ML20213A512

Forwards Evaluation of Cycle 9 Startup Physics Testing & Westinghouse Evaluation of Plant Control Rod Worth Measurements.Based on Evaluations,Addl Measurements Not Required

ML20213A512
Person / Time
Site:	Trojan File:Portland General Electric icon.png
Issue date:	01/27/1987
From:	Lindblad W PORTLAND GENERAL ELECTRIC CO.
To:	NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
TAC-64532, NUDOCS 8702030318
Download: ML20213A512 (8)

Text

,

O POHTLAND GENEMAL ELECTRIC COMI*ANY 821 S. W. S ALM ON STRFET WILLIAM J. LINDBLAo PO RT LA N o. O REGON 97204 1503}226-8875 January 27, 1987 Trojan Nuclear Plant Docket 50-344 License NPF-1 U.S. Nuclear Regulatory Conunission ATTN: Document Control Desk Washington DC 20555

Dear Sir:

Trojan Cycle S Core Physics Data Pursuant to discussions between T. L. Chan of the NRC and G. A. Zimmerman of PCE, attached is our documented evaluation of the Cycle 9 Core Physics Data.

Initially, because of differences between design predictions and test data for shutdown margin, consideration had been given to performing taiddle-of-life (MOL) core physics measurements to verify the accuracy of the antici-pated end-of-life design conditions. We have recently concluded that additional measurements are not required. That conclusion is supported in Attachments 1 and 2, which are evaluations of the Cycle 9 data by PGE and Westinghouse, respectively. At PGE, formal evaluations were performed by the Fuel Operations Department as well as the Plant Staff, and discrep-ancies in the data were reviewed by the Plant Review Board and the Trojan Wuclear Operations Board. Each of these separate evaluations and reviews i concluded that additional core physics measurements are unnecessary; with continuing reviews specified by the TNOB.

We would be pleased to discuss these results with you.

l Sincerely, i "S _

i Attachments c: Mr. David Kish, Acting Director State of Oregon Department of Energy I

l Mr. John B. Martin Regional Administrator, Region V U.S. Nuclear Regulatory Commission Mr. Gary Kellund Resident Inspector Trojan Nuclear Plant 8702030318 870127 O PDR ADOCK 05000344 P-PDR lh

. ~ . _, - . - . . . ..-- - . _

Trojan Nuclear Plant Document Control Desk Docket 50-344 January 27, 1987

-License NPF-1 Attachment 1 Page 1 of 5 EVALUATION OF CYCLE 9 STARTUP PHYSICS TESTING

Background

j During the performance of the Core Physics Testing following startup for Cycle 9 reactor operations on June 18, 1986, several measurements I

were outside the acceptance criteria. The anomalous results are categorized as follows:

i

1. Control rod worths were generally lower than predicted. l

2. The worth of all the control rods as indicated by the boron end l points for the all-rods-in-minus one (ARI-1) condition was

inconsistent with the reactivity computer measurement.

l

\

7

3. Neutron flux in the central portion of the core was higher than predicted and neutron flux on the periphery of the core was lower

, than predicted at BOL, HZP conditions.

i An evaluation was initiated to determine the root cause and effect of l

( the unexpected data, and to ensure that EOL conditions will meet the

} Technical Specification requirements for shutdown margin.

Discussion The measured control rod worths for Control Banks B and A, and the combined worth for control Banks D, C, and B were outside the

PCE-established acceptance criteria of the Reload Cycle 9 Startup Low Power Physics Test. Additional measurements were taken, usirs both the reactivity computer and boron samples. The ARI-l configuration, not l originally scheduled to be measured, was determined. Note that the

! predicted values for Control Banks ~B and A have differed from measured values more than the values predicted for other banks for the last several years. The bank worth for Control Bank B, located near the i

perimeter of the core, has normally been predicted higher than measured, ,

j and the worth for Bank A, located near the center of the core, has j normally been predicted lower than measured.

Investigation concerning the core physics data concentrated primarily in two areas:

l l 1. Validity of the reactivity computer.

+

Inaccuracles in the reactivity computer data are due to uncertainties in the delayed neutron parameters, inability to model the changes

! occurring in the delayed neutron parametero ao power distribution

changes due to rod insertion, inaccuracy of the analog reactivity i

9 I-

. _ _ - - ~ .. _ . ,

t Trojan Nuclear Plant Document Control Desk lI Docket 50-344 January 27, 1987 License NPF-1 Attachment 1 Page 2 of 5 4

i computer components, inaccuracles in interpreting the reactivity computer traces, errors due to the use of a point kinetics technique i

for a three-dimensional reactor, errors due to a background gamma signal to the flux detector (which is more significant when reactor power is extremely low), and errors due to thermal feedback if reactor power nears the point of adding heat during data measure-1 ments. The combined use of a low-leakage core loading pattern and the short duration refueling shutdown caused the background gamma j signal to be higher this cycle than in most previous cycles. Efforts 3 to attain a low-leakage core loading pattern had been initiated in Cycle 5. However, a complete low-leakage core loading pattern was not attained until this cycle. Core physics measurements during the previous two cycles had indicated that the neutron-to-ganna signal ratio was lower that in prior cycles. The fact that the refueling outage was shorter than other refueling outages did not permit the gamma flux to decay as much as in the past. These effects caused the reactivity computer to indicate reactivity values which, for the ARI-1 measurement, were lower (after attempts to compensate) than actual by approximately 2 percent. Also, the reactivity computer j checkout, which was performed by comparing the reactivity indicated by the reactivity computer when the reactor is on a stable reactor

period to the value calculated from the measured reactor period,

! indicated that the reactivity computer was reading low by approxi-

mately 3 percent. These inaccuracies contribute to an inherent error

in the reactivity computer data of at least 17 percent. Over the i course of an extended measurement program such as an ARI-1 determina-l tion, a reactivity computer bias error may tend to accumulate. We believe that the combination of uncertainties in the reactivity computer measurements caused the ARI-1 rod worth measurement to be

, low.

The calibration of the reactivity computer was verified to have been performed by experienced personnel in accordance with equipment i operating instructions. The observed indications of errors were

! within the 17 percent inherent error band and within the acceptance l criteria.

As a further check of the measured data, the reactivity computer data was sent to Westinghouse for analysis. Westinghouse obtained essentially the same rod worths from the trace data as PGE had. r f 2. Accuracy of the boron samples.

l l

To verify the results obtained by means of the boron samples, additional boron analyses were performed. Backup analyses were

! independently performed by Westinghouse with essentially the same i measurement results. It was determined that the uncertainties in the baron moaouromonto woro approximatoly 13 porcont, banad upon tho methodology used to determine the boron concentration. The errors i

h

Trojan Nuclear Plant Document Control Desk Docket 50-344 January 27, 1987 4

License NPF-1 Attachment 1 Page 3 of 5 4

associated with an ARI-l worth measurement based on two boron samples i

(before and after conditions) will tend to be limited since the boron sample values are subtracted. Thus, the errors tend to cancel.

It was concluded that the unusual difference between the rod worth l

measurements obtained by the reactivity computer and the worth indicated by the boron end point measurements was due to the particular accumula-tion of errors in the reactivity computer data. Furthermore, a review of i the reactivity computer traces indicated some boren dilution had con-tinued during the initial phases of the ARI-l reactivity measurements.

When the reactivity computer measured data are compensated for this

dilution, the results are within the acceptance criteria (shutdown margin j of greater than 1.6 percent at both BOL and EOL) in the Trojan Technical a

Specifications and are reflected in the data shown on Table 1 of Attachment 2.

Since the measured data obtained via boron analyses is believed to be accurate, an examination was made to determine if predicted data was representative of conditions in the reactor. Several areas affecting the predictive model were examined, two of which were the assumed burnup of i previous cycles and the core baffle assembly modeling treatment.

The core follow data compiled at Trojcn over the last several cycles indicates that true core power has been less than indicated power level l- and the difference is attributed to feedwater flow venturi fouling. This i

discrepancy results in a difference between actual cycle burnup and the i burnup that is calculated assuming that indicated power is accurate.

Although adjustments were made at the end of each cycle based on boron letdown data in an attempt to account for the burnup difference, the adjustments were found to be too small for Cycles 7 and 8.

i

A second item investigated was the baffle modeling. A revised core baffle assembly treatment, taking into account observed trends in power

' distribution behavior at Trojan, was developed and incorporated in the predictive models. The new model resulted in a small reduction in pre-dicted boron concentration throughout Cycle 9, as well as a beginning of cycle inboard power shift of approximately 3 percent which decreases to less than 0.5 percent at end-of-cycle.

In addition, a small error in predicted boron concentration and power distribution at BOL, HZP conditions was found to exist due to the reduced power operation at the end of Cycle 8 which affected the amount of samarium in the fuel at the beginning of Cycle 9. These changes to the models improved control rod worth predictions and power distribution

- predictions at BOL while not significantly changing predictions at EOL.

Trojan Nuclear Plant Document Control Desk Docket 50-344 January 27, 1987 License NPF-1 Attachment 1 Page 4 of 5 Conclusions Based on the above evaluations and considerations, it was concluded that errors existed in the reactivity computer data, and that the chemistry data is more accurate. Secondly, it can be concluded that adequate shutdown margin exists in the Trojan reactor at beginning-of-life and end-of-life and additional core physics measurements are not needed.

The attached table summarizes the calculated / measured rod worth and shutdown margin for the Trojan Nuclear Plant for Cycle 9.

An investigation is continuing to determine the potential impact of B-10 depletion on core physics measurements. A sample has been sent to Westinghouse to determine the quantity of B-10 in the boric acid used at Trojan. Results are expected by January 31, 1987 and will be accounted for, as required, in calculations of critical and shutdown boron concentrations. The TNOB will further review this issue upon completion of the B-10 measurements.

GAZ/mr 0995P.187

. .--.-- - . . . . - , -- - --- . ~ . . - . . . - - . - - . - . . - - - . - - - - .

j .. ~ a.

1 i

C?f

8 a "Ea

.mnm m

l Calculated / Measured Rod Worth and Shutdown Margin (in' pcm) m z

j -E??

4

?*r  ;

Worth H^*

All Rods In Required ,

]

, Except Most Available Tech Spec p ,

Reactive Design Shutdown. Shutdown S--

(ARI-1) Requirements Margin Margin Conunents 3

W Calculated Worth 5094 Beginning of Life (BOL) Less 10%* -2480 2105 1600 505 pcm > Tech spec limit

{

= 4585 i

j Measured Worth BOL 4803 -2480 2407 1600 807 pcm > Tech Spec limit Using Boron Samples i

l W Calculated Worth 5586

End of Life (EOL) Less 10%* -3300 1727 1600 127 pcm > Tech Spec limit

= 5027

?

5245 -3300 1945 1600 345 pcm > Tech Spec limit U ng r S s i

1

! 1 i i i

1 2

• Additional uncertainty applied to rod worth calculations.

?NU

%2aoo

• Q'Q 1 aa m .

w " .~ i H

1362W "

}

i .

1 l

i

• Trojcn Nuclear Plant Document Centrol Desk Docket 50-344 January 27, 1987 License NPF-1 Attachment 2 l Page 1 of 2 WESTINGHOUSE EVALUATION OF TROJAN N-1 CONTROL ROD WORTH MEASUREMENTS An extensive review of the neutronics models and measured data for Trojan Cycle 9 as well as previous cycles has been performed. The review led to improved prediction versus measurement results for the Cycle 9 core by accountiny, for less than full power operation in previous cycles. Due to feedwater flow ventuel fouling Trojan has consistently operated slightly below 100 percent power, hence actual burnup is less than calculated fuel burnup and actual reactivity is slightly higher. For Cycle 9 and future cycles, Plant calorimetric evaluation is expected to eliminate this dis-cropancy. Westinghouse has since evaluated this phenomenon and adjusted their predictions for this effect. These changes resulted in improved agreement between measured and predicted N-1 control rod worth at beginning-of-life (BOL), while not significantly changing predicted N-1 worth at end-of-life (EOL). At BOL, uncertainties in assembly burnups and fission product concentrations in reinserted fuel often lead to small IN-OUT tilts for Trojan. These discrepancies burn out with time because assemblies operating at higher power levels will deplete faster and, consequently, lose reactivity faster than those assemblies operating at lower power which tends to reduce assembly reactivity differences between predicted values and measured values with cycle operation. As a result of this effect, EOL power distributions essentially do not exhibit IN-0UT power tilts, and more than sufficient shutdown margin to meet the techni-cal specification requirements at EOL is predicted. The improved pre-dicted values with their associated revised errors are shown on Table 1, with the original predicted and measured values.

The ARI-l worth based on boron end points at BOL has been measured to be within the 10 percent allowance when compared to the predicted value.

The boron end point measurements were verified by Westinghouse chemists to give the same N-1 worth reported by PGE chemists. Westinghouse believes that these boron end points are the most reliable source of data for BOL N-1 worth. Westinghouse recommends the use of boron end points as the most accurate means for determining that the shutdown margin is met.

Based on previous cycle operation and core behavior with burnup described

above, the EOL ARI-1 worth is expected to fall within the 10 percent allowance of the predictions. Based on the current rod insertion limits, an additional margin exists in the Rod Insertion Allowance (RIA) at EOL j which reduces the RIA requirement. This reduces the EOL reactivity design requirement and increases the calculated shutdown margin to j 1.73 percent. Consequently, there is no shutdown margin concern when boron end points are used as the basis for control rod worth determination.

In summary, analysis of the Cycle 9 startup results show that sufficient shutdown margin is available throughout cycle operation and no additional data is necessary. Westinghouse will continue to oxamina and improve our models based on operational experience.

o C5Y

;L6L 85a TABLE 1 'S w2 TROJAN CYCLE 9 STARTUP

SUMMARY

E?5 7y7 ec~g Original Revised' g, Parameter W Pred. W Pred. Measured Original Error Revised Error 7 (M - P) (M - P R) 5.

P PR HZP ARO Boron (ppm) 1558 1579 1587 29 ppm 8 ppm HZP Din Boron (ppm) 1435 1454 1468 33 ppm 14 ppm Bank D worth (ppm) 123 125 119 -3.3% -4.8%

Bank D worth (pem) 1053 1056 1012 -3.9% -4.2%

Bank C worth (pcm) 909 959 850 -6.5% -11.4%

Bank B worth (pcm) 1242 1125 891 -28.% -20.8%

Bank A worth (pem) 493 575 576 17.% 0.2%

DCBA Boron End Point (ppm) 1130 1143 1150 20 ppm 7 ppm DCBA worth (ppm) 428 436 437 2.1% 0.2%

DCBA worth (pcm) 3694 3715 3329 -9.9% -10.4%

N-1 Boron End Point 942 978 1022 80 ppm 44 ppm N-1 worth (ppm) 616 601 565 -8.3% -6.0%

N-1 worth (pcm) 5324 5094 4343 to 4526* -18.4% to -15.0% -14.7% to -11.1%

N-1 worth (pem-BW)** 5236 5109 4803 -8.3% -6.0%

Boron Worth (pem/ ppm) -8.5 -8.5 -8,5 0 0

• The measured value of 4343 pcm does not account for observed boron dilution and tempera-ture change effects that occurred during the K-6 (the most reactive rod) rod swap with Shutdown Banks A and B, and is not affected by core flux distribution skewing. The measured value of 4526 pcm was derived using reactivity computer worth during the rod swap gsgg.gr of K-6 with Shutdown Banks A and B and assumes that the core flux distribution skewing gggg would not adversely affect the reactivity computer results. ,, o g g

• • BW - value obtained by conversion of ppm to pcm using the measured borun worth, o *N mgUn 82 S 0;"

5E

?

CAZ/kal $-

1354W