Improved Vehicle Crashworthiness via Shape Optimization

Advancements in high-performance computing and in nonlinear dynamic structural simulation have made it possible to virtually test potential designs prior to building and testing expensive prototypes. These tools alone, however, still require an engineer to develop a design based on intuition and numerous time-consuming and error-prone iterations. The next level of advancement is software to automate the process of iterating over a large number of design scenarios and intelligently seek optimal values for those parameters that strongly affect product performance and cost. While many design optimization approaches are limited to a small number of continuous design variables, the approach described here leads to a productive search over hundreds of variables at a time. This capability has been implemented in a software product called HEEDS (Hierarchical Evolutionary Engineering Design System). HEEDS uses multiple autonomous agents to hierarchically decompose a problem into subsets with highly decomposed overlapped relationships. Decomposition is effected by using different numbers of design variables, different levels of design variable discretization, and/or other problem-specific divideand-conquer rules. The system combines evolutionary search algorithms with local optimization 1 Copyright © 2003 by ASME techniques. Using ABAQUS/Explicit and LS-DYNA as the finite element solver within the HEEDS optimization environment, this process has been applied to several automotive rail designs, resulting in significant gains in performance along with up to 20% reductions in mass compared to baseline rails designed by experienced engineers. Two example applications of this method are described herein.