Probabilistic Prediction of Failure in Columns of a Steel Structure Under Progressive Collapse Using Response Surface and Artificial Neural Network Methods

Much attention has recently been paid to the issue of progressive collapse, which is associated with the uncertainties that may affect the accurate assessment of the safety of the structures. Probabilistic analysis can be used to quantify the probabilistic safety of structures under extreme loadings. Since the columns play a key role in the stability of the structures subjected to the progressive collapse and they are very prone to failure, this research focuses on estimation of the failure probability in these structural elements. Monte Carlo simulation is used to perform the probabilistic analysis in a steel structure. The ratio of the axial force demand to the inelastic buckling capacity in columns adjacent to the damaged column is considered as the implicit limit state function. Artificial neural network and response surface methods are used to estimate an explicit function to save computational time. The results obtained from this study can be used to rehabilitate damaged structures using the effective role of each random variable on the structural responses which have been determined by the sensitivity analysis.