Simulation of airbag sensing signals using finite element method

Using FE simulation to aid in development of the body structure for vehicle crashworthiness is a common practice in automotive industry today. However, development of airbag sensing calibration is still heavily depended on the physical tests. The airbag sensing algorithm is typically developed using a suite of signals from real crashes and immunity tests. The use of FE method to calculate the airbag sensing signals is an emerging technology with many potential advantages. Some recommendations are proposed in this study to reproduce the laboratory crash signals using FE simulations for obtaining usable FE data. In addition to the recommendations to the simulation, some recommendations to the airbag algorithm are proposed as well in order to have a FEM-compatible airbag sensing algorithm. For acceleration-based airbag sensing systems, the corresponding sensor accelerations are calculated using LS-DYNA. Two upcoming issues need to be resolved: - high frequency content of acceleration data (frequency range and accuracy) - aliasing problem when not exporting results to nodout file at every calculated time step. When the airbag algorithm is mainly based on low frequency accelerations or integral of the acceleration ( velocity), differences in the high frequency content between the simulations and the tests will not be a concern. The proposed method does use the global nodal information (th-node) from nodout file, instead of the usually used element “seatbelt_accelerometer” in LS-DYNA which outputs nodal information in a specified local coordinate system. In the proposed method the accelerometer information corresponding to its local coordinate system is obtained by post-processing the simulation results. A local default rotation of the sensor relative to the global coordinate system can also be calculated in the post-processing. With other post-processing algorithms, an optimized sensor orientation can also be determined without crashing several vehicles or running different models.