Simulation based crashworthiness design
Finite element (FE) simulations have become a very efficient tool in crashworthiness design. By evaluating simulation results, an improved design can be obtained by changing a set of design parameters. This design approach may not always lead to the "best" design, since design objectives are often in conflict. A more systematic approach can be obtained by casting the objectives and constraints into a mathematical optimization problem.
Due to the nonlinear and transient dynamic nature of an impact problem, the solution of the equations of motion completely dominate the computational effort. Furthermore, the response obtained is usually sensitive to design changes due to buckling, wrinkling and other instabilities. In classical structural optimization methods, gradients are needed to approach the optimum. However, in the problems at hand, the gradient based methods usually fails due to the illconditioned numerical approximations of the gradients.
Response Surface Methodology (RSM) is a method for constructing smooth approximations to functions in a multi-dimensional design space. With RSM a design space is selected and approximating surfaces are constructed for the objective and constraint surfaces. Thus, local effects, e.g. caused by numerical noice, are smoothened and the global optimum may be approached. Kriging is another method for construction smooth surface approximations. It originates from geostatistics and it can be introduced as a local correction to the global approximation obtained with traditional response surfaces.
The present project aims at developing the RSM and kriging further in the application of crashworthiness design. Since both these methods in general are described as global methods, it may be possible to further improve the accuracy of the optimum by utilizing a local method in sequence after the global optimization. In order to evaluate the robustness of the design, we also aim at including stochasical properties of key parameters of geometry, material, initial- and boundary conditions etc.
One interesting application for these optimization procedures is the head to head impact between a car and a truck and to be competetive in the market and due to legalisation demands the need for an energy absorbing Frontal Underride Protection system (eaFUP) is to be developed. To construct this portion of the vehicle with mainly crashworthiness properties in concern is an excellent challenge.
Page responsible: Bo Torstenfelt
Last updated: 2008-02-07