• Energy Research

Steam explosions in nuclear reactors pose significant risks to reactor safety and containment integrity during severe accidents. This study addresses the challenges of accurately simulating shockwave propagation and structural impact in such events by establishing a unified Smoothed Particle Hydrodynamics (SPH) framework. The proposed SPH model was optimized using GPU parallelization and validated against experimental results from shock tube, underwater explosion and high-velocity impact tests, demonstrating its efficacy in capturing shockwave-structure interactions. The model was applied to simulate steam explosions in the APR1400 reactor cavity under various conditions, examining factors such as the equation of state (EOS) for water, structural materials and explosive forces. The study further found that the choice of EOS and structural material greatly influenced the peak pressure and the extent of damage, with the Mie-Grüneisen EOS producing higher peak pressures and reinforced concrete showing higher damage resistance. These findings suggest that the unified SPH approach provides a comprehensive and robust framework for assessing steam explosion risks, offering crucial insights for optimizing reactor design and safety measures against such events.

Abstract Recently, a novel concept of integrating Very High Temperature Reactor (VHTR) waste heat to Forward Osmosis (FO) system has been proposed. The proposed nuclear desalination system possesses a significantly higher energy utilization rate, however at the same time introduces a tritium exposure issue. This is especially critical as the VHTR integrated FO system produces potable water since tritium is especially hazardous when ingested. In this study, a numerical code named BOTANIC is developed using a chemical process analysis code, gProms, in order to understand tritium behaviors in the VHTR-FO system and migration of tritium to downstream processes. The code involves tritium generation, sorption, leakage, purification, recombination, dissociation, permeation, trapping, release models. The developed code is verified using the analytical solutions and the benchmark code in stepwise approach. Using the developed BOTANIC code, tritium behavior in the proposed VHTR-FO system is analyzed and sensitivity analysis is extensively conducted in order to figure out the effective measures for reducing tritium level in the final product. Based on the sensitivity analysis results, two mitigation concepts are suggested and investigated; (1) protective barrier in PHX and (2) ceramic PHX.

This paper presents the finite element deformation and failure simulation of a typical Korean high-power reactor vessel under a severe accident characterized by large break loss of coolant (LBLOCA) with in-vessel retention of molten corium through external reactor vessel cooling (IVR-ERVC) conditions. Temperature distributions calculated using Modular Accident Analysis Program Version 5 (MAAP5) as thermal boundary conditions were used, and ABAQUS thermal and structural analyses were performed. After full ablation, the temperature of the inner surface in the thinnest section remained high (920 °C), but the stress remained relatively low (less than 6 MPa). At the outer surface, the stress was as high as 250 MPa; however, the resulting plastic strain was small owing to the low temperature of 200 °C. Variations in stress, inelastic strain, and temperature with time in the thinnest section suggest that the plastic and creep strains are saturated owing to stress relaxation, resulting in low cumulative damage. Thus, the lower head of the vessel can maintain its structural integrity under LBLOCA with IVR-ERVC conditions. The sensitivity analysis of internal pressure indicates the occurrence of failure in the thinnest section at an internal pressure >9.6 MPa via local necking followed by failure due to high stresses.

Abstract Lagrangian-based meshless CFD methods have recently been developed and are being gradually applied to various research areas. In nuclear engineering, the meshless methods are gaining attention in modeling natural disasters and severe accident phenomena, such as tsunami, molten corium behaviors, etc., because of its flexible computational domain and effective treatment of non-linear deformations. This paper introduces recent progress and on-going activities in Lagrangian-based CFD code development at Seoul National University (SNU). The code, named SOPHIA, is based on the smoothed particle hydrodynamics (SPH) method, one of the best-known meshless numerical methods, based on a Lagrangian framework that can easily handle various types of physics because of its simplicity in expressing and solving mathematical equations. The SOPHIA code currently incorporates basic conservation equations and various physical models, including fluid flow, heat transfer, turbulence, multi-phase, phase change, elastic solid, diffusion and so on. To handle multi-phase, multi-component, and multi-resolution flows simultaneously, the SOPHIA code formulates density and continuity equations based on the normalized-density form, which has been recently developed and is proposed in the current study. The SOPHIA code also adapts CUDA GPU architectures for parallelization and performance improvement. Based on the benchmark calculations, the parallelized GPU code shows much higher computational speed by two orders of magnitude compared to the single CPU code for 1.0 million particles. This paper summarizes the overall features of the SOPHIA code, including governing equations, algorithms, and parallelization methods, along with some benchmark simulations.

Abstract The Divergence-Free SPH (DFSPH) method is a recently proposed novel incompressible Smoothed Particle Hydrodynamics (SPH) method. The DFSPH method enforces incompressibility via two iterative solvers: the Divergence-Free (DF) solver and the Constant-Density (CD) solver. In this study, the DFSPH algorithm is implemented into the SOPHIA Plus framework to simulate a set of wave propagation under the same geometry and conditions of a laboratory-scale experiment. The experiments are conducted for wave propagation and interaction with a structure of scaled-down undersea topography near the Kori nuclear power plant in South Korea. This study compares the free surface propagation and the wave height with the experimental measurements for three test cases: low/medium/high frequency waves. Overall, the simulation shows good agreement with the experiment both qualitatively and quantitatively. However, according to the sensitivity study, more realistic water splashing behaviors are captured as the particle size is reduced. The predicted wave heights at three different locations are also in fairly good agreement with the experimental measurement. Slight differences are observed after the wave collides with the structure because of the low energy dissipation in the simulation. However, the differences are not significant.

In this study, we propose a fully parallelized adaptive particle refinement (APR) algorithm for smoothed particle hydrodynamics (SPH) to construct a stable and efficient multi-resolution computing system for nuclear safety analysis. The APR technique, widely employed by SPH research groups to adjust local particle resolutions, currently operates on a serialized algorithm. However, this serialized approach diminishes the computational efficiency of the system, negating the advantages of acceleration achieved through high-performance computing devices. To address this drawback, we propose a fully parallelized APR algorithm designed to enhance both efficiency and computational accuracy, facilitated by a new adaptive smoothing length model. For model validation, we simulated both hydrostatic and hydrodynamic benchmark cases in 2D and 3D environments. The results demonstrate improved computational efficiency compared to the conventional SPH method and APR with a serialized algorithm, and the model's accuracy was confirmed, revealing favorable outcomes near the resolution interface. Through the analysis of jet breakup, we verified the performance and accuracy of the model, emphasizing its applicability in practical nuclear safety analysis.

The in-vessel retention through external reactor vessel cooling (IVR-ERVC) strategy is a key management strategy for early termination of a nuclear severe accident that can threaten the integrity of the reactor vessel. To simulate the physical phenomena of the molten corium, the smoothed particle hydrodynamic (SPH) method is utilized in this study. The SPH method is a Lagrangian computational fluid dynamic (CFD) method that can simulate multi-fluid stratification, turbulence, natural circulation, radiative heat transfer, thermal ablation, and crust formation. To address the external vessel cooling, it is coupled with a conventional 1-D nuclear system analysis method. The 1-D system analysis code can calculate the two-phase natural circulation of cooling water and the convective heat transfer on the external reactor vessel wall. These two simulation codes exchange the temperature and heat flux of the reactor vessel outer wall. This study numerically simulated the IVR-ERVC strategy for a Korean high-power reactor and compared it with the traditional lumped parameter method (LPM). Unlike LPM, this study provides localized detailed data about the thermal hydraulic behavior of molten corium and visualization of phenomena in the IVR-ERVC strategy. This enhances our understanding of the phenomena in IVR-ERVC strategy and introduces new perspectives.

Understanding the behavior of molten corium is crucial for effective heat management and ensuring the integrity of reactor containment during severe nuclear incidents. Corium spreading is a complex multi-physics phenomenon influenced by hydrodynamics, heat transfer, and phase change. The Smoothed Particle Hydrodynamics (SPH) method is utilized to study corium flow, leveraging its ability to handle complex fluid dynamics and its compatibility with parallel computing architectures. This research examines two specific corium spreading cases: the VULCANO VE-U7 experiment, which is characterized by a wide mushy zone, and the FARO L26S experiment, noted for its narrow mushy zone. The study compares spreading lengths and temperature profiles over time with experimental data, offering a detailed analysis of the observed spreading behaviors. Variations in physicochemical properties result in distinct spreading patterns between the cases, highlighting the complexity of corium behavior under different scenarios. The findings demonstrate the applicability and capability of the SPH method in capturing various characteristics of corium spreading effectively.

Abstract Dry cask storage systems (DCSS) is a method of storing high-level radioactive nuclear spent fuel. Due to the decay heat from fission products, effective cooling of the spent fuel is one of the key roles of dry cask storage systems. This study proposes a method to improve the cooling performance of DCSS by adding a small amount of high-density inert gas into helium backfill gas and verifies the method by numerical analysis. First, the candidate group of additive gases that can improve heat transfer in the DCSS environment is selected through figure-of-merit (FOM) analysis based on natural convection heat transfer models. Then, the candidate gases are evaluated using detailed computational fluid dynamics (CFD) modeling without the porous media assumption. From the analysis, it is found that Xe and Kr can reduce the peak cladding temperature (PCT) inside the cask at pressure greater than 1.5 atm. The optimal composition of the additive gas (Xe and Kr) is estimated to be approximately 0.8 in mass fraction. The effect of the additive gases becomes more significant as the pressure increases due to the effect of increased density.