This article describes the reconstruction of an intersection collision involving a motorcycle striking a sport utility vehicle (SUV). Event Data from the SUV is incorporated into the reconstruction.
The collision involved a motorcyclist riding a 2007 Harley-Davidson (H-D) FLSTFI Fat Boy eastbound through the subject intersection. The driver of a westbound 2010 Ford Edge attempted to turn left to go southbound and the motorcycle struck the passenger’s side rear of the Ford. The driver of the Ford stated that she did not see the motorcycle prior to initiating her turn on a yellow traffic signal. A witness who observed the motorcycle prior to the collision stated that the motorcyclist accelerated into the intersection to make it through on the yellow light. The speed limit for east-west traffic through the intersection was 60 mph. An investigating officer noted that sun glare could have been a factor contributing to the Ford driver not seeing the motorcyclist.
The first photograph below depicts the H-D motorcycle and the Ford Edge at rest in the intersection following the collision. The second photograph depicts the Ford Edge at rest and shows the damage to the vehicle from the collision. The motorcycle struck and damaged the passenger’s side rear of the Ford, directly contacting the rear wheel of the Ford behind the axle. This caused the wheel to deform and rotate clockwise relative to the vehicle and the tire to go flat. The collision caused significant clockwise yaw rotation of the Ford, on the order of 180 degrees, and the Ford deposited prominent tire marks between impact and rest. The second photograph below also shows that the front tires of the Ford are steered to the left following the collision.
Mapping the Site and Struck Vehicle
The intersection was inspected and its geometry mapped with a Faro Focus3D X 330 scanner. Some of the resulting scan data is depicted in the first image below. This scan data was used to generate the terrain for the simulation described later. The Ford Edge was also inspected and its post-collision geometry documented with the same scanner. The resulting scan data for this vehicle is also depicted below. The clockwise rotation of the passenger’s side rear wheel due to deformation was measured from this scan data at 17 degrees relative to the vehicle. In the photographs of the Ford at rest in the intersection, the leftward steer angle of the right front wheel appears to be of greater magnitude than the 17 degrees the right rear wheel was angled.
The EDR Data
The Bosch Crash Data Retrieval system was used to image data from both the restraint control module (RCM) and the powertrain control module (PCM) on the Ford. Data from one event was recovered from the RCM – a side deployment event. The data indicated that the driver pretensioner and both the driver and passenger curtain airbags deployed 28 milliseconds after the system began sensing the collision. The data also indicated that the driver, who was the sole occupant of the vehicle, was buckled at the time of the collision. The RCM data included pre-crash data at 5 indicated time intervals (-4 sec, -3 sec, -2 sec, -1 sec, and 0 sec). This data is included below. Finally, the RCM data reported a lateral change in velocity of 2.49 mph. However, the reporting window for the lateral change in velocity was only 50 milliseconds. This recording window would be too short to capture the entire collision between the Ford and the motorcycle and certainly too short to capture the collision between the rider and the Ford. The graphical data for this lateral change in velocity is included below. As this figure shows, the horizontal axis of the graph starts at minus 49 milliseconds and goes to 0 milliseconds. The first non-zero acceleration reading is at -28 milliseconds. Given that the system reports that the curtain airbags deployed 28 milliseconds after the collision began being sensed, it seems likely that, in this case, the EDR only reported the timeframe from the start of the collision until the deployment of these airbags.
Data was also recovered from the PCM on the Ford. The graphical data, which is incorporated below, included the vehicle indicated speed, the accelerator pedal percentage, brake switch status, anti-lock brake status, and engine rpms for 20 seconds prior to and 5 seconds after time zero. Tabular data was also reported for each of the variables included in the graph. In addition, the tabular data added transmission status (reverse or not reverse), speed control status (on or off), traction control status (active or inactive), and stability control status (active or inactive). According to the data limitations in the CDR report, time zero corresponds to the time at which the PCM received a restraint deployment signal from the RCM. However, “The Restraint Deployment Signal (RDS) may not be recorded on the PCM immediately at impact. Time lags within the system may result in the Restraint Deployment Signal being recorded a few data samples after impact has occurred.” The Ford was equipped with Ford’s version of electronic stability control, AdvanceTrac with Roll Stability Control (RSC), and the tabular data from the PCM indicates that it was activated from +0.2 through +1.4 seconds. After that, it was inactive. The AdvanceTrac system utilizes the ABS system and the data also reports the ABS system was active during the same timeframe.
Arndt et al.  reported a study of the accuracy of the speeds reported by the Ford PCM on a 2005 Ford Explorer during high slip angle maneuvers and during acceleration and braking. This test vehicle was equipped with Ford’s electronic stability control system (ESC), AdvanceTrac with Roll Stability Control, just like the Ford Edge involved in the subject collision. Test were conducted with and without the ESC active. Arndt noted that “in testing with ESC enabled, speed error was associated with ESC intervention…A Ford PCM download which provides recorded speed every 0.2 seconds appears to provide enough data that an accurate speed trend can be discriminated.” For the subject collision, ESC intervention did occurred, but not until after the collision. The tabular data identifying this ESC-intervention is included below. The second to last column states that the ESC was intervening from +0.2 to +1.4 seconds. Based on Arndt’s results, there would likely be intervention-related error in the reported speeds during this timeframe. In addition to that error source, the significant yaw rate of the Ford following the collision would also be expected to result in errors in the reported speeds because of the discrepancy between the heading and velocity directions of the vehicle.
That said, the speeds reported by the PCM prior to the collision are likely to be accurate. These speeds were recorded when the vehicle would have a low slip angle and when there was no ESC-intervention. Ruth et al.  reported testing to evaluate the accuracy of PCM-reported speeds under steady state conditions. They compared PCM-reported speeds to speeds measured with a 100-Hz VBOX and with a speed trap and found a maximum difference of 0.61 kph (0.38 mph).
The tabular data indicated that the speed of the Ford at time zero (the time of the restraint deployment signal) was 35 kph (21.7 mph). This data also indicated that the vehicle was accelerating in the seconds leading up to the collision. The data from the RCM, on the other hand, reported a speed at time zero of 27.9 kph (17.3 mph). This is 4.4 mph lower than the speed reported by the PCM for time 0. The RCM and the PCM are obtaining their speeds from the same sensor – a speed sensor on the transmission output shaft. However, they are sampling from this sensor at different rates. The PCM obtains a speed reading every 0.2 seconds, whereas the RCM obtains a reading every 1 second. The speed reported by the RCM for time 0 would simply be the last speed reading obtained by the RCM prior to the collision. Thus, the RCM did not capture the actual speed at the time of the collision, rather it captured a speed sometime in the 1-second time window preceding the collision. Looking at the tabular data from the PCM, the 17.3 mph speed reported by the RCM would have been measured sometime between -0.6 and -0.4 seconds in the PCM data. For the analysis reported here, the collision speed was estimated from the PCM data.
Mapping the Evidence
Camera matching photogrammetry was utilized to locate the vehicle rest positions, tire marks from the Ford, and scrapes and gouges on the asphalt from the motorcycle. Camera matching involves reconstructing the location and characteristics of the camera that took the photograph being analyzed. Once the camera location and characteristics are obtained, physical evidence within the photograph can be located. This technique involves the following steps:
(1) A photograph is selected for analysis.
(2) The scene geometry depicted in the photograph is mapped.
(3) The mapping data is imported into a computer modeling software package and viewed using a virtual camera with a vantage point similar to that shown in the photograph.
(4) Lens distortion is removed from the photograph.
(5) The corrected photograph is then imported into the modeling software and is designated as a background image for the virtual camera.
(6) Adjustments are then made to the location, focal length, and viewing plane of the virtual camera until an overlay is achieved between the mapping data and the scene geometry shown in the photograph. Once a match is obtained, then the camera has been reconstructed.
(7) Once the camera location and characteristics have been obtained, the evidence visible in the photograph can be located.
This process was carried out for several police photographs that depicted evidence related to the subject collision. A sample of these photogrammetry results are included in the following figures. The first image below is one of the police photographs that depicts the tire marks deposited by the Ford following the collision. The second image below shows the scan data from the accident site overlaid on this photograph, indicating that the camera location and characteristics have been reconstructed. The third image below shows the tire marks located and traced with dark blue outlines. A set of scrapes and gouges on the asphalt are also shown in this image (with light blue lines) that were traced from another police photograph that was also analyzed.
Simulating the Collision
One indicator of a motorcycle’s impact speed with a passenger vehicle is the magnitude of the translation and rotation experienced by the struck vehicle following the impact. PC-Crash simulation software was used to simulate the subject collision, specifically to determine the motorcycle impact speed necessary to cause the documented post-collision rotation of the Ford. In setting up this simulation, manufacturer specifications were obtained for each of the vehicles to obtain the geometric dimensions and weights. The Ford Edge was determined to be a Limited edition equipped with a 3.5 liter, V6 gasoline engine, an automatic transmission, all-wheel drive, anti-lock brakes, and AdvanceTrac with Roll Stability Control (RSC). Based on the manufacturer specifications for this edition of the Ford Edge, the weight at the time of the collision was estimated at 4,385 pounds, including the weight of the driver. The Harley-Davidson motorcycle was determined to be an FLSTFI Fat Boy with a 1584cc V-Twin engine, a manual transmission, and a conventional braking system. The weight of the motorcycle at the time of the collision was approximately 714 pounds and the weight of the rider was approximately 170 pounds. The rider’s weight was included in the motorcycle’s weight within the simulation.
In the simulation of this collision, the roadway coefficient of friction was set at 0.76 and the TM-Easy tire model was used with default values for both vehicles. The integration time step was set at 5 milliseconds. Because the motorcycle struck the passenger’s side rear wheel of the Ford, the coefficient of restitution was set at 0.25, consistent with coefficients of restitution for wheel impacts I have reported elsewhere. Based on the EDR data, the impact speed of the Ford was initially set at 21.7 mph. Sequences in PC-Crash were used to steer the passenger’s side rear wheel of the Ford 17 degrees to the right immediately following the collision. The simulation was optimized using the impact speed of the H-D motorcycle, the precise location of the collision on the roadway, the intervehicular friction, the impact center height, the steering angles of the front wheels, and the brake factors for each wheel. Small changes in the impact speed of the Ford were also utilized for the final step in the optimization. Steering inputs of 23 degrees at the left front wheel and 19 degrees at the right front wheel were utilized, developing over 2.3 seconds. The Ford was an all-wheel drive vehicle, and brake factors of 5% were used for the front wheels and the left rear wheel. A brake factor of 100% was used for the wheel that was struck and deformed by the motorcycle – the right rear wheel. The exception to these brake factors was during the interval of ESC-intervention, during which a 100% brake factor was assigned to the left front wheel, and when the system indicated that the driver applied the brakes near the rest position. Brake factors were not applied to the wheels of the motorcycle because it fell on its side shortly after the collision. The coefficient of friction between the body of the motorcycle and the ground was set ultimately set at 0.4, though this parameter was also iterated in optimizing the simulation.
The video below shows the optimized PC-Crash simulation of this collision. A high-quality match was obtained with the rest positions for both vehicles and the Ford’s post-collision motion matched very well with the documented tire marks.
This analysis with PC-Crash led to the conclusion that the motorcycle was traveling 69 mph at the time of the collision, 9 mph greater than the speed limit and consistent with the statement by a witness that the motorcyclist accelerated into the intersection in response to the yellow traffic signal.
During the simulation process, I quickly realized that I could match the overall rotation of the Ford with a wide range of motorcycle impact speeds. However, the rate of rotation, as represented by the tire marks that were deposited, could only be matched with a narrow band of motorcycle impact speeds.
In this case, a high-quality match was also obtained with the motorcycle's rest position. This would not be expected in every case, but often matching at least the post-impact velocity direction of the motorcycle will help with the simulation optimization and give the analyst greater confidence in the results.
The change in velocity calculated for the Ford in this simulation was 10.7 mph with a principal direction of force of 81 degrees. Thus, the total lateral velocity change for the Ford was approximately 10.6 mph, 8.1 mph more than what was reported by the RCM. Thus, the EDR data gave me significant information about the the Ford in this accident - the vehicle's pre-collision speed, the acceleration rate leading up to the collision, and the fact that the ESC intervened during the post-impact rotation, for instance. However, the change in velocity reported by the system could not be taken at face value without further analysis.
Arndt, M.W., Rosenfield, M., Stevens, D., Arndt, S., “Test Results: Ford PCM Downloads Compared to Instrumented Vehicle Response in High Slip Angle Turning and Other Dynamic Maneuvers,” SAE Technical Paper 2009-01-0882, 2009, doi:10.4271/2009-01-0882.
Ruth, R.R., West, O., Engle, J., Reust, T.J., “Accuracy of Powertrain Control Module (PCM) Event Data Recorders,” SAE Technical Paper 2008-01-0162, 2008, doi:10.4271/2008-01-0162.