Adaptive Metamodel-Based Robust Design Optimization of RC Frame Under Stochastic Blast Load

Abstract

This paper presents robust design optimization (RDO) of a reinforced concrete (RC) frame subjected to stochastic blast-induced ground motion (BIGM). As a full simulation approach requires extensive computational time to solve such problem, the polynomial response surface method (RSM) has been applied in the present study to alleviate the associated computational burden. The least squares method (LSM) generally adopted in the conventional RSM is often reported to be a source of error in optimization. Hence, a moving least squares method (MLSM)-based adaptive RSM is explored in the RDO. Random blast load is modelled by spectral representation method. The record-to-record variation of BIGM time history is captured by applying dual RSM. The optimization problem has been developed as a cost-minimization problem subjected to the displacement constraints. The RDO is formulated by simultaneously optimizing the expected value and the variation of the performance function by using weighted sum method. The robustness in the constraint is ensured by limiting the probability of failure of limit state function. The results show that the proposed MLSM-based RDO strategy yields more accurate solutions than the conventional LSM-based RSM taking full simulation approach as reference. At the same time, the present procedure yields Pareto-front in significantly lesser computational time than the full-simulation-based RDO approach.