ML20004E997

Nonproprietary Version, Responses to Second Questions on Statistical Combination of Uncertainties,Repts CEN-139(A)-NP & CEN-124(B)-NP-Part 2.

ML20004E997
Person / Time
Site:	Calvert Cliffs
Issue date:	04/30/1981
From:	ABB COMBUSTION ENGINEERING NUCLEAR FUEL (FORMERLY
To:
Shared Package
ML19276K116	List: ML20004E996 ML20004E997 ML20004E999
References
NUDOCS 8106160331
Download: ML20004E997 (21)

Text

._ _ _ _ _ _ - _ . . _ _ _ _ _ . _ _ _ . . . _ . . _. _ _ _ _

[ . . ._g1. . . . . . - _ . . _ .

t _ .. - .

)

O  !

i 4

L i

r ,

j ,

i i

1

i. . .

{ Responses to Second Round Questions i on the Statistical Combination of Uncertainties -

1

Reports CEN-139(A)-NP and CEN-124(B)-NP-1 I.

Part 2 i l-l April, 1981  :

i I-4 I

r j COMBUSTION ENGIllEERING, INC.

1

j. I L

t i

4 I

T

l. -

e 4

1 l

}-

c i

h.

t

}

j:

-rs10616O N e- twf,.my e-- wu-9 %ww ww- , mew e wy asuvz s e wy-y -d ew w es-aw e -* wye m w'M9powwWe. ,ym*mq w = __Oe='+

lL LEGAL I;3TICE ThisL report was prepared as an account of work sponsored by Combustion Engineering, Inc. Neither Combustion Engineering nor any person acting on its behalf:

A. Makes any warranty or representation, express or implied including the warranties of fitness for a particular purpose or merchantability, with respect to the accuracy, completeness, or usefullness of the inforamation contained in this report, or that the use of any information, apparatus, method, or process disclosed in this report may not infringe privately owned rights;-or.

, B. Assumes any liabilities with respect to the use of, or for damages resulting from the use of, any information, apparatus, method or process disclosed in this report.

4 W

, -,-e - , , , , - - - , ,.-r- ,w., ,-y,r,-, - ,.,.~ , , - - - -

w - - -- ---

' 2 -' j.N Abstract

-This report consists of responses to the second round of questions asked by Battelle Pacific Northwest Laboratories as part of their review of

• the statistical combination of uncertainties reports for Arkansas Nuclear One Unit 2 and Calvert Cliffs Units 1 and 2. It specifically addresses ten questions on the dccuments CEN-139(A)-P and CEN-124(B)-P Part 2.

e-l i

l .

5 l

i l

I t

I 1-l

i. .

1 1 1

! I Table of Contents

k.

• Legal Notice i

' Abstract ii

. Table of Contents iii Response to Round Two Questions on the SCU Reports 1 CEN-139(A)-P, CEN-124(B)-P Part 2 e --

J t

i i

i i

l a

I

(. \

l

[

P iii

,-.--,,~ww.w.e,,,,-,-,-*,-n,...,n-n,, -new ,,,n,--a-e.,v-,._,,-,w-,n,,v,n-- ,w.m we,,-- mm ,n,-- ,-r-- rw m

4 L Responses to Round Two Questions on the SCU Reports CEN-139(A)-P,CEN-124(B)-PPart2 1

A ),

Question 1: Is there a typographical error in the fifth paragraph of Section 3.0? The statement " system parameters define the operational state of' the reactor..." must be incorrect.

Response: The sentence should read, "As explained in Section 2.1, system parameters describe the physical environment that the working fluid encounters".

'm S

2 f

-. ,, ,-_._.._,,_....-..._...-_..,__,r ,,.,.m._ - - . - . , . , , , , _ , , - . , . _ - . - - _ - . - . _ _ - - . . . . ~ . - _ _ . , . .

& In Question 2: a. Question 3 refers to pp. 3-3 to 3-5 and the discussion of how the most adverse state parameters are derived from the information in Tables 3-3, 3-10, and 3-7. The response to the questions helps to clarify the procedures used, but it does not quite address the central problem that caused the question to be raised in the first place. We would, therefore, like to rephrase the question as follows:

On p. 3-5, in Section 3.1.5, the report states "The magnitude and impact of the [

j those of the other system parameters. Therefore the ASI and Tin which tend to maxin.ize MDNBR sensitivity to [_ ]

are used to generate the response surface." The discussion in Section 3.2 throuah 3.8 does not explain how it was determined that the [ ]

was the most important of the system parameters affecting MDNBR. Please clarify.

Response: Engineering judgement indicates that the MDNBR is most sensitive to[ ]inthehot assembly. This is further justified by the sensitivity studies discussed in the response to Question 2b.

l 9

9 3

._ . ~ - _ _ . -.

4 d.

Question 2: b. From Table 3-4 on p. 3-22 of CEN-124(B)-P, Part 2, it appears that the enthalpy rise factor effects on MDNBR are very sensitive to the axial shape index, with the largest percent. change occurring at an ASI value ofl ,], a t nominal operating conditions. But this ASI value is not used in Table 3-5, which determines the operating conditions (pressure, inlet temperature, and percent design flow rate) at

- which MDNBR is most sensitive to enthalpy rise factor effects.

Please explain the logic behind this.

Response: BecauseMDNBRis[ ]to perzurbations in the inlet flow factor than to perturbations in the enthalpy risefactor,[ ] weight is applied to sensitivity studies in which the , inlet flow is perturbed than those in which the enthalpy- rise factor is -perturbed. This difference in sensitivities is indicated by a comparison of the response surface coefficients for the hot assen.bly inlet flow factor ( ] vs that for the enthalpy rise factor

[ ], Assuming the response rurface is characterized by the coefficients listed in Table 4-2, and by the probability distribution functions in Table 5-1, a perturbation of the hot assembly inlet flow factor by lo[ ]resultsina

[ lchangeinMDNBR,whilealo[ ] perturbation of the enthalpy rise factor results in a[ ] change in MDNBR. In Table 3-4 the greatest sensitivity of MDNBR to enthalpy rise factoroccursata-0.070ASIwhere%MDNBRchange/ah=[ ]

Adjusting this point for a la change in enthalpy rise factor resultsina[ ] change in MDNBR, which is [

] the[ ] change in MDNBR for hot assembly flow factor pertur-bations. From Table 3-3 MDNBR [

, .] Because of these sensitive the{

]were used in detemining the sensitivity of enthalpy rise factor to operating conditions (Table 3-5).

4

4 6 Question 2 c: In Section 3.1.1, page 3-3 of the ANO report is stated that MDNBR is a smoothly varying function of the state parameters. However, the data presented in Tables 3-2, page 3-16, do not support that conclusion. This can be seen in the attached plot of the data in Table 3-2. MDNBR for both the advantageous anc adverse perturbations changes very rapidly as ASI c1anges from[( ]

- The behavior of MDNBR in this region makes the selection of f ] as most adverse questionable. The dashed lines on the figure represent hypothetical values of MDNBR evaluated]at Thesea hypothetical point intermediate betweenwith values are consistent antheASI of[~

rest of the data, yet they lead to a percent change of 35%

instead of 22%.

h An explanation together of what axial with a rationale as to why flux shapes one withwere used an ASI of[ and v! y,j]

has such a marked change in MDNBR, would be helpful. .

Response: Because the value of the A.S.I. is subject to a numoer of factors (eg. the axial peaking factor, % of rod power below the centerline of the active core and % of rod power above the { of the active core) it can not be stated that the MDNBR is smoothly varying with the A.S.I. In response to the question re; the most sensitive A.S.I., several Detailed TORC (D-TORC) analyses wereperformedfortheregionaboutthe[ [(seetable below).

~~

From the table abcVe it can be seen that the[ ]'

remains the most sensitive A.S.I.

5

< o m

Question 2_d: -The selection of the most sensitive pressure and temperature in Section 3.1.3 refers to Table 3-3 for support. However, half of the data (the high temperature data) in Table 3-3 was not used because quality limits were exceeded. The effect of this is that Table 3-3 contains no information on the variation of MDNBR with temperature. If the high temperature was too extreme, an intermediate point should have been examined.

Response: Sensitivity studies performed at various inlet temperatures

, confirm the choice of[ ]asthetemperaturewhereMDNBR is most sensitive to perturbation in the system parameters.

The results of these studies are presented in the table below:

A 1

~ ~

The choice of the [ linlet temperature is also confirmed by the sensitivity studies presented in Ref.1.

Reference:

1. SCU, Part II, CEN-124(B)-P, Statistical Combination of Uncertainties.

O l

6

I I D-Question 3: Section 3-2, Radial Power Distribution, needs support and justi-fication. If this were handled in the same manner as outlined in CEN-124(B)-P, Part 2, and in the answers submitted to questions concernitig this point, it should be adequate.

Response: The response to Question 4 of the first round of question on CEN-124(B)-P is applicable to ANO-2 Cycle 2. The sole exception is that in place of the S-TORC model, CETOP-D is used as the thermal margin design model.

7

f s Question 4: Section 3-3, Inlet Flow Distribution, states that "a large part of the u'1 certainty in the flow splits results from measurement

~

uncertainty. This measurement uncertair.ty is considered random ond may be characterized by a normal probability distribution function (p.d.f.)". What is the p.d.f. (mean and standard deviation) and what is its source? Is it possible to quantify the assumption that the distribution is normal?

Response: The means and standard de.-iations for the relevant inlet flow splits are given in Table 5-1 o f Reference 1. For each fuel assembly location the flow split is derived from a set of [ ] independent measurements of relative flow obtained from model tests. Because ofthe[ ]it is generally not possible to establish rigorously that the measurement error p.d.f. for any given fuel assembly location is normal. However, qualitative tests for normality are satisfied. For example, figure 1 gives the cumulative probability distribution for the [ ] data points obtained from the most limiting assembly location (channel 16 in

. Ref. 1). The data are seen to be somewhat more centrally grouped I than expected for a normal distribution. Thus the assumption of normality is arguably conservative.

Also, by application of the Kolmogorov-Smirnov test and the X2 test [ ]

the hyrothesis for normality of the population from which data set shown was drawn cannot be rejected at approximately the 30%

significance level.

References:

1. CEN-139(A)-P, " Statistical Combination of Uncertainties",

, November,1980.

e i

l 8

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

f e 99.97

_ . _ l ._ ,

1 .

-*.M.W.-. h. hw. ,p, . e .e. $ g .9 .r -e-.

<.m . o

.L.... _.p..

.u_.3

.p; , p . .. . . . _ . . . . _

. 1_ -

993 -

_p.L ... - ._ ._ . . _. p _ .

,g

.-n .. .. .. .-y . - . - . -

--.7

--e.. 4--e.. .e-Q h... _m h ,. e.9-

• e.-.iy... t>=..4 = .e. g+wie w e, mm g.w.

C * ^ - '

T*** '.~M.'Z.~~~'

. 99. =.T*.'**~ ._ . . .

i. ... .-

..w_..

. 4 . ._. .

_.,e.g.e-.-e.ew

_~}-

.h-,_4-e a 3 6 i

ti . '

+ _ ._L. , _ . . . . . . _ _

i

• di . i; ,

95 . ._

- a

.d._- ,

. . [ . .

'.l -.

. + d ._.%

p, _,w_..--.-..e - (A_

..4_..-.. _ _. .

9.__.,__.-.

n...Oh.n..

@**- ._ M^ '+ 9 wi (ee 4i-.=i

. - - - . 4 e - - +

25^.:. . .

~~&5 --h- I

~~ ~~

80

.- ..g- . e.

._6+-y.i-

. .1 9 _.44. .. _ .. h. . L ++

. c... _ _ -. _- 4  :. l_._..

70 =____.4,_.. _

, -+

3,_._.

e .g+.i e.w i.e - 4_ e-

_;4__ew O .a._ .

Q..-.=--

.-4 - .%-,-- . . _ . -

._a.--- .b. .. -H_...

/ -1 . . .. .

W.

t _ L___.

.L_. ..

u

30. -

0 W T- t - -~ -----

\D ,

t -+ 4-7 , . * . - -. . .

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

H__ . _ - _ . .

% ~O~l19Ei _ i 1 31 ri--d__-~_~ .

p) -,. ._ __ . .

g __

_u.M.

y4.

9 ..yMone.--

-e.

9 -

_ - . gg., . _ .

+_

4.

• ~fa.

n g ~  ;

,7 h..

, N_._h m -.

• ~*

U TNNW e : i

~ - '

I~"^~-*-

._ p _ / -

..uk  :- ! CON)utg77t/s Dt5rg/Bor/W Fug

!)( ,.

g _ ,; .I .

. _. _ . _ . , '. ,' ,i ' l_

.l (2CW $}%)y~ O//pq /N LOC /7770hl ?&

~r 2*

~. u ... ..

w.== <

~2 11 Crarsr nssonssy' 21G.)

O 5 -

_. J. -.-.__ ,,M.-. _.

.- 4,, , -m El# g

. /% 9

.~. . _.-  ; R

.'.- !il

.z ' '

i l1' i \ NORMGL, O/5T2/6u?70N'

.m .. -p m q4

_ ,y _ .- -

~

f f i t -f

-3 -

2 -

7 0 1 2 3 4 2 s (x N)/0" 9

f .

4 Question 5: Paragraph 3-4 of the Afl0 report states that the exit pressure I distribution has little or no affect on liDflBR. Justification, at least a reference, should be provided.

Response: Reference 'is raade to Ceil-124(B) - P Part 2 Paragraph 3.4 where a discussion of the sensitivity of pressure distribution is presented This section concludes that, " Detailed TORC analyses performed with both bottom peaked and top peaked axial power profiles demon-

) strate that fiDNBR is extremely insensitive to variations in the 4 .

exit pressure distribution".

I e

i I

a .

5 J

t .

l i

I t

l 10 i

4

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

f .

Question 6: Paragraphs 3-5 and 3-6 point out that as-built data and tolerance deviations were used to evaluate the enthalpy rise factor. Has the ampling plan for measuring the as-built assemblies been approved by the Nuclear Regulatory Commission? Is this required? How are the variations in the deviations combined into the standard deviation of the rise factor? These combination techniques are standard, but they should be identified.

Response: Derivation of the enthalpy rise factor is done in accordance with sttndard C-E ouality assurance pro cedures as reviewed and approved by NRC in Reference 2. Input involves 100% recording of the fuel pellet lots used in each rod of each assembly.

Within pellet lots, NRC approved sampling procedures are used to determine the U235 1 ading. The enthalpy rise factor in every channel is determined by taking the ratio of the as-built value of U 235 1 ading in that channel to the nominal value. The enthalpy rise factor is calculated for each subchannel in all the a .mblies of a fuel batch. The factors are then collected in a histogram for each specific type of channel and, for SCU, the histogram for the channel type in which MDNBR occurs is inserted into the SIGMA Monte Carlo Code as described in Reference 1. Although values for

'mean and standard deviation are given for information in Table 5-1 of Reference 1 it is important to note that the relevant histogram itself is used in SILMA.

References:

1. CEN-139(A)-P, " Statistical Combination of Uncertainties",

Nevember, 1980.

l

l. 2. CENPD-210-A Rev. 3. " Quality Assurance Program",

. November, 1977.

l l

l

. 11 t

Question 7: ilow is the systematic pitch reduction given in paragraph 3-8 ,

used. Presumably it must enter into the enthalpy rise '

factor and into the equivalent diameter in the DNB correlation.

Is this true?

I

{ '

• Response: Perturbations in the systematic pitch enter into the calculation of channel flow areas and crossflows. These perturbations are also taken into account in the calculation of the matrix channel hydraulic diameter which is then used in the CHF correlation.

Enthalpy rise factor does not take into account reductions in the systematic pitch. As stated in the response to Question 6 the enthalpy rise factor accounts for variations in the U235 j

loading of the fuel pellets.

i l

~

I I

l-i l

i t

12

-T .

Question 8: Why is there such a large difference between fuel rod bow penalties given in paragraph 3-9 for Arkansas Nuclear One and Calvert Cliffs? Has the Combustion Engineering topical report outlining the methodology for calculating rod bow penalty been approved by the Nuclear Regulatcry Commission?

Response

The fuel rod bow penalty of 0.6% in MDNBR for Calvert Cliffs was calculated for C-E's 14x14 fuel having a batch average burnup of 45,000 MkiD/MTV. The fuel rod bow penalty of 2.0% in MDNBR for ANO-2 Cycle 2 was calculated for C-E's 16x16 fuel having a batch average burnup of 30,000!4fD/MTV.

In both cases the penalty was derived from supplement 3 of the C-E topical report on rod bow. Although this report has not yet been approved by the NRC, it applies methods recommended by the NRC in June 1978 (Ref. 1 ).

References:

1. Letter, D. B. Vassallo (NRC) to A.E. Scherer (C-E),

June 12, 1978 and revision 15 Sept.1978.

i I

i a i

9 t

k 15 i

~T -.

Question 9: Paragrap;. 3-11 states that since TORC code is used to calculate local coolant conditions to develop the CE-1 CHF correlation,"any calculational uncertainty in the TORC code is implicitly included in the MDilBR limit that is used with the TORC /CE-1 package." This question during the initial review and the responding answer only partly allays our concerns. A more detailed expression of our question is attached to this letter for this discussion.

3

, Response: The PNL concerns result from the hypothesis Sat the TORC code

• may systematically overpredict hot channel enthalpy in a test section by a larger amount than the amount by which it may overpredict hot channel enthalpy in a reactor core. This hypothesis depends on two premises:

(1) Unheated test section shrouds may significantly reduce local enthalpy in the CHF channels of test fuel assemblies, (2) The cold shroud effect maybe systematically underpredicted (i.e., hot channel enthalpy may be systematically overpredicted) by TORC when used to calculate local conditions corresponding to CHF test points.

Discussion in reference 1, and heated shroud test results described there, indicate that both of tne above premises are probably false. Section 7.1.3.2 of reference 1 shows that TORC /CE-1 correctly predicted the results of heated shroud test's, covering a wide range of the operating conditions for which CE-1 (obtained with cold shrouds) is valid. In addition, -

the response to question 2 on CENPD-162 (Appendix F of reference 1) shows that the heated test shroud (a) :.ubstantially flattened the channel-to-channel enthalpy distribution in the test fuel assembly, while (b) producing little change in the enthalpy of the channels in which DNB was generally obtained.

The fact that TORC /CE-1 ccrrectly predicted heated shroud test data argues strongly against the second PNL premise.

The fact that the ratio of measured to predicted CHF showed a slight upward bias for the heated shroud tests (a mean of 1.032 compared to 0.999 for the CE-1 data base test points),

indicates that any code bias present is likely to be small.

If anything, TORC is more likely to overpredict hot channel enthalpy in a heated shroud case (analogous to in-reactor

! 14

T ,

conditions) than in a cold shroud case. This results in a conservative code error, exactly opposite to that 4 which would be expected if the second PNL premise were true.

1 The fact that little change in the hot channel enthalpy a

resulted from the addition of a heated shroud to the test section indicates that the first PNL premise can be true only in a limited sense. Clearly the presence or absence of a cold test section shroud affects the enthalpy distribution in the test section as a whole. However, in the region in which DNS is likely, the effect is small, and when it does occur it is adequately represented by TORC. Further, in the absence of

- errors caused by inherently different conditions in a test section,as opposed to those in reactor core, it is clear that an argument parallel to that given by PNL leads to the conclusion that TORC code uncertainties are indeed accommodated automatically by use of the san.e code for derivation of the CHF correlation and for design DNBR computations.

It shoud also be pointed out that, although the conclusion regarding accommodation of code uncertainties was restated in CEN-139(A)-P and CEN-124(B)-P, that conclusion simply reflects a position implicit in *eference 1, which has been reviewed and approved by NRC.

i

Reference:

1. CENPD-162-P-A, "C-E Critical Heat Flux", September 1976.

t 15 t

._ ~ , m ,.-.- -,.- . . _., , . . , - . _ . - , ,, - , . , = . = . , , - . . . , . , ~ . - y

eD

• w Me wNe.%.*

+4s2> ...eE Ev.<u.rie~

TEST TARGET (MT-3) i e

1.0 5 L'A 224 yll EE <

l.l 4.. Eu 'S lE ,

1.25 !I l.4 1.6 ll _

1

/ 4 6" >

l 4

1. 4 f_ k++ gf%;

V _. --.-.) i

- ~ ~ - ~ ,

+44 meE v.<e 1.

TEST YARGET (MT-3)

I j' l.0 EIn E4 li -

-yllllE E

11 l-l \ p .: ,= M e -

!25 jg a __

• r ++sp

• r  %

f , '4*,) >

$h

++ _e. ...<e TEST TARGET (MT-3) 1,em .'*>

h t

- l.0 5 82 814 4 ll"ILE 1usi?

1 /

~ m I.8

=

' G l.25 ' I.4 i 1.6

. _ I

/ 6" =

l

. p% + sp k>,j!* )z d'g,,$lg g,,7 _

l i.

=

)"

Question 10: In Section 5-2, page 5-2, it is claimed that Figures 5-1 indicates that the MDNBR p.d.f. approximates a normal distribution. Thereafter the document proceeds as if the Figures 5-1 MDNBR p.d.f. were in fact a normal distribution.

presents a qualitative comparison and does not provide sufficient grounds for accepting the hypothesis of a normal distribution.

There are quantitative tests of2that hypothesis (e.g., a Kolmorgorov-Smirnov test or a x test) that could be used.

- Response: ThiMDNBRp.d.f.contains[ ] data points as computed by the Monte Carlo code. Using a x2 test based on 32 intervals leads to:

a 2

32 (n -np3)2

=I = 24.820 x calc j=1 np.

• J where: n. = the number of calculated data points in the jth J

interval l

np. = the number of normal distribution points in the

' J 32 jth interval for n=I n. = [

j=1 J From T3ble A-5 of Reference 1, X

• f=29 l a=.70 Thus, the hypothesis that the calculated MDNBR p.d.f. is i normal cannot be rejected at nearly the 70% significance level.

f

Reference:

1. Rdliability Technology, A.E. Green and A. J.

Bourne, Wiley-Interscience, London,1972.

i*

• 1 i

16 i'

-,,,-.n,,,,- -,,,-e-- .,....,-..,-,,.,--,----,-r...-,, -,~,.-,----~.---,-_.n.- -~~-,,-..-.._.,.,..-,r.. , , , - .