Agentic Large Language Models for Automated Structural Analysis of 3D Frame Systems

The paper presents an agentic LLM framework designed for automated structural analysis of 3D frame systems, extending previous work focused on 2D plane frames. It introduces a novel geometric representation using 2D projections and employs a multi-agent pipeline to process natural language inputs into structured outputs, achieving an average accuracy of 90% on ten 3D frame evaluations. This advancement is significant for practitioners as it enhances the automation of structural engineering tasks, allowing for more efficient analysis and design workflows.