All studies undertaken to analyse phenomena occurring in lateral collision accidents are based on two types of experimental collisions-global collisions and "sub-system" collisions.

This latter type is designed to reproduce the component variables of global lateral collisions in a simplified form with the aim of evaluating how a given vehicle performs. Simplification usually means selection. For this reason the INRETS Laboratoire de Chocs et de Biomécanique (LCB) decided to simulate the severity of global lateral collisions as accurately as possible using a partial test which nevertheless considers all the parameters.

Attempts to simulate collisions involving 2 different vehicles require data from global reference tests and are obligatorily complex as they involve a large number of variables. This situation is made even more difficult by a certain number of experimental problems which were also investigated.

Even if the low number of tests implemented so far have not yet enabled the accurate reproduction ofthe severity of the global collisions used for refersnce purposes, they have enabled the evaluation of the impact on severity of various parameters which could be very useful in the context of the preparation of a definitive simplified sub-system experimental protocol.