ML19257D625
| ML19257D625 | |
| Person / Time | |
|---|---|
| Issue date: | 01/24/1980 |
| From: | NRC - RRG# 1-14 ADVANCED SYSTEM CODE |
| To: | NRC - RRG# 1-14 ADVANCED SYSTEM CODE |
| Shared Package | |
| ML19257D614 | List: |
| References | |
| NUDOCS 8002050196 | |
| Download: ML19257D625 (36) | |
Text
,
MINUTES OF THE ADVANCED CODE REVIEW GROUP MEETING Time and Place:
September 5 and 6,1979, NRC Willste Building Silver Spring, Maryland Attendees: See Attachment 1 Purpose of the Meeting:
1.
To determine whether the current thermohydraulic models in TRAC-PlA require modification, and 2.
To determine whether additional effects are of sufficient importance, for system analysis, to require inclusion in future versions of TRAC.
Topics Discussed:
The agenda of the meeting is enclosed (see Attachment 2).
It dealt with the following topics:
A.
The Status of Problems Discussed at the January 1979 Meeting of the ACRG:
1.
Thermodynamic property routine 2.
Steady state calculations and mass conservation 3.
Rod gap conductance 4.
Decay heat routine (See Attachment 3 for details)
B.
LASL's Answers to Questions Raised at the January 1979 Meeting of the ACRG Concerning the Simplified, Fast Running TRAC:
1.
PWR vessel description simplifications including the treatment of upper head injection.
2.
Field equations and constitutive relations for droplet field (s) if such are being considered.
3.
Treatment of core quenching, quench front propagation and precursory cooling.
4.
Liquid separation in upper plenum and fallback through upper core support plate.
5.
Counter-current flow limitation at upper core support plate.
6.
Downcomer penetration model for use with 1-D noding in the PWR down-comer.
1871 192 8002050194
7.
Simplified choked flow model (See Attachment 4, for details)
C.
Thermohydraulic Models in the Vessel Module of TRAC-Pl A D.
Thermohydraulic Models in Loop Components of TRAC-PlA The last two topics followed closely the format outlined in the questionnaire enclosed as Attachment 5.
The latter was prepared by S. Fabic in order to facilitate the assessment of current models as well as to help in following the progress and improvements to be made in future versions of TRAC.
E.
TRAC Assessment at BNL Conclusiona and Recommendations The following conclusions and recommendations were made in relation to these topics:
A.
Status of problems Discussed at the January 1979 ACRG Meeting 1.
Thermodynamic Properties Routine Since the last meeting of ACRG, the routine to calculate thermodynamic properties in particular, liquid energy and vapor density, has been improved. However, further improvements at higher pressures will be needed.
LASL was advised to find out whether or not these additional improvements can be made by using the thermodynamic property routine recently made available by Professor H. Paynter at MIT.
2.
Steady State Calculations and Mass Conservation LASL is working presently on improving TRAC's capability to conserve mass and carry out adequate steady state calculations.
One of the causes for the difficulties encountered with steady state calculations was removed by converting the 1-D loop components from a drift to a two-fluid model.
Work is in progress to remove defficiencies generated by the iteration scheme used in the 3-D vessel calculations.
The reviewers stressed again, the importance of adequate steady state calculations for small break LOCA analyses.
3.
Rod Gap Conductance Following the recomendations made at the January 1979 meeting of ACRG, LASL has implemented in TRAC an improved gap conductance model which accounts for the effect of thermal expansion.
1.871 193
, 4.
Decay Heat Routine Following the recommendations made by Professor Schrock, LASL will implement in TRAC a decay heat routine based on the revised (1979) version of ANS Standards. This will be done in TRAC-PFl to be released in June of 1980, but not in TRAC-PD2 to be released by the end of December 1979.
i B.
LASL's Answers to Questions Raised at the January 1979 Meeting of the ACRG Concerning the Simplified, Fast Running TRAC.
1.
PWR vessel description simplifications including the treatment of upper head injection.
TRAC-PFl will have a coarse mesh and use an improved numerical technique to realize fast running times.
It will be based on a two-fluid 1-D model; however, the vessel will have also a 2-D and 3-D options.
The noding of the vessel has not been established.
2.
Field equations and the constitutive relations for droplet field (s) if such are used.
TRAC-PF1 will not have an additional liquid (droplet) field.
Such a field will be implemented in TRAC-PD3, scheduled for release in 10/80.
Equations and correlations which describe this field have not been finalized.
3.
Details of core quenching treatment including quench front propagation and precursory cooling.
An improved reflood package has been implemented in TRAC-PD2 scheduled for release in 12/79.
The same package will be implemented in TRAC-PFl.
The improved model accounts for the effects of quench front propagation, temperature history, steam binding, oscillations and other system effects.
4.
Liquid separation in upper plenum and fallback through upper core support plate.
Since TRAC-PF1 will not have an additional liquid (droplet) field, it will not have the capability to model these phenomena. However, LASL intends,o study the data reported by CCTF to determine whether a modeling improvement (via a correlation) can be implemented in TRAC-PFl.
5.
Counter-current flow limitation at the upper core support plate.
LASL intends to analyze GE data on CCFL with the intention of deriving a suitable model for implementation in TRAC-PF1.
i87i 194
4_
6.
Downcomer penetcation model for use with 1-D noding of the PWR downcomer.
TRAC-PFI will haie the option for modeling the downcomer.
One, based on a 2-D represer tation using the routine presently in TRAC-PD2.
The other a 1-D model based on a correlation.
The latter has not been selected.
7.
Simplified choked flow model TRAC-PF1 will have a simplified model for calculating choking conditions.
The model has not been selected.
8.
Comments and recommendations pertaining to TRAC-PFl, TRAC-PD3 and small break analyses.
a.
For small break analyses. LASL should look into the possibility of simplifying property routines in TRAC-PFl.
b.
For small break analyses TRAC-PF1 could run with large time steps provided that no discontinuities exist in the various routines.
LASL should therefore attempt to smooth all relations in TRAC-PFl. This will reduce the number of potential numerical problems and improve running time, c.
For small breaks, it is most important to calculate adequately the total water inventory and the total heat removal.
Consequently, the first priority should be given to the effort that will assure that TRAC-PF1 has this capability.
d.
For small break analyses and plant transients, detailed modeling of the secondary side and control systems are important.
LASL should therefore carefully consider these requirements while developing TRAC-PFl.
e.
In developing correlations for CCFL at the upper core support plate, LASL should make sure that these correlations are consistent with the interfacial momentum exchange terms used in the two-fluid 1-D model. Otherwise there will be a potential for inconsistencies and numerical instabilities.
C.
Thermohydraulic Models in Vessel Module of TRAC-Pl A
- 1) Heat Transfer Coefficient Correlations a) CHF correlation The CHF correlation used in TRAC was thought to be inadequate when applied to rod bundles.
Therefore, LASL was advised to use the best empirical correlation applicable to rod bundles.
This correlation should account also for the effects of space.
1871:,195
. b) Condensing heat transfer LASL proposed to use Chen's correlation for condensation.
This was thought to be most inappropriate.
LASL was advised therefore, not te do it but to use instead condensation correlatior s available in the literature which have a substantial data base to support them (see Discussion).
d) TCHF and TMIN The values of TCHF and TMIN used in TRAC calculations were thought to be inadequate.
In particular, the value of T HF was thought to be too low, whereas TMIN, calculated from Iloeje s correlation, was thought to be inadequate for low mass flow rates.
LASL was advised therefore, to use measured values of TMIN, or at low mass flow rates to use the method proposed by Henry.
- 2) Thermohydraulic Correlations in the Vessel Module a) Vertical Two-phase flow regimes Vertical flow regime transition criteria used in TRAC are based on pipe flow data. Some doubts existed when these criteria were applied to rod bundle geometrics.
However, experimental data recently reported by Dukler indicate that with one exception, the flow regi w criteria used in TRAC are valid when applied to rod bundles.
The single exception being the criterion for the onset of annular flow which needs to be nodified.
The modification will be provided by the ongoing program at the University of Houston.
b) Horizontal two-phase flow TRAC does not have flow regime transition criteria for horizontal two-phase flow.
LASL was advised to include thesc criteria in TRAC because adequate modeling of horizontal two-phase flow is important for small break LOCA calculations.
c) Flow regime modeling in downcomer At the July 1978 and January 1979 meetings of ACRG, it was recommended "that correlations and flow regimes should be selected according to 1871 196
9 a particular component z.ad process rather than using a general set throughout the system. This was thought to be of particular importance to the modeling of the downtomer where " bridging" can take place.
LASL was advised again, that a " bridging criterion" should be included in TRAC in order to model adequately the downcomer, c) Bubble diameter calculations TRAC calculates bubble diameter from a Weber number criterion.
While this method is adequate for high velocity flows, it was thought to be inadequate when applied to gravity dominated, that is, to low velocity flows.
LASL was advised therefore, that IRAC ought to have the capability to model adequately, phenomena which are characteristic of gravity dominated flows.
d) Droplet entrainment TRAC uses Wallis-Steen correlation to compute droplet entrainment in reflood, annular-droplet regime and entrainment from froth level.
This method of calculating droplet entrainment was thought to be inadequate.
LASL was advised therefore, to use instead available experimental data and correlations appropriate to a particular process (see Discussion).
e) Wall shear TRAC used Harwell's correlation to calculate wall frictional pressure drop for both cocurrent and countercurrent flows.
This was thought to be inconsistent with the two-fluid model formulation used in TRAC.
It was thought also that this incsnsistency may have led to the difficulties that TRAC has experienced sometimes, in calculating steady state.
LASL was advised therefore, to make the wall shear and interfacial shear models consistent with each other.
f) Phase separation and distribution BNL has reported difficulties when using TRAC to calculate phase separation measured in RPI slab tests.
1871 197
, BNL was advised to compare both the 3-D and 1-D versions of TRAC to the more recent data reported by RPI as well as with the Wye and Tee data reported in NUREG/CR-0557.
g) Modelir.g of interfacial heat transfer In the annular-droplet flow regime, TRAC does not rodel the effect of interfacial shear on the heat transfer ccefficient.
This was thought to be inadequate. LASL was advised to incorporate this effect into TRAC. This should not be a difficult task since TRAC already calculates the interfacial shear and uses it in the two momentum equations.
LASL was advised also that all models and correlations used in TRAC ought to be documented, referenced and checked for consistency and accuracy against the available data base.
D.
Thermohydraulic Models in Loop Components of TRAC 1)
Pressure Drop Calculations a) Frictional pressure drop TRAC has four users options to calculate the frictional pressure drop in 1-D components.
This was thought to be undesirable.
U.SL was advised therefore, that TRAC should have only one correlation which must be also consistent with the correlation used for interfacial shear.
b) Expansion and contraction pressure drops TRAC has two different treatments to calculate pressure drops in expansions and contractions.
Furthermore, the ei.ect of flow reversal on form losses through abrupt area changes are not accounted for in TRAC.
LASL was advised therefore, that the modelir.g of form losses and of flow reversal must be consistent and accounted for in TRAC.
c) Slip velocity used in pressurizer and accumulator The slip correlation used in TRAC for the pressurizer ed accumulator is not based on experimental data.
Since realistic modeling of phase separation in tk pressurizer is important for calculating mass flow rates through the relief valve, LASL was advised that the slip correlation used in the pressurizer must be supported by experimental data.
1871 198
. d)
Pump modeling TPAC does not take into account the energy contribution to the fluid by the pump.
Since this energy can exceed the energy due to decay heat, LASL was advised to include in TRAC the energy contributed by pumps.
E.
TRAC Assessment at BNL In response to NRC's request, BNL prepared the document entitled, " Constitutive Equations in TRAC Pl A" (see Attachment 11).
It contains BNL's assessment and coments concerning thermohydraulic models and correlations in TRAC-Pl A.
These were summarized in BNL's presentation (see Discussion).
A list of BNL's coments and recomendations which require further clarifications and LASL's attention is given in the Discussion.
ACRG members and consultants concurred with most of them.
In addition they recommended that a) A document be prepared setting forth the correlations used in TRAC, list their source, accuracy range of applicability and shortcomings together with a summary of the data base upon which the correlation rest.
b) The independent assessment program be continued.
In carrying this work it would be mos+ helpful to prioritize the review efforts in terms of their importance to TRAC prediction.
Furthermore, an emphasis ought to be placed on relations which will be retained in present and future versions of TRAC.
c)
Independent assessment programs should develop a method for evaluating which of the correlations play an important role in TRAC predictions in order to focus studies and improvements in such areas.
Both LASL and BNL were complimented for their work, LASL for TRAC's progress whereas BNL for assessing TRAC's models.
1871 199
, DISCUSSION R. J. Pryor gave an overview of TRAC development (see Attachment 6) and together with D. Liles, addressed the first two topics of the Agenda that is, the status of problems and answers to questions raised at the last meeting of ACRG.
A.
Status of Problems Discussed at the January 1979 Meeting of ACRG
- i. Themodynamic Properties Routine At the January 1979 meeting of ACRG, LASL attributed the mismatch between the calculated vs. measured flow rates and temperature profiles in the LOFT L2-2 test, to possible inaccuracy in the thermodynamic routine.
After a closer check, LASL found a number of errors and problems related to the least square polynominal fitts which were used in the code, to calculate liquid energy and vapor density.
These errors were corrected and the discontinuities smoothed.
The improved calculations are illustrated in.
However, LASL noted that further improvements will be needed at higher pressures.
P. Griffith comented that Professor H. Paynter from MIT, has programmed steam tables up to 5000 psi and advised LASL to find out whether or not the MIT routine could be used in TRAC. He ador.d in :riting:
"I checked with Professor H. Paynter (MIT) and the computer specification of the steam properties is practically complete.
He suggested that anyone who wants to use it contact him.
As I said, the method uses the Helmholtz free energy to calculate all the other properties ar.d should compute the state from any two independent input variables. His address is Room 3-264, MIT, Cambridge, MA 02139.
Telephone (617) 253-2218."
V. Schrock vrote:
"The evaluation of water properties by TRAC evidently needs further improvement. One of my masters students has recently completed a comparison of the TRAC equations with the ASME and Keenan and Keyes tables, a copy of which I will send under separate cover.
It shows some areas of differences."
2.
Steady State Calculations and Mass Conservation The requirement for adequate steady state calculation capability was stressed at the last meeting of the ACRG where it was reported that TRAC-PlA had some difficulties in conserving mass.
This requirement becomes particularly important for small break LOCA calculations. More recently, BNL has also reported difficulties in obtaining some steady state calculations with TRAC-PlA, which were communicated to LASL.
At this meeting, ACRG was infomed that LASL is working to resolve this probl em. Originally, LASL attributed the difficulty with steady state calculations and conservation of mass to the drift model used in the 1-D loop components.
In particular, to the ambiguity in sign that exists when the velocity changes direction.
By converting the 1-D components from a 1871 200-
. drif t to a two-fluid model, this ambiguity was resolved.
However more recently, LASL has found that the problem of conserving mass was also generated by the iteration scheme used in the 3-D vessel calculations.
At the present time LASL is working on removing this defficiency.
3.
Rod Gap Conductance It was concluded at the January 1979 meeting of ACRG, that the fuel rod gap conductance treatment was too crude in TRAC-PlA.
It was recommended therefore, that this defficiency be removed and furthemore, that this should be accomplished without resorting to a full scale integration of the FRAP-T code with TRAC.
Such an integration would introduce great complexities and therefore increase significantly the running time of TRAC.
At this meeting, LASL informed ACRG that a simple gap conductance model which accounts for the effect of thermal expansion, has been installed in TRAC.
4.
Decay Heat Routine At the January 1979 meeting of ACRG, Professor V. Schrock indicated some inadequacies of the decay heat routine used in TRAC.
In particular, he noted that TRAC decay heat calculations make use of old data, that is, those given in the 1971 version of ANS Standard.
As the recent revision of ANS Standards provides a much better basis for evaluating decay heat, Professor Schrock advised that the revised ANS Standard should be used by all best estimate advanced codes and by TRAC in particular.
At the present meeting, LASL infomed ACRG that decay heat calculations based on the recent version of ANS Standards will be implemented in TRAC-PF1 scheduled for release by the end of FY 1980, but not in TRAC-PD2 scheduled for release in 12/79.
Professor V. Schrock noted again that the routine presently used in TRAC is inadequate, and wrote:
"It is of greater importance to realize that the TRAC package, in following the older procedures of RELAP and RETRAN which had been based upon the 1971 version of the ANS Standard, is now obsolete. Thus the other comments in my letter following the January 1979 meeting of the ACRG are much more important.
Therefore I was very disappointed to learn that a new decay heat package will not be introduced into TRAC for more than a year. The new ANS Standard (1979) is not that difficult to implement.
I urged at the close of the meeting that LASL reconsider the scheduling of this item and do it as soon as possible. The 1979 ANS Standard has already been implemented in RELAP 4 MOD 7.
I would like to reiterate that it should receive high priority in TRAC."
1871 201
. B.
LASL Answers to Questions Raised at the January 1979, Meeting of ACRG Concerning the Simplified, Fast Running TRAC.
R. J. Pryor gave an overview of the various versions of TRAC and their release dates (see Attachment 6 ).
He summarized also future improvements to be incorporated in the fast and detailed versions of TRAC, that is, in TRAC-PF1 and TRAC-PD3 respectively (see Attachment 6 for details),
t To questions raised during the last ACRG meeting concerning TRAC-PF1, LASL provided the following answers:
1.
PWR vessel description simplifications, including the treatment of upper head injection.
TRAC-PF1 will have a coarse mesh option and an improved numerical technique to realize fast running times.
For example, in its present configuration TRAC-PF1 runs five times faster than TRAC-PI A.
This coarse mesh, fast running TRAC will be based on a two-fluid 1-D model with options for a 2-D or 3-D representation of the vessel. The noding for the vessel has not been yet established.
LASL intends first to conduct a noding study with a 3-D representation for a TMI type of accident and compare the results and running times with those obtained with a 1-D representation.
After such a comparison, LASL will make a decision on the noding required for the vessel and for the upper plenum and downconer in particular.
2.
Field equations and the constitutive relations for droplet field (s) if such are being considered.
TRAC-PFl will not have a droplet field which will be included in TRAC-PD3 scheduled for release in 10/80.
3.
Details of core quenching treatment including quench front propagation and precursory cooling.
An improved reflood package has been implemented in TRAC-PD2 scheduled for release in 12/79.
The same package will be incorporated in TRAC-PFl.
The improved reflood model was described by F. L. Addessio, the details are given in Attachment 8.
Briefly, the model can account for the effects of quench front motion, rod temperature history, steam binding, oscillations and other system effects.
Presently, it is being assessed against data reported by VSSLT, UCB and FLECHT.
Figures shown in Attachment 8, illustrate the satisfactory agreement between calculated and measured parameters.
4.
Liquid separation in upper plenum and fallback through upper core support plate.
LASL believes that for slow transients such as TMI, no additional liquid field, that is, droplet field is needed. However, for fast transients such 1871 202
, as large break LOCA's an additional liquid field is needed. Consequently, a droplet field will not be implemented in TRAC-PFl but it will be in-corporated in TRAC-PD3.
As the capability to model liquid separation in upper plenum and fallback through upper core support plate requires an additional liquid field, TRAC-PF1 will not have the capability to model these phenomena.
- However, LASL intends to conduct a study via TRAC-PD2 and analyze recent experimental data reported in the CCTF tests.
The intention is to gain a better under-standing of liquid distribution phenomena, which could lead to a modeling improvement (via a correlation) that could be implemented in TRAC-PFl.
5.
Counter-current flow limitation at the upper core support plate.
The capability to model CCFL at the upper core support plate has not been incorporated in TRAC-PD2, but it will be implemented in the TRAC-PFl and TRAC-PD3 versions.
LASL is presently analyzing GE data on CCFL with the intention to derive a suitable model for implementation in TRAC.
6.
Downcomer penetration model for use with 1-D noding of the PWR downcomer.
LASL believes that for TMI like accidents a 1-D representation of the downcomer suffices. However, for large breaks their experience was that a 2-D modeling is required.
Consequently, LASL plans to have in TRAC-PFl, two options to model the downcomer.
One, based on 2-D representation using the routine in TRAC-PD2, the other a 1-D model based on a correlation. The latter has not been yet established.
7.
Simplified c.,oked flow model.
TRAC-PF1 will have a simplified model to calculate choking conditions. The exact model has not been established.
LASL intends to examine more closely the model used in RELAP-5, because of the considerable effort that has been expended at INEL towards its selection.
Furthermore, the T-3 group at LASL, will also examine the models available in the literature and make its recommendations.
8.
Comments by reviewers on TRAC-PF1, TRAC-PD3 and small break calculations.
The following coments were made by the participants during the meeting:
S. Levy noted that considerable simplifications in the property routine could be made in TRAC-PF1 since conditions in small break LOCA's are quasi static.
1871 203
. L. Agee observed that for small breaks, TRAC-PFl could run with large time steps provided that no discontinuities existed in the various routines.
Smoothing all relatiuns could result in a significant reduction of numerical problems.
R. Lahey stressed the need to determine the liquid distribution in the core.
L. Agee noted that for small breaks, TRAC-PFl will need considerable details on the secondary side which was not discussed at this meeting.
He noted also that the 1-D neutronic option in TRAC-PF1 should be capable to function in the radial direction because this is the dominant direction for PWR steamline breaks.
The latter are highly dependent on the neutron kinetic option for correct representation.
S. Levy observed that for small breaks it is most important to track the inventory. He advised therefore, that first priority should be given to the effort that will assure this capability in TRAC-PF1.
The following communications were received after the meeting:
S. Levy wrote" The present NRC approach is to use the same basic TRAC model and to apply it to all LOCA events.
In order to cope with small breaks which develop over a much greater period of time, it is proposed to implement fast running versions of TRAC.
As suggested at the meeting the differences between large LOCA and small LOCA may be large enough to justify a different approach.
In the case of large LOCA, events occur very rapidly with little operator intervention and the two-phase fluid flow ano neat transfer phenomena are complex, involving sudden pressure changes, nonequilibrium conditions, and often unstable interf aces.
The emphasis here needs to be on phenomena understanding and the calculations have to be performed with very small time steps as the key variables are changing so rapidly.
For small breaks, the situation is quite different.
Events move very slowly, and equilibrium conditions prevail. The primary interest here needs to be put upon total or integrated water inventory and total or integrated heat removal.
The uncertainties are not in terms of phenomena but rather in considering the very large number of operator actions, equipment, control, and instrument failures.
It might make more sense to carry such studies in improved simulators of the training type except that their basic model should be improved to give good fluid flow and thermal prediction in the areas of natural circulation, its breakdown, presence of noncondensible, etc. The simulator also needs to be upgraded to allow study of the various operator-machine interaction and instrument failures.
Excellent control representation is also necessary.
The name of the game is not to overlook sequence of events which can lead to serious fuel damage and which have a reasonable probability of occurring."
1871 204
. G. Bankoff wrote:
"For TRAC-PFl some preliminary accident analysis should be performed for small breaks.
Based on residence times and heat transfer time constants, it should be possible to treat a number of components as being in a quasi-steady state.
Either lumped-parameter representation or a very coarse mesh can then be used, even for large components, such as the steam generator.
It is essential that the code run at least as fast as real time, and I would prefer a target of one-tenth real time.
This is needed for real-time simulators for operator training, as well as for screening of the very large number of potential man-machine initiator accident scenarios."
"The suggestion of Sol Levy is a good one that it be assumed that the system is initially in a " fragile" condition (such as the core barely covered), and then to determine what cor, trol actions should be taken.
In a sense this is an entirely different approach than the one taken up to now, since this is really an optional control problem, for which a large literature exists. Civen a bounded control effort (rate of coolant entry) and integral of control effort (total ECCS coolant avail-able), what is the optimal path to follow? To explore this approach a fast-running code is necessary."
P. Griffith wrote:
Let us now turn to some more general coments.
The first concerns the method being pursued on the small break analysis.
Somehow, the operator has to be put into the small break analysis.
For the time scales of interest we must assume that the operator will take some corrective action.
Perhaos testing operators on a reactor simulator to see how they behave with confusing or incomplete information is appropriate. They then may be put into the model as one would put in a model of the control system.
In any case, I don't thiak an extension of the large break calculation procedures is appropriate.
Reactor operator behavior has to be in the solution too.
My own suggestion would be to assume the core is just covered and at decay heat levels heat input and that the operator has all or most of his instruments working and see what he does.
The combination of operator acts, and engineered safety systems must be sufficient to save the core.
Perhaps NRC should assemble a committee to look into the definition of the small break problem more closely, I thought Sol Levy had some very worthwhile suggestions on how to handle operator inputs.
In spite of the fact that you indicated to Professor D. G. Wilson at MIT that you are not prepared to pursue some pump work, I think you need it.
This is why, If the pumps were lef t on at TMI-2, I think there is an excellent a.
chance the core would not have been damaged. At 950 psia a vapor
~1871 205
. velocity of 2 in/sec through the core is sufficient to carry out the decay heat.
Simple forced convection from vapor in the steam generators would have provided enough of a heat sink.
Had the pumps not been turned off the loop seal problem might not have developed. Using the pumps to insure good core cooling is an option that I think the operators should have. We cannot test this option if we don't know how the pumps will behave with two phases passing through them, b.
If you are interested in modeling a real reactor system on a reactor simulator, you must know how the pumps will behave. We don' t know well enough now.
c.
It may be possible to design a pump control system which gives you the benefits of pumping with much reduced energy input to the system.
Kirchner told me that on the " Savannah" the pumps auto-matically went to half speed when the reactor scrammed giving better than 1/2 rated flow with about 1/8 rated power input.
Should such an option be put on commercial PWR's? I think we should look into it.
To do so, we need to know how pumps will perform with two phases passing through them.
L. Agee wrote:
"From the opening comments I got the impression that NRC feels that the current TRAC two-fluid model development is complete i.e., that interface closure relationships are adequate.
This feeling was supported by the 1980 TRAC Activity (PD3) which includes 3-D space-time kinetics, addition of a second liquid field, and a FRAP type GAP conductance model.
I personally feel that (1) most of the interface closure relationships (constitutive relations) were (by necessity) judiciously selected by LASL to complete the TRAC model and will require continuous conscientious revision over the next few years.
(2) The three TRAC-PD3 models listed above are premature and would (at this time) be decremental to the overall TRAC program development and application.
This statement should not be taken to reflect, in any way, on LASL ability, but rather reflect the current state-of-the-art in two fluid modeling.
Specifically there exists neither the theoretical under-standing nor the experimental data base that would allow one to correctly formulate the interface closure models. The addition of the PD3 features will (1) only complicate the physics, (2) greatly increase the number of arbitrary inter-phase models, (3) drastically increase the code; running time, and (4) force the user to specify tremendous amounts of input data (which in all probability will be inconsistent).
Instead of working on the PD3 models I reconmend that a) The entire code documentation (both formal reports and internal comment cards) be greatly upgraded.
1871 206
. a) The steady state option be upgraded to initialize all dependent variables in a consistent manner.
b) A BWR version of TRAC be developed.
I realize that these recomendations may not appear to be important, however, as the result of the BNL review it was apparent to me that the documentation was both inadequate and incomplete in many areas.
w that TRAC is publically available a complete and detailed description of all models, numerical solution methods, and known limitations is essential to obtain meaningful corrnents and assistance from the scientific community at large.
Likewise, the need for complete self-initialization is vital as it (1) assures that all conserved quantities are truly con-served (it is my understanding that TRAC can not currently conserve mass, a highly undesirable feature), (2) is absolutely necessary for most operational transients and quite important for small breaks, and (3) greatly reduces input requirements. The need for a BWR version is based on the fact that this could be accomplished with a limited amount of work and would greatly increase the usefulness of the code to industry 9.
Comments by reviewers on the improved reflood package for TRAC-PD2 and TRAC-PFl.
The following communication concerning the improved reflood package were received after the meeting:
P. Griffith wrote:
"The reflood heat transfer package of Addessio looks very good to me.
I think the reflood data and analysis assembled by Loren Thompson is very valuable and should help Addessio in his comparison with data.
He should get a copy of that report from Loren Thompson."
S. Levy wrote:
"This is a substantial improvement over the preceding version. While the model appears to do well in terms of quench front time prediction, the slope of the temperature up to that time does not match the data as well and this should be looked at. The proposed mcdel needs to be checked against available data and at different test conditions as proposed at LASL."
G. Bankoff wrote:
" Generally, the latest quench front results are very encouraging.
However, the interface sharpner needs further examination.
Hopefully, it will correct only numerical effects, leaving entrainment from the frothy inte rface to be given by a suitable correlation.
Unfortunately, I do not have a good suggestion for this correlation, although the paper t y Lee and Azbel in the Proceedings of the 1978 Int.
Center for Heat and Mass Transfer Seminar, Dubrovnik (a two-volume Hemisphere Press bosk edited by Durst, Tskilauri and Afgan) may be helpful, since it r eals with break-up of the liquid cap in gas bubbles at the surf n af a froth."
1871 207 C.
Thermohydraulic Models in Vessel Module of TRAC-Pl A D. A. Mandell discussed the heat transfer correlations used in the vessel module of TRAC-Pl A and improvements planned for TRAC-PD2 (see Attachment 9).
Whereas, D. Liles discussed thermohydraulic correlations used in both the vessel module and in the 1-D loop components (see Attachment 10) 1.
Heat Transfer Coefficient Correlations Wall to fluid heat transfer correlations used presently in TRAC-PlA are sumarized in Attachment 9.
In assessing the present heat transfer package, D. Mandell noted the improvements achieved in predicting the second rewet in the LOFT L2-2 test, when Iloeje correlation was used to predict the minumum stable film boiling temperature (see Figures in ).
Following the discussion of the present package, D. Mandell discussed in some detail, the proposed changes to be included in TRAC-PD2. These were:
a.
Use equilibrium quality instead of flow quality in CHF and Chen correlations b.
Replace condensation HTC correlations with Chen correlations (S = 0) 2 c.
Remove 300 kg/m s equation cut-off in Biasi d.
Use homogeneous nucleation and Iloeje Tmin e.
Allow clad outside to be Zr02 f.
Change T
=T SAT + 5.0 statement to TSAT + 0.5 CHF g.
Delete high speed flow correction to vapor HTC h.
Make hv continuous between regimes 3 and 4
- i. Delete Lineham and Grolmes subcooled HTC (subcooling accounted for in other correlations)
- j. Add liquid natural convection to Regime 1 k.
Delete ICHF = 2 (Biasi only) and ICHF = 3 (Bowring only correlations)
Most of the comments made by the participants during the presentation were concerned with:
a) The CHF correlation, which was thought to be inadequate for rod bundles by R. Lahey, S. Levy, V. Schrock and P. Griffith.
- Instead, 1871 208
. LASL was advised to use the best empirical correlations applicable to rod bundles which account also for the effect of spacers, b.
Use of equilibrium quality in Chen's correlation, which was thought to be inappropriate by A. Dukler, S. Levy, P. Griffith and N. Zuber because Chen's correlation is based on flowing quality, c.
Use of Chen's correlation for condensation, which was thought to be most inappropriate by A. Dukler, V. Schrock, S. Levy, P. Griffith and H. Zuber because there is no basis for applying Chen's correlation (developed for forced convection boiling) to condensing flows.
The reason given by LASL for proposing this change, was to simplify the heat transfer package (by reducing the number of correlations) and thereby improve the running time of TRAC.
Instea6, LASL was advised to use appropriate laminar and turbulent film condensation correlations available in the literature which have a substantial data base to support them, d.
Values for T nd Tmin, both of which have a significant effect on CHF the shape of the transition boiling curve.
R. Lahey, S. Levy and N. Zuber commented that T is much larger than that proposed CHF value of Tsat + 0.5.
Whereas P. Griffith and G. Bankoff noted that T
calculated from Iloeje's correlation was inadequate for low min mass flow rates.
Instead, LASL was advised to use measured values of T
r at low mass flow rates, use the method proposed by Henry. All min participants agreed that further work needs to be done in order to determine adequately the transition boiling region.
N. Zuber noted that the heat transfer surface proposed by Peterson and Sullivan from INEL, could provide a most useful approach for reaching this objective.
Following the meeting the following communications were received:
R. Lahey wrote:
a)
"In equation (B.8.23) a heat transfer correlation is given.
It is referred to as the Dittus-Boelter correlation. Since it has a Prandtl number exponent of 1/3, it should more properly be referred to as the Colburn correlation.
It should be recalled that when the Colburn correlation is used the fluid properties must be evaluated at the mean-film temperature.
If the Dittus-Boelter correlation (which has a Prandtl number exponent of 0.4) is used, i should be corrected for property variations using the correction factor oroposed by Kays [6]:
b I
Nu wall
, for steam
"" bulk
[
/pg )]0.2 , fg 1871 20.9 . b) I believe the CHF correlations used in TRAC are inadequate for rod bundle predictions. Since a good prediction of CHF is quite important I recommend that LASL consider implementing the rod bundle correlation presented by Bowring [1] in Manchester. It would also be wise to contact Bob Bowring to see if he has done anything since (it is my understanding he has)." V. Schrock wrote: "a) I do not think Chen's correlation should be used for condensation as discussed at the meeting, b) The CHF evaluation should be appropriate to rod bundle geometry, c) The heat transfer correlations all implicitly presume that conditions are quasi-steady and that fully developed temperature profiles are always obtained. In addition a correlation such as Dittus-Boelter represents an average over a long channel rather a local value of Nu. These presumptions are seriously violated in parts of the transient to which the code is applied. While this coment is not unique to TRAC, perhaps the level of detail that TRAC is seeking to represent would justify a review of the adequacy of steady state coefficients in transient analysis." A. Dukler wrote: "a) Refer to Figure 4. For low liquid rates with, say, a = 0.4, the logic requires the use of the Chen correlation. However. this would be a region of well defined slug flow. In the first place, Chen probably doesn't apply in rod bundles at low rates even for uniformly homogeneous flow. With slugs it is quite unlikely to apply. Heat transfer in the liquid portion of the slug unit may follow Chen but transfer coefficient can be described by the film condensation equations (B.8.31) or (B.8.32). b) Refer to Figure 4. It was stated at the meeting that in order to simplify matters, path A leading to the condensation equations would be eliminated and the Chen correlation used for all conditions. I disagree with this plan. When films exist one should use the heat transfer equation appropriate for films. There is a substantial body of research on this question. Why not use it? c) Regarding the Condensation Equations: Laminar Film Condensation (eq. B.8.31): This equation ignores the influence of interfacial shear. The retult in the presence of shear is analytic given the shear. Since the code requires the interfacial shear at another point in its logic (see discussion below), this same value of the interfacial drag can be used to calculate a more correct coefficient. Note that the Nusselt equation predicts lower coefficients for counter flow and higher coefficients for concurrent flow. 1871 210 ... Turbulent Film Condensation (eg. B.8.32): As indicated in the BNL report (p. 38) the Carpenter-Colburn equation is used, eq. B.8.32, with Tj evaluated as if the gas were flowing along a dry wall. This is seriously in error. Again, I recommend using the interfacial drag coefficient already used in the hydrodynamic portion of the code. A still better result can be expected if the more exact solution is used (see Dukler, CEP Symposium Series 56, No. 30,1,1960). G. Bankoff wrote: "a) For low G heat transfer it may be possible to use the Jens-Lottes correlation. b) The Iloeje minimum film boiling temperature expression predicts super-critical temperatures in many cases. This is a case of compensating errors, since the transition boiling curve is faired between the CHF and MFB points on the q vs. AT diagram. Certainly, for low G it is not possible for Tmin > Tcr, and the Henry correlation should be employed. For high G it is possible, since inertial and turbulence effects destabilize the vapor film, but more documentation of the Iloeje expression in other geometries is needed. c) The use of the Chen correlation for heat transfer to slug flow needs to be checked experimentally. A possible alternative is heat transfer to a falling film in the slug region, and to a bubbly flow in the liquid region." S. Levy wrote: "a) I made the comment at the meeting that we are dealing with empirical correlations and the emphasis should be to find improved and more accurate correlations rather than unproven and simplified relations. For example, there are a variety of condensation correlations for horizontal and vertical direction and for low and high flow rates. What is needed is a correlation which pulls together the well known equations with as few discontinuities as possible. By contrast, what was proposed at the meeting was to utilize the Chen boiling correlation with S=0 to predict condensation. I doubt this is the right way to go and that it can account for vertical, horizontal laminar and turbulent flow. Also, before going to such a simplification, it would be worthwhile to compare it to available and appropriate experimental condensation data. b) I do not object to the changes made in the CHF correlations but in line with my recomendations of employing the best' empirical correlations, I believe that the time has come to develop a multirod correlation which recognizes specific fuel spacer configuration. The time to CHF is one of the key variables impacting TRAC results and this time will be specified best from multirod correlations. I suggest that the present correlations be retained to deal with broad configurations but that multirod correlations options be introduced and developed in future versions of TRAC co cover specific fuel geometries. 1871 211 . c) The use of iloeje correlation to predict mininun stable film boiling temperature is a good step forward but it is suggested that the proposed correlation be compared to available data be improved to recognize multirod configuration and, in particular, spacer effects be checked by sensitivity calculations to evaluate the shape of the transition boiling curve versus valces of minimum film boiling temperature. This is another key variable impacting TRAC results and it deserves continued and intensive investigation. d) I am against the introduction of equilibrium quality in Chen correlation unless one can show why it is preferable. e) shalldefertoothersintheConnitteeinthishka,.soI I did not completely understand the change in T f) The addition of Zr0 on the outside of the cladding is correct 2 but sone way should be developed to quantify this knob rather than leaving the value of Zr0 thickness so flexible. 2 P. Griffith wrote: " Dave Mandel is clearly confused about the meaning of quality when the velocity is low. In fact, when the velocity is low enough, the quality looses its significance and I think void fraction is a more descriptive variable. In counterflow the quality is undefined. In fact, he is having difficulty using quality because the equations which he is using are probably being extended below their region of validity. Under the circumstances I suggest that a low velocity heat transfer package be put together which is appropriate for the geometries and conditions of interest. I'd use the bounds given in Bjornard's thesis, in the absence of anything better, and pay particular attention to the surface temperature. If it's in film boiling I'd use natural convection (or forced convection) to vapor (whichever is larger). If the surface is cool enough I'd use a pool boiling equation. I'd use an interpolation scheme if they are in the transition region. In any case void fraction is the only meaningful variable in counter flow. Quality is inappropriate. Turning to some other, less important, items I think justifying the use of the Chen correlation for condensation is worthwhile. As far as the CHF correlation is concerned, Bjornard used Biasi for his heat transfer package even though it doesn't apply for a rod bundle geometry. A rod bundle correlation would have been better but a useful correlation must extrapolate to reasonable (pool boiling) values as G approaches 0 and as the quality approaches 100%. BothlinitsEIure 1871 . to be important in the transients of interest. We used Biasi because it was the only equation that extrapolated tc physical values. Without further experiments I believe we'll have to use empirical Tmin data to construct our boiling curve. I believe the best theory is Henry's however, and I'd suggest putting a calculation method into the based on surface code which uses Henry's method ard calculate a Tg thermal properties. Perhapsacurvefit,withox10ethicknessasa parameter, on the lloeje data would be good. I'd like the code user to know what he is assuming. We do need more experiments however. I would leave radiation out of regime 3 mentioned in Mandell's presentation. It is complicated, of no consequence and I don't see where it is ever used. In Fig. 6 of Mandell's presentation a two phase "K" is defined. It is either wrong or irrelevant and should be dropped. I cannot forsee where the condensation equation of Linehan and Grolms is going to be used so I wouldn't suggest dropping it (as Mandell proposes) without looking at the experiments that are to be correlated. It is a good equation for calculating condensation on jets and sometimes jets are used to inject water into the upper plenum. My final suggestion is to check the heat transfer logic so that the test of the heat transfer regime is done as economically as possible. B,iornard found that for his runs 70% of the time he was in film boiling. If he tested for that first, then he saved about 80% of the heat transfer computing time compared to starting his heat transfer regime check with forced convection to the liquid which is more " logical".
- 2) Thermohydraulic Correlations in the Vessel Module D. Liles discussed in some detail thennohydraulic correlations used in the vesse' module of TRAC (see Attachment 10). Topics which generated a considerable discussion were concerned with:
a) Flow regime map b) Modeling of interfacial momentum exchange c) Droplet entrainment d) Wall shear e) Phase separation f) Modeling of interfacial heat transfer briefly: 1871 213 . a) Flow Regime Map For vertical flows, TRAC distinguishes four flow regimes: bubbly, bubbly-slug, transition and annular-droplet. However, for horizontal flows a flow regime map has not been implemented. Liles noted that LASL would prefer to replace the flow regime map with a transport equation for interfaces or entities. The latter could accommodate also nucleation and coalescence rates which are not modeled presently in TRAC. The discussion that followed was concerned with: i) The adequacy of presently available flow regime maps D. Liles observed that the flow regime map used in TRAC, is based on data obtained with two-phase flows inside pipes and questioned its validity when applied to rod bundle geometries. Since A. Dukler is conducting a research program sponsored by NRC/WRSR to resolve this question, he summarized briefly the results of his investigations. Dukler noted that with one exception his data tend to support the flow regime transition criteria used presently in TRAC. The exception being the criterion for the onset of annular flow which needs to be modified. A topical report on these investigations is due to be issued in February 1980. ii) Flow Regimes for Downcomer At the July 1978 and January 1979 meetings of the ACRG it was recommended "that correlations and flow regimes should be selected according to a particular component and process rather than using a general set throughout the system." This was thought to be of particular importance to the modeling of downcomer where " bridging" can take place. At this meeting, the ACRG was informed that a " bridging" criterion will not be included in TRAC-PD2. Hearing this, V. Schrock and R. Lahey expressed their concern and advised LASL to include such a criterion because of its relevance to scaling. The following communication regarding flow regimes, were received after the meeting. S. Levy wrote: "The flow regime map needs to be improved as pointed out in the BNL memorandum and by Dukler at the meeting. I believe that a systematic effort should be made with time to recognize flow direction (vertical versus horizontal) types of flow (co-current versus counter current), and the effects of heat flux and multi-rod fuel geometry." 1871 214 . V. Schcock wrote: "I believe a nucleation delay model should be implemented in TRAC. Flow regime maps should be somewhat specialized to the component as suggested by BNL." R. Lahey wrote: "It was clear from the discussion we had that a " bridging criterion" was needed for the PWR downcomer region. I believe that this criterion should be a function of (at least) liquid subcooling and gap width. I think more data is needed to develop such a model and encourage the USNRC to pursue this important aspect of TRAC's ability to scale." "I believe that a complete flow regime map is needed for horizontal flow. I'd recommend the map proposed by Dukler." "I think a basic study of how to resolve two-phase interfaces should be conducted and implemented into TRAC." G. Bankoff wrote: "Another field equation for number density, using the lattice approximation of Moissis (MIT) for coalescence and a nucleation model for bubble nucleation, would allow a better representation for the detailed models." A. Dukler wrote: i) Flow Patterns for Upflow: I am sending directly to Dennis Liles our recommer.dations for modifying this flow pattern map. Specifically - The transition criteria of a = 0.25 within the most open space between rods is satisfactory for bubble to slug at low mass velocities. The transition criteria of 0.5 within the most open space between rods is satisfactory for bubble to churn at low mass velocities. A new transition criteria at higher mass flow rates which is more nearly linearly related to superficial velocity than mass velocity. A different criteria for the onset of annular flow. ii) Flow Patterns for Horizontal Pipes: I do not believe the same map should be used for horizontal flow as for vertical. For example, there are many situations where a > 0.25 where the flow is not slugging and is stratified. Furthermore, annular flow can take place with voids below 0.75. These transition criteria have been spelled out in the Taitel-Dukler 1976 paper. Each is described by a rather simple equation once the liquid level is known and this can come directly from the two fluid models. 18T1 215 . Annular / Drop Regime: Interfacial Momentum: There are two aspects which trouble me a good deal. Use of the Wallis correlation for interfacial shear is not in agreement with data, especially for counterflow. Wisman(Appl. Sci. Res., 30, 367, 1975) has recently shown that for annular flow the Siiiiilarity equations are quite satisfactory (Dukler, et al, AIChE J., 10,44,1969). Perhaps these can be tested against the data." R. Lahey wrote: "The work I have done recently using Serizawa's data indicates that interfacial drag coefficients of the form used in TRAC do agree with experimental results. In particular, I found [3] that Wallis' " dirty water" correlation did a decent job, and thus I reccmmend it be implemented into TRAC." c) Droplet Entrainment Wallis-Steen correlation is used in TRAC to compute droplet entrainment irrespective of flow regime, that is, it is being applied to reflood, annular-droplet regime and to entrainment from froth level. The validity of such an approach was questioned. S. Levy advised that the Wallis-Steen correlations should be tested against available experimenual data. N. Zuber noted that data and correlations for droplet entrainment from froth level were available and were discussed and made available to LASL during the Steam Generator Workshop held recently at NRC. Following the meeting the following communications were received: A. Dukler wrote: "Use of the Wallis/Steen correlation for entrainment leaves me very uncomfortable. I suggest that this be tested against data. R. Lahey wrote: "I am concerned about the adequacy of the correlation for liquid entrainment (E). While the work of Simpson [2] tends to support the Steen/Wallis approach (with only a viscosity modification needed to improve agreement with fluids other than water), I know of no data which supports its use for countercurrent flow, clearly more data is needed here. S. Levy wrote: "The entrainment model of Wallis-Steen should be checked against substantial Russian and European e',:am-w'ter data recently published (Toronto,1978). I belier inis will s.ow the need for an improved correlation. Some . ent work we have.9en doing for EPRI in this area might help to obtain such a correlution." 1871 216 . b) Modeling of Interfacial Momentum Exchange TRAC models interfacial momentum exchange according to flow regime (see Attachment 10). Correlations used for bubbly flow and for annular-droplet elicited several coments. For the bubbly flow regime, TRAC calculates bubble diameters from a Weber number criterion. Since this criterion accounts only for hydrodynamic effects such as bubble disintegration N. Zuber noted that bubble diamters calculated by TRAC cannot model thermal effects for exa nple bubble growth or collapse. He observed also that thermal effects are more important in gravity dominated flows than in high velocity flows where bubble disintegration is most probably the dominant factor in generating interfacial area. R. Lahey commented that future versions of TRAC should be able to model the effects of nucleation and of bubble growth. The interfacial area density for the annular droplet regime is calculated in TRAC by means of Eq. A.2.17 in Attachment 11. The adequacy and origin of this equation was questioned. G. Bankoff, S. Levy and R. Lahey advised that LASL should document sources of various correlations, explain any modification (if any) made by LASL and sumarize the data base upon which these correlations rest. Following the meeting the following communications were received: S. Levy wrote: In the annular flow regime, a Reynolds analogy relation might be superior to the constant coefficient used in TRAC. Also, it would be preferable to keep the wetted area as calculated and to put the multiplier of 5 in the heat transfer expression." A. Dukler wrote: " Bubble Flow: Interfacial Momentum: Rather than use one bubble size it is important to recognize that the critical Weber number gives the maximum stable size of the dispersion. Lots of evidence exist that the distribution can be represented by an upper limit type equation which is log normal. Thus from the distribution one can uniquely relate D to the Sauter mean, or max the number average or any other diameter. It would then be possible to use the correct diameter for the particular calculation it fits. With respect to finding the " correct" Weber number I recommend a look at the Sevik and Park paper (J. Fluids Engr., 95, 53, 1973). Again, at low flow rates, where a turbulent dispersion mechanism may not be functioning, a different critical Weber number may resul t. 1871 217 . d) Wall Shear TRAC computes the wall shear using Harwell correlation for both co-current and countercurrent flows. This appraoch was thought to be inconsistent when used in conjunction with a two-fluid model. A. Dukler noted that the wall shear correlation must be compatible with the interfacial shear correlation and advised LASL to resolve the inconsistency. S. Levy commented that the difficulties which TRAC sometimes has in computing steady state may be a consequence of the inconsistent correlations used to compute wall and interfacial shears. Following the meeting A. Dukler wrote: " Wall Shear (3D): In earlier aspects of the hydrodynamic package estimates are made of the interfacial shear for a variety of flow configurations. These must be consistent with the wall shear but completely different methods and correlations are used for the two. Given a flow pattern and the flow dimensions which are calculated (film thickness, bubble size, etc.), one can be calculated from the other. I suggest the consistency be checked. Do the wall shear calculated from the multipliers, when used to calculate the interfacial shear, compare reasonably well? If not, the discrepancy should be resolved or at least recognized as needing further study. S. Levy wrote: "I am concerned about a point which Dukler was beginning to raise at the meeting. The two-phase wall frictional losses were being calculated from empirical (Harwell) correlations; yet the interfacial shear losses were calculated by other relations. In an annular geometry it is possible to calculate the wall losses from the interfacial losses and vice versa. It is possible that the specification of both wall and interfacial losses might explain why one cannot always get steady state predictions from TRAC and this should be looked at closely." e) Phase Separation and Distribution BNL reported that it ran into difficulties when using TRAC to calculate phase separation measured in RPI slab tests. The problem was discussed with LASL, but the cause of instabilities has not been identified. R. Lahey comented on the importance to calculate realistically phase separation and distribution. He noted that TRAC does not model large eddy structure and consequently, that it cannot model phase distributions in open regions such as in the lower plenum. The latter is important since it determines the flow distribution into the core. Following the meeting he wrote: "As I discussed in the meeting, I think that BNL should compare both the 3-D and 1-D versions of TRAC with the Wye and Tee data given in NUREG/CR-0557. This data should clearly show if the 3-D two-fluid model can automatically predict the observed phase separation and how the cross flow quality of the drift-flux version of the code (1-D) 1871. 218 . needs to be modified to predict these trends. Both studies will have pay-off for the assessment of the fast and detailed versions of the code. I believe that one of the most serious deficiencies in the current generation of 3-D TRAC is the fact it has no modeling of the turbulence structure. This defect is particularly important when one is modeling the plenum and downcomer regions of LWR's. In particular, as noted in a recent EPRI study [4], the large eddy structure in the upper plenum of a BWR has a strong influence on CCFL break-down. Our 2-D data (without rods) that BNL is now analyzing should be very helpful in developing such a model. The countercurrent 2-D data we will be taking this year should also be of value in code assessment." f. Modeling of Interfacial Heat Transfer TRAC models interfacial heat transfer according to flow regimes; the correlations are shown in Attachment 10. In bubbly flow TRAC used the Weber number criterion to calculate bybble diameters which are used to determine the interfacial area density. R. Lahey questioned the validity of a Weber number criterion for low velocity flows. S. Levy asked LASL to evaluate the importance which various correlations have on the calculated results. In annular-droplet and film flow regimes, TRAC does not model the effect of interfacial shear on the heat transfer coefficient. A. Dukler noted that this effect is important and advised LASL to incorporate it. He added that this should not be a difficult task since TRAC already calculates the interfacial shear and uses it in the two momentum equations. Following the meeting, A. Dukler wrote: " Boiling: Interfacial Heat Transfer: It seems unlikely that the interfacial coefficient is independent of the interfacial drag. I suggest that Hanratty's work on absorption in sheared films (AIChE J., 25,122,1979) be examined and extended to heat transfer for an ii5iiroved approach. G. Bankoff wrote: "In determining Nu, for boiling, the Sideman-Moalem correlation for translating bubble growth might be considered, instead of the Plesset-Zwick bubble growth expression. L. Agee wrote: "The following are coments of a more specific nature and reflect my recommendations given during the meeting: Detailed verification of each individuel model in TRAC should be perfonned, preferably against relatively simple separate 1871 219 ' effects test (where detailed data exists) and the justification of the model application and limitations relative to these experiments carefully documented. All relationships should be " smoothed" (i.e. made continuous). This will eliminate numerous numerical problems and probably improve running time. All models should be documented, referenced, and checked for consistency in definition of parameters. The flow regime map should be extended for both vertical and horizontal application. The heat transfer surfaces should be extended probably in conjunction with the flow regime map. Care should be taken to use all the information from TRAC to define the most appropriate fluid conditions and then select an appropriate heat transfer correlation. The 1-D neutronic option should be capable to function in the radial direction as this is probably the dominant direction of interest for a PWR steamline break, (the steamline break is highly dependent on the neutron kinetic option for correct representation). In closing I would like to comment on the brief report given by A. Dukler (Univ. of Houston) on the work he is performing for NRC. I feel that his experiments are addressing some of the most important questions associated with two-fluid modeling. His work is highly relevant to the TRAC development activity as both the flow-regime maps and interfacial shear experiments are starting to supply the hard data required to develop the interphase models so badly lacking. I highly encourage this type of activity and hope other similar reports from other NRC contractors will be presented at future meetings." D. Themohydraulic Models in Loop Components of TRAC-Pl A D. Liles informed the ACRG that the modeling of loop components in TRAC is being changed from a drift flux to a two-fluid model. The new slip correlations will be based on those presented by Ishii. These changes will be made in TRAC-PD2. The discussion that followed dealt with: 1. Pressure Drop Calculations 2. Slip Velocity used in Pressurizer 3. Effect of Pump Efficiency 4. Modeling of Non-condensibles briefly 1. Pressure Drop Calculations TRAC has four user options to calculate the frictional pressure drop in 1-D components. S. Fabic advised that only one should 1871 220 . be available. He asked also how were the pressure drops due to spacers accounted for. D. Liles replied that these were to be specified by the user. BNL has noted that TRAC has two different treatments to calculate " form" losses due to abrupt expansion or contraction (see Attachment 11). S. Fabic asked whether or not TRAC can account for the effect of flow reversal through area expansions and contractions. D. Liles replied that flow reversal effects on form losses through abrupt area charges are not accounted for in TRAC. S. Fabic advised that TRAC should test for flow direction and calculate correctly pressure drop losses. 2. Slip Velocity Used in Pressurizer In the pressurizer and the accumulator, TRAC uses a special equation (see Eq. B.3.7 in Attachment 11) to calculate the relative velocity between the phases. N. Zuber inquired what was the experimental data basis and the justification for using it. D. Liles replied that it was introduced only to improve phase separation and that it does not have any experimental data base. He added however, that this equation will not be used in TRAC-PD2. N. Zuber commented that realistic modeling of phase separation in the pressurizer is important because of its effects on calculating the proper mass flow rate through the relief valve. 3. Effect of Pump Efficiency S. Fabic asked whether TRAC takes into account the energy contributed to the fluid by pumps. He noted that this energy can exceed the decay heat. D. Liles and R. Pryor replied that this effect is not accounted for in TRAC. S. Levy stressed that it should, since each pump can contribute up to 3MW to the fluid. D. Liles and R. Pryor agreed that LASL will include this effect in TRAC-PF1. 4. Modeling of Noncondensibles D. Liles infonned the ACRG that the effect of noncondensible gases will be modeled in TRAC through an additional continuity equation for the gas with an assumption that the gas and the vapor move with the same velocity. V. Schrock inquired how will TRAC treat the. solubility when different gases are present. S. Levy commented that source and sink terms will be needed in the continuity equation to account for hydrogen solubility which depends on pressure and temperature. D. Liles replies that LASL will start first with nitrogen. 1871 221 . E. TRAC Assessment at BNL S. Fabic infomed ACRG that independent TRAC assessment programs are being conducted at LASL, INEL and BNL. To facilitate the assessment of current models as well as to help in following the rrigress and improvements of future versions of TRAC, S. Fabic prepared the questionaire in Attachment 5. He noted that the justification of all models and relations in the TRAC series will have to be made by LASL.
- However, in order to alleviate the present heavy work load at LASL, he had asked BNL to prepare answers to the questions listed in Attachment 5.
Since BNL's role in the independent assessment of TRAC is directed at examining the treatment of basic thermohydraulic phenomena, he observed that the request will be helpful to BNL as well as the members and consultants of the ACRG. In response tc the NRC request, BNL prepared the document in Attachment 11, which was distributed to the members and participants of ACRG before the meeting and summarized by P. Saha at the meeting. Briefly, he made the following coments and recomendations regarding the items and equations enumerated in Attachment 11. Vessel Module: Item 1.2: TRAC needs to have a flow regire in the downcomer to describe " bridging." Eq. A.2.16: reference needed Eq. A.2.17: justify this equation, why is the multiplier (1-E) included, what is the data base for this equation Item 2.5: Effect of noncor%:,sible should be accounted for Items 2.6 and 2.7: Evalur.te whether or not nucleation models should be incli ded in the mass generation term Item 3.1.1: Should the crag include larger values of Re? Item 3.1.2: A. Dukler noted that the film in the slug is in free fall and therefore no accounting for the interfacial shear acting or the film is needed. Equs A.3.12 and A.3.13: Why the difference between the Eq. A.3.13 proposed by Wallis and Eq. A.3.12 used in TRAC? 1871 222 . Item 3.3: More clarification is needed as to how is the drag computed in the azimuthal direction when one control volume has bubbly flow and the adjacent film or annular - droplet flow. Item 4.3: The radial and axial pressure drops are not equal. Form losses are different. Poor description in TRAC manual as to how is AP computed. Some clarification is needed. Item 4.4: No models which account for the effect of area changes in the vessel on AP. TRAC must include this ef fect. Are the spacers included and hou? Item 4.5 and 4.6: May not be adequate for reflood Item 5: Inadequate for entrainment from froth levels. Item 6: Information needed on fall back calculations from UCSP as well as on how TRAC calculates CCFL at the UCSP. Item 8.1: What is the data base for using Eq. A.8.1 to calculate the Leiderfrost temperature? Item 8.5: The effect of radial velocity on heat transfer coefficients is not taken into account. Item 8.8: TRAC does not account for the effects of grid spacers on heat transfer. Loop Components Item 1.2: TRAC does not have the capability to model flow regimes in horizontal pipes which is important for small break analyses. Eq. B.3.7: What is the justification of this equation, what is its data base? Item 4.2: The modeling of expansion and contraction losses is not consistent. Item 4.3: TRAC does not treat void fraction distribution in T's and Y's. Similarly there is no phase separatior, model for T's. Item 8.3: Heat transfer selection criteria for the secondary side of S.G. are not included in TRAC. Eq. B.8.17, B.8.19, B.8.20: What is the experimental data base for each of these equations? Item 8.4.5: Free convection heat transfer to liquid is not modeled. Eq. B.8.27: What is the experimental data base for this equation? 1871 223 , Eq. B.8.31: Change D to L Eq. B.8.32: Is the interfacial shear Tj used in this equation consistent with the shear used in the momentum equation? Eq. B.8.35 and B.8.36: Heat transfer areas in these two equations are equal which is inconsistent. After the meeting the following communications were received concerning BNL's and LASL's efforts: A. Dukler srote: " General: a. I continue to be impressed by the comitment of the LASL TRAC group. The task undertaken is enormous. This work is still in a developmental stage. The results to date are impressive. This type of interaction with the review group on detail cannot help but improve the result. b. The Brookhaven report (Rohatgi and Saha) was extremely valuable in providing a road map to the detail of the heat, mass and momentum equations used in the code, c. I think it would be a mistake to move in the direction of using more approximate equations which ignore the actual physical phenomena. The virtue of the TRAC approach is that it has the potential to build in the details of the process. The LASL group brings to the problem a large bank account of superb skills and experience in numerical techniques. To couple that powerful feature with an inferior set of heat and momentum transfer equations will only result in a bastardized offspring. P. Griffith wrote: "I'm writing my cements on the Advanced Code Review Group Meeting of September 5-6. In general, I think they are making excellent progress on TRAC and certainly should be congratulated. In addition, I found the comments of Rohatgi and Saha exceedingly useful as they had the pungent flavor of user, rather than the philosophical bent that consultants often have. Let me begin with some specific comments on thi heat transfer package and continue with some general comments about the other parts of the code. I'm very much impressed with the progress on TRAC and want to congratulate LASL on it. V. Schrock wrote: "The assessment "Constitutives Relations in TRAC - PlA" by Rohatgi and Saha was very useful in the committee's discussion of the progress in the TRAC code. It was evident from the discussion that many of the physical representations of TRAC are essentially Ao Hoc and chosen for simplicity and "a lack of something better to use". In my view, there is a serious discrepancy between the mathematical sophistication and the physical representations in TRAC. There is a need for further discussion on the constitutive relations. Where Ad Hoc relations are chosen, it is important that they be verified to the extent possible by comparison with experimental data. If there is an inadequate data base the necessary supporting research should be defined and carried out. 1871 224 ~ . In general, the meeting was a very useful exchange of ideas and I am looking forward to further meet.ings of this type. I G. Bankoff wrote: "This letter is written in response to the request for comments concerning the Advanced Code Review Group meeting on Sept. 5-6. This was one of our better meetings, with a number of useful exchanges of ideas. The LASL presentations were uniformly excellent, and the BNL preliminary analysis was very helpful. Further work along these lines is needed to develop a document setting forth the correlations employed; explaining the empirical modifications; and summarizing the data base upon which the correlations rest. In general, I agree also with the suggestion by Levy that the tendency to reducc the number of empirical correlations by applying a few empirical expressions over a range of situations is unwise. For example, separate correlations should be employed for flow condensation vs. boiling. S. Levy wrote: "a) I believe that the sessicos of September 5-6, 1979 vere very worthwhile and I wish that more time had been allocated to discuss the various constitutive equations. I recommend a similar review after TRAC-PD2 and TRAC-PF1 become available. b) The work performed by Brookhaven was an excellent start. It pin-pointed several deficiencies and I concur with practically all their findings. It is strongly urged that such an independent assessment be continued and that Brookhaven be charged with verifying the various constitutive equations by comparison to data and proposing improvements. This is a major effort if it is to be done correctly and it is recommended that it be implemented with a schedule for the review of the various portions of the model. In carrying out this effort, it would be worth-while to prioritize the review efforts, in terms of their importance to TRAC predictions. It is also worthwhile to put emphasis on those relations which will be retained in the present and future versions of TRAC. Most of the constitutive relations are semi empirical in nature. c.It is important to document their source, accuracy, range of applicability, and shortcomings. It would be a gren' help if one could develop a way to assess which of the constitutive relations play an important role in TRAC predictions in order to focus studies and improvements in such areas. There is no need to initially spend substantial time and effort to argue about relations which do not impact the answers. Such a documentation would, also, be useful in identifying where additional tests are needed. d) I hope that the above coments are useful. I may have been somewhat overcritical in the preceding paragraphs but the only reason is that the constitutive equations are most important to the TRAC predictions. Also, in no way are the comments meant to detract from the good progress being made on TRAC. 1871 225 . R. Lahey wrote: "The recent meeting of the Advanced Coce Review Committee was one of the best we have had in recent times. We finally started getting into the physics in TRAC. In addition, I feel that the independent study by BNL was helpful and encourage more of this in the future. My cormlents on the meeting are tabulated below: As I stated during the meeting I think that the individual heat and momentum transfer package used in TRAC should be checked separately whenever possible. I think this is particularly important when the model utilized has been synthesized rather than obtained directly from the literature. As an example, the model used in TRAC for the transition boiling heat transfer coefficient to the liquid phase (equation E.8.17 in BNL review) should be compared with relevant data on transition boiling and phasic energy partition. This procedure should minimize the amount of computer time required for overall code assessment and should give us increased confidence in the individual models and/or correlations. I would now like to make a few cormients on scme of the pcints raised by the BNL study of the constitutive relations in TRAC-PlA.
- 1) The expression used for the interfacial area (A ) is obviously wrong since it does not include the relevant physics.j As noted in a study of the effect of obstacles (e.g., spacer components) on a liquid film [5], the critical liquid film flow rate (below which a dry-patch is formed) is a strong function of the geometry of the obstacle.
- 2) I don't agree with BNL's comments in section (2.6).
I don't think nucleation delay is of major importance in the analysis of LWR accidents. It would obviously improve things if implemented, and can be done in a rather straightforward manner. Moreover, if it was implemented then the code could do a much better job of predicting bubble diameter (i.e., could calculate bubble: Nucleation, growth, coalesence and breakup). Nevertheless, I feel this is an area of improvement that can wait for the future.
- 3) I am concerned that the treatment of grid spacers may not he adequate.
It is well known that spacers can have a dramatic effect on diversion cross flow, CHF and rewet (during reflood). J think that more work is clearly needed in this area.
- 4) The Harwell correlations being used for two-phase pressure drop 2
(& g ) appear to be proprietary HTFS correlations. I strongly o recommend that the USNRC obtain Hewitt's approval to use these correlations.
- 5) I do not believe that equation (B.8.33) is a valid expression for Tj, I recommend that the annular flow correlation of Wallis (or equivalent) he used instead.
1871 2pg s Attendance Members: S. Fabic, NRC/RES L. Shotkin, NRC/RES Attendees: L. Agee, EPRI S. G. Bankoff, NWU P. Griffith, MIT R. Lahey, RPI S. Levy, SLI V. E. Schrock, UCB C. Truesdell, JHU I. Catton, NRC/ACRS A. E. Dukler, NRC/ACRS P. 5. Andersen, NRC Consultant W. C. Lyon, NRC/RES W. Kato, BNL V. Rohatgi, BNL P. Saha, BNL P. Litteneker, DOE /ID Y. S. Chen, INEL A. Peterson, INEL F. Addessio, LASL D. Liles, LASL D. A. Mandell, LASL R. J. Pryor, LASL S. Woodruff, LASL L. D. Baxton, Sandia P. K. Cole, Sandia J. J. Cudlin, B&W s 1871 227