Multi-objective design optimization using cascade evolutionary computations

The consideration of uncertainties in conjunction with the probability of violation of the constraints imposed by the design codes is examined in the framework of structural optimization. The optimum design achieved based on a deterministic formulation is compared, in terms of the optimum weight, the probability of violation of the constraints and the probability of failure, with the optimum designs achieved through a robust design formulation where the variance of the response is considered as an additional criterion. The stochastic finite element problem is solved using the Monte Carlo Simulation method, combined with the Latin Hypercube Sampling technique for improving its computational efficiency. A non-dominant cascade evolutionary algorithm-based methodology is adopted for the solution of the multi-objective optimization problem encountered, in order to obtain the global Pareto front curve.