Simulation methods for bumper system development

Sammanfattning: n development of bumper systems for the automotive industry, iterative Finite Element (FE) simulations are normally used to find a bumper design that meets the requirements of crash performance. The crash performance of a bumper system is normally verified by results from standardized low speed crash tests based on common crash situations. Consequently, these crash load cases are also used in the FE simulations during the development process. However, lack of data for the car under development implies that simplified models must be used as a representation of the car in the FE simulations. Present simplified models of the car lead to uncertainties of the design even though the bumper system is modelled in a proper manner. The present work focuses on methods of how to represent the car in the FE crash simulations. The work is limited to the standardized crash tests in which the force acts longitudinally along the vehicle. Two different types of modelling perspectives are investigated. With the traditional approach, the aim is to obtain agreement of the results from the FE simulation and the physical test in terms of force from the barrier as a function of the compression of the bumper system. Here, the vehicle is represented by a point mass connected via rigid beam elements to the bumper system. The point mass, which only is allowed to translate longitudinally, is assigned with a reduced mass compared to the physical mass of the car to compensate for energy transformations in the car during the collision. In paper A, it is shown that the required mass reduction is dependent on vehicle and bumper characteristics as well as on the loading conditions. Also, the simple method of mass reduction leads to difficulties in attaining high agreement for time history of force and compression. In contrast to this, the idea with the second modelling technique is to reach a high agreement of the time history of force and compression of the bumper system. This methodology is based on a model structure that consists of mass elements, linear spring and viscous damper elements. It is shown that this model structure can provide high agreement between the FE simulation and the physical crash test in terms of force and compression as functions of time even for different loading conditions without adjusting the model parameters. Within the current thesis, a methodology of identifying parameters in the Mass Spring Damper (MSD) model from physical crash tests is presented. The methodology identifies a set of parameters that minimizes the deviation of the resulting displacements from the crash test and the simulation. This identification methodology is then used in a Design of Experiments (DOE) approach for relating model parameters in the MSD model to general properties of an arbitrary vehicle such as axial stiffness, bending stiffness and mass. For this, a public domain FE simulation model of a Ford Taurus is used. The knowledge gained from this study makes it possible to use the MSD model for representation of a coming car in the FE simulations associated with bumper development.