ML20206B321

Safety Evaluation Supporting Cycle 9 Core Physics Data. Performance of middle-of-life Core Physics Measurements Verifying Shutdown Margin for End of Cycle Not Necessary

ML20206B321
Person / Time
Site:	Trojan File:Portland General Electric icon.png
Issue date:	04/06/1987
From:	Office of Nuclear Reactor Regulation
To:
Shared Package
ML20206B282	List: ML20206B277 ML20206B321
References
TAC-64532, NUDOCS 8704090078
Download: ML20206B321 (2)

Text

~. - . = - . -. .- - -- . . . - -

I '

I* OFFICE OF NUCLEAR REACTOR REGULATION

, 5ArtIT EVAllIATIUM i

TR N T PORTLAND GTMtHC'TETCTRTC"TMPANY. ET. AL. ' -

CYCLE 9 PHYSIC 5 TEST RESULT 5 j INTRODUCTION -

! During the core physics testing for Trojan Nuclear Plant Cycle 9. several  ;

i measurements were found to be outside acceptance criteria. The control rod '

I worths were lower than predicted. Theworthofall-rods-in-minusone(ARI-1)

configuration as measured by the reactivity computer was different from that '

measured by the boron end point. In addition, the neutron flux in the central 4

portion of the core was higher than predicted and the neutron flux on the periphery was lower than predicted at Seginning of Cycle (BOC), Hot Zero Power i

(HZP) conditions. Extrapolation of the 800 data to E0C conditions did not i predict adequate shutdown margin at EOC. PGE stated (Ref.1) that additional j red worth measurements between Middle of Cycle (MOC) and E0C were being

• considered. After further analysis of the measured data and' additional j verification of the predictions, PGE and Westinghouse concluded that additional measurements were not necessary.

DISCUSSION AND EVALUATION 4

PGE and Westinghouse have investigated the physics test results. The major {

question was whether the BOC data would predict adequate shutdown margin at E00 The investigation centered primarily on two areas: the validity of the

reactivity computer measurements and the accuracy of the boron samples. In

addition, the predicted data was examined to be sure that it was representattve of conditions in the reactor. Two ma,ior areas.affecting the predictive model were the assumed burnup of previous cycles and the, core baffle assembly modeling i treatment. t l ..

l The validity of the reactivity computer data is influenced in part, by .

uncertainties in the delayed neutron parameters, inability to model the changes as power distribution changes due to rod insertion and errors due to background

~

l ganna signal due to the flux detector. The combined use of a low leakage core  !

loading pattern and the short refueling shutdown caused the background gamma 4

signal to be higher this cycle than for previous cycles. These effects caused the reactivity computer to indicate values which for the ARI-1 measurement were lower than expected. Also, a review of the reactivity computer traces indicated i some boron dilution had continued in the initial phases of the AR!-1 measurement.

, This would also affect the accuracy of the reactivity computer data. When the 1 reactivity computer measured data were compensated for this dilution, the results were within the acceptance criterion.

The accuracy of the boron samples was examined by having backup analyses

! independently perfonned by Westinghouse. Westinghouse results confirmed the j original data.

• i 1

e i

l 87040 g 8"870406*

j E" S*

4' It was concluded that the unusual difference between the rod worth measurements obtained by the reactivity computer and the worth indicated by the boron end point measurements was due to the particular accumulation of errors in the reactivity computer data.

The predictions for the Cycle 9 tests were also ex'amined. A discrepancy .in true core power vs. indicated power level as a result of feedwater flow venturi i

fouling resulted in a difference between actual cycle burnup and the burnup 4 that was used in the original calculations. In addition a revised core baffle assembly treatment which took into account observed trends in power distribution behavior at Trojan was developed and incorporated in the predictive models. When these changes were made to the models, the control rod worth and power distribution measurements vs. predictions at BOC improved.

CONCLUSION AND PFCOMMENDATIONS We have reviewed the licensee's submittal as well as additional information and' conclude that the errors exist in the reactivity computer data and that the boron data are more accurate. We also agree with the licensee that adequate shutdown margin exists in the Trojan reactor for (00. We suggest that future boron dilution measurements of shutdown banks and the N-1 configuration be done l

slower. In addition, they should be done such that boron dilution does not occur during the swapping of the highest worth rod with the shutdown bank.

PRINCIPAL CONTRIBUTOR:

M. Chatterton Reference 1: Letter A. E. Chaffee, NPC to Portfand General Electric

< Company (Att: BartD. Withers) August 20,1986.

g.

i e

B