ML20010D613
| ML20010D613 | |
| Person / Time | |
|---|---|
| Issue date: | 07/23/1981 |
| From: | Advisory Committee on Reactor Safeguards |
| To: | Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-1861, NUDOCS 8108280372 | |
| Download: ML20010D613 (8) | |
Text
.-
gQ kMS- /86 l W \\\\,
DATE ISSUED:
'e g$ f ~\\l- )
J{'Sb }VN j
7hk b I W
[C ' '\\ \\[h,h MINUTES J
\\
CLASS 9 ACCIDENTS SUBCOMMITTEE MEETING j
k 1
May 21-22,1981
),1 )
Wcshington, D. C.
O (s
PURPOSE To discuss the MARCH code and other codes used to model severe accident phenomena.
PRINCIPAL ATTENDEES:
W. Kerr, Chairman R. Curtis, NRC/RES P. Shewmon, Member P. Cybulskis, Battelle-Columbus C. Mark, Member R. Bari, BNL C. P. Siess, Member K. Pratt, BNL H. Etherington, Member J. Rivard, Sandia S. Siegel, Consultant M. Liparulo, W M. Sichel, Consultant R. Ritzman, SAI Z. Zudans, Consultant G. Thomas, NSAC P. Davis, Consultant P. Baybutt, Battelle-Columbus J. Heyer, NRC/NRR R. Denning, Battelle-Columbus No oral or written comments were received from members of the public.
Attached is the meeting agenda and a list of documents considered by the Subcommittee.
USE OF MARCH BY NRR J. Meyer (NRR) discussed the use of MARCl! for licensing purposes.
He indicated that decisions are not being based solely on MARCH. One information need NRR has is the failure characteristics of containment.
Currently, "best estimate" and " conservative estimates" of containment failure are used.
Near-Term Construction Permit plants have to be analyzed for hydrogen that would result from 100% clad / water reaction.
Dr. Mark inquired how the hydrogen mixing behavior in containment is presumed to be known.
$ $h 72 810723 1861 PDR r
Class 9 Accidents Mtg. May 21-22,1981 MARCH,does not address hydrogen mixing, rather, the code assumes a uniform concentration of hyd im throughout the containment volume.
Dr. Mark noted that the assumptions of 75% clad / water reaction for OLs and 100% clad / water reaction for cps are arbitrary.
Dr. Siegel inquired how well the output of MARCH is characterf red wit!.
respect to uncertainty. The reply was the code makes no effort to estimate uncertainty.
Dr. Kerr noted that it should be determined how well answers need to be known, that is, there should be uncertainty objectives.
As an example of how NRR is using MARCH, for McGuire, the dominant accident sequences were selected from WASH-1400. MARCH was then used to calculate the time dependent production of hydrogen and the response of containments.
Dr. Zudans asked how MARCH modeled the core meltdown behavior in the vessel. The reply was that there are three simplistic models.
The user has an option to select a particular model.
DEVELOPMENT OF MARCH BY RES R. Curtis (RES) discussed the MARCH code development. MARCH was written inthetimeperiod6/76to1/77byBattelle-Columbus (BCL). At that time the code was limited to large break accidents. Minimal verifi-cation of the coce against experimental data has been performed.
RES is not funding any work at BCL for code improvement. The INTER code is used in MARCH to treat core-concrete interactions. RES would 11ke to replace INTEP in PARCH with CORCON. MARCH does not model thr. core
~
,,.y
,,,.,.-,,r,,
,+c-.,--
-.y
,w
Class 9 Accidents Mtg. May 21-22,1981 support plate.
The code treats the reactor coolant system as a single vol ume. A simplistic model of vessel lower head failure is used. The core 'is assumed to drop rapidly into the cavity. At that poin,t there are two user options for core behavior: very rapid quench, and no quench.
DESCRIPTION OF MARCH BY BCL P. Cybulskis (BCL) discussed some of the models in MARCH. The Code does not calculate steam explosions.
The user chooses whether a steam explosion will occur and inputs the energy term. The release of volatile fission products from fuel during the meltdown process is
.O accounted for in the decay heat term.
Coolant flow out a break is always single phase, water or eteam.
Depending on user options, clad reaction during the in-vessel meltdown phase varies between 20% and 100%.
In response to a question by Dr. Siegel, Mr. Cybulskis indicated that no information is provided as to the uncertainties in code output.
Mr. Etherington indicated that the Code does not treat viscous forced convection heat transfer and that the Code misuses classical heat transfer equations.
When the core has dropped into the reactor cavity, the user selects whether the debris bed dries out or does not dry out. The Code includes the presence of rebar in the basemat. The user inputs the start and stop hydrogen concentrations in containment for hydrogen burning, e.g., commence hydrogen burning when concentration reaches 10% and stop when concentration reaches 4%. The containment failure pressure and hole size is input by the user.
e
Class 9 Accidents Mtg. May 21-22,1981 BROOKHAVEN NATIONAL LAB (BNL) USE OF MARCH R. Bari and K. Pratt (BNL) presented observations on the use of MARCH.
BNL acquired MARCH in 10/79. They replaced the Hot Drop model in MARCH, where the core is instantaneously quenched when it drops into the cavity, with a Porous Bed inodel.
BNL is dissatisfied with the use of INTER as part of MARCH. Use of INTER is nonconservative with respect to base mat attack during initial period following core drop into the reactor cavity because INTER does not allow base mat attack to begin 0
until the core debris reaches a tempar'ture of 2500 F.
MARCH models the distribution of hydrogen in the reactor coolant system as uniform at all times.
Dr. Kerr inquired how BNL made decisions as to which assumptions in MARCH are reasonable and which are not.
R. Bari gave his opinion that the phenomena of greatest uncertainty include: hydrogen mixing and combustion; core melt processes; core-concrete interaction; and ccre debris / water interaction. MARCH does not have predictive capability for core melt behavior.
Dr. Siegel noted that the user has a responsibility to scrutinize the code.
Mr. Bari indicated that the user must be very familiar with MARCH to use it responsibly.
Dr. Mark inquired as to what results and con-clusions have been provided by PARCH that would not have been known if PARCH did not exist.
No examples could be provided.
PARCH EVALUATION PROGRAM J. Rivard (Sandia) summarized a code evaluation program that is being performed by MARCH users. The program started in 1/81. The final report is due in June 1981. 100 limitations have been identified.
Class 9 Accidents Mtg. May 21-22,1981 Dr. Zudans asked whether anything was found in PARCH that was not considered a limitation.
Dr. Kerr noted that for something to be considered a limitation, there must be some goal that was not being met. Mr. Rivard could not identify any goals.
Dr. Curtis indic,ated that MARCH has been pushed by,various users well beyond its capabili-ties.
WESTINGHOUSE EVALUATION OF MARCH N. Liparulo (W) presented some observations by Westinghouse on MARCH.
MARCH has been cited as demonstrating that putting water on an over-heated core is bad. Mr. Liparulo strongly disagrees with such con-clusions.
He indicated that it would take a user at least six months to gain sufficient familisrity with the code to use it responsibly.
Another require, ment for responsible use of the code is to employ people that have a sound understanding of physical phenomena associated with severe accidents. Mr. Liparulo strongly disagrees with MARCH treat-ment of vessel lower head failure as a sudden catastrophic event, rather, failure would occur through instrument tubes.
EPRI EVALUATION OF MARCH Dr. Ritzman (SAI) and G. Thomas (NSAC) summarized their observations of PARCH.
An example of the importance of user options in PARCH is that by using various options, failure of the Limerick containment is predicted to occur after two hours or after six hours, for a particular accident sequence. Mr. Thomas noted that he has observed various users, including the NRC, misuse the MARCl! code.
Class 9 Accidents Mtg. May 21-22,1981 REVISJON OF CORRAL P. Baybutt (BCL) summarized the rewrite of CORRAL that is underway.
The work is sponsored by the Division of Systems and Reliability.
~
CORRAL models radionuclides in containment.
The current version (CORRAL-2) was prepared for WASH-1400.
Gravitational scttling is the only mode of aerosol deposition that is currently considered. The aerosols are assumed to consist of a single particle size. The code is not benchmarked against experimental data.
Molecular iodine is assumed.
The source term is a user input to CORRAL.
Containment surfaces are assumed to be dry.
There is no removal of radionuclides from flow through cracks in the containment wall.
CORRAL receives input from MARCH in terms of containment temperature and pressure, failure of the reactor vessel, and failure of containment.
The new version (CORRAL-3) will be ready by Octcber 1,1981.
No effort would be made to validate CORRAL-3 prior to publishing it.
That would be left up to the user.
Dr. Kerr inquired as to how anyone can conceive of publishing a code without making any effort to validate it. Mr. Baybutt did not view this as a problem.
OTHER CODES IN USE FOR SEVERE ACCIDENTS j
R. Curtis discused other codes in use or under development.
SCDAP is being developed by INEL and will be issued by June 1982.
It models fuel behavior inside the vessel during the t
l initial stages of meltdown.
!N-
Class 9 Accidents Mtg. May 21-22,1981
.. TPAP is being developed by BCL and will be issued in 1982.
It is a source term code to provide input to CORRAL.
CONTAIN is being modified by Sandia.
It was developed for LMFBRs.
It treats containment phenomena in more detail than MARCH.
Dr. Curtis indicated that RES will provide the necessary organization and money to upgrade MARCH. Dr. Kerr inquired as to how the code development work is being coordinated with the upcoming rulemaking activities.
The answer was that it isn't.
DISCUSSION OF KESS R. Denning and P. Cybulskis discussed the German code KESS, which is a similar type of code as MARCH.
KESS models have two options avail-able to the user: a simple model and a more complex model.
KESS is currently limited to treatment of large break LOCAs. The code does not allow water to be present in the reactor cavity.
FUTURE MEETINGS The Class 9 Accidents subcommittee will meet in Washington, D. C. on June 24,1981 to discuss the Class 9 accident research budget, and on
(
June 30,1981 in Albuquerque, NM to discuss filtered-vented contain-i ment systems.
I cc:
M. W. Libarkin G. R. Quittschreiber l
l l
l GA
LIST OF DOCUMENTS Licensing Needs for Sistems Code for Severe Accidents 5 Slides I
(
Description of MARCH Code by BCL 41 Slides BNL Activities with the MARCH Code 99 Slides MARCH Assessment Program 18 Slides Westinghouse Evaluation of MARCH 11 Slide:
EPRI/SAI Euluation of MARCH 21 Slides Revfsion of CORRAL 7 Slides CORCON Development 7 Slides CONTAIN Development 10 Slides KESS Description 38 Slides Letter from G. Thomas (WSAC) to W. Kerr on observation of use of MARCH Code Analytical Results.
I i
l l
[
l l
l
-