Determining the Actual Shape of a Bridge Pylon with Cylindrical and Conical Shapes Using Point Cloud Data and the RANSAC Shape Fitting Method

Abstract

Determining the shape and deformations of structures with complex geometries has become an increasingly common task in the construction industry. This article presents a case study on determining the shape and deformation of a slope and circular steel structure, a cigar-shaped cable-stayed bridge pylon with varying cylindrical and conical shapes during three key construction stages using terrestrial laser scanning (TLS) point cloud data and the Random Sample Consensus (RANSAC) algorithm for shape fitting. A methodology involving vertical reorientation, point cloud slicing, and RANSAC-based circle fitting was applied to cross-sections of the pylon. Optimal RANSAC parameters were determined to be 80 iterations and a 5 mm tolerance, achieving reliable circle fitting with root mean square (RMS) errors predominantly within 3 mm across all scans. After initial steel frame installation, geometric distortion analysis revealed minimal deviations within a 2 mm radius error, a slight increase after concrete pouring, and a partial recovery in the final stage, except for a localized anomaly at the pylon's top. Deformation analysis along the pylon's axis showed maximum horizontal displacements of 65 mm according to the static models in the first scan, reaching a maximum of 269 mm after pouring concrete, with a subsequent reduction and change in direction after cable tensioning. This research demonstrates the efficacy of TLS and RANSAC for evaluating structural deformation and emphasizes the importance of parameter optimization for accurate point cloud analysis in complex civil engineering structures.