Toward automatic generation of 3D steel structures for building information modelling

Building information models (BIMs) are becoming standard for new construction. Extending this trend to existing structures is complicated because of an absence of reliable documentation and the cost of generating it anew. To overcome this problem, this paper proposes a method to identify automatically structural steel members from a terrestrial laser scan point cloud and to generate that geometry in a BIM compatible format. The proper shape and dimensions of the cross-section are established by employing kernel density estimation. A method associated with measured metrics is introduced to determine the best match of various cross-sections, from a prepopulated library. The proposed method successfully identified up to 92.0% of the required cross-sections and 81.3% of structural members across two steel frames of different shapes, sizes, and configurations. [Display omitted] •Identify automatically structural steel members from laser scanning data points•Use non-parametric regression to detect the primary surfaces of a cross-section•Determine automatically shape and dimensions of the cross-section-based point cloud•Propose a method mapping selected sections onto the data point of the cross-section•Introduce a method to determine the best match cross-section of a cross-section