Evolutionary Structural Optimization with Multiple Performance Constraints by Large Admissible Perturbations.

Abstract

A LargE Admissible Perturbation (LEAP) with Evolutionary Structural Optimization methodology is developed. The LEAP methodology uses an incremental predictor-corrector scheme, which makes it possible to solve the redesign problem using data only from the finite element analysis of the baseline structure for changes on the order of 100% in performance and redesign variables without trial and error or repetitive finite element analyses. A structural topology evolution algorithm is introduced using a Cumulative Energy Elimination Rate (CEER) scheme by removing low energy elements at each iteration, while using the elastic modulus in each element as redesign variable in the LEAP methodology. Benchmark examples are used to demonstrate that static displacement, modal dynamic constraints, and simultaneous static and dynamic constraints can be achieved. Convergence is achieved in 3 to 7 iterations with two FEA’s per iteration inside the ESO/LEAP algorithm. Results of numerical applications satisfy engineering intuition and show the effect of multiple objectives on topology evolution.