ML20236A800

Forwards Description of Status of Plant Seismic Review,Per Request.Meeting Will Be Requested W/Util to Resolve Audit Finding 1 Issue

ML20236A800
Person / Time
Site:	Satsop
Issue date:	10/08/1987
From:	Philippacopoulo BROOKHAVEN NATIONAL LABORATORY
To:	Pichumani R NRC
Shared Package
ML20236A745	List: ML20236A742 ML20236A800
References
CON-FIN-A-3850 NUDOCS 8710230140
Download: ML20236A800 (6)

Text

_ _. . -

y. f N BROOKHAVEN NATIONAL LAB 6RATORY h, Q l ASSOCIATED UNIVERSITIES, INC. p4 Upton, Long Island. New York ii973 ")y.

(516) 282s '

i FT$ g y 2115 Departrnentof Nuc!eorEnergy f October 8,1987

.h R t Dr. Raman Pichumani facility Operation Branch g' Nail Stop P-1114 .g Phillips Building l U.S. Nuclear Ragulatory Commission 7920 Norfolk Avenue t Bethesda, MD 20814  !

Dear Raman:

i (1 In response to your request, enclosed please find a description of the .

1 status of the WNP-3 seismic review which is being carried out by BNL. Based on this description it should be apparent that at this time there is not a 1

.~

sufficient basis to conclude our review and prepara a TER. In order to complete our evaluation it is proposed to have a meeting with the utility.

The objaetive of the proposed meeting vill be to resolve the " Audit Finding No. 1" issue.

Sincerely youri, -

t -

Jo,4 A. Philippa'copoul.s AJPtkas ,

cet M.' Reich .

C. Costantino -

. P. Betler 8710230140 871014 PDR ADOCK 0500050s

,R PDR i

STATUS OF WP-3 REVIEW (FIN A-3850)-

1

)

The WP-3 Seismic Evaluation is based on a soil-structure interaction l

analysis which is described in Section .3.7 of the WP-3 PS AR. Specifically, l the SSI models are shown in Figures 3.7.2-1 and 3.7.2-2 and' art discussed in section 3.7.2.1.1 of the FSAR. The scil-structure interaction method used is )

of the finite eleaant type and it 'is referrad_ by the utility (WPP$$) to as finite elemer.t/boundarias approach. According to the latter, a deconvolution was performed '(by Ebasco) using the 5KAKE code from which the input at the bottom boundary of a finite element model of the rock-structure system was derived. This is a standard approach used widely to. perform soil-structure interaction ans. lysis. From a review of the results, however, the staff concluded that the deconvolution produced a reduction (over 351) in spectal accelerations at the base of the mat (difference between action at the surface and at the base). According to the staff such a reduction would not be expected for a rock site. This issue was termed as " Audit Finding No. 1". In order to resolve this issue, two options were proposed: i

a. Perform a half-space analysis and apply seismic input directly at the [  ;

base

b. Perform a fixed base analysis. 1-The utility chosen the first option. Specifically, a half-space analysis was performed which is documented in the following reports-WPPSS WP-3 "Saismic Dynamic Analysis Comparison of Pinite Boundaries Approach and Half-Space Approach", June 1984 Subsequent to the half-space avlysis by Ebasco,. an additional analysis

,1 g

was performed by the otility which is called " Owner's Astaysis". This' ~/ O analysis is documented in the following reports WPPSS WP-3: " Validation of Ebasco's Foundation Spring Modeling for Riastic flalf-Space $aismic Model", June 1906.

According to the utility, the " Owner's Analysis" was performed with the objective to fully resolve the " Audit Finding No. 1". Specifically, the utility recreated Ebasco's foundation model which was used to generate the foundation springs for the elastic half-space analysis. Using this model, the utility performed studies to validate the Ebasco's model. .

The objective of our review is to evaluate the adequacy of the utility's submittals reisted to their efforts to resolve the " Audit Finding No. 1". We reviewed the above two (i.e., June 1984 and June 1986) subaittals by the utilfty describing the half-space analysis and the validation studies for the Ebasco's foundstion springs. In order to proceed with our review, a request for additional information was prepared and send to NRC on February, 1987. We receive the response by WPPSS to our questions on August,1987. This information was subsequently reviewed. A list of comments with respect to this information is attached.

As can be seen from this attachment, although some items were resolved on the basis of the WPPS$'s submittal (kly 31,1987) there are still concerne -

which need further justification. There concerns are summarised belows I. Comparison of Finite Element Versus Halfspace Results: .

An important item which is needed in ord6e to resolve " Audit Finding No.

1" is the justification of the differences in floor response specta between finite element and half-space analyses. In the finite element celeulation which is the basis for the seismic design of WNP-3, the input at .

the base of the model was defined through a deconvolution procedure. On the

- . . , i

, l g  ;

4p7 other hand, in the half space analysis the seismic input was applied directly ,

1 to the SSI model at the base. Comparisons of' floor spects at various buildings which were ma'de by WPPSS show significant differences in spectal accelerations. In certain frequency ranges tha finite element results envelopfthecorrespondinghalf-spaceresults. For exampis, the N-S, E-W and vertical floor response spectra obtained from the finics elemenc solution at ,

the top of the Shield Building envelope the corresponding results from tha  :

f half-space analysis (Figures 8,9 and.10 of 1984 report). In this particular -l case the issue of deconvolution analysis is irrelevant since the seinie design basis _results apprear to be conservative as compared to the. corresponding i half-space results. However, this is not always the case as can be seen from stailar comparisons given by WFPSS for the top; of the Rasetor Auxiliary ~1 Building (Figures 11,12 and 13 of 1984 report), containment vessel / vertical compotent (Figure 16 of 1984 report) and other locations. In the June 1984 report the utility concludes that the half-space approach correlates well with l the finite element approach. This conclusion is not clear from the results obtained by WPP88 and needs further , justification.

II. Foundation Springs t! sed in the Ralf-Space Analysist As can be seen from our response to question No. 2 (attachment) the utility did act answer our questione retarding the foundation springe employed in the half-space analysis. Details pertinent to this issue were further amplified during our telecon with WPP88, following the submittal of the

question.

l Based on the review performed so far for the WNP-3 it appears that in order to complete our evaluation and prepare a TER it will be necessary to have a meeting with the utility. The purpose of.this meeting will be to discuss and resolve the " Audit Finding No. 1" issur..

3

a. : _ _ _

j 1

.. 4

,?

/ M. A n:: .u,f )

ATTAcIDatNT 1 l

l With respect to the information provided by the utility, the fo11oving j Comments hre Bade.

j 1

j Question No. ESI-1 l l

The data provided by the utility is satisfactory and show that the calculated shear strains are in teasonable agreement with the available  ;

rock data.

Questio_n_No. SSI-2 s

d. Themethodofanalysisusedtoobtainthesidavall. springs (itemd)is  !

,, l incomplete. Specifically, fixing the bottoa modes of the finite element model does not ensure that the proper stiffnesses result. Since the structure is embedded, it is important to properly assess these I; I

parameters. ,

a. No justification was provided for the choice of sidewall damping I i

parameters as requested in it.am e. The use of the shear asterial damping l to sieuinte sidewall compressive damping is inappropriate. 'j

f. With respect to item f, it is not true that methodologies are not j available to treat this problem. Solutions for the rigid wall case -

sppsat to be reasonable for this. application.

g. A similar response to item a was siven by the utility. j 1

)

q

i

~~~" ~

4 n

. rn

' sc, '*, sa f J

Question No.__S$1-3

a. The applicant's response appears to have misunderstood the intent of the question. In the June 1984 report, the utility has presented data The comparing the finite element approach with the half-space approach.

intent of the question (as completed by telecon following submittal fo the question) was to explain differences in floor response spectra determined by the two methods. The applicant discussed the selection of the width for the finite element mesh used to determine sidewall springs for use in the half-space analaysis. The intent of the question is to speak to the applicability of deconvolution for the site.

b. 34 solved.

c. Essolved.

_____________.______m_-.__._4____ -

.m.______. _ _ _ _ _ _ _ _ _ . . _ _ _ _ _ _ _ ._ _ __ . _ . . m _ .____