Research on BIM-to-FEM Seamless Conversion for Transportation Structural Engineering and Its Digital Twin Applications
Data silos and topological incompatibility between building information modeling (BIM) geometric models and finite element method (FEM) analysis models in transportation infrastructure engineering represent critical bottlenecks that impede real-time digital twin analysis and the intelligent transformation of the industry. Based on a critical review of existing BIM-to-FEM conversion methods and their limitations, this study proposes a “BIM-FEM” seamless conversion and dynamic twin mapping method that integrates parametric modeling with finite element meshing, with modeling and repair time reduced from 16 h to 3 h, and the maximum element aspect ratio improved from 84.78 to 16.59. In terms of geometric topology, we propose a collaborative construction method in which finite element hexahedral meshing rules drive BIM parametric modeling in reverse. By regularizing the decomposition of axis lines and cross-sectional feature points of linear transportation structures and optimizing their topology, we achieve fully automated hexahedral meshing without topological errors. In terms of mechanical analysis, an “offline pre-solution, online superposition” computational order-reduction model is proposed. This reduces the high-dimensional full-range finite element solution of dynamic traffic loads to a dot product operation between the influence line matrix and real-time load vectors, enabling sub-second computational response under high-concurrency dynamic traffic conditions—specifically, single-point mapping takes less than 0.27 ms, incremental updates are controlled within 0.2 s. In terms of spatiotemporal mapping and system applications, a high-fidelity “FEM-BIM” mapping mechanism based on inverse isoparametric transformation and AABB (Axis-Aligned Bounding Box) spatial indexing has been established, supporting real-time rendering of 3D cloud maps on the web and digital twin applications in engineering. Applications of this method in real-world bridge engineering digital twin systems have demonstrated its ability to perform automatic structural safety assessments and health condition predictions with an overall computation time reduction of approximately 73% compared to conventional approaches. This addresses the shortcoming of traditional structural health monitoring—which emphasizes sensor-based identification over mechanistic evaluation—and provides a viable path for intelligent, precise management and maintenance of transportation infrastructure throughout its entire life cycle.