Time efficiency maximization of progressive collapse analyses

Abstract

The study presented herein refers to investigation of ways to reduce investments necessary for performing progressive collapse analyses in terms of computational costs and time. The first part of this study is focused on implementation of reduced finite element (FE) modeling approach to simulate complex models featuring low computational power requirements. Good agreement with experimental results is obtained by reducing mesh sensitivity of FE model using the constant fracture energy approach. Mesh regularization on steel material is performed in analogous way by conserving total rupture energy. The use of single FE material model for both concrete and steel material applied to the whole structural frame produces good convergence with experimental results, while reducing modeling complexity and maintaining computational efficiency of the proposed approach. A parametric FE modeling program is proposed to reduce time and complexity of pre-processing stage of FE progressive collapse analysis. Object-oriented parametric modeling approach using Python programming language is demonstrated to be time-efficient in generating complex FE models with ease and flexibility. The parametric modeling software is then utilized to produce automated parametric optimization tool based on Covariance-Matrix-Adaptation Evolution Strategy (CMA-ES). The implemented algorithm is demonstrated to be applicable to objective functions defined through input-output relations involving the use of results of FE simulations. The applicability of the proposed automated parametric optimization approach is demonstrated through investigating case studies that involve large number of search space dimensions and computationally intensive evaluations of functions. Overall, good performance of automated conduction of parametric studies is observed.