In the field of accident investigation, a vehicle's delta-V is commonly used to quantify the severity of a collision. Delta-V is defined as the impulsive change in vehicle velocity caused by a collision. Often, the method used to obtain delta-Vs in real-world collisions is estimation based on post-collision vehicle crush measurements. However, these estimates have been known to result in large errors. The recent advent of Event Data Recorder (EDR) technology has provided engineers with measured delta-Vs from real-world collisions. These accurate measurements were the basis for several studies conducted in this thesis.

As EDRs are a new technology, little has been published regarding their accuracy. Therefore, the accuracy of new-generation (2000-2004) EDRs from General Motors, Ford, and Toyota in laboratory crash tests was examined across a wide spectrum of impact conditions. This study evaluated the performance of EDRs by comparison with the laboratory-grade accelerometers mounted onboard test vehicles subjected to staged crash tests. The study concluded that, if the EDR recorded the full crash pulse, the average error in frontal crash pulses was just under six percent when compared with crash test accelerometers.

An important aspect regarding the measurement of delta-V is the necessary recording length of the EDR. To determine the necessary recording duration for EDRs, a crash pulse duration analysis of real-world collisions was conducted. The dataset included Event Data Recorder data from 143 real-world collisions. It was found that an EDR crash pulse recording duration of 250 milliseconds would completely record 95 percent of all crash pulses.

Currently, the majority of delta-Vs in real-world collisions are estimated through energy-based methods. These methods use post-collision vehicle crush measurements to calculate the energy dissipated in a collision. Using EDR data, a thorough analysis of energy-based delta-V estimation was conducted. In 121 real-world collisions, energy-based delta-V estimates were on average 25 percent below EDR delta-V measurements. Further investigation into this discrepancy revealed two causes;​ categorized stiffness coefficients and the lack of a model to account for post-collision vehicle restitution. Fifty energy-based delta-Vs were re-calculated using vehicle specific stiffness and restitution coefficients, and the new estimates were on average the same as EDR measured delta-Vs.