Robust design optimizationof vehicle structures for crashworthiness under frontal impact

Abstract

Crash simulation of vehicles evaluates vehicle crashworthiness for passenger safety. Although crash simulation analysis still requires large computational cost, tremendous increases in computer speed and development of numerical methods for computer aided vehicle design enabled design optimization techniques to be applied in practical engineering problems. As a result, deterministic design optimization has been recently applied in various crashworthiness designs in order to improve safety and reduce cost. However, conventional design optimization may give insufficient and less meaningful results due to various uncertainties of design variables or material properties. In fact, when these uncertainties are considered in design optimization, they can lead significant different results from those of deterministic design optimization. To overcome this drawback, robust design optimization technique is required to obtain a design with better performances while improving the robustness of optimum design. In this research, a robust design optimization of vehicle structures for crashworthiness under frontal impact is performed. A number of vehicle performances such as weight reduction, energy absorbed during collision, structural deformation and safety criteria are considered in crash simulation. In this work, reduction of structural weight is selected as design objective, occupant load criterion and occupant compartment intrusion as constraints. However, the variances of such vehicle performances caused by input uncertainties, are also considered in the design formulation to increase the robustness of the results. Because of the large computational cost of crash simulation, design of computational experiments and surrogate model techniques are adopted to rapidly calculate the responses of the simulation. Surrogate model is a numerical function in the design domain which is constructed based on simulation results from design of experiments. Among a lot of surrogate models, kriging surrogate model is employed to calculate the responses of the simulation since it is widely used due to its good performance to predict nonlinear responses. Finally, the results of conventional design optimization and robust design optimization of the vehicle structure are compared and discussed in detail. It is concluded that a robust design considering uncertainties should be employed to improve the quality of vehicle design.