Multi-Disciplinary Optimization of a Sedan Using Size and Shape Parameterization

In vehicle development activity there are different disciplines with their own set of requirements that need to satisfied in order to get a successful product. Traditionally, the simulations were performed for different disciplines but at the time of optimization, Single Discipline Optimization (SDO) was usually performed and later confirmatory runs were done for the other disciplines. This process resulted in a time consuming loop of running several iterations with different disciplines and engaged people from these disciplines to make the optimized design meet the performance targets. Quite often these approaches led to tradeoffs made for design requirements or significant deviations from the optimized design obtained by running SDO. Multidisciplinary Optimization (MDO) addresses this shortcoming and takes into account different disciplines for optimization, thus reducing the need to iteratively evolve the design for different disciplines. Though this process has its benefits but those benefits are overridden with the setup time required for MDO. The major proportion of the setup time is consumed in defining shape variables on full vehicle FE model that has all types of connections as the process needs to be repeated for different disciplines. As a result an alternate methodology is sometimes pursued to create concept models from the full vehicle FE models thereby reducing the complexities of a full vehicle FE model. This approach not introduces approximation in the entire process by idealizing the detailed FE model but also requires this definition of shape variables separately for each discipline. In this paper, the MDO was performed on a full vehicle by considering Crash and NVH load cases. These two disciplines were considered to provide a process and to demonstrate its benefits. LS-DYNA® was used for crash simulations and Nastran was the solver used for NVH load cases. This process of MDO has been successfully applied to different vehicle programs including disciplines like durability, vehicle dynamics, occupant simulations, CFD etc. Due to IPR restrictions, the program specific work is not shared in this paper. The present process demonstrated on a sedan with different types of design variables defined for Crash and NVH load cases. The MDO is performed with the objective to reduce mass of the vehicle without any significant performance degradation. The approach presented here uses advance features of DEP’s Meshworks and provides the processes through which MDO could be carried out without sacrificing the complexity of the full vehicle FE model. It expedites the entire process by offering faster turnaround time.