Stochastic Computational Model for Progressive Collapse of Reinforced Concrete Buildings

Abstract

A two-scale numerical model is developed to investigate the probabilistic collapse behavior of reinforced concrete (RC) buildings subjected to local structural damage. In this model, a set of coarse-scale cohesive elements is used to model the failure of potential damage zones in various RC structural members. The constitutive properties of the cohesive elements and their probability distributions are determined from detailed stochastic finite element simulations of the potential damage zones by taking into account the uncertainties in various material properties. The two-scale model is validated both experimentally and numerically for different structural subassemblages. The model is then applied to study the collapse behavior of a prototype 10-story RC building subjected to sudden column removal using both deterministic and probabilistic analysis frameworks. The deterministic calculation uses the mean material properties and the factored gravity loads according to the Unified Facilities Criteria (UFC) guidelines. The stochastic calculation considers uncertainties in both gravity loads and material properties, from which the occurrence probabilities of different collapse extents are determined. The results of the present probabilistic analysis are discussed in comparison with the existing deterministic approach, which reveals the important role of probabilistic methods in the analysis of progressive collapse.