During recent years the accident simulation program PC-CRASH was developed, which allows simulating the vehicles movement before, during and after the impact. As shown in several publications, the software allows the calculation of the 3D movement of all involved vehicles. When vehicle to vehicle collisions are analyzed using a discrete kinetic time forward simulation, several simulation runs have to be performed, to find a solution, where post impact trajectories and rest positions correspond with the real accident.

This paper describes in detail a method to vary the preimpact parameters automatically and to evaluate the simulation results This optimization tool allows one to vary pre crash conditions automatically and to determine the solution, which fulfills predefined target functions best. This method has proven to determine a set of initial conditions, which fulfill the predefined target functions, within a reasonable amount of time.

As in most accident reconstruction cases, several parameters, like ground friction or inter-vehicle friction are not well known, as well as many vehicle or driver related parameters, a method was developed, which allows to vary multiple parameters based on a Monte Carlo method. The error within the fulfillment of the target function can thus easily be used as a quality indicator for the selected initial conditions.

In this way it is possible to determine how stable the solution is and what amount of variation within the input parameters must be expected.
Two algorithm s are presented, both based on the Monte Carlo approach. The first one serves as an alternative for the optimizer tool and is included in the current version of PC-Crash. It gives reasonable insight in the variation of certain parameters in reasonable calculation time. The second one was developed by the Netherlands Forensic Institute as a systematic approach to obtain a-posteriori distributions for the pre crash velocities. Both algorithms are illustrated by examples. In both algorithms, the Quality Function or Trajectory Error plays an important role, and will be discussed first.