ML20038C602
| ML20038C602 | |
| Person / Time | |
|---|---|
| Issue date: | 09/22/1981 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Boehnert P Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-CT-1376, NUDOCS 8112110266 | |
| Download: ML20038C602 (3) | |
Text
_w_
O I-O C0PY 70$
September 22, 1981 g
N y.
} v.s.sveunmag; ann O V 8 0 MEMORANDUM T0:
Paul Boehnert FROM:
Ivan Catton V
j
SUBJECT:
ECCS SUBCOMMITTEE MEETING, AUGUST 28, 1981, MONTEREY, CA General Comments on Appendix K - Appendix K was implemented to assure that sufficient conservatism would exist when large break LOCA was used as the design basis accident. The result has been a new dimension in physics called " Appendix K Space" (Jim Cermak).
Dealing with Appendix K physics has lead to a number of anomolous or intuitively incorrect results that have had an impact on operating set points, procedures and other aspects of plant operations. The present state of affairs is not solely the re-suit of Appendix K.
The vendors have put together overly simple one-dimensional codes that meet Appendix K, (the EM code) have had them blessed by NRC and use them even when they suspect they may be wrong.-
There is no strong incentive to do otherwise.
An example of a problem arising from Appendix K is the UHI system and ECC bypass. Using an Appendix K EM code results in predicting a great deal of water going directly out the break. This results because th_ downcomer is
,/
two dimensional and the EM code is one-dimensional. fis known from TFi tests that the ECC bypass criterion is overly conservative and that UHI should be helpful. A number of the steps taken to force the EM codes to demonstrate that all was well are not physically correct.
A similar problem resulting from the one-dimensional codes is_ how to get the W 2-loop plant upper plenum injection water into the lower plenum.
I don't know how this was resolved within Appendix K space.
t I
We are all familiar with the initial vendor calculations of what happeris during a small break. The fixes have been slow in coming and many predicted results may still be the result of the EM codes. There is a continuing effort to make the predictions as realistic as possible while staying within the law. As noted by Dr. Plesset, Appendix K acts as a hugh security blanket under which we hide our ignorance.
A number of training simulators still use the results of calculations found in plant FSARs.
Even though they are being encouraged to change, the process is slow or doesn't happen at all. The vendors supply the accident description to be used in the simulator. Operating procedures for accident control are, inpart, based on calculations.
Even though 8112110266 810922 PDR ACRS CT-1376 PDR
C0PY bi Paul Soehnert September 22, 1981 attempts are made to make the calculations best estimate, there are remnents of EM codes still around. Too many of the responsible people have been trained in EM physics and need to be weaned.
NRR has instituted a " Required Problem Program" where utilities are told to demonstrate their calculation expertise by reproducing LOFT and Semiscalt.
results. The utilities, for the most part, go to the vendor through their respective owner's groups. Again EM models are used for calculation and words are used to argue why they are not doing to well.
In principal the program is a good one but as long as the experiment is not conducted on a full scale plant, trying to make an EM code do well is not necessarily pro-ductive.
It would be encouraging if the utilities (vendors) would develop best estimate codes on their own.
I don't believe, however, that they will.
A great deal has transpired since Appendix K became law. A large number of separate effects experiments have been carried out.
Small scale integral facilities now exist in several countries and are producing data at a high rate.
Through the efforts of EPRI and others, full scale plant transient data is being obtained.
LOFT has reproduced a number of important reactor accidents.
The RES advanced code program has produced several best estimate codes that are now available to the public.
EPRI has sponsored the develop-ment of RETRAN.
Industry should be strongly encouraged to use best estimate codes in all aspects of nuclear power plant licensing, design and operation.
The transition from Appendix K EM codes to best estimate codes should not be piecemeal.
A broad, all encompassing, plan should be followed.
GE Request For Appendix K Change - All available experimental data show that the GE EM model has significant margin.
Unfortunately the model has even less to do with reality than the PWR vendor models. The GE EM model consists in essence of the following single node components:
Component Modeling Core Homogeneous-Equilibrium Bypass Homogeneous-Equilibrium Lower Plenum Homogeneous-Equilibrium Downcomer Homogeneous-Equilibrium Recirculation loop Homogeneous-Equilibrium Recirculation Loop Homogenecus-Equilibrium Upper Plenum Homogeneous-Equilibrium Separation Zone Input Steam Separation and Homogeneous-Equilibrium Steam Dome Homogeneous-Equilibrium A
5' COPY Psul Boehnert September 22, 1981 Data from TLTA show that subcooling (non-equilibrium) and separated flow are important if the observed phenomena are to be tracked.
The BWR TRAC (TRAC-BDI) has difficulty representing the phenomena with a large number of nodes.
CCFL taking place at the top of the fuel cannister, at the side entry onfices and within the bypass channel all have dramatic effects on the time behavior of the thermal hydraulics during an accident. This could have a dramatic effect on the accident seen from the control room. Many of these behavior are qualitatively observed with the RPI freon loop (made of glass for visual obsevation) and measured with TLTA. The GE EM model is too non-physical and simplistic even for an EM model. CH is probably fortuitous that the predicted results are so conservative.
(g A LOCA, large or small, is much easier to mitigate in a SWR than a PWR. The multiplicity of pumps and methods of water injection along with the ADS give the CWR a great deal of margin under most circumstances. Therefore, the GE request for relief from ANS + 20 should be given early consideration.
A re-duction in decay heat should not, however, be given without a commitment from GE to completely redo their LOCA model.
A commitment to do something was mentioned at the subcommittee meeting. Their SAFE code, for example, does not appear to be a satifactory way of meeting such a commitment.
The ECCS Subcommittee should review what GE has in mind for their new LOCA model and when they plan to have it approved and in place.
Pumps On or Off - Vendor calculations based on codes of unknown validity led the industry to a pumps-off action if a small break were to occur.
Subsequent experimental investigation at LOFT and Semiscale has shown that continued pump operation during a small break leads to greater primary coolant deple-tion that when the pumps are tripped early in the accident.
Time available before the pumps must be tripped appears enough for CE and W PWRs to allow it to be manual whereas B&W (2 min) most likely should use automatic trip.
The times, off course, are the results of suspicious calculations.
At one time the question of whether a particular pressure reduction was overcooling or a small break was of interest. To eliminate this question, presure set points for pump trip are made low enough that overcooling pressure decrease will not trip the pumps. This seems to be a rather simple solution to what was a perplexing problem. How the decision was made as to what is low enough was not discussed.
Further does the set point miss some small breaks (pressure hangs up at a valve).
Establishtng whether a given pressure reduction is due to a small break or overcooling seems simple enough that it ought to be given some attention.
For example, pressure decrease at fairly constant cold leg temperature implies a break whereas pressure decrease with cold leg temperature decrease implies overcooling.
System heat balances (S.G. and RPV) could dlso give an indication of what is causing the pressure decrease.