An efficient framework for structured material particle generation in multi-context modeling

Zenan Huo

doi:10.1007/s00366-025-02222-z

An efficient framework for structured material particle generation in multi-context modeling

2025-10-17

0

PRE

AI

DOI：10.1007/s00366-025-02222-z

Original

Original request for help

Save

Abstract

En 中文

High-fidelity Material Point Method (MPM) simulations are particularly sensitive to preprocessing quality. This paper presents a novel Julia-based framework for rapid generation of high-quality structured material particles for 2D/3D and complex-terrain models. The framework performs partition recognition to assign heterogeneous properties and discretizes complex terrain based on known subsurface structures or failure surfaces predicted by the sloping local base level (SLBL). To inform the selection of an appropriate initial grid spacing, theoretical upper bounds are established for the geometric discretization error and for the property error associated with the three-partition model. We validate the framework on benchmark cases and show that fine geometric features are accurately preserved. Excluding data loading and writing, up to 100 million structured material particles are generated within 10 s. The framework supports algorithm composition, requires no volumetric meshing, and enables rapid resolution sweeps during preprocessing. MPM simulations, including a landslide case and a reproducible parametric study integrated with SLBL, further demonstrate the framework’s robustness and broad applicability. The framework thus provides an efficient, general-purpose preprocessing solution for MPM, with potential for transfer to other particle-based methods.

Keywords:

MPM

Structured material particle generation

Multi-context modeling

Julia language

An efficient framework for structured material particle generation in multi-context modeling

Abstract

Journal

Organization