Scholarlink

<p>This study presents a&#xa0;physical-mathematical model and a&#xa0;numerical method for simulating detonation flows in gas suspensions of finely dispersed reacting particles. The numerical algorithm employs the Harten-Lax TVD scheme for the gas phase and the Gentry-Martin-Daly scheme for the particle phase, with time derivatives approximated using Runge-Kutta methods. The algorithm was tested on the problem of heterogeneous detonation initiation and propagation in a&#xa0;plane channel. The computational efficiency of different orders of Runge-Kutta schemes was evaluated. To enhance performance, a&#xa0;parallel programming approach using the OpenMP library was implemented for heterogeneous detonation problems. The parallelization efficiency of the code was tested on various computer systems. Performance tests on AMD Ryzen and Threadripper processors demonstrated computation time reductions up to 11-fold using 35 threads. Approximate relationships between computation time and both the number of grid nodes and the number of threads were derived, providing a&#xa0;basis for predicting computational requirements. The obtained approximation graphs are compared with Amdahl’s law.</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

Testing a numerical algorithm for calculating detonation processes in gas suspensions of aluminum particles

  • 期刊论文

  • 发表日期：

    2026年4月13日

查看全文
  • A. A. Afanasenkov

摘要

This study presents a physical-mathematical model and a numerical method for simulating detonation flows in gas suspensions of finely dispersed reacting particles. The numerical algorithm employs the Harten-Lax TVD scheme for the gas phase and the Gentry-Martin-Daly scheme for the particle phase, with time derivatives approximated using Runge-Kutta methods. The algorithm was tested on the problem of heterogeneous detonation initiation and propagation in a plane channel. The computational efficiency of different orders of Runge-Kutta schemes was evaluated. To enhance performance, a parallel programming approach using the OpenMP library was implemented for heterogeneous detonation problems. The parallelization efficiency of the code was tested on various computer systems. Performance tests on AMD Ryzen and Threadripper processors demonstrated computation time reductions up to 11-fold using 35 threads. Approximate relationships between computation time and both the number of grid nodes and the number of threads were derived, providing a basis for predicting computational requirements. The obtained approximation graphs are compared with Amdahl’s law.