ML20246F755
| ML20246F755 | |
| Person / Time | |
|---|---|
| Issue date: | 11/16/1988 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Houston D Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-CT-1943, NUDOCS 8905150080 | |
| Download: ML20246F755 (3) | |
Text
. _ _ _ _ _ _ _ _ _
{ [---
($}C f/Mf P
TO:
Dean Houston 16 Novembsr~1'988 FROM:
Ivan Catton M
D HiMME REACTOR SAFEGUE U
SUBJECT:
Severe Accident Research Partners Meeting Bethesda, Maryland NOV 211968 20-21 October 1988 MX gg the above ref erenc,&9,hkMAW 7 ed meeting.
I attended the final two days of Material I was interested in was scheduled at hke beginning or I chose the part of the meeting where Direct Containment Heating pe (DCH) was to be discussed. There have been several code studies
'di 4
of the core degradation process and the impact of natural circulation. The results point to DCH being much lessor issue than SNL would lead one to believe. The SNL experimental and analytical studies are still based on too much molten core.
The issue of how much, how hot and how fast still does not receive the attention it deserves. The Mark I liner failure issue suffers from the same problem. Convincing arguments were given that the liner will fail when large fractions of the core must be dealt with. More detailed discussion of the meeting topics follows.
Dana Powers discuemed a number of issues related to DCH. He made the point that no t itainment i s immune. The source term was not recognized in WASH.400 and as a result has not been properly quantified. To quantify the source term one must first determine how much energy gets into the containment (quantif y debris dispersal from the cavity and chemical reactions with the atmosphere) and second determine how much hydrogen production will occur. Surtsey (1/10 scale) experiments were conducted to try and understand some of the physical processes important to DCH. The first few experiments have established that increasing mass of the injected molten materials resulted in a decreasing energy release per unit mass. This is not a surprising result although it is certainly comforting to see it based on experiment rather than supposition. Contrary to the Bob Henry views, it was found that structures in the flow path can enhance energy transfer to the containment atmosphere. It is not clear how this conclusion translates to a full scale PWR as most of the structure referred to in the past is in the keyway or corridors leading from the reactor cavity to the containment volume.
Models being developed to predict the DCH induced containment pressure are CONTAIN and KIVA. CONTAIN is a lumped parameter code that has been around for a long time. It is interesting that the code that was to address issues like DCH was at one time to be HECTOR (sp?). It appears that HECTOR has fallen from favor. KIVA is claimed to be a detailed hydrodynamics code based on finite difference methods that will do a mechanistic analysis.
Considering our ability to predict such phenomena, I view the use of KIVA as fun and games. CONTAIN has been augmented with models to describe debris clouds, structural interactions and chemistry. It was claimed that the codes do reasonably well in predicting DCH.
(SN DESIGNATED ORIGINAL 8905150080 881116
I g,y
2-l
~
Accident management studies are underway at SNL. Depressurization evaluation led to the conclusion that one must get to very low' pressures to achieve beneficial effects. The experiments show that there is a critical pressure where core debris dispersal goes from essentially zero to all of it. Without understanding the physics i
one cannot begin to guess what the effect of scale will be. An interesting bottom line was reached: DCH impact is small if less than 20% of the core is involved. The basis for this was not clear.
Further there was no discussion of the rate of dispersal.
SNL has still not looked carefully at the initial conditions for DCH.
Power s presentation seemed to be based on massive amounts of the core and varying fractions being dispersed. It seems to me that CONTAIN is good enough and its time to put our efforts into determining just what it is that we must deal with. We always get back to i) whether or not the primary system will fail before the core melts through the bottom and if it does, ii) how much of it must we deal with, iii) how hot will it be, and iv) how rapidly will it be blown into the containment atmosphere. The first item requires that natural ci r c ul at i or-induced heatup of the upper parts of the primary system be properly eval uated. This is being studied and is ciscussed below.
Ginsberg described the 1/40th scale experimental work underway at BNL.
Ginsberg has injected the simulated core materials ir co a large number of fluids. These fluids include water and liquid nitrogen. I could not establish what simulant characteristics were used to select the various combinations. It seems to me that energy transfer to the atmosphere is l
dependent on the drop size of the atomi:ed core materials which means the Weber number is an important parameter yet it was not mentioned.
High pressure blowdown usually means Mach number should be another i mportant par..neter yet it was not mentioned. Energy transport and chemistry also depend on the Reynolds number (based on the relative velocity). It also does not seem to be a part of the scaling analysis.
Either I did not understand what I was being told or the BNL program is just another group of nuclear engineers having fun. Work like that described can be an important part of developing the needed models for prediction in codes like contain. One must, however, pay attention to scaling and if the conclusions are at odds with the views of others, some effort should be made to enlighten.
Bradl ey (SNL) described some recent experiments that were to address the Mark I liner failure concern. The experiments were poorly conceived and demonstrated a lack of understanding of the important physical processes.
A series of very nice basic experiments are being conducted by Greene at BNL. He has systematically addressed many aspects of the Mark I wall failure question. His work should now be used to address the problem. It is interesting that Greene is one of the few involved in.
severe accident phenomena studies who consistently publishes his work in respectable archival journals. It seems much of the NRC sponsored research is either unpublishable or they do not care. The discussions that followed his presentation put a different complexion on the issue of Mark I iiner failure. The liner that may fail is less than ten feet from the pedasti doorway. It was argued that the molten core materials will slosh out and literally run up the wall. The question always gets back to how much, how fast and how hot. If a large fraction of the core must be dealt with, then further study will not change the conclusion that one must protect the liner.
I Alsmeyer from Germany presented the results of a study of V;at happens to the molten core on the concrete basemat. He preU cts that it will penetrate in about five days or so. Containment evaluation should take this into consideration. To stop the downward propagation, one must chill and fracture the debris. Water from above will not do the job as it only fractures a thin surface layer. The results of his study sounded like the China Syndrome revisited. He ~
found that the molten core penetrated 80 meters into the earth and spread laterally to 30 meters in a period of two to three years (certainly lots of time for intervention).
In-vessel core melt studies using RELAPS/SCDAP were was described by Allison (INEL). I am continually amazed at how well one dimensional codes can describe two dimensional behavior. Allison considers crust failure (around the molten pool) a major factor in molten pool rel ocati on. How SCDAP addresses this without a detailed pool convection model is truly amazing.
A similar analysis was carried out using MELPROG/ TRAC. It sounded good but Heames (LANL) did not seem to know too much about how the code did its job.
Natural ci rcul ati on during a TLMB' incident was studied using SCDAP/RELAP5. To accommodate hot leg countercurrent single phase stratified flow they modeled the hot leg as two pipes. The core was represented by three parallel channels. The modeling was quite crude but probably adequate to give some indication of whether the lower head would be the vessel failure location. It was concluded that the surge line will fail, the steam generator tubes will not fail and that the surge line failure will proceed the lower head failure by one hour. Although I would take issue with the conclusion regarding the steem generator failure, that the failure would not be the lower head is reasonable. A number of sensitivity studies were carried out and it was found that nothing seemed to change the conclusion that the ex-vessel part of the RCS would fail and depressurire the system before the the lower head fails. Each new series of calculations of in-vessel natural circulation seems to point more convincingly to early failure of the primary system and a decreased probability of DCH induced containment failure. In this regard, I was pleased to hear that Zuber will be looking into the DCH issue.
Bergeron (SNL) described his studies of DCH using CONTAIN based on 75% of the core. Although his study was well done, using 75% of the core as an initial condition prejudices me. He found, in contradiction to Power's interpretation of the Surtsey experiments, that structure mitigates the impact by absorbing energy through heat transfer to it. He found that the biggest uncertainty is whether or not hydrogen burns. It was found that debris transport to the upper containment from the reactor cavity is not the dominant mechanism for energy transport. Rather the steam acts as a working fluid. Bergeron concluded by noting that DCH seems to be less threatening to large dry PWRs than it was earlier thought to be.
-_.