In 2003, rollover accidents caused more than 10.000 fatalities and 229.000 injuries in the US alone. In view of this statistic and in order to provide a better occupant protection, the interest in the behavior of the vehicle structure and passive restraint systems under rollover loads is continuously growing.

In order to ensure a realistic reconstruction of the vehicle behavior in development tests, four new different test setups have been elaborated according to accident analysis results. For the restraint system development, knowledge about the borderline between roll and no roll is essential. To save expensive prototypes, this borderline is determined before performing first tests by using numerical simulations. The test and simulation tools support a comprehensive development process, which allow the adaptation and optimization of protection systems for rollover.

One key component of the restraint system is the algorithm, which has the task of rollover accident detection and determination of the optimal system activation time. For the latter task, knowledge about real occupant movement is essential. The low acceleration and rotation rates over a long period, which occur during some rollover constellations, lead to considerable movement deviations between the test dummy and the human. The firing time therefore, based on the dummy movements can only be determined approximately. Great optimization potential exists for activation algorithms which are adapted to humans. This adaptation is possible with a new developed simulation tool, which takes the possible muscle work of the human against occurring rollover loads into account. It determines the occupant movement during a rollover and has been validated to the human behavior by sled tests.