Probabilistic modelling of steel column response to far-field detonations

Due to the deficiency of current design guidelines for blast loadings on steel structures, this research develops probabilistic models for steel wide-flange columns under axial and far-field blast loading on both their weak and strong axes. A total of 160 finite element (FE) simulations were conducted using ANSYS LS-DYNA, with columns subjected to different Axial Load Ratios (ALRs) and blast impulses. Validation against two experimental tests showed a strong correlation in displacement plots, with a material model accounting for strain rate effects. Probabilistic models for predicting maximum displacement and residual axial capacity were formulated using Bayesian inference and posterior statistics. These models were developed by incorporating dimensionless physics-based explanatory functions. The slenderness ratio of the column was identified as the most influential. The models account for uncertainties such as material and geometric properties, as well as strain rate effects. Graphical plots between the ALR and Damage Index (DI) were examined to assess the column's damage level. Furthermore, the probability of failure (fragility) of four columns for similar blast impulse was assessed w.r.t DI. These models along with ALR vs DI plots will be useful tools to know the level of building occupancy and retrofitting options.