An automated scan-to-FE model workflow for steel box girder exterior surfaces based on in-situ point clouds

Abstract

High-fidelity finite element (FE) models are crucial for bridge analysis, where detailed geometric information acquired by 3D laser scanning is key to ensuring model accuracy. However, direct generation of FE models from as-built point cloud data remains challenging due to computational inefficiency and data incompleteness. To address this, this paper presents a structured Scan-to-FE Model framework that reconstructs steel box-girder exterior topology from preprocessed point clouds and automatically instantiates a solver-ready FE model via command scripting. The methodology is implemented on the scanning data from a large-span bridge, covering multiple girder segments. Geometric fidelity is quantified using cloud-to-mesh distance statistics, confirming that the reconstructed mesh preserves measured surface features. Computational feasibility and the fidelity robustness of the generated FE model was validated through simulation under representative static load cases and parametric sensitivity analysis.