Prediction of residual strength of FRC columns under blast loading using the FEM method and regression approach

Abstract

This study presents a development of empirical equation which can be used for predicting the residual strength of fiber reinforced concrete (FRC) columns subjected to a TNT explosion. For that purpose, numerical investigations on the structural behavior and damage of FRC column under blast loading were carried out. Finite element (FE) modelling of FRC column and TNT explosive charges were built using a robust program, LS-DYNA. Those FE modellings were then verified using relevant, recent test data. A series of parametric analyses was conducted with such various main parameters as section, scaled distance, scaled section, fiber type, and fiber content to investigate the effect of these parameters on the structural behavior and damage of the column under explosion. Some interesting discussions were then presented to thoroughly understand the role of each parameter on the resistant capacity of the column. To develop the empirical equation for predicting the residual strength of the FRC column, Multivariate Linear Regression (MLR) and Multilayer Perceptron for Regression (MLP-R) were employed to generate the fitting models. While the MLP-R showed a superior capacity for estimating the result, MLR, on the other hand, could easily achieve the regression coefficient with fairly good estimation. As a result, a reliable empirical equation that can be used to predict the residual strength of the FRC column in practical design was proposed as an important outcome of this study.