ML20215L338
| ML20215L338 | |
| Person / Time | |
|---|---|
| Issue date: | 09/28/1986 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Houston D Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-CT-1865, NUDOCS 8610280584 | |
| Download: ML20215L338 (3) | |
Text
- - _ - -
C T-I%6 l
~
~
Qf /f$~
b I
i 28 Septembeh 1986
=.
TO:
Dean Houston
-~
- 18.
- FROM:
Ivan Catton
SUBJECT:
Se. vere Accidents and Nuclear Plant Chemistry Meeting 24 September 1986 Washington, DC
" Regulatory Implementation Plan for Severe Accident Plan" A key element to development of a " Regulatory Implementation Plan for the Severe Accident Plan" will be the " Introduction of Realistic Source Term Estimates into Licensing" (Secy-86-228). Secy-86-228 states that methodology spelled out in NUREG-0956 will be the tool used to assess the accident' sequences and their source terms. The methodology outlined in NUREG-0956 is called the Source Term Code Package (STCP). The Severe Accidents Subcommittee has raised a number of questions about the STCP during the past year or so. These questions were summarized by the subcommittee chairman in August 1985. Consultant reports during the past year amplify many of the concerns about the STCP. Although many of the concerns have been acknowledged, they have not been addressed in a meaningfull way. Until this is done, it is (in my view) premature to use the STCP to assess the various accident sequences postulated as part of the severe accident source term world. Using the STCP to decide whether or not the IDCOR codes yield results that are meaningful for licensing decisions is just asking whether or not one groups' set of assumptions somehow yield results similar to anothers. They may both be wrong.
The source term will only be calculated more accurately if our assumed phenomenological modeling is correct. The " realistic models" in both STCP and IDCOR codes are,in many cases, someones view of reality. There are some of us (the heat transfer community) who sincerely believe that the calculation of core melt temperature requires one to know how the molten core couples to its surrounds rather than to assume an upper limit on the temperature. Many of my concerns about the process are exemplified by the NRR attempt to obtain resolution of the so-called issues between NRC cnd IDCOR on modeling of severe accident phenomena. Some discussion of the
" issues" and the resolution process are given in the following m
@ @ paragraphs.
ISSUE 1-Fission Product Release Prior to Vessel Failure. The IDCOR O
conclusion, based on MAAP results, that total in-vessel fission
$m pr duct release is relatively independent of core melt temperature over the range 2500 to 3100 deg K is surprising. The reason'for nm
@y$ this result deserves amplification. As little as a year ago, it g $ was thought,that temperature was the single most important
$@$ parameter Jts value impacts everything f rom steam explosions to
-mi how the vessel fails and the question of direct heating. I don't understand how such an about face in our thinking could occur. Just because core melt temperature is difficult to evaluate, we cannot assume it is relatively unimportant. It seems to me that an explanation of the physical processes leading to such a result are needed before such a conclusion can be accepted and the issue be considered resolved.
Dr3IC"!Q ORIGIIIAL Cort 1tisa 37 fh]$
~
i
/
je a/
2
=.
ISSUE 2-Re-circulation. A fairly good picture of our state of knowledge is given. The uncoupling of fission product migration from recirculation is, however, not justified. It can, for example, Icad to steam generator rupture among other things. Further, the EPRI in-vessel recirculation modeling has demonstrated the importance"of coupled fission product transport and thermal hydraulics. The EPRI results should be viewed as more " realistic" than either the simple minded approach taken in MARCH or even that of MELPROG. A sensitivity study based on MARCH will yield little of value. It is beyond my comprehension why the published works of EPRI are not used, at least as a sta-ting point, as a guide for future analysis.
ISSUES 4 and 12-Fissien Product and Aerosol Deposition is RCS cnd Containment. The statement that there is no direct evidence that the aerosol correlations used in MAAP code are too inaccurate implies that we have an idea about how accurate they must be. That they are not too inaccurate seems to me to be an awfully weak basis for acceptance. Given the state of knowledge, a correlation seems to be a much better approach than a highly suspect computer code that is e:: pensive and dif ficult to evaluate.
ISSUES 5 and 6-Hydrogen Production and Core Melt Progression and Vessel Failure. Core melt progression is a big unknown. Both STCP and the IDCOR codes are based on a set of views as to how the process progresses. Some of the views are reasonable (my view) and some are not and some phenomena are not considered at all. The staff questions more the differences between the sets of codes than missing physics. The IDCOR " levitation based blockage" model seems to me to be pure science fiction. I agree with the staff conclusion that the IDCOR model has not yet.been adequately substantiated. I would argue that the STOP has not been adequatel y substantiated either. The strong dependence of hydrogen production on the assumed clad / fuel relocation temperature makes meaningful estimates difficult. Three codes based on three sets of developer assumptions (STCP, IDCOR codes, MELPROG) will yield little usable for uncertainty estimates unless some basis is found for some of the modeling. It is not clear why the EPRI recirculation model is not used. It has, in my view, the most physically based set of models.
The staff report discussing the issues and their resciution does not contain information on how they arrived at their position.
One is left with the unfair view that resolution is arrived at through negotiation of the physics. This leads one to conclude that resolution of an issue does not necessarily mean that we understand the process well enough to bound it. The 18 issues include those aspects of s_evere accidents we must understand if we are to make the source term-more realistic.
I 3
't The IDCOR IPE methodology looks very promising. The use of a simplified source term estimate approach sounds reasonable when one considers the successes of the more complex approaches. This is particularly true when our measure of success is 'is based on simple intui tive estimates. Answers to the staff questions about the method (letter from Speis to Buhl) should be helpful in f ully evaluating-the the IPE method. The important benefit of IPE is that judgements needed to complete the process are visible and can be evaluated by others. One is left with the feeling that IPE is more hardware based than computer based and that has got to help. On the other hand one must not loose sight of the incomplete physical descriptions of the core melt process used by both the IDCOR IPE and the STCP to arrive at risk conclunions.
e m
- P