Options
June 2025
Conference Paper
Title
Blast Simulation: Thermal and Material Analysis
Abstract
These kinds of thermal and material analyses are crucial for enhancing the safety and efficiency of detonation facilities. They provide a detailed understanding of explosive behavior, which can lead to the optimization of their composition and application in various scenarios. Additionally, they offer valuable insights for the development of advanced numerical simulation models.
This study explores the utilization of machine learning algorithms, particularly those provided by PyTorch, to optimize blast simulations in non-commercial software. The presented simulations involve afterburning processes in a detonation bunker located at Fraunhofer ICT, employing the Miller term. To model the complex thermodynamic processes and dynamic behavior of mixtures following detonation initiation, a hybrid computational framework is proposed using our in-house simulation platform ictFOAM. The process is divided into these stages: conducting simulations with and without the Miller term to capture the detailed distribution of temperature, pressure, and reaction rates, and using machine learning models to iteratively refine parameters defining the material, thereby improving simulation performance. The integration of the Miller term shows significant potential for enhancing the accuracy of afterburning models, while the application of machine learning provides an innovative approach to refining CFD simulations over time. This advanced methodology offers a more comprehensive understanding of performance and safety dynamics in detonation facilities. Numerical results are included in the study.
This study explores the utilization of machine learning algorithms, particularly those provided by PyTorch, to optimize blast simulations in non-commercial software. The presented simulations involve afterburning processes in a detonation bunker located at Fraunhofer ICT, employing the Miller term. To model the complex thermodynamic processes and dynamic behavior of mixtures following detonation initiation, a hybrid computational framework is proposed using our in-house simulation platform ictFOAM. The process is divided into these stages: conducting simulations with and without the Miller term to capture the detailed distribution of temperature, pressure, and reaction rates, and using machine learning models to iteratively refine parameters defining the material, thereby improving simulation performance. The integration of the Miller term shows significant potential for enhancing the accuracy of afterburning models, while the application of machine learning provides an innovative approach to refining CFD simulations over time. This advanced methodology offers a more comprehensive understanding of performance and safety dynamics in detonation facilities. Numerical results are included in the study.
