Parameter identification of GISSMO damage model for DOCOL 1200M : A study on crash simulation for high strength steel sheet components
Abstract: In the automotive industry there is a conflict between the need for weight reduction in order to reduce the CO2 emissions and the need for high safety. It has led to the use of high strength steel instead of the traditional lightweight materials. The increased use of high strength steels in combination with that the shortened development time in the automotive industry has led to the need of improved predictions of the actual crash behavior well since a full scale crash test is both expensive and time consuming.The damage model GISSMO is used in such crashworthiness simulations. In the present thesis the high strength steel grade DOCOL 1200M, GISSMO damage model has a number of parameters and curves that defines when necking and failure occurs, those have to be found. GISSMO is a phenomenological damage mechanics model which is based on experiments and does not consider voids and cracks thus it is only reliable to similar load cases as analyzed in the experiments. The different load cases are represented by the triaxiality which is the ratio between the mean stress and the von Mises stress.To find the parameters a number of test specimens were manufactured and tested in uniaxial tension then a FEM model was designed and the force displacement curve achieved from the simulation was mapped to match the experimentally achieved curve. The parameters were changed by the software LS-OPT® in order to increase the match. A metamodel-based optimization was run to find the curves and parameters with feedforward neural-networks and space filling point selection.The result shows that GISSMO has the potential to predict the failure behavior well, when the different specimens are optimized individually the match is good for all cases and when the simulation is examined the necking and localization of deformation is clearly seen. When the specimens are optimized together the match is not as good as the individual match.In order to improve the results more precise force displacement curves from the experiments would be of interest, for example load cells and optical measurments/strain gages could have been used to get the local displacement and the local stress and stress state. To make the model more reliable pure shear tests and compression tests could be used in future work. Running the optimizations for more iterations may also improve the result.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)