Scholarlink

Abstract <p>Due to a particle-based approach, the discrete element method (DEM) is preferable to finite difference or finite element methods for modeling brittle materials’ behavior as it can naturally simulate fracture formation and propagation without any additional assumptions. However, one of the drawbacks of DEM is its computational complexity, arising from the requirement to calculate all pairwise interactions between particles within a certain range. In this paper, we present an implementation of the algorithm for 2-dimensional DEM simulations using a combination of CUDA and MPI to enhance computational performance and memory management, resulting in more effective use of the method. We provide results of the algorithm’s strong and weak scaling analysis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Implementation of the Algorithm Using Hybrid Computing Systems to Simulate Porous Materials Fracturing with Discrete Element Method

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • V. D. Chepelenkova,
  • V. V. Lisitsa,
  • E. A. Gondyul

摘要

Abstract

Due to a particle-based approach, the discrete element method (DEM) is preferable to finite difference or finite element methods for modeling brittle materials’ behavior as it can naturally simulate fracture formation and propagation without any additional assumptions. However, one of the drawbacks of DEM is its computational complexity, arising from the requirement to calculate all pairwise interactions between particles within a certain range. In this paper, we present an implementation of the algorithm for 2-dimensional DEM simulations using a combination of CUDA and MPI to enhance computational performance and memory management, resulting in more effective use of the method. We provide results of the algorithm’s strong and weak scaling analysis.