ML20133E265
ML20133E265 | |
Person / Time | |
---|---|
Site: | Arkansas Nuclear ![]() |
Issue date: | 07/03/1985 |
From: | NRC |
To: | |
Shared Package | |
ML20133E263 | List: |
References | |
TASK-2.K.3.30, TASK-TM TAC-45817, TAC-45859, NUDOCS 8507220441 | |
Download: ML20133E265 (51) | |
Text
-s- a s - 2 - a -,_ ,. --a- - easa- =-- - -.m e
g .4
, e t .
1 s l
.l SAFETY EV&LUATION REPORT FCR THE BABC0CK AND WILC0X OWNERS GROUP SMALL BREAK LOSS-0F-COOLANT ACCIDENT EVALUATION MODEL, CRAFT 2 (REY. 3)
(BAW-10092P, REY. 3 AND BAW-10154) m
.i
-l i
l 850703 koj72hK05000313 P
PDR ,
i Table of Contents
.P,gte I. Background ...................................................... 1 II. Introduction .................................................... 2 III. Evaluation of CRAFT 2 Small-Break LOCA Models .................... 6 III.1 Condensation Heat Transfer and Noncondensable Gases ...... 10 III.1.a Steam Generator Model ........................... 10 III.1.b Pressurizer Model ............................... 12 III.2 Nonequilibrium Effects ................................... 13
! III.2.a CoreFloodTank(CFT)Model..................... 13 ..
III.2.b Pressurizer Model . .. . *. .. . . . . . . . . . . . . . . . . . . . . . 14 III.3 Hot Leg Phase Separation ................................. 14 III.4 Steam Generator Heat Transfer ............................ 18 III.5 Systems Verification and Other Experimental Verification . 18 i
III.5.a Vent Valve Model and Countercurrent Flow Model .. 19 III.5.b Pressurizer and Surge Line Model ................ 22 III.5.c Steam Generator and AFW Models .................. 22 III.5.d Two-Phase Flow and Phase Separation Models ...... 24 III.S.e Core Heat Transfer Model ........................ 25 III.5.f Integral System Benchmarks ...................... 26 III.5.g B&W Integral System Test Program'....'............ 28 d III.6 Flow Regimes ............................................. 30 III.7 Core Steam Cooling ....................................... 31 i
III.8 Metal Heat ............................................... 32 III.9 B re a k F l ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 IV. Concerns in NUREG-0565 .......................................... 34 V. Concerns in NUREG-0623 .......................................... 41 VI.
VII.
Conclusions ..................................................... 42 References ...................................................... R-1 List of Tables N
- 1. Category Classifications ........................................ 4 l i List of Figures i
- 1. Small-Break LOCA Code Interface ................................. 7 l 2. Comparison of CRAFT 2 and RELAPS ................................. 21 i
}
, .-e-. - . . . -_._ -_- -._,e - - - , , - - - -
, - . - - , , - - , . . - . . . , - - . . --,e.-.,,_._ . -.
- - - . . _ _- ~_. - _. . _ _ _ . _ - - . - _ . _ _ - _ . _ _ _ ____
u
- SAFETY EVALUATION REPORT FOR THE
' BABCOCK AND WILCOX OWNERS GROUP i SMALL BREAK LOSS-OF-COOLANT ACCIDENT 1
EVALUATION MODEL, CRAFT 2 (REV. 3)
(BAW-10092P, REV. 3 AND BAW-10154) 1 4
i I. BACKGROUND i
~
Following the accident at TMI-2, the flulletins and Orders Task Force was formed j within the NRC Office of Nuclear Reactor Regulation. The Task Force was j charged, in part, with reviewing the analytical predictions of feedwater tran- -
- sients and small break LOCAs to ensure the continued safe operation of all op-c*ating reactors, and with the determination of the acceptability of operator emergency guidelines. As a result of these reviews, the Task Force concluded that, while there were no apparent safety concerns, additional system verifica-tion of the small-break LOCA model (as required by II.4 of Appendix K to
);
10 CFR 50) was needed in certain areas. These improvements and concerns, as
! they applied to each LWR vendor's model, were documented in the various Task '
j Force reports for each LWR vendor. The review of the B&W small-break LOCA mod-
! 01 was documented in NUREG-0565, " Generic Evaluation of Small Break Loss-of-Crolant Accident Behavior in Babcock & Wilcox Designed 177-FA Operating j Plants" (January 1980). The review of the reactor coolant pump model was docu-j mented in NUREG-0623, " Generic Assessment of Delayed Reactor Coolant Pump Trip l
During Small Break Loss-of-Coolant Accidents in Pressurized Water Reactors" j
(November 1979). On October 31, 1980, the NRC issued NUREG-0737, "Clarifica-tion of TMI Action Plan Requirements." Included in NUREG-0737 is the require-ment for B&W licensees to review NUREG-0565 and -0623 and develop a program that addresses the NRC concerns therein. After a meeting between the 177-FA i
Owners Group and the NRC, the B&WOG instituted a Small-Break LOCA Methods Pro-l gram to address the requirements of NUREG-0737,Section II.K.3.30, as they were i
I identified by the staff in the meeting of December 16, 1980.
I i
05/10/85 1 -
_ _ _ . ~ . _ _ _ _ - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - ~~ ~~ ~ ~ ~ ~ ~ ~ ~
1 There were nine major areas of concern identified by the staff at the j December 16, 1980 meeting. They are:
j 1. Condensation Heat Transfer and Noncondensable Gases
- 2. Non-Equilibrium Effects
- 3. Hot-Leg Phase Separation
- 4. Steam Generator Heat Transfer l 5. Systems Verification and other Experimental Data i 6. Flow Regimes
- - 7. Core Steam Cooling
- 8. Metal Heat
- 9. Break Flow - - -
{
~
l.
Each of these concerns will be addressed in the body of this Safety Evaluation #
Report.
- i. \
l 1
II. INTRODUCTION .
} !
I
{ The B&WOG has submitted two topical reports to the NRC in response to the i NUREG-0737 (Reference 1) concerns. These are BAW-10092P, Revision 3 1
j (Reference 2) and BAW-10154 (Reference 3). A third report, for core mixture .
level analysis (F0AM2, BAW-10155), was also submitted and reviewed elsewhere, - I 3
see Section III.7 below. i i
BAW-10154 describes the features of B&W's small-break LOCA emergency core cool-l ing system (ECCS) evaluation model and is applicable to all current B&W nuclear :
I steam systems. I j l BAW-10092P Rev. 3 describes the CRAFT 2 computer program. In particular Appendix I of the CRAFT 2 report addresses the new features of the CRAFT 2 models for small-break LOCA analyses.
l At present, B&W's nuclear steam plants can be divided into three major 1
) categories:
I I
l 05/10/85 2 B&W CRAFT 2 SER i !
,, 1. 177-fuel assembly plants with lowered-loop arrangement.
- 2. 177-fuel a'ssembly plants with raised-loop arrangement.
- 3. 205-fuel assembly plants There are no significant design differences between the NSSs and ECCSs in each category. Table 1 lists the current B&W plants in each category. The plants in these categories are described as follows:
Category 1 - The plants in this category are generally referred to as the Oconee type. They are characterized by their loop arrangement, in which .the cnce-thrcugh steam generators are at a low elevation relative to the reactor vessel. These plants have eight internal vent valves and utilize the Mark 8 (15 x 15) fuel assambly. -
.).
.W '
. 7f; L Category 2 - The design is essentially identical to Category 1 except that the l
steam generators are raised in relation to the reactor vessel. The pump suc-tion leg is shorter for these plants due to the raised configuration of the
- l steam generators. Also, there are only four vent valves in these plants. This,. -
reduction in the number of vent valves is factored into the model as a reduced l v:nt valve flow area. TheHPIsystemcomprisesiow-headHPIpumps. There is '
only one plant of this design, Davis-Besse 1.
I Cateoory 3 - These plants have the raised-loop arrangement of the Category 2 plants but are larger (more fuel assemblies) and have eight internal vent valves. The Category 3 plants employ the Mark C fuel assembly instead of the Mark B. Currently the only U.S. plant of this design is Bellefonte. '
! The small break LOCA evaluation model described in this report is applicable to all three plant categories.
1 The CRAFT 2 computer program was developed by B&W to study the transient behav-ior of a Nuclear Steam Supply System undergoing a loss-of-coolant accident i
(LOCA). The program solves the conservation equations for mass and energy, the continuity equation, and the equation of state for water.
I i
l l
05/10/85 3 -
! I I
~_ _ _ _ . _ _ _ _ . _ _ , _ . _ _ _ _ _ _ , . - . . _ _ _ _ _ _ _ _ _ _ _ _
The CRAFT 2 program permits the user to select the nodal representation that re-sults in the best finite differencing of the fluid system to be analyzed. The program then solves the conservation equations for each node and the momentum cquation for each flow path between nodes. CRAFT 2 utilizes explicit solution techniques to analyze the transients. Components with different thermal-hydraulic characteristics must be simulated as different nodes.
Table 1 Category Classifications Category Plant Name Docket No. II.K.3.30 OR TACS 1 Oconee 1 -
50-269 45845 ,
Oconee 2 ,
h, 50-270 * ' 45846 Oconee 3 , ... 4 4 ,,50-287. ..
. 45847 Three Mile Island l'.J,.,f 50-289 . ,. 48286 Three Mile Island 2 - 50-320 -" ** -
Crystal River 3 50-302 45815 Arkansas Nuclear One 50-313 45803 Rancho Seco 50-312 45859 Midland 1 50-330 -
Midland 2 50-329 -
2 Davis-8 esse 1 50-346 45817
~
3 Bellefonte Unit 1 50-438 -
Bellefonte Unit 2 50-439 -
WNP-1 50-460 - -
WNP-2 50-513 -
e
{
- l. .
CRAFT 2 contains flexible models of all major Nuclear Steam Supply System com-
!, ponents 'Various options as well as user input parameters enable the program j to model the reactor core, reactor coolant pumps, steam generators, and con-necting piping in any configuration and operating mode desired. The diversity j! l of the models also allow the program to accurately model any themal-hydraulic l system containing similar components. l
\
i i The CRAFT 2 computer program has been previously reviewed by the NRC and was
) found to be in conformance with 10 CFR 50 Appendix K. The purpose of this re-port is to document the NRC review and findings of the new information provided i
by the B&WOG in response to item II.K.3.30 of NUREG-0737. This review is lim-
} (
ited to those new models related to small-break LOCA Evaluation Model analyses
! cnd to information provided to justify the new models or to demonstrate the !
lf - conservative aspects of CRAFT 2 for these analyses, as discussed in Section III j of this report.
I l With respect to LOCA analyses, a break is termed a "small-break" when its l cross-sectional area is 0.5 ft8 or less. Past experience with studies of small ,
breaks has shown that the large break concepts of bypass and reflood do not ap-ply to breaks of this size. A brief description of the behavior of small i breaks will be valuable in understanding the evaluation technique. A small ,
break accident involves a rather slow, non-violent system depressurization.
Flow conditions within the reactor coolant system change gradually and smooth-l
- ly. Temperature and pressure gradients between regions tend to be small. The !
! lack of agitation allows partial phase separation of steam and water and, in J some situations, countercurrent flow. Rather than the distinct blowdown and
) reflood phases associated with large breaks, small breaks have a smooth transi-tion from a period of relatively high core flow to one of relatively quiescent
- conditions. During the early phase, heat transfer in the core is j flow-controlled from natural circulation flow and is adequate to keep the clad-
- ding cool. Later, during the quiescent period, a two phase froth level devel- ;
ops in the reactor inner vessel. The portion of the core that remains covered l by this mixture is cooled by pool nucleate boiling, which is adequate to main-l tain the cladding temperature near that of the saturated fluid. If the entire l core is not covered by the mixture, the portion above the' froth level is cooled I
4 i
i 1
05/10/85 5 -
_ - , ,, _m-_,,-__,___,..,_,.m._ _ _ , _ _ __,_m-
- - _ _ , . - ,,___,__~.-_,.,___ _.
1
. by forced convection to steam. As the system depressurizes, injection flow in-I t creases, and gradually the core is recovered completely. The CRAFT 2 code is used to predict the hydrodynamic behavior of the reactor coolant system. If
^
CRAFT 2 predicts that the core will be covered with liquid throughout the tran- 1 sient, no' core heat-up is predicted, and therefore, no themal analysis is re-quired and compliance with 10 CFR 50.46 is ensured. Othenvise the F0AM
, computer program is used to determine the mixture height within the reactor l core and the thermal response of the hottest fuel pin is calculated using the i
THETA computer program. The computer code interface and data transfer scheme l is shown in Figure 1.
As appropriate, the NRC concerns regarding the B&W small-break LOCA Evaluation ~
~
Model as identified in NUREG-0565 (Reference,4) and NUREG-0623 (Reference 5) ~
are addressed in Section IV and V of th'is report. " "" ' ' ~
l The B&WOG has also responded to NRC questions concerning BAW-10092P Rev. 3 and j BAW-10154 (Reference 6).
i III. EVALUATION OF CRAFT 2 SMALL-BREAK LOCA MODELS l .
The reviews of the revised CRAFT 2 computer program models are provided in this i s2ction and are related to the nine major areas of concern expressed by the NRC, as mentioned earlier. The following guidelines were, in general, used for i
this review:
I (a) Lower-than-actual energy removal from the primary system and minimi-zation of cooling water injection into the primary system is judged to be conservative because higher fuel temperatures will result.
Note that the lower-than-actual energy removal results in a higher pressure level and, therefore, increased break flow out of the prima-t ry system and decreased safety-injection flow into the system.
i j (b) Models not used for SBLOCA-EM analysis were not reviewed. Included l' are models for noncondensable gases (NCG), pressurizer spray, 1
l enthalpy adjustment, and downcomer bypass. It has been shown that l
t 05/10/85 6 B&W CRAFT 2 SER f_____,._._. _ __ __. - - - _ _ _
l 9
Figure 1. Small-Break LOCA Code Interface Taken from BAY-10154 Initial RC System ;
Parameters BAW-10092P i
" CRAFT 2 Code Average Response p Of System and Core V
. l Is Core YES Covering nsured? -
i Initial Core NO :
Pa,rameters $ .
Pressure, Power and I Y Qsiet Yater Level i
' Pressure, Power and I Flow Enthalpy '
l V BAY-10155 l FOAM Code V BAW-10094 V
THETA Code Mixture Level Hot Channel
> Response 4 Average Channel Steaming Rate ;
l V '
.r l HotResponse Pin Thermal ,;
l* I 31, 7q
!10CFR50.46 j l,, D emo n s t r a t e d -
l I
~
the effects of noncondensable gases is insignificant for SBLOCA-EM analysis.
(c) B&W experience and judgment was utilized for the selection and evalu-ation of tests or plant transients for code benchmarking, for the estimation of input values for the models, and for the estimation of parameter values when some pertinent test data missing.
(d) Future benchmarking will be performed as new test and plant transient results become available.
The following modelis .have been incorporated into the CRAFT 2 computer program, '
in response to NUREG-0737: " -~ ' '
(a) Pressurizer Model A non-equilibrium pressurizer model was added for SBLOCA-EM analyses.
The model simulates pressurizer performance using two thermodynamic -
systems, one for stratified steam and one that contains either subcooledliquidoratwophasemjxturs. Models are included for simulating pressurizer sprays, heaters, safety. valves, and steam-mixture interface heat and mass transfer. Surge line flow is comput-ed using a linear momentum balance, while relief valve performance is approximated via input mass flow rate versus pressure tables,'the Moody model, or the HEM isentropic expansion model. The spray model and the heater model are not used for SBLOCA-EM analyses.
(b) Two-Phase Flow Model Two phase flows in vertical columns are modeled with the Lahey and Ohkawa drift-flux model correlation. Associated with the model is a logic to account for phase separation within the control volumes com-prising drift flux columns.
Q e /
c - () - .
s
.The drift flux model accounts for phase slip by modifying the convec-tiveterminthemixtureenergyequation. The mixture mass and mo-mentum conservation equations remain unchanged.
The drift flux model uses two equa.tions to relate steam and liquid volumetric, flux densities to the total volumetric flux density.
t' These equations model the composition of flow in the flow path based on the fluid flow conditions as well as the states of the bounding control volumes.
In the development of the basic drift flux model, it is assumed that
~
control volumes comprising vertical columns contain a homogeneous mixtureofsteamihdliquid. A drift flux level formulation option 5
is provided for the~ accurate description of two phase phenomena when
,sufficiently small control volumes are used.
/
Tb w.
.. (c) Two-Phase Pump M__od.el. .
A new pump model was developed and accounts for two phase flow degra-dation of the head and torque curves.
, a (d) Steam Generator Model
,s ,
A detailed steam generator model was added to the CRAFT 2 SBLOCA ver-sion. In this erdel, the steam generator is described by multiple l ; axial regions, each of which contains one secondary volume and one or l two primary volumes. Heat transfer between primary volumes and the corresponding secondary volume is calculated based on a regime-dependent correlation set and the results of an implicit tube calculation. This model contains all the features comprising the l
standard once-through steam generator (OTSG) and such special fea-l tures as level. rate-dependent auxiliary feedwater control, an aspira-tor'model (OTSG), and:a'model that accounts for condensation of steam in the presence of noncondansables on the primary side of the steam generator.
05/10/85 9 -
- --. . _ _ a+_. - , _ - - . - -..- -_ .-. _
. The heat transfer contains logic that can be used to account for the degradation of condensation heat transfer on the primary side due to the presence of noncondensable gases. The model accumulates the noncondensable gases (NCG) from the following input sources:
l -
initial lbmoles of NCG in the system lbooles of NCG/108 lbmoles of fill flow lbooles of NCG/10s 1booles of flood tank flow lbooles of NCG released for each fuel pin rupture radiolytic NCG production in the core as a function of time and control volume density
~
lbacles of MCG from the Ir-H2O reaction.
All of the sources are assumed to originate in the reactor vessel and '
are split between the two steam generators based on the ratio of hot leg nozzle flow rates. Inside the steam generator upper plenum, the NCGs are split between the two radial regions based on the ratio of the flow rates into each radial region. The NCGs are assumed to re-side only in the steam generator upper plenum and in primary volumes undergoing condensation.
i III.1 Condensation Heat Transfer and Noncondensable Gases Two general concerns are identified in II.K.3.30: (1) the present condensation heat transfer models have not been adequately verified against applicable data, cnd (2) the effects of noncondensable gases have not been verified.
III.1.a. Steam Generator Model The steam generator condensation heat transfer model used in CRAFT 2 is for a flat plate in a saturated atmosphere without noncondensable gases being present. B&W has justified the use of this model by comparison to a more de-tailed solution for the heat transfer coefficient in a tubular geometry. Since 05/10/85 10 B&W CRAFT 2 SER
- ., - - . . - . - . - _ - - - - - . . ~.--- -.- ._._ - - .-
l T there was.less than a 0.4% difference between the flat plate solution and the more complicated result for flow inside of the tubes, the flat plate solution is justified.
4 The flat plate solution can be used because the film thickness is small com-pared to the tube radius.
This statement will be true for all conditions (where water is the fluid) where the flow is laminar. A substantial condensate
- film thickness resulting from long condensing lengths and large temperature differences will produce a turbulent flow regime, and the flat plate equation will underpredict the heat transfer coefficient, and is therefore conservative.
The condensation heat transfer coefficient is only used when the wall tempera-ture is below the saturation temperature.- When this event occurs, the location of the steam-water interface is deteNined,' and the length used in the conden-sation heat transfer forumula is the length from the bottos, of the upper tubesheet to the steam-water interface.
Babcock & Wilcox has performed experimental measurements of condensation heat transfer with noncondensable gases present in a prototypical 0TSG tube.
The results of this study were reported in Reference 7. An analysis of this data was reported in Reference 8.
The analysis considers the diffusion of the vapor ,
t into solution in the condensate film. This model was in good agreement with the data. However, the model requires three interactive calculations to deter-cine the heat transfer coefficient; thus, the implementation of this model in the CRAFT 2 code would not be feasible. A user-input table of heat transfer 1
cultiplier is therefore provided as a means of accounting for degradation due to noncondensable gases.
B&W has not yet run any SBLOCA-EM calculations using this capability. B&W 4
SBLOCA-EM analyses have demonstrated that the presence of noncondensable gases is negligible, i.e., the Core Flood Tanks do not empty and peak cladding tem-peratures are below the metal-water reaction temperatures. When such calcula-tions are required, the input tables will be developed using the analytical codel developed in Reference 8. The NRC will institute a review of this model at that time as a plant specific issue.
05/10/83 11 B&W CRAFT 2 SER n .. . . .
Based on.the observation from the Semiscale Mod-2A Natural Circulation Cooling test program (Reference 9), the staff finds that the B&W conclusion as to the i
negligible effect of noncondensable gases on condensation heat transfer is ac-
$ ceptable, provided that the Core Flood Tanks do not empty and that the peak cladding temperatures do not reach the onset of significant Zircalloy-water reaction. s.
! The effects of noncondensable gases on natural circulation flow are discussed in Section III.3, below.
III.1.b Pressurizer Model The effects of condensation and noncondensable gases were also reviewed for the
- new non-equilibrium pressurizer model in CRAFT 2. The model uses an overall heat transfer coefficient, U ,j a steam-mixture interface area, A g, and a multiplier, f , to e determine the condensed mass. Heat transfer through the liquid is assimed to be predominately a condensation process. The variable ar-ca, Ag, is calculated based on the actual pressurizer geometry, and is the heat transfer area at the interface.
A value of 1.0 is input for f , ewhich requires an equivalent mass transfer (condensate) for all heat transferred across the interface. The value of unity j prevents desuperheating or an accimulation of a mist (quality) in the steam space. The use of unity for f isg a reasonable assumption since mass and cnergy is conserved and superheat temperatures would not be expected to be sig-nificantly altered in the steam region. This assumption corresponds to the
{ isentropic compression of an open system in which mass is removed from the re-1 maining mass.
The accumulation of noncondensables in the pressurizer is judged by B&W to be negligible due to the surge line configuration. Also, the condensation heat transfer across the liquid-steam interface is very small compared to the con-densation heat transfer at metal surface interfaces. Additionally, the conser-vative conduction model assumed for heat transfer at the liquid-steam interface 05/10/85 12 -
t is conduc, tion limited, adding to the conservatism in accounting for the pres-ence of noncondensables in the pressurizer. Therefore, a more mechanistic ac-counting for noncondensables in the pressurizer would have little, if any, 1 impact on the results predicted by the B&W SBLOCA Evaluation Model.
The condensation model for the pressurizer is acceptable. Validation of the i pressurizer model is discussed in Section III.5, below.
Accumulation of noncondensable gases in the pressurizer is not expected. The design of the surge line would limit the flow of gases to the pressurizer for breaks other than in the pressurizar itself. For breaks in the pressurizer, it is expected that noncondensable gases will exit the system through the break.
Reference 10 and Appendix C to 8AW-10154 provide additional details concerning i
the surge line design and evaluation. -
III.2 Nonequilibrium Effects i
The general concern identified in II.K.3.30 is the validation of the procedures -
used to model the physical phenomena resulting from subcooled water injection.
i Perticular effects of concern include the effects of node size, injection loca-tion, and the localized pressure. Note that these can affect the injection mass flow rate. This concern is primarily directed at Core Flood Tank (CFT) injection into the reactor coolant system.
III.2.a Core Flood Tank (CFT) Model The magnitude of system depressurization resulting from CFT injection is depen-4 dent upon control volume fluid content and control volume size. The selection of the ECC injection location can cause gross disturbances in the system re-sponse following CFT actuation. A sensitivity analysis performed by B&W con-l firms that the gross pressure disturbance is a function of modeling techniques.
, Reference 11 and Appendix D to BAW-10154 provide the results of these studies.
Results from LOFT Test L3-1 have shown that a minimal pressure disturbance will occur upon CFT actuation. CFT injection into a steam environment results in i
d the injection of a substantial amount of CFT liquid causing a rapid core recov-ery, wnich is non-conservative for transients similar to that of LOFT Test L3-1.
To maintain a conservative position for all SBLOCA transients, including those i
cutside the representation of L3-1, the B&W CFT model injects into a liquid cnvironment to minimize pressure disturbance. This conservatism is valid for all SBLOCAs of intenest, including those during which a greater pressure dis-turbance may occur.
l The CFT model used by B&W is acceptable for SBLOCA-EM analyses.
III.2.b Pressurizer Model The two-region, nonequilibrium pressiarizer model uses two homogeneous regions, of variable size, to model the mixture-steam interface. Constant heat transfer
) coefficients are used to account for condensation on the pressurizer walls.
Together with the assumption of constant mixture temperature, coupled with the
! cssumption of conduction limited heat transfer at the interface, this is found to be a conservative and acceptable pressurizer model for SBLOCA-EM analyses.
III.3 Hot Leo Phase Separation
'i II.K.3.30 recommends the use of an adequately conservative phase separation -
i model because entrapment in the candy cane of the separated vapor could inter-rupt natural circulation. Of additional concern are the hot leg model account-ing of temperature distribution as the hot leg is refilling, the energy I
cxchange with the walls, and the condensing rates at the liquid-vapor interface.
The interruption and re-establishment of natural circulation is dependent upon the mixture height in the hot leg relative to the bottom elevation of the U-bend piping. The SBLOCA-EM U-bend noding scheme accounts for the bottom ele-vation of the U-bend piping. Natural circulation will be interrupted when the sixture level falls below the bottom elevation of the U-bend node and is sus-tained above this elevation. Thus, the model accounts for spillover prior to sustained recirculation, when the hot leg is refilling.
4 05/10/85 14 -
e Two drift flux models are available. One of these correlations is a simpler model developed by Kelly, Dougall, and Cantineau. This model was used in B&W's successful prediction of the LOFT L3-6 test (Reference 12). The second is a more sophisticated drift flux model developed by Lahey and Ohkawa. This second correlation (Lahey and Ohkawa) has been shown to demonstrate better agreement i
to level swell test data and as a result is the model currently used for SBLOCA-EM calculations.
The level formation model is used with the drift flux model to provide for phase separation. Phase separation begins in the uppermost node cor.taining liquid and proceeds down from node to node as phase separation progresses. To accommodate phase separation, the upper stratified node calculation combines the Wilson bubble rise correlation a with the drift flux model to detamine the bubble escape velocity' from the mixture and, hence, the mixture elevation in the node. The phase distribution in each of the lower nodes is assumed to be homogeneous.
The drift flux model/ level formation model is used on the primary and secondary -
F sides of the steam generators. Elsewhere in the system, the Wilson bubble rise correlation is used to model two phase behavior." The Wilson bubble rise model is suitable for vertical columns in which low flow or fluid stagnation exists.
Each vertical region of the systen under Wilson bubble rise consideration is modeled as a single control volume. B&W experience demonstrates that one ver-tical node and one U-bend node together with the Wilson bubble rise correlation is sufficient to model spillover and the interruption of natural circulation.
The Wilson bubble rise model is acceptable for two phase low flow calculations in the hot leg for SBLOCA-EM.
B&W experience has demonstrated that using the Wilson bubble rise model in the hot leg for two phase flow (low flows) and for phase separation yields more conservative results than the drift flux model/ level formation option. Phase separation is predicted to occur more quickly, interrupting natural circulation I
sooner and delaying the re-establishment of natural circulation.
_ ___ _ x m_ _
4 The drift flux models have been benchmarked to applicable data.Section III.5, below, provides additional information concerning verification of the drift flux model.
4 Heat transfer from the liquid in the U-bend node to the reactor coolant system metal is modeled. Condensation of steam in the U-bend node is modeled if the metal surface temperature is less than the fluid saturation temperature. A conservative condensation heat transfer coefficient of 1.98 BTU /fts-hr *F is used.
It has been demonstrated that the effect of noncondensable gases is negligible en the condensation heat transfer in the steam generator. The other potential
~
impact of noncondensable gases for a SBLOCA-EM analysis is the interruption of
- natural circulation as a result of a sufficiently large volume of gases in the hot leg U-bend. '
In NUREG-0565 (Reference 4) the NRC staff reported the results of the B&W li-l consees evaluation of the effects of noncondensable gases.
There are nine sources of noncondensable gas whic'h are already in, or could po-tentially be introduced into, the primary system. These are:
(1) dissolved hydrogen in the primary coolant; (2) dissolved nitrogen in the CFT water; (3) dissolved air in the borated water storage tank; (4) hydrogen releases from zirconium-water reaction; (5) free nitrogen used to pressurize core flood tanks; (6) hydrogen released from radiolytic decomposition of injected water; (7) fission and fill gas in reactor fuel; l (8) hydrogen gas (free and dissolved in the makeup tank); and (9) pressurizer steam space gas.
1 With the exception of the source due to radiolytic decomposition (item 6), B&W accounted for each of these sources in their analysis. Because the CFT actua-I tion pressure is approximately 450 psig below the secondary system relief valve 05/10/85 16 -
y setpoint, the steam generators will be heat sources rather than sinks for any breaks which depressurize to the CFT setpoints and natural circulation would 1 not be a requirement for decay heat removal. Therefore, gas sources from the l CFT were not included in the analyses. The licensees have also concluded that for all small breaks considered in the design bases, peak cladding temperatures are low enough that fission gas sources due to cladding rupture or oxidation sources are negligible. Therefore, it was concluded that gas from sources identified as itans (1), (3), (8), and (9), along with fission and fill gases assuming one percent failed fuel in the core, are available to the primary system.
I j If it is conservatively assumed that all the gas comes out of solution, that no noncondensable gas is lost through the break, and that the amount of water in-jected by the high pressure injection system from the borated water storage tank is 64,000 lbe (which corresponds to 1500 seconds of injection), then the B&W estimate for noncondensable gas in the primary system is 780 standard cubic feet. At a systes pressure of 1050 psig (the secondary side relief valve setpoint), this volume would occupy 22.4 cubic feet.
~
In order to inhibit natural circulation at pressures representative of small breaks requiring secondary system heat removal, the gas would have to fill the U-bends at the top of the hot legs. There bends have a volume of 125 cubic j
} fcet. Thus, the conclusion drawn by the licensees is that the maximum amount of noncondensable gas calculated to be avail'ble a is approximately a factor of five less than the amount needed to inhibit natural circulation. This analysis conservatively assumed that no gas accumulated in the upper head or plenum of the reactor vessel, which is considered the more likely location for gas accu-culation. Thus, no reduction in natural circulation flow is predicted by the licensees due to noncondensible gas accumulation. However, as pointed out pre-viously, B&W has neglected any gas source due to radiolytic decomposition of the water.
B&W endorses the conclusions in NUREG-0565 concerning noncondensable gases. In addition, the B&W position regarding the radiolytic decomposition of the injected 05/10/85 17 B&W CRAFT 2 SER
e water is ,that this additional source of noncondensable gas does not alter the conclusions in NUREG-0565. Acceptance and reference to this Safety Evaluation i
Report by B&W and by the B&WOG affirms the conclusion regarding the insignifi-cant effect of noncondensable gases for a SBLOCA-EM analysis.
III.4 Steam Generator Heat Transfer I
II.K.3.30 expresses the peneral concern that the modeling of the steam genera- l i
tor secondary side conditions are oversimplified and too dependent on user-specified input, which could be used to dictate the desired transient result.
i Specific concerns regarding the steam generator model were the heat transfer correlations and the effects of auxiliary feedwater (AFW) on the transient re-i sponse. In response to this concern, B&W has developed a more mechanistic 4
steam generator model with multiple heat transfer correlations and more realis-l tic AFW interaction based on actual OTSG characteristics. The model includes the correlations of Dittus-Boelter for subcooled and superheated forced convec-tion, Chen for saturated nucleate boiling, modified Chen for subcooled nucleate boiling, McAdams for natural convections, Nusselt's condensation correlation as given by Kreith, and Drew's correlation for..a falling film for AFW heat trans-for. Acceptability of the new steam generator model for SBLOCA has been demon-strated through noding sensitivity studies, the benchmark of Semiscale Mod-2A natural circulation test S-NC-2, and the benchmark of a Loss of Offsite Power (LOOP) event at Unit 1 of the ANO-1. Additional infomation concerning verifi-cation of the new steam generator model is provided in Section III.5, below.
III.5 Systems Verification and Other Experimental Verification j II.K.3.30 expresses the general concern that predicted overall system perfor-4 mance is not adequately verified against applicable data. Verification of
- overall performance would test detailed code models, for example those for con-densation heat transfer and the vent valves. Also tested would be integral ef-fects, such as interruption and restart of natural circulation. In addition, sensitivity studies for integral tests could be used to determine the impor-i tance of various modeling features.
05/10/85 18 -
III.5.a . Vent Valve Model and Countercurrent Flow Model The B&W NSSS design incorporates vent valves. The vent valve is a flapper-type valve which allow steam generated in the reactor core during a cold-leg LOCA to be vented from the reactor vessel upper plenum region directly to the downcomer, cnd then out the break. The vent valve design precludes the need for loop seal clearing and unacceptable core uncovery, and therefore sitigates the consequences of a LOCA. Prototypical, sealed model tests of the vent valve have been performed by B&W to determine the vent valve flow characteristics (Reference 13).
The composition of flow in the cold leg during pump suction and pump discharge breaks plays a major role in governing the consequences of an SBLOCA. A higher quality effects the volumetric discharge out the break which tends to reduce the rate of mass inventory depletion while simultaneously reducing the specific -
cnergy of the system. The net effect is an increase in the depressurization rate and, therefore, an increase in the HPI delivery.
In the ' current SBLOCA EM, reacto'r vessel vent valves are modeled using a simple pipe momentum equation oriented so that flow from the upper plenum to the down-comer is in the positive direction. The forward flow loss factor, which is input to the model, is based on experimental data and corresponds to a fully open vent valve configuration. A very large reverse flow loss factor is used to preclude flow from the downconer to the upper plenum. The modeling of vent valve dynamics and flow characteristics as a function of opening angle has not been incluc'ed in the SBLOCA EM due to the small pressure differentials required to open the valves and hold them in a full open configuration.
As required by design and periodically verified by testing, the vent valves will reach a full open configuration when acted on by a force no greater than the equivalent of a 0.25 psi pressure differential between the upper plenum and the downcomer (Raference 14, for example).
In SBLOCA evaluation model cases, steam passes through the vent valves and flows through the cald leg to the break location. Simultaneously, liquid, which over-flows the purep suction or is injected by the HPI, will flow in the direction 05/10/85 19 B&W CRAFT 2 SER
towards the vessel. This countercurrent flow phenomenon is represented in the j
SBLOCA-EM by providing two horizontal flow paths joining adjacent cold leg I
- volumes at different elevations.
4 l The ability of the CRAFT 2 code to predict countercurrent flow using the " double path" modeling scheme described above has been verified by predictions of LOFT and Semiscale tests. The phenomenon of countercurrent flow in horizontal and sloping pipes is not unique to the B&W MSSS configuration. It has occurred in LOFT and Semiscale SBLOCAs during the draining of the hot leg piping to the i' vessel.
I The ability of the current SBLOCA model to predict steam migration in the cold leg has been verified indirectly by predictions of LOFT and Semiscale experi-ments. Precise predictions of this phenomenon are not judged by B&W to be nec-cssary for providing an appropriate representation of the overall SBLOCA scenario.
l Recently completed analyses performed by the NRC with the RELAPS/M002 computer program have been qualitatively compared to the B&W CRAFT 2 results for a 0.01 -
square feet cold-leg break (Reference 15). While this study was not intended
~
to be a direct audit analysis, input assumptions were selected to explore the
.l cffects of natural circulation, reactor system repressurization, and boiler-condenser heat transfer including the effectiveness of Auxiliary Feedwater (AFW) spray on the upper OTSG tube elevations.
The RELAP5/M002 analyses indicated that although AFW wetting effectiveness can 1
l influence the primary system response, the general trends are unaffected. The '
4 analyses demonstrate that repressurization can occur during a small break LOCA '
at a B&W lowered-loop plant. Figure 2 provides a comparison of the RELAPS re-sults to a similar CRAFT 2 analysis. This confims the B&W CRAFT 2 results for a
- similar size break. The timing of the operator action to raise the secondary system water level to 95% was found not to be critical. Some boiler-condenser heat transfer was found to occur with the level at 50% of the operating range, which terminated the increase in the reactor system pressure before the level I
was raised to 95%. Finally none of the conditions examined led to core uncovery 4
or heat up. '
05/10/85 20 -
- l. . , _ _ _ _ _ _ _ . . . _ __ _
t -
i 2500 LEGEND i
~
l .5 CRAFT 2 - EM m -
l a -
R5..C. ..a..s :
- g 2000 - '
.. ...... e. 1
. , / R5 Case _2
'm ai 1
/
e .
s 4s1 e -
8 i
g 1 ss k i . 1500
- I ~
.h. . .... .
,./ .
i "x *
'...,,~~-..I / ' N'
...2 u %...... , ,,
- O 1000 - , m.
, u. .
W .
E .
~
i i
500 ..............,.... .... ....
0 500 1000 1500 2000 2500 3000 Time in seconds
, Figure 2. Comparison of CRAFT 2 to RELAP5.
RELAPS/ MOD 2 sensitivity study on percentage of stearn generator tubes wetted by AFW.
Case 1100 %
Case 2 0%
- B&W test data shows 6 to 10% wetting, as
! reflected in CRAFT 2 results.
i
l i
This qualitative assessment between CRAFT 2 and RELAPS/M002, which employs a dy-namic model for the vent valve as well as current state-of-the-art models for flow and heat transfer, provides suitable justification for the acceptance of the CRAFT 2 models, and the B&W nodal representation used for SBLOCA-EM analyses.
Details concerning the nodal studies performed by B&W are documented in BAW-10154.
Additional justification for the B&W SBLOCA-EM models are discussed below.
III.5.b Pressurizer and Surae Line Model The nonequilibrium pressurizer model and surge line model have been evaluated against the ANO-1 Loss of Offsite Power (LOOP) event (Reference 16), the NPD surge tank insurge experiment (Reference 17), and the Syracuse University surge tank tests (also Reference 17).
These evaluations demonstrate that the models for the pressurizer and the surge line are acceptable.
III.S.c Steam Generator and AFW Models The benchmark analyses for the AFW model are provided in Reference 18. Includ cd in these studies is the LOOP test at TMI-2, and the LOOP events which oc-curred at Davis-Besse 1 and at ANO-1.
The natural circulation experiment during the hot functional tests at Oconee-1 provided AFW model development data (Reference 19). One of the Oconee-1 steam generators was heavily instrumented during the hot functional test program to cbtain the data needed to develop the AFW model. Additional data to support the AFW models was also obtained from a flow-visualization test program per-formed by B&W (Reference 20).
The auxiliary feedwater model developed for use in the CRAFT 2 computer program is based on experimental data and has been adequately verified against both I separate and integral test data. The AFW model is acceptable for use in i SBLOCA-EM analyses.
05/10/85 22 -
4 The steam generator heat transfer models have been assessed against the Alli-cnce Research Center loss-of-feedwater test (Reference 21). The CRAFT 2 SBLOCA model analysis compared favorably with the test data.
l The steam generator heat transfer models were also assessed with the Semiscale
, Mod-2A natural circulation test S-NC-2, and is documented in BAW-10154, Appen-dix G. The purpose of this analysis was to demonstrate that the revised CRAFT 2 computer program can track the various modes of natural circulation observed during a small-break LOCA.
During this analysis only the single- and two phase natural circulation modes were predicted. The reflux condenser mode of natural circulation was not con-sidered here since the relevant phenomenon was not applicable to a B&W NSS.
i The single- and two phase modes of natura1 circulation were ob'tained by drain-ing discrete amounts of liquid out of the reactor vessel lower plenum, allowing '
sufficient time for steady-state conditions to be achieved between drains. The everall loop natural circulation mass flow rate varied considerably depending en system mass inventory. The variation in loop mass flow rate with inventory
) was a result of the transition from single phase to two phase natural circula-
- tion. '
J Initially, the draining simply lowered the vessel liquid level to the top of the het leg with no significant voiding in the loop. Consequently, there was little change in loop mass flow rate. Further draining caused the loop mass flow rate to increase sharply and eventually peak. This increase in flow was caused by increased voiding in the upflow portion of the steam generator, which increased the everall loop density gradient. The peak in flow occurred as
! steam bubbles in the upflow side eventually spilled over into the downflow side of the steam generator, causing a reduction in overall loop density gradient between the upflow and downflow sides in the steam generator.
The results of the post-test predictions of test S-NC-2 show that the CRAFT 2 computer code compared reasonably well with the data. CRAFT 2 predicted the same general trends as were found in the test. For most of the data points, 05/10/85 23 B&W CRAFT 2 SER w _.
the results calculated by CRAFT 2 were within the uncertainties of the measure-ments. This analysis demonstrates that the upgraded CRAFT 2 code is capable of predicting the single- and two phase natural circulation modes observed during the small-break LOCA transient.
B&W considers Semiscale Test S-NC-2 to be a suitable benchmark because it ex-hibits relevant SBLOCA phenomena. Results of the benchmark analysis demon-strate the adequacy of various analytical correlations in the evaluation model to simulate SBLOCA phenomena as they occurred during the Semiscale test. The commonality of SBLOCA behavior between reactor designs justifies the usefuir.ess cf S-NC-2 as a benchmark. Consequently, B&W concluded that the performance of the evaluation model will be consistent and that the evaluation model will ade-quately simulate SBLOCA performance of a B&W NSSS as plant-specific features cre added to the modeling scheme such as the OTSG and vent valves.
The key issue is modes of natural circulation, governed by fluid conditions in typical PWR loops. The Semiscale facility is not scaled directly to the B&W configuration. Nonetheless, the physics of natural circulation are presumably independent of the design. Interruptions in flow, boiler-condenser heat trans-fer, vent valve effects, etc., may change the sequence and duration of certain cvents. Nonetheless, S-NC-2, as well as other benchmark analyses, confirm that ,
o CRAFT 2 can adequately predict the key physical processes associated with natu-ral circulation in commercial PWRs.
Additional comments related to Integral Systems Tests (IST) for geometries which are representative of the B&W design are provided in Section III.5.g. ,
b310w.
III.S.d Two-Phase Flow and Phase-Separation Models The benchmark analyses for the Wilson bubble rise model are provided in Refer-ence 17. The GE/Hitachi and Westinghouse level swell experiments were used to assess the Wilson bubble rise model. The application of the Wilson bubble rise codel, as part of the drift flux model, was assessed against the Mitsubishi 05/10/85 24 -
. . _ , , , - _ - , ---_--- _y.-
Atomic tests for drift flux modeling assessment. Additional verification of these models was obtained by comparing CRAFT 2 to the LOFT L3-6 test (Reference 12). i l
The purpose of these investigations was to demonstrate that the revised CRAFT 2 ,
computer program, more specifically the new steam generator model, can ade- '
quately predict the phase-separation in the hot leg and the mixture level in the steam generator and to account for bubble formation and the interruption Cnd re-establishment of natural circulation.
Nodalization studies for the hot leg and steam generator were performed by B&W, cnd provided in BAW-10154, to determine the proper ~ nodal model for their repre-sentation. !-
. . . a* '
The use of the Wilson bubble rise model and the Lahey and Ohkawa drift flux '
model, coupled with the nodal representations for the hot leg and steam genera-tor, are acceptable for use in SBLOCA-EM analyses.
III.5.e Core Heat Transfer Model The small break core heat transfer model employed by B&W was compared for steady-state conditions to six tests performed at the Oak Ridge National Labo-ratory (ORNL). These tests were designed to evaluate the steam cooling capa-bility within a core in which the fluid level was allowed to fall below the top of the active core region and stabilite at an intermediate location. Core cool-ing under these conditions would be accomplished through pool boiling below the swell or fluid level, and by steam cooling above the swell level. The compari-stns were made for that portion of the core above the swell level.
The comparison of the ORNL heat transfer test to B&W techniques indicates that utilization of the Dittus-Boelter correlation as the sole determination of heat transfer is acceptable for determining compliance to 10 CFR 50.46, no core un-ctvery. For the low flow tests, which are most representative of B&W SBLOCA conditions, the combined convective and radiant heat transfer from the ORNL l
l l
tests was higher than the Dittus-Boelter prediction when cladding temperatures
- cxceed 1000F. For all tests, the predicted heat transfer was conservative, relative to the ORNL test data, for cladding temperatures above 1400F. Due to
, the conservatism of the Dittus-Boelter heat transfer correlation at high clad-ding temperatures, the present heat transfer model is acceptable for licensing j calculations. s t If the CRAFT 2 analysis indicates core uncovery, a more detailed mixture level i
and heat up analysis is perfomed with FOAM and THETA.
Appendix A of BAW-10154 provides a detailed description of the B&W studies and 4 comparison to the ORNL data.
III.5.f Intecral System Benchmarks In addition to the ANO-1 LOOP event, the Semiscale S-NC-2 test, and the LOFT L3-6 test, previously discussed, additional integral system benchmarks have been performed.
These include LOFT test L3-1 (Reference 22) and Semiscale test -
S-07-100 (Reference 23).
The LOFT L3-1, L3-6, and Semiscale 5-07-100 analyses were performed prior to the submittal of the CRAFT 2 computer program documentation under review. B&W -
was requested to provide a discussion on the suitability of these comparisons for verification. The following information was given concerning these analyses.
1 The code versions used to benchmark LOFT L3-1, LOFT L3-6, and Semiscale 2
S-07-100 are characteristic of the code presently under review. Upgrades to these earlier versions include a two phase pump model, a non equilibrium pres-surizer model, and an improved steam generator model. These upgrades are ex-pected to have a negligible impact on the previous results of these specific benchmark analyses.
Inclusion of the updated two phase pump model will not affect the results of i i
the benchmark simulations. This is an appropriate conclusion since the RCS pumps were tripped at the initiation of each test with the exception of L3-6. '
l 05/10/85 26 -
- - , __ .~ _ _ . _.- ._ __ __ - _ . _
l Thus, two phase pump performance, for the benchmark cases other than L3-6, was not experienced in the tests. For LOFT Test L3-6, two phase pump flow degrada-tion parameters were used which were :haracteristic of the upgraded two phase pump model.
The inclusion of the non-equilibrium pressurizer model will have little or no offect on the earlier results. This conclusion was reached after investigating the hydrodynamic behavior of the pressurizer during the transients used to benchmark the SBLOCA-EM. LOFT L3-1, LOFT L3-6, and Semiscale S-07-100 are transients in which the pressurizer emptied in a very short time and non-equilibrium modeling in the pressurizer is of little consequence.
The updated steam generator model would a1so have been inconsequential to the benchmarks. In the LOFT L3h6 benchmark', secondary steam generator conditions were controlled to the actual test data. In Semiscale S-07-100 and LOFT L3-1, the primary and secondary systems were decoupled during most of the transient.
Therefore, steam generator performance predicted in the earlier code version is ~2-representative of expected results for the version under review.
In sussiary, the analytical models affecting the s'ystem hydrodynamic predictions (i.e., leak discharge, drift flux, bubble rise) are modeled identically in the
/
c:de version under review and the versions applied in LOFT L3-6, L3-1, and Semiscale S-07-100. These tests were selected as the best available benchmarks, and they represent key PWR phenomena during a SBLOCA for a cold leg pump dis-charge (CLPD) break. Accurately predicting the system hydrodynamic behavior in these tests is considered partial but substantive justification for the B&W SBLOCA-EM.
The CRAFT 2 code version under review was benchmarked against the Semiscale Nat-ural Circulation Test S-NC-2 to demonstrate the analytical capability of the upgraded CRAFT 2 code in tracking various modes (single- and two phase) of natu-ral circulation observed during an SBLOCA. The results of the analyses show that the upgraded CRAFT 2 code was capable of reasonably predicting the various modes of natural circulation.
l The CRAFT 2 computer code was benchmarked against a loss of offsite power (LOOP) transient'at Arkansas Nuclear One, Unit 1 (ANO-1). The analysis was performed in part to demonstrate the capability of the code to correctly predict the steam generator response during a B&W plant transient. .The results showed that the upgraded CRAFT 2 model was quite capable of predicting the system response !
during the ANO-1 LOOP event. !
i The verification and assessment program presented by B&W in support of the CRAFT 2 co.:puter program is judged to be acceptable in demonstrating that CRAFT 2 i
can predict the major phenomena associated with a $8LOCA. These are the inter-ruption and re-establishment of natural circulation, phase separation in the j hot-leg, and steam generator condensation heat transfer.
j While the integral system tests (LOFT and Semiscale) are not representative of
) the B&W NSSS design, the CRAFT 2 comparisons to plant transients and tests, most -
l notably the ANO-1 LOOP event, provide reasonable assurance that CRAFT 2 can pre-dict the pressurizer response, the steam generator response, and natural circu-lation (single phase) for the B&W NSSS design.
i III.S.o B&W Integral System Test Program -
i
l data for the verification of the B&W version of the best-estimate computer pro-gram RELAP5/M002 for B&W NSSS specific geometries. The IST program is not in-tended to provide integral test data for verification of the CRAFT 2 SBLOCA-EM computer program.
B&W and the B&WOG have indicated (Reference 6) that the RELAP5 code may, in the future, become a component of the ECCS Evaluation Model but until that takes place, ECCS evaluations for licensing will be performed with the then current, approved Evaluation Model as required by 10 CFR 50.46.
I
, The B&WOG position on the use of ITS data for verification of the CRAFT 2 com-puter program is documented in Reference 25.
l- i 1- -
i !
y The NRC position concerning the IST program, in particular this MIST test fa-
.; cility, is that completion of the MIST program is not needed to approve comput-or programs and evaluation models to resolve the II.K.3.30 issue. However, the MIST program must be done to confirm the conclusion that CRAFT 2 provides a con-servative representation of S8LOCA behavior in B&W PWRs.
i
! It is the intention of the B&WOG to use RELAP5/M002 for best estimate long term
, transient predictions. Future transient response predictions for ATOG and the i
Generic Technical Bases Document will be based on RELAP5/M002. It is realized
- that present ATOG guidance in the area of SBLOCA is based on experience gained through CRAFT 2 licensing analyses. In order to affim the validity of present
! guidance in the light of new best estimate codes and availability of IST data, the 8&WOG is evaluating the benefits'and effort required to perform a confirma-tion of the CRAFT 2 model capabilities'in one of three ways: (1) benchmark of y CRAFT 2 to an OTIS test, (2) comparison between predictions of the same tran-l' sient performed in a best estimate mode using CRAFT 2 and a verified RELAP5/M002 !
l cr (3) comparison between predictions of the same transient performed in an Appendix K type calculation using CRAFT 2 and a verified RELAPS/M002. The lat-ter two alternatives are considered for post MIST evaluation when the "verifi-cation" process of RELAPS has been established.
j S&W has made a long standing commitment to continually review experimental data i cnd code predictions of these data by both 8&W and other organizations, as they i
apply to the CRAFT 2 evaluation model or portions of it (see Section 7 of BAW-t l 10154). The MIST program (IST) is one such source of new information. This l commitment, along with the commitment to affirm the validity of present ATOG
) guidance, is taken to be a commitment by B&W and the B&WOG to perform one of the comparisons noted above to demonstrate that CRAFT 2 does provide a conserva-i tive representation of SBLOCA behavior in a B&W PWR.
I l The B&WOG also recognizes that test facilities other than MIST are currently i
performing tests that may be applicable to a B&W-designed NSS. They have not l
yet received sufficient information from either the University of Maryland or i
SRI-II Program to adequately assess the benefits they believe will be derived f from each of these programs. The B&WOG will follow the test programs for each l
! 05/10/85 29 B&W CRAFT 2 SER l
_.,..__,_,,_,._m.. _ _ , _ _._,.y ._.,_.,,_,___..,,.,_,,_.r_
l
=
of these facilities with the intention of determining the usefulness of the da-ta generated to address scaling issues. However, the B&WOG currently has no i firm plans to benchmark data from either of these facilities. l l
The B&W IST program is being monitored and supported by the NRC. The program l is intended to support the development of the best-estimate analysis computer program RELAP5/ MOD 2. The effort is considered by the NRC staff to be confirma-tory with respect to the licensing analysis computer program CRAFT 2. It is the position of B&W and the B&WOG that the CRAFT 2 (Revision 3) computer program, as
- an evaluation model for SBLOCA analyses, is conservative with respect to the criteria for ECCS analysis for compliance with 10 CFR 50.46. With respect to the requirement of 10 CFR 50 Appendix K, the CRAFT 2 computer program conforms to the Evaluation Model criteria.
The NRC will continue to be involved in the IST program, and will continue to monitor other experimental programs related to the B&W NSSS design. B&W has committed to the continual comparison of relevant test data for the evaluation model and portions of it, as stated in BAW-10154. Any information on future comparisons will be documented and supplied to the NRC.
III.6 Flow Regimes i
II.K.3.30 expresses concern about the general ability of codes to properly pre-dict the two phase-with-noncondensable gas flow regimes that might exist in loop piping. The modeling of the flow phenomena needs to be verified or else justi-fication needs to be given of why the models used are conservative for all breaks without modeling the phenomena explicitly.
CRAFT 2 is an evaluation model code and is not intended for detailed best-estimate calculations. Thus, detailed calculations of flow regimes are not performed. Countercurrent flow is accounted for through the modeling tech-niques of parallel flow paths and flow regime dependent drift flux models. Of primary importance in transient analyses and predictions are macroscopic, or global, behavior such as RCS pressure and temperature, natural circulation, 05/10/85 30 -
heat transfer, and inventory. It has been demonstrated through benchmark ana-lyses that the CRAFT 2 code and modeling techniques can predict SBLOCA system characteristics and transient phenomena. For a discussion on noncondensable gas modeling refer to Section III.1, above.
LOFT Test L3-6 exhibited two phase flow and phase separation and is thus appli-4 cable to the verification of the evaluation model. The drift-flux model used in the LOFT Test L3-6 analysis is characteristic of the models in the code ver-sion under review. The drift-flux model used in the LOFT Test L3-6 analysis tas based on recommendations by Kelly, Dougall, and Cantineau. The same model has been incorporated into the CRAFT 2 code. Additionally, a second drift-flux model based on recommendations by Ohkawa and Lahey has been included in CRAFT 2. ~
Based on a better representation of fevel swell test data, the Ohkawa and Lahey ,,
4 model will be used for SBLOCA-EM calculations. -
l The flow regimes encountered in the LOFT and Semiscale SBLOCA experiments are indicative of those to be expected in actual plant transients. The duration, Cxtent, and timing of events in the tests may not coincide with prototypical behavior, primarily owing to scaling limitations. Nonetheless, the test phe-I nomena are representative, and, so long as the test geometries and conditions are within the range of the formulations used in the evaluation model, the tests present valid benchmarks.
The countercurrent flow parallel path modeling technique and the flow regime dependent drift flux model are acceptable for SBLOCA-EM analyses.
1 l III.7 Core Steam Coolina II.K.3.30 expresses concern regarding the capability to predict core level and core heat transfer because the comparisons with experimental results are not challenging to the code models.
The heat transfer models available in CRAFT 2 for an SBLOCA analysis can be cat-egorized as follows: (1) fuel pin surface heat transfer, (2) steam generator 05/10/05 31 B&W CRAFT 2 SER
_. - - . . . - . - . . ~ _ . _ _ - - - _ = .- - - - _ . . . . . - -- --
primary to secondary heat transfer, and (3) primary metal (i.e. , structural metal) heat transfer.
i 1
The fuel pin surface heat transfer model is described in the CRAFT 2 topical re-
~
! port BAW-10092, Rev. 3. The heat transfer coefficient at the pin surface is calculated for fivesregimes: subcooled forced convection, nucleate boiling, l
f transition boiling, film boiling, and superheat forced convection. These re- '
j gimes are modeled by the correlations of Dittus-Boelter for subcooled and su-i perheat forced convection; Thom for pre-CHF boiling; McDonough, Milich, and King for transition boiling; and Dougall-Rohsenow/Groeneveld/and Morgan for
] film boiling. These correlations are used in the applicable regimes. Of these
! regimes, for low flow condition, the superheat forced convection is of primary
~ '
inte mst to an 58LOCA when a portion of the core is uncovered.
j . ..
As indicated previously the detailed core level analysis is performed with the
{ FOAM 2 (Reference 26) computer program if the CRAFT 2 analysis indicates core i
uncovery for a 58LOCA-EM analysis. The actual hot pin heat-up analysis is then
] performed with the THETA computer program (Reference 27). -
l The review of the FOAM 2 computer program is.being performed by the Core Perfor-mance Branch, Division of Systems Integration. A preliminary assessment indi-I '
cates that, with some additional justification, the F0AM2 level swel'1 and pin temperature calculations are conservatively evaluated for 58LOCA-EM analyses.
J j The CRAFT 2 model is acceptable for determining if core uncovery will occur and i
is judged to be conservative for 58LOCA-EM analyses.
i
{ III.8 Metal Heat i
l II.K.3.30 expresses concern that metal heat should be appropriately accounted j for.
l All metal mass is simulated in the SBLOCA-EM calculations. Since the outside
{ walls are considered adiabatic, all energy is conservatively retained in the j system, i
05/10/85 32 -
The effect of metal heat is of particular concern in the modeling of the hot leg because the vapor phase separation, natural circulation interruption, and re-establishment would be affected. Metal heat is also of concern in the mod-eling of the pressurizer because of the effect of the heat transfer on the pressurizer response to insurges and draining.
The metal heat modeling in the hot leg is described in BAW-10192 (CRAFT 2). The I
heat transfer coefficient to the froth is calculated from the Jens-Lottes cor-relation or an input constant. The input steam heat transfer coefficient is held constant. The total heat transfer coefficient is the steam and frath con-tributions. Heat flow from the primary metal is deposited in the control vol-une fluid. -
x ..--.:
The metal heat modeling in the pressurizer is~ described in section 1.2.13 - -
Appendix I of BAW-10192 (CRAFT 2). It includes two variable area slabs of uni-form thickness associated with both regions of the two-node pressuizer model.
Heat flow from the primary metal is deposited in the fluid of each pressurizer -
region. If only one region exists, then the heat flow is deposited in the re--
- gion which is present.
The treatment of metal heat, including the modeling for the hot leg and the pressurizer, is acceptable for SBLOCA-EM analyses.
III.9 Break Flow II.K.3.30 expresses concern regarding the break flow representation used for SBLOCA-EM analysis. The concerns include accounting for the break geometry (F/L), location (hot leg, cold leg, top or bottom of pipe), and the upstream thermodynamic state and flow regime.
The upstream thermodynamic state and flow regime, two phase conditions, provide the properties and state of the fluid flowing through the break.
The B&W SBLOCA Evaluation Model utilizes the orifice equation for subcooled l
{ discharge and the Moody correlation for saturated /two phase discharge. A dis- I charge coefficient of 1.0 is applied to both models. This configuration has 4
been demonstrated to be conservative in Appendix 8 of BAW-10154, and is accept-l able for $8LOCA-EM analyses.
3 s In demonstrating conformance to 10 CFR 50.46 criteria, the actual break size l cnalyses are only important with respect to demonstrating ECCS performance. To l data, 84W 58LOCA-EM analyses have demonstrated conformance to 10 CFR 50.46 cri- !
{ teria for the full range spectrum of 58LOCA transients. The full range spec-j trum has demonstrated the acceptability of the 8&W ECCS design (HPI, LPI, and i CFT) to maintain core cooling.
l
} Item II.K.3.31 of NUREG-0737 requires confirmatory analyses with the revised l 58LOCA-EM to show that the criteria to 10 CFR 50.46 criteria are still met.
j Analyses of the full 58LOCA spectrum are not necessarily required. The analys-j es must demonstrate that the previous $8LOCA-EM analyses are conservative, or l new analyses will be required. Therefore, it is necessary to choose a break i
spectrum for analysis that will bound previous analyses concerns.
i -
The proposed break spectrum to be analyzed with the revised $8LOCA-EM is cur-j rently under consideration as a specific response to NUREG-0737, Item II.K.3.31.
i IV. Concerns in NUREG-0565 l NUREG-0565, Section 4.1.1.1, identifies concerns regarding the B&W small-break 4
evaluation model:
Concern No. 1 1
] Following postulated small break loss-of-coolant accidents, a primary mechanism I
for heat removal is natural circulation. The staff is concerned about the ability of the computer programs to correctly predict the various modes of nat-ural circulation and the interruption of natural circulation, if it occurs, i
I 05/10/85 34 -
8&W CRAFT 2 SER 1
l
y Experimental data for the verification of methods for two phase natural circu-lation are currently not available.
Response
In response to this concern, the CRAFT 2 code was upgraded. Included in this modification are a non-equilibrium pressurizer model, an upgraded two phase flow model, pump model, and a new steam generator model.
To demonstrate the ability of the upgraded CRAFT 2 code to predict the various modes of natural circulation observed during a small break, a post-test analy-sis of the Semiscale Mod-2A Natural Circulation Test S-NC-2 was performed.
This was a natural circulation test exhibiting single-and two phase natural circulation modes. CRAFT 2 predicted the same general trend as found in the test. The results calculated, for most data points, were within the uncer-tainties of the measurements. This analysis demonstrated that the upgraded
! CRAFT 2 code is capable of predicting various modes of ' natural circulation ob-served during the small-break LOCA transient and the transition from one mode .
to another. .
In addition, a benchmark of the small-break model against a B&W plant transient was performed. The high pressure reactor trip incident at Arkansas Nuclear One '
(ANO-1) on June 24, 1980, was the selected transient. The results calculated
! by the upgraded CRAFT 2 code were compared with the transient data to analyze the adequacy of the new steam generator and pressurizer models. It was demon-strated that the upgraded CRAFT 2 code is capable of predicting the natural cir-culation mode observed during the B&W plant transient.
Concern No. 2 The experimental verification of small break analysis methods with systems data is currently limited. The available small-break data from Semiscale Test S-02-6, although containing a number of deficiencies, is the best information now available. The analytical methods used to predict the results of this test OL/10/85 35 B&W CRAFT 2 SER
do not correctly predict the overall system depressurization rate, and the de-pressurization rate following core flood tank infection. These are significant )
parameters in that they affect the injection rate of the core flood tank fluid. '
Analyses by B&W of Semiscale Test S-07-108 and LOFT Test L3-1, have been sub-citted by B&W and are currently being evaluated by the staff.
w
Response
- In addition to the pre-test predictions of Semiscale Test S-07-10B and LOFT Test L3-1, B&W has also performed the post-test evaluation of these tests as requested in the " Letter to All Babcock & Wilcox Licensees" from R. W. Reid, Chief Operating Branch No. 4, Division of Licensing, February 24, 1981.
The post-test evaluation of LOFT Test L3-1 was submitted to the NRC in June 1981. It was concluded in this analysis that using initial and boundary condi-I tions consistent with the actual test, the results calculated by CRAFT 2 are in good agreement with the test data, thus confirming that CRAFT 2 is capable of
B&W and the B&WOG are also committed to the. MIST integral test facility program
, (the IST program) to provide additional data to confim that CRAFT 2 provides a conservative representation of SBLOCA behavior in a B&W PWR. '
Concern No. 3 l The appropriateness of the pressurizer model for analyses of small breaks at various locations is a potential concern. The equilibrium pressurizer model j assumed in the B&W analyses give.; somewhat different results from hand calcula-l tions assuming non-equilibrium conditions. These modeling differences may be
]
significant for various postulated breaks. Also, the representation of flood-i ing in the surge line could affect draining of the pressurizer. A flooding check is not made for the surge line in the computer program.
l 05/10/85 36 -
- - _ - --.___ --. = _ .
Response
In response to this concern, a non equilibrium pressurizer model was developed and incorporated in CRAFT 2. The model simulates pressurizer performance using a steam region and a liquid region. Heat and mass transfer between the two re-gions is controlled by steam-mixture interface parameters.
The second part of the concern regarding the addition of flooding in the surge line was also assessed. The result of this evaluation is shown in 8AW-10154, Appendix C. It is demonstrated in the report that, based on the geometry of the pressurizer surge line, countercurrent flow within the surge line cannot exist to any significant degree. Consequently, the flow in the B&W pressurizer 1
surge line will be in the only one direction. There is no need to add a flood-ing check to the surge line.
Concern No. 4 ,
The calculation of core level and core heat transfer are important features of .
the small break model. Limited experimental data are currently available to
~
justify these models. Although the current couparisons have been satisfactory, the experiments are not challenging to the codes. More experimental data must be obtained for further code verification. #
. m..
Response -
/
, In response to this concern, previous studies contained in BAW-10064 showing enalytical and experimental agreement of the core mixture level evaluation
, technique are referenced. These comparisons show that the level evaluation technique employed by the B&W model is capable of predicting the core mixture level.
j In order to provide the analytical and experimental agreement of the core heat transfer evaluation method, the small break core heat transfer model employed by B&W was compared for steady-state conditions to several tests performed at Oak Ridge National Laboratory (ORNL). These tests were designed to evaluate r ,
'/ ,
05/10/85 *
1 the steam cooling capability within the core in which the fluid level was al-lowed to fall below the top of the active core region and stabilized at an in-termediate location. These comparisons demonstrated that the use of the Dittus-Boelter correlation as the sole determinant of heat transfer is acceptable for cvaluating compliance with 10 CFR 50.46, no core uncovery. Consequently, the present heat transfer model is acceptable for licensing evaluation. If CRAFT 2 predicts core uncovery, additional core mixture level and heat-up analyses are performed.
Concern No. 5 The number of nodes used to represent the primary system for small break LOCA analyses should be sufficiently detailed to model the flashing of hot fluid in various locations. This modeling detail is necessary since the calculated sys-tem pressure during the decompression process is controlled by the flashing of the hottest fluid existing at any time in the model. The assumption of thermal cquilibrium requires that the fluid combined in a single node be represented by the average fluid properties. If fluid from several adjacent regions is com-bined in one node, the calculated systes process during a portion of the tran-sient may be lower than could occur if the small'er regions of hot fluid flashed cnd maintained the system at the corresponding saturation pressure. Thus, the modeling detail could have a significant effect on the calculated times for various events, such as ECCS actuation.
Response
As a result of the Small-Break LOCA Methods Program developed to address the requirements of NUREG-0737,Section II.K.3.30, significant code modifications cnd revisions were made to the existing small-break LOCA evaluation model. Be-cause of these modifications and revisions of the existing evaluation model, it was necessary to perform noding sensitivity studies to develop the base noding scheme which demonstrates convergence with respect to spatial detail. To ac-complish this goal, noding studies were performed by B&W.
05/10/85 38 -
{ l l
l I
! A noding. sensitivity study was performed to develop a converged steam generator l i model for 177- and 205-FA plants. These studies were conducted using a break that relies on the steam generator for RCS depressurization. The spatial de-tail for modeling the steam generator was increased to the code's capacity to assess the impact of additional spatial detail on the transient response. Based cn these studies, the steam generator models that adequately accounted for all the phenomena were chosen as the appropriate models for 177- and 205-FA plants.
To ensure that the effects of local flashing were accounted for, noding sensi-tivity studies of the upper plenum and upper head of the reactor vessel w'ere performed for 177- and 205-FA plants. The converged steam generator models were used for these studies. Based on these studies, a converged model was de-
~
veloped for the upper headgnd upper plenum of 177- and 205-FA plants by evalu-cting the results of various degrees of spatial detail in these regions.
Finally a noding study was conducted for the hot leg to ensure that its spatial ,.
detail is sufficient to model any interruption in natural circulation flow due
' to the formation of a steam pocket in the top of the inverted U-bend in the hot-legs.
Concern No. 6 During the recovery period from a small-break LOCA, the thermodyanmic equilib-rium assumed in fluid control volumes could result in errors in the predicted system pressure. This could, in turn, introduce errors in both the break dis-charge and safety injection flow. The rate at which the water is refilling the system can affect steam condensation. If the condensation efficiency is less than 100%, system pressure would be higher than predicted.
l Concern No. 7 The reduction in the primary system pressure determines the rate and amount of core flood tank water injected. Core reflooding is dependent on this flow. As discussed in NUREG-0611, the sensitivity analyses performed demonstrate the in-fluence of core flood tank injection. The amount of steam present at the in-05/10/85 39 -
jection location is the predominant factor that determines the core flood tank cass delivery. The results of an analysis will be influenced by the model and the modeling assumptions used to calculate the core flood tank flow. Addition-ci studies will be required to obtain the necessary information to perform an Appendix K analysis. Additional work in this area is underway at EG&G Idaho since more recent experimental data, including LOFT Test L3-1, indicate less depressurization than Semiscale Test S-02-6.
Response (to Concerns 6 and 7)
These concern deal with the adequacy of the ECCS injection model used in small-break LOCA evaluations. During the NRC/B&W Owners Group meeting of December 16, 1980 these concerns were clarified. The concern addressed the possibility of a large pressure disturbance after CFT actuation due to the ECCS injection location. In order to respond to this concern, previous B&W small-break tran- '
sient evaluations were reviewed to determine whether they exhibit the system disturbance of concern. The review of these previous analyses showed that the downcomer liquid volume remains high throughout the transient. As a result of this high liquid content, the use of the thermodynamic equilibrium assumption does not illustrate the system disturbance of concern. The system depessuri-zation characteristics are not significantly altered. Thus, the ECC injection a
modeling employed in the B&W evaluation model provided an adequate representa-tion of the actual phenomena and the system responses.
In NUREG-0565, Section 4.2.11, the staff expresses the following concern:
Concern All sources of noncondensable gas generation in the RCS must be taken into con-sideration, including radiolytic decomposition, to determine the effect on the small-break transient. In addition, it was recommended that the licensees pro-vide " confirmatory information to verify the predicted condensation heat trans-fer degradation" in responding to this concern.
ME; Response
>(
In response to this concern, all sources of noncondensable gas, including the radiolysis have been accounted for to assess the impact'of noncondensables on the small-break transients. The condensation heat transfer degradation model used to assess the impact of noncondensables on SBLOCA transients has been de-veloped by investigating the available literature of industry data including the B&W Single-Tube Condensation Test results at ARC.
V. Concerns in NUREG-0623 The following two concerns are identified in NUREG-0623:
Concern No. 1 In NUREG-0623, Section 4.2.2, the staff expressed a concern that the two phase flow' treatment in CRAFT 2 is not adequate to calculate the distribution of liquid in the primary systes during a small break with reactor coolant pumps operating.
Response
i In response to this concern, the drift-flux model was developed and incorporat-cd in the CRAFT 2 code. The adequacy of the two phase flow model was demon-strated by the successful prediction of the LOFT L3-6 test submitted to the NRC in April 1981.
Cnncern No. 2 In NUREG-0623, Section 4.3.5, the NRC raised a concern that the two phase pump model currently used in the evaluation of small-break transients does not ade-quately model the degradation of pump head and hydraulic torque during two-phase operation.
05/10/85 41 -
_ ___y . _ _
Response
In response to this concern, a new pump model was developed and incorporated into CRAFT 2. The new pump model will account for the degradation of pump head
)
cnd torque in a two phase environment. I VI. Conclusions The Babcock and Wilcox Owners Group, through Babcock and Wilcox, have modified the CRAFT 2 small-break LOCA Evaluation Model computer program in response to NUREG-0737 TMI-2 Action Item II.K.3.30, " Revised Small-Break Loss-of-Coolant Accident Methods to Show Compliance with 10 CFR 50, Appendix K." These revi-sions are based on the NRC recommendations and concerns identified in NUREG-0565 and NUREG-0623. ^" -
The modifications to CRAFT 2 include a nonequilibrium pressurizer model, a mech-cnistic steam generator model which incorporates a condensation heat transfer
~
model, a drift flux / level fomation model to account for two phase flow and primary to secondary heat transfer, and a new pump model to account for two-phase flow degradation in the head and torque cu'rves.
i A noncondensable gas model, to account for the degradation in condensation heat transfer, was also added to the CRAFT 2 computer program. The model is based on t
data obtained by B&W for an OTSG tube geometry. Past experiences by B&W have demonstrated that the amount of noncondensable gases occurring during a SBLOCA- ,
EM evaluation are insufficient to significantly effect the calculation, and l therefore noncondensable gases are not tracked by the CRAFT 2 computer program.
Recent testing at Semi-scale has also shown that, for the expected amounts of noncondensable gases occurring during a SBLOCA, the effect on the system tran-sient are negligible. At this time the NRC has not reviewed, in detail, the noncondensable gas model.
B&W experiences with SBLOCA-EM evaluations have shown that the Core Flood Tanks (CFTs) do no empty and that the calculated peak cladding temperatures remain 05/10/85 42 B&W CRAFT 2 SER
1
\
l below the_ metal-water reaction temperatures. Therefore these potential sources of noncondensable gases may be omitted.
B&W endorses the conclusions in NUREG-0565 concerning the amount of noncondens-cble gases which could accumulate in the primary system. This evaluation con- '
servatively estimated the maximum volume of noncondensable gases from all 3 potential sources, with the exception of the radiolytic decomposition of the safety injection water. The result of this evaluation was that the amount of noncondensable gases is not sufficient to block natural circuation in the hot leg U-bend, if all the noncondensable gases were conservatively assumed to i accumulate at that location. In addition, the B&W position regarding the radi-olytical decomposition of the injected water is that this additional source of noncondensable gas does not alter the conclusions in NUREG-0565.
Acceptance and reference to this Safety Evaluation Report by B&W and by the B&WOG affirms the conclusion regarding the insignificant effect of noncondens-cble gases for a SBLOCA-EM analysis.
In support of the new models, and the CRAFT 2 computer program in general, the B&WOG has performed an extensive verification and benchmark program. Separate effects tests as well as integral system test comparisons were provided. The verification and benchmark program demonstrate that the revised CRAFT 2 computer program is capable of predicting those phenomena identified as being important to the SBLOCA-EM analysis. These are condensation heat transfer in the steam generator, hot leg phase separation, the interruption and re-establishment of l natural circulation, single and two phase natural circulation flow, counter- I current flow, nonequilibrium effects for Core Flood Tank injection, and core i steam cooling heat transfer. For the most part the new models developed and icplemented in CRAFT 2 result in a conservative evaluation for SBLOCA-EM analysis.
Other models, such as the two phase pump degradation model, provided for a better, realistic representation of the effects.
The SBLOCA integral system tests used in this evaluation program, LOFT and Semiscale, do not represent the unique B&W NSSS design feature, such as the vent valves, the hot-leg U-bend and the once-through steam generator (OTSG).
05/10/85 43 -
However these comparisons are acceptable for demonstrating the capability of CRAFT 2 to predict the SBLOCA phenomena of concern.
Additional support for the steam generator model and pressurizer model are based on B&W operating reactor data. These demonstrate the capability of CRAFT 2 to model the pressurizer response, to model the steam generator heat transfer, and to model natural circulation (single phase) flow.
Recently completed studies by the NRC using the RELAP5/ MOD 2 computer program have shown similar response characteristics to CRAFT 2 for a SBLOCA-EM calcula-tion. Primary system repressurization for a lower-loop plant was observed, the interruption of natural circulation and re-establishment of natural circulation was observed, and boiler-condenser heat transfer was observed to occur at lower secondary side water levels than CRAFT 2 'would predict. The overall response was very similar to an equivalent CRAFT 2 analysis. The RELAP5/ MOD 2 computer program employs a dynamic vent valve model as well as current state-of-the-art models for flow and heat transfer.
The verification and benchmark evaluation coupled with the RELAPS/ MOD 2 qualita-tive assessment providus suitable justification for the acceptance of the CRAFT 2 computer program and the B&W system nodal models to be used for SBLOCA-EM cal-culation to be provided in response to NUREG-0737, II.K.3.31.
' The B&W Integral System Test program (IST) will provide data concerning SBLOCA bihavior for the B&W specific geometry. It is B&W's position that this program will not result in the identification of any new phenomena related to the B&W design which will alter the conclusion that the CRAFT 2 SBLOCA-EM model is con-servative and in compliance with 10 CFR 50 Appendix K. This statement is based on a review of the test results from the GERDA and OTIS test programs. GERDA has not been used for benchmark because no reactor coolant pumps were modeled and is not representative of US-B&W PWRs. {
B&W is currently evaluating the ben-efit, if any, from benchmarking CRAFT 2 to OTIS test data. OTIS is more repre-sentative of US-B&W PWRs and can supply natural circulation verification date.
B&W and the NRC will continue to monitor the IST program results, as well as other experimental programs related to the B&W design, to confirm the accep-tance of the CRAFT 2 SBLOCA-EM computer program.
B&W has made a long standing commitment to continually review experimental data cnd code predictions of these data by both B&W and other organizations, as they apply to the CRAFT 2 evaluation model or portions of it (see Section 7 of BAW-10154). The MIST program (IST) is one such source of new information. Accept-cnce and reference to this Safety Evaluation Report by B&W and by the B&WOG affirms this commitment to perform a suitable comparision to the MIST data, as identified in Section III.5.g of this report, to demonstrate that CRAFT 2 does provide a conservative representation of SBLOCA behavior in a B&W PWR.
S The nvised CRAFT 2 computer program for small-break LOCA analysis to demon-strate compliance with 10 CFR 50.46 has been shown to be in conformance with '
the Evaluation Model criteria as specified in 10 CFR 50 Appendix K. The re- ,4. .
vised CRAFT 2 computer program is acceptable for reference in future B&W ECCS licensing evaluations. . .
It is the intention of the B&WOG to provide.geneYic analysis, by plant configu-ration, in response to NUREG-0737 II.K.3.31. which will demonstrate that the a
current FSAR small-break LOCA results are conservative. This will be accom-plished by selecting a limited break spectrum for the evaluation. The break spectrum will be selected to exercise the ECC system (HPI, LPI and CFTs) and span the previously identified limiting break size. Should this evaluation not confirm that the current FSAR results are conservative, each licensee will be r; quired to perform a plant specific ECCS SBLOCA-EM covering the complete small break spectrum.
l
- n l 05/10/85 45 -
l VII. REFERENCES
- 1. " Clarification of the TMI Action Plan Requirements," NUREG-0737, U.S. '
Nuclear Regulatory Commission, Office of Nuclear Reactor Regulation, !
November 1980.
- 2. J.J. Cudlin, M.I. Meerbaum, J.A. Klingenfus, and M.E. Mays, CRAFT 2 --
Fortran Program for Digital Simulation of a Multinode Reactor Plant During a Loss of Coolant, BAW-10092P, Rev. 3, Proprietary Babcock & Wilcox, Lynchburg, Virginia, October 1982.
- 3. N.K. Savani, J.R. Paljug, and R.J. Schonaker, B&W's Small-Break LOCA ECCS Evaluation Model, BAW-10154, Babcock & Wilcox, Lynchburg, Virginia, .
November 1982. _
- 4. " Generic Evaluation of Small-Break Loss-of-Coolant Accident Behavior in Babcock & Wilcox Designed 177-FA Operating Plants," NUREG-0565, U.S.
Nuclear Regulatory Commission, Office of Nuclear Reactor Regulation, January 1980.
- 5. " Generic Assessment of Delayed Reactor Coolant Pump Trip During Small-Break Loss-of-Coolant Accidents in Pressurized Water Reactors,"
NUREG-0623, U.S. Nuclear Regulatory Commission, Office of Nuclear '
Reactor Regulation, November 1979.
- 6. Letter from J.H. Taylor, Manager, Licensing Services, B&W, to C.O. Thomas, Chief, Standardization and Special Projects Branch, DL, NRC, dated July 25, 1984. I
- 7. C.D. Morgan and G.C. Rush, " Experimental Measurements of Condensation Heat Transfer With Noncondensable Gases Present in a Vertical Tube at High Pressure," Heat Exchangers for Two-Phase Applications, ASME Symposium, Volume HTD, 22 (1983).
- 8. C.D. Morgan, "An Analysis of Condensation Heat Transfer With Noncondensable Gases Present in a Vertical Tube at High Pressure," Heat Exchangers for Two-Phase Applications, ASME Symposium, Volume HTD, 27 (1983).
- 9. D.J. Shimeck and G.W. Johnsen, " Natural Circulation Cooling in a Pressur-ized Water Reactor Geometry Under Accident-Induced Conditions, Nuclear Science and Engineering, 88, 311-320 (1984).
- 10. N.K. Savani and R.C. Jones, Surge Line Modeling, Task 18 of NUREG-0565 Program, Document No. 51-1126077-01 prepared by the Babcock & Wilcox Co.
for the Owner's Group of Babcock & Wilcox 177 and 205 Fuel Assembly NSS Systems, July 1981.
05/10/85 R-1 B&W CRAFT 2 SER 4
T 11. Evaluation and Justification of the B&W ECCS Injection Model, Proprietary Document No. 77-1136045-00, Babcock & Wilcox, Lynchurg, Virginia, August 1982.
- 12. B&W's Best-Estimate Prediction of the LOFT L3-6 Nuclear Small Break Test Using the CRAFT 2 Computer Code, Document No. 12-1124993-01, Babcock &
Wilcox, Lynchburg, Virginia, March 1981.
- 13. Internals Vent' Valve Evaluation, BAW-10005, Rev. 1, Babcock & Wilcox, Lynchburg, Virginia, June 1970. )
- 14. Technical Specifications for Oconee Nuclear Power Station, Appendix A, BWNP-20004, Babcock & Wilcox, Lynchburg, Virginia, June 1976.
- 15. W.L. Jensen, memorandum for B.W. Sheron, Chief, Reactor Systems Bra'nch, DSI, "Small Break LOCA Sensitivity Study - B&W Lowered Loop Plants,"
April 10, 1985.
- 16. Evaluation of SBLOCA Operating Procedures and Effectiveness of Emergency Feedwater Spray for B&W-Designed Operating NSSS, Document No.
77-1141270-00, Babcock & Wilcox, Lynchburg, Virginia, February 1983.
- 17. Bubble Dynamics -- Phenomena, Experimental Benchmarks, Assessment of Sen-sitivity, Document No. 12-1132565-00, Babcock & Wilcox, Lynchburg, Virgin-ia, April 1982.
Wilcox, Lynchburg, Virginia, April 1982.
- 19. Supporting Data for AFW (EFW) Models -- Auxiliary Feedwater Axial Flow Distribution, Document No. 12-1132543-00, Babcock & Wilcox, Lynchburg, Virginia, 1982. <
- 20. Supporting Data for AFW (EFW) Models, Auxiliary Feedwater Penetration, Document No. 12-1132513, Babcock & Wilcox, Lynchburg, Virginia, April 1982.
- 21. CRAFT 2 Prediction of Alliance Research Center Loss of Feedwater Data, Document No. 12-1132544-00, Babcock & Wilcox, Lynchburg, Virginia, April 1982.
- 22. B&W's Post Test Evaluation of LOFT Test L3-1, Document No. 51-1125988-00, Babcock & Wilcox, Lynchburg, Virginia May 20, 1981.
- 23. T. E. Geer, et al. , B&W's Post Test Analysis for Semiscale Test S-07-100, Document No. 86-1125888-00. Babcock & Wilcox, Lynchburg, Virginia, May 20, 1981.
- 24. Integral Systems Testing Program for B&W Designed NSS Systems, Test Advi-sory Group Final Report, BAW-1787, Babcock & Wilcox, Lynchburg, Virginia, i June 1983. '
05/10/85 R-2 -
. 25. Letter from F.R. Miller, Chairman, B&W Owners Group Analysis Committee to P. Kadambi, NRC, dated January 3, 1985.
- 26. BAW-10155, " FOAM 2- Computer Program to Calculate Core Swell Level and Mass Flow Rate During a Small-Break LOCA," Babcock & Wilcox, Lynchburg, Virgin-ia, November 1982.
, 27. BAW-10094, " THETA 1 Computer Code for Nuclear Reactor Core Thermal i
Analysis- B&W Revisions to IN-1445 (Idaho Nuclear, C.J. Hocevar and T.W. Wineinger)," Rev-3. , Babcock & Wilcox, February 1981.
l G
e i
4 1 -
4 05/10/85 R-3 B&W CRAFT 2 SER
1 l
l Mr. J. M. Griffin Arkansas Power & Light Company Arkansas Nuclear One, Unit 1 !
cc:
Mr. J. Ted Enos, Manager Licensing Arkansas Power & Light Company P. O. Box 551 Little Rock, Arkansas 72203 Mr. James M. Levine, General Manager Arkansas Nuclear One P. O. Box 608 Russellville, Arkansas 72801 Mr. Nicholas S. Reynolds Bishop. Liberman, Cook, Purcell & Reynolds 1200 Seventeenth Street, NW Washington, D.C. 20036 Mr. Robert B. Borsum Babcock & Wilcox Nuclear Power Generation Division Suite 220, 7910 Woodmont Avenue Bethesda, Maryland 20814 Resident Inspector U.S. Nuclear Regulatory Commission P. O. Box 2090 Russellville, Arkansas 72801 Regional Administrator, Region IV U.S. Nuclear Regulatory Commission Office of Executive Director for Operations 611 Ryan Plaza Drive, Suite 1000 Arlington, Texas 76011 Mr. Frank Wilson, Director Division of Environmental Health Protection Department of Health Arkansas Department of Health 4815 West Markham Street Little Rock, Arkansas 72201 Honorable Ermil Grant Acting County Judge of Pope County Pope County Courthouse Russellville, Arkansas 72801 l
l l
1 l
- .