ML19345G723
| ML19345G723 | |
| Person / Time | |
|---|---|
| Issue date: | 06/18/1980 |
| From: | Budnitz R NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Harold Denton Office of Nuclear Reactor Regulation |
| References | |
| RIL-092, RIL-92, NUDOCS 8104220169 | |
| Download: ML19345G723 (24) | |
Text
~
e g
e ao UNITED STATES
[g n, b
NUCLEAR REGULATORY COMMtsslON o,
y W ASHINGTON, D. C. 20555 ys,;' :., p: ~ ~
M 131980 J UA)
MEMORANDUM FOR: Harold Denton, Director Office of Nuclear Reactor Regulation FROM:
Robert J. Budnitz, Director Office of Nuclear Regulatory Research
SUBJECT:
RESEARCH INFOPRATION LETTER TRAC-Pl A 1.
INTRODUCTION A.
Need for a Best-Estimate Code The NRC development of a best-estimate (BE) computer code for analysis of reactor transients and accidents, such as TRAC-Pl A (Ref.1), fulfills three related needs. The first is to provide a firm numerical and physical basis for the margin of safety built into the conservative, or Evaluation Model (EM) computer codes. These conservative codes (Ref. 2) are used extensively in the licensing process and are developed under the philosophy tnat if one doesn't know for sure how to accurately calculate a complicated pf pical process, one should use a conservative bounding approximation for that process.
For example, since there is some justifiable uncertainty as to exactly how much ECC water bypasses out the broken cold-leg during the depressurization period, the EM model assumes that all the water leaves the system during this 10 to 15 second time period. The TRAC-PIA BE code, on the other hand, treats this same partial bypass period as a complicated two-phase flow process to be calculated with a multidimensional code that has been assessed against bypass data in several scaled experimental facilities. This need to establish the margin of safety in conservative calculations was recognized by NRR (Ref. 3), the American Physical Society (Ref. 4)
- and the ACRS (Ref. 5) who all requested development of a BE code to fill this need.
The second need for a BE code is to predict, analyze and comprehend data from scaled experimental facilities. These test facilities investigate both integral system and separate effects phenomena for various LOCA conditions, as well as for non-LOCA conditions. This second need is central to the NRC/RES program, which supports a coordinated research effort of both analysis and experiment on safety-related issues. Both the American Physical Society (Ref. 4) and the ACRS (Ref. 5) support the requirement for this combined effort and, thus, recognize this second need for a BE code, like TRAC-PlA, which can assimilate all this experimental data into a comprehensive model of a reactor.
18Il 04 2 2 0
ThFs leads to the third and most important need for a BE code:
the need to analyze and predict consequences for both real and postulated accidents in full-scale LWR's. This last need permeates all activities in BE code development and assessment. The TRAC code is being developed to include all major phenomena expected to occur during a severe accident in a LWR, in sufficient detail to provide as accurate a calculation as presently possible. The development is being assessed against scaled data to provide confidence in its extrapolation to full-scale LWR's.
I l
B.
TRAC-PlA vs. RELAP-4 RELAP-4 is a BE code which has been used extensively to calculate transient behavior in full-scale LWR's.
Its well-known modeling deficiencies (one-dimensional flow gecmetry, homogeneous equilibrium flow) have been overcome by the development of special global modeling based on experience and engineering judgment.
TRAC-PIA was developed as a conscious attempt to improve the RELAP-4 modeling deficiencies. TRAC would then provide an alternate, and advanced, BE code to analyze full-scale LWR's which would be based more on local physical models rather than on global engineering models.
TRAC-PlA provides an advanced analysis capability for pressurized-water reactors. The advanced features of TRAC-P1 A include nonhomogeneous, nonequilibrium and multidinensional hydrodynamics with flow-regime-dependent constitutive relations; quench-front tracking capability for both bottom flood and falling films; consistent treatment of entire accident sequences, including the generation of initial steady-state conditions; and modular design which allows representation of a wide variety of experimental configurations, ranging from single components to multiloop systems.
Further details of the TRAC-PlA capabilities are described in Appendix I and in Reference 1.
C.
TRAC-PlA vs. Future TRAC Versions TRAC-PlA has been tested against an initial set of separate-and integral-effects experiments. Further assessment of the code through
- pretest and post-test predictions of other experiments has also occurred.
These testing and assessment activities have been quite successful, but have also pointed out areas where improvements, described below, are required in future versions of the TRAC code. These improve-ments are based on the experience of several laboratories which have used TRAC-PlA over the past year. They are being incorporated into the next version, TRAC-PD2, which will be ready for release in the sumer of 1980.
w
-n--
,,._y,
~
II. RESULTS,.
The details of the assessment of TRAC-PlA predictive capabilities are covered in Appendix II and in Reference 6.
The results can be summarized as follows:
- 1. Hydrodynamics Blowdown Phase large Break (-100%)
Very good for a wide range of experiments.
Characteristic dimensions range from 0.02m to 0.5m.
Fluid conditions range from highly sub-cooled liquid to two-pnase mixture, to saturated and superheated vapor.
It is important to note, however, that TRAC underpredicted the flow in the short nozzle Marviken tests, where nonequi' ?brium effects were important. Experimental facilities: Edwards, CISE, Marviken, Semiscale, LOFT.
Small Sreak Very good for single phase flow.
Intufficient data comparisons for two-phase flows.
Closely coupled to calculate inlet fluid conditions for vertically oriented nozzles (e.g. due to level swell). TRAC
~
underpredicted cold-leg break flow during first 5 seconds of L2-3 (large-break) and the first 150 seconds of L3-1 (small-break).
The cause for both disagreements with data was too much local void generation. Experimental facilities: CISE, LOFT, Analytical.
Refill /Byoass Phase Downcomer (3-D calculation)
Excellent for small scales (1/15-3/15).
Insufficient data comparisons at full scale. Wide range of ECC subcoolings and injection rates.
Tends to slightly overpredict delivery. Experimental facilities:
_ Creare, Battelle LOFT.
i Pioes (1-0 Calculation)
Poor for all flow regimes except dispersed flow. Tends to underpredict penetration in countercurrent flow.
Insufficient data for large (.5m) pipes. Experimental facilities: Semiscale Mod-3, INEL Air / Water Tests, Dartmouth.
Reflood Phase Strongly coupled to heat transfer. Very good results for high flooding rates; poor results for low flooding rates and lower
plenJE ECC injection. Underpredicts liquid carryover and pre-cursory cooling.
Numerical pressure spikes caused by too much vapor generation during a time-step within a given control volume.
Experimental facilities: FLECHT-SET, FLECHT-SEASET, UCB, LOFT.
- 2. Heat Transfer Blowdown Phase Nucleate boiling regime accurately modeled: Time to DNB very good for a wide range of geometries and fluid :
'itions. Peak clad temperature normally occurs during thi: phase - generally very good agreement with data.
Rod rewets in LOFT are accurately modeled using Iloeje Tmin.
Experimental facilities: CISE, Semiscale, LOFT.
Refill /Byoass Phase Heat transfer coefficients between rods and fluid calculated accurately for film toiling and superheated vapor.
Underpredicts precursory cooling due to entrained liquid and sputtering on clad. Experimental facilities: Semiscale, LOFT.
Reficod Phase Core conditions at the beginning of reflood may be considerably different than previously thought due to early rod rewets. Thus, PCT may not occur during reflood.
Reflooding rate generally underpredicted by a substantial amount for cold leg ECC; steam generation due to requenching tends to expel liquid from the core. Nuclear fuel rod model needs to include dynamic fuel gap dimension. Experimental facilities: FLECHT-SET, FLECHT-SEASET, U.C. Berkeley, Semiscale, LOFT.
- 3. Identified Needs for Improvement a) The numerical techniques need to be improved to tighten up mass conservation and to decrease the computer running time I
b) More mechanistic treatment of the reflood process is needed to allow automatic calculation of the quench front c) A more realistic model of fuel gap conductance d)
Improved heat transfer correlations e)
Improved flow regime recognition criteria e
y w
-m
e Harold Denton 5
f) -Additional models need to be incorporated to handle counter-current flow of liquid and vapor in a horizontal oipe, as occurring in the " reflux boiler" process during small break LOCA.
Other areas in need of improvement are identified in Section III below and in Appendix II.
III. RECOMMENDATIONS TRAC-PIA has been designed and tested for analysis of large-break LOCA's in PWR's. The major processes it calculates include: the blowdown process, including choked flow from pipes; ECC bypass and lower plenum refill; and reflooding of the core by the ECC.
PWR calculations with TRAC-PlA, which is a BE code, have shown that the peak clad temperature occurs early in the transient during the blowdcwn period. This contrasts with conservative EM codes which calculate peak clad temperature much later in time during the reflood period. This is despite the fact that TRAC-PlA overpredicts clad temperature for icw flooding rate tests in FLECHT, due to under-prediction of precooling by entrained drops in the upficwing core steam.
One reason why TRAC-PlA calculates lower reflood temperatures than conservative codes is because of its multidimensional BE model for downcomer flow during the ECC bypass period; a model that does well against Creare and BCL downcomer data. Although the code calculates some ECC bypass, it allows much more water to remain in the vessel than do conservative licensing calculations.
It should be pointed out that TRAC-P1 A does include heat transfer frem the downcomer walls.
Initial PWR calculations with TRAC-Pl A used between 400 and 600 computa-tional nodes in order to get a reliable base-case indication of the codes' capabilities for the entire LOCA transient. This led to long computer run time - about 15 to 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br /> of CPU time on a CDC-7600 computer.
Subsequent noding studies showed that fewer nodes could be used and still get a reliable PWR simulation. One successful study with about 150 nodes gave a running time for the entire LOCA (blowdown through reflood) of about 5 CPU hours. Even this running time is recognized as being too long for many applications, and a fast-running version of TRAC is currently being developed.
There is no kinetics feedback modeled in TRAC-Pl A, so this version of the code cannot be used for such problems as ATWS and RIA; a future version of TRAC will include models for these problems.
-= -
-w m.
m
When the code was applied to small-break tests, its choked flow rodel was found t5 overpredict void formation near the break and thus underpredict break mass flow rate. This is being corrected for later versions of TRAC which will be more applicable to small-break analysis.
The code can conceptually handle natural circulation and, in fact, did calculate natural circulation conditions for the TMI accident. However, the accuracy of its natural circulation predictions has not yet been tested against data.
r r
/p YT g,,4(
<W Robert J. Budnitz, Director Office of Nuclear Regulatory Research
Enclosures:
1.
Appendix I," Description of TRAC-Pl A Capabilities - Summary" 2.
Appendix II," TRAC-Pl A Developmental Assessment - Summary" cc w/encis:
D. F. Ross, NRR P. Check, NRR T. P. Speis, NRR R. Mattson, NRR G. W. Knighton, NRR m
,----u a
REFERENCES 1.
a) TRAC-Pl A, An Advanced Best-Estimate Computer Program for PWR LOCA AnaTysis; Safety Code Development Group, Energy Division, Los Alamos Scientific Laboratory, NUREG/CR-0665, LA-7777-MS, May 1979.
b) Development and Assessment of TRAC, J. Vigil and R. Pryor, Nuclear Safety, 21 (1), 1980. This latter article is quoted extensively in this RIL.
2.
WRAP-A Water Reactor Analysis Package, M. M. Anderson, Savannah River Laboratory, DPST-NUREG-77-1, June 1977.
3.
Letter, E. Case to S. Levine, "NRR Requirements for LOCA Analysis Computer Programs," 6/23/77.
4.
Pecort to tne 'nerican Physical Society by the study grcup en LWR Safety, Rev. of Mod. Pnys., 47 (Supp.1), Summer 1975.
5.
ACRS, " Review and Evaluation of the NRC Safety Research Program," NUREG-0392 (1977), NUREG-0*96 (1978), NUREG-0603 (1979).
5.
TRAC-P1 A Develcpmental Assessment, J. Vigil, K. Williams at al., Energy Divisicn, Les Alames Scientific Laboratcry, NUREG/CD-1059, LA-8056-MS October 1979.
7.
Constitutive Relations in TRAC-Pl A, U. Rohatgi and P. Saha, Brookhaven Nati nal Lateratcry, August 1979.
8.
J. H. Mahaffy and D. R. Liles, " Application of Implicit Numerical Metneds for Problems in Two-Phase Flow," Los Alamos Scientific Laboratory report LA-7770-MS, NUREG/CR-0763 (April 1979) 9.
R. J. Pryor, D. R. Liles, and J. H. Mahaffy, " Treatment of Water Packing Effects," Trans. ANS 1978 Winter Meeting, Washington, D. C., 30, 208 (1978)
- 10. A. R. Edwards and T. P. O'Brien, " Studies of Phenomena Connected with the Depressurization of Water Reactors," J. British Nucl. E. Soc., 9,125 (April 1970).
~~
- 11. A. Premoli and W. T. Hancox, "An Experimental Investigation of Subcooled Blowdown with Heat Addition," Submission to Committee on Safety of Nuclear Installations, Specialists Meeting on Transient Two-Phase Flow, Toronto, Ontario (August 1976).
- 12. L. Ericson, L. Gros D' Aillon, D. Hall, J. Ravnsborg, O. Sandervag, and H. Akesson, "The Marviken Full-Scale Critical Flow Tests Interim Report; Results from Test 4," Marviken draft interim report MXC-204 (May 1978).
.=
REFERENCES Cont' 13.
S. A. Naff and P. A. Pinson, "l-1/2 Loop Semiscale Isothermal Test Program.and System Description in Support of Experiment Data Reports,"
Aero' jet Nuclear Company report ANCR-lla3 (February 1973).
14.
L. J. Ball, D. J. Hanson, K. A. Dietz and D. J. Olson, "Semiscale Program Description," Idaho National Engineering Laboratory report TREE-NUREG-1210 (May 1978).
15.
C. J. Crowley, J. A. Block and C. N. Cary, "Downcomer Effects in a 1/15-Scale PWR Geometry: Experimental Data Report," Creare, Inc. report NUREG-0281 (May 1977).
16.
R. A. Cudnik, L. J. Flanigan, R. C. Dykhuizen, W. A. Carbiener and J. S. Liu,
" Topical Report on Baseline Plenum Filling Behavior in a 2/15-Scale Model of a Four Loop Pressurized Water Reactor," Battelle Columbus Laboratories report B?il-1997 (NUREG/CR-0069) ( April 1978).
17.
J. O. Cermak, "PWR Full-length Emergency Cooling Heat Transfer (FLECHT)
Group I Test Report," Westingnouse Electric Company report WCAP-7435 (January 1970) 18.
H. C. Robinson, " LOFT Systems and Test Description (Loss-of-Coolant Experiments Using a Core Simtiator)," Idaho National Engineering Laboratory report TREE-NUREG-1019 (t'ovember 1976).
19.
J. R. Ireland and P. B. Bleiweis, " TRAC Calculations of U. S. Standard Problem 8," recorted in " Nuclear Reactor Safety Quarterly Progress Report, July 1-September 30, 1978," Los Alamos Scientific Laboratory report LA-7567-PR (NUREG/CR-0522), p.12, (December 1978).
20.
K. A. Williams, " TRAC Calculation of Standard Problem 6," reported in
" Nuclear Reactor Safety Quarterly Progress Report, July 1-September 30, 1978,"
Los Alamos Scientific Laboratory report LA-7567-PR (NUREG/CR-0522), P.21 (December 1978).
21.
K. A. Willians, " Pretest and Post-test Predictions of LOFT Nuclear Test L2-2,"
reported in " Nuclear Reactor Safety Quarterly Progress Report, October 1-December 31, 1978," Los Alamos Scientific Laboratory report cA-7769-PR (NUREG/CR-0762), p.49 (May 1979).
22.
D. A. Mandell and K. A. Williams, "L2-2 Parametric Study," reported in
" Nuclear Reactor Safety Quarterly Progress Report, January 1-March 21,1979,"
Los Alamos Scientific Laboratory report LA -/867-PR (NUREG-CR/0868), p.18 (July 1979).
23.
D. L. Reeder, " LOFT System and Test Description (5.5-ft Nuclear Core 1 LOCES)," EG&G Idaho, Inc. report TREE-1208 (NUREG/CR-0247) (July 1978).
- 24. TRAC Developmental Code Assessment, K. A. Williams, LA-UR-79-2969, Seventh WRSR InfJrmation Meeting, November 1979.
44,,
APPENDIX I DESCRIPTIGN 6F TRAC-PlA CAPABILITIES -
SUMMARY
A. General TRAC can be characterized as an advanced, best-estimate LWR systens computer program. Within limitations imposed by computer running time, it incorporates state-of-the-art metinds and models.
The models in TRAC are designed to yield realistic solutions as opposed to conservative evaluation models used in licensing codes. TRAC mainly differs from other existing LWR systems codes (e.g., RELAP-4 code) in its more detailed geometrical models of system components and its more basic treatment of two-phase thermal hydraulics.
User-selected options are minimized in the basic fluid dynamics and heat transfer modeling. This approach, as opposed to that which allows modeling options, places great demands on the basic thermal-hydraulic modeling because the code must determine local flod topology and surpty appropriate constitutive relations. Thus, tae development of accurate flow-regine-dependent constitutive relations is vital to the TRAC effort. The ultimate goal of the TRAC effort is to produce comouter programs that have a demonstrated capability to adequately predict the results of a broad range of experinents with no tuning of basic physical models from cne test to another.
Because of the advanced features of TRAC, most of the physical pnencmena that are inportant in LOCA analysis can be treated.
The code can be used to obtain steady-state solutions to provide self-consistent initial conditions for subsequent transient calculations.
Both a steady-state and transient calculation can be performed in the same run if desired. Efficient solution strategies, ranging from semi-implicit to fully implicit, are used.
An important characteristic of TRAC is the ability to address the entire LOCA (blowdown, bypass, refill and reflood) in one continuous and consistent calculation. This eliminates the need to interface and combine calculations performed with different codes for each major accident phase. Trips can be specified to simulate protective system actions or operational procedures (e.g., opening or closing of a valve).
A sophisticated graphics package, including movie generation capability, is available to help analyze and digest the large amount of output information generated during a TRAC run. A dump restart feature allows the user to restart a calculation from any point in a transient. This feature is very useful in performing parametric studies and in minimizing loss of computer time due to hardware failure or input studies.
~
l e
2 TRAC is, designed to run on a CDC 7600 computer, but standard programming techniques are being used to ease its conversion to other computers.
ORNL has recently converted TRAC-PlA to the IBM computer. All storage arrays are dynamically allocated so that the only limit on problem size is the available core memory. A capacity of 60,000 words of small-core memory and 220,000 words of large-core memory is sufficient to handle most problems of interest.
B.
Component and Functional Modularity TRAC is completely modular by component and by function.
Component modules, which consist of subroutines or sets of subroutines, are avail-able to model vessels (with associated internals), steam generators, pressuri:ers, etc. Component modules currently available in TRAC are described in Table I.
The user can construct a wide variety of config-urations by connecting an arbitrary number of these components in a meaningful way. Thus, the user can solve problems ranging from a simple pipe blowdown to a LOCA in a multiloop PWR. Component modularity allows cceronent models to be improved, modified, or added without disturbing the rest of the code.
Functional modules, which also consist of subroutines or sets of subroutines, are available for the multidimensional two-fluid hydrodynamics, one-dimensional drif t-flux nydrodynnics, thermo-dynamic and transport properties, wall heat transfer, etc. These functional modules are described in Table 2.
Functicnal modularity allows the code to be easily upgraded as improved correlations and experimental information beccme available.
C.
Multidimensional Fluid Mechanics A three-dimensional cylindrical (r-e-z) or two-dimensional Cartesian (x-y) hydrodynamic calculation can be performed within the reactor vessel. Components outside the vessel are treated in one-dimensional l
geometry. A typical arrangement of components and mesh cells for one c
i loop of a FWR is shown in Fig. 1.
The vessel module is used to model all regions inside the pressure vessel, including the downcomer, lower plenum, core, upper plenum and upper head.
It is in these regions of the reactor system that signif-icant multidimensio'ial effects are likely to occur during a LOCA and other postulated ac cidents. Examples are two-dimensional and counter-current steam water flow patterns in the downcomer during the blowdown
~
and refill periods acd preferential rewetting of the cooler fuel rods in the core dur.ng reflood.
[
1 i
i L.
3 Table 1 TRAC Cocoonent Modules
' ~
Module Description VESSEL Models a PWR vessel and associated internals using either a three-dimensional (r-e-z) or two-dimensional (x-y) geometrical representation and a six-equation two-fluid model to evaluate fluid flows within the vessel. VESSEL includes rot heat transfer with reflood dynamics in one-dimensional and cylindrical geometry, slab heat transfer from structure, and point-reictor kinetics with decay heat.
PIPE Models thermal-hydrauli: flow in a one-dimensional duct or pipe using the five-equatien drift-flux model.
PIPE can treat area changes, wall heat sources, wall friction, and heat transfer across the inner and outer wall surfaces.
Both semi-implicit and fully implicit solution algorithms are available in this module. The area change correlations do not change with flow direction.
PRIZER Simulates a pressurizer using the one-dimensicnal drif t-flux model with drift velccities specified to produce a sharo liouid-vapor inter-face during discharge. The pressurizer walls are adiabatic, but energy transfer from a heater / sprayer system is simulated.
PL'MP Describes the interaction of the two-chase fluid with a centrifugal pump using tne PIPE capabilities and pump correlations for the source of mixture momentum.
ACCUM Simulates an accumulator filled with ECC water and pressurized with nitrogen gas using the one-dimensional drift-flux mcdel. The vapor-phase properties are those for :litrogen gas and drift velocities are specified to produce a sharp liquid vapor interface during discharge.
Nitrogen is not allowed to discharge frcm the accumulator because a noncondensible field is not yet available in the basic hydrodynamics model.
STGEN Models either a U-tube or once-through steam generator using the one-dimensional drift-flux model.
Primary-and secondary-side hydro-dynamics are treated separately with coupling through wall heat transfer.
TEE Models the thermal-hydraulics of three piping branches (two of which lie along a common line with the third entering at an arbitrary angle) using essentially two pipes. The momentum source modeling is improved in the PD2 version of TRAC.
VALVE Models the thermal-hydraulic flow in a valve using the basic PIPE capabilities. Valve action is modeled by controlling the flow area and hydraulic diameter between the two fluid cells.
4 Table 1 Continued Module Description BREAK Imposes a fixed or time-dependent pressure boundary condition one cell away from its adjacent component.
BREAK is not actually a system component module but is treated as such with respect to input, initiali-zation and identification procedures.
FILL Imposes fixed or time-dependent velocity boundary conditicos at the junction with its adjacent ccmpenent.
FILL is not actually a system component module but is treated as such with respect to input, initiali-
- ation and identification procedures. The user tapplies the temperature and void function for the FILL input.
t
=
5 Table 2 TRAC Functional Modules Module Descriotion DFID Solves the finite-difference equations for the one-dimensional drift-flux model using either a semi-implicit or fully implicit algorithm.
TF3D Solves the finite-difference equations for the multidimensional two-fluid model using a semi-implicit algorithm. TF3D includes a constitutive package to provide wall and interfacial shears and interfacial mass and heat transfer.
THERM 0 Provides thermodynamic properties of water and steam.
FWALL Computes two-phase wall friction factors; also calculates loss coefficients associated with abrupt area changes.
SLIC Calculates relative velocities between vapor and 1icuid phases for the one-dimensional drif t-flux model. The procedure is based on a flow regime map similar to that used in the three-dimensional vessei hydrodynamics.
ROCHT Solves the one-cimensional (cylindrical) finite-difference thermal-conduction r 'ations in the fuel rod including pellet, gap and clacding regions.
SLACHT Solves for the lunped-parameter temperature of a slab of arbitrary
~
configuration.
CYLHT Solves the one-dimensional (cylindrical), finite-difference thermal-conduction e 2ations in pipe walls.
HTCOR Provides heat transfer coefficients from wall to fluid based on local conditions.
O
-01 i.,6i....
-f PRIZER
_" r - O' rr' '
~
al _ p u-
- E STGEN m
1
]
- C ACCUM 4
\\
4 III IIIIIIII I I I I 1 1--
-1 I I IA m
i
% C g
V w
l PUMP TEE
}
j j l
k [
/
1 C 'T'T3 I II IIIIIIIIIIIIIIIIIII IIIIH j
1 q
y-v
(
T.
~
VESSEL l
' PIPE VALVE s
IIIIIIIIIIIIIIIIIIIII
]
(r,6,Z)
/
j
.- j l
~,'
Pig. 1.
Typimi arrarvjenent of IWR vessel and loop avgrments.
O
7 D.
Nonhoj::ogeneous, Nonequilibrium Hydrodynamics Two-phase flow in the various TRAC components is treated using nonhomoge-neous, nonequilibrium models; that is, liquid and vapor velocities are not assumed to be equal, and further:nore liquid and vapor temperatures are in general unequal with neither phase assumed to be at saturation conditions.
A two-fluid six-equation model is used to describe the liquid-vapor flow field within the reactor vessel. These ecuations are based on conservation of mass, momentum and energy for the separate liquid and vapor fields. Supplementing these field equations are so-called constitutive relations or closure equations that specify (1) the transfer of mass, energy and momentum between the liquid and vapor phases and (2) the interaction of these nhases with the system structure.
The nature of these interfacial transfers and interactions is dependent on flow topology, and therefore a fle.;-regime-dependent constitutive equation package is included in TRAC.
Tne flow in the one-dimensional loop components is described by a five-equation drif t-flux model. These equations are based on conservation of mass, energy and momentum for the mixture and conservation of mass and energy for the vapor.
Licuid and vapor velocities are not assumed to be equal but are expressed in terms cf a relative velocity which is dependent on flow topology.
More details can be found in Refs. (la) and (7).
E.
Comprehensive Heat Transfer Heat transfer models in TRAC include (1) conduction models to calculate temperature fields in structural materials and fuel rods and (2) convec-tion models to provide heat transfer between structure and coolant.
Heat transfer to the two-phase fluid is calculated using a generalized bofling curve constructed from a library of heat transfer correlations based on local surface and fluid conditions.
Conduction models are available for obtaining temperature fields in one-dimensional (cylindrical) pipe walls, lumped-parameter slabs, and one-dimensional (cylindrical) fuel rod geometries.
Pipe wall conduc-tion is used in the components outside the vessel, whereas the slab and fuel rod conduction models are used in tne vessel module. The fuel rod conduction analysis accounts for gap conductivity changes due to temperature effects, but not due to geometry effects, metal-water reaction, and quenching phenomena. A fine-mesh axial renoding capability is available for fuel rods to permit more detailed modeling of reflood heat transfer and tracking of quench fronts due to bottom flooding and falling films. Precooling e,ffects and consistency between quench-front propagation and stored energy considerations are included in the reflood heat transfer methodology.
h
8 Th'e RAC library of heat transfer correlations includes data for the following heat transfer regimes: laminar and turbulent forced convec-tion to a single-phase liquid or vapor and to a two-phase mixture; nucleate boiling and forced convection vaporization; pool boiling and high-flow critical heat flux (CHF); transition boiling; minimum stable film boiling; film boiling including subcooling and radiation effects; and horizontal, vertical, and turbulent film condensation.
There is no special treatment for heat transfer in any particular component, all use the same boiling curve and heat transfer coefficients.
In the case of reflood, there are specialized heat transfer coefficients used fce the core only, which are differentiated for falling film quench fronts dnd bottom reflood quench fronts.
F.
Solution Strategies The system of field and constitutive ecuations in TRAC is solved by scandard spatial finite-difference techniques. A semi-implicit time-differencing technique is normally used in most components. This technique is subject to the Courant stability limitaticn, which restricts the size of the time step in regions of high-speed flew.
A fully implicit time-differencing option is available for the fluid dynamics in most of the one-dimensional components. This option allows fine spatial resolution in regions of high velocity (e.g., in a nozzle) without restricting the time step size.
The description of numerical procedures given here is necessarily limited. More detailed descriptions can be found in Refs, la, 8 and 9.
Computer ru ning time is highly problem-dependent.
It is a function of the total mesh cells in the problem and the maximum allowable time step size. The total run time for a given transient can be estimated from a unit run time of 2 to 3 ms per mesh ccll per time step and an average time step size of 5 ms.
The TRAC steady-state capability is designed to provide time-dependent solutions which may be of interest in their own right or as initial conditions for transient calculations. Two distinct calculations are available within the steady-state capability: (1) a generalized steady-state calculation and (2) a PWR initialization calculation. The first is ~ sed to find steady-state conditions for a system of arbitrary u
configuration. The second is applicable only for configurations m
9 typ'ical of current PWR systems and is used to adjust certain loop par.2 meters to maten a set of userspecified flow conditions.
Both calculations utili:e the transient fluid dynamics and heat transfer routines to search for steady-state conditions. The search is terminated when the normalized rates of change of fluid and thermal variables are reduced below a usersp( ified criterion throughout the system. Generally, for a given problem, mucn less computer time is used for steady-state calculations than for transient calculations.
l
APPENDIX II TRAC Pla DEVELOPMENTAL ASSESSMENT - SUMPARY A. Introduction Experiments selected for developmental assessment of TRAC-P1 A, and the more important thermal-hydraulic effects occurring during these tests are given in Table 3.
Note that the first five analyses use only the one-dimensional capability in TRAC whereas the remainder involve the multidimensional capability as well. Tests selected for developmental assessment include separate effects (tests involving basically only one component), synergistic effects (several coupled components but only one LOCA phase), and integral effects (several components and more than one LOCA phase).
Detailed comparisons betwen code results and experimental measurements for the tests in Table 3 are reported in Ref. 6.
Therefore, only brief surraries and typical comparisons are given below for selected tests.
B. Edwards, CISE and Marviken Blowdown Tests referred to as Standard Problem 1, was the The Edriards experiment, ($)ht horizontal pipe (0.073 m ID x 4.1 m long) decressurization of a straig initially filled with subccoled water at approximately isothermal conditions. A glass rupture disk at one end of the pipe was broken to initiate the blowdown. TPAC best-estimate calculations are in reasonable agreement with available experimental measurements of fluid pressures and temperatures and with the single density measurement.
I In the CISE (Centro Informazoni Studi Esperienze) experiments, (l_1) subcooled water was circulated through a vertical tubular test section.
TRAC calculations of the CISE tests are in good overall agreement with j
the measured data, including fluid pressure and temperature at several i
locations in the test section, pipe wall temperature in the heater i
section, and mass holdup measurements.
- 12) are designed to detennine how well The Marviken critical flow tests (Tiig small-scale experiments actually code models that were developed us apply to full-scale systems. These tests involve the blowdown of a large (5.2 m ID x 21.5 m high) pressure vessel through a discharge pipe (0.75 m ID x 6.3 long) which protrudes 0.74 m into the bottom of the vessel.
In
(
l l
1 ww a
2
~
- Table 3 TRAC-Pl A Develocmental Assessment Analyses Experiment Thermal-Hydraulic Effects 1.
Edwards horizontal pipe Separate effects, one-dimensional critical flow, blowdown (Standard Problem 1) phase change, slip, wall friction 2.
CISE unheated pipe blow-Same as No.1 plus pipe wall heat transfer, flow down (Test 4) area changes, and gravitational effects 3.
CISE heated pipe blowdown Same as No. 2 plus critical heat flux (Test R) 4.
Marviken full-scale vessel Same as No. 1 plus full-scale effects blowdcwn (Test 4) 5.
Semiscale 1 1/2 loop iso-Synergistic and system effects, one-dimensional thermal blowdown (Test 1011, flow, ;hase cnange, slip, wall friction, critical Standard Problem 5) nozzle flow 6.
Semiscale Mod-l heated loop Same as No. 5 plus 3-dimensional vessel model blowdown (Test S-02-8, with rod heat transfer including nucleate Standard Proble~ 5) boiling, de;arture from nucleate boiling (DNB),
and post-DNS 7.
Creare quasi-steady downcomer/ Separate effects, countercurrent flow, interfacial ECC bypass experiments drag and heat transfer, condensation 8.
FLECHT forced-flooding tests Separate effects, reflood heat transfer, quench-front propagation, liquid entrainment and carryover 9.
Non-nuclear LOFT blowdown Integral effects during blowdown and refill, with cold-leg injection scale midway between Semiscale and full-scale PWR (Test Ll-4, Standard Problem 7)
\\
3
[esh 4 a nozzle with a minimum diameter of 0.51 m was attached to the bottom of the discharge pipe. The blowdown is initiated by overpressurizing the gap between two rupture disks at the downstream end of the nozzle.
TRAC best-estimate results for Marviken Test 4 are in very good overall agreement with fluid pressure and temperature measurements at various locations and with mass fluxes derived from differential pressure and Pitot tube measurements. The mass flux from the break in fiarviken Test 4 l
are shown in Fig. 2.
I I
I I
I I
I I
I I
-- FROM VESSEL N
60 PRESSURE DIFF.
g
- "N FROM PITOT l
N MEASUREMENTS
{ 50' 2
\\
TRAC
\\
\\,
m 40 9
N s
x 3o
\\s D
\\
d Nw _
- --____________y\\
20 m
Mg
\\
2 10
\\,-
0 I
O O
20 30 40 50 TIME (s)
Fig. 2.
Mass flux for Marviken bW experiment (Test 4).
-9 y
l 4
C.
55miscaleBlewdownTests Tests in the Semiscale 1 1/2 Loop Isothemal Test Facility (Em.13) provided the first system-efh i hydraulic data from a multiloop sys TRAC analysis of Test L ! included a steady-state calculation to provide self-consistent initial conditions for the blowdown transient and a transient calculation utilizing the restart-dump from the steady-state calculation. Calculated steady-state initial conditions for Test 1011 agree well with measurements of the vessel outlet temperature, intact-loop volumetric flow rate, pump differential pressure, system pressures, etc.
Agreement between calculated and experimental results for the blowdown transient was generally very good for all system variables that were compared. These included mass flow rates, system pressures, fluid densities and temperatures, and differential pressures. The comparison for the icwer plenum pressure is given in Fig. 3.
Test 1011 represents the first developmental assessment problem involving a large variety of components arranged in a multiloop configuration.
It is encouraging that the one-dimensional TRAC model is adequate, since the experiment was designed to minimize multidimensional effects. With a TRAC model containing 122 fluid cells, the steady-state and transient calculations required 0.5 and 19 min of CPU time, respectively.
The Semiscale Mod-l system (H) was very similar to the i 1/2 loop configura-tion described previously, however, the Mod-l vessel contained 39 electrically heated rods (which could be prcgramed to simulate the surface heat flux of a nuclear rod) and a 0.011-m downcomer gap. Test S-02-8 consisted of a 200 percent double-ended cold-leg break with a programmed power decay curve to simulate decay heat in a nuclear core. The transient was initiated from a steady-state temperature distribution in the core and loop at a power level of 1.6 MW.
The best-estimate TRAC model of Test S-02-8 contains a total of 263 fluid cells, including 152 cells in the three-dimensional vessel model. Although multidimensional effects are not too significant in this facil'ty, the three-dimensional vessel module was used because fuel rod heat transfer is not available in the one-dimensional pipe module. As was the case for
_ Test 1011, calculated steady-state initial conditions and transient results for Test S-02-8 agree well with measurements of system variables. The calculated cladding temperature in the high-power zone is compared in Fig. 4 with the band of temperatures measured in the same zone. Although the overall agreement in cladding temperature response is good, some detailed features were not predicted by TRAC. These include what appear to be random variations in the time to CHF and rewetting of some rods after the initial dryout. Running times for the steady-state and blow-down calculations were 50 min and 120 min, respectively.
e SEMISCALE TEST 1011 o-20.0 LEGEND O
n-TRAC CALCULATION W
EXPERIMENTAL DATA E
15.O-o i.
M M
i W
i E
1 2
10.0 -
nZ 6_
m W
1
_J O.
__' ~
l g
~
,N s
l w
5.0 -
~
3c:
5 oJ
~ ~ ~ ~ ~
i j
0.0 i
i i
0 5
10 15 20 25 30 TIME (s) i Fig. 3 Imer piernan pressure for Suniscale isotivermal blowikwn (Test 1011).
4 i
6 1200 s
i e
i 110 0 1000
. TRAC P1A CALCULATION TEST DATA 3 900 p
800
- 700, 600O 5
10 15 20 25 30 TIME (s)
Fig. 4 c1Wirs tmperature in high-power zone for Semiscale Test S-02-8.
I
- O
-y.,. -.._ _,., -,. ~
-.___.-_=pi,+.
we
7 D.
C,regre ECC-Bypass Tests The primary purpose of the 1/15-scale downcomer experiments at Creare (15) was to study the effect of countercurrent steam flow rate, ECC water subcooling, and downcomer wall superheat on the delivery of ECC water from tne downcomer to the lower plenum.
The TRAC best-estimate modei for the Creare experiments consists of a three-dimensional vessel containing 112 fluid cells and one-dimensional piping connections for the injection and break ports. The calculational procedure closely parallels the experimental procedure. Results of the Creare calculations are in excellent overall agreement with experimental flooding curves for a wide range of ECC injection rates and subcoolings.
This is shown in Fig. 5, which presents the flooding curve for high subcooling. The complete bypass and complete delivery points on the curves are well predicted by TRAC for low-and high-subcooling cases.
Computed results for the Battelle Columbus Laboratory (BCL) 2/15-scale facility ( _16) are similarly in good agreement with the data, indicating that scale effects in this range are properly treated.
E.
FLECHT Reflood Tests Assessment of the reflood heat transfer and quench-front prcpagation models in TRAC has to date focused on the forced-bottom-flooding experiments in the PWR Full-length Emergency Cooling Heat Transfer Facility (FLECHT) (17_).
The single-channel geometry of these experiments lends itself very well to the use of the slab vessel option in TRAC. As a matter of fact, a one-dimensional representation was obtained by using only one cell per axial level. The base-case model contained nine axial levels in the core, with each of these levels containing five fine-mesh axial intervals fo-the reflood heat transfer calculation. Conduction in the electrically heated rod was represented with eight radial nodes. Test conditions for the three cases calculated are given in Table 4, and a summary of the calculated and measured results is given in Table 5.
TRAC-PlA predicts the maximum temperature (and hence the temperature rise) quite well for all three tests. For the high-flooding-rate case (Test 03541), the calculated turnaround time and quench time also agree very well with the data. This is not the case, however, for the low-flooding-rate tests where the code predicts early turnaround and quenching. Under-prediction of the carryover rates results in water remaining within the test vessel and, hence, a rapid refill of the core region which partially accounts for early quenching. TRAC-PlA does not contain an explicit entrainment model. This capability, along with a better definition of a rewetting criterion, should significantly improve the code results for low flooding rates. The ratio of CPU time to transient time is about 25 for the FLECHT calculations.
1
_ _. _,