•  
  •  
 

CIB Conferences

Abstract

This study introduces a proof-of-concept agentic framework designed to assist structural engineers in managing sophisticated computational workflows for complex structures. The proposed framework automates the structural analysis lifecycle while adhering strictly to procedural controls. Unlike conventional workflows that rely on graphical user interfaces or automation scripts, this framework utilizes large language model (LLM) orchestrators to decompose complex structural problems into executable sub-tasks, such as structural data extraction, structural analysis, and results visualization and discussion. Furthermore, the data extraction agent establishes a live connection to AutoCAD to extract structural data, the physics-based agent interacts with the finite element software OpenSees to conduct analyses, and the visualization agent graphically presents the calculated results. We demonstrate the efficacy of the framework through a case study of a spatial truss structure. Results indicate that the autonomous framework demonstrates the feasibility of automating manual modeling and structural evaluation. This research provides a scalable foundation for the next generation of autonomous engineering assistants, bridging the gap between generative artificial intelligence and rigorous structural mechanics.

Keywords

Large Language Models, Agentic AI, Workflow Automation, Truss Structures

Download

Share

COinS