ML20239A550

Technical Evaluation of Bg&E Topical Rept:Retran Computer Code Reactor Transient Analysis Model Qualification

ML20239A550
Person / Time
Site:	Calvert Cliffs
Issue date:	09/09/1987
From:	INTERNATIONAL TECHNICAL SERVICES, INC.
To:
Shared Package
ML20239A543	List: ML20239A542 ML20239A546 ML20239A550
References
NUDOCS 8709170432
Download: ML20239A550 (16)

Text

-6 f

7bchnical Evalt:ation of B3&E 7bpical Report:

REDtAN Ctmputer Cbde Reactor Transient Arnlysis Model Qualification m l

< ~: ~ . .- nwr.- _j

,e; ., . . . ; - , -

i I

International Tectinical Services, Inc. I 420 IrrJ.ngton kienue New York, New York 10170 8709170432 070909 PDR ADOCK 05000317 P PDR

t TABIE OF CDtfrDTIS Page 1.0 Su:: nary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 2 3.0 Topical Objectives. . . . . . . . . . . . . . . . . . . . . . . 2 4.0 Cmputer }bdel ing . . . . . . . . . . . . . . . . . . . . . . . 2 4.1 RETRAN02/}OD03 . ..................... 2 4.2 Nodalizations. . . . . . . . . . . . . . . . . . . . . . . 3 4.3 }bdel s . . . . . . . . . . . . . . . . . . . . . . . . . . 4 5.0 Plant 'Ibst Caraparison . . . . . . . . . . . . . . . . . . . . . 5 5.1 }ilitiple Secondary Side }hlfunction Event. . . . . . . . . 5 5.2 Four Pump Coastdown fzun 20% Power . . . . . . . . . . . . 6 5.3 Reactor Cbolant Operation Punp Canbination Flw Tests at Hot Zero Power. . . . . . . . . . . . . 7 5.4 One Pu::p Coastdown fran 80% Power. . . . . . . . . . . . . 7 5.5 Total loss of FIN / Natural Ciru11ation Test frun 4 0% Power. . . . . . . . . . . . . . . . . . . . 8 6.0 'IRAC Analysis Cccparison. . . . . . . . . . . . . . . . . . . . 9 6.1 Cboldown to RHR Entry Usirg AINs and APS. . . . . . . .. 9 6.2 Runaway Main Feedwater to One Steam Generator . . . . . . 10 7.0 IDFT Test Omparison. . . . . . . . . . . . . . . . . . . . . . 10 7.1 16-1 'Ibst . . . . . . . . . . . . . . . . . . . . . . . . . 11 7.2 I4 -3 'Ibst . . . . . . . . . . . . . . . . . . . . . . . . . 11 B.0 Conclusions and Pa_- - . edations . . . . . . . . . . . . . . . . 13 9.0 References. . . . . . . . . . . . . . . . . . . . . . . . . . . 14 i

> i

i. j

~

e i 1.0 Sumarv  ;

i i /,

)!

'Ihe Baltimore Gas and Electric Company (EG&E) sutraitted a topical .

report (1) for the purpose of'documer:tation of the B3&E best-estimate REIRAN- ,

i thermal-hydraulic analysis capability. . BG&E. presented acrparison of five plant transients; these are (1) Multiple Secondary ' Side Malfunction Event, (2) Four Pump Coastdown ~frm 20% Power, (3) Faactor Coolant LOpeE5ticn Pump

~

Combination Flow Tests at Hot Zero Power, (4) One Pump Coastd? M n frm 80%' ,. '! ,

O '

Power and (5) 7btal Ioss of Flow / Natural Circulation 7%st f23m 40% Ptwer) .' .

BG&E also presented two TRAC analyses; (6) Cooldown to RHR Ent:y Using AINs ,

t >

and (7) APS and Runaway minJFeedwater to One Steam Generator. Two LOFT 4y I i,

tests si:milated for ccxqparisoa with REmW calculations are I4-1 and.I4-3. ,

f-

/

[e ,

{

j 4 l It is our general conclusion . hat the applicant has presented thorough -

analyses, considering sensitivities to nodalizations, models and . physical parameters. We feel that they have exhibited ta good un$ere@ 'of the l REIRAN code, and the ener in which its ewlications to simulata plant 'j i e i l transients is conducted, and demonstrated ~ that the . staff ., dambers who 4 1 ., a i performed these analyses h$e adequate ski.lls to' analyze the Idolts of such i 4

applications. [Although tNir use of the contru1Isyst4ms was not examined in )

./ ,; r. . .,

detail ard none'of the transients analyzai teWad tJmir ability to use the l 1

control systems ibily, their skill and understanding of the code i demonstrated in the diriassion of ' the results in the report , gives us '

)

t <

reasonable assurances that the BG&E staff also poce(us the ability to model j

- , i the control systems with t M code. i

, i I i . . 4

1. I # ,' '!

j f 3 ,, ,y ,

[ls' j b

,7 1 ,

e .<

V *

, g_ ,e t ,# i I

p j. \

, i' l

a I, '

'1 -

s

,.?

/ t

> p e i

^

+ { d ,

, a ' _  !

E, 1 2.0 Introduction

'Ihe Baltimore Gas nd- Electric Ccrpany (B3&E) subnitted a topical report [1,2,3) for t' purpose of documentation of the BG&E best-estimate 1 RETRAN thermal-hydraulic' analysis capability. B3&E pmsente:1 ccmparison of five plant transients, two TRAC analyses and two IDFT tests with RETRAN i calculations. We have reviewed the applicant's efforts to qualify RETRAN models for use in the analysis of the Calvert Cliffs plant. Our evaluation is summarized in this mport.

l

)

3.0 Toolcal Renort Obiectives It was B3&E's stated intent to de:mnstrate (i) the capability of B3&E analysts to properly develop RETRAN models of the Calvert Cliffs plant, a Ctrobustion Engineering ' design, (ii) to perfom calculations with these models to simulate realistic plant transient response, and (iii) to campare I l

the results to measured plant data, another best estinate camputer code

('IRAC), and experimental data (IDFT) . In addition, model enhancement and l appropriate sensitivity studies were performed by B3.E to provide greater l- insight and a better understanding of nodeling Calvert Cliffs with REIRAN. l l

One of B3&E's objectives in denting their analyses and results in j this topical report was to present their effort in performing a code verification in accordance with NRC Generic I.etter No.83-ll (4) i l

l 4.0 Canouter Modelirn  !

4.1 REIRAN02/)OD03 i

, 'Ihe applicant used REIRAN02/)OD03 to perform analyses presented 2

l

e u.

W . 1 l

. 1

~

in tacir topical report. This version, which has not been approved, contains corrections of code errors discovered during the review of RETPRJ02/ LOD 02 [5] which were implemented at the request of the Imc.

However, in addition to error correction, some model modifications were also mde to this version (6). The imC staff has not yet reviewed the code to I 1

assure that these model revisions were appropriate, accurate and properly ]

irpler ented. However, the use of this version by EE is acceptable in the )

context of this topical rc~'t.

4.2 Nodalization ME developed three nodalizations for Calvert Cliffs: (1) a detailed one-loop model for transients with nearly symmetric loop corditions; (2) a split-core two-loop model (two cold legs were lumped j together) for transients which result in asymmetric plant conditions; and {

(3) a "four-loop" model which in fact has only two loops, but which rodels all four cold legs discretely to study flow patterns durirg flow tests or j transients which can be impacted by one or more RC pumps on a coastdown. )

The applicant has selectively and appropriately used one of these l 1

ncdalization schemes for each transient deperriirg upon the expected plant 1 1

behavior and has, in addition, provided acceptable justifications for such '

selection, in some cases based upon sensitivity analysis.

1 l The one-loop model was used in a four pump coastdown and a j ccrparative study with TRAC for cooldown to RHR entry usirq ADVs and APS. j An obvious advantage of this ncdalization is the camputer code speed obtained by a simplified model, while an obvious disadvantage is lack of detz il. ME de.nurstrated an awareness of these trade-offs.

l 1

Use of a split-core rodel is a stat ~ af-the-art technique for an asymetric plant behavior which can infiten~' the core physics due to reactivity feedback frm the asymetric tarperature distribution caused by different cold leg fluids and the mixirg of these fluids in the core. This nodalization was used by ME for analysis of the multiple swvvt:lary side l

c. .

malfunction event, a natural circulation test fram 40% pwer, ard a runaway l

main feedwater to one steam generator event as part of a set of FTS-limitig system transient analyses.

The four-loop model explicitly models each of the four cold legs, including two cold legs per loops, a unique problem to CE plants. l This model is used when irdividual cold leg /RCP transients are analyzed such as three Calvert Cliffs start-up tests; a four pump coastdown frun 20%

power, RC coerational punp ccrnbination flow tests at hot zero pwer, and a one pump coastdown fram 80% power.

1 In all three of these plant models, the RETRAN non-equilibrium j pressurizer was used in combination with a single-node representation of the pressurizer. In order to ovexecne the inaccuracy introduced during the cer1putation of a rapid insurge/outsurge transient, B%E intends to use the ,

1 Temperature Transport Delay Time Model (TTDT) available in RETRAN. B ME has q demonstrated that they are aware of the limitations of the TTDT model and i

the non-egailibrium model and of their ranges of applicability.

The steam generator secondary side is modeled as a four volume )

recirculating steam generator with a best estimate recirculation ratio in l the two-1v.p and four-lcop plant nodalizations, while the one-loop nodalization only uses a one volume steam generator. Transients for which B%E intends to use the one-volume rodel have been discussed thoughtfully j and are listed in Table 1.

In addition, BG&E obtained frun Energy Incorporated and modified  ;

a IDFT RETRAN-01 IIs-5 deck to model the IDFT facility usirq RETRAN-02/)OD03 for sinulation of two tests; 14-1 and 16-3.

4.3 Models B%E has qualified a portion of 'their REIRAN code nrdaling for the Calvert Cliffs plant, and in a&iltion, they have provided Ma'iccions ,

4

j

. indicating their understanding of the details and limitations of certain of

• those models, including the' non-equilibrium model, the bubble rise model, the TTDT redel and they have . die'mW thoughtfully the impact of various steam generator nodalizations. 'Ihese discussions are important aspects of demonstrating their ability to use the code and analyze the results. In addition, the transients analyzed required sufficient use of the RETRMJ control systems modeling that BG&E has adegaately demonstrated its ability to use those input models to represent the plant's control systems, l

TABE 1.

BG&E Nodalization Selection .)

l 1

RE18aN SG SECONDARY SIDE }ODEL Ch.14 UPDATED FSAR EVDIT SDEE NODE MULTIPE LODE l l

RCSD X j IDF X SR X CEAD X CEAW X CEAE X EL X IDL X IDFW X FWM X 10AC X PELB X SGIR X FWIB X i

5.0 F) git Transient Otroarisons  ;

L.1 Multiole Secondary Side Malfunction Event A Isasonable agreement was obtained between the . REIRAN 5

t

calculated results ard data of the asymmetric cooldwn during the first 200 seconds of the event. B3&E, by performance of a sensitivity sttdy, irdicated that use of the TIUT model reduced the peak pressure from 2338 psia to 2310.7 psia which is cloter to the measured 2306 psia, ard increased {

the peak pressurizer level by about two inches closer to plant data. T.inung of these occurrences, hwever, were not affected by the use of this model.

Although, the results began to differ after roughly 200 seconds into the transient, B3&E g ized and discussed the fact that this was due to the uncertainty in the stuck open TBV position.

A series of sensitivity studies was conducted to determined which parameters would affect the transient results.

In the supporting document presented by the applicant [3), B3&E analysts demonstrated a thorough understardirg of the transient by providirg an explanation of each substantial change in slope of plant parameters ard l how they inpact each other. 'Ihe applicant should be encouraged to provide future subnittals in this depth.

5.2 Four PtTo Coastdown from 20% Power Meaets were made of the total RCS flw for the first 55 secords after reactor trip to obtain RCS flw data durirg a four prp coastdwn at 20% of full pwer.

l RETRAN calculated flow was within roughly 3% of reasured flw for the first 35 socords of the transient shere the flow measurement accuracy is 2%. A closer agreenent was obtained with the four-loop model than with the sirgle-lcop model due to the finer ncdalization and rore acx: urate representation of RCS flow paths and pressure losses. B3&E indicated an urderstarding of the accuracy of these nodels ard the unoartainties in plant data.

6 i

l l

i

. 5.3 Reactor Coolant Operation Pump Combination Flcu Tests at Hot Zero Power Fifteen various ccanbinations of f1cu tests were conducted at hot stand-by to evaluate operatire pump combination f1cu distribution. The g PfrRAN four loop model was redified to include a cross flow junction between the two . identical dcuncmer volumes used for this analysis, to perrlt such cross flcu to occur when the coastdown is asymmetric. B%E also irrut an explicit locked rotor reverse f1cu pressure loss coefficient to improve ncdelirg.

Statistical analysis was performed with the measured plant data and REIRAN cxrnputed results were cmpared to the mean value of the data. In twelve out of fourteen cases studied, the ocanputed results were within two  !

standard deviations (95.5% of data) of measured data. One of the remaining tests had insufficient data points to be meaningful and in the other test, the computed flow differed frtan the mean by 4.1%.

i l

5.4 One Puno Coastdcun frm 80% Power

- i The objective of this test was to reasure single RCP coastdcun data aM validate the low RCS flcu trip frun 80% power with one RCP secured.

The four-lcop model was used in the arkCysis. Model  !

enhancements including downcaner cross f1cu and realistic RCP reverse {

pressure loss coefficient were used. Other plants conditiens were adjusted to reflect cmditions at 80% power.

Althcugh the corputed pressurizer pressure an:1 hot and cold leg temperatures originally did not agree well and the_ total RCS flow under-predicted plant data by 1 to 4 % over the 60 second period, these differences were attributed by B%E to differences in TSV and AW opening times. B%E then used revised data, adjusted the TSV ard AW irput to the 7

~

)

code and obtained better agreement.

BG&E performed a sensitivity study in which RCP nament of  ;

inertia and rated torque were varied. It was found that by either increasing pump noment of inertia by 10% or decreasing punp torque by 10%,

temporal flow rate increased closer to measured data. It was also found that addirg a dwncamer cross flow path and realistic RCP reverse flow loss l factor resulted in higher RCS flow than the initial model after 18 seconds into the transient and cmpared more closely with the data (79.1% versus 78%

for the base case and 80% for plant data). For the base case with downconer j cross-flow and RCP reverse flow unial coefficients, REIPAN under-predicted RCS total flow (i.e. , over-predicts flow reduction in the shutdown loop) by 0.4% to 3.9%. "Ihese parametric studies indicate BG&E's ability to seek out the causes of differences between plant data and computational results. i l

l 5.5 Total Loss of Flow / Natural Circulation Test from 40% Power

'Ihis test was conducted to determine the power-to-flow ratio during natural circulation. Key parameters were recorded for only the first 60 auvvds of this transient although the transient lasted much longer. -l

'Iherefore, calculational result caparison was made for this 60 namnd period. For this simulation, the RUIRAN two-loop nodel was rxxiified to be .;

I initialized at the 40% power conditions. j Agrecment between the plant data and REIRAN results in RCS flow, j pressurizer pressure and pressurizer level was good; bwever, hot and cold j leg temperatures and therefore the secondary side pressures did not agree  !

well. 'Ihe temperatures differed by about 5'F ard the pressure by roughly 25 psi. (During this time pericd, the hot and leg temperatures decrouad ]

roughly 10'F and 5'F, respectively and the net change in the secondary i pressure was approximately 75 psi.)

'Ibe topical report Me'w=a= the ccrqparison of the onset and magnitude of the natural circulation flow well after 60 seconds (five 8

. 1 minutes), and B%E indicated that this discussion was based upon the record  :

of the test written by the operator rather than upon recorded reasurenents.

1 l

1 1

6.0 TRAC Analysis Coroarison he los Alamos National Laboratory (LANL) developed a finely detailed TRAC-PF1 nodalization of the Calvert Cliff plant including a three  !

dimensional vessel with modelirg of lower plenum mixing pipes. Both steam generators ard the pressurizer were rodeled usirg multi-volume multi-ro:le nodalizations. The two TRAC analyses presented by BGLE were performed by IANL. )

1 1

i For simulation of two transients performe:1 by IRE, it appears that B%E's objective was to perform their own best-estimate analysis for each of these transients, ard not to simulate the LNE computation of the transient by usirg exactly the same assumptions. Thus our review of these two analyses was not focused upon the degree of agreement or disagreement between the results obtained by the two codes, but was focused upon B%E's urderstandirg of their own results and the reasons for differences frcrn the TRAC results.

l \

l 6.1 Cooldown to RHR Entrv Usina ADVs ard APS I I I l

The REIRAN one-locp model was usal in the comparison of this I transient with the results frcn using a finely detailed TRAC nodalization.

Two major differences between the initial B%E calculations and those of LNE using these mim: are: (1) the pressurizer enptied in the RETRAN analysis while TRAC did not predict pressurizer emptying; ard (2)

TRAC predicted faster cooldown after 6000 namnds than RETRAN. Reanalysis of this transient by BG&E after the initial subnittal showed that difference (1) was due to different assumptions about the use of charging flow in the analysis ard the time at which charging was a===4 to be restored for level -

9

_ - _ _ _ _ _ - __a

~

control. With these changes, BG&E reported that RETRAN did not predict the pressurizer emptying. Difference (2) was attributed to the gecraetrical modeling of the AINs and, after B3&E altered its nodel, differences between the two calculations were said to have been " resolved".

6.2 Runaway Main Feeh'aler o One Steam Generator This analysis was performed as part of the NRC sponsored pressurized themal shock project.

The FORV reached its setpoint and the pressurizer level rose to-the top roughly 500 secords earlier in the RETRAN analysis than in the TRAC calculation because the TRAC model did not allow backup pressurizer heaters to reactivate after pressurizer liquid level returned above a programmed level setpoint. Furthe.Incre, BG&E indicated that the reason RETRAN ccanputed that the pressurizer level rose to the top over 1000 seconds earlier than IRAC was because of the large difference in HPSI flow rate which resulted from the different pressures predicted.

The minimum downcx:imr tanperature was ccmputed to be 350'F by l RETRAN, whereas TRAC cx2nputed minimum average downocraer temperature of ]

400*F. The hot leg terperaturus also shru a sirdlar difference between j EE.T RAN and TRAC. Overall, the results exhibited similar trenis despite large differences in the exrnputer code acdels and the degree of detail and capabilities in plant nodalizations.

l 7.0 IDPr 'Nst ocznoarison B3&E modified an M-5 REIRAN-01 model supplied by Energy Incorporated to wvh1 M-1 and M-3. These experiments are secondary system ir.itiated ,

events: M-1 was a loss of steam load anticipated transient; ard M-3 was an excessive load increase anticipated transient.

10

+

I 1

\

7.1 IDFT Test 1/2-1 l

The calculational results presented (steam generator dame pressure, secondary side liquid level, pressurizer pressure and level and j coolant temperature in the intact loop steam generator primary side inlet I J

plenum) in the topical report did not agree with the test data. After a l

sensitivity study with respect to the Interfacial Heat Transfer Coefficient j (DfIC) was perfomed, B3&E concluded that RETRAN does not permit a varying DfIC to simulate this sort of transient involvirq a rapid pressurizer j insurge followed by an equally rapid outsurge. The use of a large value of DfIC (27500 BIU/HR-FT 2 _.F) properly calculated the insurge portion but not j the outsurge portion because it caused excessive energy transfer frcen the I steam which resulted in a lower pressure, which in tum caused too much pressurizer liquid to flash during the outsurge. Correspondingly, the use of a low value (10 BIU/HR-FI2 _.F) of DfIt predicted plant data more closely during the outsurge portion than the original calculation which used large DfIC; however, pressurizer pressure and level were not well caputed during the insurge. BG&E fourri that 400.0 BIU/HR-FI 2 *F gave a better overall fit  !

to the data.

l l

i l In this analysis, the B3&E REIRAN prinary and secondary pressure I transient behavior did not follow the plant data precisely. We concur with BG&E's conclusion that this daviation is attributable to lack of basic data reganiing the details of Main Steam Flow Control Valve ope.mtion during the i

l transient, and that the absence of such data makes it virtually irpossible to obtain exact correlation.

1 I

7.2 IDFT Test II)-3 After BG&E AMai the use of the TIUT option, the primary side parameters, e.g., pressurizer pressure and hot leg temperatures agreed well l betwen the test data ard REIPAN results. The conputed coolant )

I 11 I

l

. 1

\

, temperature in the intact loop steam generator prirary side inlet plenum

. lagged the test data by roughly 25 seconds. 7his difference was attributed

.J to uncertainty in the data. 7he ccenputed ard test steam generator done ]

l pressures began to diverge at about 50 secorris due to the earlier RETPRJ reactor trip ard the resulting earlier closure of the MSITN. The reason for the pressurizer level divergence starting at about 50 seconds was attributed to less energy being available to be deposited in the secondary chte to the early scram arxi to a higher terminal steam flcu and feedwater ficw in the j REH W1 calculation. I I

1 i

1 1

)

l l

12

l 8.0 Conclusions and Reconnendations BG&E staff have demonstrated their ability to develop input models for I

best estimate cmputation of plant transients in the Calvert Cliffs Nuclear Power Plant, and to perfom the supporting sensitivity studies to aid in determination of the appropriate model selection and nodalization. BGLE has developed three nodalitations which it proposes to use; BG&E has presented j l

discussions arrl justifications of their nodalizations for certain of the transients (as irdicated in the foregoirg text), and such nodalizations ]

I should therefore be considered qualified for future use. It is r+> -, arded that in future licensirg applications, model selection be justified on a transient by transient basis. IE&E may, however, rely upon and reference the discussions contained in this application when makirq such inodel selection.

BG&E staff have further demonstrated their ability to thorcughly analyze a transient. It is recommended that in future licensirg subnittals BG&E be requestM to submit analyses of the detail and thoroughness provided in their analysis of Pultiple Secondary Malfunction Event.

13

1 1

• l

\

. l 1

1 i

9.0 References ]

4

1. BG&E Topical Report A-85-11, "RETRAN Ccrnputer Code Reactor System.

Transient Analysis Model Qualification," January 31, 1986. ,

1

2. Ietter fran J.A. Tiernan (BG&E) to NRC, dated February 24, 1987, "REIRAN Review - Submittal of Additional Infonnation".

)

3. Istter fram J.A. Tiernan (BG&E) to NRC, dated June 9,1987, "REIRAN  !

Review - Subnittal of A&iitional Information".

i

. .. 4. USNRC Generic letter No. 83-11,' " Licensee Qualification for - )

i Perfonning Safety Analyses in Support of LicensinJ Actions," l February 8, 1983. ]

5. Istter frun P.B. Abramson (ANL) to J. Carter (NRC), " Technical Evaluation Report: REIRAN-2AOD02," May 31,1983.

6. Letter frun T.W. Schnatz (UGRA) to C.O. 'Ihcanas (NRC), "REIRAN - A Frugam for Transient 'Ihermal-Hydraulic Analysis of Otmplex Fluid Systems," February 4, 1985.

l 1

1 14 u __ _ _ _ _ _ _ . _ . - . . _ _ _ _ _ _ _ _ _ _ _ _ - - l