Coupling of Particle-based Simulation and MARS Code for Simulation of IVR-ERVC: Preliminary Study

In-Vessel Retention (IVR) through the External Reactor Vessel cooling (ERVC) is a major severe accident mitigation strategy to confine the core melt inside the lower head of the reactor vessel during the late process of core damage [1, 2]. The essence of the IVRERVC is to stably and sustainably remove the thermal load of the core melt trapped inside the vessel by transferring to the external coolant. In this regard, the heat transfer mechanism of the corium pool is the most important consideration because it determines the safety criterion of the reactor vessel by evaluating thermal margin to Critical Heat Flux (CHF) [2]. The thermal load that the corium pool exerts on the reactor vessel is influenced by various factors such as the thermal and hydrodynamic behavior of the corium, the heat removal rate on the outer vessel wall, and the composition and chemical behavior of the corium. In addition, the stratification/mixing of the oxide-metallic corium pool or crust formation affects the heat transfer mechanism of the corium pool [1]. To understand the complicated in-vessel corium behavior and evaluate the applicability of plant scale reactors, many benchmark experiments have been conducted, but the results are rarely applied to the safety assessment of real scale accident due to the limitation of scalability and materials [1]. Currently, based on these experimental results, several studies have developed numerical models or correlations and they have been applied to plant safety analysis. These numerical methods are mainly based on the fixed grid-based method (e.g. FVM, FEM, and FDM). Due to the nature of Eulerian based method, they suffered from handling non-confined domain such as natural convection with free surface, large interfacial deformation of stratified fluids, local phase change, and etc. This drawback has been addressed by the restrict assumptions on the complicated geometry or boundary conditions. In this sense, this study develops the integrated code platform of the SOPHIA code (Lagrangian-based Smoothed Particle Hydrodynamics (SPH) code), and MARS code (Reactor-scale system code) in order to reduce the assumptions and uncertainties of the previous methods. The SPH method, a representative Lagrangian particle-based CFD method, analyzes the flow motion following the fluid mass point instead of a fixed lattice. Thus, the fluid system is discretized into a collection of Lagrangian particles carrying the physical properties and each particle moves according to the governing equations (mass, momentum, and energy) derived from the kernel-weighted summation over nearby particles. Because of the moving particles, the SPH method enables to effectively handle free surface or multiphase/multi-fluid flow by tracking the trajectories of fluid interfaces [3]. Using the SPH method, Seoul National University has developed the multi-dimensional and multi-physics CFD code, called ‘SOPHIA’ since 2015 in order to simulate the nuclear safety-related phenomena [4]. The SOPHIA code is based on the Weakly Compressible SPH (WCSPH) method that allows a slight compressibility of the fluid using equation of state (EOS). On this basis, various SPH-formulated physical models are implemented to deal with complicated phenomena; viscous force, surface tension, heat conduction, diffusion, elastic solid mechanics, etc. Since these governing equations and physical models are expressed linearly and solved by serial calculations, GPU-based parallelization becomes optimal to the SPH method. Therefore, recently, the SOPHIA code was parallelized using the multiple GPUs and it achieved dramatic improvement of the computational performance [4, 7]. However, since the SOPHIA code is a CFD-scale code, integral simulation on the IVR-ERVC phenomena encounters physical and computational limitation. For effective and efficient simulation, IVR-ERVC needs to be dealt with separately; The SOPHIA code analyzes the complicated and detailed behavior of in-vessel corium, and MARS code analyzes the external vessel cooling system. This study aims to develop an integrated code system that couple SOPHIA code to MARS code in order to simulate the IVR phenomena more realistically and provide the best estimate of the safety analysis. For demonstrating its capability, this study performed preliminary simulation on the benchmark case that show the phenomenological characteristics of IVR-ERVC phenomena.