Center for Advanced Vehicular Systems

Spotlight on Technology - Simulation-Based Design Optimization

Simulation-based design optimization has been playing an increasingly important role in the design of advanced vehicular systems ranging from automobiles to spacecraft. However, direct application of optimization techniques to simulation-based design becomes prohibitively expensive when the simulations involve high-fidelity analysis of complex computer models (e.g., crash simulations, see Fig. 1). A commonly followed practice is to use metamodels as surrogates to approximate the results of these expensive high fidelity simulations. However, the existence of so many different metamodeling techniques makes it difficult for an engineer to know which model is the best for a specific response.

An active research program in the Center for Advanced Vehicular Systems at Mississippi State University is focused on developing novel techniques related to the use of surrogate models. Recently, a new technique based on optimum ensemble of metamodels was developed (Acar and Rais-Rohani, 2007). Using a weighted sum formulation of individual stand-alone metamodels, an optimization problem is set and solved to find the optimum value of the weight factor corresponding to each participating metamodel. As an example, in a full-frontal impact problem, different metamodels are developed for the acceleration at the driver seat. As indicated in Fig. 2, a weighted-sum of metamodels with optimum weight factors is found to outperform the best stand-alone metamodel by about 18%.

The simulation-based design optimization team at CAVS is actively working to develop design technologies that can help improve the fuel efficiency of automobiles through mathematical modeling and simulation of vehicle structures and optimization of critical response characteristics that are influenced by multiscale material models and various sources of uncertainty.