On the Development of a Process Chain for Structural Optimization in Vehicle Passive Safety

Successful vehicles fulfil a tremendous number of multi-disciplinary requirements. Various attributes like safety, dynamics, vehicle rigidity, sound quality, vibrations, performance or fuel consumption, roominess, interior flexibility and styling have to be balanced to attract the customer. Reduced vehicle development cycles address the issue of changing customer demands. Increasing requirements focussing on fuel consumption and CO2 emissions on the one hand and increased vehicle quality and safety expectations on the other without increasing vehicle prices are actual challenges for car manufacturers. In order to find vehicle structures solving the mentioned conflicting requirements, smarter vehicle structures are needed. The thesis focuses on an approach to combine latest software and hardware tools to a process chain, as depicted in the Figure below, in order to find vehicle body structures fulfilling the various requirements as efficiently as possible. The ingredients of this process chain are the modules Model Build, Simulation and Evaluation forming the ”Structural Analyzer”, which can be linked to an optimization loop resulting in a ”Closed Loop” optimization, especially Genetic Algorithms (GA) are addressed. Furthermore a pre-defined set of inputs, as provided e.g. by a Design of Experiment, can be investigated. In the Model Built module, the software SFE CONCEPT is used to create a parametrical geometry of vehicle models, including spot welds. The aim is to derive Radioss finite element meshes. Beside this, panel thickness and material properties can be selected as a design variable to assess their contribution. An alternative SFE CONCEPT modeling approach is explained supporting the interactive vehicle development process and the parameterizing effort when using optimization tools. The Simulation Processing module was created to ideally utilize the available hardware environment. A process implementation is explained, separating the simulation process from the optimization loop in order get to a generic tool for parallel simulation of a large number of simulation decks. Different hardware configurations have been bench-marked. The Evaluation & Data Judgement module describes how the simulation results are reduced to the information separating a good from a bad design state. A large set of evaluation operations are implemented, similar to those used during interactive work (e.g. displacements, intrusions, plastic strains, velocities, ...). The extracted values are rated using a ”fitness function”.