A high-fidelity and efficient framework for point-particle direct numerical simulation based on multi-block overset grids

Abstract

In this work, we present a high-fidelity and efficient point-particle direct numerical simulation framework based on a multi-block overset curvilinear grid system, enabling large-scale Lagrangian particle tracking in complex geometries with high-order accuracy and low computational cost. To handle the multi-domain topological challenges inherent in such configurations, we develop an efficient particle storage and redistribution framework leveraging overset grid techniques. In particular, two optimization strategies have been proposed for particle redistribution: one is an innovative inter-block mapping within overlapping zones, and the other is a fast search-locate algorithm based on particle velocity. Together, these approaches significantly reduce the particle tracking overhead, especially for particles passing through interfaces between overlapping grid blocks. Moreover, the accuracy and robustness of the present framework are rigorously validated through various cases, including massless particle trajectories, one- and two-way coupled simulations. Specifically, we demonstrate the framework's applicability to the direct numerical simulation of particle-laden flow in a linear compressor cascade at engine-relevant conditions, showcasing its capability to resolve complex particle dynamics in turbomachinery configurations with low computational costs.