ML20024E734

Topical Rept Evaluation of NEDE-23785-1(P),Rev 1, GESTR- LOCA & SAFER Models for Evaluation of Loca,Vols I & Ii. Vol II Acceptable

ML20024E734
Person / Time
Issue date:	08/29/1983
From:	Office of Nuclear Reactor Regulation
To:
Shared Package
ML20024E733	List: ML20024E732 ML20024E734
References
NUDOCS 8309060618
Download: ML20024E734 (17)

Text

f' TOPICAL REPORT EVALUATION Report Number:

NEDE-23785-1-P, Revision 1 Report Title :

The GESTR-LOCA and SAFER Models for The Evaluation of the Loss of Coolant Accident - Volume 2 Report Date :

December 1981 Originating Organization:

General Electric Company Nuclear Energy Division i

1.0' Introduction The purpose of this report is to provide the staff evaluation of the SAFER computer code (Ref. 1, Volume 2) for use as a best estimate model for loss-of-coolant accidents (LOCA) in boiling water reactors (BWRs). The! code was submitted for review by the General Electric Company in December 1981 as licensing topical report NEDE-23785-1-P,_ Revision 1, Volume 2.

Minor changes which have been made to the model since its submittal are documented in Appendix A to reference 2.

2.0 Summary of-Topical Report The p'urpose:of the SAFER code is to calculate long' term reactor vessel inventory and peak cladding temperature.for LOCA and' loss of inventory events. SAFER is intended for use with the GESTR-LOCA code (Ref.1, Volume 1) ~which is currently under staff review.

SAFER builds directly on the SAFE and REFLOOD codes (Ref. 3) and retains many of-these models' features. Major modifications are in 8309060618 830829 PDR TOPRP EMV w

the areas of counter current flow limiting models, core heat transfer models, bypass leakage models, and the nodal

. representation of the reactor vessel.

SAFER assumes thermodynamic equilibrium in each region.

Hydraulic and heat transfer modeling is essentially one dimensiona.1, and fluid properties are calculated at a mean thermodynamic pressure.

2.1 Hydraulic Models

. For the purpose of inventory calculations the reactor vessel is divided into eight regions:

(1) the lower plenum,~ (2) the volume inside.the control rod guide tubes, (3) the active core region, (4) the. core bypass, (5) the upper. plenum.(including _theJep. ara. tor standpipes), (6) the lower powncomer below the feedwater. sparger and outside the jet pumps, (7) the upper downcomer (above the feedwater sparger) and (8) the steam-dome. A single hot fuel assembly is also modeled as a separate region for the purpose of calculating ~ peak cladding temperature. Mass and energy balances are performed for each region and for the entire vessel..A pressure' rate is derived from the constraint that the total vessel volume is constant.

. Vapor-flow from.a' region. is modeled by a drift _ flux correlation or a bubble rise model. The void fraction profile is assumed to be a linear function.of the mean void fraction in a given saturated tregion.

SER2: Collins. -

Counter current flow limiting (CCFL) is modeled by a modified Wallis correlation.

Back flow leakage from the bypass region is modeled as a mechanism for filling the lower plenum when CCFL limits liquid downflow into the core.

Reactor pressure boundary break flows are modeled as the minimum of critical flow or Bernoulli flow with a loss coefficient of 1.0.

Overall loop momentum equations are solved for the two loops

~

through the jet pumps and the core.

Each loop is between the steam dome, through the downcomer; through a bank of jet pumps, into the lower plenum, and back up through the core and upper plenum to the steam. dome.

Several special regional models are used in SAFER.

The upper

-plenum model-incorporates features such as core spray distribution, non-uniform mixing, and liquid entrainment due to core exit vapor flow.. The core. bypass region is divided into two _ parts for CCFL calculations. The core region is divided into seven subregions for a more accurate estimate of the axial void distribution. Mass and

~

energy balances are also performed on each subregion of the core.

A hot fuel-assembly model is included to calculate the fluid inventory' and-PCT.in a high power bundle.

Flows in and out of the hot channel are calculated by imposing on the hot channel the plenum to plenum' pressure drop'taken from the average core' calculation.

2.2 -Heat Transfer Models SER2: Collins,

o The fuel rod and cladding transient temperature response is determined by solving the transient heat conduction equation in cylindrical coordinates assuming no axial or circumferential heat conduction.

A maximum of 10 equally spaced radial nodes and 5 axial nodes.are used.

The reactor pressure ves'sel thermal response is simulated by considering 4 lumped parameter heat slabs characterized by specified constant values for. thermal capacitance and thermal resistance. The vessel internals are simulated with six lumped parameter heat slabs.

Surface heat transfer coefficients for these slabs are specified as a function of void

' fraction for each hydraulic region contacted by the slab. The heat source is the sum of contributions from decay heat and metal water reactions.

Gamma smearing is also modeled.

The metal water

' reaction heat source assumes a continuous-supply _of_ water _or_ steam _

and that no energy is required to bring the reactants to the reaction temperature.

Heat ' transfer from the fuel rods' to the coolant is calculated using

~

realistic-values for heat transfer coefficients based upon different-heat transfer regimes. These. include nucleate boiling, forced flow film boiling, pool' film boiling, and transition boiling. ~ Nucleate boiling heat transfer. coefficients are modeled as a function of coolant void fraction.

The applicable heat transfer regime is determined from a' logic which examines the-coolant quality, the cladding superheat, and the critical heat flux. Steam cooling, core spray heat' transfer, and radiation heat

-transfer are also considered.

SER2: Collins h

The heat transfer models are applied to different characteristic fuel rods to obtain the fuel rod. temperature distribution. The

~

temperature distribution for the average power rod in the average power _ bundle with core average water inventory is used in the system mass and energy balance calculations. The maximum cladding temperature is calculated for the hottest rod in the average power bundle in both the central and peripheral core regions. The core

~

wide PCT is determined from the temperature response of the hottest rod in the hot channel.

3.0. Summary of Staff Evaluation The staff evaluation of SAFER was performed as follows: Modeling changes relative to the previously. approved models (Ref. 3) were

-identified.

The new models were then evaluated separately against experimental data.

The composite SAFER model was then evaluated against systems test data from the Two Loop Test Apparatus (TLTA),

and the ROSA facility in Japan.

Finally the predictions from SAFER were. compared against TRAC (Ref. 4,5) predictions for the.same events.

3.1 Model Changes Relative to the Current Evaluation Model

-The changes to the SAFE and REFLOOD codes which have a major impact on the peak cladding temperature calculations are: 1)CCFLis modeled at the fuel bundle side entry orifice (SE0); 2) steam j

cooling ~ heat transfer above the two phase level is included; 3) a more' realistic correlation is used for film boiling heat transfer;

'4) additional _ leakage paths from the bypass to the bundle and lower SER2: Collins-.

w

~

e 1en

p plenum are included; and 5) the reactor vessel is modeled as 8 separate regions.

~

Changes which have a minor impact on the peak cladding temperature calculations are: 1) a transition boiling correlation has been fit be'tweentheendpointsoftilenucleateboilingandfilmboiling regimes; 2) a drift flux model option for two phase flow

^

calculations has been introduced; 3).the upper tie plate CCFL correlation has beer refined to better reflect'the data; and 4) a' simplified radiation heat transfer option has been introduced for

~

cases where the calculated PCT is relatively low.

3.1~.1

.. Counter Current Flow limiting at the Side Entry Orifice (SEO-CCFL)

'CCFL_is a phenomenon in which upward steam flow through a-restriction limits the rate at which liquid may flow down through the same restriction. _ CCFL is important at two locations in a BWR under LOCA conditions: the upper tie plate and the side entry-

orifice (SEO).

CCFL at the upper tie plate can limit the amount of

' core. spray water which drains into a fuel bundle.. CCFL at the SE0 limits the rate at which liquid can drain from the bundle into the lower plenum. CCFL-at the upper tie p. late (UTP) is non-beneficial; CCFL at 'the SE0 is very beneficial. -The model used by GE for CCFL has been previously reviewed by the staff (Ref. 6) for UTP-CCFL.

1GE has proposed a model of_ the same form for side entry orifice =

CCrl.

Data supporting the_GE proposal at the SE0 is shown'in SER2: Collins i

Appendix _C to reference 2.

The data covers limiting conditions from full liquid downflow to liquid shutoff for several orifice sizes. We have reviewed these data and conclude that the GE coefficients used in a modified Wallis correlation represent the data very well and are acceptable.

~

3.1.2 Steam Cooling Heat Transfer The heat transfer. logic in SAFER includes the effect of steam

~

cooling whenever a fue_1 node is uncovered. The Dittus-Boelter

. correlation is used as the model for single phase vapor heat General Electric performed heat transfer experiments on transfer.

~

a 4x4 rod bundle and reported the results in reference 7.

The data

covered pressures from 190-590 psia and exit steam superheats of 100-_400'F.

The Dittus-Boelter correlation'was compared with several other correlations and found to give the'best I

representation of the test data. The. SAFER steam cooling model is acceptable.

3.1.3

' Additional Backflow Leakage Water draining from the bypass region through leakage paths.to the bundle and lower plenum. contributes significantly to bundle and

' lower plenum refill.

SAFER includes the leakage paths shown in Figure 1.

Paths 7_ (finger spring flow) and 8 (holes in lower tie plate) are not included in.the currently approved evaluation model.

GE' performed backflow leakage tests for these paths in the ATLAS facility and has reported the data in reference 8. -These data SER2: Collins

-7.

indicate that the correlations used in SAFER are slightly conservative, i.e., SAFER underpredicts the amount of leakage measured.

The tests were performed on production hardware and covered a driving differential pressure range of 1-6 psid. Water temperatures of 150-400*F were used. These represent conditions which may be expected after a LOCA.

Flow reduction factors due to boiling.in the leakage paths are included in the correlation. We

~

have reviewed the' test data,and conclude that the GE correlations give a reasonable representation of the data and are acceptable.

3.1.4 Revised Film Boiling Correlation SAFER uses a combination of the Dougall-Rosenhow correlation and the Bromley correlation for film boiling heat transfer.

Experimental data supporting.the use_of_these_corre.lationLare discussed in reference 9..iThe Bromley/Dougall-Rosenhow model has been previously approved by the staff (ref. 6) for use with the

~

current evaluation methodology.

In SAFER, the film boiling heat transfer coefficient is weighted by void fraction so that the Dougall-Rosenhow correlation dominates at higher void fractions and the Bromley correlation dominates at lower void fractions. This is consistent with the data which. indicate that the Bromley correlation may overestimate the. heat transfer when the core is

. highly. voided.

The experimental data discussed in reference 9 include pressures up to 1000 psia and wall superheats up to 1500'F.

We conclude iihat'the-SAFER film boiling correlation is acceptable.

SER2: Collins.

3.1.5 Reactor Vessel Nodalization SAFER models the reactor vessel as eight' separate regions (Figure 2). The current evaluation model assumed only two physical regions. The detailed noding more accurately models mass distribution because a separate void fraction is calculated for each' region.

Flashing and boilingLmass losses are also better defined since they are determined in each individual region.

This is acceptable to us.

f

- 3.1.' 6.

Other Model Changes The remaining ~model changes discussed below have a small impact on the-calculation of peak cladding temperature and are acceptable.

Transition boiling is modeled as a smooth transition between

- nucleate boiling and film boiling. When a fuel node is covered by

- a two phase mixture but the coolant quality is below a specified critical value and the cladding superheat is also less than a specified value, transition. boiling-is assumed. A correlation due to Iloeje (ref.10) is used to' calculate the wall superheat.

The

-Iloeje correlation is based upon high quality film boiling test data.' Such conditions are expected in e BWR during a design basis LOCA.

Two modeling options are available for two phase flow calculations:

the drift: flux model and the bubble rise model. : SAFER calculates

. vapor slip flows using both and selects the-higher flow.

This 9-SER2: Collins y

e e

-e r

a i----'

v-w-me

results in the_ bubble rist model being used for low flow, low void conditions and the, drift flux model being used for high void and high flow conditions. Appendix C to reference 2 contains the results of a sensitivity study which shows that the.CT is not significantly impacted regardless of which model is used.

The upper tie plate CCFL model in SAFER is a b'est estimate correlation of the_available data (ref. 11).

The currently approved evaluation model uses a conservative bound.

' The _ radiation heat transfer model in SAFER is simplified relative to that used in the current evaluation model, and is more conservative. The radiation model is not critical however since

- ~ ~ '

calculated PCTs are low.

SAFER'includeran option to calt the more~

detailed radiation model us' d in CHASTE (ref. 3) if necessary.

e To address multidimensional effects during the refill /reflood phase of a LOCA, SAFER models the core as two bundle types, hot bundles and average _ bundles.

Steam flow to the hot bundles is subtracted "from the' total-lower plenum steam flow.

The balance of the steam flow is. equally' distributed among the average bundles. Upper

. plenu'm liquid drains to the lower plenum at a rate which maintains

'the _ level in -the upper, plenum at or below'the core spray sparger.

We have' reviewed data from tests at the Steam Sector Test Facility.

(Ref.12) and find that this model reflects the phenomenon

~

- observed.

There is little impact.on the calculated PCT however, i

SER2: Cnilins- __

since this effect does not come into play until after the cladding temperature has leveled off.

4.0 Comparisons With System Test Data Predictions using SAFER were compared with test data from the Two Loop Test Apparatus (TLTA). TLTA is a single bundle, electrically heated,l full height facility used to simulate BWR transient response.

Data from boiloff, intermediate break, large break, and small break simulations were compared with predictions from SAFER.

Vessel pressure, two phase level, mass inventory and cladding temperatures-were compared.

Detailed plots of the data and the SAFER predictions are given in Appendix B to reference 2.

We have reviewed these data and find that SAFER captures the data trends

-well for each of tests.

PCT was consistently.overpredicted by about 100*- 200'F.

i SAFER outputs were also compared with test data from the ROSA III-Facility in Japan.. ROSA III is a 4 bundle, half-height facility.

'We have reviewed the PCT predictions and data from these tests (ief.13) and find that SAFER predicts the small break PCT well and

=is conservative in its PCT predictions for the large break simulation.

9 5.0 Comparisons With TRAC

.The results of SAFER runs' were also compared with TRAC predictions for the same events. TRAC is a 3 dimensional best estimate code.

Predictions ~for-the TLTA tests were compared, as well as several SER2: Collins- '

l.

licensinc basis events.

For the BWR/6 product line, 3 recirculation line breaks were examined (100% DBA, 80% DBA, and 2

IFT ), a steam line break and a HPCS line break.

For the BWR/4 product line, a.100% DBA line break calculation was compared.

The results showed that both codes predicted similar trends, and SAFER calculated consistently higher PCTs. We noted however, that the SAFER.model for. core spray heat transfer is more conservative than

~

that used in TRAC.

In addition, SAFER defines the end of the liquid continuous region at the 90% void fraction level. This results in a lower overall heat transfer rate and thus, higher PCTs

. relative.to TRAC. The plots for the SAFER and TRAC predictions are provided in' Appendix D to reference 2.

We have reviewed these outputs and conclude that the two codes are in good agreement, with no significant discrepancies..

6.0 Reaulatory Position 1.

The staff concludes that SAFER is qualified as a code for best-estimate.modeling of loss-of-coolant accidents and loss of.

inventory events in General Electric designed boiling water

. reactors.of the BWR 3, 4, 5 and 6 class. The basis for this position is the st'aff review of licensing ' topical report NEDE-23785-1-P, Revision 1, Volume.2..

- 2.~

SAFER may be used as a licensing evaluation model only in conjunction with an application methodology which has been specifically. approved by the staff for use with SAFER.. General SER2: Collins -

^

Electric'has proposed an application methodology which tne staff is

-currently reviewing.

- 3.

SAFER may only be used as a licensing model in conjunctior, with the GESTR-LOCALcode (ref. 1, Volume 1).

The evaluation of GESTR-LOCA will be provided by Core Performance Branch.

4. -

The staff is inspecting the General Electric Company administrative' control procedures for computer code quality assurance. Responsibility for this inspection is assigned to NRC's Region _.IV office and the inspection will be initiated during August 1983 and will_ be' completed during September 1983.

If.esults of the inspection show that serious deficiencias exist in GE administrative control procedures, appropriate corrective actions

- will be required.

u

' SER2i Collins.

g-w y

4 o

4

References 1.

"The GESTR-LOCA and SAFER Models For the Evaluation of the Loss-of-Coolant Accident", NEDE-23785-1-P, December 1981

'(proprietary).

2.

Letter from J.F. Quirk to C. Thomas, " Formal Submittal of Final Set of Information on General Electric's LOCA Eval.uation Model (SAFER)"

~

dated June 28,1983.(proprietary).

3.

'" General Electric. Company Analytical Model For Loss-of-Coolant

~

Analysis in Accordance with 10CFR50 Appendix K" NEDE-20566P,

. -November 1975 (proprietary).

"4.

"BWR Refill-Reflood Program Task 4.7 - Basic Models for.BWR Version of TRAC," GEAP-22051 (Draft) to be issued August 1983.

5.

'"BWR. Refill-Reflood Program Task _4,7_. _IRAC. BWRlomAonent Models2'_

GEAP-22052 (Draft) to be is, sued Adgust 1983.

6. -Letter from R. Tedesco to G. Sherwood,," Acceptance For Referencing, of Topical-Report'NEDE-20566P, NEbO-20566-1, Revision 1, and NEDE-20566-4, Revision 4", dated February 4, 1981.

7 "A Study of Heat Transfer to Superheated-Steam in 4x4 Rod Bundle",

NEDE-13462, June 1976.

8.

" Backflow Leakage From The Bypass Region For ECCS Calculations",

NEDE-20566-5-P, June 1978 (proprietary).

~

9.

'" Calculation of Low Flow Film Boiling Heat Tranc#cr For BWR LOCA Analysis",- NEDE-20566-1-P, January 1977 (proprietary).

10.-

0.C. Iloeje, et al., "An' Investigation of the Collapse and Surface Rewet in Film Boiling-in Forced Vertical-Flow", Transactions of the "ASME,.J.H.T., May 1975.

~

'SER2:: Collins-.

- ;of

'o;

~

" Saturated Counter-Current Flow Characteristics of A BWR Upper 11.-

' Tieplate".NEDE-20566-4-P, July 1978 (proprietary).

12.

"CCFL/ Refill System Effects Tests (30' Sector) - Evaluation of ECCS

. Mixing Phenomena", NUREG/CR-2786, May 1983.

.13.

Letter from J.F. Quirk to B. Sheron', "Smtrary of June 10, 1983

- NRC/GE Meeting on SAFER Approval" dated June 22, 1983 (proprietary).

4 b

b

- SER2':'. Collins --

• A.

,e e

/

h

(( CHANNEL

, FLOWER r

TIE PLATE 8 (WTP) 1

(

)

i

/

- FUE L SUPPORT g3 CORE SUPPORT INCORE GUIDE TUBE

-)

WLEAK

%CONTRO L ROD

. G UIDE, TUB E a

J SsROuO j

i e

1.

FUEL SUPPORT - LOWER TIE PLATE 1 a, I b. CONTROL ROD GUID E TUBE - FUEL SUPPORT 2'

CONTROL ROD GUIDE TUBE - CORE SUPPORT PLATE 3.

CORE SUPPORT PLATE - lNCORE GUIDE TUBE 4

SUPPORT PLATE HOLES (OR PLUGGED HOLE LEAKAGE) 5.

CORE SUPPORT PLATE - SHROUD (SINGLE PATH) 6.

CONTROL ROD GUIDE TUBE - DRIVE HOUSING 7.

FINGER SPRING FLOW 6 (W L1) 8.

HOLES IN LOWER TIE PLATE CONTROL RCD DRIVE HOUSING Figure i Leakage Flow Paths

,4 O

l

- w pmj SATURATED j

FEEDWATER IN LET f

if! l

///

' UPfER ILENU@l/,

&t

/ / / /

f SUBCOOLED b.

\\)

o o

9 CORE BYPASS

/

f L

u

=

g/

,x p-l J

l 1

LOWER I

~

PLENUM J

G Figure 2.

SAFER Nodalization EiE l

s l

,,._-,,...._.._,.-...____m..__

,.