ML19248C824
| ML19248C824 | |
| Person / Time | |
|---|---|
| Issue date: | 05/01/1979 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Bates A Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-CT-1110, NUDOCS 7907050350 | |
| Download: ML19248C824 (3) | |
Text
.-
- ~
f* "%
DISTRosu7to
,,,9
[ 'ig N U C L E A R R E G U L ATonrco.'.1).11SS I O N
[pg.S~
,j ADVISnRY COMMITTEE ON REACTOR SAFEGUARDS, g
WASHING ton. o. C. 20555 g.,. g,g.,. ( g, maca s.:.v.
April 30, 1979 1.! A l q EJ/j Ytc;f..,.-: b M IJ S
70:
Dr. Andrew Bates si "s t y....u FROM:
Ivan Catton
/
g
SUBJECT:
ECCS SUBCOMMITTEE MEETIIG MARCII 19-20, 1979, LOS A';GELES, CALIFORNIA COPIES 70:
M. S. Plesset A great deal c f information on a number of subjects was presented at the two day meeting.
I have tried to highlight what I feel are the important aspects of each part of the meeting. Tne following paragraphs contain my comments and suggestions rather than a detailM summary.
Code Work on Transient Two-Phase Flow. Code work on transient two phase flow is primarily embodied in TRAC for system analysis and COBRA for sub-channel analysis. We codes secc to han a sophistication that our under-standing of there physical processes cannot match and may never match.
Dr. Wendroff discussed several aspects the ill-posedness question, liis definition of a well-posed sc~ition is ticularly clear:
for each data set there is a unique solution.
Unfort
._t ly there is little chance for any problem to pass the test. To get around this situation there are two schools of thought:
1.
do whatever one must do to make the equations being solved pass the test, 2.
don't worry about the problem - if the code runs count your blessings and if it doesn't renode.
It is not clear how a code like TRAC is made well-posed, if it is.
It is suspected by some that the differencing algorithm causes a great deal of damping which helps. h ere is no way of knowing how much false dissipa*. ion occurs in TRAC.
I have a general concern.
Some combinations of models and numerics seaa to overcome what appear to be insurmountable difficulties (inherent instabilities associated with flow regime changes) by numerical damping.
IT is difficult to decide where we are at.
Status of_ Physical Inputs to Codes.
Dr. Fabic gave an excellant summary of where the data for code developaent will come from. Much of it is large scale experiments and much of it is foreign.
It is possible that he will Le 7907050S50 qOr tbl uJ n
- s--
ECCS Mtg 3/19-20/79 April 30, 1979 succ'essful, but not without a great deal of effort.
Few of the programs
~
were designed to yield the information he is seeking.
The LASL presentation by Dr. Liles skirted the issue of transient flow regime behavior.
Dr. Liles indicated that flow regime data was not needed directly because the codes were developed from first principles.
There are no programs I am aware of that will yield data on phase interface area and its behavior dut ing transient conditions.
IF Dr. Liles position is the one under which TRAC is being developed, we have embarked on a life-time project.
Its my feeling that NRC's view is not so all encompassing.
The work on the reflood model for TRAC seems to be a bit behind the state of the art.
Dr. Kirchner does not seem to have all the available informa-tion.
The inter-relationship between precursor cooling and Thin resulting from the rate of change of temperature has lead to a great deal of difficulty for the code devleoper. A differcnt flow regime changes the precursor cooling and can lead one into using the wrong Thin. The result is inconsistant and undependable results.
Standard Problem Proaram.
The basic idea of a standard problem program is a good one.
It allows one to take the measure of a given vendors capability-grade him. There are, however, many difficulties associated with obtaining venaor participation.
The vendor's codes are aimed at satisfying U4 require-m7nts whereas the standard problems are actual physical problems.
There is no strong reasons for a vendor to become publicly involved with the program.
As a matter of fact one can see many reasons why he does not.
Mr. Phillips of the staff indicated that the program does not have any requirements for participation. As a result, the primary participants in the program at present are those associated with licensing in the US and foreign countries.
Little participation of US or foreign vendors is taking place.
It is my belief that certain, if not all, standard problems should be used in the certification of the vendors code.
This would maintain parti-cipation and continual code development within the vendor's technical groups.
If the vendors are to participate, and I believe they should, then some better methods of inducement are needed.
The equivalent of a standard problem pro-gram already exists within NRC as the code verification program. Without the vendors, a standard problem program is duplication of efforts.
ODYN Code Review. On the surface, the ODYN code appears to be adequate for the task it must address.
It is a finite difference code and uses a difference algorithm that results in a highly damped result.
This can be seen if one compares the measured results from Peach Bottom wit.h the ODYN predictior The measured results show pressure waves and expansion waves travelinc aack and forth in the steam line.
Fbst of this structure is missing trom the ODYN result. The response of comprcssible flow in a pipe to valve closure is usually determined using the method of characteristics G v1 90
/. u !
ca0
ECCS Mtg 3/19-20/79 April 30, 1979 to avoid problems such as encountered in OD'd. The most important part of solution is the rate of pressure rise in steam dome of the vessel.
The steam dome pressure reproduced fairly well.
Its not clear what the limita-tions of the code are.
This will not be a problem as long as the code is only used under circumstances where there is experimental verification.
Analysis of Se:aiscale Test S-07-6.
There secm to be a number of efforts underway to predict the oscillatory behavior observed in S-07-6.
Once the oscillatory process is initiated, it seems to carry forward on its own.
Ex-cess heat transfer is only a possible initiator of the process.
It is my opinion that Flashing causes high flow into the pressure suppression tank and condensation in the lower plenum causes a high rate of flow back.
lner-tial effects cause an overshoot and the resulting excess pressure starts the process again.
Tb analyze suci a process one can treat the lower plenum as a bubble with an effective volume to account for condensation and arrive at a frequency that is not ur.
asonable. The slowly increasing amplitude is probably due to the pump adding energy to the flow. Under such circumstances, two solutions to the governing equations exist and all of the codes are written to pick the non-oscillatory solution.
I believe it is a waste of time to pursue the problem further with codes like REUP or TRAC.
Westinghouse 2-Loop and UHI Plants. The planaed use of TRAC and COBRA to address the 3-D character of rPa W 7-I. cop Plant upper plenan injection and upper head injection plants is the first realistic attempt at analyzing a very complex problem.
The results of the two studies will be of interest both in the problems addressed and completing the development of some very useful tools.
b U l 707 e*
L.!,