An improved mathematical model for the residual axial capacity of blast‐damaged RC columns based on optimized FE analysis

Abstract

This paper proposes an artificial neural network (ANN) based method to improve the computational efficiency for the residual axial capacity (RAC) of blast‐damaged columns, and subsequently develops a mathematical model for the RAC with enhanced applicability and accuracy. The critical parameters that can be optimized to reduce the analysis time of finite element (FE) simulations are identified. Based on the ANN‐based method, the effects of column height, mesh size, and column weight on the optimum critical parameters are investigated. Finally, the optimized FE method was employed to conduct a parameter analysis, and a mathematical model for the RAC was developed based on dimensional analysis. The failure mode and RAC of the columns obtained using the ANN‐based method are consistent with the blast test results, and the calculation time can be reduced by more than 90%. The proposed mathematical model is applicable to predict the RAC under various explosion distances. This research provides valuable insights for improving the computational efficiency of FE analysis and presents important information for predicting structural blast responses.