CHARACTERIZATION AND ANALYSIS OF CONSTRUCTIONAL ERRORS OF LONG-SPAN PRESTRESSED STEEL STRUCTURES BASED ON IN-SITU MEASUREMENTS

Abstract

Long-span prestressed steel structures, known for its light weight and well performance, are world-wide used and developed nowadays. The constructional errors are challenges to long-span prestressed structures with considerations of the constructional precisions, prestressing degrees, and global stability respectively. The 3D laser scanning technique is applied in the structural health monitoring which is used in the long-span prestressed structures as well. However, a gap exists between measurement point clouds and structural assessments of long-span prestressed steel structures due to the complexity and volume of scanning data. The research targets at the real-time global stability assessments of long-span prestressed steel structures, including characterization of constructional errors from in-situ measurements, establishment of probabilistic model for constructional errors’ sensitivity study, and real-time constructional errors’ analysis. This work emphasizes current research progress on constructional errors’ characterization and data analysis from the in-situ measurements. The in-situ measurement data obtained from two projects of long-span prestressed steel structures charged by the researchers. The constructional errors are smartly recognized from the geometric deviations in comparison with nominal BIM models. The recognized data are then characterized from the convolutional neural network algorithm and statistically analyzed as well. The statistical data is used for the constructional-error sensitivity study where failure probabilities and collapse modes will be carefully evaluated. The research bridges the structural health information and structural stability assessments of long-span prestressed steel structures. In turn, it lays a solid foundation of real-time instant global stability assessments of long-span prestressed steel structures in a long term.