Automatic reverse modelling of box girder bridge from point cloud data

Abstract

Building Information Modelling (BIM) has over the last decades brought the physical- and digital worlds closer together. BIM implementation is now expected in construction projects, ensuring interdisciplinary cooperation from concept to demolition. In this thesis, a workflow for a reverse BIM modelling method specialized for as-built box girder bridges is presented. The method takes advantage of state-of-the-art software including both numerical and visual programming, parametric design, metrology tools, 3D laser scanning and BIM software, to create scripts, algorithms and digital data in order to digitally reproduce a physical bridge in its as-built condition. The results show that the automatically generated reverse BIM model accurately resembles the geometry of the studied box girder bridge. The created reverse modelling algorithms were found to struggle with accurately modelling quickly changing geometry and areas with much occlusion in the point cloud data. In areas with consistent geometry and complete point cloud data, the reverse BIM model is a highly accurate representation of the analysed bridge structure.