Structural collapse simulation using a hybrid FEM-rigid body dynamics approach

Abstract

Compared to continuum method, the discrete collapse simulation method, particularly the Applied Element Method (AEM), is generally more effective in modeling separation. However, it still requires a significantly large number of elements to achieve the desired accuracy compared to traditional Finite Element Method (FEM). This study introduces a novel approach that combines FEM with Rigid Body Dynamics (RBD) to achieve both realistic and computationally efficient collapse simulations using fewer number of elements than AEM. The proposed technique leverages physics engines, commonly used in game development, to perform structural collapse simulations. An algorithm integrating FEM into the backend of a rigid-body simulation model has been developed. Failure criteria are employed to detect cracking, while a Depth-First Search (DFS) algorithm determines active and inactive elements during the simulation. Based on the DFS categorization, displacements are applied to active rigid bodies, whereas inactive ones are influenced by physics engine for collisions and movements under gravity. The developed algorithm is validated by comparing its results with theoretical solutions, experimental data, and simulations from the literature for structural components including beams, columns, and 2D and 3D frames. The results show that the proposed approach outperforms AEM in both accuracy and computational efficiency, even with fewer elements. Influence of the number of rigid bodies to discretize the structure on accuracy, failure pattern, debris extent and simulation time is also investigated. The findings suggest that a limited number of elements for discretization can achieve sufficiently realistic and computationally efficient results, making the approach particularly suitable for collapse simulations where reduced simulation time is critical.