Accident Reconstruction/Simulation With Event Recorders
Kristin Bolte, Lawrence Jackson, Vernon Roberts, Sarah McComb
National Transportation Safety Board

Introduction

In 1997, 42,000 people were killed in highway accidents in the United States. The exact cause of an accident is often unknown and, therefore, conclusions relating to the safety afforded by the vehicle to the occupant cannot be made. In addition, safety hazards in the highway environment are often not discovered due to the lack of information. Accident reconstruction is a tool commonly used by the National Transportation Safety Board (NTSB) to investigate the accident sequence, but data are often lacking and accurate reconstructions are difficult and time-consuming. Because many assumptions are made in this process, the reconstruction is not exact, making it difficult to accurately predict occupant kinematics and to identify potential safety hazards within the vehicles. Vehicle recorders would eliminate much of the guesswork involved in reconstructing accidents, enabling a more accurate assessment of occupant injuries, driver performance, and safety hazards within and around a vehicle.

Vehicular Data

Much information concerning the accident vehicles is needed to reconstruct an accident accurately. This information includes both driver inputs and vehicle outputs. The driver inputs would include steering angles; application of brakes and throttle; gear selection; engine braking; and use of lights, turn signals, cruise control, wipers, and horn at various increments prior to and during the accident sequence. Information is also necessary on driver restraint use and occupant seating location. The vehicle output parameters that are needed include the pre-impact speed, engine rpm and gear selection, acceleration history, braking efficiency, the use of cruise control and anti-lock braking systems, and activation of passive restraints.

Typically in highway accidents, the driver inputs and the vehicle outputs are derived from witness statements, or physical evidence such as the impact location, tire marks, and the final resting position of the vehicles. Unfortunately, witness statements are often in conflict, biased, and perhaps based on past experience, adding confusion to variables such as initial speed or speed at impact. Further, the accident typically occurs within 0.10 seconds, a speed at which witnesses cannot always accurately comprehend vehicle interactions. Furthermore, physical evidence can be limited if road conditions are not ideal or if evidence is not collected immediately after the accident.

Methods for Accident Reconstruction and/or Simulation

Several programs have been developed to aid in accident reconstruction/simulation. These developers include McHenry Software (m-smac,m-hvosm, m-crash, ATB/cvs); Engineering Dynamics, Inc. (EDCRASH (reconstruction), EDSMAC, EDVDS, EDVSM, EDHIS, EDVTS, EDGEN); MacInnis Engineering (PC-Crash); Mechanical Simulation Corporation (TruckSim, AutoSim); AR Software (Slam, WinSMAC); Accident Reconstruction Software (REC-TEC); Fonda Engineering Associates (CRASHEX); and others. These programs use the driver inputs and the vehicle outputs to reconstruct/simulate the accident. Crash pulses detailing the acceleration-time history of a vehicle can be calculated from these inputs, the damage profiles of the vehicles, and the vehicle stiffness, as shown in Figure 1.

Figure 1: The crash pulse on a school bus from a simulated accident between a tractor-trailer and the bus.

Human Performance: Recorded information could help investigators determine the drivers' actions before and during the accident. These inputs will help to refine the vehicle and occupant dynamics during the accident. The recorded information may also help to assess if the driver was fatigued, impaired, or not making evasive maneuvers. For example, a lack of steering reversals may indicate fatigue. Such information will help to assess countermeasures for preventing accidents.

Biomechanics: Once accurate information on vehicle performance and driver response is known, the reconstruction may then be used to assess occupant kinematics (Figure 2) and occupant protection within the vehicle. The more accurate the vehicle dynamics, the more accurate the occupant simulation. Better occupant simulation will result in more effective design changes that improve occupant safety.

Figure 2: Occupant kinematics during a simulated accident between a tractor-trailer and a school bus

Accident Reconstruction Examples

Collision of a Van and a Train: Event recorders may be used to facilitate the accident reconstruction process. For example, in 1998 in Wagner, Oklahoma, a van was struck in the rear by a train. Witnesses in the van reported that the driver stopped at the railroad crossing and then proceeded into the crossing. The train engineer reported that the van did not stop at the crossing. An event recorder was present on the train, so the train speed at impact was known to be 46 mph. Based on the damage profile on the van and the final resting location of the van, reconstructionists at the NTSB varied the speed of the van at the point of impact to determine the most accurate trajectory and the related initial speed of the van. (Figure 3)

Figure 3: The simulation of the train and van in Wagner, Oklahoma.

The conclusion drawn, based on the reconstruction, was that the van had to be traveling about 35 mph at impact to sustain the damage and reach the final resting location. The train's event recorder enabled this iterative process to occur. Without the recorder, there would have been an infinite number of train and van speed combinations that would have resulted in a similar damage profile and rest location for the van. Furthermore, the cause of the accident would have been unknown and potential safety hazards might not be identified.

Collision of a Tractor Semi-Trailer and a School Bus: Another simulation being developed by the NTSB involves a school bus that was struck from behind by a tractor semi-trailer in Holyoke, Colorado, in 1998. Vehicle recorders would have greatly helped the simulation and reduced the number of simulation runs required. A bus recorder would have indicated if the bus was stopped or had just started to accelerate when struck by the truck. A truck recorder would have provided information on the truck's initial speed, gear in use and rpm, when the truck driver started to brake, how hard he braked, and which brakes (semi-trailer only or full truck) he initially used. The driver added steering input just before the crash; knowing when he started to steer and the rate at which he steered would have helped in the simulation. Furthermore, both bus and truck driver inputs prior to the crash would help determine a potential cause of the accident.

Summary

Accident reconstruction/simulation is a difficult and time-consuming process. Event recorders not only simplify the reconstruction process, but they also increase the accuracy of the reconstruction resulting in more detailed conclusions concerning safety.