ML20198N840

Non-proprietary Version of Prairie Island Unit 1 Cycle 19 Thimble Deletion Study

ML20198N840
Person / Time
Site:	Prairie Island
Issue date:	01/06/1998
From:	Lesko J, Srinilta S WESTINGHOUSE ELECTRIC COMPANY, DIV OF CBS CORP.
To:
Shared Package
ML20070L182	List: ML20198N814 ML20198N824 ML20198N830 ML20198N840 ML20198N855
References
NUDOCS 9801210231
Download: ML20198N840 (24)

Text

- . - - . _ . - - . -- . . - . - . _ __- _ . -.

I Desunghouw Non-Pmpnetary Clau 3 l

a PRAIRIE ISLAND UNIT 1 CYCLE 19 THIMBLE DELETION STUDY ,

i t

Y ~ $^^Y a (plp)

S. Srinilta Core Analysis B Date: ,

!S h i

~

Verified:

. R. Lesko 1 Core Analysis A Date: /,!( I'I C 1998 Westinghouse Elect ic Company All Rights Resened 1

l PRA!RIE ISLAND UNIT 1 AND UNIT 2 EVALUATION OF TillMBLE DELETION ON PEAKING FACTORS Introduction A study utilizing Prairie Island Unit I and 2 flux maps was undenaken to assess incremental peaking factor measurement uncenainties associated with a reduction to a minimum of 18 of the 36 of the movable detector (M/D) thimbles. Due to a large database used in the study, it is intended that the uneenainties quantified herein is to be considered of a generic nature and applie'able to subsequent cycles.

Section I of this study presents the rrrthodology and results of rA.mly deleting thimbles from actual INCORE maps to quantify the uncertainties. Section 2 quantifia the minimum number of thimbles per quadrant required in order to improve the ability to distinguish between random and systematic thimble deletion events and to establish the bounds of applicability of 5cetion 1.

For Prairie Island Unit i Cycle 19, an evaluation was perfonned to confirm applicability of this cycle to the study described herein. Review of current cycle flux maps indicate that measurement to predicted peaking facton are well within the required measurement uncenainties and indicate the core is behaving as predicted.13ased on this, it is not anticipated that the core will not perform as expected for the remainder of the cycle, it is not expected that the additional uncenainties on the peaking factors will result in any violation of the limits. Even with the increased measurement uncenainty applied as a result of the thimble detection study, the Prairie Island Unit I Cycle 19 Fn Surveillance Technical Specification will provide necessary protection.

When referring to percentages in Sections 1 and 2, they refer to the percentage from a total of 36 thimbles unless otherwise specified.

2

4 SECTION 1 hiETilODOLOGY GENERAL To assess the additional peaking factor measurement uncertainties associated with as few as 50% of the i ht/D thimbles available, ten full core INCORE flux maps from Prairie Island Unit I and Unit 2 (5 maps per unit) were used. The selection of these maps was made to cover the entire cycle bumup ranges. For each of the INCORE maps, five separate random deletions were made, giving a total of 50 thimble deletion cases with 50% of the thimbles available. Five separate random deletions were also done with 'his same set of 10 INCORE maps giving 50 thimble deletion cases with 75% of the thimbles available. The INCORE code was used to randomly delete thimble locations. The measured peaking factors for the thimble deletion maps were then compared with the measured peaking factors in the reference maps, i.e., the INCORE maps employing all or most of the 36 movable detector thimbles. Figu ; I shows the movable detector (hi/D) locations for the Prairie Island plants considered and Table I provides additional information on the 10 maps used in the study. For those maps with less than 100% of available thimbles (e.g.,80.6%), thimbla deletion cases were run deleting 50% of the total 36 thimbles (e.g. 30.6%). These comparisons yielded the additional measurement uncertainties to be applied to F,n and Fo. Thimble deletion effects on the INCORE measured axial olTset and quadrant tilt were addressed in a similar manner.

htEfilODOLOGY STATISTICAL The percent error between the reference peaking factor value, Fu,(Reference), and the thimble deletion case peaking factor value. Fu,(T.D.)is defined in Equation 1 as Fa(T.D.) (Eq.1)

% Error (T.D.) = x 100 1 Fa (Reference),

where Fu is F,n, or Fo and T.D. refers to 75% or 50% of available thimbles in the reference case. A positive value of error implies that the peaking factor from the thimble deletion map is non-conservative relative to the reference in the following paragraphs the error will be denoted Xi where i refers to one of the 10 flux maps and j refers to one of the 5 thimble-deletion cases for each map. The percent error between the reference value and h thimble deletion case value for quadrant tilt and axial offset are defined in Equations 2 and 3 as Error (T.D.) = (Ref. - Deleted) x 100 for QUAD Tilt (Eq.2)

Error (T.D.) = (Ref. Deleted) for A.O. (Eq.3) 3

m 4

i The mean error for map i, X; and the percent relative sample standard deviation for map i, S;,are defined in Equationn 4 and $, respectively.

5 1

-X = - E Xu . (Eq.4)

J=1 t

t NI!!

3 E (xy - 5)'

je S=i

~~

(Eq. 5) 5~1 After computing X' and S ifor cach map, for each parameter ofinterest, and for both 50% and 75% thimble deletion cases, the data is combined, The combined mean for all maps, Xeu,is given by Equation 6 as:

10 1

--Xas,a = g E Xs- (Eq. 6) i=1 u

4 4

The i ombined percent relative sample standard deviation of all maps is given by Equation 7 as:

f

^ '"'

" 10 E -((N, - 1) S! + No 1,')

n=1 - Nr Smu.a = - X,>,r

• gr Nr - 1 a > >

where:

N, = Number of random deletion cases of each map = 5 and Nr = Total number of datapoints = 10 maps x 5 deletions / map = 50 Equations 6 and 7 are constructed in such a manner that if one were to directly compute the mean and standard deviation for all 50 datapoints, the same numeric results would be obtained.

Aner 5caNad and Se,,,a have been obtained for each parameter ofinterest, and for both 50% and 75%

thimble deletion cases,95% confidence / 95% probability one-sided upper tolerance limits are constructed to quantify the thimble deletion uncertainty component (See Equation 8).

Thimble Deletion Uncertainty Component (%) = 5caNad + kSw,,,a (Eq.8) where k = the one-sided 95% confidence /95% probability tolerance limit factor for 49 degrees of freedom =

2.065.

Application of the above methodology is presented in the "Results" section of this report. The statistical combination of the thimble deletion uncertainty component with INCORE measumment is discussed in the

" Thimble Deletion Uncertainty" section of this report.

RESULTS Table 2a provides the peaking factors sample mean (%) for each map (see Equation 4) and the sample standard deviation (%) for each map (see Equation 5) for the 50% thimbles available case. The combined sample mean (%) and the comb lned standard deviation (%) for each parameter of interest, as calculated per Equations 6 and 7, is also shown. Table 25 presents the analogous information for the 75% thimbles available case. Tables 2c and 2d provide the sample mean and the sample standard deviation for quadrant tilt and axial offset over the same database.

5

.- Thimble deletion uncenainty components (i.e. the 95% probability,95% confidence tolerance limit) for Fsi, and Foare calculated in Appendix A using Equation 8 and are based upon the data of Tables 2a and 2b. The Thimble Deletion Uncenainty Component (%) is plotted in Figure 2 as a function of Percentage of Thimbles Available. This figure is provided for information only and is not directly used in the uncenainty application.

THIMBLE DELETION UNCERTAINTY Current flux map peaking factor measurement uncenainties include allowance for down to 75% thimble; available. Accordingly, an inerc~ mental thimble deledon uncenainty component penalty from 75% to 50% of thimbles available could be considered to be appropriate, llowever, for conservatism and simplicity, the full thimble deletion uncenainty com,nonent penalty from 100% to 50% thimbles available will be used. The Thimble Deletion Uncenty Component (50% T.D.) discussed in the preceding section is combined with the appropriate flux map measurement uncenainty to obtain a total uncenainty, r.m UNCERTAINTY, FL The appropriate equation for combining statistically in& pendent uncertainty companents is F%(50%) = 1 + Faro sw + ((FBY - 1J' + (KSJ'ra}"' (Eq.2) u For conservatism, a negative value of T.D. Bias will be treated as zero. Analogous equations apply to 1 q ,

Evaluating the above expression yields the following result (rc)

For conservatism to support generic application to subsequent cycles and to support Unit I Cycle 19, Fgi (50%) will be rounded up to 1.06. This value can be interpreted as a 95% probability tolerance limit at a high confidence level. This two percent incremental thimble deletion penalty is linearly applied from ~5%

to 50% thimbles available (i.e.,1.04 at 27 thimbles and 1.06 at 18 thimbles available).

6

i. f TO UNCERTAINTY, rg (a,c) u For conservatism to support generic application to subsequent cycles and to support Unit 1 Cycle 19, Fn (50%) will be rounded to 1.08. This 3% incremental thimble deletion penalty is linearly applied from 75% ,

to 50% thimbles available (i.e.,1.05 at 27 thimbles and 1.08 at 18 ihimbles available).

AXIAL OFFSET AND QUADRANT TILT The mean change in quadrant tilt with 18 of the thimbles available was found to be only [

]" Similarly, the mean change in axial offset with 50% of the thimbles available was also quite small at [

' ]" Note that all uncertainti., on A.O. and tilt are absolute values and not percentages of A.O. nor tilt. These values indicate that thimble deletion has a negligible impact on the core average axial power shape measurement. Changes of this nagnitude are not significant and will not adversely aPect excore detector calibration.

7

d CONSERVATIVE ASSUMPTIONS For convenience a summary of conservative assumptions employed in this study are provided below: .

1) The total thimble deletion penalty from 100% to 50% of the available thimbles was utilized ratter than the incremental penalty from 75% to 50% of the available thimbles.

2) Thimble deletion uncertainty results were rounded up and negative bias values were set to zero.

3) [

]"

4) {

5) [

ju 8

SECTION 2 This section quantifies the :Enber of thimb;es per quadrant required for Prairie Island Unit I in order to improve the ability to distinguist . between random and systematic thimble deletion events and to establish the bounds of applicability of th. incremental peaking factor uncertainties. The peaking factor measurement uncenainty analysis described in Section 1 makes the assumption that thimbles were randomly deleted from the core. If thimbles are somehow systematically deleted from the core then the calculated peaking factor measurement uncertainties will not apply.

The assumption of random deletion of thimbles is an important one. If removal ofinstrumentation thimbles in the core is completely random then each thimble in the core has an equal probability of being removed from operation. Therefore, if 50 percent of the thimbles in the core were to be deleted randomly, a random pattern of thimbles would result. On the other hand, if there were some function driving the removal of the thimbles the result would not be a random pattem of thimbles. This systematic deletion of thimbles could conceivably result in large areas of the core being uninstrumented.

The current Technical Specification requirement of a minimum of 2 M/D thimbles per core quadrant is not sufficient to distinguish between random and systematic deletion events with high confidence. To help insure that thimble deletion is random, a restriction can be placed on the number of thimbles that must remain operable in each quadrant. By defining the quadnnt in such a nunner as to essentially place a f

requirement on each 1/8th core, the ability to distinguish between random and systematic events will be

, significantly enhanced.

If, for example, for 50% thimbles remaining, the requirement of 2 or more thimbles per quadrant is satisfied, then in all likelihood a random deletwr occurred and incremental thimble deletion peaking factor measurement uncertainties are appropriate. On the other hand, if there are less than two thimbles per quahant, then it is possible that a systematic thimble deletion occurred and that the impact on measured quadrant ceaking factors, may be larger than quantified in Section 1.

9

METIIODOLOGY-COMPUTER SIMULATION A shott computer program for determining the probability distribution of thimbles remaining was written.

'Ihe program allows for different number of thimbles per quadrant and keeps track of interior, axis, and diagonal thimbles (see 2-loop description). This program has been used to determine the number of thimbles per quadrant for all of Westinghouse Thimble Deletion Analyses.

Starting with ni thimbles in the core and randomly deleting down to rr thimbles constitutes one case. After deleting nr - r, thimbles from the core, the number of thimbles remaining in each of the eight quadrants is determined. The minimum number of thimbles remaining over all 8 quadrants is then found. A large number of cases is run ir order to determine the probability distribution of thimbles remaining. .

• 2 LOOP PROBLEM DESCRIPTION The maximum possible number of available thimbles for a 2-loop Westinghouse PWR is 36. The initial distribution of these thimbles is p.evided in the following table. Figures 3 and 4 should also help in visualization.

No. ofInterior Thimbles in Qi 6 No. ofInterior Thimbles in Q2 7 No ofInterior Thimbles in Q3 7 No. ofInterior Thimbles in Q4 6 No. of Axis Thimbles Ql-Q2 1 No. of Axis Thimbles Q2-Q3 3 No. of Axis Thimbles Q3-Q4 3 No. of Axis Thimbles Q4-Q1 .3 36 Total No. ofInterior Thimbles in QA 7 5

No. ofInterior Thimbles in QB 8 No. ofInterior Thimbles in QC 7 No. ofInterior Thimbles in QD 8 g- No. of Diagonal Thimbles QA-QB 2 No. of Diagonal Thimbles QB-QC 2 No. of DiagonalThimbles QB-QD 1 No. of Diagonal Thimbles QD-QA _1 36 Total Note that all thimbles are counted as whole values even if they lie on an axis or diagonal. Provided the technical specificatian value and computer simulation are consistent this is appropriate. Eighteen (18) thimbles are randomly deleted from.each case.

10 1

l

2 LOOP PROBLEM RESULTS A 1000 case simulation was run to obtain the probability distribution of the minimum number of thimbles left after having reduced to 50% and 60% of the thimbles available. Results are summarized in Table 3.

[

}" Therefore, a requirement that 2 or more thimbles per quadrant for 50% be available is appropriate. Assuming rmdom thimble deletion, it is unlikely that with 18 thimbles remaining overall, fewer than 2 thimbles will be available over the 8 quadrants.

CONCLUSION With the inclusion of the additional peaking factor uncertainties, it is concluded that operation of the movable detector system with a minimum of 50% of the thimbles available is acceptable provided that an additional 2.0% for Fy and 3.0%_for Fq be applied to the INCORE measured peaking factors. Ho"*ver, when fewer than 75% of the thimbles are available there should be a minimum of 2 thimbles per quadrant where quadrant includes both horizonte1 vertical quadrants and diagonally bounded quadrants. This requirement increases the ability to distinguish between random and systematic thimble deletion events. In addition, the confidence on the appropriateness of the incremental thimble deletion peaking factor uncertainty values is increased provided that 2 or more thimbles per quadrant are observed to be available, and counting thimbles on the axis and diagonal as whole values.

I1

TABLE 1 INCORE DETECTOR TilIMBLE REDUCTION STUDY MAPS Bumup Core Power Percent (MWD /MTU)  % Thimble Available (Ref.)

linit 1 Cyc 15 MAP 6 210 100.0 94.4 MAP 8 2l00 100.0 94.4 MAP 13 7380 100.0 94.4 MAP 21 11590 100.0 94.4 MAP 25 15750 100.0 94.4

\

Unit 2 Cyc 15 MAP 6 304 100.0 80.6 MAP 8 2489 100.0 83.3 MAP 19 8545 100.0 77.8 MAP 24 13643 100.0 80.6 MAP 28 17734 100.0 80.6 F i

+ -

a

^

- 4..

-- r TABLE 2a SAMPLE STANDdRD DEVIATION AND MEAN FOR INCORE MAPS ' -

WITH 50% OF THE THIMBLE AVAILABLE FOR TWO LOOP REACTOR CORE PARAMETERS {

F.n Fo , ,

Unit . - Cycle MAP X'(%) S,(%) X' (%): S,(%) - ,

1 15 6 (a,c) l- 15 8 ,

i 15 13-

-1 15 21 1- 15 25 2 15 6 '

2 15- 8-2 15 19

2. 15 24 2 15 28 T :.

r >

l' (M

[N e

E i

. ' ' A _;

13.

,n_._ p ,+,,a . m q n.~.. 3 * -

,m , w , _ . . _ .4,-- p ,ge y .y _, y y.-  % ..

. . . . . . . - . - - . . - - . . ~ - - .. . . ..

, s. ,

i

. TABLE 2b .

i SAMPLE STANDARD DEVIATION AND MEAN FOR INCORE MAPS L

- WITH 75% OF THE THIMBLE AVAILABLE FOR TWO LOOP REACTOR CORE PARAMETERS i

F,n Fo .

Unit - Cycle MAP X' (Y.) S.' (Y.) -' XI (Y.) S, (Y.) 7 1 -15 6 _a,c)

(

1 15 8 1 15 13 1 15 21 1 15- 25 2- 15 6 ,

2 15 8 2 15 19 2 15 24 2- 15. 28-Sa I M g:

a 4

4 I . .

14

.. . a -

_. - . - . , . . . - . . -. .2.

j- ,

,-r.

^

TABLE 2c SAMPLE STANDARD DEVIATION AND MEAN FOR INCORE MAPS -

WITH 50% THIMBLES AVAILABLE FOR TWO LOOP.-

REACTOR CORE PARAMETERS A.0/ QUAD TILT +

Unit Cycle MAP Xi(%)-_ Si (%) - Xi(%) Si (%) -

1 15 6 (a,c) l- 15- 18 1 15 -13 1- 15 21 1 15 25 2 15 6-2 15 8

'2 15 19 2 15 24 2 15 28 Sc.

$ comb

+ Standard deviation for QUAD TILT about Atilt = (Ref. - Deleted) x 100%.

• Standard deviation for A.O. abaut AA.O = (Ref. - Deleted).

.. j 15

^i

.d3LE 2d SAMPLE STANDARD DEVIATION AND MEAN FOR INCORE MAPS WITH 75% OF THE THIMBLE AVAILABLE FOR TWO LOOP REACTOR CORE PARAMETERS -

A.O.* - QUAD TILT +

- Unit Cycle - MAP X'(%) Si (%) X'(%)- S (%)

1 15 6- (a,c) 1 -15 1- _15 13

'l 15- 21 1 -15 25 2 LI5- 6 2 15' 8 2 15- 19-2 15 24

2. 15- 28 Sc s

__Xca

+ - Staadard deviation for QUAD TILT about Atilt = (Ref. - Deleted) x 100%.

Standard deviation for A.O. about AA.O. = (Ref. Deleted).

4 16

TABLE 3.

2-LOOP CORE

SUMMARY

1000 CASE TIIIMBLE DELETION SIMULATION 50% and 60% TillMBLES AVAILABLE 1000 cases at 50% 1000 cases at 60%

Thimbles remaining # ofcase % cases Cumulative % # ofcases % cases Cumulative %

(a,c) 17

c r

FIGURE I MOVEABLE DETECTOR, TIIERMOCOUPLE AND FLOW MIXING DEVICE LOCATIONS (a,c) 18

A _.Ad_ AJ.-

4 e

FIGURE 2 PRAIRIE ISLAND UhTT 1 CYCLE 19 PEAKING FACTOR UNCERTAINTIES VERSUS TlilMBLES AVAILABLE (a,c) r 19

FIGURE 3 MOVEABLE DETECTOR, THERMOCOUPLE AND FLOW MIXING DEVICE LOCATIONS (a,c) i e

P 20

)

4 f

FIGURE 4 MOVEABLE DETECTOR, THERMOCOUPLE AND FLOW MIXING DEVICE LOCATIONS (a,c) e b

C 21

APPENDIX A THIMBLE DELETION UNCERTAINTY COMPONENTS 95% PROBABILITY AND 95% CONFIDENCE ( ~6 + KS% )-

~ NORMAL (TYPICAL) FLUX MAPS

- F,n Thimble Deletion Uncertainty Component (a,c) =

Fo Thimble Deletion Uncenainty Component (a,c) e

( e-22

APPENDIX B TWO-SIDED 95% CONFIDENCE LIMITS ON MEAN ATILT AND MEAN AA.O.

t a2s se./ (approximate t by z) tilt or tilt or A.O. A.O.

Quadrant Tilt Uncertainty Component (a,c)

Axial Offset Uncertainty Component (a,c) 23

l

. a t

Exhibit F Prairie Island Nuclear Genwating Plant License Amendment Request Dated January 15,1998 Westinghouse Autharization Letter,

- CAW-981196, and Accompanying Affidavit Proprietary Information Notice Copyright Notice 4

6

_ _ _ _ _ . _ _ _ . _ _ _ _ . _ _ _ _ . _ _ _ _ _ _