In rollover crashes there is a high risk for occupants to suffer severe injuries. The number of rollovers tends to increase at present presumably because of the increasing number of cars with a relatively high COG (Minivans, MPVs). Therefore there is a great potential for injury reduction in that area. Available dummies are designed and validated for front, side or rear impacts but not for complex events like rollovers. So the question comes up which Dummy should be used to assess safety systems for rollover accidents.

The aim of the study was to get a detailed information of the dummy behaviour compared to human behaviour in the first phase of a rollover accident.

Series of measurements with volunteers and dummies (Hybrid III and EuroSID) were carried out by using a current car seat mounted on a sled with additional tilting mechanism. Two types of motion were imposed to the sled that represented different rollover scenarios:​ a pure translational motion and a pure rotational motion. Two different acceleration levels from the range found in real world crashes were used. The kinematics of dummies as well as kinematics and muscle activity of volunteers were analysed.

The results show a significant difference between the kinematics of dummy and volunteer. In the rotational sled motion the volunteer movement was directed to the opposite side compared to the dummy. Thus, the dummies do not represent human occupants very well. Furthermore, the kinematics of both dummies is very similar, so no preference regarding the dummy type can be recommended. The EMG revealed activity of all observed muscles in all test configurations, the muscle activity influences evidently the movement of human occupants. This results are suposed to be useful for the development of rollover dummies and advanced numerical occupant models.