ML19323H795
| ML19323H795 | |
| Person / Time | |
|---|---|
| Issue date: | 06/10/1980 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Quttschreiber Advisory Committee on Reactor Safeguards |
| References | |
| FOIA-80-505 ACRS-CT-1244, NUDOCS 8006160195 | |
| Download: ML19323H795 (4) | |
Text
_ -_ - _________
).
c,./2 W T0: Gary Qui.ttschreiber 2
d FROM:
- 1. Cat}o'n USA /FRG Fuel Melt Research progra., Review and Information Exchange
~
SUBJECT:
Meeting, 11-12 April, 1980 a
The USA research sponsored by NRC-RSR was presented and no new informa-In that i.t has been presented before and my comments have pre-The work done in support of tion surfaced.
viously been expressed, I'will not reiterate.
the Rogovin report was discussed by Dr. R. Denning of BCL and will app Vol. II of the Rogovin report.
are worth noting:
Can't predict severity of fuel damage based on MARCH type' analysis, 1.
Uncertainties in boundary conditions are too great, and 2.
There are a number of modelling uncertainties.
3.
Apparently there are a number of versions of codes such as ORIGIN and RELAP was noted to as a result different people obtain different answers.Actually results seem to have done better in a T!11-2 calculation than TRAC.
have converged reasonably well considering the complexity of the problem.
Dr. Benjamin noted that containment model uncertainties are significant and may dominate vented filter design.
Its purpose is to study Dr. Barnacik, GRS, described the Beta facility.In that tb2 FRG containment design tries concrete penetration by a core melt.
Three regimes to keep the reactor cavity dry, no tests will be run with water.
of melt-concrete attack have been established:
1.
1900 < T < 2400 C Vigorous attack by a well stirred pool p
0 < t < 7 hrs of molten core debris. Vessel failure is assumed to yield a molten mixture at 2400 C.
2.
liixture Melt Temperature < T < 1900 C pA more quiescent melting attack where 7 < t < 8 hrs concrete thermal properties add possible pool segregation are important. They have some concern about layer to layer heat transfer.
J.
T 4 Mixture Melt Temperature Freezing and re-nelting will be occurring.
p.
8 Ers < t This regime is the least understood of the three The first tests will be the following:
8006180l9E
a
'o Power = 8.4 MW Induction H2ater at 1000 H3 T = 2000*C with at least 38% efficient coupling E
300 < Mass of fielt < 600 Kg Iron t = 300 min continuous An acoustic Results from.this series of tests will be available in mid 1983.
method developed by SANDIA will be used to follow the decompos surface.
This does not address the question of penetration when I
and 2 meters deep.
the cross-sectional area is very large and the gases resulting from decompo-sition can spread the melt into a thin layer.
Dr. Hoseman presented ar. overview of the FRG work on soil penetration They have chosen a soil that is (Dr., Peeks could not attend the meeting).
a mix of A1 0, CACO 3 and silica sand (SiO ).
A series of experiments is planned 2
starting with 4 kg of steel, inductively heated at 6.34 kw/hr on both wet and 23 They will also study the soil mix, wet and dry, with both steel dry sand.
and a corium mix. Tests run so far show lots of crusts and bubbles with wet A typical sand penetration experiment shows sand having less penetration.
a layering-corium melt, crust, dry steam filled sand, heated wet san an unaffected zone.
aporation at the dry zone-wet zone interface.
taking place in the wet zone and The FRG engineers believe steam explosions are a very complicated function A number of at least temperature, thermal properties, geometry and mass ratio.
of their ideas are :
The efficiency of the interaction is less than ten percent based on ideal a)
The SANDIA small shock tube tests as other geometries will be less.
scale tests showing 15% and higher are too idealistic.
Estimate work 50 to 60 ms after triggering to be 50 to 60% of the 10%
b) available.
An increase of system prassure could suppress the explosion but good rapid c) mixing will override the cressure suppression effect (this is in contrast to Henry's hypothesis).
Theories are hard to validate and there are many reservations about models.
d)
Extrapolation of results from less than 1 Kg to 10d0 Kg yields conserva-e) tive results because small experiments are more ideal for an efficient rapid interaction, FRG will rely on the SANDIA large scale tests to confinn the conservatism f) resulting from extrapolation.
Rather than attempt to develop theories, it is believed to be better to l
g) use engineering judgment and get on with the job.
An estimate must be made of the maximum expected mass that might inter-h) act so that an estimate of the maximum loads on the RPV can be made..
Factors of importance are, contact mode, coherence of the reaction, m tios.
velocity, how reaction is triggered, superheat of melt, and volume ra 1) must focus on the pre-fragmentation phase and contact modes.
e
~
One Preliminary estimates are that there will be 40 tons of melt and that j) 5 tons will interact.
Projects associated with containment hydrogen are as follows Containment thennal hydraulics, Dr. Mayinger at TUH h.
1.
dispersing, Dr. Jahn at GRS Munich Containment code including H2 2.
dispersion, Dr. Kanzlieter at Battele Frankfort 3.
Experimer.tal studies of H2
- Rather, detonation is a problem.
Presently the Germans do not believe that Hfrom a zirc-steam reaction plus e ste 2
time due the combination of H2 will almost always cause a cont inment overpressurization at someis uniformly mixe Their present modeling assumes the H 2 to a rapid burn.
It is a com-Experiments are underway to understand how a core melts.
d t on the plicated process due to the formation of eutectics that are depen enWhen the 2 and H O and heat up rate.
less than 0.5 C/ SCC the Zr0 and 002 melts and if greater, the Zr clad m enemistry of Zr, U, 0 2
This forms The melt seems to run down between the pins and refreeze.The steel and steam Below the plus is clear.The slump and refreeze will be fi rs t.
a plug that moves downward.
were noted to form a messy foamy material.
h a prime factor in determining the rate of delivery of molte t
the core has melted down to the core support structure and allow a large lower head.
It was noted by SANDIA of molten material to rapidly enter the lower head.that the re bing.
The UO -Zr0 raixture is less dense than molten steel.
head will therefore most likely fail in t he manner described by Mayinger -
2 failure at or near the upper surface of the molten UO -Zr0 pool 2
The attack of the concrete during the early stages results in a The molten steel will be below the UO -Zr0 and deal of gas being generated.
2 tant.
It is not clear that the answer to this question will freeze.
penetrate concrete.come from any of the research programs as they are now The German containment has no coolers and is dry until the molten core The lack p'enetrates the shield wall and allows the sump water d containment.
A series of codes for class IX accidents have been put together in It is composed of package called KESS.
5-a sophisticated core melt code NELSIM:
a simple single pin slumping model KAUHZ:
a detailed evaporation model for heat removed from core fragments LUECKE:
RAUHZ:
heat up of melt in dry part of RPV KAVERN:
concrete penetration code C0C0:
containment codes A workshop There are additional codes that complete the KESS code system.
for its use will be scheduled for late summer at KfK.
9 D*
9 s:.
b k