ML20217K065

Safety Evaluation of Topical Rept BAW-10193P, RELAPS5/MOD2-B&W for Safety Analysis of B&W-Designed Pressurized-Water Reactors. Rept Acceptable for Referencing in Licensing Applications

ML20217K065
Person / Time
Issue date:	10/15/1999
From:	NRC (Affiliation Not Assigned)
To:
Shared Package
ML20217K053	List: ML20217K046 ML20217K065
References
PROJECT-693 NUDOCS 9910260038
Download: ML20217K065 (15)

Text

.

,. m,

.; pf*%,

K .p " UNITED STATES g j; NUCLEAR REGULATORY. COMMISSION e ,

WASHINGTON, D.C. 30006 4001

...../

1 SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION FRAMATOME TECHNOLOGIES GROUP TOPICAL REPORT BAW-10193P "RELAPS/ MOD 2-B&W FOR SAFETY ANALYSIS OF B&W-DESIGNED PRESSURIZFD-WATER REACTORS" 1.0 - -INTRODUCTION -

' By letter (Reference 1) dated August 14,1995, Framatome Technologies Group (FTG),

. formerly B&W Nuclear Technologies (BWNT), submitted Topical Report BAW-10193P, "RELAPS/ MOD 2-B&W for Safety Analysis of B&W-Designed Pressurized Water Reactors," for staff review. FTG intended to use the RELAPS/ MOD 2-B&W code (Reference 2) to perform

. future safety analyses of non-LOCA (loss-of-coolant accident) transients and accidents for the

Babcock & Wilcox (B&W)-designed pressurized-water reactors (PWRs), replacing the currently used CADDS (Reference 3) and TRAP 2 (Reference 4) codes. Currently, CADDS is used to analyze system responses to primary system transients sucn as reactivity transients, loss of '

primary flow events, and anticipated transients without scram; and TRAP 2 is used to calculate system responses to the secondary system initiated events, such as steam line break, turbine trip, feedwater line break, and steam generator tube rupture. Consistent with how theTRAP2 and CADDS codes are currently used for safety analyses, RELAP5/ MOD 2-B&W will he used for predicting the reactor coolant system (RCS) and core power responses to non-LOCA

. events, while the LYNXT code (Reference 5) will continue to be used for calculating minimum departure from nucleate boiling ratio (DNBR) for the core hot channel, 1.1 _ Background - RELAP5/ MOD 2-B&W '

RELAP5/ MOD 2-B&W is an FTG version of the advanced system analysis code RELAPS/ MOD 2. _RELAP5/ MOD 2 was developed by the Idaho National Engineering Laboratory as a best-estimate code for analyzing a w ' ide variety of light-water reactor (LWR) system i transients.: The fundamental equations, constitutive models and correlations, and method of solution of RELAP5/ MOD 2 are described in NUREG/CR-4312 (Reference 6).- The code is designed to model the behavior of all major components in the reactor system during transients and accidents ranging from large-break and small-break LOCAs to anticipated operational transients involving plant control and protection systems. This code is organized into modules '

by components and functions to simulate the primary coolant system, secondary system, feedwater train, system controls, and core neutronicsE Special component models include  ;

. pumps, valves, heat structures, electric heaters, turbines, separators, and accumulators. 1 RELAPS/ MOD 2-B&W retains virtually all of the features of the original RELAPS/ MOD 2 code, while making certain modifications either to add predictive' capabilities of the constitutive models l

9910260038 991015' PDR l

TOPRP ENVBW l 1

> 1

_PDR

cc ..

y or to improve code ' execution.L More significant modifications are the addition of models and features to meet the 10 CFR Part 50; Appendix K requirements for emergency core cooling

^ system (ECCS) evaluation model (EM) calculations. The details of RELAPS/ MOD 2-B&W are described in Topical Report BAW-10164P-A, Revision 3.  !

RELAP5/ MOD 2-B&W uses a one-dimensional (axial), transient, two-fluid hydrodynamic model to calculate the flow of a steam-water two-phase mixture. This two-fluid model uses six field

equations (i.e., two phasic-continuity, momentum, and energy equations each), which provide the capability to calculate the characteristics of non-homogeneous, non-equilibrium flow. The

hydrodynamics model also contains options for invoking simpler models, such as homogeneous flow, thermal equilibrium, and frictionless flow models, which can be used independently or in combination. The system model is solved numerically using a semi-implicit finite difference technique 'The user can also select an option for solving the system model using a nearly implicit finite difference technique that allows for violation of the material Courant limit, and is suitable for steady-state, and slowly varying, quasi-steady-state transient calculations.

' The code uses a point-kinetics model with six delayed neutron groups to calculate reactor power as a function of time. . it contains provisions for fuel temperature,- moderator temperature, {

and density reactivity feedback.. Other reactivity feedbacks such as those caused by boron j concentration changes and tripped-rod reactivity are provided with input tables for generalized )

reactivity with respect to time.

.The constitutive models include models for defining flow regimes, and flow-regime-related models for calculating wall friction, interphase mass transfer, heat transfer, and drag force.

'Also included are a core structure heat transfer model and a fuel pin heat conduction model with a dynamic fuel cladding gap conductance model. The core heat transfer package can calculate heat transfer coefficients for various heat transfer regimes from single-phase convection, nucleate boiling, to post-critical heat flux (CHF) heat transfers.

~ Other special features include dynamic pressure loss models associated with abrupt area change for single-phase and two-phase flows, a centrifugal pump performance model with two-phase degradation effects, choked flow models with treatment for horizontal stratification, non-homogeneous two-phase flow, countercurrent flow models, crossflow jun:: tion, decay heat models, a fine mesh renodalization scheme for heat conduction, liquid entrainment, a motor valve model, a relief valve model, control system, and trip system.

The B&W-designed nuclear steam supply system (NSSS) is unique in that it employs once-through steam generators (OTSGs), in contrast to the U-tube recirculating steam generators (RSGs) in the PWRs designed by other vendors. The OTSG is a counterflow, single pass, tube and shell heat exchanger that produces superheated steam at a constant secondary pressure over the entire load range. The boiling heat transfer area and secondary inventory vary with )

, . load, requiring special modeling capabilities to properly predict plant response. For application

.of RELAP5/ MOD 2-B&W to the B&W-designed plants, Revision 2 of RELAP5/ MOD 2-B&W added the BWUMV CHF correlation, the Wilson model for determining interphase drag, and a  :

countercurrent flow limit model for performing small-break LOCA analyses. Revision 3 of 1 RELAP5/ MOD 2-B&W included enhancement to the EM fuel pin model, EM heat transfer model, )

and models to support use of the code for analyses of OTSG plants. These models included  ;

i l

i

p 1 4 -..

L h 3- l l

the Becker CHF correlation, the BWNT slug-drag model, the high auxiliary feedwater (AFW).

L model, and the Chen' nucleate boiling heat transfer coefficient void ramp.

3 The NRC has approved RELAP5/ MOD 2-B&W for LWR LOCA and non-LOCA transient . l analyses (Reference 2). The staff has also approved specific application of RELAPS/ MOD 2- I t

. B&W for performing LOCA and non-LOCA analyses on PWRs with RSGs (References 7 & 8)

and for performing LOCA analyses on B&W-designed PWRs (Reference 9). The purpose of l BAW-10193P is to obtain NRC approval to extend the application of RELAP5/ MOD 2-B&W for safety analyses of non-LOCA transients of the B&W-designed .PWRs.-

^

2.0 L EVALUATION

To support the use of RELAP5/ MOD 2-B&W for safety analyses of non-LOCA transients of.

B&W-designed PWRs, BAW-10193P presents the following benchmarks of the

RELAP5/ MOD 2-B&W calculations against various data and code predictions:

.. benchmarks of_OTSG test facility data to demonstrate adequacy of RELAPS/ MOD 2-B&W modeling of OTSG in predicting boiling length in the. steam generator (SG)

L

.. secondary side, and primary-to-secondary heat transfer during upset conditions, benchmarks of B&W-designed PWR plant data to demonstrate the ability of RELAP5/ MOD 2-B&W to predict the phenomena exhibited during non-LOCA events, and '

'a comparisons to the CADDS and TRAP 2 calculations of non-LOCA events to.

demonstrate the similarity in the predictions of the system and core power responses

between RELAP5/ MOD 2-B&W and these approved ccdes.

The sections that follow describe the staff evaluations of these benchmarks.

l

.g 2.1  : Benchmarks Against OTSG Test Data Section 4 of BAW-10193P discusses the benchmarks of RELAP5/ MOD 2-B&W against the 19-tube model OTSG tests performed at the Alliance Research Center Nuclear Steam Generator Test Facility. The 19-tube full-length model OTSG is a single-pass, counterflow, tube and shell heat exchanger. It can be used to simulate OTSGs with either an aspirator or

l. integral economizer, referred to herein as aspirator-OTSG (AOTSG) and integral economizer-

' OTSG (IEOTSG), respectively, . Primary inlet flow entered at the top of the SG, flowed downward through the tube bundle, and exited at the bottom. Secondary feedwater flow entered by way of an external downcomer, through the bottom of the tube bundle' boiled as it passed by the outside of the tube bundle, and exited at the top as superheated steam. When 1

run in the AOTSG mode, steam bled from the tube region, which simulated the' aspirator and raised the feedwater temperature to saturation conditions by mixing the water with steam. In ths IEOTSG mode', the steam bleed was closed and the subcooled feedwater entered at the bottom of the tube nest.-.

l o

7. - s m-L.

'4 , -4l-a

The following four sets of tests were simulated with RELAP5/ MOD 2-B&W

a two sets of steady-state tests performed in 1969 and 1971 to determine the thermal i performance of the AOTSG and IEOTSG, respectively, and i o 2 two sets of loss of feedwater flow (LOFW) tests from scaled full power conditions, performed in 1977, for the AOTSG and IEOTSG designs, respectively.

The steady-state OTSG tests were performed for a range from 20 to 100 percent of the scaled full power with primary pressure and inlet conditions, feedwater conditions, and secondary pressure held constant for each test.' The boiling lengths (dryout locations) as a function of 4 I scaled power level were det' ermined from ' primary tube and secondary-side thermocouples.

The AOTSG LOFW test was initiated from the scaled full-power conditions by simultaneously l

~ tripping the feedwater pump and closing the feedwater isolation valve. The SG was allowed to boil dry and then the feedwater was restarted. Secondary steam flow and temperature and

. primary outlet temperature were measured during the tests. The IEOTSG LOFW test had the same procedure, except for the closure of the downcomer isolation valve rather than the

- feedwater isolation valve. However, the test data suggested that the feedwater isolation valve was actually closed during the IEOTSG LOFW test instead, thus allowing feedwater to trickle into the tube region of the downcomer.: Therefore, as described in FTG's response to staff Question 2A (Ref.10), the boundary conditions used in the RELAP5/ MOD 2-B&W benchmark

, for the IEOTSG LOFW test contained an estimate'of the average rate of liquid displacement from the downcomer.i 2.1.1 : RELAPS/ MOD 2-B&W Model Description

- The RE _AP5/ MOD 2-B&W model of the 19-tube OTSG test facility utilized 11 axial volumes in j

'the primary tube region and in the secondary shell region. Primary-to-secondary heat transfer 1was modeled using 11 heat structures between the primary and secondary sides. The external downcomer was modeled with five axial control volumes, representing the piping from the i steam /feedwater mixing region to the tube bundle inlet. A feedwater aspiration was provided by a single junction component that connected the tube bundle region to the external downcomer.

A junction connection between the shell side of the heat exchanger and the control volume ,

representing the steam / water mixer was included.. Time-dependent volume and junction l components were used to set the primary- and secondary-side coolant inlet flowrates and temperatures.

1 The following benchmarks were performed with certain features available in RELAPS/ MOD 2-

B&W for the OTSG shell side

use'of a specific CHF correlation on the shell side of the tube heat structure to provide a better prediction of the dryout point in the OTSG, e use of the BWNT slug flow drag model with default multipliers (in RELAPS/ MOD 2-B&W) to reduce the interphase drag in the slug and annular-mist flow regimes, and i

p p.- .

1 p

t use of a multiplicative weighting factor to force the boiling suppression factor of the Chen nucleate boiling correlation to zero as the steam void fraction approaching 1.0.

These features, incorporated i.n Revision 3 of BAW-10164P-A,'have been approved by NRC ,

1

-(Reference 2).

2.1.2 - Compansons with Test Results

. I Tables 4-2 and 4-4, and Figures 4-4 and 4-8 of BAW-10193P compare the boiling lengths i above the lower tube sheet at various power levels predicted by RELAP5/ MOD 2-B&W to those ,

p - measured during the AOTSG and IEOTSG steady-state tests. They show good agreement I L between the RELAP5/ MOD 2-B&W predictions and the test data. Table 4-2 also shows the AOTSG steady-state boiling lengths calculated with the base RELAP5/ MOD 2 code, Cycle 36.05.

. These boiling Imgths calculated with the RELAP5/ MOD 2 base code, however, differed l significantly from the test data below 80 percent scaled power. FTG attributed the improvement f of the RELAP5-MOD 2-B&W calculations over the RELAPS/ MOD 2 base code to the use of the

specific CHF correlation. This CHF correlation was developed from heated rod bundle dryout )

data and will be used in future safety analyses.

Table 4-3 and Figures 4-6 and 4-7 of BAW-10193P present the benchmark results of the AOTSG LOFW test. The RELAP5/ MOD 2-B&W calculations of the'ir,;tial conditions agree well i with the measured values of the primary and secondary system fluid temperatures preceding the initiation of the LOFW test. For the LOFW test progressing, the RELAPS/ MOD 2-B&W calculations of steam flow agreed well with the data except for some sharp step changes in the

, calculated steam flow. These step changes occur as the secondary-side liquid-steam mixture l -level crosses the control volume boundaries, resulting in sudden changes in heat transfer as l the control volumes in the tube region systematically dry out and, later, refill. FTG stated that

. the addition of control volumes in the nucleate boiling region would decrease the magnitude of the step changes, but the number of steps would increase. The resulting predictions of heat transfer and primary outlet temperature would be approximately the same as the current prediction.

Figure 4-7 shows differences between the predicted and observed primary-side outlet temperatures. FTG attributed these differences to the heat capacity of the resistance thermal 1

detector (RTD) used in the test. The RTD heat capacity caused a lag in the measured temperature response such that the actual fluid temperatures were higher than the recorded

. values during heatup and lower than the recorded values during the refill. In response to staff question 2 (Ref.10), FTG performed a Laplace transform of the calculated primary-side outlet temperature to account for the RTD time constant The adjusted temperature results show good agreement with the actual RTD output, demonstrating that, if a thermal lag were applied to the code' prediction to account for RTD capacity, a good match would be obtained with the measured data This indicates that the RELAP5/ MOD 2-B&W prediction of transient heat transfer is in good agreement with the test, as demonstrated by the good agreement between the predicted and measured steam flow. The benchmarks demonstrate that RELAPS/ MOD 2-B&W with the control volume arrangement used in this benchmark can predict the shell-side l.

f- I

)

i L boiling length at various power levels, as well as primary-to-secondary heat transfer of the AOTSG design. i Table 4-5 and Figures 4-10 and 4-11 of BAW-10193P show the benchmark results of the I

IEOTSG LOFW transient. The RELAPS/ MOD 2-B&W predictions of the steam temperature and the primary outlet temperature agree well with the test data for the first 20 seconds of the transient. After 20 seconds, a deviation occun, when the predicted primary outlet temperature rapidly approaches the inlet temperature as the IEOTSG dries out, but the observed {

temperature remains much lower than the inlet temperature, indicating continued heat transfer. )

FTG stated that the continued heat transfer is not supported by the measured steam flow, and that the deviation between the predicted and observed temperatures is primarily due to the RTD heat capacity.

Figure 4-10 shows the code prediction of dryout time to be 2 seconds less than the observed time. This is because the code overpredicted the steam flow from the IEOTSG during the ,

dryout period. This results in early dryout and a low prediction of primary outlet temperature  ;

caused by an overprediction in heat transfer. The overprediction of steam flow arose as the j mixture level crossed the control volume boundaries. FTG concluded that the noding detail  !

used to predict the IEOTSG test data is too crude to produce an accurate result and that additional modefing is required. FTG, in its response to staff question 2 (Ref.10), indicated that the benchmark will not be refined at this time, because it has no plans to perform IEOTSG plant j safety analysis. FTG stated that before (or concurrent with) the licensing submittal of an l IEOTSG plant safety analysis, it will submit an updated benchmark of these test data for NRC i review. The staff concludes that RELAPS/ MOD 2-B&W may not be used for the safety analyses j of PWRs with IEOTSGs until FTG submits and the staff accepts an updated benchmark of the IEOTSG LOFW test case.

2.2 Benchmarks to Plant Data  ;

Section 5 of BAW-10193P describes benchmarks of RELAP5/ MOD 2-B&W against the data of the following four transient events or tests of B&W-designed PWR plants:

- Three-Mile-Island Unit 2 (TMI-2) LOFW event of March 26,1979 )

- Rancho Seco loss-of-ICS (integrated control system) power event of December 26, 1985

- Four-pump coastdown data from Oconee Unit 1 and Crystal River Unit 3

- Three-Mile-Island Unit 1 (TMI-1) natural circulation test of October 7,1985 The benchmarks were performed with a generic B&W lowered-loop 177 fuel assembly plant i model depicted in Figure 5-1 of BAW-10193P. The special RELAPS/ MOD 2-B&W features described in Section 2.1.1 of this safety evaluation (SE) were employed for the OTSG model. i The B&W high AFW model with AFW injection from high elevation location was used in conjunction with a two-region SG model. This allowed for the heat transfer in the tube region wetted by AFW to be calculated separately from the heat transfer in the tubes that are unwetted by AFW, l

In each benchmark, the initial conditions were set to the plant conditions that existed preceding the event or test, and the transient was simulated by imposing the plant boundary conditions,

E. . .

q

(?

h n

.-7--

l

- taken from the data recorded by the plant recall computer, or estimated using available data.

~ When required, the core decay heat input was calculated from the plant power history using L 1979 ANS (American Nuclear Society) 5.1 methodology. The predicted values of primary pressure, secondary pressure, primary system fluid temperatures and pressurizer level were compared with the plant values.

2.2.1 Benchmark of TMI-2 LOFW Event 2 l l

l - The TMl-2 LOFW event occurred as a result of the loss of both main feedwater (MFW) pumps

! while the plant was operating at 97 percent power. This event caused a coincident turbine trip, l- resulting in the secondary pressure increase and primary-to-secondary heat transfer reduction.

l- The mismatch between the core heat generation and SG heat removal caused the RCS pressure to increase, the power-operated relief valve (PORV) to open, and the reactor to trip on high RCS pressure. .During the post-trip RCS cooldown and contraction as the core power n dropped to the decay heat level, however, the PORV failed to close when the RCS pressure fell

. below the low-pressure setpoint, and the RCS continued to depressurize. Approximately

' 40 seconds into the event, the SG water level dropped to the low-level setpoint and the AFW control valves opened automatically to supply AFW to maintain minimum SG levels. However, the AFW block valves between the control valves and the SGs were closed, preventing the -

AFW from being delivered. Consequently, the SGs dried out, and the RCS began to reheat.

Eventually, at about 8 minutes into the event, the AFW was restored to the SGs. This i . benchmark was performed for the first 2 minutes of the event to focus on predicting the plant

!. - behavior during the LOFW period. .

l L Tables 5-1 through 5-6 and Figures 5-2 through 5-10 of BAW-10193P showed the benchmark results of the TMI-2 LOFW event. The code properly predicted the primary- and secondary-system pressurization rates following the turbine trip, and also predicts the timing of PORV lift

. and reactor trip; The calculations of the post-trip RCS pressure and temperatures, and SG l i

liquid levels and dryout time agreed with the plant data. The only significant deviation from the plant data occurred in the calculated pressurizer liquid level, which agreed with the plant data until the reactor trip. After the reactor trip, the plant data appeared to indicate a much greater outsurge than was predicted by the code. At 50 seconds into the event, the' pressurizer liquid

, level was calculated to be 191 inches compared to the recorded plant data of 159 inches. l

. However, FTG stated that the plant data were probably not reliable, and the level should be about 189 inches. This was based on experience with operating B&W plants that the

. pressurizer level should decrease by. 5.5 inches for every 1 *F decrease in average system p ' temperature. Therefore, the predicted pressurizer level was very close to the actual level. l

~

l The benchmark of the TMI-2 LOFW event shows that RELAPS/ MOD 2-B&W is appropriate for ]

l- analyzing overheating events on B&W-designed PWRs. '

2.2.2 L Benchmark of the Ranch Seco Loss of ICS Power Event q Lin 1985,' while operating at 76 percent power, the Rancho Seco plant experienced a loss of dc

' power for the integrated control system (ICS). The loss of ICS power caused a reduction of the MFW flow, and an increase of total steam flow from the opening of turbine bypass valves (TBVs) and atmospheric dump valves (ADVs), resulting in an RCS overcooling.

(i .

L j .-

t

P l

Tables 5-8 through 5-10 and Figure 5-11 of BAW-10193P showed the transient boundary conditions, including the core power, MFW flow, AFW flow, main steam safety valve (MSSV) flow, and the primary makeup and high-pressure injection (HPI) flows. Table 5-9 shows that,

, after an earlier termination, the AFW flow to SG-A was restored at 976 seconds into the event, l t following damage to the isolation valve for AFW flow to SG-A. In addition, the recorded AFW

{

flow to SG-A went off-scale for a portion of the transient; thus the AFW flow during this time j was estimated from system conditions and the AFW pump head / capacity curve. j The benchmark results are shown in Figures 5-12 through 5-18 and Tables 5-11 and 5-12 of i BAW-10195P, The loss of ICS power caused a reduction of the MFW flow, and an increase of total steam flow resulting from opening of the TBVs and ADVs. Initially, the RCS temperatures and pressure increased because the increased heat removal by the increased steam flow cannot overcome the heat removal reduction from the loss of MFW flow. At approximately 15 seconds into the event, the reactor tripped on high RCS pressure. The code predictions of this scenario were consistent with the plant computer data.

When the reactor tripped, the turbine also tripped, causing the secondary pressure to increase to the MSSV lift setpoint. The secondary pressure decreased subsequently as steam was relieved through the MSSVs. This caused the RCS to undergo a post-trip cooldown and l contraction as the reactor power fell to decay heat, emptying the pressurizer. As the SGs {

continued to depressurize from the open ADVs, full AFW flow was started, and the RCS continued to cool and depressurize to the threshold for engineered safety features actuation

. system (ESFAS) actuation at about 200 seconds. The actuation of the ESFAS initiated the HPl flow, and slowed the RCS contraction rate caused by the continued feeding and depressurization of the SGs. RCS depressurization continued until the fluid flashed in the reactor vessel upper head occurred at approximately 400 seconds. At approximately 500 seconds, the RCS pressure stabilized as the upper head liquid flashing and HPl addition compensated for the contraction of the primary system. The code prediction of the RCS temperatures and pressure, pressurizer level, and secondary pressure and SG levels agreed well with the recorded values during this period.

In the next period, as the HPl volumetric flow exceeded the RCS contraction rate, the RCS started to repressurize, ending the flashing in the upper head, and the pressurizer started to refill. By 700 seconds, the secondary relief valves were closed, and the AFW flow to SG-A was terminated. This ended the RCS overcooling, and caused the pressurizer level and the RCS pressure to increase at a greater rate. At 976 seconds, however, the SG-A AFW isolation valve was damaged, resulting in a restoration of the AFW flow to SG-A. Consequently, the RCS cooldown resumed, and the RCS contraction rate increased, so that the RCS pressure and the pressurizer level stabilized. At 1150 seconds, pressurizer spray flow was actuated to decrease RCS pressure. At about 1550 seconds, ICS power was restored, and all ICS demand signals were reduced to zero percent, ending the AFW flow and overcooling of the RCS.

The code underpredicted the RCS repressurization, which FTG attributed to the uncertainty in l measured HPl flows that were input in the benchmark calculation. In response to staff

! question 4 (Ref.10), FTG reanalyzed RCS pressure response in the loss of ICS power event L using an increased HPl flow. The results demonstrate that an increase in HPI flow has little effect on the pressure prediction during the depressurization because of the dominant effect of the RCS contraction caused by overcooling. However, once the cooling of the RCS was ]

n - ,

. . j

. i

. .t.- . ,

l y >

I

!J reduced, the effect of HPl flow on the pressure and pressurizer level increased significantly. f

' The result provided also agreed more closely with the data.

l DThe code prediction of secondary pressure during the entire transient is excellent. The code

, predicted secondary liquid levels very well, given the uncertainties in the AFW flow estimate.

The predicted RCS temperatures, RCS pressure, and pressurizer level matched up well with i _ the recorded values.. This benchmark demonstrates REl AP5/ MOD 2-B&W's capability for j

. _ analyzing secondary-system-initiated events. )

l 2.2.31 Benchmark of Flow Coastdown' Data Four-pump coastdown tests were performed from hot, full-pressure, zero-power conditions during the startup tests at Oconee Unit 1 and Crystal River Unit 3, both of which are of the E - B&W. lowered-loop 177-FA design. RELAP5/ MOD 2-B&W was benchmarked against the flow -

and pump speed data recorded from these tests. The results of these comparisons, shown in

. Figure 5-19 of BAW-10193P, shows that the predicted pump response essentially overlays the L _ plant data. This demonstrates that the pump inertia, pump frictional torque values; and reactor L ~ coolant loop flow resistance input to the RELAP5/ MOD 2-B&W plant model yield an accurate l -calculation of the system flow rate.

L 2.2.4 Benchmark of TMl-1 Natural Circulation Test

, 1 A low-power natural-circulation (NC) test was conducted on October 7,1985, at the TMI-1 plant L

4 to demonstrate the NC heat removal capability of B&W-designed PWRs. The NC test was j'

initiated by tripping the reactor coolant pumps while the unit was operating at approximately 1

! 3 percent power with full RCS flow and SG liquid levels controlled to the NC setpoints. 1 Throughout the NC test, SG pressures were controlled to within 11 psi of the initial value, and i the SG levels were maintained at about 50 percent on the operating range by a control system I

~

using AFW. The RCS pressure was regulated by the intermittent us s of the pressurizer spray j ano adjusting letdown flow during pump coastdown, and.by the use of the pressurizer heaters l and letdown flow after NC was established. These test conditions were used as boundary  !

conditions for the RELAPS/ MOD 2-B&W benchmarks.-

1The B&W 177-FA lowered-loop plant model was used for the benchmark analysis. There are a few differences in the boundary condition inputs to the code relative to the test data. The

reactor power input to the code, shown in Table 5-13 of BAW-10193P, was equal to the measured power multiplied by a correction factor of 1.12. The correction factor of 1.12 (see FTG's response to staff question 5, Reference 10, for the derivation) was necessary because

' the out-of-core neutron detectors used for power measurement were calibrated at a temperature higher than the reactor vessel downcomer fluid temperature during the test. The inputs of the setpoints of pressurizer heaters 3 and 4, shown in Table 5-14 of BAW-10193P,

- were slightly lower than the plant data.' The SG level boundary condition in the benchmark, shown in Figure 5-20 of BAW-10193P, was maintained near the 50 percent level by a simplified control system as opposed to the larger variation observed in the test data. However, the  ;

~effects of these differences on the predictions were small (see FTG's response to staff question 75, Reference 10).

L

w .

s..

m Thh benchmark results are shown in Table 5-16 and Figures 5-21 through 5-23 of

' BAW-10193P. The hot leg temperature, RCS pressure, and pressurizer liquid level rose during l the RCS pump coastdown, and decreased subsequently with the decline in core power until they stabilizedi The code predictions of these parameters agreed well with the test data, although they were overpredicted during the pump coastdown period. The equilibrium RCS fluid temperature difference was calculated to be 34*F compared to 35*F in the test.

jTherefore, the calculated NC flow was within 3 percent of the test result. . BAW-10193P

, attributed the slightly higher NC flow predicted by the code to the plant model used in the analysis. This plant model had only three axial control volumes for the core, so that the core thermal center in the model was 2 feet below the mid-core elevation. Consequently, the

. SG-to-core' thermal center differace in the model was greater than in the plant, yielding an -

RCS NC flow slightly greater than ine test. . However, the code properly predicted the SG

. thermal center during NC, indicating accurate calculations of the heat transfer in the tube region

. wetted by AFW, and the heat transfer to the secondary pool. This benchmark demonstrates -

that RELAPS/ MOD 2-8&W is suitable for analyzing the response of B&W-designed PWRs for NC events.

- 2.3 Benchmark of RELAP5/ MOD 2-B&W Against CADDS and TRAP 2 Section 6 of BAW-10193P describes comparisons of the RELAPS/ MOD 2-B&W predictions of the control rod withdrawal and the main steam line break (MSLB) transients against the predictions of the NRC-approved codes,'CADDS and TRAP 2, respectively. CADDS has been used for the analyses of the primary system response of B&W-designed PWRs to such transients as reactivity insertion and loss-of-primGry-flow events.' TRAP 2 has been used for the calculations of the core power and system responses to the secondary-system-initiated events, such as MSLB, turbine trip, loss of feedwater, and steam generator tube rupture accidents.

The intent of the RELAP5/ MOD 2-B&W comparisons with the CADDS and TRAP 2 predictions is to show that, with the same conservative initial and boundary conditions used in the safety analyses, RELAP5/ MOD 2-B&W will predict system and core power responses of various non-

, LOCA events similar to those predicted by CADDS and TRAP 2.-

2.3.1 ' RELAP5/ MOD 2-B&W-CADDS Comoarison of Startuo Events

' The following three startup events of control rod withdrawal from a low power condition were

, analyzed with both the RELAPS/ MOD 2-B&W and CADDS codes for comparisons:

the withdrawal of a single control bank from hot zero power condition,

+~

the withdrawal of all control rods from hot zero power condition, and an intermediate rod withdrawal with a spectrum of reactivity insertion rates.

These. control rod withdrawal events resulted in a reactivity insertion into the core and an RCS

. overpressurization. Different reactivity insertion rates were used to simulate these cases.

The'CADDS analyses of the rod withdrawal events used lumped single-loop modeling of the

- RCS consisting of the hot leg, SG,' cold leg, reactor core and bypass, and pressurizer. The RELAPS/ MOD 2-B&W analyses used a simplified generic B&W 177-FA plant model shown in Figure 6-1 of BAW-10193P, which is a more detailed representation of the plant than the CADDS model. The 'model consisted of two hot legs and SGs, four cold legs with reactor

g. * ,

'- I 7 f  ;

p- 1

(

L w 11. j L coolant pumps, the reactor core,'and the pressurizer. The RCS pumps were explicitly modeled  ;

L as compared to the CADDS model that used a specific flow rate and loop time ~ delay input. l

However, the following features of the RELAP5/ MOD 2-B&W model were purposely altered to match the CADDS model formulation' '

l i e l deletion of the upper reactor vessel mad region, l

exclusion of all heat structures but the fuel pin, SG tubing, and pressurizer shell metal,-

a I

use of a single control volume and heat structure modeled for the core region, and use of a constant heat demand to model the SG heat transfer.

L I

l Table 6-1 of BAW-10193P compares the RCS initial and boundary conditions between the. l l CADDS and RELAP5/ MOD 2-B&W calculations. The reactor core kinetics parameters and the reactor trip setpoints are shown in Table 6-2. l The comparisons between the REl.AP5/ MOD 2-B&W and CADDS predictions of the " single i i control bank withdrawal" and "all-rods withdrawal" cases are shown in Tables 6-3 and 6-4, and

- Figures 6-2 through 6-15 of BAW-10193PJ In both cases, the reactor power increased as a ,

, result of reactivity insertion from the control rod withdrawal. The heat addition to the core caused the increases of the RCS pressure and temperature. ~ln the single-bank-withdrawal y case, Doppler reactivity feedback terminated the power excursion before a high flux trip was reached, and the reactor subsequently tripped on high RCS pressure. In the all-rods- -

- withdrawal case; the reactor tripped on the high' neutron flux.- After the reactor trip, control rod  ;

insertion sharply reduced the reactor power and the pressurization rate. The overpressure ended after the lift of the pressurizer safety valves.

In both cases, the RELAP5/ MOD 2-B&W's predictions of the neutron power, thermal power, and l' fuel temperature responses agreed well with the CADDS predictions, except for the time delays of the sharp declines of the power and fuel temperature for the single-bank-withdrawal case.

These delays are due to the slower system pressurization predicted by RELAPS/ MOD 2-B&W

' that resulted in the reactor trip on high RCS pressure to be later than that predicted by CADDS.

1The CADDS prediction of pressurizer pressure response closely' approximated an adiabatic

- compression of the pressurizer steam region, whereas the RELAP5/ MOD 2-B&W prediction

! _. considered the real effects of condensation at the steam-liquid interface and on the surface of the pressurizer shell metal. The pressurizer model difference resulted in later reactor trip with attendant greater peak thermal power, but lower peak RCS pressure than predicted by CADDS. ,

b After the reactor trip, the rates of RCS fluid expansion, pressurizer insurge, and pressurization

, i reduced as the core power decreased. Although the CADDS prediction showed that the RCS L  ; pressurization rate remained iJnchanged after reactor trip until the lift of pressurizer safety

valves, the RELAPS/ MOD 2-B&W predictions of these sequences of events are more consistent

. with the reduction'of thermal power after reactor trip.

Intermediate rod withdrawal with a spectrum of reactivity insertion rates was also analyzed with l both RELAPS/ MOD 2-B&W and CADDS 'codesi Figures 6-16 and 6-17 of BAW-10193P show the comparisons of peak neutron and thermal powers as a function of reactivity insertion rate L predicted by the two codes. These figures showed good agreement between the two codes, j i

m-l

F V

The overall comparisons demonstrate RELAPS/ MOD 2-B&W to be suitable for analyzing the system response during reactivity transients on B&W-designed PWRs.

2.3.2 - RELAPS/ MOD 2-B&W-TRAP 2 Comparison of Main Steam Line Break The RELAPS/ MOD 2-B&W and the TRAP 2 calculations were compared for two MSLB accidents: a 6.28-square-foot double-ended rupture and a 2.0-square-foot split break of a steam line in the steam generator B (SG-B). The RELAP5/ MOD 2-B&W MSLB analyses used the generic large-detail B&W lowered loop 177-FA plant model, shown in Figure 6-18 of BAW-10193P. This model was altered to be consistent with the TRAP 2 model by:

deletion of the upper reactor vessel head region, exclusion of all heat structures except for the fuel pin and SG tubing,

+ addition of secondary steam and feedwater piping, and a feedwater pump simulation,

• use of the same break geometry and critical flow model, AFW flow table for the unaffected SG, and HPl flow versus pressure table.

The reactor core neutronics parameters and reactor protection trip setpoints are shown in Table 6-6, and the ESFAS setpoints and delay times are shown in Table 6-7 of BAW-10193P.

Both the double-ended rupture and split-break cases were initiated from the same plant conditions. Table 6-5 of BAW-10193P shows the initial conditions . established by the RELAPS/ MOD 2-B&W and TRAP 2 codes. The comparisons between the RELAP5/ MOD 2-B&W and TRAP 2 transient analysis results are shown in Table 6-8 and Figures 6-19 through 6-28 of BAW-10193P for the double-ended rupture case, and in Table 6-9 and Figures 6-29 through  !

6-38 for the split-break case.

The plant system responses to the double-ended rupture and split break cases were very similar. Each MSLB caused decreases in the SG pressure and saturation temperature, an increase in primary-to-secondary heat transfer, and attendant decreases in the RCS pressure I and temperature. Because of negative moderator temperature coefficient (MTC), the RCS cooling caused the core fission power to increase. In the double-ended rupture case, the RCS i depressurization proceeded at a faster rate, resulting in the reactor trip on low RCS pressure.

In the split-break case, the power surge resulted in a reactor trip on high neutron flux at a later time than the double-ended break. After the reactor trip, the RCS pressure continued to decrease because of overcooling, resulting in the actuation of the ESFAS on low RCS pressure. The ESFAS actuation initiated closures of the main steam isolation valve (MSIV) and main feedwater isolation valve (MFIV), and actuations of the AFW flow to the unaffected SG and the HPl flow with respective time delays. Following the MSIV closure, the unaffected SG repressurized and became a heat source, while the affected SG continued to depressurize at a faster rate as it became the sole source of the break flow. After the MFIV closure, the unaffected SG continued to fill with the AFW flow. Meanwhile, the affected SG started to dry i out, which, however, was delayed as liquid in the feedwater pipe began to flash, pushing liquid l into the SG.

The RELAPS/ MOD 2-B&W predictions of the break flow, RCS pressure and temperature, secondary pressure, core reactivity, and power agreed well with those of TRAP 2. There are differences between the calculations of the two codes in the SG secondary mass inventory, cold leg temperature, total reactivity, and neutron power after the MFIV closure. BAW-10193P 1

l

a: , ,

f.- )

attributed these prediction ' differences to t_he' differences between the two codes in the calculation of steam-liquid phase slip in the secondary system. The TRAP 2 bubble rise velocity

, inputs on the secondary side were typically a constant 0.5 foot per second, which means that

the control volume fluid conditions were effectively assumed homogeneous. On the other hand, RELAP5/ MOD 2-B&W used an NRC-approved mechanistic model (B&W slug flow drag model) to calculate the steam-liquid phase slip in the SG and feedwater piping control volumes, thus -

- providing more realistic calculations of the_SG boiling length and dryout. Because RELAP5/ MOD 2-B&W calculated higher phase slip in the SG,' the primary-secondary heat L transfer was lower as the break flow continued through the break, resulting in higher cold leg ,

temperature prediction and earlier minimum cold leg temperature than the TRAP 2 predictions.

Higher phase slip in the feedwater piping control volume predicted by RELAP5/ MOD 2-B&W

. also resulted in less liquid transported into the SG as the piping liquid flashed. Therefore, it

calculated earlier SG dryout and a shorter boiling length than did TRAP 2, resulting in less heat

, . transferc Consequently, RELAP5/ MOD 2-B&W predicted higher RCS temperature, and less severe core reactivity and powerjust preceding SG dryout than the TRAP 2 predictions.

The RELAP5/ MOD 2-B&W and TRAP 2 comparisons of the MSLB events demonstrate that, given ccervative initial and boundary conditions, RELAP5/ MOD 2-B&W produces conservative s results, similar to those predicted by TRAP 2.

2.4 _ REl.AP5/ MOD 2-B&W Non-LOCA Safety Analysis Methodology in response to staff question 1'(Ref.10), FTG_ added an appendix to BAW-10193P, "Non-LOCA LAnalysis Methodology for B&W Designed Plants." The appendix gives guidance on using -

LRELAP5/ MOD 2-8&W for safety analyses of various transients and accidents, The guidance covers (1)'the NSSS model noding details, (2) the options for the constitutive models and correlations, and (3) determining input assumptions, initial conditions, and boundary conditions. .

'In general, FTG intends to conform to the accident analysis methods and licensing bases

described in the Updated Final Safety Analysis Reports (UFSARs) of the B&W-designed plants, except for the use of RELAP5/ MOD 2-B&W in place of CADDS and TRAP 2, and other exceptions noted in the appendix.

The staff reviewed the appendix and finds that the guidance is consistent with those l methodologies chosen for the benchmark analyses. Figures A.1 and A.2 in the appendix show

+

the noding details of (1) the large detail plant model for the analyses of those transients

' dominated by the performance of the SG or secondary plant systems, or both, and (2) the

reduced detail plant model for those transients dominated by the core response, respectively.

L Table A.1 in the. appendix specifies which plant models should be used for various transients, r . These noding' details are consistent with those used in the benchmarks. The user input options

!' for the constitutive models and correlations for the safety analyses described in Section A.2 of J the' appendix are consistent with those used in the benchmark analyses.-

? Section'A.3 of the appendix presents guidance on safety analysis input assumptions regarding

/ the initial conditions, bounds.., conditions, reactivity coefficients, effects of control system, loss of offsite power, and single failure assumptions. These assumptions are consistent with those H described in the UFSARs, except for (1) the initial condition of pressurizer level and (2) the core -

decay heat calculation." Although the guidance still follows the original UFSAR safety analyses

- in setting the ' initial pressurizer level to the nominal value for most transients, it also advises r

n si,3 w <-

, b, '

y .a

setting the initial pressurizer level 16 a value greater than or equal to nominal value plus measurement uncertainty for those events that cause an increase in pressutizer liquid level or

, iRCS pressure. This assumption is 'more conservative than the nominal value used in UFSARs,

, f and is therefore acceptable. '

'Regarding the core decay heat, the guidance advises the use of the ANS 1971 decay heat

. standard plus actinide decay for analyzing non-LOCA transients. This is different from the -

original UFSAR analyses of using'1.2 times the ANS 1971 decay heat standard, or later, the use of the ANS 1979 decay heat standard. The ANS 1979 standard more accurately predicts

- core decay heat following a reactor trip, and presents a method to conservatively apply f uncertainties; but the ANS .1971 decay heat model is conservative and does not require verification for every reload core design. The 1971 decay heat standard plus heavy isotope actinide contribution has been used in Sequoyah Units 1 and 2 (Ref.11). ' For the design calculations of B&W designed PWRs including LOCA analyses, the' decay heat contribution of the heavy isotope actinides has been calculated with a heavy isotope decay heat.model, _

referred to as the B&W heavy isotope model (Ref.12). Figures A.3 and A.4 of the appendix show that the ANS 197.1 decay heat standard plus the B&W heavy isotope actinide contribution

~ bound the ANS 1979 decay heat plus 2 sigma uncertainty for a wide variation in fuel assembly fenrichment and burnup. Therefore, this model will be used for all non-LOCA transients of the B&W-designed PWRs, except for MSLB. For.MSLB at end of cycle for.which it is desired to minimize heat input to the RCS,0.9 times the ANS 1971. decay heat standard will be used. The .

staff finds this decay. heat calculation for non-LOCA safety analyses acceptable.

3.0 REFERENCES

1. Letter from J.' 'H.. Tayk,. (BWNT) to U. S. Nuclear Regulatory Commission, " Submittal Of

. Topical Report BAW-10193P, RELAP5/ MOD 2-B&W for Safety Analysis of B&W-

. Designed Pressurized Water Reactors, August 1995," JHT/95-85, August 14,1995.

. 2. . BAW-10164P-A, Rev. 3, "RELAP5/ MOD 2-B&W - An Advanced Computer Program for Light Water Reactor LOCA and Non-LOCA Transient Analysis," July 1996.

3.- BAW-100'38, Rev.1, "CADDS - Computer Application to Direct Simulation of Transients

' in PWRs With or Without Scram," January 1978.

4. - BAW-10128, " TRAP 2 - FORTRAN Program for Digital Simulation of the Transient

- Behavior of the Once-Through Steam Generator And Associated Reactor Coolant System," August 1976.-

5. BAW-10156-A, Rev.1, "LYNXT - Core Transient Thermal-Hydraulic Program," March  !

1991J

6. .V. H. Ransom, et al, "RELAP5/ MOD 2 Code Manual," NUREG/CR-4312, EGG-2396,

August 1985; l

1 '

7. BAW-10188P-A, Rev. 3, "RSG LOCA - B&W Loss-of Coolant Accident Evaluation Model I for Recirculating Steam Generator Plants," December 1996.  !

l v ,

i' R 1

7 -

. 1 l

E I l

8. BAW-10169P, "RSG Plant Safety Analysis," October 1987.

9. BAW-10192P-A, "BWNT LOCA - BWNT Loss-of-Coolant Accident Evaluation Model for f Once-Through Steam Generator Plants," June 1998. '

10.'. Letter from J. J. Kelly (Framatome Technologies) to U. S. Nuclear Regulatory j Commission, " Response to NRC's Request for Additional information Regarding Topical I Report BAW-10193P," May 4,1999, FTI-99-1523.

11. Letter from J. H. Taylor (Framatome Cogema Fuels) to U.S. Nuclear Regulatory Commission, " Submittal of Topical Report BAW-10220P, ' Mark-BW Fuel Assembly Application for Sequoyah Nuclear Units 1 and 2,' March 1996," March 5,1996, JHT/96-20.

l 12. Letter from J. J. Cudlin (Framatome Technologies) to U.S. Nuclear Regulatory l

Commission, " Additional Information on the Actinide Decay Heat Model in BAW-10193,"

August 30,1999, FTI-99-2750.

4.0 CONCLUSION

l The staff has reviewed the benchmarks of the RELAP5/ MOD 2-B&W code against various data

~ from the OTSG tests and plant transients, and the calculations of the approved safety analysis codes. The good agreements of these benchmarks demonstrated acceptability of RELAP5/ MOD 2-8&W for performing safety analyses of non-LOCA events for the B&W-designed PWRs.

As discussed in_ Section 2.1.2 of this SE, the noding detail used for the benchmark of the IEOTSG test data was not sufficient to produce an accurate prediction of the primary-to-secondary heat transfer for the IEOTSG. FTG stated that before (or concurrent with) the licensing submittal of an IEOTSG plant safety analysis, an updated benchmark of the IEOTSG i test data will be submitted for NRC review. Therefore, RELAP5/ MOD 2-B&W may not be used l

for analyses of PWRs with the IEOTSGs until FTG submits and the staff accepts an updated benchmark of the IEOTSG LOFW test.

(

Principal Contributor: Y. Hsii Date: October 15, 1999 b

i I

r L )

E

• J. J. Kelly October 15, 1999 the continued effective applicability of the report without revision of the respective documentation.

Sincerelv.

ORIG. 3IGNED BY Stewart N. Bailey, Project Manager, Section 2 Project Directorate Ill Division of Licensing Project Management Office of Nuclear Reactor Regulation Project No. 6w DISTRIBUTION:

Central Files

Enclosure:

Safety Evaluation PUBLIC PDIV-2 Reading cc w/ encl: SRichards Mr. Michael Schoppman JWermiel Licensing Manager YHsii Framatome Technologies, Inc. OGC 1700 Rockville Pike, Suite 525 ACRS Rockville, MD 20852-1631

• concur by memo To receive a copy of this document, indicate "C" in the box OFFICE PDill/PM C PDIV-2/LA L SRXB/BC* PDIV-2/SC NAME SBailey EPdl'of F JWermiel SDembek p DATE 10 / /4 /99 10 / t3 /99 9 / 14 /99 10// 1 /99 DOCUMENT NAME: G:\PDIV-2\B&Wog\SE93346.wpd OFFICIAL RECORD COPY I: _