When vehicle collisions are reconstructed using a discrete kinetic time forward simulation program, many trials have to be performed to reach a point where the simulation results are close to the actual ones. The pre-impact speeds and directions of travel of each vehicle are the main variables that affect the post-impact motion of the vehicles. However, other factors, such as the exact resultant position where the impact forces are exchanged, the vehicle positions with respect to each other at impact, and the coefficient of restitution are important variables that also affect the results. When a number of impact parameters are unknown, a reconstruction can take a great deal of time.
An optimizer tool in PC-Crash is designed to minimize reconstruction time and error by automatically varying a selected number of impact parameters, comparing the resulting simulation for each combination of parameters with the actual incident. For each simulation, it calculates a weighted total error that is based on differences between the actual vehicle positions and angles and those obtained in the simulation. In each subsequent simulation, it changes values in an attempt to minimize the weighted total error. Hundreds of different combinations can be compared in a few minutes.
Twenty staged collisions were reconstructed with the optimizer tool in PC-Crash. These staged collisions had been previously reconstructed with a combination of manual linear momentum calculations combined with the trajectory model in an earlier version of PC-Crash. Differences between the reconstruction simulations and the actual collisions, and methods for minimizing reconstruction errors, are discussed.