Car-To-Car Side Impacts: Development and Validation of Mathematical Models and their Usability for Protective System Design
Doctoral thesis, 1996
The objective of the present work was to develop and use mathematical models to establish principles for protective systems for occupants in car-to-car side impacts. The efforts in this work have concentrated on the protection of the thorax and abdomen since these body parts are most frequently injured in side impacts. The chosen approach was to develop general-purpose mathematical models since developing occupant protection systems by means of mechanical tests is very costly and time consuming. The models were subsequently used to evaluate a large number of design parameters.
The mathematical models developed and implemented were: 1) a model of the BIOSID dummy, and 2) a hybrid model consisting of occupant, vehicle and impacting barrier. The BIOSID model was used in sled simulations to evaluate the potential injury-reducing benefits of padding or airbags in side impacts. The BIOSID dummy model was validated by means of mechanical pendulum tests at two impact velocities. The sled model, including the BIOSID, was validated by means of mechanical sled tests at two impact velocities: with padding or with airbags mounted to the door.
The hybrid model was used in crash simulations to evaluate the effects of barrier type and vehicle modifications on injury response. Models of the EUROSID-1 and US-SID dummies and a model of the human body were used as occupants. Models of the EEVC foam and NHTSA honeycomb side-impact barriers were the impacting object. The EUROSID and US-SID models were validated by means of pendulum tests at one velocity, and the barrier models were validated by means of rigid-wall impact tests at one velocity. The baseline vehicle model including dummies and barriers were validated by means of crash tests according to the NHTSA and EEVC side impact test methods.
The sled model, including the BIOSID dummy model, was used to establish the occupant protection system that would result in the efficient protection of the occupant. The lowest TTI was obtained with the airbag with 0 kPa initial over-pressure and 1500 mm2 ventilation area, while the lowest chest deflection and chest VC were obtained with an airbag with 40 kPa initial over-pressure and 2000 mm2 ventilation area. Significantly reduced pelvis acceleration was obtained with 75 mm of compliant padding. The risk of the head impacting the side window was significantly reduced with an airbag mounted on the door. It was found that the airbag must be at least 120 mm thick when fully inflated in order to significantly reduce injury risks. It was also found that if the arm and shoulder of the occupant are not engaged by the intruding structures in the impact, significantly higher injury measures would be obtained than if they are engaged.
The hybrid model was used to establish the most violent barrier for the occupant. In addition, occupant responses to vehicle modification were evaluated. Generally the TTI predicted by the EUROSID were higher than those predicted by the US-SID. The chest deflection and chest VC predicted by the EUROSID were similar to those predicted by the human body. It was found that significantly reduced injury measures were obtained with the addition of thick padding at higher impact velocities. At lower impact velocities, padding produced no injury-reducing benefits. Impact with the EEVC barrier was more violent for the occupant than impact with the NHTSA barrier; thus, higher dummy readings were obtained with the EEVC barrier. It is shown that vehicle improvement to reduce injury risks in side impacts did not necessarily entail the stiffening or strengthening of the structure. Modifying the structure to minimize the dummy readings for one side-impact barrier may at other impact velocities and other impacting objects result in increased injury risks. The advantage of the mathematical modeling approach is that it provides a quick and economical means of developing principles for occupant protection systems.