The objective of the work underlying this thesis was firstly to develop a neck for a new rear impact dummy, to evaluate the complete dummy and to specify test conditions for a consumer test with attention to AIS 1 neck injuries in rear impacts. In the development of the dummy neck, a mathematical neck model was developed and evaluated. Furthermore, impact severity and seat designs were also investigated.

Rear collisions can result in AIS 1 neck injuries. These injuries, which are becoming more frequent, occur mostly at low changes of velocity (less than 30km/h). Since AIS 1 neck injuries can result in long-term symptoms, it is of major importance to devise protection from these injuries. When testing the safety performance of seats and head restraints, an essential tool is the crash test dummy. However, the standard crash dummy of today, the Hybrid III, has had limitations in its interaction with the seat and head restraint.

The new dummy neck developed was evaluated by using data from crash tests involving volunteers as well as post mortem human subjects. For comparison, the Hybrid III frontal impact dummy was also tested under the same conditions. The new neck was found to have more human-like motion than that of the Hybrid III in low velocity rear tests when compared to both volunteers and post mortem human subjects. This was found to be the case for the head relative to upper torso horizontal and angular displacement. The new dummy neck became a fundamental part of the new, low-velocity rear impact crash dummy, the BioRID. The BioRID was found to have more human-like motion than that of the Hybrid III in low velocity rear impact tests when compared to both volunteers and post mortem human subjects. This result was observed for angular, vertical and horizontal displacement of the upper torso.

The variations in acceleration pulse characteristics in different vehicle models in identical impact conditions was shown to be substantial. A similar delta-V could be generated in a large variety of ways in terms of mean acceleration and acceleration pulse shape in a rear impact. The variation in crash pulse characteristics for the same car model from different real-world crashes of similar delta-Vs was also shown to be significant. This served as a background for the specifications of the test conditions for a proposed consumer test.

Real-world rear impact collisions with crash recorder-equipped vehicles, were reconstructed on a sled reproducing the real-world crash pulse. The results illustrate the risk of sub-optimisation when using only a single test in assessing neck injury protection. Further, five different seat configurations were evaluated in a series of sled tests at four impact severities. Identical vehicle seats were found to perform differently in tests with of different severities. Changing the mean acceleration (from 4.2g to 7.6g) influenced key dummy readings more than changing the delta-V (from 15km/h to 25km/h). Therefore, it should be expected that different real-world rear collisions at similar delta-Vs imply highly differing loading conditions to the occupants. As a consequence, the test conditions for the proposed consumer test program included specifications for several levels of change of velocity and mean acceleration.

The results of this thesis are expected to become important input in the definition of future rear impact test procedures for neck injury risk assessment.