The development process of crash management systems (CMS) involves a lot of iterative work that often is based on experience and intuition from the engineer. Today that is the methodology used to design a system that fulfils all the demands and requirements required to have good crash performance. The development of the CMS involves dynamic, non-linear load cases and the goal is to design a system that has plastic deformation predictably.

In this master thesis, topology optimization was investigated as a method to improve the performance of the CMS and make the designing process less repetitive and instead make it more creative for the developer. If this is done right it would also be less time consuming and lead to new ideas and innovations. The process of this thesis was to first do a comparison of two different topology optimization softwares and then evaluate methods and efficiency for designing extruded aluminium sections through topology optimization.

The two softwares compared are the LS-Dyna-based software LS-TaSC and the other is Altair Inspire. The softwares are compared by doing a feature comparison and comparing how well both softwares integrate into Gestamp's simulation environment and the other softwares used in the organization. To learn the method and the optimization routine some models are created and optimized in both LS-TaSC and Inspire.

The topology optimization models created are based on a few different components of the CMS. The components tested to optimize are a patch for the cross-beam, the cross-beam itself, a reinforcement inside the beam and then some different crashboxes in both Inspire and LS-TaSC. The models use both solid and shell meshed design spaces and the optimization method changes depending on the element type. The shell meshed models reduce the element thickness until a given value where the element is deleted. In the models with solid elements, the optimization reduces the element density until it reached a certain threshold where the element gets deleted.

In the project different objectives, constrains, load cases and element lengths are tested. The models and results are compared when those parameters are changed and the method is trial and error based where the models are changed constantly in the desire to receive the best possible results. In the later half of the work a limitation is determined and that is to design a crashbox with topology optimization. The crashbox is made out of aluminium and extrusion is set as a manufacturing constrain.

The project resulted in a few different components optimized for the CMS. The first result was a shell meshed cross-beam optimized to see where the element thickness could be reduced. The next model was an optimized patch which resulted in a lighter component with equivalent performance. An inner reinforcement for the cross-beam was optimized with and without an extrusion constrain. The limitation to optimize a crashbox resulted in a crashbox optimized with several different models. It started with a simple model, with only one load case and it then moved on from there to more complex models with more load cases and different initial design spaces.

The project concludes that the use of topology optimization for dynamic crash simulations is possible but difficult. Since the softwares object functions are to maximize the stiffness and minimize the mass the models have to be manually controlled by different constrains and load cases to optimize for deformation. The softwares compared can both be used for different models, LS-TaSC for dynamic loading and Altair Inspire for reinforcements when linear-static loading can be used. A conclusion is also that the use of solid elements for thin-walled structures is problematic. The element lengths are controlled by the wall thickness which often is between 1-3 mm and therefore the elements can not be any larger than that. It is also possible to use shell elements, but there are some problems with that as well, one of them being the limit of possible design spaces.