severe accident research program plan update · 2012-06-21 · in august 1989, the staff published...

NUREG-1365Rev. 1

Severe Accident ResearchProgram Plan Update

U.S. Nuclear Regulatory Commission

Office of Nuclear Regulatory Research

AVAILABILITY NOTICE

Availability of Reference Materials Cited in NRC Publications

Most documents cited in NRC publications will be available from one of the followingsources:

1. The NRC Public Document Room, 2120 L Street, NW., Lower Level, Washington, DC20555

2. The Superintendent of Documents, U.S. Government Printing Office, P.O. Box 37082,

Washington, DC 20013-7082

3. The National Technical Information Service, Springfield, VA 22161

Although the listing that follows represents the majority of documents cited in NRC publica-tions, it is not intended to be exhaustive.

Referenced documents available for inspection and copying for a fee from the NRC PublicDocument Room include NRC correspondence and internal NRC memoranda; NRC bulletins,circulars, information notices, inspection and investigation notices; licensee event reports;vendor reports and correspondence; Commission papers; and applicant and licensee docu-ments and correspondence.

The following documents in the NUREG series are available for purchase from the GPO SalesProgram: formal NRC staff and contractor reports, NRC-sponsored conference proceed-ings, international agreement reports, grant publications, and NRC booklets and brochures.Also available are regulatory guides, NRC regulations in the Code of Federal Regulations,and Nuclear Regulatory Commission Issuances.

Documents available from the National Technical Information Service include NUREG-seriesreports and technical reports prepared by other Federal agencies and reports prepared bythe Atomic Energy Commission, forerunner agency to the Nuclear Regulatory Commission.

Documents available from public and special technical libraries include all open literatureitems, such as books, journal articles, and transactions. Federal Register notices, Federaland State legislation, and congressional reports can usually be obtained from theselibraries.

Documents such as theses, dissertations, foreign reports and translations, and non-NRCconference proceedings are available for purchase from the organization sponsoring thepublication cited.

Single copies of NRC draft reports are available free, to the extent of supply, upon writtenrequest to the Office of Administration, Distribution and Mail Services Section, U.S. NuclearRegulatory Commission, Washington, DC 20555.

Copies of industry codes and standards used in a substantive manner in the NRC regulatoryprocess are maintained at the NRC Library, 7920 Norfolk Avenue, Bethesda, Maryland, foruse by the public. Codes and standards are usually copyrighted and may be purchasedfrom the originating organization or, if they are American National Standards, from theAmerican National Standards Institute, 1430 Broadway, New York, NY 10018.

NUREG-1365Rev. 1

Severe Accident ResearchProgram Plan Update

Manuscript Completed: October 1992Date Published: December 1992

Division of Systems ResearchOffice of Nuclear Regulatory ResearchU.S. Nuclear Regulatory CommissionWashington, DC 20555

ABSTRACT

In August 1989, the staff published NUREG-1365, "Re-vised Severe Accident Research Program Plan." Since1989, significant progress has been made in severe acci-dent research to warrant an update to NUREG-1365.The staff has prepared this SARP Plan Update to:(1) Identify those issues that have been closed or are nearcompletion, (2) Describe the progress in our understand-ing of important severe accident phenomena, (3) Definethe long-term research that is directed at improving ourunderstanding of severe accident phenomena and devel-oping improved methods for assessing core melt progres-sion, direct containment heating, and fuel-coolant inter-actions, and (4) Reflect the growing emphasis in two ad-ditonal areas--advanced light water reactors, and supportfor the assessment of criteria for containment perform-ance during severe accidents.

The report describes recent major accomplishments inunderstanding the underlying phenomena that can occur

during a severe accident. These include Mark I liner fail-ure, severe accident scaling methodology, source termissues, core-concrete interactions, hydrogen transportand combustion, TMI-2 Vessel Investigation Project, anddirect containment heating. The report also describes themajor planned activities under the SARP over the nextseveral years. These activities will focus on twophenomenological issues (core melt progression, andfuel-coolant interactions and debris coolability) that havesignificant uncertainties that impact our understandingand ability to predict severe accident phenomena andtheir effect on containment performance. The SARP willalso focus on severe accident code development, assess-ment and validation. As the staff completes the researchon severe accident issues that relate to current generationreactors, continued research will focus on efforts to inde-pendently evaluate the capability of new advanced lightwater reactor designs to withstand severe accidents.

111 NUREG-1365, Rev. 1

CONTENTS

Page

A bstract ....................................................................................... iii

Executive Sum m ary .............................................................................. E-11 Severe Accident Research Plan ................................................................ 1

1.1 B ackground ................ I ........................................................... 11.2 D iscussion .............................................................................. 11.3 Benefits of Additional Research ........................................................... 2

1.4 Criteria for Termination of Research ....................................................... 31.5 Organization of the Report ............................................................... 3

2 R esearch Plan .............................................................................. 52.1 Direct Containm ent H eating .............................................................. 6

2.1.1 R esearch N eeds ................................................................... 62.1.2 Direct Containment Heating-Current Research Program ............................... 102.1.3 Anticipated Results ................................................................ 13

2.2 Core Melt Progression Research .................................................. 14

2.2.1 R esearch N eeds ................................................................... 142.2.2 Current Research Program ........................................................... 172.2.3 Anticipated Results ................................................................ 25

2.3 Fuel-Coolant Interactions and Debris Coolability ........................ .................... 26

2.3.1 R esearch N eeds ................................................................... 262.3.2 Current Research Program .......................................................... 272.3.3 A nticipated Results ................................................................. 30

2.4 Severe Accident Codes ................................................................... 30

2.4.1 SCD AP/RELAP5 ................................................................. 312.4.2. CO NTA IN . .. ..... ............................................................ 332.4.3 M ELCOR .................................................... ...... 342.4.4 COMMIX ............................................................. 402.4.5 V ICTO R IA ....................................................................... 412.4.6 Integrated Fuel Coolant Interaction .................................................. 42

2.5 BW R Mark I Containment Liner Failure .................................................... .43

2.5.1 Current Research Program .......................................................... 432.5.2 Future Plans ............................. ........................................ 43

2.6 Hydrogen Combustion and Research ....................................................... 43

2.6.1 Current Research Program .......................................................... 43

2.7 Source T erm ........................................................................... 45

2.7.1 Status ............................................................................ 452.7.2 PH EBU S-FP ..................................................................... 452.7.3 Other Research Activities ........................................................... 46

v NUREG-1365, Rev. 1

CONTENTS (continued)

Page

3 Research Plan for Advanced Light Water Reactors ............................................... 49

3.1 Introduction ............................................................................ 493.2 In-Vessel Severe Accident Phenomena ............................ . ........................ 49

3.2.1 Core Melt Progression and Reactor Vessel Lower Head Failure .......................... 493.2.2 Fission Product Transport and Behavior .............................................. 50

3.3 Ex-Vessel Severe Accident Phenomena .................................. .................. 50

3.3.1 Research Approach ................................................................ 503.3.2 D iscussion ........................................................................ 50

3.4 Sum m ary ............................................................................... 52

APPENDICES

A Severe Accident Issues, Status, and Progress to Date ........................................... A-1

B Closure Of Severe Accident Issues ............................................................ B-1

FIGURES

2.1.1 Entrainment Versus Pressure Ratio For Steam Blowdown ....................................... 7

2.1.2 Initial Material Conditions For Direct Containment Heating ................................... 9

2.2.1 Core Melt Progression Sequence Showing Blockage or Drainage Paths ............................ 15

2.2.2 Axial Temperature Profiles for the Control Blade, Channel Box and Fuel Rods,Indicating Maximum Lateral Temperature Variations .......................................... 18

2.2.3 XR1 and XR2 Test Bundle Cross Sections .................................................... 19

2.2.4 Melt Drainage Pathways Through the Fuel Canisters Which Bypass the Lower Core Plate ........... 20

2.2.5 Test Bundle Cross-Section for the NRU BWR Test ............................................ 22

2.2.6 MP-2 Test Section Showing Major Components and Thermocouple Locations ...................... 24

2.4.1 NRC Severe Accident Codes ............................................................... 32

2.4.2 CONTAIN Code Validation and Assessment ................................................. 35

2.4.3 MELCOR Code Validation and Assessment .................................................. 38

A.2.1 TM I-2 Core End-State Configuration ....................................................... A-5

TABLES

2.2.1 Test Matrix for the Ex-Reactor Experiments on Metallic Melt Relocation and Blockage Formation

Under BWR Dry Core Accident Conditions ................................................... 21

A.2.1 Sources of Current Integral Experimental Information on Melt Progression ....................... A-4

A.2.2 Core Melt Progression: Status of Current Understanding ....................................... A-6

NUREG-1365, Rev. 1 vi

EXECUTIVE SUMMARY

In May 1988, the staff prepared an "Integration Plan forClosure of Severe Accident Issues" (SECY-88-147) forthe Commission. Many of the elements of this integrationplan require a basic understanding of severe accidentphenomena. In fact, one of the major elements of theplan is a Severe Accident Research Program (SARP).Followingthe issuance of SECY-88-147, the NRC staff,with input from DOE laboratories and consultants fromuniversities and industry, identified and prioritized sig-nificant technical issues in order to focus the researchefforts needed to close severe accident issues.

In August 1989, the staff published NUREG-1365, "Re-vised Severe Accident Research Program Plan." Thisplan was organized in two parts. One was oriented towardthe short-term resolution of issues related to early con-tainment failure (i.e., direct containment heating, BWRMark I liner melt-through) and methodologies to evalu-ate and analyze these phenomena (i.e., scaling analysis,severe accident codes). The second part of the SARP Planwas oriented toward providing long-term confirmatoryirnformation to support the assessments of a broad spec-trum of severe accident phenomenology associated withthe major elements of SECY-88-147.

The overall near-term goals of the plan were to providethe technical bases for assessing containment perform-ance over the range of risk-significant core melt eventsand to develop the capability to evaluate the efficacy ofgeneric containment performance criteria. The long-termgoals of the plan were to provide an improved under-standing of the range of phenomena expected duringsevere accidents and to develop improved methods forassessing: fission product behavior and release in theevent of containment failure during severe accidents.Since 1989, significant progress has been made in severeaccident research to warrant an update to NUREG-1365.The staff has prepared this SARP Plan Update to:

1. Identify those issues that have been closed or arenear completion,

2. Describe the progress in our understanding of im-portant severe accident phenomena,

3. Define the long-term research that is directed atimproving our understanding of severe accidentphenomena and developing, improved methods forassessing core melt progression, direct containmentheating, and fuel-coolant interactions, and

4. Reflect the growing emphasis in two additional ar-eas-advanced light water reactors, and support forthe assessment of criteria for containment perform-ance during severe accidents.

The focus of the SARP remains on regulatory research toaddress technical issues of concern regarding contain-ment performance and release of fission products in theevent of containment failure.

The priority for severe accident research depends onwhether there is a high likelihood that additional researchwill result in a significant reduction in uncertainty and onthe knowledge necessary for the conclusions to be widelyaccepted in the technical community. For some issues itmay not be necessary to reduce uncertainties further,since other issues may dominate overall uncertainty. Forthese and other issues for which it may not be practical toattempt to reduce uncertainties further, the staff recog-nizes that some regulatory decisions or conclusions willhave to be made with full awareness of existing uncertain-ties.

Table 1 provides a brief description of the program,status, and major milestones for each of 11 major severeaccident issues in the SARP. One of the goals of theSARP is to complete all the major severe accident experi-mental programs within the next 2 to 3 years. Assuming arelatively constant level of funding for the SARP, closureof all severe accident issues is anticipated in 4 years.

Major Accomplishments

Considerable progress has been made in recent years inunderstanding the underlying physical and chemical phe-nomena that can occur in a severe accident. The staffdeveloped programs consistent with the Integration Planfor Closure of Severe Accident Issues (SECY-88-147) toobtain key data and information needed to make an in-formed decision on issues and phenomenological uncer-tainties associated with accident sequences that couldpotentially lead to early containment failure. This infor-mation is essential for assessing potential safety improve-ments and for making decisions on whether or not par-ticular improvements are warranted. As pointed out inthe Commission's Severe Accident Policy Statement,such decisions should be based on a combination of engi-neering judgment (i.e., a deterministic method of settingand assessing safety margins) and the application ofprobabilistic risk assessment techniques to evaluate thelikelihood of the occurrence of low probability events.

A summary of the recent major accomplishments andfuture efforts under the SARP is given below.

1. Mark I Liner Failure

Completion of the initial study of BWR Mark I con-tainment shell failure was documented in NUREG/CR-5423, "The Probability of Liner Failure in aMark-I Containment" (July 1991). An extensive

E•-I NUREG-1365, Rev. 1

Executive Summary

peer review indicated that the methodology em-ployed to resolve the Mark I liner issue was soundand no major deficiencies or problems were identi-fied that would invalidate the results. The peer re-view also identified the need for confirmatory evalu-ation of a number of subjects. That follow-on work isunder way in fiscal year 1992.

2. Severe Accident Scaling Methodology

A severe accident scaling methodology (SASM) wasdeveloped to guide the formulation of experimentalprograms and analytical methods. Documentationof the SASM and application of the methods to di-rect containment heating (DCH) was addressed inNUREG/CR-5809, "An Integrated Structure andScaling Methodology for Severe Accidents Techni-cal Issue Resolution" (Draft for comment, Novem-ber 1991). Work on this issue is now complete. Ap-plication of the results of SASM to scaling andoperation of experimental facilities and modelingdevelopment for reactor analysis are pursued underthe direct containment heating (DCH) research pro-gram described in Section 2.1.

3. Source Term Issues

The NRC has sponsored numerous experimentaland analytical research projects on fission productrelease and transport. Early experiments and ana-lytical work tended to focus on release from fuel ma-terial under severe accident conditions. Later, ex-perimental data were obtained .on the behavior ofaerosols in reactor coolant systems and contain-ment. These data were used to develop aerosoldeposition and transmodels to analyze fission prod-uct behavior in the reactor coolant system (RCS) andcontainment. Currently, fully integrated model(VICTORIA) are in the process of being completedto analyze in-vessel release from fuel and retentionin RCS. The CONTAIN code is being developed toanalyze for the ex-vessel source term, including thetransport of fission products, condensation of va-pors, agglomeration and settling of aerosols, andchemical reactions in the containment.

Technical evaluations to support the revision to cur-rent methods for specifying the chemical form of theiodine entering the containment following a severeaccident (TID-14844, "Calculation of Distance Fac-tors for Power and Test Reactor Sites," March 1962)were completed. The Oak Ridge National Labora-tory performed analyses using a chemical kineticmodel to determine the equilibrium distribution ofthe iodine, cesium, hydrogen, and steam species en-tering the containment. The results indicate that io-dine entering the containment is at least 95% CsI(Cesium Iodide) with the remainder I (elemental io-

dine) or HI (hydrogen iodide). Once in the contain-ment, CsI is expected to deposit onto surfaces anddissolve in water pools forming I- (iodide) in solu-tion. Subsequently, iodine behavior within the con-tainment depends on the time and pH of the watersolution. If pH control is available and maintained,little of the dissolved iodine will be converted to ele-mental iodine.

One remaining area to complete the test series is re-lated to severe accident situations where air in-gressed into the core either by natural circulation orfrom the residual heat removal system will react ex-othermically with the cladding producing high tem-perature in the fuel. Large vaporization of ruthe-nium, tellurium, and molybdenum will occur.

4. Core-Concrete Interactions

The NRC has conducted an extensive program ofanalytic and experimental research to obtain an im-proved understanding of core-concrete interactions.In FY91, an interim version (CORCON MOD3) ofthe core-concrete interaction code, CORCON, wasreleased. The final version of CORCON MOD3 isexpected to be released by fall 1992. In addition toimproved thermal-hydraulic modeling, CORCONMOD3 incorporates the VANESA model of aerosolgeneration and radionuclide release during core-concrete interactions. Also, the NRC's experimen-tal program on core-concrete interactions has beencompleted. The remaining work on this issue is tovalidate CORCON MOD3 against the experimentaldata. This validation is currently underway.

5. Hydrogen Transport and Combustion

Computer codes on hydrogen transport and com-bustion have been developed to evaluate the staticor dynamic pressure loads from hydrogen combus-tion and detonation in containment. Results of theseevaluations should enable the staff to make regula-tory decisions to assess the potential threat to con-tainment integrity. Since combustion processes arecomplex, many aspects are still not well understood.For example, for a premixture of hydrogen-air-steam in containment at elevated temperatures, butbelow the auto-ignition temperature, flame accel-eration and high-speed combustion may lead to atransition to detonation. Research on some of thesephenomena is continuing. Although the current re-search is intended to reduce uncertainties, reduceduncertainties are not required in some cases to makeregulatory decisions. A joint agreement was reachedfor a cooperative program with the Ministry of Inter-national Trade and Industry of Japan and theNuclear Power Engineering Center. Under this pro-gram, high-temperature, high-speed hydrogen com-bustion research will be conducted for next 4 years.

NUREG-1365, Rev. 1 E-2

Executive Summary

6. TMI-2 Vessel Investigation Project

The objectives of the TMI-2 vessel investigationprogram are to investigate the condition and proper-ties of material extracted from the lower head of theTMI-2 reactor pressure vessel, to determine the ex-tent of damage to the lower head, and to determinethe margin of structural integrity that remained inthe pressure vessel. Significant progress has beenmade on these overall objectives. Examinations ofthe vessel samples from the TMI-2 lower head indi-cated that a small region of the lower head (approxi-mately 2 feet in diameter) experienced inner surfacetemperatures of about 1350 *K which exceeded thetransition temperature of the steel. The examina-tions also indicated that the temperature 2 inchesinto the wall was about 100 °K lower than the innersurface temperature. (The lower head thickness was5 inches.)

7 Direct Containment Heating

Since the publication of NUREG-1365, consider-able research on direct containment heating hasbeen undertaken to provide new insights and an im-proved data base to answer the questions, What isthe nature of the DCH threat, and what mechanismsand configurations exist ex-vessel that will mitigateor eliminate it? In order to quantify the pressure andtemperature loads generated by DCH, phenomenaand processes that mitigate or augment DCH loadsto the containment must be considered.

To assist the NRC and its contractors in developingan experimental program and interpreting and ana-lyzing test results, the staff convened a peer reviewgroup to evaluate the NRC's program to resolve theDCH issue. The peer reviewers meet regularly to as-sess progress and ensure that the research programobjectives are being met.

Integral testing was initiated in fiscal year 1991 to in-vestigate the containment loadings resulting fromDCH. The experimental program will explore inte-gral DCH phenomena at different scales for repre-sentative reactor designs. Separate-effects testing toconfirm the validity of the assumptions employed inthe scaling analysis was initiated in FY92. Details ofthe testing are provided in Section 2.1.

Major Ongoing ActivitiesOver the next several years, the major share of NRC'ssevere accident research program will focus on twophenomonological issues and on code development, vali-dation and assessment. These issues, core melt progres-sion, and fuel-coolant interactions and debris coolability,have significant uncertainties that impact our under-

standing and ability to predict severe accident phenom-ena and their effect on containment performance. Severeaccidents codes will continue to be developed to reflectthe current understanding of severe accident phenomenaand will be validated against experimental data. Decisionson when code development is completed require a bal-ance between the level of precision needed and the levelof uncertainty or variability in models of severe accidentphenomena that are acceptable for regulatory decisionmaking.

1. Core Melt Progression

In-vessel core melt progression describes the stateof an LWR reactor core from core uncovery up toreactor vessel melt-through in unrecovered acci-dents, or through temperature stabilization in acci-dents recovered by core reflooding. Melt progres-sion provides the initial conditions for assessingloads that may threaten the integrity of the reactorcontainment.

A great deal of information has been obtained on theprocesses involved in the early phase of melt pro-gression that extends through core degradation andmetallic (but not ceramic) material melting and relo-cation. This information has come from integraltests in the following test facilities: (1) the PowerBurst Facility (PBF); (2) the Annular Core ResearchReactor (ACRR); (3) the Canadian National Reac-tor Universal (NRU); (4) the Japanese NuclearSafety Research Reactor (NSRR); (5) the FrenchPhebus test reactor; (6) the Loss of Fluid Test(LOFT) facility; and (7) the German CORA ex-reactor fuel damage test facility. Laboratory sepa-rate effects experiments have also contributed infor-mation on significant phenomena. Most of theavailable information on late phase melt progressionhas come from the post-accident examination of theThree Mile Island, Unit 2 (TMI-2) reactor.

A comprehensive draft, "Research Plan for MeltProgression Issue Resolution," was prepared in fis-calyear 1991 to address the remaining core melt pro-gression issues. The plan was subjected to expertpeer review, and is currently being revised. This planand comments of the reviewers were used as a basisto develop the plan discussed in Section 2.2 of thisupdate.

Reactor pressure vessel lower head failure mapshave been developed for local penetration failureand for local and global creep rupture failure as afunction of the characteristics of the lower head de-bris and of the lower head structure. The analyticalbasis for these maps is supported by the results ofthe metallurgical analyses of the samples in theTMI-2 lower head program. A draft report,NUREG/CR-5642 "Light Water Reactor Lower

E-3 NUREG-1365, Rev. 1

Executive Summary

Head Failure Analyses," on this analysis was issuedfor comments in March 1992 and has undergonepeer review. The final report will be issued in thelater part of 1992.

2. Fuel-Coolant Interactions and Debris Coolability

There are three specific topics under this issue: fuel-coolant interaction (FCI) energetics, quenching inwater pools, and adding water to a degraded core.Because of the large variability of scenarios and pa-rameters that impact FCIs and debris coolability, thetask of closing this issue is difficult. The plan in-volves fundamental, long-range elements, as well asassessments applying these fundamentals to specificreactor designs and accident scenarios. Over the last10 years, the emphasis has gradually shifted from"energetics" aspects to "coolability" aspects rele-vant to accident management. However, the needfor assessing containment integrity for advancedlight water reactors (ALWRs) indicates that a bal-anced approach with continued research on ener-getic aspects is still warranted. These fundamentalenergetics aspects are being pursued at the Univer-sity of California at Santa Barbara and at the Univer-sity of Wisconsin. Results of this research are alsoexpected to contribute to the understanding of thecoolability experiments at FARO facility in Italy thatNRC is sponsoring.

Two experimental research programs addressing ex-vessel debris coolability were initiated during FY91.The Advanced Containment Experiment (ACE)program, conducted at Argonne National Labora-tory (ANL), is an internationally sponsored programwith NRC participation. One phase of this program,Melt Attack and Debris Coolability Experiment(MACE), deals with melt coolability issues. So far,three tests, including a scoping test, have been per-formed under the MACE program. While theMACE tests involve prototypic debris composition,a separate NRC-sponsored program WETCOR,conducted at Sandia National Laboratories (SNL),supplements the MACE program and investigatesthe coolability issue of metallic and oxidic core de-bris under heated side wall conditions to determinethe limits of coolability. Two tests were conducted inFY91 and in FY92 under the WETCOR program.

3. Severe Accident Codes

The principal integrated computer codes that sup-port the evaluations of nuclear plant responses topostulated severe accident scenarios are being inde-pendently reviewed for their adequacy to meet NRCobjectives. The first such comprehensive peer re-view was completed in FY91 for the MELCORcode; SCDAP/RELAPS peer reviews started in

FY92; and the CONTAIN and the VICTORIA codewill be undertaken in the second quarter of FY93.The results of the peer reviews will focus on criticalneeds for completing code development.

Advanced Light Water ReactorsResearch programs have been initiated to independentlyevaluate new ALWR (AP600 and SBWR) features, inparticular the adequacy of these features to withstandsevere accidents. These efforts will provide the technicalbasis for NRC's support of ALWR plant design certifica-tion.

Long-Term PlanEven though we anticipate that closure of all major severeaccident issues will be accomplished in 4 years, additionalwork will continue on severe accident research; althoughat a reduced level of effort. Residual issues will still needto be addressed. In addition, results of severe accidentexperiments and research that are being conductedworld-wide will be used to continue to update and validatethe severe accident codes. Severe accident research onALWRs will continue as designs for these plants continueto evolve. It is likely that additional severe accident issuesmay arise in the future. Therefore, the NRC will continueto support severe accident research, at a somewhat re-duced level, in order to improve our technical under-standing and maintain the level of expertise needed toaddress future issues in this area.

Criteria for Termination of Research*The degree to which a severe accident technical issuemust be resolved depends on the needs of the relatedregulatory decisions and on the necessary knowledge forthe conclusions to be widely accepted in the technicalcommunity. Ideally, an issue is considered closed whenthe NRC can pronounce that sufficient experimental andanalytical research has been completed to allow, for thepurpose of IPEs (individual plant examinations), accidentmanagement studies, or containment performance evalu-ations, for the prediction of reactor plant response. Inaddition, uncertainties have been reduced sufficientlythat regulatory decisions are either insensitive to furtherreductions in uncertainty or that the residual level of riskis considered low enough that further work is not justifi-able.

In practice, deciding when work is completed requires asubjective assessment of the potential benefits of furtherresearch. Although these judgments are not easy, twodifferent approaches are possible. The first approach isbased on developing the analytical capability for predict-ing the complete evolution of severe accidents (undergiven initial and boundary conditions) in some reasonable

NUREG-1365, Rev. 1 E--4

Executive Summary

level of detail. The other is focused on particular proc-esses, relevant to specific containment integrity issues,assuming that these processes can be adequately charac-terized within a rather general assessment of the overallsequence. Accordingly, the emphasis in the first approachis on computer code development at the system level (i.e.,MELCOR, CONTAIN, SCDAP/RELAP) and associ-ated code validation or verification activities. In the sec-ond approach, the emphasis is on addressing specificphysical processes and through them identify specificsafety concerns, then focus the research to address thespecific concern. An example of the second approach isthat taken for the assessment of Mark-I liner attack inNUREG/CR-5423.

These two approaches are not, certainly, mutually exclu-sive; the two approaches will work well together. There isalready a lot of momentum on the first approach; proto-typic testing has been used to validate computer models,which in turn would be exercised over the range of condi-tions and uncertainties associated with the data and acci-dent analysis. In fact, this approath may be even neces-sary for a widely available capability for severe accidentassessments. However, it is also true that such use is only

possible after the main issues have been resolved, and aresearch effort that is driven by this approach tends to berather inefficient in addressing the issues themselves.

The second approach provides for self-correcting focus-ing on specific technical issues (such that they are well-posed for the research program), and it provides for grad-ual synergism and eventual convergence. Thisconvergence constitutes closure of an issue and therebyprovides the basis for terminating further research. Weare expecting to use such an approach on all containmentintegrity issues. This latter approach on issues relating tocontainment integrity and will temper the approach bycognizance of the issue significance in terms of its overallrisk impact.

An important element in the issue resolution process isthe role of peer review. For all of its major programs, theSARP routinely incorporates the feedback from peer re-view groups both: in the area of experimentation and incode development. The breadth of experience and exper-tise brought to bear on the problem by the review groupsassists both the staff and its contractors in achieving clo-sure.

E-5 NUREG-1365, Rev. 1

z

C0

Table 1 SARP Status and Milestones

Expected ReferenceIssue Description Status Completion Section

Major Program

Mark I Liner Develop probabilistic method to integrate SARP results Complete Appendix A.1Failure into conditional liner failure probability. Results indicate (NUREG/CR-5423)

that without water, liner failure is nearly certain; withwater, liner failure is physically unreasonable.

Residual Issues: 2.6

1. Liner failure criteria Complete2. Melt superheat Complete3. Melt spreading analysis Complete4. Recalculation of the liner failure probability Anticipated December 1992

Major Program

Direct Contain- Develop analytical models for predicting entrainment ofment Heating core debris from reactor cavity and its interaction with

the containment atmosphere.1. Integral-effect tests to simulate fundamental Ongoing December 1992 2.1.

synergetic effects of high-pressure melt ejection.2 Separate-effect tests to confirm rate or detailed Ongoing February 1994 2.1

spatially dependent information employed in thescaling analysis (e.g., corium jet disintegration, liquidfilm formation, entrainment, capture of corium incompartment).

3. Development and validation of system-level code Ongoing December 1992 2.1, 2.4

Residual Issues:

Development of generalized assessment criteria for Anticipated February 1994 2.1different cavity and compartment designs employedin LWRs.

Major Program

Scaling Develop severe accident scaling methodology to guide Complete Appendix B.3the experimental program. (NUREG/CR-5809)

Residual Issues:

None

Table 1 SARP Status and Milestones (continued)


Major Program

Source Term Obtain better understanding of fission product release and Complete Appendix B.1transport mechanisms in LWRs under severe accidentconditions.

Residual Issues:1. Complete analysis of V16 fission product release test Ongoing September 1992 2.7

conducted in FY922. Complete VICTORIA code development Ongoing September 19923. Validate source term codes with Phebus data Anticipated 19994. Complete the fission product release data for severe In planning September 1994

accident associated with air ingression condition

Major ProgramCore-Concrete Obtain better understanding of the amounts of noncon- Complete Appendix B.2Interaction densable gas generated from the interaction between

molten core debris and the concrete and radioactive aerosols.

Residual Issue:

Compare code predictions to test results. Ongoing January 1994 Appendix B.2

Major Program

Hydrogen Develop data base and computer codes to assess the con- Complete Appendix A.3Combustion tainment performance due to hydrogen combustion. Dataand transport were obtained on subsonic and sonic combustion modes.

Data were also obtained on hydrogen transport and mixing.

Residual Issues:Data obtained to date is limited to hydrogen-air-steam Ongoing 1996 2.6mixtures at ambient or low temperature. Obtain betterunderstanding of the combustion phenomena of hydrogen-air-steam mixtures at high temperature represen-tative ofsevere accident conditions.

z

Z

CD

z Table 1 SARP Status and Milestones (continued)


Major Program

TMI-2 Vessel 1. Extract specimens from the TMI-2 vessel Complete Appendix A.4Investigation 2. Investigate the condition and properties of material Ongoing June 1993Project extracted from the lower head of the TMI-2 reactor,

determine the extent of damage, and determine themargin to failure.

Residual Issues:

None

Major Program

Core Melt Provide initial conditions (melt mass, rate of release, coin-Progression position, and temperature) for assessing containmentand Hydrogen performance; describe the state of LWR core from coreGeneration uncovery up to reactor vessel melt-through.

1. Early phase: core uncovery, clad ballooning, oxida- Complete Appendix A.2tion and hydrogen generation, eutectic materialinteractions, Zircaloy relocation.

2. Late phasea. Conditions for formation of metallic melt block- Ongoing 1995 2.2

age in BWR vs. metallic melt drains from corewhen formed.

b. Conditions for ceramic pool melt-through from Ongoing 1995 2.2blocked core; determines the melt mass andcharacteristic of the melt released to the reactorvessel lower head.

3. Mode of vessel failure; determine rate of melt Complete 2.2release into containment.

Residual Issues:1. Application of experimental results to models of Ongoing 1996 2.4

governing phenomena in severe accident codes.2. Model development for severe accidents recovered Ongoing 1993 2.2

by core reflooding.

!.rj0



Fuel-CoolantInteractionsand DebrisCoolability

Major ProgramDevelop models or correlations and the technical bases forthose models for predicting the broad range of resultantphenomena of interactions of degraded core materials withcoolant (including steam and hydrogen generation rates)for various reactor geometries, meltdown scenarios, andtiming of coolant addition.1. FCI energetics-relevant to in-vessel interactions in

PWR.a. Premixing experiments to determine the extent of

fuel and coolant intermixing that causes the regionto become pressurized.

b. Fragmentation experiment to determine the rapidincrease of fuel surface area causing vaporization ofmore coolant and increasing local pressurization.

c. FCI yield experiment to determine the expansionwork associated with the triggering and explosionof premixture.

2. Fuel melt quenching experiment to determine theconditions under which FCI leads to coolableconfiguration as occurred at TMI-2.

3. Reflooding-relevant to in-vessel and ex-vessel issuesrelated to operator adding water to achieve stablecoolable configuration.a. In-vessel reflooding issues-includes recriticality

and hydrogen generationb. Ex-vessel reflooding to mitigate core-core

interactions

Appendix A.5

Ongoing

Ongoing

Ongoing

Ongoing

September 1992

September 1992

September 1993

January 1994

2.3

2.3

2.3

2.3

2.3

2.3

Complete

Ongoing March 1993

Residual Issues:The models or correlations devel oped under the FCI research

Z program will be validated against available experimental data.Perform selective plant analyses.

0•

LA

Anticipated December 1993

z0ý



Major Program

Severe Accident 1. Develop, validate, and document computer codes to Ongoing December 1994 2.4Codes analyze severe accident phenomena and issues for light

water reactors. The code development program is anongoing iterative process; codes are developed andassessed against selected experimental data; modelimprovements are identified; new models for newphenomena are implemented and assessed.

2. Extensive assessment and validation against experimental Ongoing 1999results from domestic and international programs willcontinue to increase the confidence in the codes.

Major Program

Advanced Light 1. Assess the methodologies used to evaluate containment Ongoing September 1992 3Water Reactors cooling concepts, including natural circulation cooling

and external spray cooling of containment.2. Assess the methodologies used to evaluate the effects Ongoing December 1992 3

of phenomena resulting from ALWR. fuel design onsevere accidents.

3. Apply severe accident codes (e.g., SCDAP/RELAP, Anticipated 8-12 months following receipt 3MELCOR, CONTAIN) to the analysis of ALWRS. of vendor's design information

Residual Issues:

None.

0D

Cn

1 SEVERE ACCIDENT RESEARCH PLAN

1.1 Background

In May 1988, the staff presented to the Commission anIntegration Plan for Closure of Severe Accident Issues(SECY-88--147). The Integration Plan consists of six ma-jor elements:

1. Examination of existing plants for severe accidentvulnerabilities (individual plant examinations).

2. Development of generic containment performanceimprovements (CPI) with respect to severe acci-dents to be implemented if necessary for each of thesix containment types.

3. Upgrading of staff and industry programs to improveplant operations.

4. A severe accident research program (SARP).

5. A program to define how and to what extent vulner-abilities to severe accidents from external eventsneed to be included in the severe accident policyimplementation.

6. A program to ensure that licensees develop andimplement severe accident management programsat their plants.

All these elements are mutually supportive and interre-lated, but the severe accident research program is thecommon source of information, and hence of central im-portance. Elements 1, 2, and 6 particularly depend on ourunderstanding of severe accident phenomena and phe-nomenological sequences, and it is the purpose of thisprogram to provide this understanding.

Following the issuance of SECY-88-147, the staff estab-lished several groups of experts to help identify and de-fine the status of SARP activities (item 4 above). Eachgroup was composed of contractor personnel from DOElaboratories, consultants from universities and industry,and NRC staff. These groups considered detailed techni-cal issues to define their status and the need for furtherresearch, to identify and focus research necessary forsound regulatory decisions to be made within the frame-work of the integration plan, and to prioritize the researchactivities needed to close severe accident issues. The staffused the experts' input as a basis for preparing the revisedSARP plan that was published in August 1989 asNUREG-1365, "Revised Severe Accident Research Pro-gram Plan." This plan was structured in two parts: oneoriented to the short-term resolution of certain pressingissues, and the other oriented to the long-term confirma-tory support, and perhaps refinement, of current assess-

ments in a broad-spectrum-coverage of the whole severeaccident phenomenology. The specific issues identified inthe short-term part of the plan were: Mark I liner attack,direct containment heating, severe accident codes, andscaling-the last two being primarily of methodologicalthrust.

As stated in NUREG-1365, the overall goals of the re-vised SARP plan are to provide the technologicaI base forassessing containment performance over the range ofrisk-significant core melt events, develop the capability toevaluate the efficacy of generic containment performanceimprovements, provide an improved understanding of therange of phenomena exhibited by severe accidents, andreduce the uncertainties in the source term sufficiently toenable the staff to make regulatory decisions on severeaccident issues. Since the issuance of NUREG-1365, theSARP has generated a large amount of data and insightsinto the progression of severe accidents. According to therevised SARP plan, the research efforts in the past 3 yearsfocused on the short-term issues listed above; however,significant progress was also made'on certain elements ofthe long-term plan, including in particular corium-con-crete interactions and the chemical form of radioiodine insevere accidents. In addition to the improved understand-ing of severe accidents achieved in the past 3 years, thepresent plan (referred to as SARP plan update) was moti-vated by the following considerations:

1. Identify those issues that have been closed or nearcompletion,

2. Describe the progress in the maturation and evolu-tion of our understanding of important severe acci-

dent phenomena,

3. Define the long-term research that is directed atimproving our understanding of severe accidentphenomena and developing improved methods forassessing core melt progression, direct containmentheating, and fuel-coolant interactions, and

4. Reflect the growing emphasis in two additional ar-eas-advanced light water reactors, and support forthe assessment of criteria for containment perform-ance during severe accidents.

1.2 DiscussionAs in NUREG-1365, the focus of the SARP remains onresearch efforts to address technical issues of concernregarding the health and safety of the public. Technicalissues regarding containment performance are of particu-lar regulatory significance. The two issues identified inNUREG-1365 that could potentially lead to early

I NUREG-1365, Rev. 1

1 Severe Accident Research Plan

containment failure are the Mark I liner attack and directcontainment heating (DCH). At this time, the Mark Iissue is near resolution; a plan to bring closure to theDCH issue will be in place by the end of CY92. Withrespect to the long-term research, it is important here toachieve a definition that will produce the sought-afterunderstanding within a reasonable time schedule. At thesame time, it is important that this understanding is di-rectly relevant to safety to adequately support elements 1,2, 5, and 6 of the Integration Plan. Carrying out thelong-term research plan requires a lot of judgment and anatmosphere of open communication among all partiesinvolved. It also requires clear communication and a de-liberate approach in planning, reviewing, and conductingresearch. One purpose, then, of this plan is to address themethodological and practical aspects of meeting theserequirements in future research efforts.

Consistent with this philosophy, the staff has proceededwith the implementation of the revised SARP, makingextensive use of debate and deliberation of expert peerreview to address a good number of issues. The expertreviews include the technical program group on severeaccident scaling methodology and review panels for coremelt progression research, for direct containment heatingexperiments, and for severe accident codes. In addition,the Mark-I liner attack study (NUREG/CR-5423) wassubjected to extensive peer review and to smaller special-ized panels for follow-up activities. These activities haveproven very fruitful in sharpening the staff's judgmentconcerning the practical aspects of implementing the re-vised SARP and will be continued in the implementationof this SARP update.

1.3 Benefits of Additional Research

The phenomena under investigation and their combina-tion into severe accident scenarios are highly complex-usually involving multiphase or multicomponent interac-tions, rapid transients, and multidimensional behavior.Since full-scale experiments are not possible, the in-depth understanding presents quite a formidable task. Onthe other hand, experience has shown that dominantmechanisms and evaluation methodologies can be foundto provide an understanding quite adequate for the in-tended purposes. The special challenge in planning andconducting the severe accident research plan is to identifyand use effective approaches, focusing quickly and study-ing in depth all the essential aspects of the behavior. Infact, a continuing focusing and refocusing process is re-quired, as new research results provide the basis for im-proved judgments as to where to expend future efforts.

Which severe accident phenomena are studied, how con-clusions about these phenomena are reached and de-fended, and when research on specific phenomena andissues should be stopped have traditionally been highly

charged areas of debate. The impact of improved under-standing in a certain area of phenomenology on risk re-duction is not always readily apparent. One example iscore melt progression research, which is one of the mostcomplicated technical areas in severe accident phenome-nology. Core melt progression impacts hydrogen genera-tion (timing and quantity) and core debris relocation intothe lower plenum (timing, quantity, rate, and compositionof debris). The latter aspect, in turn, impacts fuel-coolantinteraction (FCI) during and after the relocation (ener-getic, coolability), thermal loading on the lower head(mode and timing of failure), and release of core debrismaterials to the containment. Eventually there are im-.portant implications for containment integrity (i.e., ener-getic FCI, DCH or Mark I liner attack). As NUREG/CR-5030, "An Assessment of Steam-Explosion-InducedContainment Failure" (February 1989), and NUREG/CR-5423 have shown, conservative treatment of uncer-tainties is feasible to yield bounding and acceptable re-sults for the alpha-mode containment failure and Mark Iliner attack issues, respectively. For these two issues, themain benefit of learning more about core melt progres-sion phenomena will be related primarily to improvedunderstanding of the margin of conservatism in the analy-ses to resolve these issues.

From a containment integrity standpoint, learning moreabout core melt progression provides the input for con-tainment loading for DCH. If DCH were amenable to thesame approach as the alpha-mode containment failureand containment liner attack issues (see section 2.1), thevalue of core melt progression research would be in im-proved knowledge in the area of accident managementwhen core melt progression is considered in the presenceof water as it would in a recovered scenario. The payoffwould be in improved estimates of risk, even though theaccident scenario is of exceedingly low probability and therisk is already low, if we can demonstrate that accidentscan be terminated even at relatively advanced states oftheir progression. The excellent safety record of nuclearpower plants leaves only such exceedingly low probability,high consequence events to be of concern. Again, eventhough the accident sequence leading to vessel failure isof low probability, lack of adequate understanding of coremelt progression leads us to use conservative assumptionsto provide answers to a set of rather intangible questionsthat we are often asked. For example, for ex-vessel debriscoolability, a core debris coolability criterion of 0.02 m2/MWt was proposed. This criterion is intended to be usedin sizing the reactor cavity for ALWRs. The existing database for sustained melt-concrete-coolant experimentssuggests that debris depth and the formation of crust onthe surface of the melt may inhibit water ingression intothe melt, which is necessary to produce rapid cooling. Ifmore realistic assumptions are used, the debris bed depthwould be substantially smaller than is currently investi-gated experimentally (75%-100% of total core). The coremelt progression research can provide valuable

NUREG-1365, Rev. 1 2


information on the amount of molten material that canenter the containment at the time of vessel breach.

As was the case when NUREG-1365 was developed, thestaff recognizes that for some issues it may not be practi-cal to attempt to reduce uncertainties further, and someregulatory decision or conclusion will have to be madewith full awareness of existing uncertainties. Expert elici-tation (as done in NUREG-1150) in the Level II PRAserves only to reveal rather than to resolve issues since itdeals with phenomenological issues at a high level. Quiteoften experts rely on parametric codes in which theauthor has programmed his opinion of what happens; realimportant phenomena may never be discovered. Further-more, the degree of judgment, and therefore the degreeof confidence associated with it, depends on the particularissue and the nature of the phenomenology involved. Thepriority and manner in which severe accident research isconducted depend on whether there is a high likelihoodthat such research will result in a significant reduction inrisk uncertainty and on the depth necessary for the con-clusions to be widely accepted in the technical commu-nity.

This research plan is structured to examine the issues indepth and develop the necessary methodological toolsthat can soundly address these issues. Therefore, one ofthe goals of the SARP is to complete all the major severeaccident experimental programs within the next 2 to 3years. Results of severe accident experiments and re-search that are being conducted world-wide will be usedto continue to update and validate the severe accidentcodes.

1.4 Criteria for Termination ofResearch

The degree to which a severe accident technical issuemust be resolved depends on the needs of the relatedregulatory decisions and on the necessary depth for theconclusions to be widely accepted in the technical com-munity. Ideally, an issue is considered closed when theNRC can pronounce that sufficient experimental andanalytical research has been completed to allow, for thepurpose of IPEs, accident management studies, or con-tainment performance evaluations, for the prediction ofreactor plant response. In addition, uncertainties havebeen reduced sufficiently that regulatory decisions areinsensitive to further reductions in uncertainty.

In practice, deciding when work is completed requires asubjective assessment of the potential benefits of furtherresearch. Although these judgments are not easy, twodifferent approaches are possible. The first approach isbased on developing the analytical capability for predict-ing the complete evolution of severe accidents (under

given initial and boundary conditions) in some reasonablelevel of detail. The other is focused on particular proc-esses, relevant to specific containment integrity issues,assuming that these processes can be adequately charac-terized within a rather general assessment of the overallsequence. Accordingly, the emphasis in the first approachis on computer code development at the system level (i.e.,MELCOR, CONTAIN, SCDAP/RELAP) and associ-ated code validation or verification activities. In the sec-ond approach, the emphasis is on addressing specificphysical processes and through them identifying specificsafety concerns, then focusing the research to address thespecific concern. An example of the second approach isthe one taken for the assessment of Mark-I liner attack inNUREG/CR-5423.

These two approaches are certainly not mutually exclu-sive; the two approaches work well together. There isalready a lot of momentum on the first approach; proto-typic testing has been used to validate computer models,which in turn would be exercised over the range of condi-tions and uncertainties associated with the data and acci-dent analysis. In fact, this approach may be necessary for awidely available capability for severe accident assess-ments. However, it is also true that such use is possibleonly after the main issues have been resolved, and aresearch effort that is driven by this approach tends to berather inefficient in addressing the issues themselves.

The second approach provides for self-correcting focus-ing on specific technical issues (such that they are wellposed for the research program), and it provides for grad-ual synergism and eventual convergence. This conver-gence constitutes closure of an issue and thereby providesthe basis for terminating further research. We expect touse such an approach on all containment integrity issues.

An important element in the issue resolution process isthe role of peer review. For all of its major programs, theSARP routinely incorporates the feedback from peer re-view groups both in the area of experimentation and incode development. The breadth of experience and exper-tise brought to bear on the problem by the review groupsassists both the staff and their contractors in achievingclosure.

1.5 Organization of the Report

Our intent is to make this research plan scrutable; thus,when pertinent technical details cannot be given by refer-ence, they are included in appendices at the end.

Chapter 2 points out the main directions for future severeaccident research, and Chapter 3 discusses planned re-search efforts for ALWRs. Appendices A and B of thisreport discuss all the major phenomenological areas ofsevere accidents and progress to date on these areas.

3 NUREG-1365, Rev. 1


This report has benefited from reviews by the AdvisoryCommittee on Reactor Safeguards, Nuclear Safety Re-

search Review Committee, and the Office of NuclearReactor Regulation.


2 RESEARCH PLAN

IntroductionThe programs described in this chapter of the updatedSARP are the high priority items that need to be ad-dressed in the next few years, namely

1. Core melt progression,2. Direct containment heating,

3. Fuel-coolant interactions and debris coolability, and

4. Severe accident codes

This chapter presents the research needs, our currentresearch programs, and anticipated results for each of theabove issues. Research programs on other residual issuesare also discussed in this chapter. Appendices A and Bpresent a brief summary of the current state of knowledgeon the core melt progression and fuel coolant interactionissues listed above as well as other severe accident issues.The focus of this research is to provide validated codes toassess the performance of nuclear power plants with se-vere accidents and to gain a more quantitative under-standing of the overall probabilities of core damage andfission product releases. NUREG-1150 and other PRAshave shown that events that lead to early containmentfailure (such as several hours after severe core damageprogression starts) have the greatest risk significance. Asthe accident progresses through its various stages, various.kinds of containment loads are a consequence of theprevalent material interactions. The intensity of theseloads depends on the quantities, composition, and tem-perature of the material involved and the overall configu-ration of contact.

While the NUREG-1150 study did not produce quantita-tive measures of the risk importance of specific melt pro-gression phenomena, the results show that both the mag-nitude and uncertainty of melt processes are important.For example, the observation that the risk significance ofDCH has decreased relative to earlier studies can in partbe attributed to an improved understanding of melt pro-gression that led to improved DCH testing procedures.Further exploration of the NUREG-1150 informationbase together with a sensitivity study has produced quan-titative information on the risk importance of melt pro-gression phenomena.'

Several of the conclusions from this study are:

1. In-vessel issues directly influence the low contain-ment failure probabilities and the low risk valuesobtained for Surry (depressurization of the primary

'Letter from Fred Harper, SNL, to Brian Sheron, NRC, dated Janu-ary 24, 1992.

system and arrest of core damage before vesselbreach). The regression analyses would not revealthis sort of insight.

2. Although the ,regression analysis had limitations,and care should be taken in interpreting the results,the regression analysis showed that the followingin-vessel issues are important to the uncertainty inNUREG-1150 results: the fraction of each radionu-clide group in the core released to the vessel beforevessel breach,. pressure rise in the containment atvessel breach (implies that pressure of the reactorcoolant system (RCS) at the time of vessel breach,the mode of vessel breach, and the fraction of thecore released at vessel breach are important), in-vessel hydrogen production, and alpha-mode fail-ure.

3. The probability of failure of the Surry containment islow. However, within the conditional containmentfailure probability that does exist, the contributionsfrom different modes of vessel breach are extremelyvaried-the conditional probability of containmentfailure at vessel breach for vessel failures at highpressure is several orders of magnitude higher thanthe conditional probability of containment failure atvessel breach for vessel failures at low pressure. Thisindicates that in-vessel issues could be critically im-portant to containment failure probability at plantsthat are more susceptible to overpressurization thanSurry.

4. The magnitude of source term releases at Surry aresignificantly impacted by the mode of vessel breach.The risk of prompt fatalities from a high-pressuremelt ejection mode of vessel breach and an earlycontainment failure is several orders of magnitudehigher than the risk of prompt fatalities from vesselbreach at low pressures with an early containmentfailure. This difference is due to the high rate ofaerosol generation from the molten debris during ahigh-pressure melt ejection. This risk at Surry is low,but could indicate a high source term sensitivity tothe mode of vessel breach at the other plants.

5. In-vessel issues are especially important from anaccident management point of view. The timing ofvessel breach is important for planning operatoractions. Also, during.recovery stages, the predictedresponse of the molten core to water injection wouldbe useful information for the operator to have.

In summary, the evidence in NUREG-1 150 indicates thatin-vessel melt progression issues are important to risk andthe uncertainty in risk.


2 Research Plan

In general, core melt progression research provides im-portant input for all containment failure issues, such asthe amount, composition, and temperature of the meltreleased to the containment and the mode of vessel fail-ure. In addition, since accident management is consideredto be a practical measure for mitigating the consequenceof severe accidents, an understanding of degraded corecooling and other phenomena related to fuel-coolant in-teractions is needed to assess accident managementstrategies. Sections 2.2 and 2.3 provide detailed NRCresearch programs to address these issues.

At the time the revised SARP was issued in 1989, theissues of high-pressure melt ejection and DCH had beenthe subject of considerable study over the preceding sev-eral years. The results of these previous studies increasedconcerns about these issues, and it did not then appearlikely that further research could lead to their resolution.Recent understanding of the core melt progression, thecompletion of the lower head failure analysis program,the completion of the severe accident scaling methodol-ogy, and restructuring of the DCH experimental programare all contributing significantly to the near-term closureof this issue. Section 2.1 provides the detailed NRC re-search program to address the direct containment heatingissue.

Section 2.4 discusses the codes being developed andmaintained by NRC. The MELCOR code has become theNRC-supported systems analyses code that replaced thesource term code package. A few detailed mechanisticcodes will also be maintained that are applicable to spe-cific severe accident phenomena.

Sections 2.5,2.6, and 2.7 discuss the BWR Mark I contain-ment liner, hydrogen combustion, and source term issues,respectively.

2.1 Direct Containment Heating

2.1.1 Research Needs

The ultimate needs from research on DCH are analyticalmodels or correlations and the technical basis for thosemodels for predicting the entrainment of core debris fromthe reactor cavity and its interaction with the containmentatmosphere and blowdown gas during a high-pressuremelt ejection.

Thus, in addition to predicting the dispersal of debrisfrom the reactor cavity, DCH modeling must also becapable of evaluating or suitably accounting for:

1. Debris fragmentation-the surface area for heattransfer and reactions, translated into a particle di-ameter

2. Debris to gas heat transfer-the embedded debrisvelocities, residence time

3. Debris trapping-the capture of debris within sub-compartment, volumes or on intervening structures

4. Hydrogen generation-resulting from oxidation ofmetallic debris constituents

5. Hydrogen combustion-resulting from reaction ofany preexisting hydrogen in the containment alongwith hydrogen generated during the high-pressuremelt ejection

6. Water vaporization-as a consequence of debris-water interactions in the cavity or other containmentregions.

Ideally, modeling should also address parameters, proc-ess times, and spatial dependencies as well as their rela-tionship to, in the SASM parlance, control parameters(e.g.; RCS pressure).

While not strictly related to DCH, the precursor proc-esses that influence melt ejection from the reactor vessel,gas blow through, and hole ablation are also necessarycomponents of the overall calculation.

Modeling of DCH must also reflect the effects of thediversity of plant designs insofar as plant-specific featuresinfluence the phenomena. There are some 18 differentcavity and lower containment subcompartment designs inuse in the U.S. commercial PWR nuclear power plants.

Past research has indicated that while the cavity geometryinfluences the extent of dispersal from that region atelevated RCS pressure, differences in design configura-tion (at least for the designs experimentally investigated)have less impact on the entrainment fraction. As part ofthe SASM technical program group efforts to develop ascaling methodology for DCH testing, a correlation forentrainment fraction (i.e., the fraction of debris dispersedfrom the reactor cavity) was developed and assessedagainst a range of experimental data at different scales(1/42, 1/25, 1/10 linear scales) for different designs (Surry,Sizewell, Watts Bar, Zion). This work suggests that theentrainment fraction can be correlated against a numberof parameters with a notable dependence on the initialEuler number, and thus, the pressure ratio defined by theinitial RPV pressure/cavity pressure. Figure 2.1.1, takenfrom NUREG-5809, illustrates the dependence, as pre-dicted by the correlation of the entrainment fraction onthe initial RPV to cavity pressure ratio for two differentdesigns, Zion and Surry.

The general conclusion that debris dispersal from thecavity is nearly complete at elevated RCS pressures hasbeen qualitatively demonstrated in many of the past DCHexperimental programs, regardless of the direct

NUREG-1365, Rev. 16 6

2 Research Plan

oh

Entrainment Fraction, y

.I I --' 0

-Q

&4

C

a0

y.

0

Lii

-II

,, 01 C,,

- &I I j I I IU' • o 0110.

CC

0 - .N C "CO' 4

990 0 O..A 00 - • ILi - 0 00 I0

Entrainment Fraction

Figure 2.1.1 Entrainment Versus Pressure Ratio for Steam Blowdown.


2 Research Plan

scalability of individual tests. While this preliminary con-clusion needs to be confirmed by additional testing andscaling analysis, the results of work to correlate entrain-ment fraction culminating in the plant extrapolationshown in Figure 2.1.1 indicate that research is needed toaddress the other major processes, e.g., subcompartmenttrapping and water interactions, which may have a signifi-cant effect on DCH loadings. Furthermore, correlation ofentrainment fraction data indicates that testing need notrepresent the full range of possible RCS pressures (up tothe power-operated relief valve setpoint), since completedispersal out of the cavity is potentially achievable atmuch lower pressures. Interestingly, the work to correlateentrainment fractions also suggests that relatively simplemodels, or even quantitative empirical criteria, may besubstituted in lieu of detailed mechanistic models.

While the details of the cavity configuration, for the lim-ited range of variations considered, have been shown tohave less effect on debris dispersal at elevated pressures,it is intuitive and has long been argued that compartmen-talization and structures downstream of the cavity wouldtrap or deentrain debris from the flow stream exiting thecavity on its way to the bulk containment. Since the freevolume of the lower containment compartmentalized re-gion of the containment is a small fraction of the totalcontainment volume (-15% in the case of Zion) and thecompartmentalized region immediately adjacent to thecavity exit is yet a smaller fraction of the total volume, it isapparent that trapping of debris in those regions wouldsignificantly decrease the ability to rapidly heat, by debris-gas heat transfer, the bulk atmosphere.

In addition to trapping or deentrainment of disperseddebris, the other first-order mitigative effect on the high-pressure melt ejection (HPME) process is the potentialinteraction of debris with water, either in the cavity or inthe containment. Large quantities of water present in thecontainment during certain severe accident sequencescan potentially quench or cool debris and thereby reducecontainment pressurization.

While direct containment heating research has princi-pally been concerned with the basic mechanisms by whichmolten debris could give up its latent and sensible heat tothe atmosphere, a significant contribution to the pressuri-zation of the containment arises from the potential oxida-tion of the metallic constituents of the core debris and thesubsequent combustion of any hydrogen generatedthrough that reaction. Assuming roughly 15% (by mass)of the core debris is unreacted zircaloy cladding, the co-rium thermal energy (sensible plus latent heat) is roughlyequivalent to the chemical energy, on the order of 1 MJ/kg. Energy released from the combustion of the hydrogenproduced by oxidation of that unoxidized cladding is alsoequivalent to roughly 1 MJ/kg of melt. If the hydrogen

generated is consumed at the same rate at which it isreleased and mixed into regions containing sufficient oxy-gen, .no hydrogen accumulates and the resulting combus-tion mode is a diffusion flame, either a buoyant plume orjet. Similarly, another requirement for this scenariowould be the presence of an ignition source, hot debrisparticles in the case of an HPME, or the release andmixing of hydrogen at temperatures above an autoignitiontemperature. If the hydrogen is released at a rate fasterthan which it can be mixed with oxidant and burned or ifthere is no mechanism for initiating combustion uponinitial release, the hydrogen may accumulate and thesubsequent combustion, which can be presumed to occurafter eventual mixing of gases to combustible levels, maytake the form of detonation.

In order to provide the answers to the research needsdescribed above and resolve the DCH issue, the NRC hasdeveloped a research program composed of three majorelements, all of which are judged necessary to achieveclosure: (1) integral testing, (2) separate effects testing,and (3) development and validation of a systems-levelcode (CONTAIN).

As described in Appendix B.3, "Scaling," a necessaryprecondition for the resumption of DCH testing was thedevelopment of a scaling rationale to guide the design andoperation of planned tests. It was further determined thatscaling evaluations would be based on the evaluation ofconditions for a specific accident sequence and for a spe-cific plant design in order to assure a more scrutable andmeaningful comparison of the scaling results with reactordesign and analysis,

The starting point for this scaling evaluation of appropri-ate initial conditions was the technical program group(TPG) review of existing core melt progression analysesfor a station blackout at the Surry plant in order to assessthe amount of core debris present in the lower head at thetime of reactor pressure vessel failure. Three synthesizedsets of initial conditions for DCH were developed; a set ofconditions associated with a penetration failure of thebottom head and two sets of conditions (for both lowpressure and high pressure) associated with creep failureof the bottom head. The quantities, compositions, andamounts of material involved are summarized in Fig-ure 2.1.2., taken from NUREG/CR-5809. The evalu-ation of melt conditions for a station blackout at the Surryplant yielded the following general insights.

* About 40% of the core mass would be ejected inmolten form.

" From 30 to 70% of the molten material could be in ametallic form. If failure of the crust or flow blockagefrom structural weakness is considered more realis-tic than crust or flow blockage melting, about 30 to


2 Research Plan

-L CANDLING, FUEL CANDLING,oT FORMATION LIMITED BLOCKAGE

AND FAILURE OF GRIDFAILURE PLATE

FUEL CANDLING,LIMITED BLOCKAGEFAILURE OF GRID

PLATE

INITIAL CONDITIONSTIME OF FAILUREEJECTED KGSMOLTEN KGSOXIDIZED ZREJECTED METALLIC

CONTENTTEMPERATURE

-3 HOURS-40,000-40,000-40%

-30-40%-2500 K

-4 HOURS-70,000-40,000-40-50%

-40-60%*Z2500 K

-10 HOURS-80,000-40,000-60-70%

-50-70%*-2500 K

*INCREASED METALLIC CONTENT DUE TO MOLTEN STEEL

Figure 2.1.2 Initial Material Conditions for Direct Containment Heating.


2 Research Plan

50% of the molten material will be in a metallicform; (In both instances, the majority of the metalliccomponent is steel; Zr is 10-25% of the moltendebris.)

0 A significant amount of solid debris (over one-half ofthe molten mass) will be present in the lower headfor scenarios in which vessel failure is delayed untilcreep failure occurs. The majority of the solid debrismay be retained in the vessel, at least in the timescale related to HPME. The mass of molten materialfor a creep failure scenario was virtually identical tothat associated with a penetration failure.

To proceed with testing and validation of analytical mod-els, the staff has assumed initial conditions associatedwith a station blackout sequence and a penetration failureof the reactor pressure vessel (RPV) bottom head. It isfully recognized that there is uncertainty surrounding theprecise initial conditions for DCH; nonetheless, for pre-scribing scaled test conditions and conducting integraltests, we conclude that the best available informationindicates adopting this approach. Simulation of creep fail-ure, or for that matter multiple penetration failures, canbe addressed by variation of the hole size used for testing.

2.1.2 Direct Containment Heating- CurrentResearch Program

As summarized in section 2.1.1, the staff has outlined aresearch program consisting of (1) integral testing,(2) separate effects testing, and (3) analytical model de-velopment and validation.

The objective of integral testing is to simulate the funda-mental synergetic effects of high-pressure melt ejectionthrough experiments that include entrainment orsweepout of corium from the cavity, transport and trap-ping of debris in the lower containment subcompart-ments, oxidation of metallic constituents in the corium,combustion of hydrogen produced as a result of that oxi-dation of metals, and vaporization or heat transfer to theexisting water inventory in the reactor cavity and contain-ment. This integral testing is. designed to investigate boththe mitigative processes and phenomena that lessen theimpact of an HPME as well as the inherent mechanismscontributing to the pressurization by DCH.

In terms of scaling, we sought to simulate the pressuriza-tion of the containment to the extent practicable withoutintroducing distortions that significantly impact the simu-lation of the dominant phenomena influencing the funda-mental behavior. As an example, tests conducted at lessthan full scale will produce conditions in which trans-ported debris has a shorter residence time in the atmos-phere; if debris velocities are preserved, a test at 1/10scale would result in a much shorter residence time fordebris to heat the atmosphere. Thus, preservation of de-

bris-gas heat transfer in scaling may lead one to distort oreven neglect principal mitigative mechanisms in an inte-gral test. When those conflicts arose, scaling was driven bythe objective of simulating the transport of debris to thebulk containment as opposed to direct simulation of pres-surization. As a check on the possible distortion of phe-nomena, including concomitant atmosphere pressuriza-tion, counterpart integral tests using the same scalingprinciples are run at different scale.

Following the basic objective and guidelines for integraltesting outlined above, the simulated RCS and contain-ment features and volumes were geometrically (i.e., line-arly) scaled, the RPV aspect ratio as well as the lengthscale from the RPV to the cavity was preserved. In keep-ing with our objective to simulate fundamental or globalbehavior in a consistent fashion and as a consequence ofusing a simulant material (iron-alumina thermite withchromium metal) rather than reactor materials, the meltmass employed in the tests was scaled based on the equi-librium energy/volume ratio. Because of the lower den-sity of simulant relative to corium, scaling on an energy/volume basis results in a distortion of the amount of melt,volumetrically scaled, by approximately 48%. The meltmass employed in the 1/10 scale representation of Zionwas 43 kg.

The staff has convened an independent peer review groupto evaluate the program to resolve the DCH issue andspecifically assess the program of integral testing. Withthe concurrence of the peer review group on our basicapproach, the staff has proceeded to conduct integraltests at Sandia National Laboratories (SNL) in the 1/10thlinear scale Surtsey facility and at Argonne NationalLaboratory (ANL) in the 1/40th linear scale COREXITfacility. The integral experimental tests currently underway at SNL and ANL are considerably different from thetests proposed prior to the development of the SASM.Specifically, these tests are improved because (a) the in-itial and boundary conditions for the tests are scaled forspecific accident scenarios for specific p' mnts, (b) impor-tant scaling groups have been matched or the distortionshave been minimized for those tests in which matchescannot be achieved, (c) scoping tests have been conductedand instrumentation and procedures required to carry outHPME/DCH experiments are reliable and reproducible,and (d) test conditions include more prototypic conditions(i.e., steam-driven melts, realistic containment com-partmentalization, sources of water, potential for hydro-gen combustion, etc.). Integral effects tests are based ondetailed modeling of the Zion and Surry plants, designsthat are representative of two large classes of reactordesigns. Scaled models of the reactor pressure vessel,reactor cavity, in-core instrument tunnel, and subcom-partment structures were constructed. The subcompart-ment structures and equipment modeled included thecrane wall, steam generators, reactor coolant pumps, seal

NUREG-1365, Rev. 1 10

2 Research Plan

table room, biological shield wall, refueling canal, radialbeams and grating, and the operating deck.

On February 14 and June 1, 1992, the staff reconvenedthe DCH peer review group with the purposes of (1) de-scribing the progress and results of the initial counterparttests at 1/10 and 1/40 scale, (2) discussing preliminaryconclusions and observations regarding scale effects, and(3) proposing the balance of our integral test program.Table 2.1.1 presents that portion of our integral test pro-gram proposed to be carried out at SNL. For complete-ness, tests that have already been conducted are listedalong with the data of the test. Note that Table 2.1.1identifies tests at a third and larger 1/6th scale that areproposed to be conducted. These tests would be per-formed using the existing Containment Technology TestFacility at SNL that has been used in the past to investi-gate containment integrity (containment leakage and fail-ure beyond design pressure). As evident from the testmatrix, the approach in integral testing is systematic; ma-jor synergetic effects are added in the progression oftesting. For example, in order to experimentally measurehydrogen generation directly and to measure and observethe incremental effects of hydrogen combustion, the firstintegral test was conducted in an inert (99% N2) atmos-phere with later tests (IET-3,4,5,6) allowing combustionwith an oxygen-bearing atmosphere. In similar fashion,the effects of cavity water inventory and containmentwater heat removal are included in the test program.Proposed counterpart testing in the 1/40 scale COREXITfacility at ANL was originally designed to mirror theSurtsey tests through IET-7. Other options under consid-eration include use of the ANL facility to explore issuesrelated to the use of a simulant melt-the ANL facilityhas the capability for conducting tests with U0 2 -basedthermitic material. At this stage three tests have beenconducted in the 1/40 scale facility, counterparts toIET-1, IET-3, and IET-6.

Relative to the research needs identified in subsection2.1.1, the integral tests provide direct indication of thefraction of debris swept from the cavity (entrainmentfraction), subcompartment trapping, hydrogen genera-tion, and hydrogen combustion. Indirect information isderived for water vaporization, debris-to-gas heat trans-fer, and debris fragmentation. In the case of debris frag-mentation, posttest analysis provides a direct measure ofrecovered debris, however, this is not a measure of thedebris particles upon inception of entrainment nor is it ameasure of characteristic debris geometry or surface areafor melt that can not be recovered in particulate form(because of agglomeration of molten or semi-molten ma-terial).

While the integral tests provide valuable information andallow for a global evaluation of DCH, because of thedifficulties associated with use of high-temperature reac-

tive melts (instrumentation limitations) and their addedcomplexity, these tests do not provide the most efficientmechanism for determining rate processes or detailedspatially dependent information. Though integral testsprovide a posttest measurement of entrainment fraction,there is no ready means to measure entrainment rateswithin the cavity or to determine the characteristic proc-ess by which debris is swept from the cavity. Integral testsalso do not provide for measurement of the characteristicparticle size range associated with the entrained debris inthe cavity. Resolution of subissues related to HPME andDCH, such as the location and rate at which material isoxidized, depend in part on this information. Further-more, the examination of rate parameters provides forthe confirmation of behavior observed over a time inter-val, e.g., measurement of an entrainment rate allows theconfirmation of an entrainment fraction.

In order to obtain the general information identified inthe preceding discussion, the NRC has initiated a sepa-rate effects program to be conducted at Purdue Univer-sity. Six specific phenomena will be studied in detail:

1. Corium jet disintegration immediately after dis-charge

2. Liquid film formation upon impact of the jet3. Entrainment and drop formation by streaming gas4. Liquid film transport from pressure and shear5. Liquid film ejection and disintegration6. Capture of liquid mass in compartment

Each phenomenon will first be studied experimentallyusing the preliminary scaling parameters. The proposedmechanism and model for each phenomenon will be sys-tematically tested at more realistic scales than the originaldata base. A preliminary study indicates that a 1/10 linearscale experimental facility for the separate effects tests isadequate when water and woods metal are used assimulant material. This is also attractive since this scale iscompatible with the SURTSEY integral test facility atSandia. Flow visualization and detailed instrumentationare possibleby using the simulant fluid pairs. Either ofthem is quite difficult or impossible in the integral facilityusing more prototypic high-temperature material. Theexperimental data are then used to verify or scale-up theproposed correlations. For this purpose, several basicexisting models and correlations, as well as the ones de-veloped by previous NRC programs, will be systematicallyexamined. The study will be carried out by a detailedmechanistic modeling study, together with visual observa-tions through high-speed photography and the data fromthe local instrumentation.

Overall, the separate effects testing and model assess-ment is projected to be complete within 2 years, withcompletion of the first phase of testing, using a water-airsystem, scheduled for the end of calendar year 1992.

11 NUREG-1365, Rev. 1

2 Research Plan

Table 2.1.1

Test

IET-

IET-2A

IET-3

Test Initial Conditions

Melt Mass: 43 kg Fe oxide/Al/Cr thermiteDriving pressure: 7.1 MPa (IET-1), steamHole size: Tube ejection with ablation, 3.5 cmAtmosphere: inert, N2Cavity water: condensate level, 0.9 cmContainment water: none

Thermite temperature measurement

Atmosphere: reactive, 0.2 MPa air/N2The rest of the initial conditions are the same as IET-1.

Repeat of IET-1 except driving steam pressure of 6.2MPa

Containment water: condensate levelThe rest of the initial conditions are the same as IET-3.

Repeat of IET-2

Atmosphere: 0.2 MPa air, C0 2, H 2The rest of the initial conditions are the same as IET-4.

Table

2.1.1

Target Date

9/13/91

11/1/91

12/13/91

lET-1R

IET-4

2/7/92

3/20/92

IET-2B

IET-5

4/28/92

5/13/92

IET-6 Atmosphere: reactive, 0.2 MPa air, N2, H2 (scaled) 6/18/92The rest of the initial conditions are the same as IET-3.

IET-7 Atmosphere: reactive, 0.2 MPa air, N2, H 2 (scaled) 7/09/92The rest of the initial conditions are the same as IET-4.

IET-8 Cavity water: 1/2 fill.with water 7/30/92

The rest of the initial conditions are the same as IET-3. \

Conversion to Surry geometry and subcompartment 9/92

IET-9 Melt Mass: #kg (scaled) - Fe oxide/Al/Cr thermite 9/24/92Driving pressure: 13 MPa steamHole size: Tube ejection with ablation, 3.5 cmAtmosphere: reactive, 0.16 MPa, air/steam/H 2 (scaled)Vessel without annular gapCavity water: condensate level (scaled)Containment water: condensate level (scaled)

IET-10 Counterpart to IET-9 in 1/6 scale CTI facility 10/15/92.Scale parameters accordingly.

IET-11 Atmosphere: reactive, 0.16 MPa, air/steam/H 2 (scaled) 11/12/92Vessel with annular gapThe rest of the initial conditions are the same as IET-8.

IET-12 Counterpart to IET-11 in 1/6 scale CTT facility 12/10/92Scale parameters accordingly.

NUREG-1365, Rev. 1 12

2 Research Plan

The third major element of our DCH research program isthe development and validation of a systems-level code,i.e., the CONTAIN code. The need for a systems-levelcode, and specifically CONTAIN to assess the conse-quences of a high-pressure melt ejection, reflects severalconsiderations.

1. The diversity of reactor designs and the range ofpossible accident conditions leading to an HPMEpreclude the experimental confirmation of plant re-sponse for all reasonable combinations. Therefore,some modeling approach is necessary to address thepotentially wide range of outcomes.

2. A highly complex detailed analytical treatment ofDCH by a multidimensional finite difference or fieldcode is not practical, given the capabilities of avail-able models (e.g., KIVA, CONCHAS-SPRAY,HMS, COMMIX) and the effort required to applythese codes even in their present incomplete form.Furthermore, application of such codes at this stagewould introduce the additional uncertainty associ-ated with general validation of the codes themselves.

3. HPME is only one phase in the overall progressionof a severe accident that proceeds to vessel failure athigh pressure, but the resulting DCH has an influ-ence on the subsequent plant response by alteringcontainment atmosphere conditions from whichlater phenomena, e.g., long-term pressurization orcore-concrete interactions, will influence contain-ment performance. Thus, calculation of the inte-grated containment response to a severe accidentover the entire course of the event dictates that asevere accident code serve as the vehicle for theassessment. CONTAIN, the NRC's severe accidentcontainment response code, is the logical choice formodeling. As a practical consideration, past effortsto model DCH have enabled the NRC to modify thecode with a reasonable level of effort.

4. Interpretation and analysis of integral tests requirean analytical model capable of calculating the syner-getic effects associated with DCH. Analysis of theintegral tests with the tool to be used in plant analy-sis is also essential to most directly address thescalability of test results and models.

The principal challenge in applying the CONTAIN codeto DCH is to incorporate models that are sufficientlymechanistic to predict the influence of major plant andaccident parameters, yet are consistent with the overallengineering, correlational modeling used in CONTAIN.It is also necessary to allow some flexibility in modelinggiven present uncertainties.

In accordance with our program objectives, the NRC hasinitiated efforts at SNL to quantitatively assess, imple-

ment, and document CONTAIN models for RPV andcavity phenomena, relying wherever possible on past ac-tivities including the SNL preparation of a DCH modelsand correlations document and those tasks of the TPG inwhich model assessment was undertaken. Given the rela-tive abundance of modeling approaches that have beenproposed in the past, no new model development is beingundertaken; the individual model selected will be madefollowing expert peer review of these models.

For RPV phenomena, models will be incorporated thatare consistent with those models recommended by theTPG and used by SNL in scaling the integral tests. For thecavity processes such as entrainment rate, a variety ofmodels will be included that reflect the variety of pro-posed modeling approaches (Whalley-Hewitt, Ricou-Spalding, Kataoka-Ishi correlations). Incorporation ofmore than one model for a process allows for a compara-tive assessment of model adequacy without a significantpenalty since most correlations rely on similar variablescalculated internally in CONTAIN. The implementationand initial assessment of CONTAIN DCH models isscheduled to be completed by the end of CY 1992.

2.1.3 Anticipated Results

Our planned research on direct containment heating is.directed to produce generalized modeling, incorporatedinto the CONTAIN code, capable of predicting the con-sequences of a high-pressure melt ejection for the rangeof applicable accident conditions. The modeling approachwill have been validated against available integral andseparate effects tests results. Scaling issues will have beenaddressed by comparison and analysis of test results forexperiments conducted under this program at three dif-ferent scales, as well as experimental data generated fromother research. Selected analysis of full-scale reactorplant analysis will be performed to gauge the expectedresults for some subset of plant and accident conditions.Plant analysis will be performed in part during the pro-gram as an iterative process in order to assess the de-mands for accuracy in modeling, i.e., to determine howgood or accurate an answer is needed and the effect onplant response to variations in modeling.

Another expected outgrowth of this research will be thedevelopment of generalized assessment criteria that maybe used in lieu of more detailed analysis for a more spe-cific set of accident and plant parameters. Such assess-ment criteria can serve as the basis for evaluating theadequacy of specific calculations based on other modelsand can serve as the basis for the development and assess-ment of simpler models adopted for use in full plantsevere accident codes such as MELCOR and MAAP.

While our planned research is anticipated to provide forresolution of the issues that have long been associatedwith DCH, it is important to note that our research pro-

13 NUREG-1365, Rev. 1

2 Research Plan

gram will not fully resolve questions surrounding the phe-nomena related to corium-water interactions that mayresult from the ejection of debris at high pressure into adeep water pool that may be in the cavity in some designsin certain severe accident sequences. While our testingprogram includes a very limited investigation of the ef-fects of cavity water inventory for the accident conditionssimulated, a complete resolution of this general issue, tothe extent it bears on plant performance, relies on ourresearch program on fuel-coolant interactions.

2.2 Core Melt Progression Research

In-vessel core melt progression describes the state of anLWR reactor core from core uncovery up to reactor ves-sel melt-through in unrecovered accidents, or throughtemperature stabilization in accidents recovered by corereflooding. Melt progression provides the initial condi-tions for assessing the loads that may threaten the integ-rity of the reactor containment. The significant results ofmelt progression are the melt mass, rate of release, com-position, and the temperature (superheat) of the meltreleased from the core and later from the reactor vessel atvessel failure. Melt progression provides the in-vesselhydrogen generation and the conditions that govern thein-vessel release of fission products and aerosols and theirtransport and retention in the primary system. Melt pro-gression also provides the core conditions for assessingaccident management strategies.


Understanding core melt progression phenomena, in-cluding an assessment of uncertainties, is important indetermining the margin of conservatism for those severeaccident issues that have been resolved (e.g., BWR MarkI liner failure) and in resolving the remaining severe acci-dent issues (e.g., DCH). The objective of the research onmelt progression is to provide quantitative tools for as-sessing the mass and other characteristics of the meltreleased from the core and from the reactor vessel, as wellas hydrogen generation and the conditions for fissionproduct release, transport, and retention. This research isfocused on four general issues or technical areas:

1. Determining whether a metallic blockage of the core(or core plate), similar to that which occurred atTMI-2, would occur in BWR accidents, or alterna-tively, Whether the metallic melt drains from thecore (and core plate) when formed. The core block-age or melt drainage branch point and the differentmelt progression sequences for the two differentpathways are shown in Figure 2.2.1.

2. The conditions for ceramic pool melt-through froma blocked core that determine the mass and other

characteristics of the melt released from the coreinto the lower plenum.

3. The mode of vessel failure that determines the rateof melt release into the containment and the timingof the release.

4. The application of experimental results in models ofthe governing phenomena for severe accident safetyassessments. These models are also to be used, usu-ally in simplified form, in the severe accident sys-tems codes SCDAP/RELAP5 and MELCOR.

If during the course of this melt progression research theneed arises for special research results outside the aboveareas of focus, for example, specific material properties,then such research will be undertaken after a peer reviewof the need. A strong effort has been made in planningthis research program to focus on the key areas of uncer-tainty in issue resolution, and not to try and cover a largelist of unprioritized technical uncertainties.

The research needs in each of these areas are discussed inthe remainder of this section.

Whether a metallic blockage develops across the coreduring a severe accident, as occurred at TMI-2, or themetallic melt (and later the ceramic melt) drain promptlyto the lower plenum from the core (and BWR core plate)has a major impact upon subsequent melt progression, asshown in Figure 2.2.1. The mass, rate of release, andother characteristics of the ceramic melt released fromthe reactor core into the vessel lower plenum impact theoverall consequences of the accident. Analysis indicatesthat in BWR accident sequences with automatic depre-ssurization, the blowdown lowers the water level belowthe core (and core plate) so that core heatup occurs underdry core (negligible steam flow) conditions. All the avail-able experimental data on melt progression are for wetcore conditions (i.e., water in the bottom of the core). Nodata currently exist for the dry core conditions that wouldresult from the automatic depressurization that is calledfor by the emergency operating procedures for U.S.BWRs. Analysis indicates that core heat up, particularlythe rapid oxidation transient, and metallic melting, relo-cation, and blockage formation are significantly differentfor dry core and for wet core conditions.

The current experimental and analytical research pro-gram to determine whether core blockage occurs or doesnot occur under BWR dry core accident conditions isdiscussed in Section 2.2.2. Analytical modeling of theprocess of melt relocation and blockage formation isneeded to provide a tool for extending the experimentalresults to the full range of risk-significant severe accidentconditions for both BWRs and PWRs.

The second research area concerns the characteristics ofthe ceramic melt that drains from the core (and BWR

NUREG-1365, Rev. 1 14

2 Research Plan

Rapid HydrogenGenentlon - PWR

slow lHydringmGenwaion - BWR

%As'

mie Uft nroeMing flmItr ..Cor Gore ftlt______ _____ II

II C~uws EmitDralningIOrncoremlpas I- i

Ip 1

Mht-Water Intsr- - oneLower Plenum Bolfs Dry I

"e-Mae intem-lowsI Lmwe Plenum 3.11. Dry IVemse FailureI

Figure 2.2.1 Core Melt Progression Sequence Showing Blockage or Drainage Paths.

15 NUREG-1365, Rev. 1

2 Research Plan

core plate) into the lower plenum in a blocked core acci-dent. The significant melt characteristics (mass, rate ofrelease, temperature (superheat), and composition(metal content) are largely determined by the thresholdand location of melt-through of the metallic core block-age (and secondary ceramic crust) by the growing ceramicmelt pool. These characteristics are very important inassessing the impact of core-melt accident sequences oncontainment loads. Most of the information currentlyavailable on the late (ceramic melt) phase of melt pro-gression, including blockage melt-through, has comefrom the TMI-2 core examination. This information,however, is not sufficient to validate modeling of themelt-through threshold and location and the characteris-tics of the released ceramic melt.

Ceramic pool melt-through from a blocked core is awhole-core phenomena and is not amenable to directexperimentation. Therefore, experimental researchneeds to be conducted on separable key parts of theproblem with integration of the results by analysis andcodes. For example, the surface heat flux distribution ofthe internally heated ceramic melt pool that results fromnatural circulation in the pool is an important phenome-non in determining the meltthrough threshold and loca-tion in blocked-core accidents. An assessment will beundertaken to determine whether or not correlations de-rived from the existing data at lower Rayleigh numbersand in different geometry than expected in a core meltaccident are adequate for characterizing the surface heatflux distribution of the ceramic melt pool. The researchprogram to determine the threshold and location of melt-through is discussed in Section 2.2.2.

The third research area concerns the mode and timing ofvessel failure in core melt accidents. A primary question iswhether, for a given reactor type and accident scenario,vessel failure is by penetration failure or by global or localcreep rupture failure. These different failure modes givevery different results on the rate of melt ejection from thevessel and also on the failure threshold. The research onlower head failure models reported in NUREG/CR-5642, "Light Water Reactor Lower Head FailureAnalyses," is discussed in Section 2.2.2.

The fourth area concerns severe accident systems codes,such as SCDAP/RELAP5 and MELCOR. These codesmodel the key physical processes that occur during in-vessel melt progression, often in simplified form, in orderto determine the global response of the reactor during acore melt accident. Detailed phenomenological modelsare also used to perform analyses of the experiments inorder to gain a firm understanding of the important physi-cal processes. This understanding forms the foundationfor the simplified models in the severe accident codes.The detailed phenomenological models and the thoroughanalyses of the experiments form the basis for assessingthe adequacy of the simplified models in the systems

codes. Section 2.4 presents the NRC research to developand validate severe accident codes.

While it is recognized that appropriate scaling analysesare important to ensure that the experiments are per-formed under the proper conditions and ,in the properparameter range for application to full-scale reactor acci-dents, it may be that a specific issue is not completelyamenable to scaling analysis. In these cases, experimentswill not be used to develop empirical correlations, butrather to identify important dominant physical/chemicalprocesses and assure that they are properly representedin the analytical models. Analytical models of the keyphenomena are also necessary for the results to be appli-cable over the range of severe accident conditions.

The primary current development activities in the analy-sis of melt progression processes involve the DEBRISporous media late-phase melt progression model and theMERIS model of metallic melt relocation and blockageformation that is an extension of the porous media frame-work of DEBRIS. Rather than trying to describe thedetailed geometry changes during melting, melt reloca-tion, and the freezing and remelting of crusts, DEBRISand MERIS use a much simpler porous media frameworkthat keeps track only of material relocation and state(liquid, solid, composition, temperature). DEBRIS pro-vides a 2D(r,z) mechanistic treatment of the key phenom-ena involved in late-phase (ceramic melt) melt progres-sion in blocked core accidents (Refs. 1,2). Specifically,DEBRIS treats the melting dynamics of a particulateceramic debris bed supported by a metallic crust acrossthe fuel rod stubs in the lower region of the core. Themodeling includes ceramic crust formation around themelt pool and pool melt-through of the supporting crustsystem and melt drainage from the core into the lowerplenum. MERIS uses a 3D(xy,z) formulation to treat thecomplex BWR core geometry and a time-varying aniso-tropic effective permeability to account for melt reloca-tion and freezing (Ref. 3). DEBRIS and MERIS respec-tively, have shown good agreement with the very limitedexperimental data currently available, on ceramic and onmetallic melt relocation and crust formation.

An assessment is needed of the ability of the DEBRISporous-media model to predict the late-phase melt pro-gression behavior in the TMI-2 accident. An assessmentis also needed of DEBRIS with the experimental datafrom the melt progression (MP) experiments in ACRR onthe relocation and failure dynamics of the ceramic andmetallic crust system that supports the growing ceramicmelt pool. Also needed is an assessment of the MERISporous media type model of metallic melt relocation andblockage formation with the experimental results of theplanned ex-reactor experiments on this subject and withany other relevant available information. If the assess-ments of DEBRIS and MERIS, with possible improve=ments, are successful, their modeling, probably in simpli-

NUREG-1365, Rev. 1 16

2 Research Plan

fied form, can be incorporated into SCDAP/RELAP5and MELCOR.

2.2.2 Current Research Program

As described in Appendix A.2, a great deal of informationhas been obtained on the processes involved in the earlyphase of melt progression that extends through core deg-radation and metallic material melting and relocation.Except for the BWR experiment in the Canadian NRUreactor, which has been delayed because of a leak in theheavy water system of NRU, current NRC research onmelt progression is concentrated on two major uncertain-ties or issues. The first issue is the BWR accident condi-tions, if any, in'which a metallic core blockage, similar tothat in the TMI-2 accident, would not be formed. This isthe major branch point in the melt progression sequenceshown in Figure 2.2.1. The second issue concerns theconditions for the meltthrough of the growing pool ofceramic melt that is supported by the metallic blockage.Research on the mode of vessel failure in core melt acci-dents, which in general is plant-and accident-sequence-specific, is limited to the validation of the models devel-oped and presented in NUREG/CR-5642. Issues relatedto severe'accident core melt codes are discussed in Sec-tion 2.4.

2.1.2.1 Core Blockage or Melt Drainage in BWRAccidents

An ex-reactor experimental program has been started toresolve the question of metallic melt blockage or drainageunder BWR dry core accident conditions. Unlike the pre-vious ACRR and CORA BWR experiments, the ex-reactor experiments do not start from the initial intact rodgeometry. Instead, the ex-reactor experiments accuratelyreproduce the core conditions at the onset of metallicmelt relocation, and the system behavior under these welldefined conditions is then measured.

Since internal heat generation in the metallic melt isnegligible in reactor accidents, these experiments can beperformed ex-reactor without internal heat generationwith pours of metallic melts of prototypic compositionand flow rates (dribbles). The test assemblies have proto-typic dimensions, materials, structures, heat capacities,and temperature distributions. Because of the axial tem-perature distribution, illustrated in Figure 2.2.2, blockageformation is only possible in the lower quarter of the coreand in the core plate region. The metallic melt comesfrom the upper three-fourths of the core and the metallicpour represents that part of the core in these experi-ments. The ex-reactor experimental apparatus representsthe lower quarter of also the core and also the core-plateregion. The pour of metallic melt represents the incomingmetallic melt from the upper three-quarters of the core togive overall full-length simulation. The mass, flow rate(dribble), and composition of the metallic melt poured

into the experiment test sections are obtainedby analysisof the flow of metallic melt from the upper three-quartersof a full-scale BWR core under dry core accident condi-tions. This analysis is consistent with the phenomenologyobserved in the ACRR DF-4 and CORA BWR tests.These tests were performed, however, for BWR wet coreconditions. For most of the tests, there is pre-oxidation ofthe Zircaloy (and stainless steel) surfaces by furnace-heating in steam to provide oxide film thicknesses that arerepresentative of operating BWRs. The initial tempera-ture distribution in the core and core-plate structure isproduced by preheating the system with downward flow-ing argon. There is compensation for radial heat losseswith external heaters on the insulation of the test assem-bly. A detailed description of these experiments and theirobjectives is given in Reference 4.

A cross-section of the test fuel bundle is shown in Fig-ure 2.2.3. This has full-scale radial simulation of half achannel box unit cell in a BWR core. This simulationincludes the' control blade in the gap between channelboxes, the roughly equal horizontal length of unbladedgap that is potentially important for melt drainage, and anadditional two rows of fuel rods outside the gap. Theseadditional rods provide a prototypic temperature distri-bution in the channel box walls, in the gap, and in thecontrol blade, and they also provide a volume for diver--sion of the melt flow following channel box failure.

This simulation has 64 fuel rods, channel box walls, andbladed and unbladed sections of the gap between thechannel box walls. In initial operation of the experimentand for debugging, it is planned to perform two XR1experiments in the simplified geometry of the blade, gap,and channel box walls shown in the upper schematicdrawing of Figure 2.2.3. These experiments will also pro-vide basic information on melt relocation flows beforechannel box failure and on the effects of pre-oxidation onthe interaction with and failure of the channel box wallsby the control blade melt.

The XR2 experimental test assemblies will include a fullradial-scale mockup of a corresponding half unit cell inthe core plate region below the core that includes theprototypic complex flow paths, materials, and heat ca-pacities, as well as simulation of the interior section of theBWR channel boxes with fuel rods and of the core platestructure below the core. The complexities of the struc-tures and flow paths in the core plate region are shown inFigure 2.2.4. The core plate itself occupies only 34% ofthe horizontal cross-section in the core plate region, andmost of this cross-section is relatively open with complexflow paths defined by thin, low-heat capacity structures oflow-melting stainless steel. Because of these geometricalcomplexities, rivulet flow complexities, and the effects ofeutectic interactions, analysis of the potential forblockage or drainage of eutectic metallic melts of control

17 NUREG-1365, Rev. 1

2 Research Plan

'00000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 0000000000000000 00000000

1600 k0000000 00000000

100000000 '000000000000000 000000600

1400 --------------------------------------- --0000000 000000000000 000 00000000000 000 0000000000 000 0000000000 0000 00000000

0~0

Ea,8 0 0 . . . ..-- --- ---- --- --- ---- ----- ---- ---.- - --

600 -- .- .--------------------- .-- .------------

400,• "

0 1 2 3 4

Axial Location, m

Figure 2.2.2 Axial Temperature Profiles for the Control Blade, Channel Box and Fuel Rods, IndicatingMaximum Lateral Temperature Variations

NUREG-1365, Rev. 1 18

2 Research Plan

Channel- Box Wall

ZircaloyWedgeDeflector

XR1 BundleControlBlade

FuelRod

Simulators

ChannelBox

Walls

BypassFlow

Region

ZlrcaloyWedge

Deflector

nControl Blade

XR2 Bundle

Figure 2.2.3 XR1 and XR2 Test Bundle Cross Sections.

19 NUREG-1365, Rev. 1

2 Research Plan

Relocation PathwayThrough Canistersbypassing theLower Core Plate

- SPACERGRID

TIE PLATE

NOSE PIECE

FUEL SUPPORT-PIECE

I CORE PLATE

CONTROLSBLADE DRIVE

. GUIDE TUBE

Figure 2.2.4 Melt Drainage Pathways Through the Fuel Canisters Which Bypass the Lower Core Plate.

NUREG-1365, Rev. 1 20

2 Research Plan

blade materials and Zircaloy in the core plate regionneeds guidance from experimental data. It is planned toperform the XR2 experiments in a steam environment.

Analysis with MELCOR has indicated that two separaterelocations of metallic melt occur under BWR dry coreconditions, the first a melt of B4 C and stainless steelcontrol blade materials with a melting (eutectic) point ofabout 1500°K, and the second a melt of cladding andchannel box Zircaloy with a melting point of about2200'K. The two simplified gap-geometry XR1 experi-ments will use a single pour (dribble) of control blademelt. The XR2 experiments, which include the fuel rodregion within the channel boxes and the core plate regionbelow the core, will use two separate pours, one withcontrol blade material melt with some dissolved Zircaloyand the other with a higher temperature Zircaloy melt.The melt mass, composition, temperature, pour rate, andthe time of separation between the two pours in the XR2experiments will be determined by analysis and by theresults of the dry core test soon to be performed in theGerman CORA ex-reactor fuel damage facility.

The effect of the pre-accident oxide film on the coreZircaloy (and stainless steel) from normal reactor opera-tion in delaying material interactions and eutectic forma-tion will be investigated in the XR1 experiments. Theseresults will be factored into the XR2 experiments. Addi-tional data on this effect will also be also obtained fromCORA experiments. Current plans are for nearly all theXR2 experiments to be pre-oxidized.

The test matrix for the ex-reactor experiments has beenformulated to determine, with as few tests as possible,whether there are any conditions within the range of thegoverning parameters (including uncertainties) for BWRdry core accident conditions under which metallic meltdrainage, rather than core or core plate blockage, canoccur. The governing parameters are the metallic melttemperatures, the core axial temperature gradient, andthe presence of an initial oxide film on the Zircaloy struc-ture corresponding to that in normal BWR operations.

After the two initial simplified XR1 experiments to checkout the experimental techniques and to determine theeffect of pre-existing oxide films on the channel box Zir-caloy, the first of the four planned XR2 experiments willbe performed with all the above parameters at the end oftheir range that favors melt drainage. If core blockagedoes occur under these conditions, it will then have beendetermined that the blocked core sequence is to be ex-pected in all dry core BWR accidents. The experimentalmatrix would then be shortened to focus on the character-istics of the metallic blockage formed under the bestestimate conditions. This is needed for assessing the dura-bility of the blockage with continued core heat up and alsofor assessing blockage melt-through by the growing ce-ramic melt pool in late-phase melt progression. If meltdrainage does occur in the initial XR2 experiment, thepreliminary test matrix given in Table 2.2.1 will be fol-lowed step by step, increasing the potential for blockageto see if blockage is achieved under these dry core condi-tions.

The conditions for this test matrix will be reviewed asexperimental results are obtained, particularly thosefrom experiment XR2-1. Currently it appears that theseexperiments and corollary analyses with additional inputfrom the CORA BWR tests and from the BWR FLHT-6test in the NRU reactor will provide sufficient data toresolve the question of metallic melt blockage or drainagefor the relevant range of BWR dry core accident condi-tions. The last of a series of full-length tests on fueldamage and melt progression during coolant boildown(wet core) is scheduled to be performed in the CanadianNRU reactor in late 1992. The results will provide uniquelength-effect data for BWR core geometries. Two high-burnup fuel rods in the 14-rod test assembly that is shownin cross-section in Figure 2.2.5 will provide unique data onfission product release in the boron-containing BWR coreenvironment. Post-irradiation examination (PIE) of thetest fuel bundle should also provide some information onthe effects of high-burnup fuel on early-phase melt pro-gression.

Table 2.2.1 Test Matrix for the Ex-Reactor Experiments on Metallic Melt Relocation and Blockage FormationUnder BWR Dry Core Accident Conditions

Test* Melts* Grad T** Melt T Zry Oxide Objectives

XR1-1 C.B. Prototypic Prototypic Yes Oxide FilmXR1-2 C.B. Prototypic Prototypic No No Oxide FilmXR2-1 C.B., Zry High High Yes Favor DrainageXR2-2 C.B., Zry High Low Yes Decrease Melt TXR2-3 C.B., Zry Low Low Yes Decrease Grad TXR3-4 C.B., Zry Low Low No Delete Zry Oxide

1C.B. means control blade materials plus some dissolved Zircaloy.*High grad T means that the AT occurs over a short distance, while low grad T means that the same AT occurs over a long distance.

21 NUREG-1365, Rev. 1

2 Research Plan

Zr-2.5% Nb LOOP LINER TUBEZIRCALOY-2 PRESSURE TUBEOUTER TUBE

INNER TUBE

SADDLE

BYPASS FLOW ANNULUS

INSULATION (ZRO2)

SHROUD LINER

BUNDLE COOLANT MAKEUP LINE

CHANNEL BOX

CONTROL BLADE

TIME DOMAIN REFLECTOMETER(LIQUID LEVEL)

mw I aw ISPND

IFRADIATED RODS 2C & 3C

Figure 2.2.5 Test Bundle Cross-Section for the NRU BWR Test

NUREG-1365, Rev. 1 22

2 Research Plan

Experiments in the German CORA ex-reactor fuel dam-age test facility with electrically fuel rods have providedmuch of the current information base on core degradationand early (metallic melt) phase melt progression. Thisinformation includes materials interaction (eutectic) ef-fects and the effects of reflooding damaged cores. TheCORA results are available to NRC as part of the inter-national Cooperative Severe Accident Research Program(CSARP) and will be used to further the data base forcode validation and assessment. The forthcoming CORAdry core BWR test will provide data on the initial condi-tions of the incoming melt for the ex-reactor experimentson metallic melt relocation and blockage formation andwill complement the results of these experiments.

2.2.2.2 Ceramic Pool Melt-through from a BlockedCore

In blocked core accidents, the primary determinants ofthe mass and other characteristics of the mostly ceramicmelt released into the lower plenum are the thresholdand location of meltthrough of the metallic and ceramiccrust system that supports the growing, mostly ceramicmelt pool. The failure threshold and location are them-selves primarily determined by the relocation and failuredynamics of this complex fuel-rod-supported system ofceramic and metallic crusts. Uncertainties as to the sur-face heat flux distribution of the internally heated meltpool because of natural circulation are also important,particularly for the failure location. An assessment will bemade and peer reviewed to determine whether or not thecorrelations derived from existing data at much lowerthan prototypic Rayleigh numbers and in differentgeometry than in those occurring in blocked core acci-dents are adequate to characterize the surface heat fluxdistribution of the ceramic melt pool. The TMI-2 meltpool had a Rayleigh number of about 1016, turbulent flow,and hemispheric geometry. Nearly all of the availabledata are for a Rayleigh number of 1011 or less, laminar orunsteady laminar flow, and thin, semi-circular geometry.In addition, there are a few data in thin, rectangulargeometry at Rayleigh numbers up to 3 x 1013 that agreereasonably well with the low Rayleigh number data. Ifdata at a Rayleigh number of 1016 should prove to beneeded, an experiment with water as a simulant fluidwould be feasible. The time constant for establishingnatural circulation flow and heat transfer in the melt poolalso needs to be examined. In addition, analysis should beperformed to assess the possibility that the metal in themostly ceramic melt pool can significantly affect the natu-ral circulation in the pool and its surface heat flux distri-bution.

Very little experimental information currently exists onthe melting dynamics of the ceramic melt pool and on therelocation and failure dynamics of the fuel-rod-supportedsystem of metallic and ceramic pool-supporting crusts in

the late-phase blocked core. Nearly all the informationcurrently available on these processes has come from theTMI-2 core examination. In the TMI-2 accident, poolmeltthrough occurred at the side of the core and reflectedthe core reflooding in that accident. It is currently notknown whether in unrecovered accidents, meltthoughwould occur at the side or at the bottom of the melt pool.The failure location makes about a factor of two differ-ence in the released melt mass.

The melt progression experiments in the ACRR test re-actor have been designed to furnish phenomenologicalinformation on the governing mechanisms involved in therelocation and failure dynamics of the fuel-rod-supportedmetallic and ceramic crust system that supports the grow-ing internally heated mostly ceramic melt pool. The meltattack on the metallic crust involves both thermal attackand possibly chemical (eutectic) interactions among thecrust materials and the Zircaloy cladding and U02 in thefuel rod stubs. The results of these experiments are to beapplied at full scale to blocked-core reactor accidentsthrough a mechanistic model or models of the melt-through process. The DEBRIS porous media model,which treats the governing processes involved mechanis-tically, will be the primary tool used in analysis and inter-pretation of the experimental results (Refs 1, 2). Thesimplified late-phase melt progression modeling inSCDAP/RELAP5 and MELCOR will also be checkedagainst these results.

A schematic drawing of the test assembly for the meltprogression experiments, actually that for MP-2, is shownin Figure 2.2.6. Detailed descriptions of these experi-ments are given in experiment plans for the melt progres-sion experiments and for the MP-2 experiments specifi-cally (Refs. 5, 6). In these experiments, a pre-cast metalliccrust with a prototypic low U0 2 content (for the dissolvedUO2 in the metallic crust) bridges a 32-rod array of cladfresh fuel rod stubs and supports a particulate ceramic(U0 2 ,ZrO2) debris bed.

The purpose of the MP-1 experiment was to providebase-line information on ceramic melt dynamics in theblocked core system and on the failure mechanics of thesupporting metallic crust. MP-1 was conducted in Octo-ber 1989, and the experiment was terminated earlier thanplanned because a temperature safety limit had beenreached internally in the experiment package. The MP-1experiment provided basic information on heat transfer,densification, and melting in the particulate ceramic de-bris bed and on the formation and downward propagationof the secondary ceramic crust or blockage below thegrowing ceramic melt pool. Peak temperatures of about3200'K were reached in the ceramic melt pool, whichgives about 400'K superheat above the UO2-ZrO 2eutectic temperature.

23 NUREG-1365, Rev. 1

2 Research Plan

U02-ZrO2 DEBRIS

WATER JACKETSTEEL JACKET

ZIRCONIA INSULATIONTANTALUM LINER

OUTER THORIA LINERINNER THORIA LINER

TUNGSTEN LINER

U-Zr-O-Ag-In-S.SMETALLIC CRUST

ZPZ=ZRP2162

__ ZRC1852ZF151882

ZRS1867

____ZRK1 872ZRK1877

TSKPCSI

ZRIS1562-ZR3S1567

!TKC32

ZRK1572A ZRK15T7

GflCN a-CRC82-1

FUEL ROD ARRAY Cm

FFUQ714

ZRC1362ZRC1 152ZRS0852

ZRK0772ZRK0787ZRK0762ZRK0757ZRK0652

STEEL HEAT SINK N

ZRK0062

TSKPC3

Figure 2.2.6 MP-2 Test Section Showing Major Components and Thermocouple Locations.

NUREG-1365, Rev. 1 24

2 Research Plan

MP-2 is to have a metallic crust of TMI-2 composition,which includes PWR control rod materials (Fe, Ni, Cr,Ag, In, Cd). A purpose of the MP-2 experiment is todetermine the effects of the eutectics of these control-rod-materials on the crust failure mechanisms andthresholds and to investigate the effects of the fuel rodstubs on the behavior of the secondary ceramic crust. It isnot currently known whether eutectic (chemical) interac-tion of the control-material-containing metallic crust withthe fuel rod stub Zircaloy cladding itself is significant. It isplanned to run MP-2 to metallic crust failure. The ce-ramic insulating shrouds and metallic shields in MP-2have been modified to prevent the premature termina-tion of the experiment that occurred in MP-1 and to allowtemperatures in the melt pool well above U0 2 melting(3100'K) as well as the U0 2 -ZrO2 eutectic (2800'K)without reaching the safety limit of the internal tantalumshield that required the shutdown of MP-1.

With the results and:interpretation of experiment MP-2and with the results of melt progression sensitivity stud-ies, an expert peer review group on melt progression Willreview the need for and the nature of further research inthe area of late-phase melt progression. In particular, adetermination will be made of the need for further meltprogression experiments.

2.2.2.3 Mode of Vessel Failure

The mode and timing of the reactor vessel lower headfailure have controlling effects on the subsequent con-tainment loading events in severe accidents. A programthat reviews and extends the numerous studies of reactorfailure modes made during the past decade is nearingcompletion. The major potential failure mechanisms in-clude penetration tube heatup and failure, tube ejection,lower head global creep rupture, and localized thermaland mechanical loads on the lower head. This last in-cludes nonuniform distribution of a the debris bed andcoherent jets of molten core material that can directlycontact and ablate the vessel wall.

Because there are a large number of debris conditions,lower head designs, and accident scenarios, analyticaltechniques using key dimensionless parameters wereused to develop failure maps that indicate the relativepotential for failure of the various modes as a function ofthe dimensionless parameters. Limited numerical finiteelement analyses were then utilized to benchmark thefailure maps.

An analytical method for bulging analysis (local creeprupture) was also developed. The analytical solution isbased on an axisymmetric plastic analysis with a pseudo-plane strain treatment of deformation normal to ameridional plane. This and other failure analysis methodsrequire high-temperature creep data. Therefore, a seriesof tests to acquire high-temperature creep rupture data

was carried out for pressure vessel steel. Additional creeprupture data acquisition is planned for the penetrationmaterials (Inconel, stainless steel, and SA105 or SA106steel). The results of this analysis have, been reported indraft NUREG/CR-5642, "Light Water Reactor LowerHead Failure Analysis" (Ref. 7).

The analytical models in this report are undergoing a peerreview to ensure that the report is on a firm technicalbasis. Model validation will be undertaken when resultsbecome available from the Swiss CORVIS program ofintegral tests on the melt pool thermal attack on- thevessel lower head and the head penetrations, which willbe made available to NRC through the CSARP program,and from the cooperative research program between theUSNRC and the I.V. Kurchatov Institute in Russia. Anextensive series of experiments and associated analysishas been started at the I.V. Kurchatov Institute on theintegrity of the vessel lower head under melt pool attack.A major purpose of this research is to investigate theefficacy of ex-vessel water cooling in preventing vesselmeltthrough. The research plans include large scale inte-gral tests with up to 200kg melts of U0 2 or molten salt, aswell as smaller scale separate effects experiments andanalysis on melt pool thermal hydraulics.


The first major area of investigation in the current meltprogression research program is the determination ofwhether blockage of the core by metallic melt or drainagefrom the core and core plate occurs in BWR dry coreaccidents. It is anticipated that the currently plannedex-reactor experiments on metallic melt relocation andblockage formation under BWR dry core conditions alongwith modeling will provide sufficient' information to re-solve this question. The full-length BWR test in NRU andresults from recent BWR tests in CORA should contrib-ute increased confidence to the conclusion. The results ofthese experiments will be used to validate models in theSCDAP/RELAP5 and MELCOR systems codes for usein predictions of the severe accident behavior of nuclearpower plants.

The situation regarding the second major area of investi-gation, ceramic pool melt.4hrOugh from a blocked core,however, is far more complex. The results and interpreta-tion of the ACRR MP-2 experiment should substantiallyincrease our understanding of the key processes involvedin ceramic melt pool growth and in the dynamics of therelocation and failure of the pool-supporting metallic andceramic crust system. With the results and interpretationof MP-2 in hand to augment the results of the TMI-2core examination, sufficient information should be avail-able for the expert review group on melt progression toassess the need for and nature of future work for late-phase melt progression issue resolution. Such future re-search may or may not include further melt progression

25 NUREG-1365, Rev. 1

2 Research Plan

experiments. The need for an experiment on the thermal-hydraulics of melt pools with internal heat generationmay be established, although this currently appears un-likely.

The report on lower head failure analysis will providefailure maps in terms of dimensionless parameters thatmay be used to predict failure modes. In many cases, themost probable failure mode may be predictable withoutfurther analysis. If further analysis is required, only alimited effort with minimal uncertainty ranges should benecessary.

Section 2.2 References

1. S. S. Dosanjh, "Melt Progression, Oxidation, andNatural Convection in a Severely Damaged ReactorCore," NUREG/CR-5316, SAND 88-3476, SandiaNational Laboratories, Albuquerque, NM (1990).

2. R. D. Gasser, S. S. Dosanjh, and R. 0. Gauntt, "TheDEBRIS Module: An Effective Tool for the Analy-sis of Melt Progression in LWRs," Eighteenth WaterReactor Safety Information Meeting, Oct. 22-24, 1990,NUREG/CP-0114, Vol. 2, p. 74

3. R. C. Schmidt and S. S. Dosanjh, "Core StructureHeat-Up and Material Relocation in a BWR Short-Term Station Blackout Accident," Sixth Proceed-ings of Nuclear Thermal-Hydraulics, 1990 ANSWinter Meeting, Washington, D.C., p. 42,Nov. 11-15, 1990.

4. R. 0. Gauntt, R. C. Schmidt, and S. S. Dosanjh, "AProgram Plan for the Ex-Reactor Metallic Melt Re-location Experiments," letter report to NRC, June1990.

5. R. 0. Gauntt and J. W. Fisk, "An Experimental Planfor the MP (Melt Progression) Experiments," letterreport to NRC, September 1989.

6. R. 0. Gauntt and R. D. Gasser, "Experimental Planfor the MP-2 Melt Progression Test in the SNLACRR," letter report to NRC, September 1990.

7. J. L. Rempe, S. A. Chavez, G. L. Thinnes, C. M.Allison, G. E. Korth, R. J. Witt, J. J. Sienicki, S. K.Wang, C. H. Heath, and S. D. Snow, "Light WaterReactor Lower Head Failure Analysis" (Draft,NUREG/CR-5642, EGG-2618, December 1991.

2.3 Fuel-Coolant Interactions andDebris Coolability


Since the WASH 1400* quantification of a steam-explosion-induced missile as a possible mode of contain-ment failure (alpha mode), significant progress has beenmade as reflected in current risk studies, e.g.,NUREG-1150 where alpha-mode failure does not seemto be a dominant contributor to early containment failure.However, the progress in understanding this area hasbeen mainly directed at the conditions for in-vessel mol-'ten fuel pouring into a coolant pool and its likelihood ofcausing containment failure by energetic interactions.

'The shift in emphasis to accident management for a vari-ety of reactor geometries and meltdown scenarios, cou-pled with the wide uncertainties in the NUREG-1150expert elicitations of fuel-coolant interactions (FCI), sug-gest that confirmatory research is needed to learn moreabout the fundamental mechanisms of FCI to be able todetermine the conditions under which FCI is important tosevere accident risk and accident management. The em-phasis of this research is now shifted to providing theappropriate methodological and analytical tools forevaluating major aspects of the accident sequences, in-cluding quantification of steam and hydrogen production,the mode and timing of vessel (or reactor coolant system)failure, and ex-vessel events of potential significance todebris coolability and containment loading. Although allthese issues are discussed elsewhere in this research plan,there are fundamental aspects of FCIs that are germaneto all these issues.

Three specific issues that require additional informationeither from experimentation or from analysis are:

1. FCI energetics

2. Fuel melt quenching in water pool

3. Water added to a degraded core (in-vessel as well asex-vessel)

The FCI characteristics are briefly described below, andthe experimental/ analytical data base needed for morecomprehensive resolution of these issues is discussed fordeveloping the overall research to address the abovethree areas.

2.3.1.1 Research Needs - Fuel- Coolant Interaction(Energetics)

The fundamental aspects of FCI are the evolution ofliquid interfacial (fuel-coolant) area and associated heattransfer during the contact. When the two liquids first

*U.S. Nuclear Regulatory Commission, "Reaclor Safety Study: An As-sessment of Risks in U.S. Commercial Nuclear Power Plants,"NUREG/-75/014 (WASH 1400), December 1975.

NUREG-1365, Rev. 1 26

2 Research Plan

come into contact, the coolant begins to vaporize at thefuel-coolant liquid interface as a vapor film separates thetwo liquids. The system remains in this state for a delayperiod ranging from a few milliseconds up to a few sec-onds. During this time, the fuel and coolant liquid inter-mix (sometimes called premixture) because of density andvelocity differences as well as vapor production.

The vapor film destabilization occurs next, triggering fuelfragmentation. This rapidly increases the fuel surfacearea, vaporizing more coolant and increasing the localvapor pressure. The vapor formation propagates spatiallythroughout the fuel-coolant mixture, causing the macro-scopic region to become pressurized. Subsequently, thehigh-pressure coolant vapor expands against the inertialconstraint of the surroundings and the mixture itself. Thevapor explosion process is now complete, transformingthe fuel's internal energy into the kinetic energy of themixture and its surroundings. Experiments to validatemixing calculations and fragmentation are needed to ad-dress in-vessel and, to a large extent, ex-vessel fuel-cool-ant interactions.

2.3.1.2 Research Needs-Fuel-Melt Quenching

The TMI-2 accident indicated that under certain condi-tions the fuel melt may be quenched at the time of pour-ing into a water pool in the RPV lower plenum. Previouslyit had been assumed that a fuel pour into the RPV lowerplenum would result in either settling of the unquenchedfuel (and eventual RPV wall failure) or a vapor explosion.Although there have been many integral FCI experi-ments, there is no data available under these particularconditions. Because of this lack of data, the FARO-LWRexperiments are planned. They involve a prototypic fuelmass (50-150 kg of U0 2/ZrO2 /Zr at 3000 °K) poured intosaturated water at high pressures (5-50 bar for 1-3 metersdepth with a chamber diameterof 0.5 to 0.7 m)..

2.3.1.3 Research Needs-Adding Water to a DegradedCore

Severe accident management is a natural outgrowth ofpast emphasis on risk assessment. In fact, there are anumber of particular issues that must be addressed whenaccident management is the main objective. Timing isimportant to accident management, and only recentlyhave PRA studies considered it in some systematic fash-ion; e.g., NUREG-1150 considered the effect of the op-eration of engineered safety features (containment fancoolers or sprays) before core heatup, before vessel fail-ure, or after failure. Inclusion of this "timing" behaviorcan indicate where opportunity exists for operator inter-vention to help reach a stable coolable state. Since wateris the primary accident management tool and the FCI canalter the course of the accident, it is important to investi-gate the benefits of adding water to the degraded corewith consideration of the possible adverse side effects.

This is particularly true for the FCI because many of thefundamental mechanisms are not well understood.


This section describes the current research program foreach of the three specific issues discussed earlier-FCIenergetics, fuel-melt quenching in a water pool, and add-ing water to a degraded core.

2.3.2.1 Research Program to Address FCI Energetics

The objective of this research is to determine under whatconditions vapor explosion energetics must be consideredand what are reasonable estimates for the energetic yield.As a first priority, we are testing the hypothesis that mix-ing will be limited by local steam formation and high voidfractions, causing water removal from within the fuelcoolant mixture. With such data, the experimental resultscan be compared to the computational models (e.g., IFCIor PM-ALPHA).

To isolate the water depletion phenomenon from particlesize distribution effects, experimental premixing simula-tions are carried out at the University of California atSanta Barbara (UCSB) using clouds of hot solid particles.A new instrument developed specifically for this purposeis used to measure the local liquid fractions in the three-phase mixing zone. The experiments are scaled(1/8-scale), using numerical simulations (PM-alpha) tocreate water depletion regimes similar to full-scalepoursof fixed-size 1-2 cm corium particles. These experimentsare currently in progress, and the results are expected toprovide the first experimental confirmation of the waterdepletion phenomena as well as a basis for assessing theaccuracy of its predictions in numerical simulations.

Additional melt breakup will occur as a consequence ofthe hydrodynamic interactions in the mixing zone. Twoexperimental programs are currently addressing thistopic. The one at UCSB is examining the fundamentalcomponent of an explosion, a single melt drop, underconditions that effectively simulate the role of such a dropin a real explosion environment. This is accomplished byusing a hydrodynamic shock tube capable of generatingpressure pulses similar to those of an explosion (designpressure 20,000 psi). With x-ray diagnostics, the detailedtime-history of fragmentation can be composed, and theresults are the key input in numerical simulations of esca-lation/propagation. This work is now completed; a reportwill be published by the end of CY92.

A second set of vapor explosion experiments is beingplanned at the University of Wisconsin. The purpose ofthese experiments is to produce a well-controlled one-dimensional geometry in which a fuel simulant (e.g., 2-20kg tin at 1300'K) pours into a water column, mixes withthe coolant, an explosion is triggered, and the explosionexpansion work is measured. These experiments are

27 NUREG-1365, Rev. 1

2 Research Plan

aimed at providing benchmark data to examine the effectof fuel-coolant initial conditions and mixing on explosionenergetics. The hypothesis for these tests is that fuel-coolant mixing and explosion processes occur under con-trolled conditions. Data from these experiments enablecomparison among the fuel-coolant mixing and explosionmodels used to make the case for applicability at reactorscale. This program is expected to be completed by theend of CY93.

It is expected that the above-described work will providethe basis for improved assessments of alpha-failure aswell as for several other special-effects assessments asthey arise in BWR ex-vessel sequences, where the rele-vant "sizes" of the explosion and the corresponding levelof energetics may be considerably smaller than those in-volved in alpha-failure considerations and where the needfor best estimate, rather than bounding, analyses is ofgreater importance. For example, the sensitivity calcula-tions in NUREG-1150 suggested that ex-vessel explo-sions in a Mark-II and Mark-IlI drywell may lead todrywell failure. However, more detailed analysis isneeded to verify that this threat exists.

2.3.2.2 Research Program for Quenching

To calculate quenching, the melt-coolant interface areaand the constitutive laws that define the interphase trans-fers (heat and momentum) must be determined. On theconstitutive laws, a reasonable estimate can be made byextending 2-phase formulations. The program involvingthe determination of interfacial area requires thebreakup history of the melt as it descends through thecoolant. Quite clearly, such experimental data are next toimpossible to obtain; our approach, therefore, has to belargely empirical. Available experimental data in this areaare very limited, thus we have joined the program at theFARO facility in Ispra.

The objective of the FARO tests to be performed at theJoint Research Center in Ispra is to observe the integralbehavior of fuel-melt quenching at high pressures underlikely severe accident conditions. What makes these ex-periments particularly attractive from a technical stand-point is that they can be performed with real reactormaterials (U0 2 , ZrO2 , Zr) at temperatures and pressuresthat are prototypic of actual severe accident conditionsand with the proper full-scale water depths for in-vesselaccident situations. The instrumentation within theFARO-TERMOS facility is substantial and the data col-lected is extensive. Thus it is likely that computer codecomparisons will be made to gain modeling insights intofuel-melt mixing and melt quenching in water.

Because these experiments are using prototypic materialsunder realistic initial and boundary conditions at theproper vertical length scales (e.g., water depth) the ques-tion of scale only becomes an issue relative to the size of

the fuel-melt pour and the lateral dimension of the facil-ity. The FARO facility has the capability to deliver a largemass of oxide melt under a variety of conditions. The fuelpour rate is planned to be within reasonable ranges foraccident conditions, i.e., jet diameters 5-10 cm and entryvelocities of a few meters per second.

Second, the FARO experiments could be considered rep-resentative of two types of geometric situations: (1) asingle jet in a large water pool or (2) a unit cell of amultiple jet pour into the lower plenum. In either case,the adequacy of a FARO facility vessel to provide a prop-erly scaled lateral dimension depends to some extent onthe degree of melt quenching. It is important to note thatthe chamber cross-sectional area will be varied by a factorof two in the tests to be performed to specifically addressthis point. If the results of the scoping test (50 kg of melt)and the base case experiment (150 kg of melt) indicatethat melt quenching is minimal (e.g., < 10% of the fuelquenched during the pour), the steaming rate will be low,level swell will be minimized, and the steam superficialvelocity will be small. Under these conditions, the lateraldimension of the vessel will not be an important concernregardless of which scenario is considered. Conversely, ifthe results of the scoping test and the base case experi-ment indicate significant melt quenching, the scaling ofthe experiments considering the lateral dimension will beproblematic for either scenario. Qualitatively, one wouldstill expect the experiments to be quite informative andvalid for reactor safety implications. However, quantita-tive interpretations of the tests must then account for thelateral dimensions and the likely large steaming rate,superficial velocity and all other consequences associatedwith substantial fuel and liquid water interactions.

Based on these considerations, it seems clear that thelogical sequence is that the scoping test and the base casetest should be performed in the FARO-LWR experimen-tal program, then the scaling rationale should be reas-sessed, and finally, the parameters determined for futureexperiments. Currently eight experiments are plannedfor the test series.

2.3.2.3 Research Program for Reflooding

During a severe accident situation, there is little doubtthat the primary efforts of the operators will be directedtoward achieving a stable, coolable configuration by mak-ing water available to the reactor vessel (if loss of coolingwas believed to be the cause of accident). An importantquestion that must be considered is the likely conse-quences of these actions. Given the uncertainties in coremelt phenomena and the state of the core during anactual severe accident, and given the intuitive drive to putwater into the core in the event of an undercooling acci-dent, it is likely that the operators would inject water intothe reactor vessel should water become available duringthe course of an accident. However, along with the poten-tial benefit of achieving a stable, coolable configuration,

NUREG-1365, Rev. 1 28

2 Research Plan

restoring water to a core that has been severely damagedcan have effects of which the operator should be aware.The operator should be fully cognizant of possible symp-toms and responses of the plant to added water undersuch circumstances (e.g., molten core coolant interaction,increased hydrogen generation, increased containmentpressure). The following addresses reflooding researchrelating to the in-vessel and ex-vessel parts of an accident.

2.32.4 Research Program for Reflooding (In-vessel)

The working hypothesis is that adding water would behighly beneficial and likely to terminate the accident atthis stage. The fuel-coolant interactions associated withsuch addition would be benign, yielding quench, with thepossibility of hydrogen production in a narrow time win-dow when cladding is hot enough and still in a highlyundistributed geometry.

The current research approach to this issue is to reviewpast investigations in which water (or its simulant) hasbeen added to a degraded core and to determine what theadverse effects could be and what the current state ofknowledge (data and analysis) suggests. An initial reviewof past experiments suggests that a few tests have beenperformed as part of the CORA and PBF experimentalprogram as well as the LOFT-FP2 test. In the past,simulant tests of coolant added to fuel debris were per-formed at Brookhaven, Argonne, and UCLA. Althoughlimited in scope, these tests address water addition duringin-vessel core degradation. The possible adverse effectsshould be understood and their impact minimized. Possi-ble adverse effects include:

* Hydrogen generation and fuel heatup from exother-mic metal-water reactions,

* Energetic FCIs that may adversely affect attainmentof a stable coolable state.

The major variables for determining whether water addi-tion would have adverse effects would be the rate andcharacter of water addition and the state of the fuel at thetime it is added. To help in focusing this work, qualitativescenarios of the accident were developed with referenceto water availability and its effect. On the basis of qualita-tively different fuel-coolant contact configurations, thefuel states are:

1. Initial heatup and core degradation (rods intact),

2. Advanced core degradation (core rubble, melt, andrelocation).

In order to better understand the accident managementissues for this stage of the accident, we are collectingavailable information on core degradation morphologies

and implied permeabilities as affected by fuel rod disinte-gration or clad relocation.

2.3.2.5 Research Program for Ex-Vessel DebrisCoolability

One of the most important phenomenological issues inthe progression of severe accidents after the reactor pres-sure vessel has failed is whether sufficient energy can beremoved from the discharged molten debris that the plantcan be brought to a stable condition and the challenge tocontainment integrity, whether by basemat penetrationor by containment overpressurization, is avoided. Themost commonly available mechanism for removing en-ergy from the ex-vessel molten debris in LWR contain-ments is water addition. Issues that must be resolved todevelop debris coolability criteria are (1) the nature andconfiguration of debris, (2) the heat extraction mecha-nism from debris by water, and (3) the molten debris-con-tainment structure interaction during the cooling process.

In order to obtain data to support the development ofcoolability criteria, an experimental program calledMACE was developed under the sponsorship of NRC,EPRI, and several OECD countries. The program is in-tended to determine the ability of water to cool moltencore debris during MCCI and to enable characterizationof the resulting debris for assessment of permanentcoolability.

In August 1989, a scoping test was conducted in theMACE series of tests in which approximately 109 kg ofU0 2 -ZrO 2-Zr melt in a 30 cm x 30 cm x 15 cm poolinteracting with limestone-common sand concrete wasflooded with a 50-cm head of water. The melt decay heatwas simulated by direct electric heating; however, opera-tional problems caused the input power to significantlyexceed the decay heat (by a factor of about 3). The resultsfrom the test indicate heat transfer rates of 2.4-3.5MW/m 2 during the initial period of water interaction. Therates decreased to 0.6 MW/m 2 and down to 150 KW/m 2

for the more quiescent period. The results support theconcept of a thickening crust with periodic access of waterto the melt and partial melt quenching.

A second MACE test, conducted in November 1991, em-ployed approximately 430 kg of U0 2-ZrO2 -Zr-Fe2 O3-CaO melt mixture in a test configuration of 50 cmx 50 cm x 25 cm. As in the scoping test, the melt pool hadan overlying water with 50 cm head and an underbed oflimestone-common sand concrete. The test was termi-nated after 25 cm of concrete ablation. The results indi-cate a heat transfer rate of 1 MW/m2 during the initialperiod down to about 30-60 KW/m2 during the quiescentperiod. The results further indicate formation of a sus-pended crust thereby preventing continuous meltquenching. The molten material underneath the crustablated 25 cm of concrete at which point the test wasterminated. Based on the post-test analysis of the second

29 NUREG-1365, Rev. 1

2 Research Plan

MACE test, it was concluded that the initial and boundaryconditions of the test were not prototypic. Thus, a thirdtest was performed in April 1992 with corrected initialand boundary conditions, but otherwise in an identicalconfiguration. Preliminary results from the test indicatesome degree of coolability although any firm conclusionmust await further analysis of the test data.

Currently, three more tests are planned in the MACEprogram. These tests will use variable depth ofU0 2 -ZrO2-Zr melt mixture in two different geometricalconfigurations (50 cm x 50 cm x 75 cm) interacting withlimestone-common sand concrete and siliceous concrete.Water will be added shortly after the MCCI has started.The objective is to measure the rate of heat removal bywater; study the melt cooling process, including crustformation, stability, and growth; and to investigate theeffect of various parameters (e.g., debris depth, cavitygeometry, power density, concrete composition) on avail-ability.

Concurrent with the MACE program and under thesponsorship of NRC, a second experimental program ondebris coolability was initiated at the Sandia NationalLaboratories in February 1991. The goals of this program,called WETCOR, are to complement and augment theMACE program and to provide a means to support theassessment of existing as well as ALWR 'designs. TheWETCOR program is designed to address two specificissues: (1) the comparative coolabilities of oxidic and me-tallic debris and (2) the limits of debris coolability in termsof debris composition and depth.

In the first WETCOR test (WETCOR-1) run in Septem-ber 1991, 35 kg of charge material 80 w/o A120 3-20 w/oCaO was heated to melting at 1850'K within a 32-cmdiameter tungsten ainnulus. The charge and the annuluswere'surrounded by a cylindrical MgO crucible and sup-ported on a 40-cm diameter by 40-cm high concretebasemat. After 2 to 3 cm of concrete ablation, the moltendebris was flooded with approximately 25 cm head ofwater at a rate of 60 lpm. The results indicate approxi-mately 5 to 6 cm of concrete ablation after 30 minutes ofwater addition. The results further indicate formation ofrelatively thick and layered crust, and partial meltquenching with voids between crust and quenched melt.No evidence of fragmentation is apparent from the posttest examination of crust and frozen debris.

A second test, called WETMET, was performed in De-cember 1991 to evaluate the potential for prolonged mol-ten debris-water interactions which might result in bulkfreezing or quenching of the debris. The charge materialfor the test was 92 kg of 304 stainless steel and thebasemat was made of limestone concrete. The test wasexecuted in two stages. In the first stage, the metal debriswas heated to melting at 1990k, ablation began and waterwas added at 50 1pm for approximately 1 hour before the

power was cut off to let the debris solidify and cool. In thesecond stage, water was added before power was appliedto the new quenched meltpool configuration. Power wasthen increased until melting and concrete ablation werere-established. The results indicated an initial period ofvigorous melt-water instability followed by a stable crustgeometry with substantially reduced rates of energy trans-fer.

Results from the limited number of tests described aboveled the experts to believe that the debris coolability (acoupled FCI and MCCI phenomenon) may be an ex-tremely complex process that demands a more carefulexamination of various factors contributing to the proc-ess. This gave rise to the concept of "morphological"testing, which deals with the determination of debris mor-phology as a function of experimental variables. The firstof such morphological tests currently in the planningstage involves a prototypic charge material on a nonreac-tive (MgO) basemat.


The results of this research will identify the basic vari-ables governing heat transfer and hydrodynamics of melt-water interaction, including the effect of water injectionon debris configuration. Models supplemented by appro-priate correlation to address all possible phases describedabove would be validated with available and planned ex-perimental data. These models then can be used to assessthe potential for and consequences of fuel coolant inter-actions, assess the efficacy of accident managementstrategies, assess the effect of FCI on altering accidentsequences, and provide estimates for the rate of steamand hydrogen generation following reflooding by water.

2.4 Severe Accident CodesBecause of the difficulty in performing prototypic experi-ments and the variety of scenarios possible, substantialreliance must be placed on the development and valida-tion of complex computer codes for analyzing severeaccident phenomena or planning accident managementstrategies. A number of codes (e.g., SCDAP/RELAP5,MELCOR, CONTAIN, CORCON, COMMIX,HECTR, MELPROG/TRAC, VICTORIA, andBWRSAR) have been developed for various stages insevere accidents, both in-vessel and ex-vessel, for bothPWRs and BWRs. However, as a result of a review ofNRC-supported codes and associated documentation,the staff has terminated support for three severe accidentanalysis codes, HECTR, MELPROG, and BWRSAR.BWR-specific models developed under BWRSAR spon-sorship are being incorporated into MELCOR andSCDAP/RELAP5. The HECTR and HMS codes weredeveloped to test models of hydrogen mixing and combus-tion within reactor containments. MELCOR and CON-TAIN have incorporated the models that were part of

NUREG-1365, Rev. 1 30

2 Research Plan,

HECTR for assessing hydrogen challenges from severeaccidents. The HMS code is being documented, and nofurther development is planned. Support for MELPROGwas terminated on the basis that it would be duplicative ofSCDAP/RELAP5 as a detailed mechanistic in-vessel se-vere accident code, however, the IFCI module will beextracted from MELPROG to be used for fuel-coolantinteractions analyses. SCDAP/RELAP5 is a less detailed,more flexible code and has been validated with consider-able experimental data. The development and documen-tation of the original integrated risk analysis code, theSource Term Code Package (STCP) has been completed.The STCP, a collection of various codes with theMARCH code serving as the cornerstone, will no longerreceive developmental or maintenance support.

The development of severe accident phenomenologicalmodels and computer codes continues to have a role inachieving the objectives of the severe accident researchprogram. The codes now under NRC's sponsorship andsupport are SCDAP/RELAP5, MELCOR, and CON-TAIN. In addition, several other codes such as VICTO-RIA, COMMIX, and IFCI are being developed and main-tained to perform specific functions that require detailedmodeling These codes will be used to benchmarking thesystem level codes discussed earlier. Also, the core con-crete interaction code, i.e., CORCON, will be incorpo-rated into the CONTAIN and MELCOR codes. The rela-tionship of these codes to various severe accidentprogression phenomena is shown in Figure 2.4.1. Addi-tional discussion on the status of each of these codes isprovided below.

2.4.1 SCDAP/RELAP5

2.4.1.1 Research Needs

Since the 1979 Three Mile Island-2 (TMI-2) accident,the NRC has conducted a broad-based research programto develop an understanding of severe accident behavior.As far as practicable, much of this understanding, frominitial core uncovery through reactor vessel failure, iscontained in the SCDAP/RELAP5 computer code.SCDAP/RELAP5 is designed to model the coupled inter-actions that occur between the reactor coolant system(RCS), the core, and the fission products during a severeaccident.

The objective of this research program is to develop ananalytical tool (i.e., the SCDAP/RELAP5 code) that willprovide the NRC with the capability to perform detailedanalyses of in-vessel core melt progression phenomenaduring various severe accident conditions and scenarios.

2.4.1.2 Current Research Program

The development of SCDAP/RELAP5 has been basedon a combination of factors, including model assessment

and validation activities, insights and results of separateeffects and integral experiments, and user needs to sup-port resolution of various severe accident issues, and acci-dent management. Since 1984, the code has been used tosupport the analysis of several major severe accident ex-periments such as PBF SFD 1-3/1-4, LOFT FP-2,ACRR DF-4, and CORA experiments. The code hasbeen used for severe accident analyses, including naturalcirculation studies and the analysis of lower plenum de-bris and lower head heatup. Accident management stud-ies have been performed which include the analysis ofstrategies or phenomena that minimize direct contain-ment heating.

Systematic assessment of the newest version of the code[i.e., MOD3(8x)] will be performed to define modelinguncertainties in important severe accident phenomena.These uncertainties, along with uncertainties in systemthermal-hydraulics from natural circulation experimentsand design basis accident (DBA) experiments, will bepropagated through plant-and accident-specific condi-tions to assess the uncertainties in conditions of RCSfailure caused by natural circulation, a damaged core, andfission product and hydrogen releases during a severeaccident. Some late-phase core-melt-related experimentswill be planned and performed as discussed in Section 2.2.Once these experiments have been performed, models totreat late-phase core melt progression that are not ade-quately modeled in the code will be developed, and sys-tematic assessments will be performed.

Preliminary assessment of earlier versions of SCDAP/RELAP5 that began in 1991 have identified the relativestrengths and weaknesses of many models in the code. Ingeneral, many predicted early-phase phenomena, such assystem temperature, fission product release, and the in-itial change in the core geometry caused by ballooningand melting of core structures, have been within the ex-perimental uncertainties for the individual experiments.Assessment results have also revealed several areas thatwill require model improvements. Specific areas include(1) renewed bundle heating, melting, hydrogen produc-tion, and fission product release during reflood, (2) initialrelocation of molten material as droplets and rivulets, (3)interaction between fuel rod cladding and Inconel gridspacers, and (4) the diversion of flow from damaged fuelassembles. Removing these modeling deficiencies willsignificantly reduce the calculated phenomenological un-certainties in the early-phase core melt progression. As aresult, the end state of in-vessel core melt progression canbe more accurately determined. This in turn should con-tribute to a better estimate of containment loading, fis-sion product release, and perceived overall risk for severeaccidents. Thus, efforts devoted to removing these codedeficiencies are the main focus of current SCDAP/RELAP5 research activities. Other research activities in-clude (1) incorporation of ORNL-developed BWR model

31 NUREG-1365, Rev. 1

2 Research Plan

(a00

z

-0

II

iiILI~ .s

iiII

Sw

iiI

I!d

ii

CO)w0)0

0U

0

z

0)LU0

U)

0

I-

LU

Cm

0

0

zlii

0zuJI

z

0~z0J

0

0Ui0

x

00

Figure 2.4.1 NRC Severe Accident Codes

NUREG-1365, Rev. 1 32

2 Research Plan

improvements for current generation BWR as well as forthe advanced SBWR, (2) SCDAP/RELAP5 peer review,and (3) model extension to treat Westinghouse AP600core design changes, including core structures and mate-rials that differ from current PWR designs. The purposesof the SCDAP/RELAP5 peer review effort are (1) toprovide an independent, high-quality review of the code,(2) to help the NRC determine future code developmentdirection, effort, and priority, (3) to provide informationconcerning code deficiencies and limitations that need tobe improved or corrected, and (4) to determine the de-gree of technical adequacy of the code.

As for the late phase of severe accidents (e.g., from theformation of bundle-size blockages through the growth ofmolten pools and the relocation of molten material intothe lower head), the core melt progression phenomenaare still poorly understood because of the lack of experi-mental data. As the late-phase core melt progression databecome available, models will be developed and incorpo-rated into the code. Modeling assessments against experi-.mental results have been made (see Table A.2.1). Addi-tional modeling assessment and validation efforts willcontinue to be made to ensure that SCDAP/RELAP5meets the code's design objectives and targeted applica-tions.

2.4.1.3 Anticipated Results

Uncertainties in predicting vessel and RCS failure timesand vessel failure modes using SCDAP/RELAP5 will besubstantially reduced because of the extensive data base(including late-phase experimental results) and modelimprovement efforts. The end product of this researchprogram is to provide NRC with a computer code that iscapable of performing (1) plant (both PWR and BWR)analysis for the in-vessel core melt progression phenom-ena for various severe accident scenarios and (2) experi-mental analysis and support for in-vessel severe accidentexperiments. Other uses of the code include (1) assess-ment of the efficacy of accident management strategies,(2) MELCOR benchmarking and assessment, (3) TMI-2accident evaluation, and (4) support for ALWR designcertification as defined in SECY-91-161.

2.4.2. CONTAIN


The CONTAIN code is a detailed mechanistic code forthe integrated analysis of containment phenomena. It hasbeen developed under NRC sponsorship to provide thecapability to predict the physical, chemical, and radiologi-cal conditions inside a nuclear reactor containment in theevent of a severe reactor accident, as well as fission prod-uct releases to the environment in the event of contain-ment failure.

The CONTAIN code models intercelU flow, hydrogen andcarbon monoxide combustion, heat and mass transfer (ra-diation, convection, conduction), aerosol behavior, fissionproduct behavior (decay heating, transport), engineeredsafety systems (sprays, fan coolers, ice condensers),BWR-specific systems (suppression pools and safety re-lief valve discharge), core-concrete interaction, and sim-ple treatment of direct containment heating. The codeprovides the capability to analyze a wide variety of LWRplants and accident scenarios.

One of the safety issues that is currently under extensiveinvestigation is direct containment heating (DCH) andpressurization of the reactor containment atmosphere bymolten core materials ejected following the lower headfailure of the reactor vessel under pressure. DCH in-volves a large array of complex processes, many of whichare only now becoming better understood and for whichlittle or no experimental data previously existed.

Several research activities are ongoing in an attempt toquantify the effects of DCH (See Section 2.1). Histori-cally, the NRC DCH experimental program was guidedby calculations and sensitivity studies performed with theCONTAIN code. Although these sensitivity studies andpretest and posttest predictions were useful in identifyingimportant processes, there was concern over the degreeof confidence in the code's treatment of the DCH proc-esses to determine the relative importance of these proc-esses in the tests. In FY90, a comprehensive scaling meth-odology was developed and implemented for the purposeof ensuring a properly focused and technically defensibledirect containment heating experimental program. Thisprogram will guide the development of models for incor-poration into the CONTAIN code.

Another key research need is related to containmentanalyses for ALWR designs namely AP600 and SBWRplants. Containment designs are being developed forALWRs that incorporate passive cooling and decay heatremoval features for protection against long-term con-tainment overpressure in accident situations. The passivenature of these containment systems poses unique chal-lenges to containment analysis codes for predicting con-tainment response in both design basis and beyond designbasis events. Such challenges which require new or im-proved models include natural circulation air flow in thechannel outside the containment shell, behavior of anevaporative flowing water film on the containment shell,stratification of gases within the containment, potentiallyunique heat transfer and condensate film behavior notadequately represented by existing correlations, and nu-merical challenges arising from the need to efficientlyperform long-term containment response calculations.Similarly, the performance of the passive containmentcooling system (PCCS) in the SBWR will be modeled.

33 NUREG-1365, Rev. 1

2 Research Plan


In FY92 and FY93, selected DCH models will be incorpo-rated into the CONTAIN code and evaluated against therelated experimental data. Then, actual plant cases willbe performed with the updated CONTAIN code to deter-mine the impact of DCH on containment pressurization.

Another primary objective of the current research pro-gram is to develop and validate models related to theALWR performance for both design basis accidents andsevere accidents. The CONTAIN code will be modifiedaccordingly, and industry experimental programs will beevaluated and used to validate the CONTAIN code.

Other research activities include efforts to update theCONTAIN code consistent with the improvements inmodeling already developed under other programs. Cur-rently, only iodine washout by containment sprays is mod-eled mechanistically in CONTAIN. The TRENDS codedeveloped at ORNL is able to provide mechanistic mod-eling of iodine chemistry and other fission product physi-cal and chemical processes. Therefore, incorporation ofthis package into the CONTAIN code would further ad-vance the overall capability to predict fission productrelease constituents and fission product spatial heatingeffects. This effort will be completed in FY93. Core-con-crete interaction phenomena modeled in CONTAIN arebased on the CORCON-MOD2 code. Release of theCORCON-MOD3 code, which includes the improve-ments accumulated over approximately 6 years, necessi-tates the update of the corresponding CONTAIN model;this effort will be completed in FY93. Also, Figure 2.4.2illustrates past and present validation and assessmentactivities to compare experimental tests against codemodel predictions.

As a result of incorporating the various selected models inCONTAIN, the code manual will be updated. Also, toensure the CONTAIN code meets its design objectivesand need for regulatory application, a peer review will beundertaken by independent experts. The peer review willbegin in CY93, and will be completed in about 12 monthsafter initiation.


As a result of upgrading the CONTAIN code, actual plantanalyses will be able to be performed to determine theimpact of DCH on containment loading. Also, the codewill be capable of performing ALWR containment analy-ses for both design basis accidents and severe accidentsfor the AP600 and SBWR plants. In concert with theseefforts, the CONTAIN peer review will provide an over-all independent assessment of the code. This assessmentwill assist in steering future efforts on this program.

2.4.3 MELCOR


Research needs for the MELCOR code fall into a num-ber of areas that can generally be divided into develop-ment or assessment. The development areas include (1)adding calculational capabilities (or models) into thecode, (2) correcting or improving existing calculationalroutines (or models) in the code, (3) correcting code er-rors, and (4) improving code documentation. The assess-ment area includes (1) comparing calculational resultswith experimental data or output from other code calcula-tions, (2) varying input parameters to characterize thesensitivity of results for specific types of calculations, and(3) determining reasons for differences between the re-sults of calculations and experimental data.

A comprehensive independent review of the MELCORcode was undertaken by a peer review committee. Theobjectives of this review were to (1) provide an independ-ent assessment of the MELCOR code, (2) determine thetechnical adequacy of the code, and (3) issue a final reportdescribing the technical findings. The review committeerecommended improvements in five areas: (1) MELCORnumerics, (2) missing models, (3) revisions to existingmodels, (4) expanded assessment, and (5) documentation.They concluded that when these recommendations aresatisfied, MELCOR will be technically adequate for PRAapplications, although the code may not always be suffi-cient for some parametric accident management studies.The peer review committee issued its finding in 1992(report LA-12240).


The current focus of the MELCOR research efforts areon (1) resolving time-step size and machine-type depend-encies (numerics problems) as urged by the peer review-ers, (2) correcting inadequacies in some of the phenome-nological models that were pointed out by the peer reviewcommittee, (3) adding missing models 'nto MELCOR,and (4) expanding the technical assessment program toprovide assurances that the code is reliable,or, if deficien-cies are found, to identify those deficiencies so that reme-dial actions may be taken.

In regard to the five areas of recommendations by thepeer reviewers the following actions are planned.

2.4.3.2.1 MELCOR Numerics: Efforts have been directedtoward investigating problems and sensitivities associatedwith code numerics including time-step and machine-typedependencies, identifying their underlying causes, andeliminating or mitigating them. Efforts to remedy thenumerics problems started in FY91 and those problemsidentified through 1991 have been settled; others arebeing investigated in FY92 and FY93.

NUREG-1365, Rev. 1 34

CONTAIN Validation andAssessment Strategy

Surc SNL and ANL NTS tests and HDR T31. W PCCS LACE ABOOVE PNL lce HDR ElieaCHtss TI2nlys adT3 6L4 AB&A7CoesradEi 1.2i~ •lWtoBeta DCH tests TMI-2 analysis and T3!.6 tests; LA AB5 & A87 Condenserl and El1.4

a

CORCON DCH HECTRStandalone Scaling StandaloneValidation Models Validation

Core Concrete Direct Contain- Hydrogen Heat & Mass Two Phase Engr Safety IntegralInteractions Jment Heating] urnigng Transfer T-H and Flow Behavior Features Analysis

."' •;',' ' :".-:: .. i • ;i O N A ;. IJ i .; .,,•-• ..... , •.•. r;•CON T A IN . -. ,' , ,. •-•-. . - -..- -.-- , -

... ,CD : ;*: , . ; : - • ,, € , .. ... . . . . . .. . .• . . .• . .

2 Research Plan

2.4.3.2.2 Missing Models: The review committee con-cluded that models for the following phenomena shouldbe given the highest priority for incorporation in MEL-COR:

0 PWR primary system natural circulation in compo-

nents with countercurrent flows,

* high-pressure melt ejection and DCH,

* ice condenser containments,

* nonexplosive interactions between debris and water,

* fission product vapor scrubbing,

" additional reactor coolant system fission productdeposition processes, and

" fission product reactions with surfaces.

Model development activities were initiated in FY91 forthree missing models: natural circulation, direct contain-ment heating, and ice condenser containment. Work onthe remaining models dealing with fission products isbeing integrated into the current workscope. The effort toincorporate a model into MELCOR for handling directcontainment heating phenomena was based on modelinginsights derived from the DCH experimental and analyti-cal programs developed through interactions between theexperimental group and the code developers. Incorpora-tion of appropriate DCH modeling into MELCOR wascompleted in FY92. Completion of the implementationof a relatively fast-running model for calculating naturalcirculation that is consistent with the MELCOR architec-ture and execution time requirements is scheduled for.FY92.

Three fission product behavior models, identified as miss-ing by the MELCOR peer review committee, are beinggiven consideration for inclusion in the source term calcu-lation, especially for PRA use. These are described asfollows: Although MELCOR has a relatively sophisti-cated model for scrubbing aerosols in a water pool, insome sequences gas temperatures in the primary systemmay be sufficiently high that volatile fission product va-pors, rather than aerosols, would be discharged intopools. MELCOR does not represent the removal of thesevapors as they condense to aerosols and attempt to passthese aerosols through the pool. Examples of sequencesin which this process is important include low-pressureBWR sequences with discharge of vapor through thesafety relief valve lines to the suppression pool or low-pressure PWR containment bypass sequences with dis-charge to a water pool in the auxiliary building. In suchsequences, this phenomenon could dominate the calcula-,tion of the source term, especially when other fissionproduct removal mechanisms are weak.

MELCOR's fission product deposition models areadapted from the MAEROS containment model. Assuch, certain processes that are not generally important inthe containment have been neglected. These include im-paction and turbulent deposition of aerosols. Experimen-tal data on containment bypass sequences performed forthe Electric Power Research Institute, as well as calcula-tions using more comprehensive aerosol deposition mod-els, indicate that the neglect of these processes may resultin a significant underestimate of the retention of aerosolsin the primary system, especially for low-pressure se-quences in which gas velocities are high. Correlationswhich have been developed to represent these effects areincluded in the VICTORIA code and consideration willbe given to adopting those or simpler versions in theMELCOR code.

Chemical reactions between settled aerosols and vaporsand heat sinks in the primary system can greatly affectdeposition (chemisorption) and revaporization rates. Thereviewers state the lack of explicit modeling in the codeapplies to all accident sequences and is particularly seri-ous for cesium hydroxide and tellurium compounds. Sim-ple models for these effects which capture the most im-portant effects during accident sequences, as determinedby the detailed VICTORIA code, will be considered forMELCOR in the future.

2.4.3.2.3 Revisions to Existing Models: The Committeerecommended that the following concerns with existingMELCOR models be given the highest priority.

* Condensation is treated independently in the codeshydrodynamic behavior modeling from those calcu-lations of aerosol particle growth and deposition inthe radionuclide modeling portion of the code. Thevalidity of this approach should be demonstrated bycomparison with more exact models or data.

* Inconsistencies in the treatment of chemical reac-tions between CORCON and VANESA should beresolved, and improvements should be made to theCORCON/MOD2 phase diagrams.

* The model for condensation in containment (masstransfer) should be revised.

" There seems to be a disagreement between the pooldecontamination factors computed with the currentpool scrubbing model and those calculations madeusing other codes. A general feeling is that theremay be an implementation error, however, no sucherror has been located.

The acceptability of using the hydrodynamics packagewater condensation/ evaporation model for calculationsof aerosol particle growth and deposition fri the radionu-clide package is being addressed.

NUREG-1365, Rev. 1 36

2 Research Plan

Improvements in the oxidic and metallic phase diagramsare needed because the effects of eutectic formation ondebris layer melting and solidification behavior are im-portant. Current incompatibilities in material propertiesbetween the cavity modeling package and other MEL-COR packages are being eliminated, and duplicative andinconsistent chemistry between CORCON andVANESA are being eliminated by the implementation ofCORCON/MOD3 into MELCOR.

The model for condensation in the presence of noncon-densables is being revised to address identified deficien-cies, including modeling of the film thermal resistanceand high mass transfer effects.

The pool scrubbing model discrepancy will be resolved.

2.4.3.2.4 Expanded Assessment: The committee also con-cluded that the ability of MELCOR to calculate severeaccident phenomena has not been demonstrated suffi-ciently. Such a demonstration should be based on (1)sensitivity studies, (2) benchmarking activities using ex-perimental data, and (3) code-to-code assessments.

A plan for a more comprehensive integral assessment ofMELCOR is under development. The technical assess-ment planning involves a number of related activities,including verification, validation, and quantification ofuncertainties. This process involves reviewing models andcomparing analytical results to experimental data, includ-ing small-scale and full-scale experiments. Only a smallportion of the comprehensive plan has been accom-plished so far. It is a high priority to have at least somevalidation results as soon as possible for each of the majorphenomena treated by the code. There is a need to de-velop a standard code package demonstration/test prob-lem for many of the MELCOR code packages. At leastone assessment problem for each major package or foreach major phenomenological area will be prepared (e.g.,the ice condenser model being added to the general HSpackage). Also, the program is being carefully designed toprovide more substantive assessment of the COR and RNpackages, which were major areas of concern to the peerreviewers. A graphic description of the assessment pro-gram comparing analytical results and experimental datais given in Figure 2.4.3 which relate code phenomenabeing validated to applicable experiments. Additional as-sessment activities involve code to code comparisons andanalysis and evaluation of full plant transient calculationsfor various plant and accident sequences.

Consistent with the draft assessment plan, activities con-centrate on PRA risk-dominant sequences, such as sta-tion blackout, steam generator tube rupture, and V-sequences. The Semiscale test selected for primarysystems thermal-hydraulic and heat transfer assessmentcomplement and support these selected demonstration

plant analyses. The work in this area is intended to becombined with work on other plant geometries (e.g.,B&W PWRs, BWRs) for similar sequences in both ex-perimental analyses (using MIST and FIST test facilities,from Semiscale-sized equivalent facilities for those othergeometries) and for full-sequence demonstration analy-ses for other plant types. Demonstration calculations toanalyze B&W and Combustion Engineering PWRs arebeing run at Brookhaven National Laboratory.

In addition, the NRC is in the process of initiating aninternational cooperative effort for technical assessmentof the MELCOR code, the MELCOR Code AssessmentProgram (MCAP). The objective is to accelerate the tech-nical assessment, consistent with the peer reviewers com-ments, by employing the expertise of many of the codeusers both inside and outside the U.S. Thus, many casescan be run in a shorter time. This will also allow expansionof the user community and at the same time improve theunderstandings and abilities of the users to run the MEL-COR code.

2.4.3.2.5 Documentation: The body of existing MELCORdocumentation is significant. However, the Committeefelt that detailed descriptions of the models and correla-tions were lacking in some cases, and documentation onthe applicability and benchmarking was either inadequateor missing. The Committee also recommended that aprocess for collecting, documenting, and distributing userguidelines to the MELCOR user community be devel-oped.

Improvements to the MELCOR documentation areplanned. However, documentation equivalent to theTRAC "Models and Correlations" code document wouldbe highly resource intensive and somewhat duplicative ofinformation already available in the MELCOR manuals.Nevertheless, in FY93 some resources are directed toupgrading reference manuals to cover the most criticalneeds. Further, there is a task to develop a practical userguidebook for MELCOR users.

2.4.3.2.6 Additional Research

Not all of the code development and assessment work is aresult of the peer review. Other research activities in-clude: (1) incorporation of ORNL-developed BWRmodel improvements, (2) incorporation of an upgrade ofthe CORSOR fission product release model includinguse of the Booth model, (3) improving core debris reloca-tion modeling to include spreading, (4) improvement ofcore melt modeling to allow simple material (eutectic)interactions to be treated on a parametric basis, (5) fur-ther expansion of the technical assessment program toinvolve other DOE laboratories engaged in severe acci-dent research for the NRC, and (6) testing of an inputmodel for AP600.

37 NUREG-1365, Rev. 1

•Z N-?

0

CD

'J2

:. FLECHT- GE HFIR LOFT PHEBUS CORA FLHT PBF ACRR Marviken TMI

SEASET Level LBLOCA LP-FP-2 B9+ 13 2,4 SFD ST-1,2 ATT-2b,4 2

NC Tests Swell & S.C. (ISP-28) (ISP-31) 1-1,1-4

00

CalCL

<

0.

T11.2(ISP-29) 2,6,10,19 A87 16-11

0

c. CORCON

Standalone

Validation

Core-Concrete Containment DCH Hydrogen Aerosol ContainnmentZ Interactions T/H Combustion Behavior ESF

C)

2 Research Plan

Material interaction models for the in-core region will beincorporated into MELCOR. ORNL has arranged theprogramming of the BWRSAR eutectics model for incor-poration into MELCOR. This model is limited to BWRapplications and to lower plenum debris bed behavior.

In another important developmental area, work is ongo-ing and planned for the evolutionary light water reactor(ELWR) and ALWR applications. MELCOR is currentlycapable of handling evaluations of transients in ELWRprimary systems and containments. It is also capable oftreating most features of ALWRs. MELCOR will bebenchmarked against results of detailed codes to deter-mine its adequacy and areas of development needed forALWR applications. In FY92, an activity is under way atBrookhaven to develop and test an input model for theWestinghouse AP600 design. Similar activity is under wayat ORNL for SBWR. Based on the results of the inputmodel effort and on initial runs with the model, additionalrecommendations will be developed for further code im-provements to integrate MELCOR capabilities forALWRs.


Within the context of the current research program, theresults of the MELCOR peer review, and the continuingplans for MELCOR development and assessment, theanticipated results of the MELCOR program are:

1. Development of a second-generation integrated se-vere accident code that should (1) appropriatelymodel phenomena essential to the understanding ofsevere core damage accidents, (2) provide predic-tions of the progression and consequences of severecore damage accidents, (3) permit estimates of theuncertainties associated with such predictions, and(4) have a structure that facilitates the incorporationof new or alternative phenomenological modelsbased on the ongoing experimental research pro-gram.

2. Provide a code that includes major phenomenologi-cal developments from severe accident research inadequate detail to address the phenomenology andalso has a practical running time for severe accidentanalyses.

3. Provide continuing maintenance and user supportfor the MELCOR code.

Within a few years, all the important models for severeaccident analyses will have been added to the MELCORcode and the code will have been assessed for its ade-quacy.

2.4.4 COMMIX


In the past years, research in the area of code develop-ment for reactor safety analysis was mainly focused ondeveloping and validating one-dimensional system codessuch as SCDAP/RELAP5 and MELCOR. However, anumber of phenomena encountered in postulated severeaccidents are inherently multidimensional in nature.Typical examples are: natural circulation, flow stratifica-tion, countercurrent flow in a pipe, hydrogen distributionand mixing processes, and the effect of noncondensiblegas distribution on local condensation and evaporation.

The unique features of the passive containment coolingsystem in the proposed Westinghouse AP600 plant arespecifically designed to prevent damage to the contain-ment during design-basis events or severe accidents. Thepassive cooling functions are carried out via natural circu-lation inside or outside the containment. The COMMIXcode (originally developed for analysis of transient fluid•flow and heat transfer in the reactor coolant system) isbeing modified so that it can be used for the analysis andunderstanding of transient fluid flow and heat transferphenomena in the containment. The research needs forthis program are to assess and evaluate the capability ofCOMMIX for use in analyzing the new and unique fea-tures of ALWR plants during design-basis events or se-vere accidents, and to apply the COMMIX code to per-form audit calculations.


The main objectives of this program is to improve and'extend COMMIX's capability and then in turn use COM-MIX to assess the adequacy of the unique passive contain-ment cooling system designed specifically to prevent dam-age to the AP600 containment system during design-basisevents or severe accidents.

To satisfy the objective above, some code assessmentefforts are being carried out, and a number of new modelsare being developed and implemented into COMMIX.The following is a list of new models, needed for theanalysis of the Westinghouse AP600 design:

1. Multicomponent capability to compute the distribu-tion of steam, air, and hydrogen throughout thereactor containment.

2. Liquid film tracking model to compute liquid filmthickness, velocity, and temperature on the internaland external containment steel shell to calculatetransient heat removal from the AP600 contain-ment. The AP600 steel shell consists of semi-elliptical domes at the top and bottom of a cylinder.The internal liquid film is formed as the result ofcondensation of steam on the liner wall of the steel

NUREG-1365, Rev. 1 40

2 Research Plan

shell. The external liquid film on the steel shell wallis formed from water flooding at the top of thedome. Evaporation of this liquid film then occurs asthe buoyancy-driven air stream passes through theannulus outside the steel shell.

3. Heat and mass transfer models which must be vali-dated with the Westinghouse AP600 Passive Con-tainment Cooling System (PCCS) small scale and1/8-scale test data.

4. Radiation model to account for heat transfer fromthe steel shell to the air baffle wall in the AP600design.

The multicomponent capability has been implemented inthe COMMIX code, and a limited validation of this capa-bility was carried out using steam blowdown data from afull scale vessel in Germany (i.e., HDR ISP-23). Furthervalidation effort is planned. The development of the liq-uid film tracking model has been completed. Both theliquid film tracking model, and heat and mass transfermodels implemented in COMMIX are being validatedwith the Westinghouse PCCS small scale data. The vali-dation effort with 1/8-scale test data will soon be carriedout. The need to develop a radiation model to treat theradiation from the dry patch of the containment steelshell to the baffle wall will be made in FY93.


With upgrading of the COMMIX code, it will be capableof performing containment analysis of ALWR for bothdesign-basis events :and severe accidents. Also, COM-MIX can provide multidimensional information and serveas a benchmarking tool for the CONTAIN lumped pa-rameter system code.

2.4.5 VICTORIA


The analyses for NUREG-1150 have shown that bypassaccidents, such as steam generator tube ruptures to be thedominant risk accidents for some classes of plants (sub-atmospheric PWRs .and ice condenser PWRs). There issubstantial uncertainty in the risk associated with theseaccidents because the accident analysis tools available forthe NUREG-1150 analyses omit models critical toproper calculation of radionuclide retention and revapo-rization in these sequences. The accident analysis toolsappear to underpredict retention in the reactor coolantsystem. The VICTORIA code was developed to addressthis uncertainty related to the bypass accident.

Current state-of-the-art models of core degradation indi-cate that much of the reactor fuel will remain in the vesseland the core region at the time of vessel failure. Tests in

Canada (CRNL) indicate that, especially for PWRs, radi-onuclide release from fuel retained in the vessel in an airenvironment will be radically different from release dur-ing early stages of core degradation. Plant configurationduring shutdown situations may also lead to the possibilityof air ingress into the core either by natural circulation orfrom the residual heat removal system. The experimentaldata has to be obtained for the fuel-release model ofVICTORIA before one could estimate the effects on riskassociated with the radionuclide releases of these types.

Deposited radionuclides may be resuspended in the reac-tor coolant system when the system is depressurizedeither as a natural event of the accident or as a deliberatemeasure to mitigate the possibility of direct containmentheating. VICTORIA has incorporated such a model. Cur-rently, there are no suitable experimental data to validaterevaporization model, but the PHEBUS-FP tests mayprovide some.

2.4.5.2 Current Research Plan

The VICTORIA code is developed under an interna-tional collaborating efforts including representativesfrom Sandia National Laboratories (SNL), Argonne Na-tional Laboratory (ANL), Oak Ridge National Labora-tory (ORNL), Battelle Columbus Laboratory (BCL),Chalk River Nuclear Laboratory (CRNL) and WinfrithTechnology Centre (WTC). SNL is the principal devel-oper of the code while other establishments are providingspecific model(s) in their area of specialty. The code and auser's manual was first released in October 1990. Anupdated version of the code has been completed in May1992, and the revised user's manual and the code arescheduled for release by the end of FY92. The VICTO-RIA code can now provide predictions adequate for reac-tor accident analyses of:

1. Radionuclide release during the early stages of core

degradation,

2. Aerosol processes in the reactor coolant system,

3. Vapor deposition in the reactor coolant system,

4. Radionuclide release during later phases of coredegradation, especially once the reactor vessel hasfailed,

5. Resuspension of deposited materials at the time ofreactor coolant system depressurization,

6. Radionuclide entrapment and revaporization in rup-tured steam generator tube accidents and other by-pass accidents.

Distinct classes of experimental data are being used tovalidate the various aspects of the code:

41 NUREG-1365, Rev. 1

2 Research Plan

1. Release During Core Degradation

Separate-effects tests such as the out-of-pile HI and VIunder steam and hydrogen conditions tests done atORNL and the ST-i, 2 in-pile tests have been used formuch of the validation of VICTORIA. French out-of-piletests (PHEBUS-SFD) and in-pile tests (PHEBUS-FP)with irradiated fuel will be examined in the future. Moreintegral tests such as the PBF tests, the LOFT-FP test,and the FLHT tests may be used for validation of therelease models once clearer portrayals of thermal-hy-draulic processes are available for these tests.

2. Release During Late Phases of Core Degradation

There are no suitable test data to validate models ofrelease once fuel has melted and the core geometry hasbeen lost. Integral experiments to study fission productrelease from late-phase core melt progression and rev-aporization are not being planned in the United States.However, the PHEBUS-FP project could provide someof the needed experimental data.

3. Transport in the Reactor Coolant System

Results of the Marviken and LACE tests are being usedto validate the aerosol transport models in VICTORIA.Results of the Argonne and Sandia National Laboratorytests as well as tests now under way in the United King-dom (WTC) are being used to validate models of chemicalprocesses affecting radionuclide transport. More integralvalidation will be provided by results of the PHEBUS-FPtests.

4. Revaporization of Deposited Radionuclides from theReactor Coolant System

There are not suitable data to validate the models ofrevaporization. The PHEBUS-FP tests may providesome.

5. Release and Transport during Shutdown or AfterVessel Failure

Separate-effects tests done at CRNL are available tovalidate some chemical aspects of models of release andtransport in the reactor coolant system after vessel fail-ure. In addition separate-effects tests on fission productin the presence of air will be performed at ORNL. Thereare no integral data on core degradation during this phaseof an accident.

In FY92, developmental assessment will be completedand documented. A peer review of the code will be initi-ated in FY93; after peer review systematic assessment ofthe code will be carried out. Meanwhile, the code is usedfor the planning of post-test samples analysis and pre-testcalculations for integral test for the PHEBUS-FP pro-

ject. In addition, VICTORIA is also used for the Interna-tional Standard Problem (ISP) 34 exercise for the FAL-CON fission product chemistry experiments in theUnited Kingdom.


The end product of this research is to provide NRC with acomputer code that models the radionuclide and non-radionuclide materials release, transport, and depositionwithin the reactor coolant system under severe accidentconditions. It will be capable of performing (1) plant (bothPWR and BWR) analysis for the in-vessel fission productbehavior for various severe accident scenarios, and (2) ex-perimental analysis and support for in-vessel fission prod-uct experiments. Specific uses of the code include(1) PHEBUS-FP post-test samples analysis; and pre- andpost-test analysis, and (2) FALCON ISP-34 exercise. Thecode is also being used by CRNL for the Blowdown TestFacility pre-and post-tests analysis and by WTC for as-sessment of full plant behavior under severe accidentconditions.

2.4.6 Integrated Fuel Coolant Interaction


In the event of a severe accident leading to core melt,molten fuel materials can come' into contact with water,producing a fuel coolant interaction (FCI). FCI's canoccur for a variety of in-vessel and ex-vessel conditions,including: reflood of a partially molten core, melt pouringinto the lower plenum, and melt pouring out of the vessellower head into a water-filled reactor cavity under low orhigh pressure. In these situations non-explosive or explo-sive FCI's may occur. The mode and resulting energeticsof FCI's depend on the complex interaction of variousthermal, physical, and chemical processes, including:coarse mixing, particle fine fragmentation, and heattransfer.

The goal of the Integrated Fuel Coolant Interaction(IFCI) effort is to provide a stand-alone code that embod-ies an integrated models for these processes so that FCIsevere accident events can be calculated for full scaleplants. Presently the IFCI code has phenomenologicalmodels for the pre-explosion mixing phase of FCI's, andincludes multi-phase, multi-dimensional, three fluid hy-drodynamic equations required to represent non-explo-sive events and the thermal detonation and expansionphases of explosive FCIs. Mechanistic models for trig-gers that initiate explosive FCI events are not included;however, there are provisions for representing intention-ally imposed triggers, such as those in FCI experiments.

The fundamental physics of FCI's are being experimen-tally investigated as described in Section 2.3 of this docu-ment. This work is largely focused on understanding trig-gers and how they affect FCI's, and the development of

NUREG-1365, Rev. 1 42

2 Research Plan

separate effects models. There is a need .for a code thatwill embody the models that result from this work andothers as our understanding of FCI's increase. This needstems from the fact that realistic modeling of FCI's re-quires that the complex interactions between the govern-ing processes be captured as well as the processes them-selves. The code is intended to fulfill the need for a tool toquantitatively describe non-explosive FCI's, and explo-sive events when the triggering mechanism is known.Such a predictive tool is needed to help answer questionsregarding the impact of FCI's for present and futurenuclear power plants.


FY92 and FY93 efforts will focus on making an IFCI astand-alone predictive tool for FCI events. PresentlyIFCI is a module of the inactive MELPROG systemscode. The thrust of the present program is to extract IFCIfrom MELPROG. Only these modules needed to repre-sent FCI's and the interaction of the governing processesof FCI's will be retained. This effort will include testing ofthe resulting stand-alone code and correction of any mod-els or algorithms required to obtain a robust analysis tool.The code will be documented in a draft NUREG reportand peer reviewed in FY93 prior to issuing the final re-port.


The anticipated result of this work will be a stand-aloneIFCI code that can be used on workstation to analyzenon-explosive FCI events, and explosive events when thetriggering mechanism is known. An operational reportthat demonstrates the applicability of the code to fullscale plants will be provided along with the peer revieweddocumentation report.

2.5 BWR Mark I Containment Liner

Failure


As stated in Appendix A. 1, the NUREG/CR-5423 peerreviewers identified three areas that warranted additionalresearch to confirm the appropriateness of the analysis inNUREG/CR-5423. These three areas are liner failurecriteria, melt superheat, and melt spreading phenomena.The staff has initiated necessary research in each of theseareas as discussed below.

2.5.1.1 Liner Failure Criteria

The NUREG/CR-5423 analysis assumed that the linerwould fail when the temperature of the liner reached themelting point of the liner steel. A member of the peerreview group suggested that the steel liner could dissolve

by chemical interactions with the melt prior to reachingthe steel's melting temperature. Further examination ofthis issue indicated that the dissolution and the data basecited by the peer reviewer are not applicable to this issue.It was also noted by several peer reviewers that thestrength of the steel liner would be greatly reduced atelevated temperatures. This, together with pressurizationof the drywell following vessel failure might produce acreep-rupture failure of the containment liner prior toreaching its melting temperature. An NRC contractor iscurrently performing a structural analysis of the linerunder these conditions to determine if creep-rupture fail-ure will precede liner failure by melting.

2.5.1.2 Melt Superheat

A question was raised as to whether the NUREG/CR-5423 analysis of melt superheat was based on the bestunderstanding of core-concrete interactions to date. San-dia National Laboratories performed the necessary con-firmatory calculations using CORCON MOD3. The re-sults indicate that the melt superheat duration inNUREG/CR-5423 is very conservative (i.e., CORCONMOD3 calculation of superheat duration is one-third ofthe values used in NUREG/CR-5423). Therefore, thethermal loading of the liner used in NUREG/CR-5423 isdeemed appropriate.

2.5.1.3 Melt Spreading Phenomena

A concern was raised that in underwater volcanic lavaflows, crusts have been observed to form at the lava-waterinterface, forming an annulus inside of a lava crust withinwhich molten lava could flow without interacting with thesurrounding water. The concern was that this phenome-non, should it occur, could insulate the molten core mate-rial flowing out of the pedestal region from the overlyingwater pool, and it also could preferentially channel theflow to the liner, eliminating the benefits of spreading.This is a complex problem not amenable to simple experi-mental resolution. The MELTSPREAD-1 code has beenused to perform melt spreading sensitivity analyses toaddress this issue.

2.5.2 Future Plans

The staff plans to incorporate the results of the above-mentioned confirmatory activities into a NUREG report.A final peer review of the NUREG report will then beconducted.

2.6 Hydrogen Combustion and

Research


Appendix A.3 describes the state of knowledge of hydro-gen combustion and transport and identifies processes

43 NUREG-1365, Rev. 1

2 Research Plan

that are not well understood. The NRC has recently spon-sored two programs for experimental investigation of is-sues that heretofore have received little attention. Whilehydrogen research has considered diffusive flame behav-ior, flame acceleration, and detonation behavior, the re-search to date has been limited to tests involving hydro-gen-air-steam mixtures under ambient or relatively low(1001C) temperature conditions. In the absence of reli-able ignition devices, auto-ignition of jets and plumesreleased at high temperatures during a severe accidentcould result in the continuous burning of hydrogen as it isreleased into the containment. For premixtures ofhydrogen-air-steam in containment at elevated tempera-tures (but below the auto-ignition temperature), flameacceleration and high-speed combustion may be morelikely to lead to a transition to detonation. Thus, ourcurrent and future experimental research is directed atconfirming the treatment of this behavior.

To address the effects of elevated temperatures on flameacceleration and detonation transition, the NRC initiateda high-temperature, high-speed hydrogen combustionprogram under a joint agreement (signed in June 1991)for a cooperative program with the Ministry of Interna-tional Trade and Industry of Japan and the Nuclear PowerEngineering Center. Under this agreement, a high-tem-perature high-speed hydrogen combustion research pro-gram, extending over 5 years, has been developed. Twocombustion vessels will be used for this research programat Brookhaven National Laboratory. The largest of thetwo vessels will be approximately 30 cm in diameter and20 m long. High-temperature hydrogen combustion mix-tures will be used with a pre-ignition temperature as highas 700 *K and an initial pressure of 1 atmosphere. The testgases that will be used will be mixtures of hydrogen, air,oxygen, nitrogen, steam, carbon dioxide, and carbon mon-oxide. The facility will be able to accommodate testingwith and without venting. The smaller of the two vesselswill be approximately 10 cm in diameter and 6.7 m long.This vessel will permit use of the apparatus for learningthe effects of high temperature on the smoked foils andtesting the instrumentation. It will also provide prelimi-nary data to assess the Shepherd/ZND model for calcu-lating cell size.

In the low-speed hydrogen combustion research program,the aspects of diffusion flames scalability and transienthigh-temperature combustion will be investigated. Theresults will be used to help resolve outstanding issues insevere accidents, i.e., hydrogen combustion aspects ofDCH; high-temperature combustion phenomena, anddetonation initiation by high-temperature steam-hydro-gen-particle laden jets. These experiments will use a tech-nique of combusting premixed rich and lean mixtures inseparate vessels followed by deliberate mixing to initiatecombustion. This will enable the creation of much highertemperatures (1000-3000°K) than could be readily ob-

tained by conventional heat exchanger techniques. Thenatural heat transfer processes within the explosion ves-sels and the timing of combustion and mixing processeswill be used to vary the temperatures in the gases. Dis-persing metal or inert dusts within one vessel prior tocombustion will allow the creation of particulate-ladenatmospheres, which will simulate the process of high-pressure melt ejection during a DCH event. Pressure andtemperature measurements and photographic recordingswill be used to determine the nature of 'the combustionphenomena.

Evaluation of hydrogen transport in reactor containmentsremains a long-term research issue owing to a limited setof experimental data and thus limited validation of exist-ing codes used in such analysis (i.e., CONTAIN, HMS).Current and near-term future activities include a modestprogram to document the development and assessment ofthe HMS code. The staff also expects to benefit fromDOE-sponsored research to upgrade the HMS code suchthat the code may be readily applied to different configu-rations; the earlier version of HMS used in NRC-sponsored research, an analysis of plume mixing and dif-fusion flame combustion of Mark III reactors, requiredcode modification for different geometries. At the com-pletion of the DOE-related activity and our sponsoredactivity to document the code, the NRC will be betterpositioned to apply the code for reactor analysis and vali-dation. With regard to the CONTAIN code that utilizesthe traditional control volume technique for containmentanalysis, validation of the flow and mixing models againstHDR project data is continuing with analysis of the inter-national standard problem ISP-29.

As a result of our recent cooperative agreement withJapan in the area of hydrogen research, the NRC now hasaccess to ongoing hydrogen mixing and distribution test-ing in the Tadotsu facility, a large-scale mixing and distri-bution test facility that simulates at one-fourth scale aPWR' 4-loop reactor containment 'with compartmen-talization. Additional data from hydrogen mixing andcombustion testing in the Takasoga facility, which roughlysimulates a RESAR SP-90 type design,will also becomeavailable during FY92-93. Test results from these facili-ties will provide a greatly expanded and improved database for the validation of our analytical tools. It is antici-pated that long-term confirmatory research will focus onthis newly available data.

An important supplement to our hydrogen research pro-gram is that work being carried out under an arrangementwith the Russian Academy of Sciences in collaborationwith the I.V. Kurchatov Institute. Under this cooperativeprogram, the HMS code, which has been used to predicthydrogen mixing and combustion, will be assessed andcompared against Russian experimental data. Under thisagreement, the NRC is also provided with the results ofKurchatov hydrogen deflagration test data and the results

NUREG-1365, Rev. 1 44

2 Research Plan

of a program to develop a scaling relationship for thecritical conditions of turbulent jet initiation of a detona-tion. This research includes elements of experimentation,numerical simulation and theoretical analysis. An im-proved understanding of these phenomena will enable amore definitive evaluation of the detonation potential tobe performed for both U.S. and Russian nuclear powerplants.

2.7 Source Term

2.7.1 Status

At the present time, the NRC is pursuing several regula-tory initiatives to incorporate insights from updated se-vere accident source terms. Updated source term insightsarising from the technical update of TID-14844 are ex-pected to be made available for voluntary use by existinglicensees. A revision of 10 CFR Part 50 to incorporateupdated source term and severe accident insights willthen be undertaken, with a proposed rule for commentexpected to be issued by early CY93. Although regulatorypositions arising from updated source term insights re-main to be developed, some preliminary implications canbe seen at this time. It is clear that updated source terminsights indicate the need for consideration of nuclides(e.g., cesium) in addition to iodine and the noble gases. Inaddition, revised insights on iodine chemistry call intoquestion the need for high-efficiency charcoal absorbers(assuming that the pH is controlled, post accident). Theiodine chemistry can, in turn, impact such important plantsystems as fission product cleanup systems, control roomhabitability, and allowable containment leak rate. Finally,and most importantly, the above discussion and all recentrisk studies have shown the importance of maintainingcontainment integrity under severe accident conditions inorder to ensure low risk. This strongly suggests that theappearance of a severe accident source term within con-tainment and challenges associated with such releasesshould be more closely linked with the temperatures,pressures, and containment loads, rather than an arbi-trary linkagewith a single sequence such as a large-breakloss-of-coolant accident. The disconnect in present prac-tice is not so much that temperatures and pressures camefrom only one sequence, but that they came from a se-quence that was terminated Without perceptible damageto the core, since the analysis that gave those tempera-tures and pressures was required to show that the peakclad temperature remained below 2200'F. The revisedsource term, is believed to be consistent with the sourceterm expected to release into the containment resultedfrom core melt under low pressure severe accident se-quences. It also provides the basis for the evaluation ofthe effectiveness of containment mitigation features un-der severe accident conditions. The revised source term isnot intended to address by-pass accident source term,

since the release is direct to the environment. However,analytical tool like VICTORIA has been developed spe-cifically (section 2.4.4.) to address source term for by-passaccidents.

Although additional physics and chemistry research canbe performed to reduce uncertainties in source term phe-nomena, it is important to consider the need for suchresearch and the potential that this research could signifi-cantly improve our risk perspective on severe accidents.Hence, future research is oriented to assess the NRCsevere accident codes and to address residual source termissues related to plant configurations during shutdownsituations which would lead to the possibility of air ingressinto the core either by natural circulation or from theresidual heat removal system. The air will interact ex-othermically with the cladding remaining on the fuel,producing high temperature in the fuel. Vapor of ruthe-nium and molybdenum will be produced because of thestrongly oxidized conditions. A test will be conducted atORNL at these cbnditions in FY93 if the ongoing riskstudy sponsored by NRC indicated that accidents duringshutdown conditions are major contributor to the coredamage frequency. The NRC's participation in thePHEBUS-FP project is to obtain integral effects experi-.mental data to validate NRC severe accident codes.

2.7.2 PIIEBUS-FP

2.7.2.1 Objective

The objective of the PHEBUS-FP Project is to performintegral effects experiments in an in-pile test facility, un-der sufficient prototypical conditions, on the processesgoverning the transport, retention, and chemistry of fis-sion products under LWR severe accident conditions.The processes to be investigated are those taking place inthe core region, in the reactor coolant system, and in thecontainment building. The experiments will also study thedegradation of high burnup fuel, typical of the laterphases of the accident.

The Commissariat a l'Energie Atomique (CEA) ofFrance and the Commission of the European Communi-ties agreed to undertake the PHEBUS-FP Project inclose collaboration, using the experimental facilitiesavailable at the Research Centre of Cadarache, France.NRC entered into an agreement with CEA4 under thisagreement fission product generation; transport anddeposition generated in the PHEBUS-FP program willbe made available to the NRC.

2.7.2.2 Facility Layout and Testing

PHEBUS-FP is a loop-type test reactor with a low-enriched driver core of 20 to 40 MW power, using fuel rodelements. Core cooling and moderation is achieved bydemineralized light water, and light water and graphiteare used as reflectors.

45 NUREG-1365, Rev. 1

2 Research Plan

A cluster of 20 fuel rods, 1 m long, in a PWR configura-tion, is inserted in a test train and located in the centralhole of the driver core of the PHEBUS-FP reactor.

Before a test, the test fuel from the BR3 reactor (a Bel-gium reactor that use 1 m-long fuel rods) is re-irradiatedin the PHEBUS-FP in-pile section for 2 weeks using theexisting pressurized water loop in order to generate asufficient inventory of short- and medium-lived fissionproducts. The loop is then slowly blown down with simul-taneous reduction of the reactor power, with the in-pilesection isolated from the loop. After these steps, testingmay begin. During the test phase, the in-pile section isconnected to a circuit and vessel that simulate the primarycircuit and containment building of a PWR.

During the test phase, the in-pile fuel bundle is heated byfission power from the driver-core at a rate typical of asevere accident up to temperatures at which the fuel isdamaged. The test bundle is pushed to conditions inwhich fission product release takes place, and controlrods and structural materials are vaporized, producingsufficient quantities of aerosols. The fuel bundle will bedamaged to the extent necessary not only to release fis-sion products and aerosols, but also to study the mechani-cal behavior of the fuel during extensive degradation.

The released fission products and aerosols are swept by aflow of steam and H2 into the circuit that simulates theprimary cooling system up to the point of pipe break.Then the flow enters a vessel that simulates the contain-ment building.

2.7.2.3 Instrumentation

Instrumentation is used for process control, safety, andinterpretation of the experimental results (scientificanalysis). During the design of the PHEBUS-FP facility,a large effort was devoted to the instrumentation forscientific analysis. The success of the PHEBUS-FP ex-periments depends largely on the capability for measur-ing the parameters of interest.

The fuel bundle region will be instrumented with tem-perature, pressure, and flow sensors. The primary RCScircuit will be provided with two main instrumented sec-tions: one in front of the large components (steam gen-erator, pressurizer, etc.) and one just before entering thecontainment vessel. The main measurements of interestare (1) thermal-hydraulic conditions, (2) fluid composi-tion, especially the fission product content, (3) aerosolconcentration and granulometry, (4) deposits on the pipewall, and (5) composition of the gas phases, including H2 .These will also be measured in the containment vessel atseveral positions in order to characterize their spatialdistribution. In addition, the sump water will be moni-tored with respect to dose rate, isotope concentrations,

etc. An extensive program for post-test examination, toback up and complement the on-line measurements, isbeing put in place at Cadarache with collaboration fromseveral qualified laboratories of the European Commu-nity.

2.7.2.4 Test Matrix

The main objective of PHEBUS-FP is to obtain integraleffects experimental data on fission product transport inthe RCS and in the containment. This implies studies onretention and revaporization of fission products in theRCS. Revaporization could be enhanced either by decayheating or by a sudden steam spike in the circuit. Theformation of gaseous species in the containment is also ofa great importance. Studies in PHEBUS-FP are expectedto yield information related to volatile iodine comingfrom radiolysis of the sump or organic iodine from paintsin the containment.

The current test matrix consists of six tests. The intent ofeach test is to capture the key processes and phenomenaassociated with a particular severe accident sequence(e.g., V sequence) and not to simulate a sequence indetails. The first test is scheduled for 1993, and subse-quent tests are scheduled at yearly intervals.

2.7.3 Other Research Activities

In addition to the PHEBUS-FP project, the followingresearch activities are planned.

a In FY92, complete the analysis of the ORNL VI-6fission product release test and a technical report onthe analysis and interpretation of the fission productrelease experiments performed at ORNL. BeyondFY92, additional tests (VI-7) will be conducted toinvestigate fission product release at high tempera-ture for (1) fuel exposed to air under shutdown con-ditions or residual fuel remaining in the reactor coreafter vessel bottom melt-through, and (2) highburnup (>60 MWd/kg) and low-power-densityALWR fuel.

In FY92, complete model development and docu-mentation for the VICTORIA code, and validatethe code against test data. In FY93, complete a peerreview of the VICTORIA code. Further code im-provement or development will depend on the out-come of the peer review. The code will also be usedfor PHEBUS-FP pre- and post-test analysis and forthe International Standard Problem exercise on theFALCON project in the Winfrith Technology Cen-ter (WTC) U.K. Benchmarking of the CONTAINcode will also be carried out with the PHEBUS-FPdata.

NUREG-1365, Rev. 1 46

2 Research Plan

* A small experimental program will be continued atB3attelle to provide the capability to utilize massspectrometry under various oxidation states (reduc-ing and oxidizing) at different temperatures(10006 K to 2700-K) to obtain chemical speciation

and thermochemical data for a combination of mate-rials (fuel, reactor structural materials, etc.).

Appendix B.1 provides a more detailed discussion onsource term issues.

47 NUREG-1365, Rev. 1

3 RESEARCH PLAN FOR ADVANCED LIGHT WATER REACTORS

3.1 Introduction 3.2.1.1 Research Approach

The role .of severe accidents, despite their exceedinglylow probability, has been recognized for consideration inadvanced light water reactors (ALWR). In fact, the im-proved ALWR design features that reduce the likelihoodand consequences of severe accidents, and the excellentsafety record that continues to accumulate from operat-ing plants, leaves only such exceedingly low-probability,high-consequence, events to be of concern. The objectiveof the ALWR severe accident research program is toexamine the issues in depth and develop the necessarytools to address these issues. While it is recognized thatthe Westinghouse AP600 and the GE SBWR are suffi-ciently different from existing LWRs that different severeaccident issues or variations on existing issues may exist,much of the phenomenological understanding of accidentprogression, containment loading, and source term analy-sis developed in relation to existing reactors will alsoapply to ALWRs. The readiness and applicability of thesevere accident codes (e.g., SCDAP/RELAP5, MEL-COR and CONTAIN) to ALWR plant-specific designsand phenomena must be addressed. Important phenom-ena addressed in accident progression analyses includetiming of core melt and vessel breach, in-vessel hydrogenproduction, fission product transport and behavior, fuel-coolant interactions, release of fuel from the vessel, core-concrete interactions, hydrogen burns, and containmentloading. We will assess the state of knowledge for phe-nomena that arise in ALWR designs, that have not hadimportance for severe accidents in current plants. Con-tainment heat transfer, mixing, and hydrogen distributionin the passive containment are examples. As discussed inSection 2.4 of this report, NRC's severe accident codeswill be modified to be capable of analyzing ALWR de-signs, and input decks will be developed specifically forthe AP600 and SBWR designs.

This plan is based on an assessment of the applicability ofcurrent knowledge of severe accident phenomena to theALWR designs. The plan identifies studies needed forspecific ALWR designs, where the reviews for currentreactors may not be sufficient. We do not foresee a needof new experimental research beyond what has been ac-complished or is underway for current reactors.

3.2 In-Vessel Severe AccidentPhenomena

3.2.1 Core Melt Progression and ReactorVessel Lower Head Failure

Assess the methodologies used to develop or evaluate theeffects of phenomena resulting from ALWR fuel andcore designs on severe accidents, and the effect of designfeatures (i.e., cavity flooding) to prevent lower head fail-ure.

3.2.1.2 Discussion

The ALWR fuel design typically involves a lower powerdensity than current reactor designs. For example, theAP600 plans to employ a power density of 73 to 79 kw/liter compared to 100 to 110 kw/liter for current PWRs. Alower power density results in an increase in the mass ofthe active core for a given power level, and correspond-ingly, the mass of zircaloy as well. This lower power den-sity provides additional thermal margin during transients,and there is the potential to form a relatively thick protec-tive oxide layer that could delay the onset of fuel melting.Nevertheless, if the core does melt, there is the potentialfor greater H2 generation owing to the greater amount ofzircaloy present. The ability to evaluate the effect of thisadditional non-condensable gas should be assessed, in-cluding but not limited to its effect on events involvingsteam generator tube rupture and events with assumedcredit for natural circulation in the primary system. Al-though there is no reason to expect any new or additionalphenomena to be important to in-vessel severe accidentscenarios as a result of this fuel design, natural circulationeffects may need to be modeled more accurately in ouranalytical codes such as MELCOR and SCDAP/RELAP5. Hence, adequate assessment and documenta-tion of these codes are vital to provide the confidence inusing them for ALWR analyses. Other than the lowerpower density, there appear to be no major differences inthe core design between the present generation reactorsand the ALWRs. However, some differences in the corestructures and materials do exist (e.g., stainless steel re-flector rods at core periphery, flow distribution grid, ex-tended burnup fuels, and burnable poison rods inAP600). The severe accident code SCDAP/RELAP5 willbe modified to account for these design differences. Ourcurrent understanding, with the knowledge that will beacquired under the revised severe accident research plan,is adequate to address core melt progression for ALWRs.

The AP600 design includes the capability to, flood thereactor cavity during a postulated severe accident in orderto cool the reactor vessel with an external water pool priorto core debris penetration of the vessel. The SBWR de-sign also includes a system to flood the cavity. Analyses ofpotential lower head vessel failure for ALWR designs willneed to consider the impact of accident managementstrategies for flooding the cavity.

49 NUREG-1365, Rev. 1

3 Research Plan

3.2.2 Fission Product Transport andBehavior

The existing data base on fission product release shouldbe applicable to the new low power density fuel used inadvanced reactor designs. When the current assessmentand development plan for the VICTORIA code is com-pleted, the VICTORIA code will be adequate to addressthe fission product transport and behavior in the reactorcoolant system for ALWRs. Since ALWRs rely heavily onnatural circulation in the primary system for heat removalduring a severe accident, the effects of natural circulationon fission product transport and. behavior could be as-sessed with codes such as SCDAP/RELAP5 and MEL-COR, provided the assessment discussed in Section 3.2.1above revealed no major deficiency.

3.3 Ex-Vessel Severe Accident

Phenomena

3.3.1 Research Approach

Assess the methodologies that are used to evaluate designfeatures incorporated in ALWRs to mitigate the effectsof core-concrete interactions, high-pressure melt ejectionand DCH, and hydrogen combustion, including but notlimited to methodologies that evaluate cavities designedto mitigate the effects of DCH and methodologies thatevaluate cavity floors designed to provide sufficient areato allow coolable debris geometry.

3.3.2 Discussion

Within this context, the assessment is on the methodolo-gies used to evaluate design features incorporated in theALWRs to mitigate the effects of core-concrete interac-tion, high-pressure melt ejection and DCH, and hydrogencombustion. It includes, but is not limited to, methodolo-gies that evaluate cavities designed to mitigate the effectsof DCH and methodologies that evaluate cavity floorsdesigned to provide sufficient area to allow a coolablegeometry.

Another area for assessment is evaluating the new con-tainment cooling concepts, including passive natural cir-culation cooling using air and containment dome coolingusing water sprays.

3.3.2.1 Molten Core-Concrete Interactions

The available knowledge for predicting molten core-concrete interactions (MCCI) appears to be both applica-ble and adequate for ALWR accident analyses. The as-sessment of the criterion that core debris be spread overan area of 0.02 m2/Mw thermal to assure coolability needsto be evaluated. This is part of the current severe accident

research program (i.e., ACE consortium MACE tests atANL and the WETCOR tests at SNL).

The SBWR cavity design was indicated to be similar to theABWR design. In this design, molten debris in the lowercavity is allowed to spread to a shallow bed and then bequenched by water from thesuppression pool. The appli-cability of the MACE data to the SBWR design should beevaluated. It is recognized that there are differences be-tween the experiments and the SBWR design (e.g., debrisdepth and composition).

3.3.2.2 Direct Containment Heating

The AP600 has incorporated an automatic depressuriza-tion system for the RCS and eliminated instrument pene-trations in the bottom head of the reactor pressure vessel.These design changes, however, do not completely elimi-nate the DCH threat because repressurization of theRCS during core degradation, particularly as a result ofcore debris-coolant interaction, is still possible. Currentresearch on lower head failure mechanisms and DCHphenomena are expected to provide information thatwould allow the NRC to provide a preliminary assessmentof the mitigative features of the ALWR designs. How-ever, final conclusions on the retentive capabilities of aspecific cavity design to preclude DCH may require fur-ther attention depending upon the credit required forsuch features.

For the'SBWR, the low primary system pressure, theinerted containment atmosphere, and the passive de-pressurization system should significantly reduce thethreat of a high-pressure melt ejection. Again, the meth-odology being developed in the current research programis adequate for application to SBWRs.

3.3.2.3 Hydrogen

3.3.2.3.1 Research Approach. The methodologies that areused to evaluate acceptability of the vendors' proposedhydrogen concentration criteria will be assessed, includ-ing evaluation of the maximum allowable H2 concentra-tion and the hydrogen control features proposed withinthe containments.

3.3.2.3.2 Discussion. For AP600, the current design crite-ria proposed by Westinghouse for hydrogen is to have acontainment volume large enough that the bulk averagehydrogen concentration could not exceed about 13%,assuming good mixing and reaction of 100% of the activecladding. Local accumulation of hydrogen will be con-trolled by dc-powered igniters. Westinghouse's choice of13% as a maximum allowable hydrogen concentration ismost likely based on existing correlations, which indicatethat this is the highest practical concentration for whichstable detonations could be ruled out. However, a con-centration of 13% is too high to avoid combustionaltogether. If we assume the containment volume is

NUREG-1365, Rev. 1 50

3 Research Plan

limited to about 100°C by steam condensation on thecontainment dome, incomplete combustion can occur forhydrogen concentrations as low as 4%.- Flame balls cangrow, rise to the ceiling, and quench, leaving a stratifiedregion with hot combustion products along the ceilingandunburned hydrogen in air below. Rapid, almost complete,combustion can occur for hydrogen levels above about8%, which would leave little unburned hydrogen. For acontainment atmosphere containing some water vapor,stable detonations are highly unlikely for 13% hydrogenand below. Accelerated flames with damagingIoverpres-sures could occur inside this containment (-10 percent ormore H2) in cluttered regions that promote flame accel-eration, but when the flame transits into the region ofopen containment volume, it may decelerate again. Someresidual shocks may strike the containment, but with farless energy than for a full-containment detonation. TheNRC position taken on ALWRs is that the containmentdesign should limit the hydrogen concentration. to nogreater than 10%, and that containment-wide hydrogencontrol should be provided to preclude the formation oflocal detonable mixtures and lessen the accumulation ofhydrogen on a global basis.

Although the above criterion goes a long way towardmitigating the hydrogen problem, two problems remainto be addressed. They are (1) loads from low-pressurecombustion, and (2) local detonation. It appears that low-pressure combustion does not generate overpressuresadequate to threaten the containment. With 13% hydro-gen, overpressures of 4 to 7 bars may be expected, de-pending on the initial temperature and steam concentra-tion. Presumably, the containment can withstand thisoverpressure.

To address the local detonation problem, Westinghouseis depending on dc-powered hydrogen igniters to burn thehydrogen early. If the igniters are successful at triggeringcombustion before the hydrogen concentration builds upto a locally detonable level, even local detonations can beavoided. This depends on the number and reliability ofthe igniters as well as the strategic placement of theseigniters. While the reliability of the igniters could beaddressed, the number and placement of them requiresthe prediction of hydrogen concentration, stratification,and distribution in the containment. Because the passivedesigns lack active mixing capability, assessment of con-tainment mixing becomes a more challenging task. Theexisting codes used for containment analysis utilize a con-trol volume approach, and the constituents within eachvolume are assumed to be well mixed. This may be inade-quate to address the issue on hand. Special flow fieldmodels, e.g., HMS may be used to address this issue.

Current design criteria proposed by GE would inert thecontainment region with nitrogen. Thus for the SBWR,all the hydrogen generation issues appear to have beendealt with by assuming that inerting the containment en-

vironment will take care of any problems. One unresolvedissue for SBWR is the treatment of noncondensable gasesin the isolation condenser and the passive containmentcooling system (PCCS). Additional work may be foundnecessary to assess the hydrogen effect on the isolationcondenser and PCCS performance.

3.3.2.4 Containment Cooling

3.3.2.4.1 Research Approach. The methodologies that areused to evaluate new containment cooling concepts willbe assessed, including natural circulation cooling using airand containment dome cooling using water sprays.

3.3.2.4.2 Discussion. For certain classes of accidents, theAP600 is expected to exhibit performance characteristicsthat are different from existing PWRs. These phenomenainclude,

(1) Natural circulation within the containment,

(2) Hydrogen mixing and development of high localconcentrations,

(3) Natural circulation cooling of the containment, and

(4) Water cooling of the external shell of the contain-ment.

For the external shell, flow patterns and correct heattransfer correlationsare needed for water cooling of theexternal shell. Inside the containment, the modeling ofnatural circulation and mixing processes over large vol-umes will pose a challenge to existing codes. For theSBWR, the behavior of the suppression pool can be read-ily modeled. Like the AP600, for certain accidents theSBWR may pose new challenges in modeling naturalcirculation and mixing processes for existing containment,analysis codes.

Natural circulation flow and related mixing processes arekey issues for both the AP600 and the SBWR designs. Forthe AP600, this will primarily be a containment issue. Forthe SBWR, it will be an issue for the passive core coolingsystem and the containment long-term heat removal sys-tem, since they are closely related.

Existing NRC severe accident codes CONTAIN andMELCOR are control volume (i.e., lumped-ýparameter)codes that do not take into account finite gas velocities ormomentum convection in the control volumes, or cells.Momentum is considered in the junctions, or flow paths,between cells, but this momentum is considered to becompletely dissipated in the downstream cell. The atmos-pheres in the control volumes are consequently assumedto be stagnant. Such modeling is appropriate for largecontrol volumes connected by flow paths whose cross-sectional area is small compared to the cross-sectionalarea of the control volume itself. In certain situations,

51 NUREG-1365, Rev. 1

3 Research Plan

however, the neglect of velocities and momentum con-vection within control volumes is not justified, and thedistribution of flows can be adversely affected by thisassumption.

Since the ability of the control volume codes to calculatethe distribution (i.e., stratification) of hydrogen is limitedin some cases, and since the principal NRC severe acci-dent analysis codes are all control volume codes, a betterunderstanding of the limitations of these codes and possi-bly improved modeling capabilities based on that under-standing could be important with respect to resolvingsevere accident issues.

It is possible to modify lumped-parameters codes by im-plementing velocities within the control volumes. TheCOMMIX code (see section 2.4.4) is being modified toevaluate ALWR designs for mixing characteristics.

In addition to the general issues of mixing and naturalcirculation, the ability to evaluate ALWR designs willstrongly depend on understanding natural convection andsteam condensation processes, particularly in the pres-ence of large quantities of noncondensables. The vendorshave proposed testing to address the question.

3.3.2.5 Fission Product Transport and Behavior

The existing knowledge on source terms is applicable tothe AP600 and the SBWR. Clearly, source term predic-tions will be driven by containment behavior, particularly

aerosol transport, deposition, and suppression poolscrubbing. The CONTAIN code will be adequate to ad-dress the source term in the containment. Given that theissues related to natural circulation and mixing processesare adequately addressed, the existing methods for sourceterm analysis are adequate.

3.4 SummaryWestinghouse and GE have not presented the NRC withdetails of their severe accident research program or theextent to which vendors' research Will satisfactorily ad-dress issues related to severe accidents. The NRC is plan-ning to hold meetings with the vendors to understandtheir research programs. We will then develop a programto resolve all severe accident licensing issues. In consulta-tion with NRR, RES will identify the research programthat will be carried out by NRC. We do not foresee a needfor new severe accident phenomena experimental pro-grams. However, NRC will develop analytical capabilitiesto independently assess the vendors' analysis. To accom-plish this, we plan to:

1. Complete the research delineated in this updatedreport.

2. Perform assessment or modification and documen-tation for NRC codes (MELCOR, SCDAP/RELAP5, CONTAIN, COMMIX, VICTORIA) sothat they could be used to perform analyses to auditlicensee calculations.

NUREG-1365, Rev. 1 52

APPENDIX ASEVERE ACCIDENT ISSUES, STATUS, AND PROGRESS TO DATE

Introduction Table A. 1 Severe Accident Research Program Issues

During the past few years, the severe accident researchprogram has focused on generating information that cannarrow the broad range of uncertainties that have beenidentified (e.g., NUREG-0956, "Reassessment of theTechnical Bases for Estimating Source Terms, 1986," andthe Kouts report, NUREG/CR-4883, "Review of Re-search on Uncertainties in Estimates of Source Termsfrom Severe Accidents in Nuclear Power Plants, 1987") aslimiting the ability to accurately calculate source termsand to provide an improved degree of assurance in esti-mating risks from severe accidents. The original eightareas of uncertainty in NUREG-0956 were:

1. Natural circulation in the reactor coolant system,

2. Core melt progression and hydrogen generation,

3. Steam explosions,

4. High pressure melt ejection,

5. Core-concrete interactions,

6. Hydrogen combustion,

7. Iodine chemical form, and

8. Fission product revaporization.

In the "Revised Severe Accident Research ProgramPlan," NUREG-1365, published in August 1989, similarareas of research were combined (e.g., iodine chemicalform and fission product reevaporization), whereas otherareas that are either important to accident scenarios thatmight lead to early containment failure (e.g., Mark I con-tainment shell melt-through and direct containmentheating) or are important to assessment of accident man-agement strategies (e.g., adding water to degraded core)were presented separately, and research needs were iden-tified. Also in NUREG-1365, research plans addressingthe issue of scaling of severe accident experiments werepresented.

Appendices A.1-A.5 provide summaries of the severeaccident issues listed in Table A.1, along with their cur-rent status, progress to date, and future plans. Appendi-ces B.1-B.3 provide the technical bases for closure of theissues related to source terms, core-concrete interactions,and scaling methodology. Note that Items 4 and 8 are not"issues" in the same sense as the other issues that repre-sent technical phenomena associated with severe acci-dents; however, they are major programmatic areas in thesevere accident research plan, and therefore, are listedand discussed as separate items.

Appendix

1. Mark I Containment Shell Melt-through A.12. Core Melt Progression and Hydrogen

Generation A.2

3. Hydrogen Transport and Combustion A.3

4.

5.

TMI-2 Vessel Inspection Program

Fuel-Coolant Interactions andDebris Coolability

A.4

A.5

B.1

B.2

B.3

6. Source Terms

7. Core-Concrete Interactions

8. Severe Accident Scaling Methodology

A.1 Mark I Containment ShellMelt-through (Liner Failure)

A.1.1 Background

An accident sequence leading to early containment fail-ure has been postulated for BWR Mark I containments.This sequence involves the direct attack of the contain-ment steel liner by molten core material following vesselfailure. In SECY-89-017, "Mark I Containment Per-formance Improvement Program," the staff addressedthe issue of severe accident challenges to the Mark Icontainment and proposed a balanced approach utilizingaccident management and mitigation as the optimum wayto reduce overall risk in BWR plants with these contain-ments.

SECY-89-017 stated, "there is a growing consensus thatwater in the containment (from an alternate supply to thedrywell sprays) may help mitigate risk by fission productscrubbing and possibly by preventing or delaying contain-ment shell melt by core debris. Research is continuing inorder to confirm and help quantify these initial conclu-sions."

NRC research over the past several years has addressedkey phenomena associated with the liner meltthroughissue, such as melt conditions at the time of vessel failure;melt spreading characteristics; thermal-hydraulic charac-teristics of molten core-concrete interactions both withand without an overlying water pool; heat transfer charac-teristics at the interface of the molten core, overlyingwater pool, and liner; and fission product attenuation inthe presence of an overlying water pool.

A--1 NUREG-1365, Rev. 1

Appendix A

Integration of the research information derived fromthese programs into an assessment of the conditionalprobability of liner failure both with and without an over-lying water pool in the drywell, given a core melt accidentthat proceeds to vessel failure, was completed. A descrip-tion of this methodology and its conclusion is provided inNUREG/CR-5423. In summary, by developing probabil-ity distributions for important parameters that factor intothe analysis from data (where available), computer analy-ses, or other insights, developing causal relationships be-tween phenomena, and convoluting these distributionfunctions and causal relationships, the authors ofNUREG/CR-5423 obtained estimates of the likelihoodof liner failure both with and without a water pool overly-ing the molten corium in the drywell. When water wasassumed to overlie the molten core material as it spreadson the drywell floor toward the containment liner, it wasconcluded that the liner failure would be physically un-reasonable. In the absence of water, however, the sameconservative approach led to the conclusion that failurewould be certain.

A.1.2 Status

NUREG/CR-5423 has been subjected to an extensiveand thorough peer review. Nineteen experts knowledge-able in the subject matter were asked to review the analy-ses. In addition to providing detailed written comments,the peer reviewers were given the opportunity to discusstheir comments with the authors in a public workshopthat was also attended by members of the Advisory Com-mittee on Reactor Safeguards and the Nuclear SafetyResearch Review Committee.

The main conclusions of the peer review and the work-shop were:

1 The methodology employed in NUREG/CR-5423was considered basically sound-no major deficien-cies or problems were identified that would invali-date the results.

2 The sensitivity study performed in NUREG/CR-5423 showed there was no single process orparameter that had a controlling influence on theoverall failure probability.

3 There was also a general consensus by the peer

review group that three areas warranted additionalresearch to confirm the appropriateness of theanalysis in NUREG/CR-5423. These three areasare liner failure criteria, melt superheat, and meltspreading phenomena. The staff's plan to conductthe necessary research in each of these areas is dis-cussed in Section 2.5.

For each of the three areas, a small (three or four) groupof experts was convened, many of whom served on thelarger peer review panel, to assist the NRC in addressingthe concerns. In addition, in order to ensure that theinitial melt quantity and composition were appropriate, ameeting was held to assess the adequacy of the report'squantification of the initial melt conditions at vessel fail-ure. Additional analyses were performed using theAPRIL-MOD3 code in which the heat-up, collapse, andmelting of the BWR steam separator or dryer were explic-itly modeled. The basic conclusion was that the boundary.conditions used in NUREG/CR-5423 seem to be ade-quate.

NUREG/CR-5423 identified three scenarios that couldchallenge the containment integrity. Scenario I is basedon an initial sudden massive core slump (50% of the core)leading to localized lower head failure. Scenario II isbased on initially quenched debris and a subsequent local-ized lower head failure owing to water depletion andremelting of the debris in the lower plenum. Scenario IIIis similar to Scenario H, except that the debris heats upthe lower head uniformly, resulting in weakening andeventual creep failure. Detailed analyses were performedfor Scenarios I and II. For the NUREG/CR-5423 peerreview, the peer reviewers were informed that the evalu-ation of Scenario III would await the completion of theNRC lower head failure analysis program at Idaho Na-tional Engineering Laboratory. This program addressesthe likelihood of creep rupture of the lower head as wellas other potential failure modes of the lower head. TheINEL program is now complete, and draft NUREG/CR-5642 has been issued. The preliminary results indi-cate that for a depressurized reactor vessel, global vesselfailure is not likely to occur. Therefore, the NRC is notplanning to perform any analyses to qualify the potentialfor containment shell meltthrough resulting from acci-dents that follow Scenario III.

While there are residual issues related to the uncertaintyof analysis in NUREG/CR-5423 in predicting liner integ-rity in the presence of water, it is generally recognizedthat the presence of water will sharply attenuate themagnitude of aerosol production and radionuclide re-lease. Aerosol production is affected by water becauseaerosol-laden gases produced by the core-debris attack onconcrete must sparge through the water. A recently com-pleted NRC-sponsored study investigated the detailedprocesses involved in aerosol trapping by water pools anddeveloped a simple model of water effects on aerosolgeneration during core-debris interaction with concrete.The model provides the probability distribution functionsfor decontamination factor (a measure of aerosol trap-ping) based on the uncertainty analysis using the MonteCarlo simulation technique. Bounding case calculationsusing this simple model indicate that water has a

NUREG-1365, Rev. 1 A-2

Appendix A

profound mitigative effect on aerosol production andradionuclide release.

A.2 Core Melt Progression

A great deal of information has been obtained on theprocesses involved in the early phase of melt progressionin core uncovery accidents that extends through core deg-radation and metallic (but not ceramic) material meltingand relocation. This information has come from integraltests in the PBF, ACRR, NRU, NSRR, and Phebus testreactors, from the LOFT FP-2 test, from tests in theCORA ex-reactor fuel-damage test facility, and fromseparate-effects experiments on significant phenomena.These tests have provided core degradation informationon fuel failure, Zircaloy oxidation by steam with attendanthydrogen generation, Zircaloy-clad melting and reloca-tion, and the effects of PWR Ag-In-Cd control rods,BWR B 4C control blades, high burnup fuels, and thereflooding of severely damaged cores.Most of the avail-able information on late-phase melt progression hascome from the post accident examination of the TMI-2reactor. Despite the core reflooding that successfully ter-minated the TMI-2 accident, the general late-phase meltprogression phenomenology of that accident, althoughnot the detailed behavior, appears to be applicable tounrecovered as well as to recovered accidents and possi-bly to some BWR accidents as well. The integral experi-ments that have provided most of the current informationbase on melt progression and an outline of the informa-tion obtained from these experiments are given in TableA.2.1.

The results of these integral tests and the TMI-2 coreexamination have provided a consistent picture of meltprogression. This is illustrated in the end-state configura-tion of the TMI-2 core, which is shown schematically inFigure A.2.1. Figure A.2.1 shows the development of adebris-supporting metallic blockage above the water levelin the lower portion of the core during coolant boildown.This blockage is produced by the relocation and freezingof metallic melt, mostly from unoxidized Zircaloy clad-ding. Fission-product decay heating produces a growingpool of mostly ceramic U0 2 fuel and oxidized zircaloyabove the metallic core blockage. The growing pool meltsdownward and radially outward through the supportingmetallic blockage and the secondary ceramic crust thatsurrounds the pool. During pool growth, the crust systemmelts and reforms, relocating downward and outward. Itappears that meltthrough occurs when the crusts do notreform to continue containment of the melt pool. AtTMI-2 with a reflooded core, pool melt-through was outthe side of the core. The mass and other characteristics ofthe ceramic melt that drains from the core into the lowerplenum in blocked-core accident sequences are deter-mined by the threshold and the location of themeltthrough of the supporting crust. The TMI-2 configu-

ration illustrates essentially the general melt progressionphenomenology thought to apply to both recovered andunrecovered blocked-core accident sequences in PWRs,and also. to any BWR accidents that have blocked coresequences. Differences in specific phenomenological be-havior from TMI-2, for unrecovered accidents are possi-ble, however, particularly with regard to the question ofsideways or downward meltthrough from the core.

The similarity of the results of the many integral tests oncore degradation and early phase (metallic) melt progres-sion show that the overall behavior in this regime is repro-ducible and is not strongly stochiastic in nature. Suchassurance is not available for the late phase melt progres-sion, however, with Figure.A.2.1 the TMI-2 core examin-ation providing nearly all the available information. Theprocesses of melt pool growth in a particular ceramicdebris and melt through of the supporting crust system,however, do not appear to be stochastic processes.

Metallic melt relocation leaves behind free-standingcracked U0 2 fuel pellets and ZrO2 oxidized claddingshards that have melting points, including eutectics, in therange from 2800°K to 3100'K. During late-phase meltprogression in unrecovered blocked core accidents and invery severe recovered blocked core accidents such asTMI-2, a mostly ceramic melt pool forms and grows in themostly ceramic debris bed. Thus, the metallic and theceramic debris with melting points that differ by 600 *K or'more become separated in space, and, as the TMI-2 coreexamination shows, they behave quite differently as coreheat-up and melt progression continue. The metallic andthe ceramic materials need to be treated separately inaccident analysis codes if melt progression is to be repre-sented realistically. This is not done in the older simpli-fied codes that treat the core melt as a single fictitious"corium" fluid with a unique (high metallic) compositionand a relatively low melting point (usually 2550'K).

In BWR accidents with automatic depressurization, pri-mary system blowdown lowers the water level below thecore and the BWR core plate. Under these conditions,heat-up occurs in a dry core with very low steam flow. Thecontribution of zirconium oxidation to the core heat-up issmall under these conditions, and a large fraction of theZircaloy cladding melts over a short period of time. It hasbeen hypothesized that, under these conditions, the me-tallic melt (including eutectic alloys) drains from the coreand the BWR core plate into the water-filled lowerplenum instead of freezing to form a blocked core similarto that at TMI-2.

The reflooding to the top of the core in the TMI-2 acci-dent, however, did not prevent continued core meltingbecause the water did not penetrate into the hot moltenpool region of the core. The accident was only successfullyterminated when hot ceramic core melt that constitute

A-3 NUREG-1365, Rev. 1

Appendix A

Table A.2.1 Sources of Current Integral Experimental Information on Melt Progression

Experiments

PBF Severe Fuel Damage TestsSFD-ST, 1-1, 1-3, 1-4

NRU Full-Length TestsFLHT 1, 2, 4, 5

ACRR Damaged Fuel TestsDF-1, -2, -3, -4

CORA Ex-reactor Tests and RelatedExperiments

Phebus SFD Tests

NSRR Reactivity Initiated Accident(RIA) Tests

LOFT FP-2 Large-Bundle(101-rod) Test

ACRR Late Phase (Ceramic Melt) TestsDC-1, DC-2, MP-1

TMI-2 Core Examination

Key Information

Integral information base on core degradation and melt progression

Control rod and high burnup fuel effects

Data on length effects and the absence of a cut-off to hydrogengeneration

Separate-effects data on core degradation and melt progression

Basic BWR information from DF-4

Basic information base on material-interaction effects and metallicmelt relocation

Core degradation information for BWR and PWR geometries,including reflood effects, using electrically heated, simulated fuelrod bundles of up to 57 rods

Core degradation and early phase melt progression phenomena

Fuel failure thresholds in reactivity initiated accidents

Unique results on metallic melt relocation and the absence of acutoff to hydrogen generation with a large flow-bypass area

Significant results on the effects of reflood on core degradationand hydrogen generation

Unique test results with fissionproduct decay heating

Dry U0 2 debris-bed thermal characterization and melting behavior

Dynamics of pool and crust growth in particulate ceramic debris beds

Major source of significant information on late-phase melt progression

Results applicable to basic phenomenology for both recovered& unrecovered accidents


Appendix A •

Upper griddamage -

Coating of previously-molten material onbypass region interiorsurfaces

Hole inbaffle plate

Ablated incoreinstrument guide

CL,Figure A.2.1 TMI-2 Core End-State Configuration


Appendix A

Table A.2.2 Core Melt Progression: Status of Current Understanding

" Early (Metallic Melt) Phase, Reasonably Well Understood Phenomena:

- Clad ballooning

- Intact-core-geometry oxidation heating and hydrogen generation.

- U0 2 liquefaction (dissolution) by molten Zircaloy.

- Eutectic material interactions among U0 2, ZrO2, Zr, and control materials and their oxides. Ratelimitations are less well understood.

- Opening up the compartmentalized BWR core early in a BWR accident by the eutectic interaction ofcontrol blade material with zircaloy channel box walls.

- Molten zircaloy relocation is a noncoherent, noncoplanar, rivulet-flow process, not a film flow process. Themelt forms an incomplete blockage that does not cut off steam flow and hydrogen generation.

* Late (Ceramic Melt) Phase, General Understanding:

- Information primarily from the TMI-2 core examination.

- Results are also generally applicable to PWR unrecovered accidents.

- Ceramic melt pool growth and meltthrough from a blocked core.

- Reflooding probably stopped downward pool and crust relocation to give side melt-through at TMI-2.

- Limited melt mass released from core (20% at TMI-2).

- Low metal content in ceramic melt pool.

- During Reflood: much hydrogen generation and strong heating of uncovered core from Zircaloy oxidationby reflood steam (LOFT FP-2 and CORA), is not well understood.

about 20% of the core mass drained from the melt poolinto the water-filled lower plenum (an additional heatsink). The melt was cooled by the lower plenum water anddid not fail the vessel lower head. The core reflooding,however, did stop the previous downward relocations ofthe metallic crust (by melting and refreezing) and mayhave been the cause of the melt-through of the melt poolout the side, rather than the bottom of the core. Thisresulted in the drainage of only about half the melt poolrather than the entire pool.

A major finding from all the integral tests and also fromthe TMI-2 core examination is that the unoxidized Zir-caloy, the control rod materials, and their eutectics meltbefore or during the rapid temperature transient fromsteam oxidation of the core zircaloy and at temperaturesranging from as low as 1200'K for the eutectics up to2200'K for the Zircaloy. Molten metal, which includessome dissolved U0 2 fuel, relocates downward by gravityto refreeze and form a porous partial core blockage, atleast in PWR accidents with water in the bottom of thecore.

The question of metallic melt drainage or core blockage isa major branch point for in-vessel core melt progression,and it has a large effect upon the characteristics of themelt released from the core into the lower plenum. In thecore blockage case, a large mass of mostly ceramic melt atabout 300°K drains rapidly into the water-filled lowerplenum, as happened at TMI-2. In the drainage case,layers of quenched melt are formed under the lowerplenum water in the order of their time of melting anddrainage from the core, with the low- melting metals atthe bottom and the ceramics at the top. These differencesalso have a major effect on the vessel failure process andon the characteristics of the melt released into the con-tainment upon vessel failure. In the drainage case, thereleased melt has a lower temperature and a high metalcontent.

In the TMI-2 accident, the melt released on melt-through from the blocked core into the water-filled lowerplenum by the ceramic melt pool contained only about20% of the core mass and had a very low metal content.The low metal content is important because metal


Appendix A

provides the potential for oxidation heating and hydrogengeneration after vessel melt-through. A more quantita-tive understanding of the key processes involved in deter-mining the melt-through threshold and location of failureare needed, however, in order to generalize the applica-tion of these relatively benign results. Acquiring thisinformation is a major objective of the current melt pro-gression research. A second major objective is determin-ing whether metallic, and later ceramic, melt drainagefrom the core without core blockage may occur in BWRaccidents in which the blowdown from automatic depre-ssurization (ADS) lowers the water level below the core,and core heat up occurs in a "dry core" with very lowsteam flow.

A summary of the current state of phenomenologicalunderstanding of melt progression is given in Table A.2.2.There is reasonable understanding of the significant phe-nomena involved in fuel damage and early phase (metallicmelts) melt progression, except for the processes of me-tallic melt relocation and blockage formation. For the latephase (ceramic melts), there is, in contrast, only a generalunderstanding that is based mostly on the TMI-2 coreexamination and analysis. Significant information on re-flood effects has also been obtained.

During a severe accident, hydrogen and heat are gener-ated during the exothermic chemical reaction (oxidation)of steam with core Zircaloy (and possibly some stainlesssteel). The integral severe fuel damage (SFD) tests thathave been performed have provided data on Zircaloyoxidation and hydrogen generation during severe acci-dents in addition to the data on core degradation and meltprogressionthat were discussed earlier. These tests wereconducted over a wide range of conditions in both BWRand PWR core geometries. In addition, separate-effectsexperiments on Zircaloy oxidation in steam have fur-nished basic reaction rate data and correlations. The ratesof Zircaloy oxidation and hydrogen generation are wellknown as long as intact core geometry is maintained. Atlower temperatures (below about 1700'K), the reactionrate is limited by oxygen diffusion through the growingprotective ZrO2 layer formed by cladding oxidation, andthe rate is limited by the availability of steam and Zircaloyat higher temperatures. The rate limit from diffusion iswell described by a parabolic rate law with an Arrhenius(exponential) temperature .dependence. Correlations ofthe experimental data on oxidation rates are incorporatedinto the MELCOR and SCDAP/RELAP5 codes as wellas into other severe accident codes.

Oxidation continues as the intact geometry is lost throughrelocation of molten unoxidized core Zircaloy downwardto cooler regions of the core, but the oxidation rate be-comes less well known. This is because of a lack of knowl-edge of the actual geometry, the thickness of new protec-tive oxide layers, and the locally available steam flow. Inall the integral severe fuel damage tests, substantial oxi-

dation rates continued after the start of molten zircaloyrelocation so long as supplies of steam and unoxidizedhigh temperature Zircaloy were available. In tests withBWR core geometry, eutectic interactions among thecontrol blade melt of stainless steel and. B4C and theZircaloy channel box walls failed the walls and opened upthe geometry to the cross-flow of steam and continuinghydrogen generation as in a PWR. During late-phase meltprogression in unrecovered accidents, oxidation and hy-drogen generation cannot be significant because of thelow metal content and the compacted geometry of thehotter region of the core around the growing ceramic meltpool.

The severely damaged fuel bundle was reflooded in theLOFT FP-2 test, and most of the oxidation and hydrogengeneration in the test was produced by steam duringreflooding. A quantitative understanding of the rates in-volved, however, does not yet exist. When Zircaloy-containing melt drops into lower plenum water, there canalso be a small amount of oxidation and hydrogen genera-tion from the unoxidized Zircaloy in the melt, particularlyif a steam explosion occurs.

As core melt material relocates into the lower head of thereactor vessel, the major concern of severe accidentanalysis becomes the mode and timing of lower headfailure. The research program in this area includes theanalysis and examination of samples from the TMI-2lower head. Failure mode analyses have been conductedto examine failure by the ejection of a vessel penetration,failure of a penetration outside the vessel shell, andglobal and local creeprupture failure of the vessel shell.The failure mechanisms have been evaluated for debrisand thermal-hydraulic conditions estimated for currentBWR and PWR designs. Results of these analyses arereported in draft NUREG/CR-5642, "Light Water Reac-tor Lower Head Failure Analysis," that was issued inDecember 1991.

Typical severe accident scenarios have been used for eachof the reactor types to develop lower head failure maps.The failure maps show, as a function of system pressureand vessel inner wall temperature, the mechanisms bywhich a vessel is most likely to fail. The studies haveshown that important parameters inciude not only thesystem pressure and vessel temperature, but also theeffective flow area and wall thickness of the penetrationsand the size of the annular gap between a penetrationtube and the vessel wall.

Under conditions that result in low heat-up of the vesselwall, the most likely mode of failure is ejection of a pene-tration because of failure of the vessel seal weld. In thiscase, the friction between the penetration tube and vesselopening is so low that tube ejection can occur withouttube rupture or damage. As the temperature of the vesselwall rises sufficiently to restrain the penetration tube byfriction, failure may occur by melt material penetration of


Appendix A

the vessel through the tube, with subsequent melt-through or rupture of the tube outside the vessel bound-ary. Two modes of melt penetration have been studied forthis case: conduction-limited penetration that is gov-erned by a freezing annular shell and bulk freezing pene-tration. Conduction-limited penetration has been foundto give the greater melt penetration in all cases. Moltendebris was also found to be more likely to penetratethrough a tube with a large effective diameter, such as aBWR instrument tube or a BWR drain nozzle, ratherthan through smaller diameter PWR instrument tubes.

Failure of the vessel lower head by creep rupture may bepossible when the system pressure and vessel tempera-ture are sufficiently high. The failure mode can be either aglobal failure of the hemispherical head, circumferen-tially around the vessel below the debris surface, or localbulging and membrane rupture of the shell. Vessel fail-ure analyses to date have used assumed debris conditionsto drive simple thermal analyses. Dimensionless groupsderived from these analyses may now be used in furtherdetailed analyses with severe accident codes such asSCDAP/RELAP5. An analysis for the local bulging casehas also been developed in which localized vessel wallheating may occur as a result of jet impingement of mol-ten core material on the vessel wall.

No additional research is planned on hydrogen genera-tion during in-vessel melt progression. However, analysesthat are under way of oxidation and hydrogen generationfrom reflooding in the LOFT FP-2 test will be completed.

Plans for lower head failure analysis include a peer reviewin FY92 of draft NUREG/CR-5642. The results of thepeer review will be incorporated into a final NUREGreport.

There are some significant uncertainties related to coremelt progression phenomena, primarily in the late (ce-ramic melt) phase. Future plans to address these uncer-tainties are discussed in Section 2.2.

A.3 Hydrogen Combustion and

Transport

A.3.1 Background

The safety significance of hydrogen combustion during asevere accident for non- inerted containments is that theconcomitant energy release manifested as pressurizationand heating of the containment atmosphere could pose athreat to containment integrity or to the survival or func-tioning of essential safety equipment. When hydrogencombustion alone is insufficient to threaten containmentintegrity, combustion may still represent a significant con-tribution to containment loadings when considered con-

junctively with direct containment heating or steam pres-surization. Hydrogen combustion can also impact thesource term by altering fission product chemistry andresuspending fission products in aerosol form. Hydrogentransport is a safety issue for operating reactors primarilyinsofar as the mixing of hydrogen determines the natureof subsequent combustion. In the event hydrogen re-leased into the containment during a severe accident ac-cumulates without igniting but mixes rapidly throughoutthe entire volume, the global concentrations in most in-stances will remain below the limits for detonation. Ifmixing does not occur because of stratification or pocket-ing in enclosed areas, those richer mixtures that occur, atleast locally, present a greater likelihood for flame accel-eration and detonation. For advanced reactor designswithout active mixing systems that rely on passive contain-ment heat removal from the containment atmosphere tothe containment shell (AP600) or to an external isolationcondenser (SBWR), the transport of hydrogen within thecontainment may influence the overall heat removal ca-pability of those related safety features.

A.3.2 StatusResearch conducted world-wide over the past 12 yearshas extensively investigated a number of issues related tohydrogen combustion and transport during severe reactoraccidents. Much of the work, performed to experimen-tally investigate the design and evaluation basis for reac-tor containment performance, focused on global defla-grations of premixed volumes of hydrogen, air and steam.Since containment analysis generally presumed globaldeflagrations were the mode of combustion (this assump-tion could be contrived to produce appropriately conser-vative loadings), research results were instrumental inestablishing the dominant parameters (flame speed andcombustion completeness) influencing the peak pressurefrom volumetric deflagrations. Diffusive burning of hy-drogen has also been the subject of experimental researchconducted by both by the NRC and the industry. In acooperative hydrogen research program conducted at theDOE Nevada Test Site (NTS), hydrogen and hydrogen-steam jets and plumes in ratios intended to representsevere accident blowdowns were ignited by thermal ig-niters to study diffusive burning behavior. Diffusion flameresearch has also been carried out at Sandia NationalLaboratories and at other research facilities, includingthe large scale facility at Factory Mutual Research Corpo-ration, used to investigate hydrogen mixing and combus-tion in a Mark III containment.

Recognizing that combustion modes include supersonicas well as subsonic flame propagation, the NRC has alsosponsored considerable research on the detonation ofhydrogen-air and hydrogen-air-steam mixtures. The in-ternational reactor safety research community has alsocontributed to the data base on the detonability of variousgaseous mixtures. This research clearly identified the sen-


Appendix A

sitivity of scale in establishing the limits for detonability ofmixtures as well as providing insights to the mechanismsfor flame acceleration and transition to detonation. Therange of detonable concentrations for hydrogen- air mix-tures has been shown experimentally to be much widerthan the classic limits of approximately 18 to 60% estab-lished in small-scale testing at ambient temperatures.The range of detonable concentrations has been shownexperimentally in our research to be as wide as 11.6 to74.9% hydrogen at 20'C and 9.4 to 76.9% hydrogen at1001C with the limits depending on scale, geometry, andtemperature. The view has been that steam greatly re-duced the mixture's detonability, but there is analyticalevidence that increasing temperature may make steammitigation less efficient. In conjunction with hydrogencombustion research, the research community has alsoexperimentally explored the 'issue of hydrogen transportand mixing. While the NRC has not exclusively sponsoredsignificant experimental research on hydrogen mixing,our joint research agreement with Germany has providedmixing test data in the complex geometries of theBattelle-Frankfurt and HDR test facilities. Industry datafrom the HEDL facility and the FMRC one-fourth scalefacility has supplemented that data; although this datawas specifically related to ice condenser and Mark IIIcontainments under certain accident conditions. To com-plement the experimental or phenomenological re-search, the NRC has sponsored the development andapplication of computer codes for hydrogen mixing andcombustion analysis, most notably the HECTR and HMScodes. The HECTR code, which employs control volumemodeling of the containment, and the HMS code, a finitedifference code, were developed to provide varying levelsof resolution of the containment volume for mixing analy-sis. HECTR combustion modeling has been subsumedinto the more general CONTAIN and MELCOR codes.

A.4 TMI-2 Vessel Investigation Project

A.4.1 Background

Most current knowledge of in-vessel severe accident be-havior has come from experiments such as the series ofsevere fuel damage tests performed in the Power BurstFacility, ACRR, NRU, LOFT FP-2, Phebus test reactor,CORA ex-reactor experiments, and from the extensivepost accident core examination of the Three Mile IslandUnit 2 -"1I-2) reactor, performed by the Department ofEnergy (DOE). In 1988, the NRC, in cooperation with 10foreign countries under the auspices of the Organizationfor Economic Cooperation and Development's (OECD)Nuclear Energy Agency (NEA), undertook a follow-onprogram to the DOE TMI-2 examinations. The objec-tives of this program, called the TMI-2 Vessel Investiga-tion Project (VIP), are to (1) investigate the condition and

properties of materials extracted from the lower head ofthe TMI-2 reactor pressure vessel, (2) determine theextent of damage to the lower head by chemical andthermal attack, and (3) determine the margin of structuralintegrity that remained in the pressure vessel.

A.4.2 Status

Under the VIP program, 15 reactor vessel steel speci-mens, 14 incore nozzles, and 2 incore guide tubes weresuccessfully extracted from the lower head over a 30-dayperiod ending March 1, 1990. The vessel steel samplesthen were decontaminated, sectioned, and distributed tothe United States and seven other participating countriesfor mechanical and metallographic examinations. Also,the nozzles and guide tubes were cut and distributed forexamination to INEL, ANL, and the CEA in Saclay,France.

Since the extraction of the test specimens in 1990, sub-stantial metallographic examinations of the vessel steelsamples have been completed, including microstructuralexaminations and hardness measurements. Results ofthese examinations have provided preliminary estimatesof temperature histories of the lower head samples.These results show that the maximum inner surface tem-perature of at least four samples reached 1050-1100'Cfor about 30 minutes. These samples were extracted fromthe lower head from a small region, approximately 2 to 3feet in diameter. The examinations also indicated that thetemperature 2 inches into the wall was about 100'C lowerthan the inner surface temperature. (The lower headthickness was 5 inches.) Mechanical testing is under wayto determine tensile and creep properties of the vesselsteel at high temperatures (up to about1100 0 C).

Metallographic and scanning electron microscope (SEM)examinations of the instrument tube nozzles are beingperformed to assess the nozzles' axial temperature profileand interaction of the Inconel 600 nozzle material withmolten core materials. Examinations of previously mol-ten thermocouples in the nozzles may be an importantindicator for determining the axial temperature profile inthe nozzles.

In September 1991, the VIP Management Board decidedto amend the original agreement extending the VIP pro-gram from September 30, 1991, to March 31, 1993. Theobjectives of the amended program are to (1) performmore detailed testing and examination of the in-core in-strument tube nozzle penetrations and the in- core instru-ment guide tubes that were extracted. from the lowerhead, (2) perform additional analyses of potential reactorvessel failure modes based on data from sample examina-tions, and (3) assess the margin-to-failure of the lowerhead of the reactor vessel.


Appendix A

A.4.3 Future Plans

Results of the ongoing TMI-2 lower head examinationsare expected to provide additional information on thephysical properties of the specimens, temperature distri-butions in the instrument nozzles, and interactions be-tween the molten core material and the vessel. Theseresults then will be used to perform scoping analyses ofpotential reactor vessel failure modes, such as penetra-tion tube failures and global or local failure of the reactorvessel lower head. More detailed analyses of the mostlikely failure mechanisms will be performed to estimatethe margin-to-failure of the lower head. A final projectreport, integrating the results of all the sample examina-tions and analyses, will be issued at the completion of thisprogram in June 1993. June 24, 1992

A.5 Fuel-Coolant Interactions (FCI)and Debris Coolability

A.5.1 Background

One of the fundamental phenomena in the course of asevere accident is the interaction of degraded core mate-rials with coolant. In the absence of coolant, the corematerials overheat, melt, and relocate. The opportunityfor fuel- coolant interactions arises not only as a naturalconsequence of this relocation process (into areas occu-pied by coolant), but also as a consequence of accidentmanagement actions that add water to regions previouslydepleted of coolant. Considering the variety of reactorgeometries, meltdown scenarios, and timing of coolantaddition, a broad range of resultant phenomena is possi-ble.

The intensity of the interactions and resulting materialstate and configuration depend on: (1) the geometry ofthe region, (2) the masses of core material and coolantinvolved, (3) the thermodynamic state of the materials,and (4) the rates of pouring of molten core materials intoa pool of coolant or the rate of flooding a degraded ormolten core. For example, at one extreme, a slow pour ofmolten core debris in a large, deep pool of water can leadto complete quenching and formation of a coarse particledebris at the pool bottom. At the other extreme, a rapid,massive release could yield a highly energetic explosionwith significant mechanical consequences on the systemand its surrounding structures. Initially, the subject ofFCIs emphasized the phenomenon of a vapor-explosion-induced missile as a possible mode of containment fail-ure. With increased emphasis on accident management,interest in FCIs broadened to include core degradationregimes that can be quenched by flooding with water. Forsuch scenarios, it is important to identify potential cir-cumstances in which FCIs can lead to coolable configura-

tions or can significantly alter the core melt progressionscenario.

Certain fundamental design differences between PWRsand BWRs require different areas of emphasis. Duringthe later phase of core melt progression in PWRs, thepotential exists to accumulate large quantities of melt inthe core region "crucible." Because of the largely openlower plenum geometry, in-vessel FCIs can span thewhole range from energetic to relatively benign. ForBWRs, on the other hand, if a metallic or ceramic block-age does not occur, large melt accumulations in the coreregion could be excluded. (The research effort on coreblockage or melt drainage in BWR severe accidents isdiscussed in Section 2.2.2.1 of this report.) In addition,the lower plenum is crowded by the control rod drives.Therefore, small energetic FCIs are more likely, and theemphasis for BWRs is on coolability.

During the ex-vessel stage of a severe accident, water maybe added to the reactor cavity as a deliberate measure tomitigate the consequences of core debris-concrete inter-actions. Water may be present in the cavity as a result ofnatural processes such as the condensation of steam, flowfrom containment sprays, or discharge from accumulatorsin the reactor coolant system. For all containment geome-tries, various FCI concerns exist that span the wholerange from energetic interactions to benign coolability.

A.5.2 Status

There are three specific issues addressed in this section:

1. FCI energetics,

2. Fuel-melt quenching in water pools, and

3. Adding water to degraded core (in-vessel and ex-vessel).

A.5.2.1 Status of FCI Energetics

This topic is of primary relevance to in-vessel interactionsin PWRs. The first quantification of the potential foralpha-mode containment failure was offered inWASH-1400. Additional quantification of this failuremode can be found in the expert opinion efforts, i.e.,NUREG-1116, "Steam Explosion Review Group, A re-view of Current Understanding of the Potential for Con-tainment Failure Arising from In-Vessel Steam Explo-sion, 1985," and NUREG-1150, "Severe Accident Risks:An Assessment for Five U.S. Nuclear Power Plants,1990." It is generally agreed that the probability of alpha-mode containment failure is negligible, although the needto improve the quantification basis for this conclusion isgenerally acknowledged. NUREG/CR-5030 presents aprobabilistic framework that arrives at the same conclu-sions. This report highlights "premixing" as the process of

NUREG-1365, Rev. 1 A-10

Appendix A

primary importance in limiting the magnitude of ener-getic FCIs.

The issue of premixing arises because there is a possibilitythat core relocation into the lower plenum can occur inlarge massive pours in PWRs. The geometry of the PWRdiffuser plates leads to the breakup of the melt intoseveral smaller jets. Because of extensive steaming, largepremixtures, which are expected to develop in the abovecircumstances, would be largely void of water, and there-fore, the magnitude of energetics is limited.

A.5.2.2 Status of Fuel-Melt Quenching

Fuel-melt quenching in severe accident evaluations hasseveral important aspects. For the in-vessel core meltportion of severe accident sequences in current genera-tion PWRs and BWRs, as well as their ALWR counter-parts, quenching considerations are fundamental in pre-dicting the mode and timing of lower head failure and theresultant impact on direct containment heating (DCH).In accident management, increasingly more considera-tion is given to the possibility of flooding the cavity regionexternal to the lower head of the reactor vessel in order toprevent vessel failure and retain core debris in the vessel.For the ex-vessel portion of severe accident sequences,quenching plays an important part in determining theex-vessel relocation behavior of the melt, and hence,long-term coolability.

The fundamental difficulty in addressing the role ofquenching in severe accident scenarios relates to uncer-tainties in (1) the flow characteristics of the pour (i.e., sizeand number of melt jets, pour rates) and (2) the composi-tion of the melt (i.e., metallic vs. oxidic components) andits temperature (superheat). Conservative evaluations ofsuch uncertainties and of the quenching process itself,particularly on assessment of containment integrity, arepossible. However, a better understanding of quenchingwill contribute to the depth of understanding such thatbetter judgments, especially on new plant designs, can bemade in the future.

A.5.2.3 Status of Adding Water to Degraded Core

The reflooding mode refers to two different configura-tions:

1. Degraded core configuration in which water is sup-plied from above or below the core, or

2. A relatively shallow molten corium pool (at the bot-tom of the lower plenum or on the concretebasemat) with water on top of the molten pool.

The phenomena and technical issues for these configura-ti6ns are quite different and are addressed separatelybelow.

A.5.2.3.1 Adding Water to a Degraded Core(In.vessel)

A review of past experiments suggests that a few testshave been performed with water addition following bun-dle degradation, mainly, CORA, PBF, and the LOFT-FP2 test. Also, simulant tests of a coolant added to fueldebris have been performed at BNL, ANL, and UCLA.Although limited in scope, these tests address water addi-tion during core degradation. The major variables affect-ing the consequences of water addition are the rate andcharacter of water addition and the state of the fuel at thetime water is added. Reference states for the fuel are:

1. Initial heat up and early core degradation

This first stage of a severe accident sequence is de-fined as beginning with the onset of core uncoveryand continuing up to the clad melting (not eutecticdissolution) and relocation. Coolant is still easilyaccessible to the degraded portion of the core be-cause of efficient lateral flows through the openPWR lattice. Key modeling difficulties include va-por chimney effects and permeability.

2. Advanced core degradation (core rubble, melt, andrelocation)

This stage of the severe accident sequence is definedas beginning with the first significant fuel meltingand relocation and extending through core debrisrelocation onto the lower plenum. The key events inthe late stage are the formation of a molten pool,pool growth, and breakthrough of the crust support-ing the molten pool. The behavior following wateraddition during this stage is governed by crust stabil-ity, which is dominated by the heat transfer proper-ties of the (porous) medium surrounding the crustsand of the crusts themselves, and the heat load dis-tribution on the inner surface of these crusts bynatural convection of the molten core materialwithin the crust.

A.5.2.3.2 Adding Water to Degraded Core (Ex-vessel)

The continued progression of a severe accident can leadto the expulsion of reactor core debris into the reactorcavity. Debris in the reactor cavity can interact with thestructural concrete of the containment and even, in somecases, directly with the pressure boundary of the contain-ment.

The configuration of core debris that is expelled from thereactor vessel does not ensure that the mere presence of

A-11 NUREG-1365, Rev. 1

Appendix A

water will result in cooling the debris sufficiently to elimi-nate core-concrete interactions. The core debris must befragmented into coolable rubble, or it must spread over alarge enough area that heat extracted by overlying waterwill cool the debris.

At this time, a technically justified criterion for debriscoolability during the ex-vessel phases of severe reactoraccidents cannot be defined. Issues that must be resolvedto define such a criterion can be broadly categorized as

issues of debris configuration and issues of heat removalfrom core debris by water.

The NRC is participating in the MACE test program,which is examining the effects of water on uranium-dioxide-rich melts interacting with concrete. In the past,the NRC has sponsored tests of high temperature oxidicand metallic melts with water and concrete (the WET-COR tests) to determine limits of coolability.

NUREG-1365, Rev. 1 A-12

APPENDIX BCLOSURE OF SEVERE ACCIDENT ISSUES

B.1 Source Term

B.1.1 Background

Radionuclide releases to the environment, that is; thetype, quantity, timing and energy characteristics of therelease of radioactive material from reactor accidents("source terms") are deeply embedded in the regulatorypolicy and practices of the NRC. For almost 30 years theNRC's reactor site criteria (10 CFR Part 100) have re-quired that for licensing purposes, an accidental fissionproduct release from the core into the containment bepostulated to occur and that its radiological consequencesbe evaluated assuming that the containment remains in-tact but leaks at its maximum allowable leak rate.

Evaluation of the consequences is used to assess bothplant mitigation features such as fission product cleanupsystems and the suitability of the site. The characteristicsof the containment "source term," which must be distin-guished from a release to the environment, are describedin Regulatory Guides 1.3 and 1.4, but are derived fromthe 1962 report TID-14844 (Ref. 1). The source termconsists of 100 % of the core inventory of noble gases and50% of the iodines (half of which are assumed to depositon interior surfaces very rapidly). Regulatory Guides 1.3and 1.4 also specify that the source term is instantane-ously available for release and has significantly affectedcontainment isolation valve closure times. The guidesalso specify that the iodine is predominantly (91%) inelemental (12) form.

In addition to plant mitigation features and site suitabil-ity, the regulatory applications of this release also estab-lish (1) the post-accident radiation environment for whichsafety-related equipment should be qualified, (2) post-ac-cident habitability requirements for the control room, and(3) post-accident sampling systems and accessibility.

In contrast to a specified source term for design basisaccidents, severe accident source terms first arose inprobabilistic risk assessments (e.g., Reactor Safety Study,WASH-1400, Ref. 2) in examining accident sequencesthat involved core melt and possible containments fail-ure. Severe accident source terms represent mechanisti-cally determined "best estimate" releases to the environ-ment, including estimates of failures of containmentintegrity. This is very different from the combination ofthe nonmechanistic release to containment postulated byTID-14844 coupled withthe assumption of very limitedcontainment leakage used for Part 100 citing calculationsfor design basis accidents. The worst severe accidentsource terms resulting from containment failure (espe-

cially early failures, i.e., within a few hours from onset ofan accident) or containment bypass can lead to conse-quences that are much greater than those associated witha TID-14844 release into containment when the contain-ment is assumed to be leaking at its maximum leak rate forits design conditions. Indeed, some of the most severesource terms arise from some containment bypass events,such as "event V" and multiple steam generator tuberuptures.

Source term estimates under severe accident conditionsbegan to be of great interest shortly after the Three MileIsland 2 (TMI-2) accident. The objective of a major NRCresearch effort on source terms that began about 1981 isto obtain a better understanding of fission-product trans-port and release mechanisms in LWRs under severe acci-dent conditions. This research effort, which involved anumber of national laboratories as well as nuclear indus-try groups, has resulted in the development and applica-tion of several new computer codes to examine core meltphenomena and associated source terms. Work has alsoincluded significant review efforts by peer reviewers, for-eign partners in NRC research programs, industrygroups, and the general public. Current risk assessmentmethods, including the latest research efforts on severeaccident source terms, are reflected in NUREG-1150(Ref. 3), which provides an assessment of severe accidentrisk for five U.S. nuclear power plants. Finally, the occur-rence of the accident at Unit 4 of the Chernobyl reactor inthe Soviet Union on April 26, 1986, and the large acciden-tal release. of fission products resulting from it has pro-vided further impetus to understand severe accidentsource terms as well as to prevent such occurrences.

B.1.2 Status

Since shortly after the accident at TMI-2, the NRC hassponsored numerous experimental and analytical re-search projects on fission product release and transport.Early experiments and analytical work tended to focus onrelease from fuel material under high temperatures andsevere accident environments. Later, experimental dataon the behavior of aerosols in the RCS and the contain-ment were obtained. These data were used to developaerosol deposition and transport models to analyze fissionproduct behavior in the reactor coolant system and thecontainment. Currently, fully integrated models are be-ing assembled into individual codes, the VICTORIA code(Ref. 4) and the CONTAIN code, (Ref. 5), for the analysisof in-vessel and ex-vessel source terms, respectively.

The issue of revaporization of deposited radioactive ma-terials in the RCS is of particular concern because of theinstability of the deposited radioactive material at high

B-1 NUREG-1365, Rev. 1

Appendix B

temperatures induced by decay of nuclides. Recent se-vere accident calculations have predicted that, for somesequences, natural circulation of gases through the reac-tor core may also heat structures in the RCS to substan-tially higher temperatures than had been previously pre-dicted. The high temperatures may also result in thefailure of RCS piping at certain locations. This failure ofRCS piping is of some significance since current analysesof core degradation indicate substantial fractions of thecore could be retained within the RPV after vessel fail-ure. Evidence from TMI-2 suggests that as much as halfthe core material may have stayed within the originalconfines after the rest of the core had melted and drainedinto the lower plenum. This remaining fuel in the coreregion could be exposed to air once the plenum has beenbreached. Air will react exothermically with the claddingremaining on the fuel, producing high temperatures inthis fuel. Once the cladding has been oxidized, vapors ofradionuclides not usually considered highly volatile, nota-bly ruthenium and molybdenum, will be produced be-cause of the strongly oxidized conditions. The VICTO-RIA code has incorporated a model to addressrevaporization caused by an increase in temperature ofthe RCS. The VICTORIA code also models other impor-tant severe accident phenomena such as the release andtransport of fission products, condensation of vapors,aerosols behavior, and chemical reactions in the RCS.

In low-pressure accident scenarios in which the reactorvessel fails, high-temperature core debris may fall intothe reactor cavity where it interacts with the concrete. Athigh temperatures (approximately 1,300-1,500 °C), con-crete decomposes, and the ablation products commonlyinclude water vapor and carbon dioxide as well as therefractory oxides CaO and SiO 2. The liquefied oxidiccomponents of the concrete nmx with the uranium oxidefuel and metallic oxides of the debris. Typically, the coredebris is initially all or partially molten; gases released atthe debris-concrete interface bubble through the debrispool reducing some low-vapor-pressure oxides such asLa2 O3 to high-vapor-pressure forms such as LaO. Thesemore volatile forms then vaporize into the bubble vol-ume, thus releasing fission product species that were notreleased in the vessel. Aerosols are formed when thebubbles exit the upper surface and fragment. Among thefactors that influence the magnitude of the ex-vessel re-leases are the composition and temperatures of the coredebris. Concrete composition also has a major impact onthe amount of aerosols entrained into the containmentatmosphere. Limestone concrete 'produces larger gasflows and is more oxidizing than basaltic concrete. Anextensive experimental data base has been obtained oncore-concrete interactions with no overlying water pool.If an overlying water pool exists, a considerable amount ofthe aerosols may be scrubbed and kept out of the contain-ment atmosphere. Research effort is under way to obtainexperimental data for this case. Core-concrete interac-

tions and associated radionuclide releases are modeled bythe stand-alone code, CORCON-MOD3 (Ref. 6). Thesemodels will also be incorporated into the CONTAIN codeto allow comprehensive evaluation of containment be-havior during severe accidents. In addition, the CON-TAIN code models the thermal-hydraulics (pressure,temperatures, etc.), aerosols behavior, fission products.behavior and transport, and hydrogen behavior in thecontainment. Codes such as VICTORIA, CONTAIN,andCORCON-MOD3 incorporate mostly mechanistic mod-els for severe accident phenomena, thus, allowing theexamination of complex source term issues in a detailedand systematic fashion. On the other hand, the MELCORcode (Ref. 7) was developed as an integral tool for analysisof fission product transport in the RCS and in the contain-ment. MELCOR employs simpler models for severe acci-dent phenomena in order to facilitate the fast runningtime requirements for the repetitious calculations used inPRA analysis.

With respect to the potential release of iodine from sup-pression pools and reactor cavity water, the ACE programand the ORNL iodine chemistry research have providedextensive experimental data to address this issue. AtORNL, the research included iodine partition coefficienttests, hydrolysis chemical kinetics tests, radiolysis chemi-cal kinetics tests, hydrogen bum/iodine chemistry tests,and TRENDS models development. These efforts werefurther enhanced by the ACE program, which includedhygroscopic aerosol/iodine chemistry tests, and hydrogenbum/iodine chemistry tests. Many iodine chemistry mod-els were developed from these data bases. These modelsare now being incorporated into the CONTAIN code.Further validation of the CONTAIN code could be doneusing the PHEBUS-FP data.

B.1.3 Source Term Uncertainties and PresentResearch Efforts

With respect to source term uncertainties, NUREG-1150 has identified specific source term issues as con-tributors to uncertainty in risk estimates. For fission prod-uct release from fuel and retention in the RCS, a keyquestion is, "How significant is fission product releaseduring the late stages of core melt vs. the early phase?"

The present theoretical model for the late-stage rubblebed (significant relocation and melting of ceramics) as-sumes that release is governed by gas-phase mass trans-port. For the molten pool, the main mechanism for fissionproduct release is governed by diffusion and surface con-vection. Both of these theoretical models need experi-mental data for validation. The predominant sources ofuncertainty in these theoretical model are:

" geometry of the core debris," magnitude of gas fluxes through the debris, and

NUREG-1365, Rev. 1' B-2

Appendix B

* thermo-chemical properties of the high-temperature vapor species that vaporize from thedebris.

The VICTORIA code incorporates models to addressthese questions. Uncertainties in our current understand-ing of the evolution of accidents are handled in a para-metric fashion.

Integral experiments to study fission-product releasefrom late-phase core melt progression and revaporizationare not being planned in the United States. However, thePHEBUS-FP project could provide some of the neededexperimental data.

For late-phase revaporization of fission product from theRCS, ,the key questions are:

1. After the reactor vessel has been breached and airingress occurs, what are the release rates from thefuel remaining in the vessel? Likewise, for shutdownaccidents, what are the release rates from the fuelexposed to air?

2. What chemical forms are important during thetransport and retention of aerosols and vapors?

As mentioned earlier, if the vessel is penetrated, air fromthe containment atmosphere will circulate over retainedfuel. The air will react exothermically with the claddingremaining on the fuel, producing high temperatures inthis fuel. Once the cladding has been oxidized, vapors ofradionuclides (notably ruthenium and molybdenum, ra-dioactive species not usually thought to make major con-tributions to severe accident source terms) will be pro-duced because of the strongly oxidized conditions. Testsperformed at Chalk River Laboratory in Canada withuncladded and cladded fuel, and at ORNL with fuel frag-ments under highly oxidized conditions, showed that alarge VI-7 fraction of the ruthenium, tellurium, and mo-lybdenum was released. The predicted release rates aredependent on a highly oxidized condition for the fuel.However, it is necessary to conduct tests for radionuclidereleases for typical LWR fuel, as opposed to the thinnercladding used in the CANDU fuel. Plant configurationsduring shutdown situations may lead to the possibility ofair ingress into the core either by natural circulation orfrom the residual heat removal system. That air will reactexothermically with the cladding on the fuel, producinghigh temperatures in the fuel, resulting in massive vapori-zation of ruthenium, tellurium, and molybdenum. A test(VI-7) will be conducted at ORNL for fission-productrelease at high temperature in an air environment, theexperimental data could be readily incorporated into thecurrent VICTORIA fuel-release model.

The chemical form assumed for iodine affects late-phaserevaporization. If cesium iodide or other iodides decom-

pose to produce HI or 12, there will be little or no retainediodine in the RCS to revaporize after containment failure.If CsI is stable, substantial amounts will be retained tem-porarily in the RCS and will be able to revaporize, creat-ing an iodine source term after containment failure. Testsat Winfrith and SNL have shown that CsI will react withboron oxides vaporized into the RCS atmosphere to yieldcesium borate, I, and HI. Tests at SNL have shown ther-mal instability of CsI in the RCS environment. These testshave not clarified whether I or HI produced by reactionsof CsI can subsequently react to form other iodides suchas NiI 2(g,c). Analyses done in the U.K. suggest that highvapor fractions of iodine (CsOH and CsI) at the time ofRCS failure could yield aerosols in containment that donot settle rapidly and are only slightly affected by contain-ment sprays. Hence, a source of airborne fission productwould be available for leakage out of the containment.

Currently, there are no suitable experimental data tovalidate revaporization models, but the PHEBUS-FPtests could provide some. It is likely that continued exa-mination of the chemical form of I or HI produced byreaction of CsI could alter the perceptions of risk byaltering the predicted amount of suspended radioactivityin containment at the time of containment failure. Thiscould be accomplished by sensitivity analysis using VIC-TORIA.

It is also important to utilize risk perspectives regardingthe uncertainties in late-stage core melt and late-phaserevaporization. Generally, risk is increased by the earlyrelease of fission products into containment and earlycontainment failure or bypass. Hence, additional fissionproducts released later in an accident phase will denotelesser releases at an earlier stage. Modeling sensitivitycan be made in risk assessments that can test uncertain-ties and their implications.

A key question regarding the ex-vessel source term is,"What effect does hydrogen combustion have on aerosolssuspended in the containment atmosphere?"

Aerosol materials containing CsI must be dehydrated andvaporized before chemical interactions of cesium iodidecan be expected. Tests conducted at ORNL, as part of theACE program, found that vaporized CsI was unstable inhydrogen flames with the iodine redistributing as iodide,12, and iodate. An excess of metal cations (Cs) reduced theextent of 12 formation. Oxidation to 12 was consistent withthermal decomposition of CsI, but iodate production wasrelated to the nonequilibrium OH and 0 radical concen-trations found in hydrogen/air flames. Data are believedto be adequate to address this question. However, thepresence of other aerosol species in the containmentatmosphere, such as aerosols produced by core-concreteinteractions, could affect the stability of CsI during hydro-gen combustion events, even though it is a small contribu-tion. For instance, silica could trap cesium to form cesiumsilicate so that it cannot recombine with iodine. Other


Appendix B

basic species could react with iodine produced in thecombustion to reform iodides:

Na2 O(c) + 2HI(g) - > 2 NaI(c) + H20.

Currently, the TRENDS model (Ref. 8) treats this in aconservative fashion (i.e., it assumes that 12 will be formedin a hydrogen bum). Such a model has been developed forinclusion in the CONTAIN code.

In conclusion, although additional physics and chemistryresearch to reduce uncertainties in source term phenom-ena can be performed, it is important to consider the needfor such research and the potential that this researchcould significantly improve our risk perspective associ-ated with severe accidents. NUREG-1150 indicates thatuncertainties in overall estimation of risk are largelydriven by uncertainties in containment performance, pri-marily those associated with estimation of containmentloads, estimation of containment performance at loadlevels beyond the design basis, and estimation of the prob-ability and location of containment.bypass.

B.1.4 Regulatory Applications/Implications

B.1.4.1 Development of Updated Source Term

Design basis accident source terms have been used in theUnited States for licensing purposes in three distinctways, namely:

1. For siting evaluations as required by 10 CFR Part100,

2. For defining the radiological environmental condi-tions for certain plant systems, and

3. For assessing the effectiveness of plant mitigationsystems.

The NRC is presently preparing an update of the sourceterm contained in Regulatory Guides 1.3 and 1.4, makinguse of current severe accident research insights. Thiseffort is expected to result in changes in data for thetiming of the release, the composition and magnitude offission product releases into the containment, and thechemical form of the iodine fission products. A draft of anupdated report replacing TID-14844 is expected to beissued for comment by the first half or CY 1992.

Rather than an instantaneous release into the contain-ment, the revised formulation is expected to be stated as aseries of fission product releases into the containment,each one associated with a particular stage of an accidentor group of accidents. Hence, the revised formulation isexpected to begin with the release of coolant activity,followed by the release of activity in the fuel gap, therelease of fission products associated with gross fuel de-

gradation prior to reactor vessel failure, and finally, re-lease of fission products from core-concrete interactions.

Additional nuclides other than the noble gases and iodineare expected to be released. For example, preliminaryindications are that the fraction of core inventory of ce-sium released into the containment is generally compara-ble to that of iodine. In addition, some tellurium andsmaller fractions of the remaining nuclides are also ex-pected for release.

A recent study on iodine chemical form and behavior onentering the containment from the RCS, and the subse-quent revolatilization of iodine from water pools in con-tainment, has been completed at ORNL. ORNL exam-ined a group of severe accident sequences used inNUREG-1150. These accident scenarios were for bothhigh- and low-pressure sequences that are risk significant.For the RCS, the analysis considered the chemical kinet-ics of 20 reactions of iodine with water, hydrogen, andcesium, and determined the temperature and time whenchemical equilibrium was established. Once chemicalequilibrium was established, an analysis determined theiodine chemical forms present. In most calculations, io-dine was released from the RCS into the containment ascesium iodide (CsI) with very small amounts of I or HI.Since the ORNL's study considered a limited set of reac-tions related to CsI, iodine in the form other than CsIcould be released from the RCS. In order for the RCS torelease iodine in the form other than CsI, a significantfraction of the Cs has to be removed within the RCS.However, for the accident conditions analyzed, only asmall fraction of Cs was removed within the RCS. Ifdesired, more sophisticated treatment of CsI reactionscould be undertaken with VICTORIA to confirm theORNL's findings. The ORNL results indicate that theiodine entering containment is at least 95% CsI, 5% as Iand HI, with not less than 1% as either I or HI. This is incontrast to the iodine chemical form specified by the TIDsource term, which is predominantly (91%) elemental.

The ORNL iodine research and the ACE program haveprovided data that address revolatilization of iodine fromwater pools. A comprehensive model to estimate therevolatilization of iodine from water pools was developedby ORNL. Once iodine enters containment, it dissolves inwater pools or plates out on wet surfaces as I-. Subse-quently, the iodine behavior within the containment de-pends upon time and the ph of the water solutions. If theph is maintained at a value of 7 or greater, the amount ofiodine in solution that converts to 12 and organic iodinelater in the accident sequence will be very low. If ph is notcontrolled, radiation levels in water pools are sufficient toconvert much of the dissolved iodine to elemental iodinefor release into the containment atmosphere. Experi-mental data on the revolatilization of iodine from waterpools resulted from the evolution of Ph and irradiationunder severe accident conditions are expected from the

NUREG-1365, Rev. 1 B--4

Appendix B

PHEBUS-FP project (section 2.7.2). These data will pro-vide confirmatory assessment of models for iodine revola-tilization from water pools.

B.1.4.2 Regulatory Implications

At the present time, the NRC is pursuing several regula-tory initiatives to incorporate insights from updated se-vere accident source terms. A revision of the NRC's reac-tor site criteria (10 CFR Part 100) is being carried out inparallel with an interim revision of 10 CFR Part 50. Thereactor site criteria will be revised to remove source termand dose calculations and to add requirements in Part 100for exclusion area size and population density based onthose from Regulatory Guide 4.7. At the same time, Ap-pendix A to Part 100, containing site seismic criteria, isalso being revised to reflect the latest understanding.Source term and dose criteria will continue to be impor-tant for plant design; consequently, an interim revision of10 CFR Part 50 will be carried out in parallel and willcontain the present source term (i.e., that fromTID-14844 and Regulatory Guides 1.3 and 1.4). Theseproposed rule changes are expected to be issued for com-ment by early CY93.

Updated source term insights arising from the technicalupdate of TID-14844 are expected to be made availablefor voluntary use by existing licensees. A final revision of10 CFR Part 50 to incorporate updated source term andsevere accident insights will then be undertaken, with aproposed rule for comment expected to be issued by early1993.

Although regulatory positions arising from updatedsource term insights remain to be developed, some pre-liminary implications can be seen at this time. It is clearthat updated source term insights indicate the need forconsideration of nuclides (e.g., cesium) in addition toiodine and the noble gases. In addition, revised insights oniodine chemistry call into question the need for high-effi-ciency charcoal absorbers (assuming that the Ph is con-trolled postaccident). These can, in turn, impact suchimportant plant systems as fission product cleanup sys-tems, control room habitability, and allowable contain-ment leak rate.

Finally, and most importantly, the above discussion andall recent risk studies have shown the importance of main-taining containment integrity under severe accident con-ditions in order to assure low risk. This strongly suggeststhat the appearance of a severe accident source termwithin containment should be more closely linked withthe temperatures, pressures, and containment loads andchallenges associated with such releases, rather than anarbitrary linkage with a single!sequence such as a large-break loss-of-coolant-accident.

References

(1) J. J. DiNunno et al., "Calculation of Distance Fac-tors for Power and Test Reactor Sites," U.S. AtomicEnergy Commission, TID-14844, March 1962.

(2) U.S. Nuclear Regulatory Commission, "ReactorSafety Study-An Assessment of Accident Risks inU.S. Commercial Nuclear Power Plants," WASH-1400 (NUREG-75/014), October 1975.

(3) U.S. Nuclear Regulatory Commission, "Severe Ac-cident Risks: An Assessment for Five U.S. NuclearPower Plants," NUREG-1150, December 1990.

(4) T. J. Heames et al., "VICTORIA: A MechanisticModel of Radionuclide Behavior in the ReactorCoolant System under Severe Accident Condi-tions," NUREG/CR-5545, October 1990.

'(5) K.K. Murata et al., "User's Manual for CONTAIN1.1: A computer code for Severe Nuclear ReactorAccident Containment Analysis".

(6) D. R. Bradley and D. R. Gardner, "CORCON-MOD3: An Integrated Computer Model for Analy-sis of Molten Core-Concrete Interactions, UsersManual," NUREG/CR-5843 (draft for comment),March 1992. Available from the NRC PDR.

(7) R. Summers et al., "MELCOR 1.8.0: A computercode for Nuclear Reactor Severe Accident SourceTerm and Risk Assessment Analyses," NUREG/CR-5531, January 1991.

(8) E. C. Beahm et al., "Chemistry and Mass Transportof Iodine in Containment," pp. 251-266,Proceedingsof the 2nd CSNI Workshop on Iodine Chemistry inReactor Safety, Toronto, Canada, June 1989.

B.2 Core-Concrete Interaction

B.2.1 Background

Core-concrete interactions would occur during a severeaccident only after penetration of the reactor vessel andflow of the core debris onto the concrete basemat. De-composition of the concrete from this interaction resultsin the release of steam and carbon dioxide, which may bepartially reduced to the combustible gases hydrogen andcarbon monoxide. As the gases pass through the hot mol-ten debris, they sparge small but potentially importantquantities of radioactive elements from the debris. Theseradioactive aerosols can be released to the containment,thereby adding to the accident source term. The majorareas of concern associated with core-concrete interac-tions during a severe accident are the complete penetra-tion of the basemat and the generation of radioactive


Appendix B

aerosols and combustible gases. Another related concernis the overheating of important structures inside the con-tainment.

The NRC has conducted an extensive program of analyticand experimental research to obtain improved under-standing of core-concrete interactions. The analyticalresearch focused on the development of models for study-ing phenomenological aspects of core-concrete interac-tions such as heat and mass transfer, while the experimen-tal research focused on conducting scaled-downexperiments simulating prototypic reactor accident sce-narios. These studies have recognized the variety of con-cretes used in nuclear power plants in the U.S. and thewidely diverse accident scenarios that lead to core-concrete interactions. The effort to understand core-concrete interactions was also broadened to include areassessment of the models used to predict radionucliderelease.

The efforts to resolve the severe accident issues associ-ated with core-concrete interactions have culminated inthe development of the CORCON computer code. Thepredictions of the early versions (CORCON-MOD 1 andCORCON-MOD2) of this computer code have beenvalidated against many large-scale tests of the interac-tions of both metallic and oxidic core debris and withconcrete.

B.2.2 Status

The experimental data base on core-concrete interactionsis extensive. This data base is predominantly the result ofresearch sponsored by the NRC and research sponsoredin Germany at the Kernforschungszentrum in Karlsruhe.Data on the interactions of uranium dioxide melts withconcrete have been obtained from the ACE programsponsored by the Electric Power Research Institute, inwhich the NRC is a partner. Additional data on core-con-crete interactions may come from the ALPHA program inJapan and studies under way in Russia.-

The first consideration regarding the range of conditionsthat could affect core-concrete interactions during a se-vere accident is the type of concrete. The data base nowavailable for core-concrete interactions includes testswith both of the general classes of concretes in use in U.S.commercial nuclear power plants (i.e., siliceous and cal-careous concretes).

Another important consideration deals with the composi-tion of the core debris itself, particularly the content ofmetallic zirconium in the mixture. The oxidation of metal-lic zirconium in the core debris can elevate the tempera-ture of the debris during the interaction with concrete,which would result in an increase in the release of nor-mally refractory fission product elements to the contain-

ment atmosphere. Also, the presence of metallic zirco-nium produces a chemically reducing environment thatcan increase the release of certain key elements.

Many of the early tests of high-temperature melts inter-acting with concrete were of an exploratory nature thatwere undertaken to identify phenomena or to developexperimental techniques. More recent tests have beenconducted to explicitly validate the models of core-con-crete interactions. Such tests have been heavily inst-rumented to obtain heat balances and data on gas genera-tion, gas composition, and aerosol production.

The available data base spans a broad range of conditionsas well as concrete types. The data base includes tests withboth oxidic and metallic debris. There are some limita-tions to the data base (i.e., the data base is of a genericnature and may not address specific issues that arise atparticular nuclear power plants). However, model predic-tions do not indicate that there are substantive uncertain-ties or issues affecting core-concrete interactions. Somediscrepancy has been traced to the material phase rela-tionship used in the CORCON code. Measurements ofthe melting properties of UO2-ZrO 2 concrete mixturesare being sponsored by the NRC, and improved modelsare being incorporated in the most recent version of thecode CORCON-MOD3.

With regard to the radionuclide release during core-con-crete interactions, an effort has been made in the devel-opment of CORCON-MOD3 to include a detailedmechanistic model of aerosol generation and radion-uclide release (the VANESA model). The model is basedon the assumption that aerosols are produced by themechanical entrainment of melt when gas bubbles burstat the surface of the melt and by vaporization of volatilemelt constituents into gases sparging through the melt.Predictions of the total aerosol generation rate comparedto results of the BETA test show that the model predic-tions are well within the expected accuracy limits of theavailable thermodynamic data base.

B.2.3 Future Plans

Refinements to the CORCON-MOD3 code are beingmade with regard to phase relations and models of non-ideal solutions. An effort to compare the code predictionsto test results will be completed. Part of the validation ofthe CORCON MOD3 code will be carried out under anarrangement with the I.V. Kurchatov Institute. The vali-dation will include comparison with data from the BETA(KfK), SURC (SNL) and ACE program. Once validationefforts are finished, the development of CORCON-MOD3 will be complete as a stand-alone model of core-concrete interactions. CORCON-MOD3 will then beincorporated into the CONTAIN and the MELCOR sys-tem level codes for severe accidents.

NUREG-1365, Rev. 1 B-6

Appendix B

B.3 Closure of Severe AccidentsScaling Methodology

B.3.1 Background

In many areas of severe accident research, the experimen-tal investigation has evolved or progressed to the pointthat experimental programs seek to resolve very specificissues of uncertainty for particular geometries and reac-tor plant configurations. The aim of these programs isoften to produce results that can be characterized asdirectly applicable to reactor behavior, or at the leastsuitable for the development of models that allow forextrapolation to severe accident reactor analysis.

As part of the Revised Severe Accident Research Pro-gram Plan that was published as NUREG-1365 in August1989, the NRC identified initiation of a severe accidentscaling methodology (SASM) development program as aprogrammatic element. Experimental investigation of se-vere accident phenomena poses a serious challenge owingto the fact that many processes involve a complex synergyof fluid flow across varied flow regimes, combined heattransfer modes, high temperature material interactions,and chemical reactions. Development and application ofan SASM, representing a structured methodology that issystematic, comprehensive and scrutable, provides theconfidence that scaled experiments faithfully reproducethe phenomena that will occur in a nuclear power plant.Further, application of an SASM provides the basis forapplication of analytical models validated against smallerscale experimental facilities to full scale.

B.3.2 Status

To address the scaling problem, the NRC implemented aSASM development program at the Brookhaven Na-tional Laboratory (BNL). A technical program group(MPG) was formed by the contractor to guide the develop-ment of the SASM and to demonstrate its efficacy byapplying the methodology to the direct containment heat-ing problem.

The results of the TPG activities culminated in the docu-mentation of the SASM and its application in NUREG-5809, "An Integrated Structure and Scaling Methodologyfor Severe Accident Technical Issue Resolution," pub-lished in November 1991.

The SASM consists of a number of steps that are groupedin three key elements:

Element 1: Specification of Experimental Require-ments, in which the experimental objectives are de-fined in terms of the technical issues.

Element 2: Evaluation and Specification for Experi-ments and Testing, in which the experimental objec-ztives are reflected in terms of scaling rationales thatare necessary to ensure that both separate-effectstests and integral-effect test data are applicable tofull-scale reactors, and that the tests include thephenomena important to the specified accident sce-nario.

Element 3: Data Acquisition and Documentation, inwhich the data base appropriate to issue resolution isestablished and documented for subsequent use.

To perform scaling analyses that satisfy the objectives ofSASM, a hierarchically based, two-tiered scaling method-ology was developed. The complex physicochemical proc-esses that characterize severe accident scenarios and theirassociated synergetic effects mandate a hierarchical ap-proach to the problem in order to make it tractable. Thetwo-tiered scaling methodology involves a top-down sys-tem scaling tier and a bottom-up or process scaling tier.The top-down system scaling analysis provides the basicconservation equations as well as the scaling rationale andsimilarity groups to be preserved in experimental design.Additionally, it is through the system scaling analysis thatquantification of the effects of any distortions is achieved.The bottom-up process scaling tier focuses on the modelsfor the important processes that drive the system re-sponse.

In the application of the SASM to the DCH issue, RPVconditions were examined to evaluate the initial andboundary conditions appropriate for experimental simu-lation. Scaled model laws for RPV discharge phenomenaand for reactor cavity phenomena were then derived us-ing the two-tiered scaling methodology. The SASMevaluation of initial and boundary conditions for DCH ina PWR for a station blackout scenario with failure of alower head penetration served as the basis for the DCHintegral-effects tests initiated in September 1991.

The scaling of important processes indicates that debrisdispersion or entrainment of corium is a strong functionof the gas velocity after conditions for the onset ofentrainment are satisfied. Expressed in terms of pressure,entrainment is a function .of the pressure ratio (reactorvessel/reactor cavity) to the 4.6 power as well as a functionof gas and fluid properties. The practical effect of thesedependencies, at least as analyzed for the Zion and Surryplants, is the very rapid increase of entrainment fractionsfrom low initial reactor pressures up to effectively com-plete dispersal at reactor pressures of approximately 400psi.

B.3.3 Future Plans

NUREG-5809 was issued as a draft for public commentin November 1991. The authors of the report acknowl-edged that the SASM application to DCH was not


Appendix B

intended to represent complete technical resolution ofthe scaling questions, as some of the scaling relationshipsderived are based on specific assumptions that requireexperimental confirmation and additional analysis. Reso-lution of those technical issues as they affect both scalingof experimental facilities and operation, as well as model-ing development for reactor analysis, will be pursued

under the DCH research program described in section2.1. The discipline of a SASM will be applied to otherexperimental programs in the SARP as necessary. Thestaff also intends to incorporate comments received onNUREG-5809 into a final version of the report, tenta-tively scheduled to be issued by December 1992.

NUREG-1365, Rev. 1 B-8

NRC FORM 335 U.S. NUCLEAR REGULATORY COMMISSION 1. REPORT NUMBER(2-89) (Assigned by NRC, Add Vol.,NRCM 1102, Supp., Rev., and Addendum Num-3201, 3202 BIBLIOGRAPHIC DATA SHEET bers, If any.)

(See instructions on the reverse) NUREG-1365Rev. 1

2. TITLE AND SUBTITLE3. DATE REPORT PUBLISHED

Severe Accident Research Program Plan Update MONTH YEAR

December 19924. FIN OR GRANT NUMBER

5. AUTHOR (S) 6. TYPE OF REPORT

Technical

7. PERIOD COVERED (Inclusive Dates)

8. PERFORMING ORGANIZATION - NAME AND ADDRESS (If NRC, provide Division, Office or Region, U.S. Nuclear Regulatory Commission, and

mailing address; if contractor, provide name and mailing address.)

Division of Systems ResearchOffice of Nuclear Regulatory ResearchU.S. Nuclear Regulatory CommissionWashington, DC 20555

9. SPONSORING ORGANIZATION - NAME AND ADDRESS (If NRC, type "Same as above"; if contractor, provide NRC Division, Office or Region,U.S. Nuclear Regulatory Commission, and mailing address.)

Same as above

10. SUPPLEMENTARY NOTES

11. ABSTRACT (200 words or less)

In August 1989, the staff published NUREG-1365, "Revised Severe Accident Research Program Plan." Since1989, significant progress has been made in severe accident research to warrant an update to NUREG-1365.

The report describes recent major accomplishments in understanding the underlying phenomena that can occurduring a severe accident. These include Mark I liner failure, severe accident scaling methodology, source term issues,core-concrete interactions, hydrogen transport and combustion, TMI-2 Vessel Investigation Project, and directcontainment heating. The report also describes the major planned activities under the SARP over the next severalyears. These activities will focus on two phenomenological issues (core melt progression, and fuel-coolant interactionsand debris coolability) that have significant uncertainties that impact our understanding and ability to predict severeaccident phenomena and their effect on containment performance. The SARP will also focus on severe accident codedevelopment, assessment and validation. As the staff completes the research on severe accident issues that relate tocurrent generation reactors, continued research will focus on efforts to independently evaluate the capability of newadvanced light water reactor designs to withstand severe accidents.

12. KEY WORDS/DESCRIPTORS (List words or phrases that will assist researchers in locating the report.) 13. AVAILABILITY STATEMENT

Unlimited14. SECURITY CLASSIFICATION

nuclear power plants source term 1Thi T Page)

severe accidents fission product transport Unclased

direct containment heating Mark I liner failure Unclassifiedcore melt progression core-concrete interaction (This Report)

fuel coolant interaction TMI-2 Vessel Investigation Project Unclassifieddebris coolability severe accident codes 15. NUMBER OF PAGES

hydrogen combustion advanced light water reactorshydrogen generation severe accident scaling methodology 16. PRICE

NRC FORM 335 (2-89)

Federal Recycling Program

SPECIAL FOURTH-CLASS RATEUNITED STATES

NUCLEAR REGULATORY COMMISSIONWASHINGTON, D.C. 20555-0001

SPECIAL FOURTH-CLASS RATEPOSTAGE AND FEES PAIDUSNRC

PERMIT NO. G-67 IOFFICIAL BUSINESS

PENALTY FOR PRIVATE USE, $300