This research investigates the uncertainty in the calculation of the change in velocity, ΔV, and the crush energy, EC, due to variations in the computed values of crush stiffness coefficients, A and B (d₀ and d₁), and due to variations in the measurements of the residual crush, Ci, i = 1,...6, using the CRASH3 damage algorithm. An understanding of the nature of such uncertainties is of particular importance as both the ΔV and EC are frequently used as inputs to reconstruction methods and become variations in the reconstruction process. These variations lead to uncertainties in the results of the reconstruction which are generally the preimpact speed of one or both of the vehicles involved in the collision. This paper consists of three parts. The first investigates the uncertainty associated with the calculation of the stiffness coefficients A and B (d₀ and d₁). The second part looks at the uncertainty of the CRASH3 process of calculating the velocity change, ΔV, of both vehicles and the energy loss, EC, of the system. The third part examines the effect of such variations on the reconstruction of vehicle speeds in frontal collisions.
The magnitudes of the uncertainty of the manual measurements of residual crush values Ci, i = 1,...6, are determined from results presented in the literature. The uncertainty of the stiffness coefficients, A and B (d₀ and d₁), were determined by accessing six different light vehicle models for which experimental data from multiple NHTSA frontal barrier tests was available. The stiffness coefficients were computed for each set of test data for each vehicle model. These values, with a minimum of six frontal barrier tests per vehicle, give an average and standard deviation for each coefficient for each vehicle model. These values then provide the variation of the stiffness coefficients for these vehicle models. These variations then are used in the CRASH3 damage algorithm, through the use of Monte Carlo analysis, to calculate the uncertainty in the ΔV and EC for each of the six vehicle models colliding with a fixed, rigid barrier. Comparison of the results of the analyses shows that the uncertainty in ΔV and EC due to the experimentally determined stiffness coefficients is considerably more significant than the uncertainty due to the variation in the measurements of the residual crush.
The paper then examines the use of these results in the reconstruction of two in-line collisions. Here the uncertainties in ΔV and EC become variations in the reconstruction process. The results of this section present a range of values for the preimpact speed of one of the crash vehicles based on the range of values for the crush energy loss determined in the first part of the paper.