Vehicle crash pulse optimisation

This investigation involved optimisation of a vehicle crash pulse. The objective of the optimisation was to minimise the harm suffered by the vehicle occupant during a full frontal crash. Two different restraint models were used in the analysis – a detailed occupant restraint model was analysed using MADYMO; and a simple analytical occupant restraint model was analysed using MATLAB. Common methodologies were developed to optimise two different crash pulses: 1. Acceleration-time crash pulses – optimised in order to minimise harm for a specific impact speed 2. Acceleration-displacement crash pulses – optimised in order to minimise harm weighted between different impact velocities. The acceleration-time crash pulse methodology was applied to both the MADYMO and MATLAB models. The acceleration-displacement methodology was only applied to the MADYMO model. For each model, three impact velocities were considered - 28, 48 and 56 km/hr. Each optimisation was performed for a fixed crush displacement. All of the optimisation methodologies were successfully executed. It was found that occupant harm (a function of injury criteria, expressed in terms of societal cost) was a minimum for crash pulses which produced square seatbelt loading curves (approximately constant loading with time). The results suggested that minimum occupant harm is achieved by targeting seatbelt loads alone. The acceleration-displacement crash pulse methodology was used to optimise the vehicle for multiple impact velocities of 28, 48 and 56 km/hr. This optimisation investigation significantly improved the occupant injury response for each of these impact velocities. Only acceleration-time crash pulses were optimised using the simple analytical model constructed in MATLAB. Crash pulses were optimised for minimum occupant peak acceleration, for the same range of velocities as the MADYMO model. For each velocity, minimum peak occupant accelerations corresponded to square occupant acceleration curves. It was found that there was remarkable correlation between the MADYMO and MATLAB restraint models, despite the differences in model complexity. The results from the MADYMO and MATLAB optimisation investigations were used to develop a general model of an optimised crash pulse. Based on the hypothesis that an optimised crash pulse was one which produces a square seatbelt loading curve (an assumption considered valid for restraint systems where the seatbelt remains the principle load path to the occupant), it was proposed that an optimised crash pulse has three phases – a fast loading phase, a load stabilisation phases, and a load control phase. Finally, using the analytical model, crash pulses were de-optimised in order to achieve maximum peak occupant acceleration. It was found that the rear-loaded crash pulse is the worst possible shape for a crash pulse, and will result in maximum occupant loads.