ML20210U894
| ML20210U894 | |
| Person / Time | |
|---|---|
| Issue date: | 09/22/1986 |
| From: | Speis T Office of Nuclear Reactor Regulation |
| To: | Buhl A INTERNATIONAL TECHNOLOGY CORP. |
| References | |
| NUDOCS 8610100286 | |
| Download: ML20210U894 (34) | |
Text
p y
0 i
SEp g, Dr. Anthony Buhl International Technology 575 Oak Ridge Turnpike Oak Ridge, TN 37830
Dear Tony:
During the NRC/IDCOR technical exchange meetings concerning integrated containment analyses, it became apparent that several major modelling differences between NRC and IDCOR play a significant role in the plant analyses.
Key areas of disagreement were subsequently identified and consolidated into the 18 NRC/IDCOR Technical Issues.
It was recognized that further study of these issues was necessary before proceeding with the plant reviews.
Over the past several months draft NRR position papers were prepared for each of '
the NRC/IDCOR Technical Issues by DSR0 staff in coordination with our Office of Research. These papers summarize our assessment of the status of the issue and how remaining differences between IDCOR and NRC will be accounted for in the plant reviews and the individual plant evaluation methodology. The draft papers were circulated to our contractors and to IDCOR for comment.
Comments received were taken into consideration in the preparation of the final version of each paper. Position papers are now available in final form for the following issues and are enclosed:
Issue 1
- Fission Product Release Prior to Vessel Failure Issue 2
- Recirculation of Coolant in Reactor Vessel Issue 3
- Release Model for Control Rod Materials I:: sue 5
- In-Vessel Hydrogen Generation Issue 6
- Core Melt Progression and Vessel Failure Issae 11 - Revaporization of Fission Products in the Upper Plenum Issue 13A - Amount and Timing of Suppression Pool Bypass Issue 17 - Hydrogen Ignition and Burning
(^ /
\\
, 30 s
d'y' 8610100286 860922 PDR ORG NRRC PDR
n r
o l
Dr. Anthony Buhl 2
Position papers for the remaining issues will be forwarded to you as they become available. This will complete our resolution of the IDCOR/NRC Technical Issues.
Sincerel : # 3Ded $
BdBI W. Shf05 Themis P. Speis, Director F
Division of Safety Review and Oversight Office of Nuclear Reactor Regulation
Enclosures:
As Stated cc:
Z. Rosztoczy B. Morris G. Arlotto M. Silberberg DISTRIBUTION tentral FiTe DSR0 r/f RIB r/f TSpeis BSheron FCoffman RPalla PDR
- see previous concurrance OFC : RIB
- RIB
- RIB
- DSRO
- JSP
- NAME
- *RPalla/rr :*FCoffman
- ZRosztoczy :*BSheron DATE :9/ /86
- 9/ /86
- 9/ /86
- 9/ /86
- 9/
0FFICIAL RECORD COPY
ISSUE 1 - FISSION PRODUCT RELEASE PRIOR TO VESSEL FAIL,URF, 1.0 ISSUE DEFINITION Differences exist between IDCOR and NRC in the modelling of fission product release from the primary system. These differences were grouped, during the IDCOR/NRC technical exchange meetings (Reference 1), into three related issues namely, Issue 1 - Fission Product Release Prior to Vessel Failure, Issue 4 -
Fission Product and Aerosol Deposition in the Primary System, and Issue 11 -
Revaporization of Fission Products in the Upper Plenum. The major differences identified as pertinent to the subject issue concern (1) the temperature of the core material at which relocation is assumed to begin, (2) the modelling of the initial release of fission products from the fuel, and (3) the modelling of tellurium (Te) retention in-vessel. Uncertainties in these three areas do not significantly affec't the timing of containment failure, but can affect both the timing and the total quantity of fission product release for all plant types and all sequences.
2.0 APPROACH TO RESOLUTION IDCOR and NRC agreed to a number of code modifications (Reference 2) which, when correctly implemented in MAAP, would result in treatment of the aforementioned phenomena in a manner consistent with that in the Source Term Code Package (STCP). These were:
1.
The addition of a new fuel relocation slump model and user input "eutectic" temperature, in place of the MAAP model which assumed relocation only after core material reached the melting point cf UO '
2 2.
The addition of a model to treat Te-retention by unoxidized zircaloy, in place of the MAAP model which assumed the Te to be released from the fuel and deposited in the reactor vessel prior to vessel failure.
It was agreed that the IDCOR model for in-vessel release of fission products l
from fuel (FPRAT) be retained without modification on the basis of additional l
information which shows very good agreement with the empirically-based release l
model provided in NUREG-0772 and used in the STCP.
IDCOR has implemented the above modifications, and documented the results of model comparisons in technical report T85.2 (Reference 3).
3.0 STAFF ASSESSMENT To address the matter of temperature requirements for fuel relocation, IDCOR has developed a melt progression model (See Issue 6) and integrated it into MAAP.
In the model, the constituents of the core are assumed to form a eutectic which melts at a user-specified temperature with a user-specified l
J
2 assuming a eutectic melt temperature (2500K) ported in T85.2 for calculat latent heat. A comparison of MAAP results reand a melt temperature corresponding to the melting point of UO2 (3100K) indicates that the melt temperature has a IDCOR relatively small impact on the total in-vessel fission product relefse.
has stated that a melt temperature corresponding to the eutectic temperature will be used in all future MAAP calculations (Reference 4).
To address the matter of Te-retention by unoxidized Zircaloy, IDCOR has added a user option to MAAP which binds Te with Zircaloy in-vessel, and releases the material ex-vessel during core-concrete interactions. Sensitivity studies presented by IDCOR in T85.2 indicate that ex-vessel release of the Te results in a substantially greater early release of Te, i.e., within 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> following vessel breach. With the original IDCOR model, significant releases of Te are also predicted (as a result of revaporization of the fission products initially deposited in the primary system) but at much later times.
IDCOR has stated that future MAAP calculations will be based on the assumption that Te is released ex-vessel (Reference 4).
4.0 CONCLUSION
S The staff has considered the MAAP code modifications implemented by IDCOR and concludes that use of the revised models satisfactorily resolves the issue of fission product release prior to vessel failure within the defined scope.
The resolution of related issues regarding Fission Product Deposition in the Primary System and Revaporization of Fission Products is described in Issue Papers 4 and 11, respectively. The staff notes that future calculations, as well as the risk uncertainty analysis performed by IDCOR, should be based on the revised models described above rather than the original IDCOR models and plant analyses.
It should also be noted that subsequent to the IDCOR/NRC technical exchange l
meetings, results of additional experiments have suggested that iodine may be released in the form of the more volatile iodine molecule, rather than as the less volatile salt (C I) as assumed by both IDCOR and NRC, Additional tests and analyses are curreiltly underway at Sandia and Oak Ridge National Laboratories to further address this matter. Resolution of Issue 1 is contingent upon this research confirming our present understanding of the chemical form of iodine. Pending completion of the work, the issue regarding the chemical form of iodine will be addressed as part of the uncertainty analysis performed for NUREG-1150.
5.0 REFERENCES
l 1.
"NRC/IDCOR Meeting on Outstanding Technical Issues for Severe Accidents,"
Memorandum from T. P. Speis to Distribution, dated March 7, 1985.
l 2.
" Minutes of the NRC/IDCOR meeting on March 26, 1985," Memorandum from Z. Rosztoczy to T. Speis, dated May 9, 1985.
3.
Fauske and Associates, Inc., " Technical Support for Issue Resolution,"
IDCOR Technical Report 85.2, July 1985.
4.
M. Plys, Private Comunication.
1
ISSUE 2 - REACTOR COOLANT SYSTEM NATURAL CIRCULATION 1.
ISSUE DEFINITION It is recognized that effects of multi-dimensional natural circulation in the reactor vessel and the reactor coolant system (RCS) may influence the Flow recirculation in the vessel induced by the accident progression.
buoyancy effect causes some of the upper plenum fluid to flow back into the core. The buoyancy forces can be of such a magnitude (if the primary system is at high pressure) that counter-current flow may be set up in the hot legs leading to the steam generators.
It is possible that the hot gases of the core may reach the steam generator tubes and cause consider-able heating of the tube walls (Ref. 1). Also, the reactor vessel upper plenum structures and connected RCS piping and components may heat up to temperatures sufficiently high to challenge the structural integrity of the RCS during high pressure sequences such as station blackout (TMLB')
sequences (Ref._2). Rupture of the RCS prior to core debris melting through the reactor vessel bottom head could depressurize the primary This would in turn system and thus prevent high pressure melt ejection.
preclude the adverse effects of direct containment heating (Issue 8).
The natural circulation issue also affects the issues of in-vessel fission product deposition and revaporization (Issues 4 and 11, respectively), and Retention and revaporization of fission hydrogen generation (Issue 5).
products from the RCS surfaces will depend on the natural circulation flow as it tends to raise the temperature of the upper plenum structures and l
the connected RCS piping. The effects on fission product revaporization are addressed as part of Issue 11.
i Natural circulation flow in the reactor vessel could also potentially increase the magnitude of hydrogen generation by recirculating the steam from the upper plenum back to the core to react with the fuel cladding.
The As a result, the upper plenum could reach even higher temperatures.
effect would promote further structure oxidation producing an additional amount of hydrogen. The effect of natural circulation on hydrogen generation are addressed in inore detail as part of Issue 5.
Based on the existing analyses, it appears that natural circulation flow significantly affects high pressure sequences in PWRs (Ref. 3), but may not have significant impact on PWR low pressure accident sequences or on BWR sequences (Refs. 4,5).
Nevertheless, the staff believes that with only limited information presently available, it would be difficult to realistically assess the impact on BWR TMLB' sequences.
Further research efforts are needed to support the conclusions previously suggested.
This issue deals with the analytical models and assumptions used to predict the natural circulation flow in the reactor vessel and the possible effects on vessel failure.
l i
2 2.
APPROACH TO RESOLUTION During an NRC/IDCOR meeting held on March 26, 1985, the NRC staff proposed that the natural circulation issue be addressed and resolved by IDCOR basically agreed with the NRC sensitivity studies.
proposal, but indicated that they would perform calculations,wi.th consequential steam generator tube rupture only if the analysis indicated potential tube failure (Ref. 5). To justify this position, IDCOR-developed and implemented a simplified upper plenum-core, natural circulation model in MAAP code with supporting models for clad balloning,A core barrel / core baffle heat sink, and radial radiation heat transfer.
description of the IDCOR analytical models and the results of the analysis are presented in IDCOR Program Report, Technical Report 85.2 (Ref. 3).
3.
STAFF ASSESSEMENT The IDCOR natural circulation model in the MAAP computer code assumed two For Westinghouse type reactor vessel geometries, different flow patterns.
the flow pattern was assumed to consist of one large loop coupling the core to the. upper plenum. The return leg from the upper plenum was assum-ed to flow down the outer cooler flow channels and occupied half the total core flow area.
In B & W reactors, the flow pattern was assumed to con-sist of a significant flow area through the core baffle. There was a tendency for the return flow to pass down the core barrel-baffle annulus and through the baffle into the core (Ref. 3). IDCOR fndicated that the modified MAAP models provide a good prediction of mass flow due to natural circulation when compared to very limited experimental results obtained from the Westinghouse 1/7th scale test model using low pressure water and SF gas for the experiments. Since there are only limited experimental data available to compare with MAAP's predicted flow rate at present, the g
staff believes that it would be prudent if the predicted mass flow rate could be confirmed by other analytical models, e.g. COMMIX, under similar initial and boundary conditions.
By incorporating an additional supporting model for core melt progression into MAAP, IDCOR has also performed simulations of the Zion TMLB'/ seal The accident sequences were analyzed with a,HF/ drains blncked accidents.nd without upper pl LOCA and Sequoyah S The calculations indicated a slight decrease in hydrogen production flow.
and fission product retention for the sequence with natural circulation Although a higher upper plenum temperature was predicted, IDCOR flow.
concluded that the higher temperature (approximately 1350 F for Zion and 1100 F for Sequoyah) would not be sufficient to fail the reactor coolant However, an INEL study on system prior to reactor vessel head failure.
structural integrity (Ref.9) suggested that at a temperature of 1350 F the j
vessel hot leg nozzle would fail in a few minutes under high-pressure TMLB' conditions.
In contrast to IDCOR's calculated temperature, EPRI predicted a much higher temperature, approximately 2060 F, for the upper plenum (Ref. 6).
Calculations j
The core exit gas temperature ultimately reaches 3500 F.
performed by LANL and SNL using the two-dimensional version of the MELPROG code for a similar analysis support EPRI's results (Ref.7 and 8).
.n 3
^
4.
CONCLUSIONS The staff believes that existing analytical models for natural circulation flow phenomenon need to be improved to realistically predict the thermal hydraulic coupling with core-melt progression, hydrogen generation, and fission product release and deposition.
Based on available information it is the staff judgement that the uncer-tainties in the prediction of temperatures within the RCS are sufficiently large that RCS depressurization via temperature-induced failure in the Accordingly, primary piping system can be neither assured nor ruled out.IDCOR shou sure sequences in PWRs and in the development of the individual plant examination methodology.
Available analyses also suggest that natural circulation may strongly influence the issue of fission product revaporization by altering the spatial distribution of fission product initially deposited in the RCS, as Uncertainties in well as the temperature distribution within the RCS.
predicted RCS temperatures are large, for example, estimates of upper Since the effect plenum structure temperatures range from 1300 to 2100*F.
of temperature differences on this order, e.g. 800 F, would substantially affect fission product deposition and revaporization, IDCOR should address this matter in their sensitivity study.
5.
REFERENCES V. E. Denny and B. R. Sehgal, " Analytical Prediction of 1.
Core Heat-Up/Liquifaction/ Slumping," Proc. Int. Mtg., LWR Severe Accident Evaluation, 1983.
B. R. Sehgal, V. E. Denny, W. A. Stewart and C. C. J. Chen, "Effect 2.
of National Convection Flow on PWR System Temperature during Severe Accidents," National Heat Transfer Conference, 1985 Fauske and Associates, Inc., " Technical Report 85.2, Technical 3.
Support for Issue Resolution," July 1985.
D. F. Ross, "NRC/IDCOR Technical Issues," March 5, 1985.
4.
Zoltan Rosztoczy, " Minutes of the NRC/IDCOR meeting on March 26, 5.
1985," May 9, 1985.
B. R. Sehgal, A. T. Wassel, M. S. Hoseyni and J. L. Farr, Jr.,
6.
"Thennal-Hydraulics of the Primary Coolant System of Light Water Reactions during severely Degraded Core Accidents," International Conference on Reactor Thermal Hydraulic, October 1985.
R. J. Henninger, J. E. Kelly, and J. F.
Dearing,
" Preliminary 2-D 7.
MELPROG Calculation for the TMLB' Accident in Surry, "Los Alamos and Sandia National Laboratory, March 17, 1986.
8.
J. T. Han, "A Sumary of March 17, 1986 meeting on RCS Natural Circulation Studies," May 1986.
Y. N. Shah, " Structural Failure Studies of RCS," presented at Direct 9
Containment Meeting on April 23, 1986, Bethesda, Maryland.
__}
~
ISSUE # 3 - RELEASE MODEL FOR CONTROL R0D MATERIALS 1.0 ISSUE DEFINITION In PWR plants with silver-indium-cadmium control rods, this issue relates t
to the fraction of control rod material that would be released as aerosol in the primary system during a core melt accident. The NRC sponsored BMI-2104 suite of codes has calculated large aerosol releases, resulting in an intense concentration of inert aerosols in the reactor vessel during core melt. The BMI-2104 codes predicted significant deposition of volatile fission products on the inert aerosol, and consequently, enhanced retention of fission products in the primary system. By contrast, the IDCOR model allows control rod material to rapidly melt and run off to cooler regions of the core, thereby removing it as a potential source of inert aerosol.
For BWR plants, the issue relates to possible chemical reactions of B,C control material that could increase hydrogen production and alter the chemical form of fission product species.
IDCOR does not model the chemical reactions of B C based on their assertion that blockage and 4
channel box confinement will divert steam from contact with the B C, 4
2.0 APPROACH TO RESOLUTION This issue was discussed at a March 26, 1985 NRC/IDCOR technical exchange NRC and IDCOR agreed that experiments planned for the ensuing meeting.
six month period would probably provide a basis for resolving this issue for BWR and PWR plants. The NRC also agreed to provide IDCOR with the results of SASA calculations to be performed by the Oak Ridge National Laboratory (ORNL) regarding the effect of B C-steam reactions during BWR 4
dCCident sequences 3.0 STAFF ASSESSMENT The ORNL staff (reference 1) has conducted an assessment of their experi-l l
mental data regarding the behavior of silver-indium-cadmium control rods, and concluded that the BMI-2104 models overestimate the fraction of mate-l rial released as aerosol. Subsequent in-pile tests in PBF (SFD-1) and ACRR (DF-3) under more prototypical accident sequence conditions support control rod melting and relocation models leading to low aerosol product-ion. The reference memorandum described several alternative code modifi-(
cations, ranging from rigorous model changes to purely empirical adjustments, which would remedy this situation. The option which was selected and incorporated in the Source Term Code Package (STCP) is to reduce the silver and indium releases to 10% of their previous values l
and to reduce the cadmium release to 70% of the previous value.
In subsequent calculations the aerosol concentrations from control rod materials have been predicted to be similar to those from other sources l
l w---
2 (structural materials, fission products). This result puts the STCP calculations in qualitative agreement with the results predicited by the IDCOR methodology regarding control rod aerosol releases.
Calculations performed by ORNL for the SASA program have examined the be-havior of B C control rod material in hypothetical core melt accidents at theBrownsherryNuclearPlant. The predicted affect of B,C chemical reactions on the timing and progression of the accident were minimal. The inclusion of B C reactions resulted in a marginal (10%) increase in hydrogen produ,ction compared with the case with no B C reactions. There 3
are two postulated mechanisms whereby B,C could affect the chemical form of fission products. One mechanism inv51ves the possible formation of borates of cesium. The borates would precipitate rapidly, leaving less cesium available to convert gaseous iodine to the more benign form of C I.
8 The other effect is the potential for formation of small concentrations of methane. Methane, even in very dilute quantities, can react to form methyl iodide, a highl, volatile compound which would enhance the release of radioactive iodine.
The ORNL SA'SA ' calculations for a station blackout sequence showed minimal effect of borate fomation. This result cannot be extrapolated to other sequences.
The methyl iodide formation is not yet modelled by ORNL. Methane produc-tion is estimated to be quite small, but bench experiments of ORNL in-dicate that small methane concentrations in a suppression pool in the presence of high radiation fields can result in significant production of methyl iodide. The DF4 experiment scheduled this Summer in ACRR was originally designed to examine the effect of BWR control rods on fuel behavior, but will provide information on the extent of B C oxidation. Oak g
Ridge is in the process of modelling methyl iodide production. When this work is complete, we will have a more quantitative basis for evaluating the effect of B C on fission product chemistry.
4
4.0 CONCLUSION
S The issue of vaporization of silver-indium-cadmium control material has been resolved by the aforementioned modifications to the NRC Source Term Code Package. The NRC and IDCOR models are in agreement, in so far as neither model predicts a significant impact of these aercsols on in-vessel fission product behavior.
For BWR plants, the NRC and IDCOR models are in agreement that the B C 4
l.
control rods have no significant effect on core melt progression or hydrogen production.
The impact of B,C on the chemical form of iodine is less certain. The NRC SARRP analyses for NUREG-1150 treat the chemical form of iodine as a major source of uncertainty, although the specific impact of B,C is not modelled explicitly. The IDCOR models do not currently consider Todine chemistry an important source of uncertainty.
Risk estimates for plant analyses as well as regulatory decisions should account for uncertainties in the chemical form of iodine, including the effect of B C.
4
3 With regard to the specific effect of B,C, there is insufficient evidence to conclude that it is not a major contributor to risk. Forthcoming ex-periments from the ACRR of Sandia National Labs and ongoing modelling work at ORNL should be completed in order to arrive at a firm conclusion.
5.0 REFERENCE Letter from M. Silberberg, RES, to B. Saffell, BCL, October 28, 1985.
1.
)
i l
l i
~
1 ISSUE 5 - IN-VESSEL HYDR 0 GEN GENERATION ISSUE 6 - CORE MELT PROGRESSION AND VESSEL FAILURE '
~
1.0 ISSUE DEFINITION Hydrogen generation in-vessel is primarily due to the chemical reaction of water with the Zircaloy cladding of fuel rods (and Zircaloy canisters in BWRs). Substantial agreement exists between IDCOR and NRC in the modelling of cladding oxidation and hydrogen generation prior to core geometry However, after cladding melts and slumps, significant changes.
differences exist in the modelling of mechanisms which affect hydrogen As uncertainties in hydrogen generation are intimately generation.
related to the modelling of core melt progression, the IDCOR issues regarding hydrogen generation (Issue 5) and core melt progression and Key vessel failure-(Issue 6) are addressed jointly in the present paper.
items which influence hydrogen generation and the state of the molten j
debris (mass, composition, and temperature) at vessel failure involve:
(1) cladding / fuel relocation and blockage formation, (2) multi-dimensional natural circulation flow and oxidation / failure of upper plenum structures, 4
and (3) melting and failure of below-core structures. Additional factors which can affect in-vessel hydrogen production include water additions from make-up systems, oxidation of B C control blades in BWRs, and oxidation of steel and UO inallLWks. The effect of B C is addressed as 2
4 part of Issue 3.
The concerns regarding in-vessel hydrogen generation center on the rate and quantity of hydrogen production, and the associated hydrogen-steam These mass and energy release rates from the reactor coolant system.
parameters strongly influence the flammability of the break flow and con-tainment atmosphere; the magnitude, timing, and location of potential hydrogen burns (particularly for ice condenser and Mark III containments);
and the magnitude of the containment pressure rise prior to and at reactor vessel failure (particularly for ice condenser and Mark I and II containments). Given the NRC estimates of in-vessel hydrogen production, l
failure of an ice condenser containment by hydrogen burn at vessel failure is possible for high pressure sequences. Failure of Mark I and Mark II i
containments at vessel failure (due to hydrogen partial pressure and sensible heat) is also possible for certain sequences.
The concerns regarding core melt progression and vessel failure center on the influence of associated phenomena on hydrogen production and the state of the debris at vessel failure. The state of the core debris can affect all aspects of the core-melt threat to containment integrity, particularly the issue of direct heating for high pressure sequences, and the release i
of refractory fission products due to core-concrete interactions.
This issue deals with the identification of those in-vessel models and assumptior.s which need to be considered in establishing estimates of hydrogen production and the state of the debris at vessel failure.
i
, ~
i..
2.0 APPROACH TO RESOLUTION IDCOR continues to believe that its in-vessel models are consistertt with available experimental evidence and TMI. To further justify this position, IDCOR has developed and implemented in the MAAP code new models for previously neglected phenomena which can affect hydrogen production, including core-upper plenum natural circulation (in PWRs), core inter-channel flows, clad ballooning (in PWRs), radial radiation heat transfer, and core barrel / core baffle heat transfer. A description of the new models and the results of sensitivity studies and comparisions between code calculations and data from Power Burst Facility-Severe Fuel Damage (PBF) experiments and TMI are presented in IDCOR report 85.2 (Reference 1).
The NRC and its contractors have also attempted to resolve differences in in-vessel modelling through further model development and additional com-parisons betwe'en codes and available data. This effort has included further development of the SCDAP and MELPROG codes, which employ more mechanistic in-vessel melt progression models, and comparison of MARCH 2 and the detailed mechanistic codes with data from the PBF tests and TMI.
3.0 STAFF ASSESSMENT 3.1 CLADDING / FUEL RELOCATION AND BLOCKAGE FORMATION IDCOR models assume complete flow blockage upon cladding / fuel relocation, and correspondingly lower hydrogen production due to steam starvation; NRC models do not assume blockage and thus predict enhanced hydrogen production as molten portions of the core slump into the lower plenum and supply steam to the intact portion of the core. Additional code modifications and analyses were performed by both IDCOR and NRC to further address this matter.
l l
3.1.1 COMPARISONS WITH PBF TESTS l
Using a stand-alone version of tne BWR/HEATUP module in the MAAP code which includes the previously mentioned code modifications, IDCOR has performed simulations of SFD tests 1-1, 1-3, and 1-4.
Comparisons of l
predicted and experimental cladding temperatures and hydrogen production rates show good agreement prior to cladding / fuel melting and relocation.
Beyond that point, similar trends but considerable differences in numeri-cal results are observed.
In spite of the differences, HEATUP predictions for total hydrogen production indicate generally good agreement with the experimental results when steaming rates input to the code are based on measured boildown rates.
The agreement between the HEATUP predictions and PBF test results may be partly due to the revised models in HEATUP, including a new mechanistic blockage criterion. The HEATUP code now assumes that when a computational node in the core reaches a user-specified eutectic temperature, oxidation in the node is terminated due to a combination of factors including reductions in flow area, decreases in surface-to-volume ratios, and increases in effective diffusion lengths due to addition of core materials from higher location
l 3
t In the PWR version of HEATUP, complete blockage of the node in the core.
is assumed with bypass flow into the nodes above and below the tlockage permitted; in the BWR version the fuel channel is assumed to remain open.
In both versions, molten material in the node at or subsequent to"that time is allowed to relocate to the node below, with its associated mass and internal energy. The downward progression continues until the molten material reaches a node which is either frozen or completely full. The BWR version of HEATUP includes a model by which molten cladding is maintained in a levitated position if the system pressurization due to steam flow through the bundle would exceed the hydrostatic pressure exerted by the cladding.
In this case, the molten cladding would be prevented from relocating and blocking the fuel channel. A similar model is not included in the PWR version since the system pressures required to levitate cladding are not expected for the open-lattice core of PWRs. The rationale provided by IDCOR for the use of a levitation-based blockage criterion for the simulations was that, unlike actual reactor geometries, the PBF tests provide little or no bypass flow around the fuel bundle during certain periods of the tests, and force steam generated in the pool to find its way through the degraded fuel bundle.
Although BWR/HEATUP appears to provide reasonable agreement with the total hydrogen generated in the PBF tests, these tests do not constitute verification of the IDCOR contention that in LWR accidents cladding / fuel melting and relocation will necessarily result in sufficient flow resistance in a degraded fuel assembly to effectively block steam flow through the channel and terminate further oxidation in the fuel assembly.
For example, essentially complete flow blockage (greater than 98 percent) would be required to result in steam pressurization sufficient to divert coolant flow from a degraded BWR fuel assembly to adjacent fuel assemblies (Reference 2). Significant blockages were observed in the PBF post-test bundle examinations, but not to this extent.
Furthermore, using the actual boildown rates, the SCDAP code was also able to predict with reasonable accuracy the total amount of hydrogen produced in SFD 1-1; SCDAP mechanistically models cladding / fuel relocation and attendent hydrogen production rather than arbitrarily assuming complete flow blockage and oxidation termination upon relocation, and predicts continued oxidation following relocation.
It should be noted that it is the molten Zircaloy relocation and not molten fuel relocation that limits the hydrogen generation in the early intact-geometry oxidation transient; this Zircaloy relocation is properly modeled as a relocation process separate from fuel slumping in the mechanistic MELPROG and SCDAP codes, as well as in the newer version of HEATUP.
In summary, the PBF tests provide additional insights into core melt progression but considerable uncertainty remains regarding the validity of the IDCOR assumption of blockage-induced flow diversion and oxidation termination.
3.1.2 COMPARIS0NS WITH TMI-2 Simulations of the TMI-2 accident have been performed by IDCOR and NRC in an attempt to further validate models and assumptions regarding hydrogen
It must be recognized, however, production and core melt progression.
that there is considerable uncertainty regarding the boundary cbnditions 4
for the accident and the interpretation of accident parameters such as the time dependent configuration of the core and associated hydrogen '
Also, code comparisions must be based on the estimated total production.
release of hydrogen since there were no direct measurements of hydrogen production at TMI-2.
IDCOR's simulation of the TMI-2 accident was performed using a version of MAAP which includes the new models for core-upper plenum natural circulation, core inter-channel flows, clad ballooning, and radial radiation heat transfer. This calculation was terminated prior to reflood of the damaged core because the MAAP code lacks sufficient models toAs a mechanistically treat key phenomena in a submerged, degraded core.
result, comparisons regarding hydrogen production up to the time of the hydrogen burn were not possible.
IDCOR contends, however, that the primary system pressure response prior to termination of the calculation is consistent with their core blockage / oxidation termination model, f.e.,
if oxidation were not terminated, primary system pressures greater than observed during the accident would have resulted.
Battelle Columbus Laboratories (BCL) has analyzed the TMI-2 accident using the MARCH 1 code and has estimated the extent of metal-water reaction that would be predicted by MARCH 2.
Battelle estimates that they would predict approximately 35-40 percent metal-water reaction using MARCH 2 which is somewhat less than what has been estimated for the accident (Reference 3).
Source Term Code Package (MARCH 3) calculations for TMI-2 are planned but have not yet been completed.
3.1.3 STAFF POSITION On the basis of the MAAP and SCDAP code comparisons with PBF tests, the l
PBF post-test examinations, and the limited code comparisons with TMI-2, the staff concludes that the IDCOR assumption that complete channel blockage occurs following cladding / fuel relocation has not yet been The staff recognizes that the formation of adequately substantiated.
I significant blockages during relocation is very likely; this is evidenced
~
by the results of several of the PBF tests and also appears to be supported by examinations of the central region of the TMI-2 core.
Considerable uncertainty remains, however, concerning the degree of blockage, flow patterns around and above the blockage, and the extent of blockage effects on hydrogen generation. In this regard, the staff continues to believe that models which allow oxidation to continue in degraded fuel channels following cladding / fuel relocation (such as the models used in MARCH and MELPROG) would provide more realistic estimates of in-vessel hydrogen production.
Additional comparisons between MARCH, MELPROG, and MAAP code results provide some insight into the large uncertainties inherent in the In-vessel hydrogen production estimates modelling of in-vessel phenomena.
predicted by) MARCH 2 and an earlier version of MAAP were com (Reference 4 and were found to differ by a factor of 2 or more, with MARCH predicting the larger values. More recent MELPROG calculations for
)
L a station blackout sequence at Surry (Reference 5) resulted in total in-vessel hydrogen production comparable to that predicted by MAAP when a relocation temperature of 2200K was assumed in MELPROG. However, later MELPROG calculations.in which cladding / fuel relocation was assumed to occur at 2500K instead of 2200K, resulted in total hydrogen production approximately twice that.in the original calculation, i.e., comparable to that predicted by MARCH (Reference 6). The relocation temperature referred to in the cited MELPROG sensitivity analysis calculations, it is
)
very important to note, is a molten Zircaloy (and dissolved U0relocat temperature. Relocation temperature is jur.t one of several parameters that are considered to contain large uncertainties.
BCL, in Reference 4, concluded that while some of the predicted differences between MARCH and..MAAP results are due to modelling differences between the two codes, many are due to user selected input or model parameters.
In BCL's view, given the present state of knowledge both the IDCOR and NRC modelling approaches must be considered as plausible.
The staff believes that actually, neither the MARCH nor the MAAP treatment is completely in accord with our current knowledge of the governing In the early. rod-geometry phase, MARCH is extremely physical processes.
simplistic in its treatment of Zircaloy relocation, whereas MAAP is inconsistent with existing PBF data in assuming that hydrogen generation is cut off in the initial stages of this molten Zircaloy relocation by block-age formation.
In the later stage of molten corium slumping into the lower plenum water, MARCH, by a parametric treatment (particle size and fraction of unavailable Zircaloy), allows for steam generation / debris cooling and oxidation with hydrogen generation that ranges from essentially zero to 100 percent of the unoxidized Zircaloy in the slumped corium. MAAP, on the other hand, assumes essentially no interaction between the molten debris and water in the lower plenum and consequently no hydrogen generation from molten corium slumping. While the MELPROG code represents a marked advance in modelling capabilities, it also is subject to large uncertainties inherent in the modelling of in-vessel It is likely that such uncertainties will always exist to the phenomena.
extent that the calculation of in-vessel phenomena could not be considered precise.
Accordingly, it is the staffs position that a range of in-vessel hydrogen production estimates, encompassing the results of MARCH and MAAP calculations, should be considered by IDCOR in establishing uncertainty bounds on risk. Such estimates should be developed through parametric variation of key input and modelling assumptions governing hydrogen production, as well as through sequence variations including recovery actions. The effect of in-vesse'i hydrogen production significantly greater than predicted by MAAP will also be considered as part of the uncertainty analysis performed for NUREG-1150.
b.
3.2 MULTI-DIMENSIONAL NATURAL CIRCULATION FLOW AND OXIDATION /FAII.UU OF PLENUM STRUCTURES l
IDCOR has modified the MAAP code to include among other things a model for multi-dimensional natural circulation flow in the reactor vessel; the staff assessment of this model is provided in Issue Paper 2. (Also addressed as part of Issue 2 is the question of whether the primary system boundary will survive the high structure temperatures promoted by natural circulation.) Natural circulation flow is not modelled in the MARCH 3 code, but it is modelled in the recently developed mechanistic codes, i.e., SCDAP/RELAP5 and MELPROG.
Multi-dimensional natural circulation flow in the reactor vessel, specifi-cally between the core and the upper plenum of the reactor vessel, can
}
influence the timing and extent of in-vessel hydrogen production, as well as the core' melt progression. This phenomenon may not significantly affect i
low pressure sequences in PWRs because of the dependence of natural circu-lation on vapor density. BWRs, because of their segmented fuel design which divides the core into isolated subassemblies, would not undergo significant natural circulation effects in the core until the water level drops to below the bottom of the channel boxes.
3.2.1 SUPPORTING CALCULATIONS PWR calculations based on the revised MAAP models were performed by IDCOR for a high pressure (station blackout) sequence and an intermediate f
For l
pressure (small break LOCA) sequence and documented in Reference 1.
the high pressure sequence, these calculations indicate a slight reduction-in total hydrogen production (by about 50 pounds) with the natural circulation model; for the intermediate pressure sequence little change l
in hydrogen production was noted.
PWR calculations for a station blackout sequence were also performed by Sandia National Laboratories (SNL) using both 1-dimensional and 2-dimensional versions of the MELPROG code (Reference 5).
In contrast to the IDCOR observation of reduced hydrogen production with natural circulation, the SNL calculations indicate a slight increase in total hydrogen production (by about 50.nunck) with multi-dimensional natural n
circulation. A further increase in hydrogen production would be predicted by both MAAP and HELPROG if the oxidation of steel were modelled in the codes (particularly with the higher upper plenum structure temperatures observed in the 2-dimensional analyses), but based on the results of recent MELPROG calculations this increase is expected to be limited by the availability of steam in the upper plenum.
In both the IDCOR and SNL calculations, natural circulation results in a delay in the onset of hydrogen production by about 10-15 minutes, but little difference (between 1-and 2-dimensional calculations) in the hydrogen production rate once oxidation begins. Also, both IDCOR and SNL calculations indicate that natural circulation contributes to a more uniform heating of the core and structures, and higher upper plenum temperatures.
SNL calculations using the 2-dimensional version of the MELPROG code indicate upper plenum temperatures sufficient to fail the upper core plate IDCOR calculations using the modified and cor, trol rod guide tubes; version of the MAAP code (with natural circulation models) also produce high upper plenum temperatures (close to the melting point of steel) but these temperatures are not sufficient to fail the upper plenum structure.
Similarly, calculations performed using MARCH and the 1-dimensional version of MELPROG do not predict failure of the upper plenum structures.
It should be noted that in TMI-2, the upper plenum structures are intact and the steel content in the rubble within the core or lower plenu'n is negligible.
For BWRs, recent preliminary MELCOR calculations at SNL suggest that a major natural circulation loop could be established if the water level falls to below the exit of the jet pumps. The observed flow path is up through the. core and the steam separators then down through the jet pumps The natural circulation induced flow appears to delay to the core inlet.
the onset of core damage by more than an hour in the sequence analyzed.
3.2.2 STAFF POSIT;0N With regard to the effect of natural circulation on in-vessel hydrogen NRC calculations production, both the IDCOR and presently available suggest that natural circulation has only a minor influence on the timing and quantity of hydrogen release for PWRs. Additional calculations using the MELPROG code are planned over the next several months to assess the issue of natural circulation and its effect on RCS fission product and These calculations will consider the effect of temperature distribution.
hydrogen production by steel oxidation, and will provide a more defensible basis for concluding on the importance of natural circulation on hydrogen -
production. Similar calculations using SCDAP/RELAP5 are also planned for Based on the a low pressure sequence for Bellefonte, a B&W reactor.
currently available information, it is the staff's judgement that in-vessel natural circulation does not have a major effect on the timing or total amount of hydrogen generation for PWRs, and that computer codes which do not account for natural circulation, e.g., MARCH, can continue to be used to develop estimates of in-vessel hydrogen production. The staff will reassess this position upon completion of the MELPROG and SCDAP calcu-For BWRs, the significance of natural circulation on the timing 1ations.
and hydrogen generation will be assessed upon the completion and review of additional calculations using MELCOR.
With regard to the effect of natural circulation on core melt progression, the existing analyses suggest that the melting and failure of steel structures in the upper plenum of the reactor vessel, and hence the amount of steel which may be present in the molten debris'at vessel failure, can It is the staff's be significantly increased with natural circulation.
judgement that'the failure of upper plenum structures due to natural circulation in-vessel is sufficiently uncertain at this time that failure of the structures cannot be ruled out for high pressure sequences in PWRs.
In this regard, upper plenum structure failure and its effect on core debris composition and the issue of direct containment heating should be
considered by IDCOR in establishing uncertainty bounds on risk. The steel
~
content of the melt can also influence the issue of core-con'crecte interactions, however, sensitivity studies reported by IDCOR in Reference 1 indicate that the presence of large quantities of steel in the melt has only a minor affect on core-concrete attack. Also, the results of a CORCON sensitivity study (Reference 10) show that increases in the quantity of steel in the melt lead to a marked reduction in fission product releases.
3.3 MELTING AND FAILURE OF BELOW-CORE STRUCTURES Both the IDCOR MAAP code and the MARCH module of the NRC Source Term Code Package contain greatly simplified and essentially arbitary representations of core slump, core collapse, and reactor vessel failure processes; the mechanistic MELPROG code gives the best available analysis but is also
~
The subject to lar'ge uncertainties in the modeiling of these processes.
principal effect of uncertainties in the melting and failue of below-core structures would be on the time and mode of vessel failure, and the mass and composition of debris at vessel failure. These matters are of crucial importance in high pressure sequences as they influence the debris dispersion potential for direct containment heating (Issue 8), the hydrogen concentration in containment at vessel failure (Issue 17), and the time-dependent melt addition rates for core-concrete interactions (Issue 9 and 10) 3.3.1 MODELLING APPROACHES The revised version of MAAP described in IDCOR report 85.2 includes a simplified model to track the candle-like downward progression of cladding / fuel. The molten nterial is assumed to accumulate in the lower most node until that becomes completely molten; at that time the material and any molten material in adjacent nodes falls to the lower head. The l
MAAP models provide for only minimal interactions between the molten material and the water in the lower plenum, and hence the debris does not Subsequent heatup and attack of the reactor vessel lower head by quench.
the molten debris is calculated, and produces vessel breach within tens of seconds to a few minutes. Vessel breach is always assumed to initiate as a hole entres,nnnM nn to the diameter of a single in-core instrument penetration in PWR analyses and a single control rod drive penetration 1n BWR analyses, hence, the melt ejection rate into the containment is low initially. Ablation of the surrounding steel is modelled in MAAP, however, resulting in calculated vessel blowdown times ranging from 4 to 80 seconds for high pressure and low pressure sequences, respectively.
In the latest version of MAAP, the mass of steel added to the core debris by the melting of below-core structures is considered to be equivalent to the mass of the structures (excluding contol rod drive structures in BWRs) located directly below each failed fuel channel. The IDCOR model generally results in release at vessel failure of approximately 20-40% of the mass of the core and the below-core structures (excluding control rod drive structures) for both PWRs and BWRs (Reference 7).
In contrast to the IDCOR model, MARCH delays debris contact with the lower head until core collapse. This occurs when either a specified fraction of the core is melted (typically 75%), or the calculated temperatures of the
a,
lower core support structures or the core barrel exceed user-specified values.
(MARCH contains a core slump model which assumes relocation of molten portions of the core to the below-core structures; it is this slump model that causes the lower core support structures to heat up and fail, causing core collapse.)
In general, the core melt fraction criterion is the controlling mechanism, however, in numerous BWR scenarios core When the MARCH collapse occurs as a result of lower core support failure.
core collapse criteria are met the total core, including both melted and solid fuel material, is assumed to relocate instantaneously to the lower head. Gross lower head heatup is calculated and failure is considered to At vessel occur upon overheating of the vessel at a specified depth.
failure all core debris is released over one calculational timestep (generally about 15 seconds).
In the MARCH calculations, the core debris is considered to include all below-core steel structures including the lower core support plate in PWRs and control rod drive components in BWRs.
It is important to note that there is sufficient water in the large BWR lower plenum to freeze the entire core mass if it enters the plenum in a If this occurs, it is likely that the core debris would coolable manner.
be in the process of reheating at the time of reactor vessel lower head failure.
Differences between MAAP and MARCH models for core melt progression and vessel failure do not appear to lead to significant differences in the Approximate comparisons between predicted time to vessel failure.
calculations performed using MARCH 2 and an earlier version of the MAAP code show reasonable agreement for the PWR and BWR sequences considered (Reference 4). These comparisons are considered approximate in that the accident sequence definition and plant parameters were not identical in all cases. MARCH and MAAP predictions of the time between core uncovery and vessel failure differ by less than I hour for nearly all of the An additional comparison between MARCH 2 and MAAP sequences considered.
results for an S HF sequence in Surry (Reference 8 and 9, respectively) indicates simila. agreement. On the basis of this information, it is the E
staff's judgement that the effect of differences between IDCOR and NRC in l
the modelling of core melt progression and vessel failure have only a secondary influence on time of vessel failure.
Oiffercnce:: between 10COR and NRC models for core melt progression and I
vessel failure do, however, result in significant differences in tne mass j
and composition of core debris at vessel failure. As previously mentioned, IDCOR models result in release of at vessel failure 20-40 percent of the total core mass with a proportional steel content, whereas, MARCH 3 calculations result in release of the entire core mass along with a significant amount cf steel. The debris is molten in MAAP but in MARCH can be solid in the process of remelting. Preliminary MELPROG calculations for a station blackout sequence at Surry (Reference 6) appear j
to also indicate that a significant mass of debris (approximately half of the core mass) can be released at vessel failure.
In contrast, only a small fraction of the core (less than 20%) relocated into the lower plenum t
j at TMI-2, and the damage to the lower core support structures appears to t
be minimal.
3.3.2 STAFF POSITION It is the staff's judgement that the existing models for core melt progression and vessel failure are not sufficiently advanced to provide reliable predictions of the mass and composition of core debris present at the time of vessel failure. Furthermore, uncertainties in these parameters can have a significant effect on the issue of direct containment heating and a lesser effect on in-vessel hydrogen production and core-concrete interactions.
It is the staff's position that the IDCOR models represent a reasonable central estimate of the mass and composition of debris released at vessel failure, but that because of large uncertainties in the understanding and modelling of core melt progression phenomena, release of a larger mass of core debris containing various amounts of steel should be considered by IDCOR in establishing uncertainty bounds on ri.sk., particularly in addressing the direct containment heating phenomenon. The range of debris mass and composition considered should encompass the results of MARCH calculations, or alternatively could be based on the results of calculations using mechanistic codes with conservatisms applied to account for uncertainties.
4.0 CONCLUSION
S The staff has reviewed the key differences between the IDCOR and NRC in-vessel hydrogen production and core melt progression models, and has reached a tentative position regarding the acceptability of the various models for plant analyses and risk assessments.
With regard to cladding / fuel relocation and blockage formation, the staff -
continues to believe that the IDCOR blochge assumption is not substantiated by the available experimental data and that oxidation will continue in degraded fuel channels following relocation. However, given the present state of knowledge and inherently large uncertainties in the phenomena, the staff acknowledges that the IDCOR modelling approach is plausible and that the quantity of hydrogen produced in-vessel may be considerably less than the staff's estimates. Accordingly, the staff's position on this matter is that a range of in-vessel hydrogen production estimai.es, enwnipessing the result: Of.". ARCH and ga.P. should he considered by IDCOR in establishing uncertainty bounds on risk and in fomulation of the Individual Plant Evaluation Methodology (IPEM).
Such estimates should be developed through variations in user specified parameters governing hydrogen production, as well as through sequence variations, including recovery actions.
It is also our judgement that core melt progression phenomena, including multi-dimensional natural circulation effects and failure of steel structures, are sufficiently uncertain that the mass and composition of core debris released at vessel failure should be treated parametrically in plant analyses. Accordingly, it is the staff's position that the release of a larger mass of core debris (than presently assumed by IDCOR) containing various amounts of steel should be considered by IDCOR in t
l establishing uncertainty bounds on risk, particularly with regard to the
e.
issues of hydrogen combustion and direct containment heating. The range of debris mass and composition considered should encompass the Tesults of MARCH calculations, or alternatively could be based on the results of calculations using more mechanistic codes such as MELPROG with conservatisms applied to account for uncertainties.
These staff positions will be reassessed upon completion and review of planned calculations using the MELPROG and MELCOR codes.
5.0 REFERENCES
IDCOR Technical Report 85.2, " Technical Support for Issue Resolution,"
1.
July 1985.
Cronenberg, A.W., "An Assessment of Hydrogen Generation Noted from 2.
Integral Severe Fuel Damage Experiments," Engineering Science and Analysis, to be published.
3.
Letter From R. Denning, BCL to R. Meyer, RES, July 29, 1985.
Cybulskis, P., et al, Letter to USNRC on MARCH-MAAP Comparison 4.
Calculations, April 29, 1985.
Kelly, J.E., " Preliminary MELPROG Insight into NUREG-1150 Issues",
5.
Presentation at Staff Meeting Regarding Status of TMLB Studies, March 17, 1986.
Kelly, J.E., Sandia National Laboratories, letter to J. Han, NRC, dated 6.
June 24, 1986, Transmitting "MELPROG-PWR/ MOD 1 Analysis of a TMLB' Accident Sequence," June 1986.
7.
Gabor, J., Private Communication Gieseke, J.A., et al, "Radionuclide Releases Under Specific LWR Accident 8.
Conditions," Volume IV, BMI-2104, July 1984.
9.
IDCOR Technical Report 23.1, "Sequoyah Nuclear Plant, Integrated containment Analysis." December 1984.
Bradley, D.R., and Shiver, A.W., " Uncertainty in the Ex-Vessel Source Term 10.
caused by Uncertainty in In-Vessel Models," International ANS/ ENS Topical Meeting on Thermal Reactor Safety, San Diego, California, February 3-6, 1986.
Resolution of IDCOR Issue #11 Revaporization of Fission Products i
1.
ISSUE DEFINITION Retention in the reactor coolant system (RCS) occurs as volatile fission products, released from the core during heat-up, condense on cooler RCS surfaces. For accidents involving long RCS-transit times, the fraction of material retained can be large.
A phenomena which counteracts this retention is revaporization.
Revaporization occurs due to increasing RCS surface temperatures, caused by the decay heat associated with deposited fission products. It is agreed, both by IDCOR and NRC, that revaporization occurs; however, the rate and i
timing of this revaporization is uncertain.
Timing of revaporization impacts the source term. Early revaporization of fission products will, to some extent, be mitigated through physical interaction with aerosols provided by the corium-concrete interaction.
Late fission product revaporization, after this interaction is complete, will not be mitigated, as the amount of aerosols significantly reduced.
Timing is determined by vapor pressure, which is a function of chemical species and temperature. The higher the vapor pressure, the earlier reva-porization occurs. Species deposited include Cs0H and CsI. These species may react with the RCS surfaces to form low vapor pressure compounds.
Fission product-RCS surface reaction may reduce and delay revaporization.
This issue paper focuses on the modelling of the surface reaction chemistry.
f In addition to the lack of revaporization chemistry, there is great un-certainty in the temperatures at which fission product-RCS surfaces com-pounds will revaporize. These temperatures may approach the melting point of steel, poten tially decreasing the amount of revaporization. There is also much uncertainty about the revaporization process during the vessel failure process and after vessel failure. Thermal hydraulic uncertainties f
include heat losses through piping and natural convection patterns. Ex-vessel releases may occur after melting occurs. Other IDCOR issues impact revaporization, such as natural circulation and suppression pool bypass.
nup to these uncertainties. there is potential for this issue to alter l
risk perspectives.
l This issue impacts, to an extent, all plants and sequences. Significant retention in the RCS is required for revaporization impact. Sequences which l
are susceptible to this impact are transients, small break LOCAs and the PWR interfacing LOCA. Flow through the RCS is also required. LOCA se-quences involve a break in the RCS. When the melt fails the reactor vessel, a second break forms in the bottom head, producing a " chimney" effect, resulting in flow through the RCS. Transients lack this " chimney" effect.
However, transients with induced failures or large breaks in the bottom I
head can achieve a similar effect. The interfacing LOCA, due to tempera-l ture gradients in the RCS, may result in revaporization at one point in the system and recondensation further down the flow path prior to release.
i
2 APPROACH TO RESOLUTION 2.
At the present time, the Source Term Code Package (STCP) is the tool the NRC uses to calculate source terms. There are no models in.the STCP to account for revaporization after vessel failure and because development of MELCOR is nearing completion, there are no plans to incorporate any.
Both MELCOR, an integrated risk code and MELPROG, a detailed mechanistic code, include revaporization models. However, both codes must undergo verification and validation benchmarking before their use is acceptable.
Models in MAAP allow revaporization of Cs0H and Csl species but do not include reaction between the deposited fission products and RCS surfaces.
Deposited fission products revaporize at their nominal vapor pressures.
The exclusion of the chemical reactions, which could delay revaporization, due to the formation of low vapor pressure compounds, is an important parameter missing from the code.
3.
STAFF ASSESSMENT In an effort to resolve this issue, IDCOR conducted sensitivity studies on vapor pressure using MAAP. Reducing vapor pressures to.01 of the orig-inal values, IDCOR recalculated source terms for selected plants and se-Revised IDCOR calculations using reduced vapor pressure produced quences.
a modest amount of revaporization, however, this approach, i.e., varying vapor pressure to simulate extent of revaporization, results in release of fission products early in the accident. NRC believes that the use of fission product surface chemistry models in lieu of the vapor pressure approach used by IDCOR would tend to predict slower but continuous revaporization. Because delayed revaporization is not mitigated as effectively as early revaporization, the IDCOR model is considered to have non-conservative effects on source terms.
4.
CONCLUSION It is the belief of the NRC that neither the STCP nor MAAP adequately account for the impact of chemistry on revaporization, and that both NRC and IDCOR reference plant assessments should address the uncertainties associated with this issue via uncertainty analyses. Recognizing the limitations of the STCP and the lack of a thorough understanding of l
revaporization chemistry and thermal-hydraulics during and after vessel Tailure, this issue had been included in the NUREG-1150 uncertainty analysis. Similarly IDCOR should also consider revaporization in their uncertainty studies. Furthermore, the IPEM should contain screening criteria that will highlight any plant-specific deviations from generic assumptions on revaporization, such as the magnitude of heat losses from the RCS.
l 5.
REFERENCES IDCOR Technical Report T85.2 Technical Support for Issue Resolution. July, 1985
)
ISSUE TITLE #13A - AMOUNT AND TIMING OF SUPPRESSION P0OL BYPASS 1.0 ISSUE DEFINITION IDCOR maintains that aerosols will not leak through drywell penetrations, thus bypassing the suppression pool, because the aerosol particles will deposit within the leak paths and clog them.
In addition, IDCOR would predict that any leakage would occur only at early times in an accident.
The staff is concerned that large openings will not be plugged by aerosol particles and will persist as bypass leakage pathways throughout any accident.
2.0 APPROACH TO RESOLUTION The staff has investigated the literature cited in Reference 1~ supporting the Vaughan model of aerosol plugging used by IDCOR.
IDCOR has performed a sensitivity study of the importance of the Vaughan model and its adjustable parameter to releases to the environment. This study shows that the Vaughan model results in source term predictions over an order of magnitude less than those predicted without aerosol plugging.
3.0 STAFF ASSESSMENT In, Technical Report 85.2, IDCOR incorrectly describes the Vaughan model as a'model for predicting the mass of aerosol entering a duct prior to plugging; the model predicts the mass carried though the ducts as originally proposed by Vaughan. The statement of the model was later changed by Morewitz to obtain a better fit to experimental data.
The original data cited by Morewitz are very convincing for small diameter ducts. The experiments using larger ducts, over a few centimeters in diameter, however, all involved very long ducts and small pressure dif-ferences driving the aerosol flow, and passed hundreds of kilograms of aerosol particles prior to plugging. The conditions of interest in severe accidents are those in which the pressure difference between the drywell l
and the wet well or containm'ent is large enough to drive a rapid flow through a bypass pathway. hut insufficient to clear the water leg from the
(
suppression pool, i.e., pressure differences of a few tenths of an l
atmosphere. Some experiments using ducts of less than one centimeter diameter were conducted with large driving pressures, but all of the l
experiments with larger diameters were at pressure differentials much below the range of interest in severe accidents.
i f
4.0 CONCLUSION
l For driving pressures of a few tenths of an atmosphere, the Vaughan model can be reasonably trusted to be applicable to ducts of less than a l
centimeter in diameter.
In such applications, the Vaughan model should be i
l l
2 taken to predict the mass of aerosol escaping through the duct prior to plugging.
Ducts having diameters larger than one centimeter may be rapidly plugged if the flow induced through them is sufficiently slow, or if the, ducts constitute long and tortuous paths. At present, however, there is insufficient experimental evidence to demonstrate that plugging would occur rapidly enough to significantly decrease suppression pool bypass.
For such large bypass ducts, no plugging by aerosols should be considered.
To resolve the general issue of aerosol plugging of leakage paths, IDCOR should recompute aerosol leakage for paths greater than one centimeter diameter, while the NRC should adopt the Vaughan model for computing leakages through pathways less than one centimeter in diameter.
5.0 REFERENCE 1.
Fauske and Associates, Inc., " Technical Support for Issue Resolution, "IDCOR Technical Report 85.2, July 1985.
~
ISSUE 17 - HYDR 0 GEN IGNITION AND BURNING 1.0 ISSUE DEFINITION Substantial differences exist in the way that IDCOR and NRC model combustion of hydrogen-air-steam mixtures. The major differences between IDCOR and NRC models lie in the areas of (1) ignition criteria, (2) rate and completeness of
~
burn, (3) gas transport by natural convection, and (4) hydrogen recombination in the reactor cavity. Differences also exist in the containment building nodalization scheme used in IDCOR and NRC plant analyses, particularly in the This issue reactor cavity region, and in the treatment of flame propagation.
deals with the pressure and temperature loads imposed on the containment and equipment by the accumulation and combustion of hydrogen during a severe accident, and whether the differences between the IDCOR and NRC treatment of combustion are of sufficient magnitude to challenge containment integrity early
~
in an accident.
The differences between the IDCOR and NRC treatments of hydrogen ignition and burning are of principal concern for pressure-suppression type containments such as the ice condenser and Mark III containments because hydrogen combustion events represent a greater challenge to containment integrity for these The NRC models tend to produce burns at higher hydrogen designs.
concentrations, thereby leading to the prediction of higher containment NRC's treatment of hydrogen combustion also pressures and temperatures.
indicates greater likelihood of flame propagation into the upper compartment of the ice condenser, where impact on containment pressurization is greatest, and the development of potentially detonable mixtures in the ice condenser ice bed and the MARK III wetwell region in certain sequences, which IDCOR's treatment appears to preclude.
For large dry containments, differences in ignition and natural convection modelling influence the hydrogen concentration at which combustion will occur, and the amount of hydrogen and carbon monoxide which can accumulate in containment during core-concrete interaction, respectively.
Mark I and II containments are inerted, and are not affected by this issue except as it relates to combustion in secondary containment buildings.
2.0 APPROACH TO RESOLUTION Soth IDCOR and NRC have developed comDuter codes to analyze nuclear reactor accidents involving the transport and combustion of hydrogen.
IDCUR analyzes hydrogen combustion events using the MAAP code.
NRC uses the HECTR code for detailed stand-alone containment response calculations involving hydrogen In NRC Source Term Code Package calculations, hydrogen combustion events.
transport and combustion is modelled in lesser detail in the MARCH module.
Many of the HECTR combustion models have been incorporated into MARCH, but the code retains a simplified pressure-equilibration flow model, and in practice provides a less compartmentalized representation of containment than either HECTR or MAAP.
In order to evaluate the significance of modelling differences between these codes, it was agreed that IDCOR and NRC would define and calculate a standard problem which would involve all pertinent phenomena, e.g., hydrogen combustion
y 2
Based on a comparison of the MAAP results with those and natural circulation.
obtained using HECTR, the standard problem would aid in highlighting and If assessing the importance of differences between the IDCOR and NRC models.
the differences were found to be insignificant, than MAAP might be considered It was comparable to HECTR (and perhaps superior to MARCH) for that sequence.
recognized, however, that findings based on the results of a single Jtandard problem calculation may not be applicable to certain accident sequences, e.g.,
results obtained for a sequence with fans operable may not be directly-applicable to sequences without fans. Consequently, the review would still rely largely on a comparison of IDCOR and NRC modelling approaches.
The S,,HF (small-break LOCA with failure of both ECC and containment sprays upon switch-over from injection to recirculation) drain closed sequence for the Sequoyah Plant was selected as the basis for comparison. The transient was divided into two parts to facilitate interpretation of results. The first part covers hydrogen behavior during the period of in-vessel hydrogen production, and is intended to address the effect of hydrogen ignition and burning models.
The second part covers hydrogen and carbon monoxide behavior during the ex-vessel period of the transient, and is intended to address the effect of gas transport by natura.1 circulation and continuous recombination behavior in the reactor cavity region.
The results of calculations for Part I have been completed and are summarized below; preliminary calculations for Part 2 are expected to be available in late July.
STANDARD PROBLEM RESULTS -- PART 1
--- HECTR ---
MAAP 6 VOLUME 15 VOLUME 6 VOLUME NUMBER / DURATION OF BURNS Lower Compartment 3/2-3s 6/5-8s 1/842s Upper Plenum 3/0.4s 5/2s 5/20-2050s Upper Compartment 0
0 6/60-626s Annular Compartnent 0
0 5/26-269s PEAK PRESSURE (psia) 23.5 20.7 20.5 PEAK TEMPERATURE (DEG F) 1020 996 199 Based on the standard problem calculations completed to date and the results of additional review of the IDCOR models, the staff has reached a tentative position regarding the resolui. ion of each of the key modelling differences.
The resolution is described below.
3.0 STAFF ASSESSMENT 3.1 IGNirION CRITERIA IDCOR analyses assume combustion occurs if either of two ignition criteria are met. The first criterion, termed the flame temperature criterion (FTC), applies regardless of whether igniters are available.
It involves calculation of the adiabatic flame temperature based on the atmosphere composition and temperature, and assumes that if the flame temperature exceeds a critical value (which is a function of the compartment steam concentration) then combustion of the entire inventory of hydrogen in the compartment (global combustion) will result. The second criterion, termed the burn velocity criterion in this paper, is invoked only if hydrogen igniters are available.
It involves the l
calculation of laminar burn velocity based on the atmosphere composition and
o 3
temperature, and assumes that if the velocity exceeds a minimum value (0.03 ft/sec) then ignition and incomplete combustion will occur.
In contrast, to the two IDCOR ignition criteria, NRC analyses (both HECTR and MARCH) base ignition on user-specified values for hydrogen concentration within each compartment. Hydrogen concentration values are generally selected on the basis of the type of sequence under study, e.g., higher values fo'r s~equences l
without igniters, and the degree of conservatism required for the particular application.
The IDCOR MAAP code also includes a model to treat the spontaneous ignition of hot vapor jets. The staff's assessment of the IDCOR ignition models is presented below.
3.1.1 THE FLAME TEMPERATURE CRITERION (GLOBAL COMBUSTION MODEL) l The IDCOR implementation of the FTC in MAAP accounts for the effects of steam by adjusting the critical temperature, and explicitly requires that the compartment oxygen concentration be greater than 5% as a condition for ignition. These two features represent improvements over previous versions of MAAP.
The FTC was not invoked in Part 1 of the standard problem because the hydrogen was consumed in accordance with the incomplete combustion model prior to reaching a concentration which would satisfy the FTC.
It would however, be invoked in analyses of large dry containments and secondary containment buildings and those sequences in ice condenser and Mark III containments in which the igniter systems are inoperable.
j Review of the FTC indicates that for atmosphere conditions expected during typical accident sequences, the criterion would cause combustion to occur at a hydrogen concentration of approximately 8% hydrogen -- even in plants and sequences in which specific ignition sources (thermal igniters, sparks at i
motors / relays) may not exist.
Implicit in the IDCOR model is an assumption that random ignition sources, of sufficient energy to initiate combustion, will always exist at the time the FTC is first satisfied.
4 l
The staff recognizes that the probability of random ignition increases commensurate with hydrogen concentration, and that use of the FTC as an l
ignition criteria reflects this increased probability. However, we are not persuaded that the probability is.so high that combustion at hydrogen concentrations richer than existing when the FTC is first met should be ruled out iur plants and sequcnces without igniters. The reason fnr this is that l
lack of sufficient and timely random ignition sources can act to delay ignition to some time later than when the mixture first satisfies the FTC. The hydrogen concentration can increase substantially during this period, and influence the likelihood of early containment failure.
It is the staff's position that unless IDCOR can demonstrate that a reliable source of ignition will exist in containment, such as an operable deliberate ignition system, the effect of a delay in ignition after first satisfying the FTC should be considered by IDCOR 1
in all future MAAP analyses and in establishing uncertainty bounds on risk for all plants. This should include consideration of delays in ignition until (1) the time of reactor vessel failure, and (2) various times following vessel failure, up to 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. The staff will address the impact of delayed ignition and the effect of burns at higher hydrogen concentrations as part of the uncertainty analysis performed for NUREG-1150.
1
.,,_m-._
w_mp.--.
.,,-_.-._.__...,__,,-__,.-..,.w,-.
,-.y__.,...c_
y
4 THE BURN VELOCITY CRITERION (INCOMPLETE COMBUSTION MODEL) 3.1.2 A further review of the IDCOR burn velocity criterion and MAAP results indicates that the model initiates combustion when the hydrogen concentration in a compartment reaches about 4.8 and 5.0% for steam concentrations of 0 and 60%, respectively. The MAAP model appears to be reasonably well su@orted by experimental data for steam concentrations less than about 15%, but inaccurate for richer steam concentrations.
For example, the data indicates that'for steam concentrations of 30% the hydrogen concentration required for combustion would be about 6.5%.
Furthermore, the model does not account for the effects of turbulence or suspended water droplets. At low levels, turbulence can have a beneficial effect by promoting combustion at lean hydrogen concentrations, i.e., 5%, but at high levels turbulence can act to quench flames as they propagate from the ignition source. Suspended water droplets, in sufficiently dense concentrations, can significantly shift the lower flammability limit of The fonnation of water fogs in the various hydrogen-air-steam mixtures.
compartn.&ats of an ice condenser containment has been analyzed by both the ice condenser utilities and the staff. Results of these studies suggest that the flaninability limit can increase to 8-9% hydrogen in certain compartments.
It is the staff's belief that for plants and sequences with igniters, combustion is likely to initiate at hydrogen concentrations higher than predicted by the burn velocity criterion in MAAP.
If the containment pressure prior to combustion is substantial (e.g. 35 psia), the difference in the hydrogen concentration at ignition can influence the likelihood of containment failure for weaker containments.
In this regard, the ignition criterion used in the IDCOR incomplete combustion model, in its present form is considered To be considered unacceptable, and cannot be granted approval by the staff.
acceptable, the model would need to account for the effects of the previously mentioned phenomena on the flammability of hydrogen-air-steam mixtures.
Specific modifications required by the staff include the addition of explicit checks for steam inerting and insufficient oxygen (i.e. steam less than 55% and oxygen greater than 5% in order to ignite) and a capability to account for the effects of steam and suspended water on the flammability limit.
3.1.3 SPONTANE0US IGNITION The MAAP code includes a model to treat the spontaneous ignition of hot vapor jets; the model assumes that ignition occurs whenever the calculated jet temperature exceeds a critical value of 1400 F.
NRC models (in HECTR) are capable of simulating autoignition of hot mixtures as a user option, but on the l
basis of bulk average compartment temparature rather than.iet temperature. In j
IDCOR plant analyses, the spontaneous ignition model appears to be invoked only I
for Mark I and II containn'ents upon failure of the drywell.
Ignition occurs as a result of two factors (1) high drywell temperatures calculated by IDCOR during core-concrete attack, and (2) the assumption by IDCOR that containment failure occurs as a small leak, thereby creating a jet.
The available database regarding spontaneous ignition of hot mixtures is l
sparse, but appears to support the IDCOR assumption under conditions in which sufficient oxygen is available. There is significant uncertainty, however, as to whether spontaneous ignition will occur reliably for all situations, particularly cases in whicn hot hydrogen-rich mixtures are released from the drywell of a Mark I and 11 containments. This is due to the fact that autoignition is dependent upon break (i.e., containment failure) size and configuration, gas composition and flow rates, jet expansion effects, and l
5 chemical kinetics. The dependence of autoignition on these factors is not well understood and is not modelled in MAAP. Furthermore, NRC calculations for a Mark I containment do not confirm that the temperatures necessary for autoignition are produced in the drywell, nor does the staft necessarily concur in the IDCOR characterization of the containment failure mode.
It is the staff's position that spontaneous ignition is sufficiently uncertain to require that IDCOR consider lack of ignition and corresponding accuniulation and burning of combustible gases in establishing uncertainty bounds on risk.
The present staff analyses using MARCH are considered adequate in this regard.
In the uncertainty analyses performed as part of NUREG-1150 the effect of accumulation and subsequent burning of combustible gases in the secondary building of a MARK I containment will be addressed.
3.2 RATE AND COMPLETENESS OF BURN The IDCOR incomplete combustion model appears to predict continuous burning in essentially all plant analyses, whereas the NRC treatment tends to predict a number of discrete burns. The NRC approach can lead to a buildup of higher hydrogen concentrations than MAAP and hence larger burns and higher contain-ment pressures, although this behavior was not observed in the standard For global burns predicted by the FTC model, the IDCOR models predict problem.
substantially higher flame speeds and shorter burn times, in approximate agreement with HECTR models and experimental data.
In Technical Report 85.2, IDCOR has compared results of MAAP predictions with results of selected combustion experiments. The agreement is favorable but appears to be due to judicious selection of user-specified, non-physical scaling tactors which influence tha rate and completeness of burning.
Review of the IDCOR models indicates that the burn rate calculation in MAAP is very similar to that in other combustion codes, i.e., burn times are determined by dividing a characteristic burn length (in MAAP a function of compartment radius for global burns, or number and location of igniters and compartment height for incomplete burns) by an empirically determined flame speed.
However, a major difference between MAAP and other codes is that MAAP permits combustion to begin as soon as the calculated laminar burning velocity exceeds a threshold value of 0.03 ft/sec.
For burn lengths typical of containment buildings, this results in burn times in MAAP plant analyses of several minutes or more, or essentially continuous burning (see standard problem results in section 2). Once the burn velocity criterion is met, the burn time is establishcd and rc= int constant for the duration of the burn.
(It should be noted that if the conditions for a global burn, i.e., the FTC, are satisfied while incomplete burns are in progress, the burn time in MAAP is recomputed and the burn is assumed to proceed as a global burn.)
Uncertainties in flame speeds can be large and can result in significant variations in calculated burn times. Nevertheless, the preponderance of experimental data at several scales indicates that burn times for pre-mixed atmospheres will be on the order of several seconds rather than several minutes as predicted by MAAP, even for lean mixtures. This would result in higher pressure rises than predicted by MAAP. Although the MAAP burn rate model is not substantiated by experimental data for premixed atmospheres, the near-continuous burning observed during certain portions of the MAAP calculation may not completely misrepresent the transient situation in containment, in that the MAAP result can be thought of as approximating a diffusion flame in the burn compartment. Diffusion flames have been observed in most transient
1 6
- However, hydrogen combustion experiments and result in benign pressure rises.
the formation and stability of a diffusion flame is highly dependent on the break area, hydrogen / steam release rates, and composition of the surrounding atmosphere, and therefore cannot be assured.
It is the staff's position that the IDCOR burn rate model grossly'urftlerpredicts flame speed and can underpredict containment pressurization for combustion Combustion at a rate events which occur as discrete incomplete burns.
consistent with experimental data could influence the likelihood of containment failure for weaker containments under certain conditions, e.g., a high pre-burn containment pressure combined with an initial hydrogen concentration higher than presently allowed in MAAP. Also, use of excessively long burn times in 2
the IDCOR analyses reduces the amount of hydrogen which might otherwise accumulate and burn elsewhere in containment and can also alter the temperature distribution in containment (which is one factor in assessing equipment survivabilfty).
In this regard, the IDCOR burn rate model, in its present form, is considered to be unacceptable. The IDCOR model should be modified to provide better agreement with available experimental data for flame speed and burn times, without the need for user-specified scaling factors. The effect of greater burn rates and containment pressurization will be considered as part of the uncertainty analysis performed for NUREG-1150.
1 With regard to completeness of burn, the IDCOR model assumes complete combustion of the compartment hydrogen inventory if the flame temperature criterion (FTC) is satisfied, and combustion of some fraction of the hydrogen when the incomplete combustion model is invoked. The volume of hydrogen burned in the latter case is calculated as a function of burning velocity, drag coefficient, number and location of igniters, and compartment geometry. The assumption of complete combustion when the FTC is satisfied is l
consistent with experimental data at the hydrogen concentrations at which the FTC would be satisfied, and is therefore considered to be acceptable. The incomplete combustion model on the other hand, may tend to either underpredict or overpredict experimental data for combustion completeness depending on the particular compartment geometry and igniter configuration. An assessment of the adequacy'of the combustion completeness model should therefore be made as part of the plant specific application of MAAP, and to some extent will be l
dependent upon the resolution of the matter regarding the burn rate model. The effect of burn completion on containment pressurization will also be addressed l
as part of the uncertainty analysis performed for NUREG-1150.
j 3.3 GAS TRANSPORT BY NATURAL CONVECTION l
The MAAP code simplistically models natural convection driven flow between compartments in containment; natural circulation flow is not modelled in the MARCH code but is modelled in HECTR. Natural circulation flow can alter the location of hydrogen burning by affecting the hydrogen, oxygen, and steam distribution in containment.
s Results of HECTR calculations for Part 1 of the standard problem provide little This insight into the effect of natural circulation on containment response.
is because the containment air return fans are operable in the standard An problem, and likely overwhelm the effects of natural circulation.
i additional calculation, with fans off, would provide an appropriate basis for addressing the effects of natural circulation. The staff will require that such a calculation be performed as an addendum to Part 2 of the standard i
7 problem, in order to provide verification of the adequacy of the MAAP model in predicting natural circulation flow.
Although the Part I calculations do not adequately address the matter of natural circulation, the HECTR calculations provide some measure of the effect The HECTR calcu}atJons were of compartmentalization on containment response.
performed using 3 different nodalization schemes (two different 6-compartment representations with volumes and heat sinks apportioned differently, and a 15-compartment representation.) Results of the calculations indicate that the burn sequence was not sensitive to compartmentalization, but that the use of fewer volumes to represent containment results in fewer discrete burns.
Relatively good agreement among the HECTR analyses may be due to the fact that various compartmentalization schemes considered did not result in a shift in
~ the location of hydrogen burning to the upper compartment. Significant differences in burn temperatures between MAAP and HECTR were noted, and are still under further study (800K with HECTR vs. 350K with MAAP). Nevertheless, l
for this particular calculation, it would appear that key differences between 1
the IDCOR and NRC models center on combustion rather than on transport models j
or nodalization scheme.
The staff's assessm'ent of the adequacy of MAAP in predictiong natural
+
circulation flow will be made upon completion of Part 2 of the standard l
problem.
3.4 RECOMBINATION For plants and sequences in which the reactor cavity is dry, the temperature of l
the reactor cavity atmosphere and structures following vessel failure will be sufficiently high to promote recombination of the combustible gases within the cavity. Specifically, hydrogen and carbon monoxide would react with available Most oxygen near the heated structures to form steam and carbon dioxide.
affected by this phenomenon are sequences in large dry, ice condenser and Mark III containments in which both the reactor cavity is dry and the igniter systems are unavailable.
In the IDCOR plant analyses, the lower reactor cavity is modelled in MAAP as a separate computational cell connected to the containment ; roper via vapor flow paths; combustion of hydrogen and carbon monoxide in the cavity is modelled using the flame temperature criterion.
In the NRC (MARCH) analyses the reactor cavity is not modelled separately and combustion phenomena specific to this l
region is ignored.
In Part 2 of the standard problem, the effect of recombination in the reactor cavity will be investigated using HECTR.
i Natural circulation induced flow through the reactor cavity, in conjunction i
with the high reactor cavity temperatures, can result in significant l
recombination of the combustible gases produced by core-concrete interactions with available oxygen in the cavity. However, recombination may in practice be limited by the rate at which oxygen is supplied to the reactor cavity region (relative to the rate at which combustibles are produced) and be precluded by 3
the high steam concentration and/or insufficient temperatures in the reactor Two additional factors cavity which would exist in flooded-cavity sequences.
would tend to diminish or negate the effect of recombination: (1) the combustible gases would compete with hot steel structures for the available oxygen, and (2) the steam produced by recombination could react with hot steel i
structures producing hydrogen.
i
}
i
8
'It is-the staff's belief that for plants and sequences (and portions of sequences) in which the reactor cavity is dry and in which significant convective flows through the region are predicted, credit for recombination The basis for during the ex-vessel portion of the accident is appropriate.
this judgement is that the reactor cavity temperatures would be in excess of 2500*F and gas phase kinetics at this temperature would proceed t6 equilibrium.
However, this judgement is inade in the absence of directly applicable,
experimental data.
For such plants and sequences, the IDCOR recombination models are considered to represent a reasonable estimate of combustion behavior provided that IDCOR modifies their models to account for the reverse rea' tion c
of steam with hot steel in the reactor cavity, and the associated energy It should be noted that the validity of the IDCOR natural transfers.
circulation flow model, which supplies the oxygen necessary for combustion, remains to be confirmed as part of the standard problem. The MARCH model, which provides no modelling of or credit for recombination, is considered to provide more conservative estimates of containment pressure for these sequences since the combustibles are forced to accumulate in containment and burn later.
For plants and sequences (or portions of sequences) in which the reactor cavity is flooded, or in'which insignificant convective flows through the reactor cavity are predicted, no credit for recombination should be assumed; combustible gases generated during core-concrete interactions must be assumed to be added to the containment atmosphere where they will burn at a later time.
For such sequences, the modelling assumptions of MARCH, i.e., no recombination, should be considered applicable.
The staff will require that the Individual Plant Evaluation Methodology adopted by IDCOR include acceptance criteria to ensure that adequate flow paths exist for the reactor cavity designs in all plants and sequences in which credit is taken in the reference plant analysis for recombination in the reactor cavity.
This will include consideration of the availability of fans or other systems which might be required to promote the convective flows.
4.0 CONCLUSION
S The staff has reviewed the key differences between the IDCOR and NRC combustion models, and has reached tentative positions regarding the acceptability of the various models for use in performing risk assessments.
With regard to hydrogen ignition models, the staff believes that the effect of ignition delays beyond the time at which the IDCOR FTC is first satisfied and should be considered by IUCOR in all future MAAF analyses for plant:
l sequences in which igniters are not available, and in estimating uncertainties in risk for all plants. This should include consideration of delays in ignition until (1) the time of reactor vessel failurtand (2) various times 4
1 For plants and sequences in which following vessel failure, up to 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />.
igniters are available, the IDCOR ignition criterion (incomplete combustion) is unacceptable in its present fonn, as it does not adequately account for the i
effects of steam and suspended water droplets.
In additun, the burn rate model results in burn times inconsistent with experimental data.
For these i
plants and sequences, future analyses should incorporate revised models to The staff also believes that spontaneous rectify the noted deficiencies.
ignition is sufficiently uncertain that the effect of accumulation and burning of combustible gases in the secondary buildings of MARK I and II containments should be addressed in uncertainty studies.
4
9 With regard to recombination in the reactor cavity. -the staff believes that the IDCOR models are appropriate for dry cavity sequences in which significant recirculation flows are predicted to occur, provided that IDCOR modifies their model to account for the reverse reaction of steam with steel in the cavity.
This position is contingent upon satisfactory verification of the adequacy of the MAAP model in predicting natural circulation flow. For flood d cavity sequences no recombination should be assumed.
The staff will include in the uncertainty analysis performed for NUREG-1150, consideration of the effects of each of the items mentioned above.
1 l
.