Title: An End-to-End Finite Element Analysis Framework for Solid Mechanics Powered by Multi-AI Agents

Abstract: As the premier numerical method for solid mechanics, Finite Element Analysis (FEA) faces significant hurdles, including a steep learning curve for novices and the risk of erroneous simulations stemming from misconfigured critical components like boundary conditions, load cases, and solution variables. Typically, extensive engineering expertise is required to solve practical problems effectively. To mitigate these challenges, we introduce AbaqusAgent, a multi-agent framework based on Large Language Models (LLMs) designed for solid mechanics analysis. This system streamlines the creation and execution of analysis cases within Abaqus—a leading FEA software—by translating user instructions from natural language into executed FEA models and visualized results. AbaqusAgent comprises six distinct agents: an interpreter, architect, input writer, runner, reviewer, and visualizer, which collectively manage the essential pre- and post-processing stages of standard FEA workflows. The framework was validated against a diverse set of 50 solid mechanics problems, achieving an overall success rate of 86%. Beyond enhancing FEA efficiency and reducing the barriers to computational mechanics education, AbaqusAgent transforms human-simulation interaction paradigms and facilitates integration with AI-driven optimization and material characterization processes. The source code is accessible at https://github.com/LIRAM-LIN/AbaqusAgent

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC