PANDAROSA Probabilistic ANd Deterministic Analysis for Risk-Oriented Severe Accidents Vapor Explosion Risk in Nordic LWRs Introduction Proposed by Park, Hyun Sun, Ph.D. SINEWSTEK Gustav III:s Boulevard 41, 3TR 16973, Solna, Sweden sun@sinewstek.com Severe accident analyses for light water reactors (LWR s) have always considered the likelihood and consequences of vapor explosions [1-4] which produces an energetic dynamic load to the structures (importantly containment, a final barrier of the defense-in-depth safety system) when high temperature corium and cold water comes in to contact during the actions of terminating severe accident progression. Based on experiments and related analyses, a vapor explosion is not considered a credible threat to direct containment failure; i.e., alpha-mode failure, containment failure by an upper PRV head missile generation due to in-vessel vapor explosion (IVE) [5, 6]. IVE threat has been thoroughly examined by a group of experts in the SERG-I, SERG-II [5] and OECD/FCI Specialist meetings [6], recommending that the probability of the threat is below 10-3. However, dynamic pressures resulting from the vapor explosion can damage the RPV integrity which increases the probability of vessel failure when the accidents progress future to the ex-vessel scenarios. Thereby the dynamic loads must still be evaluated to determine the effect on the reactor pressure vessel (RPV) [7]. In particular, ex-vessel vapor explosion (EVE) has been a long pending issue in Nordic LWRs [8] since (a) in some PWRs, water discharged from the reactor primary system accumulates in the reactor cavity under the vessel; and (b) in some BWRs, a deep water pool is established under the vessel, prior to vessel failure: an accident management strategy employed in the Swedish BWRs. The strategy is based upon an assumption that a risk of EVE due to the flooded water is negligible. It is feasible that without EVE hazard potentials the flooded water below RPV can provides an ultimate cooling of corium. Unfortunately, however, it is acknowledged that the strategy purveys favorite conditions for energetic vapor explosions which may produce a significant threat to the integrity of surrounding supporting structures and reactor containment. For instances, the ex-vessel water is generally highly subcooled and the extensive voiding, that develops in the premixture in a saturated pool, may not occur in the subcooled pool. Additionally, it has been found that the median particle size, obtained during the break-up process, may be much smaller for the subcooled water than for the saturated water. It is clear that if a relatively large vapor explosion occurs near the bottom of the 7 to 11 meters pool; the Swedish BWR containment (in particular the pedestal) can fail. Therefore precise evaluation of potential risk of the threat is not readily achievable since uncertainty involved in EVE is vast. Often performs the evaluation in a conservative bounding analysis providing seldom practical results for rational decision-making on the issue. Much experimental and analysis-development work is in progress, presently, on in- and ex-vessel vapor explosions. Experiments have been performed with less than gram quantities to several kilogram quantities of heated particles and molten materials. Elaborate three-field analysis codes: TEXAS-V [9-11], JASMINE [12], MC3D [13], and PM-ALPHA/ESPROSE.m [14,15] have been developed. Those codes are the main tools to evaluate the vapor explosion hazard potentials for new design and existing rectors. Interestingly, the nature of high-risk and large uncertainty vapor explosion phenomena moves new reactor designers like EPR (AREVA) to consider a design concept of a vapor-explosion free dry corium catcher ultimate safety measure of termination of severe accident progression. However, all the other reactor designs in new and existing power plants still have a high risk due to vapor explosions. Therefore, vapor explosion risk has been continously studied in a broader framework, like the OECD project, called SERENA-I (completed) and SERENA-II (in progress). The projects aims to develop more reliable steam explosion models with well-controlled PANDAROSA2 1 / 5
prototypic melt-coolant internation experiments (new KROTOS and TROI) and to examine the code capability of reactor applications. The code applications for reactor applications primarily aimed to assess the uncertainties among codes which have different approaches and models for simulating vapor explosion phenomena. No scrutinized effort was provided to the vapor explosion risk to specifically reactor buildings such as containment in the projects. In recent years, vapor explosion analysis has performed for new reactors designs and their design-certificate applications such as AP600[16], AP1000[17], ESBWR, APWR, APR1400 [18], etc. In particular, we are currently analyzing the potential EVE loads [18] in cases of two scenarios; (1) the partially flooded cavity and (2) failure of in-vessel retention with external reactor vessel cooling (IVR/ERVC). Therefore, the project designed for three years aims to a comprehensive analysis of the hazard potentials involved in vapor explosions (in-vessel and ex-vessel) in specifically Nordic LWRs by developing a tool to conduct an integrated methodology of probabilistic and deterministic analysis, verifying vapor explosion models and codes with most updated well-controlled experimental database, assessing vapor explosion risks for specific Nordic LWRs and finally synthesizing the analysis results into a quantitative assessment for the Nordic LWRs. The first year effort focuses on the development of window-based software to perform stratified random sampling based Monte-Carlo simulation with a mechanistic phenomenology code for vapor explosion, like TEXAS-V and the comprehensive verification of the TEXAS-V codes with experimental database with uncertainty analysis using the software. Project Work Plan and Schedule Figure 1. PANDAROSA Platform for Vapor Explosion Analysis Work Plan The framework of the analysis is based upon a conventional uncertainty analysis with a stratified random sampling based Monte-Carlo with integrated with deterministic mechanical computer model such as TEXAS-V. In this analysis, a stratified sampling technique, the Latin-Hypercube Sampling (LHS) method, is employed to reduce the number of code analysis without scarifying statistical characteristics. The TEXAS-V code, one of main computer codes for vapor explosions in reactor applications, is used for the deterministic analysis in the project. The TEXAS-V code employs a 1-D Eulerian-Lagrangian models for mixing and explosion. The new TEXAS-V encompassed with ANASYS-CFX3D module to evaluate the explosion shock pressure propagation in a 3-D geometry. PANDAROSA2 2 / 5
In this project, the analysis follows as shown in Figure 1; (1) the selection of random input sets by LHS method, (2) the execution of multiple code calculations with the selected input sets, and (3) the analysis of resulting probability distribution functions of EVE loads or risk. Large uncertainties involved in EVE analysis for the Nordic LWRs is also examined by testing uncertainties in (1) vapor explosion model parameters of the TEXAS-V code, (2) key input parameters with "best-estimated" probability density distribution, and (3) selected probability density functions for key input parameters. Project Tasks The project is designed for three years: In this proposal, TASK-I will be carried out in year 2011. TASK 1 (2011): Development of PANDAROSA-VERNs Framework Milestone 1-I: Development of PANDAROSA platform for Latin Hypercube Sampled Monte-Carlo Analysis for Vapor Explosion Risk (6 Months) Development of LHS based Monte-Carlo analysis with a mechanistic computer code Modification of the TEXAS-V code for automatic multiple execution Development of Window software platform (called PANDAROSA) for executing the PANDAROSA platform Milestone 1-II: Validation and uncertainty analysis of the TEXAS-V codes with the PANDAROSA software against existing experimental database (KROTOS, TROI etc.) (4 Months) Validation of mixing models Validation of explosion models Validation of code specific parameters Milestone1-III: Demonstration of the applicability of PANDAROSA for reactor application with a standard problem selected in the SERENA-I project (2 Months) Analysis of the EVE standard problem in the SERENA-I project TASK 2 (2012): Plant Specific Analysis of EVE Risk with the PADAROSA-VERNs Framework Milestone 2-I: PANDAROSA-VERNs Analysis of EVE for a Specific Nordic BWR (6 Months) Selection of a specific Nordic BWR for EVE assessment with PANDAROSA-VERNs Identification of EVE risk-significant severe accident scenarios and associated conditions for EVE analysis (Results of MELCOR, MAAP analysis and PSA analysis) (2 Months) Assessment of EVE risk for the specific BWR with PANDAROSA-VERNs (4 Months) Milestone 2-II: PANDAROSA-VERNs Analysis of EVE for Specific Nordic PWRs (6 Months) Selection of a specific Nordic PWR for EVE assessment with PANDAROSA-VERNs Identification of EVE risk-significant severe accident scenarios and associated conditions for EVE analysis (Results of MELCOR, MAAP analysis and PSA analysis) (2 Months) Assessment of EVE risk for the specific PWR with PANDAROSA-VERNs (4 Months) TASK 3 (2013): Plant Specific Analysis of IVSE Risk with the PADAROSA-VERNs Framework and Synthesis of Overall Risks involved in Vapor Explosions for Nordic LWRs Milestone 3-I: PANDAROSA-VERNs Analysis of IVSE for Specific Nordic BWRs (4 Months) Milestone 3-II: PANDAROSA-VERNs Analysis of IVSE for Specific Nordic PWRs (4 Months) Milestone 3-III: Synthesis of the PANDAROSA-VERNs Analysis on vapor explosion risks for Nordic LWRs (4 Months) Expected Contribution to Nordic Nuclear Safety Network The project aims to build an assessment methodology, development analysis platform, and PANDAROSA2 3 / 5
demonstrative analysis (first year), plant specific analysis for EVE (second year) and plant specific analysis for IVE and synthesis of vapor explosions risk in Nordic LWRs (third year). The methodology will apply to Nordic LWRs with dry-well or cavity flooding strategy for their SAMG which may be vulnerable for ex-vessel vapor explosion risk since no comprehensive application of the methodology was applied in the evaluation of risk of in-vessel and ex-vessel vapor explosion. The project contributes the improved competence of evaluation capability of risk-significant analysis to Nordic network. The analysis applied for the Nordic nuclear power plants in operation will provide scrutinized information in Nordic network for better decision making on safety issues related to the current SAMG triggered concern of vapor explosions. Existing Experiences and Resources Experiences The project is deeply rooted to several other projects in NKS projects, Nordic national research programs and EU programs which a SINEWSTEK member, (H.S., Park) was involved. In NKS projects, the Pre-Melt-Del Project (Sehgal and Park 2004) [8] provided a guideline of a direction to research severe accidents addressing in terms of issue resolutions. The EXCOOLSE Project (Park, 2004-2006 [19-21], Dinh, 2007-2010) has been conducted to focus on ex-vessel severe accident phenomena including, coolability and vapor explosions. In these projects, as for vapor explosions, the focus was given to investigate the scientific questions on the mechanisms of vapor explosions including studying on the mechanical behaviors of vapor explosions among different melt compositions using one-of-kind photo-radiographic high speed visualization facility [22]. The proposed project is in the line of the projects, extending understanding of fundamentals to implementing the understandings to practical plant applications. In national projects, a Nordic (Swedish, Finnish, and Swiss) research program called APRI [23] sponsored by nuclear regulatory bodies and nuclear utilities) has performed extensive research on severe accidents. In the European Projects, research on severe accidents have been very extensive and gravely contributed to the current knowledge and understanding on severe accident analysis. SARNET-I and II [24, 25] As one of outcomes, Areva developed EPR with severe accident safety measure employing a vapor-explosion free dry corium catcher completing their innovative catch-spreading-quenching strategy for ultimate termination of severe accident progression. However, all the other reactor designs in new and existing power plants still have a high risk due to vapor explosions. Therefore, vapor explosion risk has been continuously studies in a broader framework, the OECD project, called SERENA-I (completed) and SERENA-II (in progress). The projects aims to develop more reliable vapor explosion models with well-controlled prototypic melt-coolant interaction experiments (new KROTOS and TROI) and to examine the code capability of reactor applications. The code applications for reactor applications primarily aimed to assess the uncertainties among codes which have different approaches and models for simulating vapor explosion phenomena. No scrutinized effort was provided to the vapor explosion risk to reactor buildings such as containment. SINRESTEK is currently providing technical consultancy on the TORI experimental program in the OECD/NEA SERENA-I and II projects. SINEWSTEK is involved in an EVE analysis for Korean APR1400 reactors which is under construction in Korea (Shin Gori 3 and 4) and soon in UAE as well as is under pre-application process in US-NRC DC process. The reactor with relatively long cavity depth similar to Nordic BWRs employs the invessel retention strategy with external reactor vessel cooling (IVR/ERVC)for their SAMG when severe accident occurs. This strategy creates similar concerns on EVE to Nordic LWRs. TEXAS-V Code SINEWSTEK has extensive experiences in the TEXAS-V code. The latest version of the code improved its capability to evaluate the dynamic pressure effects of a triggered vapor explosion involve: (1) Model corium-water mixing and vapor explosion propagation with the TEXAS model; (2) Use the explosion dynamic pressures predicted by the TEXAS model at the structural boundary by a finite-element ANSYS model; (3) Develop a failure criterion for the structural wall (RPV or containment) to compare to the dynamic pressures computed by steps (1) - (2). Models for vapor explosions have been developed for alpha-mode failure analyses [9]. Specifically, the PANDAROSA2 4 / 5
TEXAS model has been under development in the 1990 s by Corradini et al [10, 11]. The basic elements of TEXAS are: (1) Eulerian fields for the coolant liquid and vapor and a LaGrangian field for the molten corium melt. (2) The Lagrangian fuel model considers the fuel to be injected as parcels of molten material with a characteristic size, velocity and temperature. These parcels can dynamically fragment due to fluid instability mechanisms induced from the relative velocity between fuel and coolant. (3) Given a trigger, rapid fuel fragmentation occurs by thermal-fragmentation mechanisms of the parcels. Reference 1. NRC, WASH-1400, NUREG-75/0114 (Oct. 1975). 2. Corradini, M.L., D.V.Swenson, SAND80-2132 (June 1981). 3. Corradini, M.L., and D.V.Swenson, SAND81-1092 (October 1981). 4. Theofanous, T.G., B.Najafi, and E.Rumble, Nuclear Science and Engr, 97, pp259 (1987). 5. NRC, SERG2, 1995. NUREG-1529. 6. Theofanous, T.G., W.W. Yuen, NUREG/CP-0127, (January 1993). 7. Theofanous, T.G. et al., Nuclear Science and Engineering 97 (1987), pp. 259 326 Parts I-IV 8. Sehgal B.R. and H.S. Park, 2004, Final Report of the PRE-DELI-MELT project, NKS. 9. Chu, C.C. Corradini, M.L. Nuclear Science and Engineering, V 101, No 1, p 48-72 (Jan 1989). 10. Corradini M.L. et al, A Users Manual for TEXAS-V: (Aug. 2000). 11. Corradini M.L. et al, FCI Analyses with the TEXASV Vapor Explosion Model,(Dec. 1999). 12. Moriyama, K.et. al, NUTHOS-6) Nara, Japan, (CD-ROM: Paper ID. 264) (2004). 13. Brayer C.and Berthoud, G. JAERI-Conf 97-011, NEA/CSNI/R(97)26. 14. Theofanous, T.G. et al., Nuclear Engineering and Design, 189:59-102, 1999. 15. Theofanous, T.G. et al., Nuclear Engineering and Design, 189:103-138, 1999. 16. NRC, The AP600 Standard Design, NUREG-1512, (1998) 17. Esmaili, H., et. al., The AP1000 Standard, NUREG/CR-6849, ERI, 2004 18. Park, H.S., Consulting contract with KOPEC for APR1400 EVE Risks, Korea (2009,2010) 19. Park, H. S., and Dinh, T-N, NKS-156, ISBN 978-87-7893-220-4, (April 2007). 20. Park, H. S., and Dinh, T-N, NKS-132, (April 2006). 21. Park, H. S., et al., NKS-112, NKS, (October 2005) 22. Park, H. S., et. al, J. of Exp. Thermal and Fluid Sci., V. 29/3 (2005). 23. Park, H. S., et al., SKI Report 2006:28, (April 2006) 24. Lindholm, I. et al., 1st European Review Meeting on Several Accidents, Aix-en-Provence, 2005 25. Bonnet J.M. et al., 1st European Review Meeting on Several Accidents, Aix-en-Provence, 2005 PANDAROSA2 5 / 5