This paper presents a two-scale numerical model for assessing the risk of progressive collapse of reinforced concrete (RC) frame structures. In this model, the potential damage zones in various structural components, such as beams, columns and joint panels, are represented by a set of coarse-scale cohesive elements. The constitutive behavior of the cohesive elements is formulated in the effective traction-separation space, where the general mixed-mode failure can be well represented. The model parameters for the cohesive elements with their corresponding probability distributions are calibrated from the fine-scale stochastic finite element simulations of the potential damage zones. With incremental Latin Hypercube sampling of both random cohesive properties and external loads, the model is used to investigate the probabilistic collapse behavior of a two-dimensional 30 story RC structural frame under different column removal scenarios, where the occurrence probabilities of various possible collapse extents are calculated. Finally, the present probabilistic approach is compared with the conventional deterministic analysis method.
Bibliographical noteFunding Information:
B. Xue acknowledges the financial support from the University of Minnesota through the 21st Century Fellowship. The authors thank Dr. Steven F. Wojtkiewicz for helpful discussions on the use of DAKOTA. The authors also acknowledge the Minnesota Supercomputing Institute for providing computing facilities for the numerical simulations.
© 2014 Elsevier Ltd.
- Cohesive element modeling
- Multi-scale modeling
- Stochastic simulation
- Structural collapse