A Stabilization Strategy of Multi-Resolution Multiphase MPS Method

Abstract CFD simulations of multiphase behavior are essential in the safety analysis of nuclear reactors. As a particle method, the MPS method inherently possesses the capability of capturing interfaces or free surfaces. The PArticle method for Nuclear reactor thermal hyDraulic Analysis (PANDA) is a software platform based on particle methods, aiming to provide a safety analysis of key thermal-hydraulic phenomenon in nuclear reactors. The integrated PANDA-BD adopts the specially designed Multiphase MPS method (MMPS method), and shows great ability in difficult multiphase simulation. However, due to the semi-implicit algorithm and the uniformity of particle size, MMPS incurs a relatively high computational cost. To improve the efficiency, some multi-resolution techniques have been developed for the MPS method. Another challenge with the MPS method could be the stabilizing issue. A seriously bias particle distribution could lead to the failure of the simulation. Most of the current rearrangement methods assume a uniform particle size, and are therefore not suitable under a multi-resolution scheme. In this study, the concept of local compressed ratio (LCR) is proposed to measure the uniformity of the particle distribution under multi-resolution. An optimized particle-shifting technique based on the novel interface recognition and LCR is developed. A series of regularization tests are performed, showing the regularization ability under both single and multiple resolution cases. Simulations of Rayleigh-Taylor Instability are presented, the novel interface detection is tested and shows good accuracy, the robustness and efficiency of the proposed method for multiphase simulation is proven.