Automated prediction of load-bearing capacity of existing spatial grid structures based on BIM and point cloud reconstruction

To enable automated prediction of the load-bearing capacity of large-span stadium spatial grid structures, thsi study proposes an algorithm integrating Building Information Modeling with 3D point cloud reconstruction. First, a 3D point cloud fitting algorithm for welded spherical joints is established using the least squares and RANSAC methods, while a combined RANSAC-PCA algorithm is developed for circular steel pipe members. Subsequently, taking the Hainan University Comprehensive Gymnasium as a case study, the BIM and finite element enable of its grid structure are reconstructed via Dynamo visual programming. Finally, the static load-bearing capacities of the designed model and the reconstructed models are compared. The results show that the vertical position deviation of the welded spherical joints ranges from (−70～+60) mm, and the diameter fitting deviation ranges from (−20～+10) mm, with a maximum deviation rate of 6.67%. The maximum fitting deviation rate of members is 21.19%, accounting for 0.51% of all members, while 78.85% of member diameters fall within a deviation range of (−6～+6) mm. Considering both joint position deviations and member bending deflection, the actual load-bearing capacity of the existing grid structure is 13.00% lower than the design value. Futhermore, the influence of joint deviation defects on static load-bearing capacity of the structure is much greater than that of member bending deformation. This findings provide critical references for the safety assessment and reinforcement design of existing spatial grid structures.