ML20196G632

Safety Evaluation of Topical Rept EMF-2087(P),Rev 0, SEM/PWR-98:ECCS Evaluation Model for PWR LBLOCA Application, Rept Acceptable

ML20196G632
Person / Time
Issue date:	06/15/1999
From:	NRC (Affiliation Not Assigned)
To:
Shared Package
ML20196G615	List: ML20196G632
References
NUDOCS 9907010221
Download: ML20196G632 (6)

Text

,

pnneo%

y & UNITED STATES l

l s*

f j NUCLEAR REGULATORY COMMISSION WASHINGTON, D.C. 20066-0001 l

\...../ .

ATTACHMENT SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION TOPICAL REPORT EMF-2087(P). REVISION 0 "SEM/PWR-98:ECCS EVALUATION MODEL FOR PWR LBLOCA APPLICATIONS" SIEMENS POWER CORPORATION (TAC NO. MA3457) 1 INTRODUCTION AND BACKGROUND Topical Report XN-NF-82-20(P), Revision 1, Supplement 5, "EXEM/PWR Large-Break LOCA .

ECCS TOODEE2 Updates," provided a description of and supporting technicalinformation for I all changes made to the TOODEE2 hot rod heat up computer code made since the code's approvalin 1986. Topical Report XN-NF-82-20(P), Revision 1, Supplement 6, documented the updates related to a change in the implementation of the fuel cooling test facility (FCTF) reflood ,

heat transfer correlation in the TOODEE2 hot rod heat up code and a correction to the Z- l Equivalent model. The staff subsequently approved the modifications to the EXEM/PWR Large-Break LOCA code (Reference 1).

Siemens Power Corporation (SPC) informed the staff of the potential for excessive variability in its EXEM/PWR LBLOCA evaluation model because of excessive variability in the RELAP4 code which forms a base in the EXEM/PWR code (Reference 2). Topical Report EMF-2087(P),

"SEM/PWR-98, Revision 0: ECCS Evaluation Model for PWR LBLOCA Applications," which presented modifications to the EXEM/PWR code to correct the excessive variability problems was submitted for staff review (Reference 3).

Review of the EMF-2087(P), Revision 0, topical report resulted in requests for additional information from the staff (Reference 4), and responses to those requests from SPC (Reference 5).

2 PROPOSED MODEL REVISION ,

While using the EXEM/PWR LBLOCA code, SPC found that some input changes that were expected to be inconsequential would result in large calculated changes in the peak cladding temperature (PCT). Revisions to the code to correct some of the PCT variability resulted in a reduction of more than 50* F for some plants. When PCT calculations result in a drop of 50' F or more, further changes are required to remove non-conservatism in the Dougall-Rohsenow correlation. Further, a 1997 inspection conducted at SPC by the NRC resulted in commitments for revisions of the emergency core cooling system (ECCS) models. The following modet revisions have been presented to correct the PCT variability problem, Dougall-Rohsenow correlation non-conservatism, and commitments made to the staff for ECCS model corNetion.

9907010221 PDR 990615 '

TOPRP EPfVEXXN PDR

1  :

.2 Numerical Scheme SPC found that there were several root causes to the " excess variability" problem and made several modifications to the numerical solution method and calculation methodology in order to  !

fix the problem. RELAP4-EM originally used the Porsching method (Reference 6) to advance  !

the solution in time. SPC modified the Porsching method to use more new time information in the numerical solution scheme effectively making the solution more implicit in time. '

Time step variatiori has been noted to cause calculational variability in large computer codes.

SPC modified the solution method to automatically control the timestep size based on convergence of the numerical scheme. Previously the user had to manually determine time step convergence.This procedure not only assures convergence, but also helps eliminate

" water packing" which makes the calculation oscillatory. The revised numerical scheme has i been used successfully in the boiling-water reactor (BWR) version of the code, giving more

. consistent, converged, and reliable results.  !

- Volume Flow Definition -

The code uses volume average flow in both the momentum equation and for heat transfer and critical heat flux The momentum equation calculation has been modified to use a simple l average of the inlet and outlet junction flows to avoid a discontinuity that occurs under certain conditions when the volume average flow exceeds the outlet flow but the inlet flow density is closest to the volume density. The volume averaged flow has been defined for use in the heat ,

transfer correlations as an integrated average of the absolute value of the flow through the ,

volume. This is always non-negative. Since the Richert-Franz correlation uses a drift flux  !

model based on the direction of flow with respect to gravity, the volume averaged flow is  ;

assigned a positive or negative value depending on whether it is upward or downward.

Combined System and Hot Channel Calculations  !

' Historically, the computers available at the time of the development of the original version of the l SEM/PWR-98 code (the Exxon Water Reactor Evaluation Model) had insufficient memory and i storage to permit complete system blowdown and core hot channel calculations. Current

- generation computers now make it possible to perform the complete calculation set with a single input model. Thus, the previous procedure of performing separate calculations for system blowdown and core heat up have been combined through redefining the parameter ,

dimensions. This modification removes smallinconsistencies between the system calculation  !

and the hot channel calculation that existed in the uncoupled calculation. It has not been necessary to modify the analytical methodology to effect this change.  !

Elimination of Enthalov Transoort Model i As discussed in the performance of single combined system and hot channel calculations, advances in computer technology no longer necessitate use of coarse nodalization and

- enthalpy transport models to model a reactor coolant system. It is now possible to use denser nodalization schemes that avoid the discontinuous behavior of the enthalpy transport model.  ;

Now, the extent of core nodalization required to reach convergence is determined based on sensitivity studies. ,

f

u ...-

)

3 Consistency Between RODEX2. RELAP4-EM. and TOODEE2

The original EXEM-PWR code package applied the RODEX2 fuel rod input only to the hot channel calculation.1This has now been expanded to permit the RODEX2 model to be applied to the average fuel rod, the average rod in the hot assembly, and the hot ' rod: A further modification was made to permit a gadolinia rod should a need arise for analysis of gadolinia bearing rods.

1 Revision of Douaall-Rohsenow Correlation A requirement of 10 CFR 50, Appendix K, is that when a calculated PCT reduction of 50' F occurs for a plant, the known non-conservatism associated with use of the Dougall-Rohsenow l film boiling correlation must be removed. The Dougall-Rohsenow correlation was developed to

l. be an extension of the Dittus-Bolter single-phase turbulent flow heat transfer correlation to two-phase flow. The problem with nonconservative heat transfer predictions by the Dougall-Rohsenow correlation has been shown to be caused by the equilibrium assumptions used in

. development of the correlation (Reference 7). SPC developed the Richert-Franz heat transfer correlation to correct the deficiencies of the Dougall-Rohsenow correlation in non-equilibrium

. conditions..  !

The Richert-Franz correlation is derived from the same basis as both the Dittus-Boeiter and Dougall-Rohsenow correlations (single phase liquid or steam, and film boiling). The Richert-Franz correlation is designed to take into account both fluids (steam and liquid) interacting at -

the wall. The two fluids are treated with their own properties. While the basic form of the correlation is the same as the two from which it was derived, the Richert-Franz correlation adds together two heat transfer coefficients. One coefficient represents superheated steam at the

wall while the other represents saturated steam at the wall. The two fluids are based on their superficial velocities in the calculated two-phase flow. The superficial velocities are based on the Ohkawa-Lahey drift flux model used in the SPC BWR methodology (Reference 8)

. The Richert-Franz correlation has been shown to be conservative relative to the measured experimental data from tests performed by Edgerton, Germeshausen & Grier at the Idaho National Engineering Laboratory, tests performed by Combustion Engineering, and thermal hydraulic testing facility (THTF) experiments performed at Oak Ridge National Laboratory.

Revised End-of-Bvoass Model Appendix K to.10 CFR Part 50 provides a definition of end-of-bypass and means by which the end-of-bypass may be calculated. The SPC model has been changed so that instead of

discarding both the ECCS and non-ECCS water at the end-of-bypass, just the ECCS water

. remaining lin the system is discarded. The primary system water that is calculated to remain in the system is retained. The model defines the end-of-bypass as the minimum of the time that.

sustained positive flow occurs from the upper to the lower downcomer volume, or the time that sustained positive flow occurs from the broken cold leg to the upper downcomer volume, less the time required to fill the portion of the cold leg from the ECCS injection point to the reactor vessel.

l 1

i

=

[:-

j .-

t 4

. Revised Pumo Dearadation Model The original EXEM/PWR evaluation model used a two-phase pump degradation model based

- on.Semiscale pump data. Since that time, data more representative of large commercial .

' nuclear power plant pumps have become available. The two-phase pump degradation model ,

has been revised to use data obtained in the CE-EPRI Pump Two-Phase Performance Program j

'(Reference 9). The model is the same one developed for use in the SPC realistic loss-of-coolant accident (LOCA) model The new modelis pressure dependent. Two-phase pump degradation increases as pressure is reduced due to the increase in the ratio of vapor to liquid specific volume. Test cases using the two different pump degradation models showed little difference in computed PCT.

Inertial Flow Estimate for' Critical Flow Mgdgj The numerical method for flow predicti;n was modified to use an inertial flow model to determine if the flow reached the critical flow velocity before the end of the time step. If it does the flow is limited to the critical flow velocity. This modification prevents the over prediction of flow from the break and will only have a small effect concentrated at the beginning of the  ;

calculation.-

]

kbble Mass Intearation Model The numerical solution method of the bubble mass integration model was improved. This change removes an approximation used in the previous integration model and will result in a more accurate solution of the equations.

Pumo Model Numerics The pump model numerical integration was modified to become implicit with respect to the pump flow. Since the pump head can depend strongly on the pump flow this improvement will improve the stability and accuracy of the calculation.

Claddina Creen Modelin RODEX2 SPC made several revisions to the cladding creep model in RODEX2. The revised model includes a term for thermal creep (MTYPE=4 option). The functional form of the model was changed to be of the same form as the approved RODEX3 model. The empirical coefficients were chosen to give more conservative results than the RODEX3 implementation which was designed for best estimate applications. The model was reviewed by Pacific Northwest National Laboratory and documented in a technical evaluation report (Reference 10). The staff finds the model acceptable for use in large-break LOCA analyses with rod-average bumups up to 62 GWd/MTU.

3 MODEL ASSESSMENT The assessment of the model changes can be broken down into sensitivity studies and benchmark cases. The benchmark cases are comprised of separate effects tests performed at I

l I

n .

L 5

l the thermal hydraulic testing facility (THTF) facility and integral system test L2-6 performed at the loss-of-fluid testing facility (LOFT).

I Sensitivity Studies -

In keeping with the requirements of 10 CFR 50.46, and 10 CFR 50, Appendix K, that sensitivity j calculations be performed for LOCA codes, SPC has determined appropriate nodalization to l

represent the reactor core and steam generators through sensitivity studies. In addition, i sensitivity studies have been performed on time step convergence, axial nodalization, radial fuel rod noding, break spectrum, worst single failure, and pump degradation model. Since the Dougall-Rohsenow correlation has been replaced with the Richert-Franz correlation, validation against three sets of film boiling heat transfer data was also performed.

A full break spectrum analysis was performed on a three loop Westinghouse plant using four break sizes (0.4,0.6,0.8 and 1.0) for both the double-ended cold leg guillotine (DECLG) and cold-leg split (CLS), and three axial power shapes, bottom of core (BOC), middle of core (MOC), and top of core (TOC). The limiting break was found to shift from the 0.8 DECLG MOC case to the 1.0 DECLG MOC. The PCT decreased from 1988F to 1925 F.

Assessment of the RODEX2 (MTYPE=4) model changes is described in Reference 10.

Benchmark Cases 1

Large system analysis computer codes are not only to be assessed against separate effects  !

data, but also by comparison with integral systems test data so that the overall performance of l the code and the interaction of the various models can be determined. SPC has assessed the l SEM/PWR-98 code against a series of THTF tests as well as against the LOFT L2-6 test. The l THTF test assessments demonstrate the conservative nature of the code in predicting both fluid conditions and heat transfer in film boiling heat transfer; The LOFT test was performed in a i scaled, integral nuclear reactor and demonstrates the ability of the code to predict the system i blowdown and recovery phenomena. The LOFT calculations were performed both with the i Appendix K features tumed off and with them tumed on. I i

Comparisons between SEM/PWR-98 and THTF tests indicate the THTF data are virtually all bounded by the SEM/PWR-98 predictions. The few data points that are not bounded exceed  ;

' the predicted values by a small amount, and are not limiting.

The SEM/PWR-98 predictions of the LOFT L2-6 test are reasonably close to the data for  ;

system primary pressure, break flow, cold leg flow, accumulator level and pressurizer level. In ,

both the EM and non-EM mode the SEM/PWR-98 prediction does not pick up the LOFT fuel l rod quench that occurs following blowdown because of the CHF locking required in Appendix K

~ Models. Both cases predict a higher blowdown peak cladding temperature, and a much higher PCT at the start of reflood. Thus, the code is conservative relative to the LOFT test PCT, 4 CONCLUSION L The staff has determined that the modifications made by SPC adequately resolve the previous ,

code calculation problems. SPC has performed adequate testing and validation of the new

F . ,

s' l-6 models to show that they conservatively predict the peak cladding temperature for a LBLOCA.

The revised model complies with the required model features of 10 CFR 50, Appendix K.

5' REFERENCES

1. Letter from T. H. Essig, NRC, to J. S. Holm, SPC, " Acceptance for Referencing of the Topical Report XN-NF-82-20(P), Revision 1, Supplement 6, 'EXEM/PWR Large Break -

LOCA ECCS TOODEE2 Updates," dated June 5,1998.

I

2. Letter from J. F. Mallay, SPC, to NRC, "RELAP4 Excessive Variability," dated March 17,

~1998.

' 3. Le' tter from J. F. Mallay, SPC , to NRC, Request for' Review of EMF-2087(P), Revision 0,

. SEM/PWR-98:ECCS Evaluation Model for PWR LBLOCA Applications," dated

'4. August 31,1998.

5. Letter from E. Y. Wang, NRC, to J. Mallay, SPC, Request for Additional information to the Topical Report EMF-2087(P), Revision 0, "SEM/PWR-98: ECCS Evaluation Model for PWR LBLOCA Applications" TAC NO. MA3457, December 4,1998.

1

6. Letter from J. Mallay, SPC to NRC, Request for Additional Information to the Topical j Report EMF-2087(P), Revision 0, "SEM/PWR-98: ECCS Evaluation Model for PWR i

' LBLOCA Applications" TAC NO. MA3457, December 18,1998.

{

i

7. Porsching, T. A., Murphy, J. H., Redfield, J. A., " Stable Numerical Integration of Conservation Equations for Hydraulic Networks," Nuclear Science and Technology, Vol. 43 (February 1971), pp. 218-225.

= 8. Morris, D. G., Mullins, C. B., Yoder Jr. G. L., "An Expsrimental Study of Rod Bundle Dispersed-Flow Film Boiling with High-Pressure Water," Nuclear Technology Vol. 69, {

. pp. 82-93.,1985 '

9. Ohkawa, K., Lahey Jr., R. T., "The Analysis of Proposed BWR Inlet Flow Blockage Experiments," Department of Nuclear Engineering, Rensselair Polytehnic Institute, Troy, 1 New York, 12181,~ 1978.

10. " Pump Two-Phase Performance Program," EPRI NP-15556, Volumes 1-8, September 1980. -;

11. Beyer, C. E., " Technical Eva!uation Report of the Topical Report EMF-2087, Entitiled )

_'SEM/PWR-98: ECCS Evaluation Model for PWR LBLOCA Applications'," Contract DE-  !

AC06-76RLO'1830, May 1999 i 12.' Letter from J. Mallay, SPC to NRC, " Response to Request for Additional Information l

t regarding EMF-2087(P)," Revision 0, "SEM/PWR-98: ECCS Evaluation Model for PWR LBLOCA Applications" TAC NO. MA3457, NRC:99:010, Dated April 20,1999.

j i