ML20128M173

Summary of ACRS Subcommittee on Class 9 Accidents 850327 Meeting in Washington,Dc Re Review of NRC Current Suite of Severe Accident Codes,Including Containment Codes & Melcor

ML20128M173
Person / Time
Issue date:	04/03/1985
From:	Advisory Committee on Reactor Safeguards
To:	Advisory Committee on Reactor Safeguards
References
ACRS-2297, NUDOCS 8507110594
Download: ML20128M173 (8)

Text

i ACRS- Qa99 PDR 66ctsd l g.

IM5 DATE ISSUED: 4/3/85 hlog5 ACRS MEETING

SUMMARY

/ MINUTES FOR THE CLASS 9 ACCIDENTS SUBCOMMITTEE MEETING MARCH 27, 1985 - WASHINGTON, DC

Purpose:

The Subccrtr.ittee met with the NRC on March 27, 1985 to continue its review of the Staff's current suite of severe accident codes. The codes discussed were the containment codes and MELCOR.

Principle Attendees:

ACRS NRC W. Kerr, Chairman R. Curtis P. Shewmon, Member T. Lee I. Catton, Consultant J. Han M. Corradini, Consultant M. Cunningham M. Bender, Consultant P. Davis, Consultants A. E. Bergeron, SNL A. Wang, Staff, DF0 P. Powers, SNL C. Wong, SNL J. Travis, SNL J. Thurgood, NAI J. McGlaun, SNL J. Sprung, SNL D. Alpert, SNL S. Haskins, SNL g71g4850403 2297 py

Class 9 Accidents Minutes March 27, 1985

Introduction:

W. Kerr opened the meeting by stating his objectives for this meeting were to:

1. determine the purpose of the codes,

2. determine who are the potential users,

3. what is the stage of development of the various codes and their projected completion dates,

4. is the code to be benchmarked or validated and what is the difference between these two, and

5. is the use of the code for plant specific analysis or generic analysis?

R. Curtis stated he hoped most of these questions would be answered. He noted that approximately half of the 18 issues identified as areas of disagreement between IDCOR and the NRC will be addressed by the containment codes. W. Kerr asked how are the 18 issues to be resolved with IDCOR. By what process will the Staff determine whose modelling and analysis is c.orrect. M. Bender asked if there are standard sequences for comparison of results. P. Shewmon stated he is interested in knowing how this information will be used in the regulatory arena.

R. Curtis stated the NRC and IDCOR are having a series of meetings to delineate an approach to resolve the differences to a point to where at least they feel a decision can be made. He stated there currently is not a standard set of sequences for benchmarking results. He noted NRR is preparing a companion document to NUREG-0956 which would provide the Coninission with a list of possible regulatory initiatives related to the source term reassessment. R. Curtis proceded to name the containment codes with a brief description of their purpose. M. Bender requested

Class 9 Accidents Minutes March 27, 1985 that RES supply a short written sumary of each code. W. Kerr asked if the Staff would explain what they meant by validation. M. Bender stated the Staff must use caution when validating results by experiments because he feels in many instances the experiments do not represent the conditions in a severely damaged core. M. Corradini stated he is concerned about the waste of resources to what appears to be overlap of the code work efforts.

K. Bergeron stated CONTAIN is a state-of-the-art code for predicting physical and radiological conditions inside the containment building during and following a severe accident. He stated the major areas of uncertainty are in the areas of scaling and synergistic phenomena which tend to plague all large codes. W. Kerr asked when will CONTAIN be validated. K. Bergeron stated he does not like to use " verification or validation" but prefers the word " testing". SNL has developed a standardized test program for CONTAIN. This program contains comparisons: (1) with hand calculations, (2) with other codes, and (3) with experiments. All results are docunented and archived. More than 120 tests cases have been documented to date. Every new version of the code must pass a " standard" subset of these test. K. Bergeron believes they are the first to develop a systematic approach for the " validation" of a code. W. Kerr inquired if K. Bergeron was aware if other NRC code developers may be taking this approach. K. Bergeron noted the NRC encourages this type of quality control but it is not an NRC standard practice. However, he believes several other codes have followed this procedure. K. Bergeron noted an important experiment used for comparison with CONTAIN w:s the HDR Blowdown Experiment in Germany. He was very satisfied witn their " blind" calculational results as compared with the experimer. cal results.

CORCON/VANESA D. Power stated CORCON describes the attack on the concrete by core debris and predicts concrete erosion, gas generation, gas composition

• Class 9 Accidents Minutes March 27, 1985 and debris temperature. VANESA describes the radionuclide release and aerosol generation during core debris / concrete interactions and predicts aerosol material density, composition and particle size. He stated VANESA will be integrated into CORCON and CORCON then will be integrated into CONTAIN. I. Catton asked how CORCON handled partial solidified debris. D. Powers stated he would supply a CORCON manual to the Subcommittee. M. Bender asked how meaningful is an experiment that last a few minutes for studying an event that may last hundreds of hours.

D. Power stated SNL believes the long-term interaction can be divided into distinct phases and experiments of finite duration can be designed to describe each phase. He stated the major thrust of their current work is the comparison against current data. They also have test planned which involve sustained uranium melts interacting with concrete and the extension of CORCON into the long-term where the material would solidify.

HECTR (Hydrogen Event Containment Transient Response)

C. Wong stated HECTR is a relatively fast-running, lumped- volume containment analysis program used to predict transient temperatures, pressures, gas composition and wall temperature in a multicompartment reactor containments for hypothetical accidents involving hydrogen combustion. W. Kerr asked how will you decide when your code is adequate? M. Bender asked what are the effects of scaling. He is concerned that if one does not understand the modelling or phenomena, how can one assure himself that one's assumptions are always conservative. I. Catton had a similar corraent regarding extrapolating results. W. Kerr c'ked does C. Wong believe HECTR can be used for licensing decisions? C. Wong stated he felt fairly confident about the use of the multiple compartment analysis for licensing decisions.

HMS (Hydrogen Mixing Studies)

J. Travis stated HMS is a state-of-the-art code to model phenomenology for hydrogen distribution and control for:

1

_ _ _ _ _ _ _ _ _ . _ _ _ _ ~ _ _ _ _ _ _ _ . _ _ _ . _ _ _ _ , ___ -

. Class 9 Accidents Minutes March 27, 1985 (1) complicated containment systems, (2) low-speed buoyancy flows, (3) diffusion dominated flows, (4) transition from deflagations to detonations, and (5) detonations.

He stated HMS is true multi-dimensional code using a finite difference method. M. Corradini stated their seems to be substantial overlap between the HMS and COBRA codes. W. Kerr asked if J. Travis could state what HMS will be used for. J. Travis stated HMS will be used to benchmark HECTR because experiments would be too expensive to run. It will also be used to analyze available experimental data.

COBRA-NC M. Thurgood stated COBRA-NC will provide multidimensional-detailed thermal-hydraulic analysis in the areas of steam / water blowdown, reactor cavity pressurization, hydrogen distribution, fission product and core debris distribution, core boil off and metal / water reaction, steam ,

distribution with ice condenser containments and steam distribution within BWR suppression pools. It is a three-dimensional finite difference code but can be run in the lump parameter mode. P. Davis asked how is COBRA-NC different from CONTEMPT and what models are used for clad ballooning. M. Thurgood stated this code is a converted primary system analysis code (approximated a i man year effort) and accounts for the effects of drop carryover. He stated their is no accounting for blockage or ballooning by COBRA-NC. M. Bender stressed the Staff's needs to demonstrate the validity of the assumptions in the codes and how they relate to each other, as the codes will ultimately be used in regulatory application.

MELCOR M. Cunningham statea MELCOR is a second generation risk code. It is not a best estimate code because the primary concern was speed of the code

f. Class 9 Accidents Minutes March 27, 1985 and models were simplified wherever possible to achieve this. MELCOR is a highly modular, integrated code and will eventually replace the BMI-2104 suite of codes. One of the advantages of an integrated code will be its capability to do uncertainty analysis throughout the accident progression. He noted several detailed mechanistic codes, such as MELPROG, will be used to benchmark MELCOR. M. Corradini asked has the Staff considered bypassing the benchmarking by detailed mechanistic codes and perhaps going directly to simple and integral experiments for validation of MELCOR. W. Kerr stated the severe accident policy implies that mechanistic codes will be the primary tools for decision making with risk codes used as a back-up. For example, if one has a difficult decision MELPROG would be used rather than MELCOR. Since all severe accident decisions are difficult, when will the Staff ever rely on the results of a risk code. What will be the application for MELCOR?

P. Shewmon asked will codes such as MARCH, MATADOR and CRAC be replaced by MELCOR. M. Bender questioned whether the Staff is actually performing uncertainty analyses as it seemed to be more of a sensitivity analysis. The MELCOR presentation was divided into six categories: ,

a. Code Development,

b. Thermal Hydraulics,

c. Fission Product Behavior,

d. Consequences,

e. Sensitivity, and

f. Validation and Verification.

(a) Code Development J. McGlaun stated the code developers have tried to produce a code with the following attributes:

1. Written in ANSI FORTRAN 77, l 2. Easy to Modify and Distribute,

3. Sensitivity Analysis,

4. Modular and Structured,

f. Class 9 Accidents Minutes March 27, 1985

5. User Friendly,

6. Appropriate and Integrated Physics,

7. Appropriate Runtime, and

8. Well Documented.

W. Kerr asked how does SNL decide what the appropriate physics is? A group of experts in the field were asked to coment on a phenomenon assessment performed by SNL. Based on these comments and other existing papers models.were identified and developed for MELCOR.

(b) Thermal-Hydraulics J. McGlaun stated approximately 60% of the thermal hydraulic models are completed. He roted their models lagged the first principle codes.

(c) Fissioh Product Behavior J. Sprung noted the processes modelled in MELCOR with a brief description of each modelling effort.

(d) Consequences D. Alpert noted they have produced a new code to replace CRAC-II. They have made improvertents in the areas of time dependent releases, atmospheric dispersion modelling and mitigating action modelling.

(e) Sensitivity D. Alpert stated by uncertainty, they are referring to the variation or imprecision in the output as a result of variations or uncertainties in the input. For sensitivity, they are referring to the contribution of an individual variable to that uncertainty. Sensitivity analysis will identify the most important variables. D. Alpert stated uncertainty does not include the uncertainty in modelling just the uncertainty in' the variables that go into the model. W. Kerr stated if one does not know the accuracy of the model, how can one put any co1fidence in the

f.

Class 9 Accidents Minutes March 27, 1985 results. The most important factor in the accuracy of the codes is in the models used.

Validation and Verification S. Haskins stated there are three basic types of code assessments: (1) conceptual assessments, (2) phenomenological assessments and (3) review of documentati)n and coding. He noted they have patterned their validation and verification program after CONTAINs prcgram.

The meeting was adjourned at 5:30 p.m.

NOTE: Additional meeting details can be obtained from a transcript ,

of this meeting available in the NRC Public Document Room, 1717 H Street, N.W., Washington, D.C., or can be purchased from ACE-Federal Reporters, 444 North Capitol Street, Washington, D.C. 20001,(202)347-3700.

1 i

{

. - -__-__