent of the estimated mean, and the top 34 estimates of the SAR exceed a rate of one event per year.

e A practical perspective on the extreme LLNL estimates of the SAR can be developed by considering the number of earthquakes that would be required to sustain these values for a time window equivalent to the period of record. Assuming a period of record of 200 years, Table 2 lists for the SARs shown, the number of earthquakes of m A 5.0 that would be s

9

expected to occur. Recall, since there were 34 estimates of the SAR that es.ceed one event per year, there are a corresponding number that predict 200 or more earthquakes of magnitude 5.0 or greater. The extreme estimates of the SAR and the conesponding expected number of earthquakes signals a significant departure of the LI.NL expert assessments from the historic record. In fact, many of the SAR values in the analysis for Plant Farley predict more events than have been recorded in the !!US as a whole.

Clearly, the task of assessing carthquake-occurrence rates in low-seismic regions is difficult at best. Consequently, it is reasonable to expect that expert assessments will reflect a considerable degree of uncertainty, given the available data. At the same time it is also reasonable to believe that estimates of earthquake occurrence rates have realistic limits, at least in a broad, regional sense, in this context, consideration eculd be given to regions of higher or comparable rates of seismicity. For example, should the assessment of seismic activity in the EUS far exceed the rate of earthquake occurrences in high-seismic regions?

This is a practical question in view of the extreme SARs that are estimated in the LLNL study. For example, there are a number of estimates of the SAR in the LLNL study that rival observed rates in California (if not the entire U.S.). Realistically, the assessment of seismicity in a region must balance the information that is available both locally (in the immediate region of interestl and globally, with the need to provide a practical measure of uncertainty. It appears that as part of the LLNL assessment of seismicity parameters there was a failure to address these issues in the process of soliciting expert input. (Note, based on conversations with Dr. J.li. Savy oi LLNL, this issue is being addressed as part of the seismic hazard evaluation currently Ming performed for the DOli New Production Reactor (NPR) at the Savannah River Site (5)).

Ground Motion Expert 5 Attenuation Models As noted in NUREG-1407 a great deal of concern has been expressed regarding the impact of the ground motion models selected by LLNL expert 5 on the assessment of scismic hazard in the liUS. In parucular, concern has focused on the fact that at many sites the 85"'

10

percentile and the mean seismic hazard curve are strongly affected by the models selected by expert 5. This subsectica pi.:sents a compilation of argunients to support the position that the ground motion models selected by LLNL expert 5 should not be used to estimate seismic hazard results at the Parley site and generally in the EUS. Consequently, it is our position that the LLNL SGX seismic hazard results should not be used as part of the IPEEE binning process.

The presentation in this section is a summary of information regarding LLNL expert 5 ground motion models and the results of recent work that has addressed this issue. The topics discussed are:

comparison of LLNL expert 5 ground motion models with available strong-motion data, including the 1988 Saguenay earthquake, procedure used to derive these attenuation models, seismic hazard estimates derived from expert 5's models, and LLNL NPR seismic hazard assessment.

In the remainder of this subsection a summary of available information is provided.

Comparison With Strong Motion Data - A comprehensive database of strong-motion recordings is not available for the EUS. Nonetheless, it is useful to assess whether the data which is available supports the predictions made by the ground motion models used in the LLNL seismic hazard study and in particular the models selected by expert 5 which he weights 100 percent over all other models.

In 1988 the Saguenay earthquake occurred in castern Canada, producing the largest dataset of strong-motion recordings in eastern North America. In addition, the Saguenay data was not available at the time the LLNL study was performed and thus provides a basis for an interes'ing comparison with model predictions. Two recent studies have made detailed comparisons between the available strong-motion database and EUS attenuation models (6,7).

I1 i

This subsection presents the basic findings of these studies, in a comparative evaluation of the LLNL ground motion models with available Frong-motion data, Reference 6 estimated the mean error (bias) in the model predictions. The assessment was performed for two cases; one using the data available at the time the LLNL study was performed and the second was the same database including the strong-motiet data from the Saguenay earthquake. Figure 4 shov s a plot of the mean, model-to-data ratic for PGA and spectral velocity at 1.0,5.0,10. and 25.0 Hz for the EUS strong-motion database recorded at distances of 100 km or less (excluding the Saguenay earthquake data). Figure 5 shows a similar plot that includes the Saguenay data. The identiGer, G16-A3/TL, corresponds to the models selected by LLNL expert 5.

For the data in Figures 4 and 5, the following obsers ations are made:

for frequencies ranging from 1 to 10 Hz, tb models selected by e

LLNL expert 5 have the highest systematic deviation from the data.

For 1 and 5 Hz these models overestimate ground motion by a factor of approximately 10 and 4, respectively, at 25 Hz the model preferred by LLNL expert 5 is generally e

consistent with the available data, and for PGA, the model preferred by LLNL expert 5 over estimates a

~

ground iaotion by a factor of about 1.5, which is similar to the results for the Nuttii PGA models (SEl and SE2).

Since these observations apply to both datasets, it is coacluded the Saguenay earthquake, despite it's particular unique characteristks, is not significantly different from other information that is available.

Reference 7 conducted a comparative evaluation of the EPRI and 1.LNL attenuation models and the available strong-motion data, including the Saguenay earthquake. The conclusions in Reference 7, which are based on a visual comparison of model predictions and the data are consistent with those statcd above. Based on a direct comparison of the EPRI 12

and the LLNL ground motion models with the Saguenay data, Reference 7 concluded that no single model consistently predicts the ground motion over the 40-200 km distance range.

With the exception of the model selected by expert 5, it was concluded that the set of EPRI and LLNL models, as a whole are consistent with the Saguenay data.

To summarize, studies that compare the EPRI and LLNL ground motion models to available strong-motion data conclude that, overall, the set of ground motion models are generally consistent with the data to within a factor of 2 (except for models selected by LLNL expert 5). This conclusion holds as well when the data from the 1988 Saguenay carthquake data is considered. With respect to the ground motion models selected by LLNL expert 5 the following observations are made:

for spectral accelerations of 1.0 and 5.0 Hz the bias in the model predictions exceeds a factor of 3 and is as high as a factor of approximately 10.

at large distances,40 km and greater, these models tend to over-estimate ground motions at most frequencies (the exception being 25 Hz).

These empirical observations suggest that available strong-motion data does not support the ground motion models selected by LLNL expert 5.

Model Estimation - The ground motion attenuation models (for PGA and spectral velocity) selected by LLNL expert 5 are based on three parameter transformations; conversion of the Modified Mercalli Intensity at a site, I,, to instrumental ground motion (e.g., PGA or spectral velocity), epicentral intensity, l, to I, and I, to mi. In the past, the process of o

estimating ground motion through the use of such transformations has been used when limited instrumental data was available to empirically derive magnitude-distance attenuation l

relationships. However, as a number of studies have pointed out (5-7) there are difficulties t

associated with such procedures, which, if not properly addressed can lead to results that are biased and unrealistic from the perspective of the physical process being modeled.

l l

13

In a recent study the issue of mode 1 estimation as applied by LLNL expert 5 was examined in detail (11). The findings of this study were:

the assumption that intensities (1, and 1) in California and the EUS are similar is not substantiated, intensity data is biased since it does not represent average conditions, e

the transformation (or substitution) procedure used by LLNL expert 5 results in ground motion models that are biased (even given unbiased data), and differences in the spectral shape of ground motions in the western and eastern U.S. were not accounted for.

The drawbacks in the development of the models used by LLNL expert 5 reflect fundamental concerns with the basic data and the process used to derive them. These broad ranging concerns reflect a weak technical foundation for these attenuation models.

Seismic HazanLA_itessments Based On Ewert 5 Models - As illustrated previously in this subsection, the highest estimates of the LLNL seismic hazard at Plant Farley are attributed in

. part, to the ground motion models selected by expert 5. Many of these hazard estimates suggest a high probability of exceedance and corresponding short-recurrence intervals for ground motions of engineering interest (PGA _> 0.13g) in a region of low seismicity, where no evidence exists that motions of this magnitude have occurreo historically.

Reference 2 (Vol. 3, p. 77) reports that the seismic hazard at Plant Farley (for the SGX case) is dominated by the occurrence of large-magnitude earthquakes (r. _>_6.5) in the New Madrid and Charleston seismic zones. Plant Farley is over 600 km from New Madrid and over 400 km from Charleston The fact that these seismic 7.ones contribute to the hazard at Plant Farley is attributed to the low rate of attenuation of the ground motion models selected by expert 5. As a result, since the seismicity near the site is low (e.g., within a distance of 300 km), the expert 5 ground motion models predict that distant seismic sources with a potential to generate large-magnitude carthquakes dominate the site hazard. This 14

observation has been made at other, low-seismicity sites as well (2). Consequently, the impact of the LLNL expert 5 models is greater for low-seismicity sites, such as Plant Farley, that are located a considerable distant from the primary source of seismicity in a region.

LLNL NPR Seismic Ha7ard Assessment - Based on discussions with Dr. J.B. Savy of LLNL, it is our understanding that the ground motion models selected by expert 5 in the USNRC sponsored project are not being used in the current study (5). From this, it is reasonable to conclude that the credibility (as reflected by approximately a 0.20 weight) assigned to expert 5's models is not defendable.

Based on the issues raised above, it is our conclusion that the use of the ground motion models selected by expert 5 in the LLNL study should not be used to estimate the seismic hazard in the EUS. The models have a weak-technical basis and have little ot no support in the scientific community.

Summary of LLNL Seismic Ilazard This subsection summarizes the review of the LLNL seismic hazard assessment for Plant Farley. The review focused on two factors that contribute to extreme estimates of seismic hazard at the plant site: the assessment of the SAR and the ground motion models selected by expert 5. Overall, there are two primary findings of this review, the LLNL estimate of seismic hazard at Plant Farley includes a number of extreme estimates of the rate of earthquake occurrences, and which are inconsistent with historic experience, and the attenuation models selected by LLNL expert 5 are not supported by available strong-motion data and do not have a strong technical foundation and as a result are not believed to be appropriate for purposes of estimating ground motion in the EUS.

The combined effect of these factors are extreme estimates of the probability of exceedance of ground motions of engineering interest at the Farley site.

15

Based on these findings it is concluded that the LLNL seismic hazard results, in particular the SGX case, does not provide a realistic measure of seismic hazard at the Parley site and should not be used as part of the IPEEE binning process. From a much broader perspective, the results of this review apply to the LLNL hazard results at other sites as well.

PLANT FARLEY IPEEE BINNING REEXAMINATION In view of the findings reported in the previous section, a reexamination of the Plant Farley IPEEE binning should be based on a revision of the LLNL seismic hazard for all sites in the EUS. A revised analysis should address the expert assessments of the SAR and the use of the 5GX ground motion models. Since such a revision could not be performed for this :tudy, the LLNL 4GX hazard results are used here as an appropriate alternative.

As a demonstration, Appendix C to this report summarizes the results of a re-evaluation of the LLNL seismic hazard at Plant Farley. The revised assessment is based on the LLNL 4GX case and the results of a statistical screening assessment that provides a basis to screen extreme estimates of the SAR. The results of this assessment are applied to revise the Plant Farley CP used in the IPEEE binning.

This section presents the results of an evaluation of the IPEEE binning, based on the LLNL 4GX huard results. The purpose of this evaluation was to re-assess the binning for Plant Farley. Given the set of RS and FS plants as defined in NUREG-1407 (minus Plant Farley), a comparative evaluation was performed to re-assess which group Plant Farley belonged to. The comparison was performed using the EPRI and LLNL 4GX median, mean and 85* percentile hazard curves. For each parameter, the seismic hazard for pitnts in the RS and the FS (minus Plant Farley) bins was summarized in terms of the range (low to high) and the midpoint of the CP values. The CP (for a given parameter) for Plant Farley was then compared to this summary of the RS and FS plants. Table 4 presents a summary of this data. Figures 6 to 11 are plots of the data in Table 4.

In the figures, Plant Earley is illustrated by a dagger.

16

EPRI Seismic Ilazard Figures 6 to 8 show the CP plots based on the EPRI seismic hazard evaluation. From these figures, the following observations are made:

for the three hazard parameters (e.g., median, mean and 85*

percentile hazard), the cps for the plants in the RS and FS bins define two distinct (i.e., mutually exclusive) groups, and the Plant Farley CP lies outside the range of the FS plants and furthermore it is closest to the midpoint of the RS p! ants (see columns D and G of Table 4).

Fmthermore, if Plant Faricy is included in the RS bin, the RS and FS bins remain as two distinct, mutually exclusive groups.

LLNL 4GX Seismic Ilazard Figures 9 to 11 show the comparison of the Plant Farley CP and the data for the RS and FS plants for the LLNL 4GX seismic hazard. In this case the CP values for the RS and FS plants do not define two distinct groups. Nonetheless, in every case, the CP for Plant Farley lies closer to the midpoiM of the RS plants than to the FS. In all cases the Plant Farley CP lies at the extreme lower end of the FS range and for the mean hazard it is less than the RS midpoint (see Fig.10).

From the results of the EPRI and LLNL 4GX seismic hazard studies, it is concluded t*1at the hazard at Plaat Farley (as characterized by the CP) consistently fits within the RS bin. Based on an examination of the CP for other plants in the FS bin, no other site demonstrates the same level of consisuncy. From this it is concluded that Plant Farley is in a unique position. It is located in the same low-seismic region as other plants in the RS bin.

However, a combination of its location and the extreme seismic hazard estimates produced in the LLNL SGX hazard assessment, adversely impacted the site ranking and the assignment of the plant into the FS bin.

17

Based on the position that the LLNL 5GX hazard assessment does not provide a realistic measure of the ground motion hazard at Plant Farley, the EPRI and LLNL 4GX hazard studies should be used in the assignment of this plant to the RS bin.

'ONCLUSIONS This report presents the results of an reexamination of the Plant Farley IPEEli binning.

The reexamination was motivated by the fact that Plant Farley is favorably kicated ge' graphically in the Gulf Coastal Plain, as are most of the plants assigned to the RS bin (see Fig.1). Despite this. Plant Parley was assigned to the FS group. A review of the n

IPEEE binning process concluded that the assignment of Plant Farley to the FS bin is attributable to the LLNL 5GX seismic hazard assessment for the site. As a result, this review concentrated on the LLNL 5GX seismic hazard analysis.

4 Based on the results of a close examination of the LLNL hazard analysis, it was concluded that the assessment includes a number of extreme estimates of the probability of exceedance of ground motion that significantly impact the metm hazard and the 85th percentile results. Two factors were found to contribute to these extreme seismic hazard estimates:

extreme estimates of the SAR which are inconsistent with historic experience, and the attenuation models selected by LLNL expert 5.

e Based on a close examination of the LLNL seismic hazard results for Plant Farley it was determined that each of these parameters, individually, contributes to hazard results that are inconsistent with historic experience. Their combined impact on the hazard assessment produces extreme estimates of the probability of exceedance of 0.13g PGA corresponding to average-recurrence intervals of less than 50 years.

A review of the LLNL expert 5 ground motion models concluded that they should not 18

I.

be used in the assessment of seismic hazard in the EUS. In addition, due to the large deviation l>etween the LLNL expert assessments of the SAR at Plant Farley and the historic record, it was concludeJ that a reevaluation of the LLNL seismic hazard for all plants should be performed in order to reexamine the IPEEE binning.

Within the constraints of the current study, a revision of the LLNL seismic hazard at all sites in the EUS was not possible. Alternatively, the LLNL 4GX seismic hazard results (which accounts for 1 of the 2 major factors that contribute to extreme estimates of seismic hazard at Plant Farley) were used to reassess the binning for Plant Farley. From this reassessment based on the EPRI and LLNL 4GX seismic hazard results, it was concluded that Plant Farley is more closely aligned with the group of RS plants. As a result it is concluded that for purposes of the IPEEE binni'ig, the LLNL SGX hazard results should not be used (at least for losv-seismicity sites such as Plant Farley) and that Plant Farley should be assigned to the RS bin.

m 19

i

!i Table 1 Summary of the LLNL Seismic Ilazard For PGA of 0.13g For Plant Farley -

Mean Probability of Exceedance = 2.78x10$

Contribution to the Cumulative

= Ground Hazard Curve Mean Hazard (percent)

Mean

- Motion Solemicity Probabil6ty Probabitty of Prel,ebility of Ank Espert Expert Wolght CDF' Individual Cumulative Exceedence Exceedence 1

5 5

3.789E 04 1.0000 5.63 5.63 4.125E-01 1.563E-04 2

5 5

3.789E 04

.9996 4.62 10.25 3.392E 01 2.N.9E 04 3

5 5

3.789E 04

.9992 4.25 14.50 3.118E 01 4.030E 04 4

5 7

3.141E 04

.9989 2.98 17.48 2.635E 01 4.858E*04 5

5 7

3.141E t,4 9985 2.94 20.42 2.597E-01 5.674E 04 6

5 5

- 3.789E 04

.9982 3.18 23.60 2.333E-01 6.558E 04 7

5 5

3.789E 04

.9979 3.11 26.71 2.281E 01-7.422E 04 8

5 9

4.170E 04

.9975 3.03 29.74 2.020E 01 8.264E 04 -

9 5

9 4.17CE 04

.9971 2.52 32.26 1.681E 01-8.965E 04 10-5 9

4,170E 04 9966 2.52 34.78 1.678E 01 9.665E 04 11.

5 7

3.1=1E-04

.9962 1.67 36.45 1.480E 01 1.013E 03 12 5

7 3.141E 04-

.9959 1.66 38.11-1.473E-01

-1.059E 03 13 5

9 4.170E 04'

.9956 2.15-40.27 1.435E 01 1.119E 03 14.

5 5-3.789E 04

.9952 1.69 41.96 1.243E-01:

1.166E-03 15 5

5 3.789E 04

.9948 1.43 43.39 1.050E 01 1.M6E-03 16-5 7

3.141E 04

.9944 1.18 44.58 1.046E 01 1.239E 03 17 5

5-3.789E-04

.9941 1,41 45.99 1.036E-01 1.278E 03 18

_5 5'

3.789E 04

.9937 1.40 47.39 1.029E-01 1.3171f 03 19

-5 9

4.170E 04

.9934 1.06 48.45 7.065E 02 1.347E-03 20 5

5 3.789E 04-9929

.95 49,40 6.977E 02 1.373E 03 21 5

9 4.170E 04

.9926 96 50.36 6.369E 02 1.400E 03-

~

22

.5 5

3.789E 04

.9921 -

. 73 51.09.

5.359E 02 1.420E-03 23 5

5 3.789f 04-

.9918 -

.70 51.79 5.139E 1.439E 03 24 -

5 2

3.361E-04

.9914

.62-52.41 5.110E-02 1.456E 03 25 5-

~5 3.789E 04

.9910

.69 53.09 5.025E 02 1.476E 03 26 5

5 3.789E 04

.9907

.68 53.78'

$.022E-02 1.495E 03 27 5

9 4.170E 04

.9903

.74 54.52 4.923E-02 1.515E 03 28 5

5 3.789E 04

.9899~

.63 55.15..

4.646E-02

-1.533E 03

=29 5[

5 3.789E 04

.9895

.61 55.76 4.464E 02 1.550E 03 30 5

7-3.141E 04

.9891

.48 56.24 4.259E-02 1.563E-03

- 31

'5.

7 3.141E 04

.9888

.47 56.71 4.118E-02 1.576E-03 32 5

2 3.361E 04

.9885

.50 57.20 4.104E-02 1.590E-03

?33

'5 9

4.170E-04 9881

.61 57.81 4.067E 02 1.607E ' - CDF = cumulative probability distribution function 20

.... _.~ -.

Table 2 Summary of the 11NL Seismic-Activity 1 tate In the Vicinity of Plant Farley Number of Events in Parameter / Percentile Events / Year 200 Years 0.15 7.81-3 1.6 0.50 2.54-2 5.1 0.85 8.96-2 17.9 hican 1.02-1 20.4 hiaximum 13.9 2770 21

Table 3 Summary of the LLNL Selsinic Activity Rate For Plant Farley Mean Seismic Activity Rate = 1.02x10' Contribution to the Mean Seismic Activity Cumulative Ground Rate (percent)

Seismic Mean seismic Motion Selsmicity Probability Activity Rate,.

Activity Rate Renk Expert Expert Weight CDF' Individual Cumulative (EventsIYearl (Events /Yearl 1

1 5

4.210E 04 1.0000 5.74 5.74 1.385E*01 5.833E-03 2

1 5

4.210E-04 9996 3.80 9.54 9.177E +00 9.697E 03 3

2 5-4.000E-04

.9992 3.09 12.63 7.854E+00 1.284E 02 4-3 5

3.789E-04

.9988 2.76 15.39 7.393E+00 1.564E 02 5

3 5

3.789E-04 9984 2.66 18.04 7.126E*00 1.834E 02 6

2 5

4.000E 04

.9980 2.11 20.15 5.361E+00 2.048E 02 7

2 5

4.000E-04

.9976 1.54 21.69 3.909E+00 2.205E 02 8

5 5

3.789E 04

.9972 1.32 23.01 3.529E+00 2.339E 02 9

1 5

-4.210E-04

.9968 1.44

24. ?.5 3.480E*00 2.485E 02 10 4

5

.2.947E 04

.9964 98 25.43 3.389E+00 2.585E 02 11 2

5 4.000E-04

.9961 1.28 26.72 3.264E+00 2.716E-02 12 1

5 4.210E 04

.9957 1.35 28.07 3.257E+00 2.853E 02 13 5

5 3.789E 04

.9953 1.19 29.25 3.186E+00 2.973E 02 14 3

5

.3.789E-04

.9949 1.06 30.32 2.852E+00 3.082E-02 15 4

5 2.947E 04

.9945

.82 31.14 2.842E+00 3.165E-02 16 2

5 4.000E-04

.9942 1.11 32.25 2.828E+00 3.278E 02 17-2 5

4.000E 04

.9938 1.10 33.36 2.796E+00 3.390E-02 18 1

5 4.210E 04

.9934 1.15 34.50 2.770E+00 3.507E 02 19 5~

5 3.789E-04

.9930 1.03 35.53 2.766E+00 3.612E-02 20 1

5 4.210E 04

.9926 1.01 36.54 2.432E+03 3.714E-02 21 3

5 3.789E-04 9922

.87 37.41 2.326E+00-3.802E 02 22 4

5 2.947E 04

'9918

.62 38.03 2.12 TE +00 3.865E-02 23 4

5 2.947E 04 9915

.59 38.62 2.036E+00 3.925E-02 24 5

5 3.789E 04 9912

.75 39.36 2.005E+00 4.001E-02 25 4

5 2.947E-04 9909

,51 -

39.88 1.766E+00 4.053E 02 26 3

5 3.789E.9906

.66 40.53 1.761E+00 4.120E 02.

-27 4

5 2.947E 04 9902

.50 41.03 1.729E+00 4.171E 02-28 3

5 3.789E-04

.9899

.62 41.66 1.672E+00 4.234E 02 i '

29 1

5 4.210E 04

.9895

.58 42.~4 1.398E+00 4.293E 02

!?

30 1

5 4.210E-04

.9891

.56 42.80 1.351E+00 4.350E-02

~

31 2

5 4.000E 04

.9887

.52 4*.32 1.329E*00 4.403E 02 h

32 1

5 4.210E;04

.9883

.52 43.83 1.245E+00 4.455E 02 33 4

5 2.947E-04

.9879

.32

.4.16 1.118E+00 4.4BSE 02 34 1

5 4.210E 04

.9876

.45 44.61 1.091E+00 4.534E 02 l

35 4

5 2.947E 04

.9871

.28 44.89 9.793E-01 4.563E 02 l

' CDF = cumulative probability distribution function L

22 L

s m

. y; s

']

k-.-

Table 4 -

Comp' site Probability Comparison:

o

' Reduced-Scope Plants Focused-Scope Plants 4

' A' B'

C D.

E F'

G

Ratio cf Ratio of.

~ Farley Mid-Point to Mid-Point to Cer.sposite Farley Value Farley Value Mid-Point (C/A)

Range Mid-Point

'(F/A)

Pn@ ability Range tO i

EPRI Median Value l

1.2:E-07 1.04E-08 to 2.05E-07.

6.88E 0.57-4.22E-07 to 9.66E-06 2.24E4 18.01 Mean Value 6.79E-07 1.78E-07 to 4.54E-07 3.05E-07 0.44 1.66E-06 to 2.38E-05 8.95E-06 12.14,

85th Percentile 8.09E-07

- 1.59E-07 to 7.13E-07 4.13E.07

' O.51 2.57E-06 to 4.32E-04 1.37E-05 16.44 s

I' LLNL4GX Median Value.

1.73E-06 8.06E-07 to 3.07E-06 1.10E-06 0.64 1.33E-06 to 1.68E-05

. 5.43E-06 3.13 Mean Value'-

2.76E-05 1.49E-05 to 2.92E-04 3.42E-05 1.24 1.91E-05 to 5.25E-04 8.73E-05 3.14 85th Percentile 1.66E-05 1.07E-05 to 4.07E-05 1.61E-05 0.97 1.26E-05 to 1.84E-04 6.19E-05 3.72 1

4

.?-c.e-m gy z

l

/t i

1

/

5v l

r D

T

__lwq D

a D

D D

I 3

l a

M

{

a a

a o

U t

O O O

O O

m' D

i l

se femas go, p 1E Thrtt r M gme.->

O

• c s--

a --

Figure 1 Location of Plant Farley and the IPEEE Reduced-Scope and Focused-Scope Plants.

{

t

l l^

9 10-1

---

85th fractile W

median og 10- -

--- - 15th fractile,

--- mean m

z\\

Od10-3:%

O

' \\>i N

D10-' k. \\.5\\

s a

gs li3 i \\\\

NN m

@10-5 Ns s t

a v,

q<s a<

$10 '

"4 z

r

'ss~

10-7 O. '

200.'

400.

600.

800.'

1000.

ACCELERATION (cm/sec" )

Figure 2 EPRI seismic hazard curves for PGA for Plant Farley.

25

e 4

.! fi

-1.0

'""3

'""I

'"i

'"'I

~ 0.9 ?

~

0.8 }

~

0.7. i i

x 0.6 7

_~=

j-O.5

._a-

~

~

~ O 0.4

- O_ -

0.3 r.-

-e 0.2 ?

i 0.1 i

5 0.0

~

i

' ' ' ~

10' "

10 "

10 -*

10"'

1-10 Seismic Activity. Rate (m3 5)-

Figure 3 LLNL cumulative probability distribution function for all seismicity experts on the seismic-activity rate in the vicinity of Plant Farley.

26

essee RV1 X1 0000o NH-SE1 124 10 A

M NH-SE2 X14 3:

O

• • • • • 16-3/TL (XS :

g g

N O

p,

+

+,

B'1 3

------------------4--------*------------------*-------------~---'~----:

v D*

e 4

o

-;s

~

' Not a BE model for XI

~

O 8

Z

• Not o HE model for X1, X4 Not a BE model for X4 0.1 1 Hz 5 Hz 10 Hz 25 Hz PGA Ground Motion Measure Figure 4 Plot of the mean, model-to-data ratio comparing the LLNL ground motion models tc EUS strong-motion data recorded at distances less than 100 km, excluding the Saguenay data (6).

seese RV1 (X134 '

cocco SE1 (X124 *-

10 :-

M SE2 (X14 A

O

• * * *

• G16-A3/TL (XS :

a Q

o a

n N_

k

~

D Q

p O

y 3 _.......-----..---.---...-...g-----.....*-----,.----------------------------

+

v e

O

~

3

~

O

~

' Not a BE model for X1 8 Not a BE model for X1, X4 -

Z 8 Not a BE model for X4 1

O.1 1 Hz 5 Hz 10 Hz 25 Hz PGA Ground Motion Measure Figure 5 Plot of the mean, model-to-data ratio comparing the LLNL ground motion models to EUS strong-motion data recorded at distances less than 100 i

km, including the Saguenay data (6).

27

i i ii or i i iii;'i

, i iriin i, i.....

Focu vJ Scope E Plants - Midpoint Reduced Scope

1. • Plants - Midpoirt 7 Forley Reduced Scope Range 1

Focused Scope Range l

=

l

,,,, i iii

,,i,at

,,.,,i,,i

,,,,ii.,

10 -*

10

• 10 "

10 ~*

10 -'

Composite Probability Figure 6 EPRI median hazard composite probability results for Plant Farley and the Reduced-Scope and Focused Scope Plants.

.. ii

.oi i i

.. oi i.

,iini

.e iiso Focused Scope-M Plants - Midpoint Reduced Scope t

Plants - Midpoint A

f Farley Reduced Scope Range Focused Scope Range

-m l

i,,,,ini

,,,.....i

, i

,,,,ot

,,iiii, 10 '*

10

• 10

• 10 -*

10 "

Composite Probability Figure 7 EPRI.nean hazard composite probability results for Plant Farley and the Reduced-Scope and Focused Scope Plants.

28

i i i. iiiii

. i..i iii..i.,

Focused Scope E Plants - Midpoint Reduced Scope A

Plants - Midpoint

-- Farley Reduced Scope Range b

i!

Focused Scope dange

=

l

. i i iii.,i

,,,..i

,,,,it 10

• 10~'

10 "

10

• 10 -*

Composite Probobility Figure 8 EPRI 85'" percentile hazi d composite probability results for Plant Farley and the Reduced-Scope and Focused Scope Plants, i i i i..

....ii.i i i i isii.i i iii..

Focused Scope E Plants - Midpoint Reduced Scope A

l Plants - Midpoint i Farley Reduwd Scope Range

! ^-

l Focused Scope Range

=

,,,,,,,,i

,,,,..i

,,,,iiiis

,,iii 10~'

10 "

10 "

10 -'

10 -8 Composite Probability Figure 9 LLNL median hazard composite probability results for Plant Farley and the Reduced-Scope and Focused Scope Plants.

29

... i i iiii i i iiiini i i

>iiini i i i iiin Focused Scope E Plants - Midpoint Reduced Sco[ point e

A Plants - Mi Forley Reduced Scope Range I

i I

t focused Scope Range

=

l

.,,,,,,,i

,,,,,,,,i

,,,,,,,,i 10 "

to" to "

10 "

10 -'

Composite Probability Figure 10 LLNL mean hazard composite probability results for Plant Farley and the Reduced-Scope and Focused Scope Plants.

.. iiiini i i iiiini i i i n ii ni i i iiiin Focused Scope E Plants - Midpoint Reduced Scope A Plants - Midpoint u

i Foriey Reduced Scope Range i1 l

Focused Scope Range l

=

l

,,,,,i

,,,,,,,,i

,,,,,,,,i

,,,,,ii, 10 "

10

• 10

• 10 -'

10 -'

Composite Probability Figure 11 LLNL 85'h perce cile hazard composite probability results for Plant Farley and the Reduced-Scope and Focused Scope Plants.

30

REFERENCES 1.

U.S. Nuclear Regulatory Commission, "Procedurai and Submittal Guidance for the individual Plant Exainination of External Events (IPEEE) for Severe Accident Vulnerabilitics," Ofnce of Nuclear Regulatory Research, Washington, D.C.,1991.

2.

Bernreuter D.L., J.B. Savy, R.W. Mensing, and J.C. Chen, 'Scismic Ilazard Characterization of 69 Plant Sites East of the Rocky Mountains," NUREG/CR-5250, UCID 21m Prepared by l2wrence Livermore National Laboratory for the U.S. Nuclear Regulatory Commission, Washington, D.C.1989.

3.

Electric Power Research Institute,

• Seismic Ilazard Methodology for the Central and Eastern United States," Vols 1-11, NP-4726A, Palo Alto, CA 1988.

4.

U.S. Nuclear Reguhtory Commission,

• Development of Criteria fu Seismic Review of Selected Nuclear Power Plants," NUREG/CR-0098, Washington, D.C.,1978.

5.

Savy, J.B., Personal Communication with M.W. McCann, Jr.,1991.

6.

Jack R. Benjamin and Associates, Inc. " Seismic Ilazard For The Savarmah River Site, A Comparative Evaluation of the EPRI and LLNL Assessments," prerated for Westinghouse Savannah River Corporation, Draft,1991.

7.

Risk Engineering, Inc. " Assessment of the 1988 Saguenay Earthquake-Implication:

On Attenuation Punctions For Seismic Hazard Analysis," Revised Draft Report, Prepared for Pickard, Lowe and Garrick, Inc.,1991.

F.

Cornell, C.A, H. Banon, and A.F. Shakn!, " Seismic Motion m,d Response Prediction Alterr.atives," Eaithquake Engineering and S tructural Dynamit ' Vol. 7, pp. 295 315, 1979.

9.

Veneziano, D and M. Heidari, Statistical Analysis of Attenuation in the Eastern United States," In, " Methods for Estimating Ground-Motion Prediction for the Eastern United States," EPRI Project RP2556-16, Electric Power Research Institute, Palo Alto, California,1986.

L 10.

Veneziano, D., Letter dated December 24,1986 to D.L. Bernreuter. In, Bernreuter l

D.L., J.B. Savy, R.W. Mensing, and J.C. Chen, " Seismic Hazard Characterization of 69 Plant Sites East of the Rocky Mountains," NUREG/CR 5250, UCID-21517, Prepared by Lawrence Livermore National Laboratory for the U.S. Nuclear Regulatory Commission, Washington, D.C.1989.

L l

l 31 L

l

l 11.

McGuire, R.K., " Ground Motion Estimation in Regions With Few Data,"

Proceedings, Eighth World Conference on Earthquake Engineering, San Francisco, California, Vol. 11, 1988.

12.

Veneziano, D., "The Use of Intensity Data in Ground Motion Estimation,"

Proceedings," Workshop on Estimation of Ground Motion in the Eastern United States," EPRI Report NP-5875, Electric Power Research Institute, Palo Alto, California,1987.

y'j%

- {

}t 9

4 4

32

Appendix A RANK ORDER LISTING OF Tile LLNL PGA SEISMIC IIAZAl(D FOR PLANT FARLEY 5

l

i i

Table A-1 l

Summary of the LLNL Selsmic IInzard For PGA of 0.13g For Plant Farley Mean Probability of Exceedance = 2.78x103 l

Contribution to the Curnulative l Gioimd Harord Curve Mean Hazard (percent)

Mean Motion.belemicity Probabikty Probabihty of Probablaty of

sse _

bnk Exped I' Emport Weight CDF' indivkluel Cumuletive Enceedence Enceedence matema usume e -

1 5

5 3.789E-04 1.0000

$.63 5.63 4.125E 01 1.563E 04 2

5 5

3.789E 04

.9996 4.62 10.25 3.392E 01 2.849E 04 Z

5 5

3.789E 04

.9992 4.25 14.50 3.118E 01 4.030E 04 4

5 7

3.141E 04 9989 2.98 17.48 2.635E 01 4.858E 04 5

5 7

3.'41E 04

.9985 2.94 20.42 2.597E 01 5.674E 04 6

5 5

3.7.9E 04

.9982 3.18 23.60 2.333E 01 6.55BE 04 7

5 5

3.789E 04

.9979 3.11 26.71 2.281E 01 7.422E 04 8

5 9

4.170E 04

.9975 3.03 29.74 2.020E 01 8.264E 04 9

5 9

4.170E 04

.9971 2.52 32.26 1.681E 01 8.965E 04 10 5

9 4.170E 04 9966 2.52 34.78 1.678E 01 9.665E 04 11 5

7 3.141E 04

.9962 1.67 36.45 1.480E 01 1.013E 03 12' 5

7 3.141E 04

.9959 1.66 38.11 1.473E 01 1.059E 03 13 5

9 4.170E 04

.9956 2.15 40.27 1.435E 01 1.119E 03 14 5

5 3.789E 04

.9952 1.69 41.96 1.243E te 1.166E-03 15 5

5 3.789E 04

.9948 1.43 43.39 1.050E 01 1.206E 03 16 5

7 3.141E 04

.9944 1.18 44.58 1.046E*01 1.239E 03 17 5

5 3.789E 04

.9941 1.41 45.99 1.036E-01 1.278E 03 18 5

5 3.789E 04

.9937 1.40 47.39 1.029E 01 1.317E 03 19 5

9 4.170E 04 9934 1.06 48.45 7.06$E 02 1.347E 03 20 5

5 3.789E 04

.9929 95 49.40 6.977E 02 1.373E 03 21 5

9 4.170E+04

.9926

.96 50.36 6.369E-02 1.400E 03 22 5

5 3.789E 04

.9921

.73 51.09 5.359E 02 1.420E 03 23 5

5 3.789E 04

.9918 70 51.79 5.139E 02 1.439E 03 24 5

2 3.361E-04

.9914

.62 52.41 5.110E 02 1.456E 03 25 5

5 3.789E 04

.9910

.69 53.09 5.025E 02 1.476E 03 26 5

5 3.789E 04 9907

.68 53.78 5.022E 02 1.495E 03 27 5

9 4.17DE 04

.9903

.74 54.52 4.923E*02 1.515E 03 28 5

5 3.789E 04 9899

.63 55.15 4.646E*02 1.533E*03 29 5

5 3.789E 04 9895

.61 55.76 4.464E-02 1.550E 03 30 5

7 3.141E 04 9691

.48 56.24 4.259E 02 1.563E-03 31 5

7 3.141E 04 9888

.47 56.71 4.118E 02 1.576E 03 32 5

2 3.361E-04

.9885

.50 57.20 4.104E 02 1.590E 03-33 5

9 4.170E-04

.9881

.61 57.81 4.067E 02 1.607E 03

' CDF = cumulative probability distribution function A-2

Table A-1 Suminary of the LLNL Selsinic Ilazard For PGA of 0.13g For Plant Farley (continued)

Contribution to the Cumulative Ground Hazard C urve Mean Hesord (percent)

Moon Motion Geismicity Probability Probability of PsobablHty of Rank Expert Expert Weight CDF' Individual Cumu!stive Exceedence Exceedence i

l 34 5

5 3.789E 04 9877

.52 58.34 3.846E 02 1.621E 03 35 5

9 4.170E 04

.9874

.58 58.91 3.843E 02 1.637E 03 36 5

5 3.78st 04 9869

.51 59.42 3.704E 02 1.651E 03 37 5

7 3.141E 04

.9866 42 59.84 3.703E 02 1.663E 03 38 5

2 3.361E 04

.9862 44 60.28 3.669E*02 1.675E 03 39 5

2 3.361E J4

.9859

.44 60.72 3.632E 02 1.687E 03 40 3

5 3.789E 04

.9856 49 61.21 3.609E 02 1.701E 03 41 5

7 3.141E 04 9852 40 61.61 3.565E 02 1.712E 03 42 5

9 4.170E 04

.9849

.52 62.14 3.485E-02 1.727E 03 43 5

9 4.170E 04 9845

.52 62.66 3.468E 0,2 1.741E 03 44 5

9 4.170E 04

.9840

.51 63.17 3.426F-02 1.756E 03 45 5

9 4.170E 04

.9836

.50 63.68 3.355E 02 1.770E 03 46' 9

4.170E 04

.9832

.47 64.15 3.162E 02 1.783E 03 47 5

7 3.141E 04

.9828

.35 64.50 3.118E 02 1.793E 03 48 5

5 3.789E 04

.9825 42 64.92 3.069E 02 1.804E 03 49 5

2 3.361E-04 9821

.37 65.29 3.068E 02 1.815E 03 50 5

5 3.789E 04

.9818 42 65.71 3.065E 02 1.826E 03 51 5

7 3.141E 04

.9814

.34 66.05 3.045E 02 1.836E 03 52 5

2 3.361E 04

.9811

.36 66.41 2.984E 02 1.846E 03 53 5

9 4.170E 04

.9807

.45 66.86 2.982E 02 1.858E 03 54 5

2 3.3616 04

.9803

.34 67.20 2.793E 02 1.868E 03 55 5

2 3.361E 04

.9800

.33 67.53 2.750E 02 1.877E-03 56 5

9 4.170E 04

.9796 41 67.94 2.704E 02 1.88BE 03 57 5

5 3.789E 04

.9792

.37 68.30 2.678E 02 1.898E 03 58 5

2 3.361E 04 9788

.31 68.61 2.558E 02 1.907E 03 59 5

9 4.170E 04

.9785

.38 68.99 2.545E-02 1.91TE 03 60 5

11 4.027E + 04

.9781

.36 69.36 2.499E 02 1.927E 03 61 5

5 3.789E 04

.f777

.33 69.69 2.t.41E 02 1.937E 03 62 5

5 3.789E 04

.9773

.32 70.01 2.363E 02 1.946E E3 63 1

9 4.633E 04

.9769

.38 70.40 2.306E 02 1.956E-03 64 2

9 4.401C 04

.9765

.34 70.74 2.166E 02 1.966E 03 65 5

2 3.361E 04 9760

.26 70.99 2.110E-02 1.973E 03 66 5

7 3.141E 04

.9757

.23 71.23 2.054E 02 1.979E 03 67 5

9 4.170E 04 9754

.29 71.51 1.914E 02 1.987E 03 68 5

2 3.361E 04 9750

.23 71.74 1.907E 02 1.994E-03 69 5

2 3.361E-04

.9746

.23 71.97 1.895E 02 2.000E 03 70 5

5 3.789E 04 9743

.26 72.23 1.873E-02 2.007E 03 i

A-3

Table A 1 Sununary of the LLNL Seismic liazard For PGA of 0.13g For Plant Farley (continued) 4' Contribution to the Cumulative Ground Harard Curve Mean Hazard (percent)

Mann Probability of Probabliity of Motion Selsenicity Probabilit; e

Rank Expert Empert Weight C DF' individual Curnulative Enceedence Exceedence i

I

.9739

.22 72.45 1.845E 02 2.014E 03 71 5

2 3.361E *'.

72 5

2 3.361(

-4 i

• • T36

.22 72.67 1.817E 02 2.020E 03 73 5

5 3.789E C' i. <' d j

.24 72.91 1.73BE 02 2.026E 03 74 5

7 3.141E # !

.GH)

.19 73.10 1.676E-02 2.031E 03 75 5

7 3.141E-04 F,

.18 73.28 1.584t 02 2.036E 03 76 5

3 3 798t 04

~f/22

.21 73.49 443E 02 2.042E 03 77 5

9 4.1.T*. - C '

.9719

.23 73.71 1.503E 02 2.049E 03 78 5

7 3.141E 04

.9714

.17 73.88 1.482E 02 2.053E 03 79 5

11 4.027E 04

.9711

.26 74.08 1.408E-Q2 2.059E 03 80 5

9 4.170E-04

.9707

.21 74.29 1.391E 02 2.065E 03 81 5

11 4.027E 04 9703

.20 74.49 1.381E 02 2.070E 03 82' 5

7 3.141E 04

.9699

.15 74.65 1.361E 02 2.075E 03 83 5

2 3.361E 04

.9696

.16 74.81 1.337E 02 2.079E 03 84 5

7 3.141E-04

.9692

.15 74.96 1.33it 02 2.083E 03 85 5

5 3.789E 04

.9689

.18 75.14 1.329E 02 2.088E-03 86 5

7 3.141E-04

.9686 15 75.29 1.327E 02 2.092E-03 87 5

11 4.027E 04 9682

.19 75.48 1.292E 02 2.09BE 03 88 5

7 3.141E 04

.9678

.14 75.62 1.253E 02 2.102E 03 89 5

6 2.768E 04

.9675

.12 75.74 1.237E 02 2.105E 03 90 5

7 3.141E 04

.9672

.14 75.88 1.233E 02 2.109E 03 91 4

5 2.947E 04

.9669

.13 76.01 1.219E 02 2.11?E 03 92 5

2 3.361E 04

.9666

.15 76.16 1.211E-02 2.117E 03 93 5

7 3.141E 04

.9663

.14 76.29 1.207E 02 2.120E 03 94 5

2 3.361E 04

.9660

.15 76.44 1.204E 02 2.124E-03 95 5

2 3.361E*04

.9657 1;

76.58 1.180E-02 2.12BE 03 96 5

5 3.789E 04 9653

.16 76.74 1.179E-02 2.133E 03 97 1

9 4.633E 04

.9649

.19 76.94 1.163E 02 2.138E-03 98 5

5 3.789E 04

.9645

.16 77.09 1.152E-02 2.143E 03 99 5

5 3.789E 04

.9641

.16 77.25 1.149E 02 2.14'E-03 100 5

2 3.361E 04

.9637

.14 77.39 1.147E-02 2.151E 03

.a A-4

Appendix 11 RANK ORCER LISTING OF TIIE LLNL MAGNITUDE FitEQUENCY ESTIMATES FOlt Tile ItEGION NEAll PLANT FAltLEY e

i d

Table 151 Sununary of the LLNI, Selsmic Activity Rate For Plant Farley Mean Seismic Activity Rate = 1.02x10'8 Contribution to the Mean seismic Activity Cumulative Otound Rats (percent)

Solemio Mean Selsmic Motion Gelsmicity Probability Activity Rete Activity Rate Rank Empert.

Empert Weight C DF' Individual Cumulative (Events / Yeast (Evente/Yearl a,. na 1

1 5

4.210E 04 1.0000 5.74 5.74 1.385E+01 5.833E 03 2

1 5

4.210E 04 9996 3.80 9.54 9.177E+00 9.697E 03 3

2 5

4.000E 04 9992 3.09 12.63 7.854E + 00 1.284E 02 4

3 5

3.789E 04

.9988 2.76 15.39 7.393E+00 1.564E 02 5

3 5

3.789E-04

.9984 2.66 18.04 7.126t+00 1.834E-02 6

2 5

4.000E 04 9900 2.11 20.15 5.361E+00 2.04BE 02 7

2 5

4.000E 04

.9976 1.54 21.69 3.9J9E*00 2.20$E 02 8

5 5

3.789E 04

.9972 1.32 23.01 3.529E*00 2.339E 02 9

1 5

4.210E 04

.9968 1.44 24.45 3.480E+00 2.485E 02 10 4

5 2.947E*04

.9964

.98 25.43 3.389E+00 2.585E 02 11 2-5 4.000E 04

.9961 1.28 26.72 3.264E+00 2.716E 02 12*

1 5

4.210E 04

.9957 1.35 28.07 3.257E+00 2.853E 02 13 5

5 3.789E 04

.9953 1.19 29.25 3.186E+00 2.973E 02 14 3

5 3.789E 04

.9949 1.06 30.32 2.8$2E+00 3.082E 02 15 4

5 2.947E 04

.9945

.82 31.14 2.842E*00 3.165E-02 16 2

5 4.000E 04

.9942 1.11 32.25 2.828E+00 3.278E 02 17 2

5 4.000E 04

.9938 1.10 33.36 2.796E+00 3.390E 02 18 1

5 4.210E 04

.9934 1.15 34.50 2.770E+00 3.507E 02 19 5

5 3.789E 04

.9930 1.03 35.53 2.766E+00 3.612E 02 i

20 1

5 4.210E 04 9926 1.01 36.54 2.432E+00 3.714E 02 21 3

5 3.789E 04

.9922

.87 37.41 2.326E+00 3.802E 02 22 4

2.947E 04 9918

.62 38.03 2.127E+00 3.865E 02 23 4

5 2.947E 04

.5715

.59 38.62 2.036E+00 3.925E 02 24 5

5 3.789E 04

.9912

.75 39.36 2.005E+00 4.001E 02 25 4

5 2.947E 04

.9909

.51 39.88 1.766E+00 4.053E 02 26 3

5 3.789E 04

.9906

.66 40.53 1.761E*00 4.120E 02 27 4

5 2.947E 04

.9902

.50 41.03 1.729E+00 4.171E-02 28 3

5 3.789E 04 9899

.62 41.66 1.672E+00 4.234E 02 29 1

5 4.210E-04

.9895

.58 42.24 1.398E+00 4.293E 02 30 1

5 4.210E 04 9891

.56 42.80 1.351E+00 4.350E-02 31 2

5 4.000E 04

.9887

.52 43.32 1.329E+00 4.403E 02 32 1

5 4.210E- 04 9883

.52 43.83 1.245E+00 4.455E 02 33 4

5 2.947E 04

.9879

.32 44.16 1.118E+00 4.488E 02 34 1

5 4.210E 04

.9876

.45 44.61 1.091E+00 4.534E-02 35 4

5 2.947E-04

.9871

.28 44.89 9.793E-01 4.563E 02 8 CDF = cumulative probability distribution function L

B-2

~

I Talite 11-1 Sununary of tiie LLNL Seismic Activity Rate For Plant Farley (continued)

Contribution to the Mean $sismic Activhy Cumulative Ground Rito (percent)

Seismic Mean Selsmic Motion Seismicit y Probability Activity Rete Activity Rete Renk E mpert import Wolght CDF' Indiv6 dual Cumulative (Events /Yeerl (Events /Yeari 36 2

5 4.000E 04 9868

.38 45.27 9.606E 01 4.602E 02 31 5

3.789E 04

.9864

.36 45.63 9.592E-01 4.638E 02 38 3

5 3.7t;9E 04

.9861

.35 45.98 9.450E 01 4.674E 02 39 4

5 2.947E 04 9251

.27 46.25 9.276E 01 4.701E 02 40 3.789E 04

.9854

.34 46.59 9.172E b1

4. 736E -02 41 4

5 2.947E 04 9850

.26 46.86 9.058E-01 4.762E 02 42 1

5 4.210E 04 9847

.38 47.23 9.052E-01 4.801E 02 43 2

5 4.000E 04

.9843

.34 41.51 8.740E-0?

4.836E 02 44 3

5 3.789E 04

.9839

.33 47.90 6.725E 01 4.869E 02 45 5

5 3.789E 04 9835

.32 48.22 8.647E 01 4.901E 02 46, 4

5 2.947E*04

.9831

.25 48.47 8.616E 01 4.927E 02 47 3

5 3.789E-04

.9828

.28 48.75 7.539E 01 4.955E 02 48 2

5 4.000E 04 9825

.28 49.04 7.223E 01 4.984E 02 49 2

5 4.000E 04 9821

.28 49.32 7.155E+01 5.013E 02 50 3.789E 04 9817

.25 49.51 6.769E 01 5.038E 02 51 4

5 2.947E 04

.9813

.19 49.77 6.691E 01 5.058E 02 52 3

5 3.789E 04

.9810

.25 50.01 6.653E 01 5.083E 02 53 5

5 3.789E 04

.9806

.25 50.26 6.589E 01 5.108E 02 54 4

5 2.947E 04

.9802

.19 50.45 6.498E 01 5.12BE 02 55 4

5 2.947E-04

.9799

.19 50.64 6.464E 01 5.147E 02 56 4

5 2.947E 04

.9796

.19 50.82 6.381E 01 5.165E 02 57 2

5 4.000E-04

.9793

.25 51.07 6.352E 01 5.191E 02 58 1

5 4.210E 04

.9789

.26 51.33 6.350E 01 5.218E 02 59 3

5 3.789E 04

.9785

.23 51.57 6.275E 01 5.241E-02 60 1

11 4.475E 04

.9781

.27 51.84 6.151E 01 5.269E 02 61 5

5 3.789E 04

.9777

.23 52.07 6.149E*01 5.292E 02 62 1

5 4.210E 04

.9773

.25 52.32 6.042E-01 5.318E 02 63 5

5 3.789E 04

.9769

.22 52.54 5.924E 01 5.340E 02 64 1

5 4.210E 04

.9765

.24 52.78 5.892E 01 5.365E 02 65 2

9 4.401E-04

.9761

.25 53.03 5.822E 01 5.390E-02 66 3

5 3.789E 04

.9757

.22 53.25 5.780E 01 5.412E 02 67 2

9 4.401E 04

.9753

.24 53.49 5.63BE-01 5.437E 02 68 2

9 4.401E 04

.9748

.24 53.74 5.592E 01 5.46?E 02 69 4

2 2.614E-04

.9744

.14 53.88 5.552E 01 5.476E 02 70 2

5 4.000E-04

.9741

.21 54.09 5.427E-01 5.498E 02 71 1

5 4.210E 04

. 9737

.22 54.32 5.409E 01 5.521E 02 72 3

5 3.789E 04

.9733

.20 54.52 5.34BE 01 5.541E 02 B-3

Table 11-1 Summary of tiie LLNL Selsmic Activity Rate For Plant Farley (continued)

Contelbution to the Mean seismlo Activity Cumulative Ground Rate (percent)

Seismic Mean Seismic Motion Salemicity Probability Activity Rete Activity Rete Rank Emport Expert Weight CDF' individual Cumulative (Events / Year)

(Events /Yearl 73 1

5 4.210E 04

.9729

.22 54.74 5.335E 01 5.564E 02 74 2

5 4.000E 04

.9725

.20 54.94 5.184E 01 5.584E 02 75

-1 5

4.210E 04

.9721

.21 55.16 5.179E 01 5.606E 02 76 5

5 3.789E 04

.9717

.19 55.35 5.144E 01 5.626E 02 77 4

5 2.947E 04

.9713

.15 55.50 5.115E 01 5.641E 02 78 2

9 4.401E 04

.9710

.22 55.71 5.041E 01 5.663Ec02 79 1

5 4.210E 04 9706

.21 55.92 5.040E 01 5.684E 02 80 3

2 3.361E 04 9702

.17 56.09 5.Dt9E 01 5.701E-02

' 81 1

2 3.734E 04

.9605

.18 56.27 4.951E 01 5.719E 02 82, 1

5 4.210E 04

.9695

.20 56.47 4.899E 01 5.740E 02 83 5

5 3.789E-04

.9690

.18 56.65 4.767E 01 5.75BE 02 84 2

5 4.000E 04

. 961.7

.19 56.84 4.758E 01 5.777E-02 85 1

5 4.210E 04

.9683

.20 57.04 4.750E 01 5.797E 02 86 1

5 4.210E 04

.9678

.19 57.23 4.696E 01 5.817E 02 87 3

9 4.170E 04

.9674

.19

$7.42 4.668E 01 5.836E 02 88 3

11 4.027E 04

.9670

.18 57.61 4.661E 01 5.855E 02 89 2

9 4.401E 04

.9666

.20 57.81 4.653E 01 5.876E 02 90 3

5 3.789E-04

.9661 17 57.98 4.643E 01 5.893E 02 91 5

9 4.170E 04 9458

.19 58.17

- 4.573E 01 5.912E 02 92 1

5 4.210E 04 9654

.19 58.36 4.536E 01 5.931E 02 93 4

2 2.614E 04 9649

.12

$8.47 4.520E 01 5.943E 02 94 1

5 4.210E 04 9647

.18 58.6A 4.=5BE 01 5.962E 02 95 3

5 3.789E 04

.9643

.17 58.82 4.443E 01 5.979E 02 96 1

5 4.210E 04

.9639

.18 59.00 4.285E 01 5.997E 02 97 2

5 4.000E 04

.9635 17 59.17 4.270E 01 6.014E 02 98 3

11 4.027E-04

.9631 16

$9.33 4.110E 01 6.031E 02 99 3

9 4.170E 04

.9626

.17 59.50 4.093E 01 6.048E 02 100 2

5 4.000E 04

.9622 16 59.66 4.070E 01 6.064E 02 L

B-4 b

t-l

4 Appendix C REASSESSMENT OF TIIE LLNL SEISMIC IIAZARD FOR PLANT FARLEY e

2

INTRODUCTION In the main body of this report it was demonstrated that as part of the Lawrence Livermore National Laboratory (LLNL) seismic hazard assessment for Plant Farley there were a number of hazard estimates that predict relatively short recurrence intervals for ground motions of engineering interest. In many cases, the estimated hazard values are high enough that they can be challenged by the relatively short historic record. This appendix presents the results of a reassessment of the LLNL seismic hazard for Plant Parley.

EXTREME ESTIMATES OF SEISMIC II AZARD Without the benefit of a well-documented, long record of the ground motion that has occurred at a site, engineers and earth scientists have developed and applied a probabilistic modeling procedure to estimate scismic hazard. The seismic hazard model is well-established and the basis for the Electric Power Research Institute (EPRI) and LLNL studies (1,2). References 1 and 2 provide a detailed description of the constituent parts of the seismic hazard methodology. Figure C-1 illustrates the steps in the analysis.

When the elements of the seismic hazard model are combined, a measure of the randomness of earthquake and ground motion occurrences is obtained, as is an assessment of uncertainty. The random occurrence of ground motion is measured by the annual probability of exceedance and the uncertainty is expressed in terms of percentiles of the uncertainty distribution. The aggregation of the various components in the hazard analysis produces thousands of hazard curves; 2750 curves are generated in the LLNL analysis. Figure C-2 shows the LLNL SGX scismic hazard curves for PGA for Plant Farley.

The review of the LLNL SGX hazard results revealed that extreme estimates of the annual probability of exceedance of ground motions of engineering interest can be attributed to the combination of extreme estimates of earthquake occurrence rates and attenuation models that predict high levels of motion, in some cases the estimated rate of earthquake occurrences (m 2. 5.0) exceeded those observed in California.

C-2

Based on the conclusions of the review of the LLNL 5GX seismic hazard, a revised estimate of the LLNL seismic hazard for Plant Farley was determined. This is described in the next section.

REVISED LLNL SEISMIC IIAZARD ASSESSMENT To reassess the LLNL seismic hazard at Plant Parley, the following steps were taken:

eliminate the ground motion models selected by LLNL expert 5, and screen out seismic hazard estimates that have a scismic-activity rate

=

(SAR) for earthquakes of magnitude 5.0 or greater that cannot be supported statistically by the historic record.

To accomplish the first step, the LLNL 4GX hazard results are used, in the second step, a statistical screening analysis is performed that screens the estimate of the SAR associated with.cach hazard curve. The screen is a statistical hypothesis test that evaluates whether the expert assessments of the SAR can be supported by the historic record. For estimates of the SAR that fail the test at the 99 percent confidence level, the associated hazard curves are removed. After normalizing the probability weights of the hazard curves that remain, the final uncertainty distribution on the site hazard is determined.

Statistical Screening Evaluation Due to the relatively short-historic record, there are uncertainties associated with the assessment of the SAR in a region. The uncertainty in the SAR of low-frequency events (e.g., SAR < 102) is due largely to the limited period of record. In order to estimate the statistical uncertainty in the rate of earthquake occurrences in the region near Plant Farley, a bootstrap simulation on the EPRI earthquake catalog was performed. The bootstrap technique was used in the EPRI seismic hazard program as a means to estimate the statistical uncertainty in the assessment of seismicity parameters (1). This method is a robust statistical technique to estimate the uncertainty in parameters derived from a limited database (3). The bootstrap simulation is particularly useful in cases where the uncertainty distribution is not known'or cannot be assumed beforehand.

C-3

The bootstrap simulation is performed on the earthquake catalog, which is the database used to estimate the scismicity parameters. A short coming of the database is the fact that it covers a relatively short-time period, approximately 300 years. Other shortcomings include the problem of incompleteness, uniformity of reporting carthquake size and errors in the j

reported location and magnitude of earthquakes (1,2), in this evaluation the EPRI earthquake catalog and probabilities of detection were used to address some of these issues.

The bootstrapping evaluation is performed by simulating multiple carthquake catalogs.

Each simulation is determined by randomly selecting events from the original catalog. The simulation is performed with replacement (3). Once a set of random carthquake catalogs have been determined, the SAR for the region around Flant Farley was estimated. The result of this process is a sample of SAR values from which sample statistics and a nonparametric distribution can be determined. For purposes of the screening cvaluation, the, nonparametric distribution was used to determine the 99 percent confidence bounds.

Figure C-3 shows the cumulative distribution on the SAR developed in the LLNL study. As noted in the main report and listed in Appendix 11, there are a number of extreme estimates of the SAR in the upper tail of the distribution.

Screening Results A total of 1000 bootstrap simulations were performed in this analysis. Based on the I

results of the bootstrap simulation, the 99 percent confidence bounds on the SAR are; 0.0 to 2.73x10-2 Each hazard curve in the LLNL 4GX case is checked to determine whether its associated SAR is within these bounds, if not it is removed. Figure C-4 shows the revised cumulative distribution function on the SAR. As compared to the original distribution (see Fig. C-3), the upper tail has been truncated and the distribution re-normalized. Although the tail of the distribution has been truncated, there was only a small decrease in the median 2

SAR from 2.54x102 to 1.41x10 events per year. However, with the truncation of the upper-tail, the revised mean SAR is approximately an order of magnitude lower.

l l

C-4

=- -

Figure C-5 shows the revised estimate of the LLNL PGA seismic hazard. Figure C-6 shows a comparison of the revised PGA hazard results and the LLNL 4GX case. Once the extreme values are removed, the revised estimate of the mean is much lower. There is, however, only a small change (less than a factor 2) in the median.

REVISED COMPOSITE PROllAlllLITY FOR PLANT FAltLEY The LLNL seismic hazard results were revised for PGA and spectral acceleration.

Based on these results the composite probability (CP) for Plant Farley was reassessed.

Figures C-7 to C-9 show the comparison of the revised CP for Plant Farley and the range and midpoint data for the Reduced-Scope and Focused Scope plants for the median, mean and 85* percentile hazard. Each figure shows the original CP for Plant Farley as well as the revised value.

The range and midpoint data shown in Figures C 7 to C-9 for the Reduced-Scope and Focused-Scope planis are based on the original LLNL 4GX seismic hazard results. As noted in the main report, ideally the LLNL seismic hazard at each site would be revised to account for the findings in this study. However, it was not possible in this study to reassess the.

LLNL seismic hazard and CP for all plants in the EUS. Nonetheless, the comparisons in Figures C-7 to C-9 are still informative.

If the LLNL seismic hazard assessments for all plant sites was revised, it is reasonable i

to assume that the CP value for each site would decrease. With respect to the Reduced-l Scope plants, the reduction in Plant Farley CP provides an indication of the degree to which the range and midpoint would vary.

l For plants in the Focused-Scope bin it is anticipated that a revision to the LI.NL seismic hazard at these sites would lead to lower estimates of the CP, but not to the same l

degree as sites in the Reduced Scope bin. It is anticipated that the uncertainty in the SAR l

will be larger in low-seismicity regions, in a relative sense, than in higher seismic areas. As l

l a result, revised LLNL seismic hazard results for all sites will lead to a greater reduction in C-5

1 the uncertainty ir. seismic hazard results in low-seismicity regions and a plant binning that better defines the range of Reduced Scope and Focused-Scope plants. In any case, there is no reason to expect that given a revised set of LLNL hazard results, that the CP estimates for the Focused Scope plants will' decrease more than the revised CP values for the Reduced-Scope plants (in fact the opposite likely will oc true).

To summarize, a revised estimate of the LLNL seismic hazard for Plant Farley, results in a revised, lower estimate of the CP. It is reasonable to expect, based on the discussion

- presented here, that a revision of the LLNL seismic hazard at all EUS plants would likely reinforce the conclusion presented in the main report that the revised seismic hazard at Plant Farley is consistent with the revised hazard at Reduced Scope plant sites.

i a

4 f

s F

I l'

f l:

l

[L

!~

l-1-

l C-6

REFERENCES 1.

Electric Power Research Institute, " Seismic Hazard Methodology for the Central and Eastern United States," Vols.1-11, NP-4726A, Palo Alto, CA 1988.

2.

Bernreuter D.L., J.13. Savy, R.W. Mensing, and J.C. Chen, " Seismic Hazard Characterization of 69 Plant Sites East of the Rocky Mountains," NUREG/CR-5250, UCID-21517, Prepared by Lawrence Livermore National laboratory for the U.S. Nuclear Regulatory Commission, Washington, D.C.1989.

3.

Efron, D. "The Jackknife, the Bootstrap and Otiier Resampling Plans," Society for Industrial and Applied Mathematics, Philadelphia, Pennsylvania,1982.

l l

l C-7 L

Sourcei Site l e.

r Distribution on distance Magnitude distribution, source i f (d) for source I, site j fu(m). Ai, p

I(d) fy(m)

D e

Z Distance d m'

n:mn Magnitude m

. Ground motion attenuation Prp5 ability analysis:

' G l m.o(a')

P[A>a' in time t)/t A

Ground gAlm,d(a.)

motion

. level a. A' ~ma6 P(A>a' in time t)/t s

N

-(log scale)

(log scale)

N g

\\'

\\

\\

r i

d Ground motion Distance

' level a (log scale)

.(109 scale) l'igure C-1 Illustration of the steps in a seismic hazard analysis.

+

l l;

g V

C-8 f

.,e

10 -i 5

i 3

1 8

I 3

i E

66**6 0.85

~

0000a 0.50 10~' r

• ~ 0.15 m

j

---

Moon j

0 C

s O

0 0 10 -' r e

o o

x w

. ' r m

o x-o

\\

I E 10 -5 a

5 5

o-o r

L.

10 -' r i

10 -'O 200 400 600 800 1000 Peak Ground Acceleration (cm/s/s)

Figure C 2 LLNL 5GX PGA seismic hazard results for Plant Farley.

l l

C-9

?

4 t

1.0

"'I

'""3

'""i

"'I

t

~

0.9 9:

0.8 i

~

0.7 }

{

MO.6 h h

=

$.0.5 j

~

..o 00.4 L.

CL

~ 0.3 0.2 i

i 0.1 }

5 1

~ ' ' " ' ' ' " " '

0.0 10 -*

10 **

10 ~*

10 "'

1 10 Seismic Activity Rate (m3 > 5)

B Figure C LLNL cumulative probability distribution function for all seismicity experts on the seismic activity rate in the vicinity of Plant Farley, n.

..a r

r C-10

~ ' " ' ' ' ' '

'I 1.0 :

0.9 i 0.8 9 0.7 }:

,p 0.6 }

j

=

O 0.5 1 O

.O O 0.4 u

O.

0.3

\\

0.2 }

0.1 i

0.0

"===':=" ~;

10 -*

10 -*

10 -8 10 -'

Seismic Activity Rate (m3 5)

Figure C-4 Revised LLNL cumulative probability distribution function for all seismicity experts on the : seismic-activity rate in the vicinity of I'lant Farley.

C-11

\\

1 I

i 10 _i 3

i i

i i

i 3

~

• • • • 6 0.85 nosso 0.50 10 -2 r

• • +++ 0.15 r

0 5

---

Mean 5

l 0

1 c

l o

x)O10-' r e

i e

=

O d

,\\,

+ 10 -4

=-

=

o 5

E x)

L C

0 10 -5 N'-

=

=

E a

s e

tr L

u_

10 -* r

~., -

=:

10 -'0 200 400 600 800 1000 Peak Ground Acceleration (cm/s/s)

Figure C-5 Revised LLNL seismic hazard curves for PGA.

i C-12

10  :

i i

i i

i i

i i

i E

E 00000 Mean Median ooooo Mean (Rev%d)d) 5 c) 10 -s

---

Median (Revise 0

C O

T)

O c) 10 -' r i

a)

O X

-G tu g

• 10 "

?

= * 'b o

E

\\,'s y

's U

C N

'8 '.

Q) 10

N T

=

-y N

g-

's, s,%

q)

's',

L.

in

,,8"'..'

10 -'

N

?

=

E

's.

5

.'O..,

s.,,,-

's.

-o-10 7

, I i

l i

I i

i l

i 0

.200 400 600 800 1000 Peak Ground Acceleration (cm/s/s)

Figure C-6 LLNL mean hazard composite probabiley results for the revised

,Omt Farley seismic hazard and the Reduced-Scope and Focused-Scope Plants.

C-13

,iii.,,,,

,,iii.,

i.....ii g focused Scope Plants - Midpoint Reduced Scope A

Plants - Midpoint f

f f Farley (original) h Farlev (Revised)

Reduced scope Range

!^

l Focused Scope Range l

=

l

,, iiii;i i,,..i

,,,,,,,,i 10-'

10 ~*

10 ~'

10 ~'

10

Composite Probability

. Figure C-7 LLNL median hazard composite probability results for the revised Plant Parley seismic hazard and the Reduced-Scope and Focused-Scope Plants.

,,,iiiii i i iiiiii

, i iiiini i i i i i s ii I

Focused SeU[ point e

E Plants - M Reduced Scope g

Plants - Midpoint h

i f Foriey (Originol) h Farley (Revised)

Reduced Scope Range l

l A

Focused Scope Range.

I a

.,,,,iiii

,,, iiiiii i i., iiii i i i i i. ii 10 -*

10 -*

10 "

10 -'

10 ~'

Composite Probability Figure C-8 LLNL mean hazard composite probability results for the revised Plant Farley seismic hazard and the Reduced-Scope and Focused-Scope Plants.

C-14

p.-

u 7

i e

i i iisisi i

. iiii e

i i iiiiii i

i i i i s ii g Focused Scope Plants - Midpoint A Reduced Scope Plants - Midpoint h

Forley (Original)

Farley (Revised)

Reduced Scope Range l -1.

l Focused Scope Range

=

l

.,,.,,,,i

,,,,iiiii

,i,iii,i i

i..i, 10

• 10 -'

10 -+

10~'

10~'

Composite Probability Figure C-9 LLNL 85* percentile hazard composite probability results for the Plant Farley seismic hazard and the Reduced-Scope and Focused-Scope Plants.

C-15