Arc far Side Impact Collaborative Research Program – Task 5b: Test Procedures Crash Tests and Sled Tests for the Far-side Environment

6.3Methodology

The kinematics of far-side occupants exposed to impacts that involve vehicle rotation are largely unknown. As of 2007 there are over 5000 crash tests in the National Highway Traffic Safety Administration (NHTSA) database, but only three tests were found to have been performed with a far-side dummy. These tests were unable to be used in this study because they were not instrumented with accelerometers in the lateral direction and poor video quality prevented accurate yaw data to be obtained through video analysis. In addition, US Government research has not conducted crash tests in configurations that are responsible for most of the far-side injuries. Consequently, the far-side occupant motion in a crash test configuration must be determined in order to design a sled test for each considered case. The cases chosen in this study were based on three full-scale crash tests that were available in the GW film library. None of these crash tests contained a dummy on the far-side of the vehicle.

The methodology involved running the MADYMO model in two steps [MADYMO 2004]. The MADYMO human facet model was selected based on the validation of the model for far-side occupant simulations by Alonso [2005].

In the first step, the actual vehicle acceleration and yaw experienced by the subject vehicle in each crash test were used to create the acceleration environment for a MADYMO model to determine the motion that a far-side occupant would undergo in each crash mode. The computer model permits the vehicle to undergo both linear and angular accelerations that vary with time. Consequently the actual rotation of the vehicle can be simulated. However, it is assumed that the sled angle will remain constant. The challenge is to determine the appropriate sled angle to simulate the far-side crash environments represented by injury producing events.

The second step was to determine a sled test with a constant angle of rotation that best simulates the far-side occupant motion and contacts. In this step, runs with different sled angles were conducted and the far-side occupant motion was compared with the response in the crash test from step one.

The input variables of the sled test model were altered until the occupant motion matched that of the crash test model. The reference condition for the evaluation of occupant motion was the occupant restrained only by a lap belt. The objective was to match the trajectory of the head as closely as possible to the occupant in the crash pulse model. The lap belt configuration was chosen to maximize the amount of distance the occupant’s head would travel unimpeded. Errors that may be indistinguishable when comparing 3-point belted occupants will be exacerbated due to the longer travel distance. The proper configuration was then simulated and compared for occupants using 3-point belt restraints. The final sled test configuration can then be used in an actual sled test to mimic the occupant motion of a far-side occupant exposed to the original crash environment.

6.4Vehicle Model Development

The NHWA/NHTSA National Crash Analysis Center’s vehicle modelling laboratory has created a library of 14 vehicle finite element models that are available for download and public use. The database contains 3 sedan models, the most detailed of which is a 2001 Ford Taurus. This model was chosen for the MADYMO simulations because all 3 crash tests were performed with sedans. A single vehicle geometry was used to model each pulse in this study in order to isolate the effect that the sled angle has on the occupant motion. This simplification allows for a baseline comparison between crash tests by eliminating differences due to different vehicle interior geometries.

The Taurus model was reduced to the components that would be interacting with the occupant during the tests in order to minimize the simulation time. Once the model was reduced to the necessary components, several preliminary simulations were run and determined that a 200 ms crash simulation would take over two hours to complete. The model was simplified further by using larger elements to capture the interior geometry of the vehicle. This lowered the number of elements from 75675 elements to 4089, thus reducing the simulation time to approximately 7 minutes. The model is shown in Figure 1.

Figure 1. Taurus Cabin Finite Element Model (left) and Reduced MADYMO Taurus Cabin (right)

The dummy model seating position was determined from the 2000 Ford Taurus SNCAP NHTSA Test Report (Test No. 3263). Although more far-side accidents occur with an occupant in the driver seating position, the passenger position was chosen to allow for a greater travel distance before interacting with the vehicle instrument panel.

In this study the sled angle is defined as the angle that the buck is rotated with respect to its position in the crash pulse tests. The pulse is applied to the buck in the negative x-direction of the global coordinate system. The sled angle for a frontal impact test is 0° and is 90° for a side impact test. The initial sled angle for each pulse was set to the impact angle between the bullet vehicle/barrier and the longitudinal axis of the target vehicle. The simulations were then run using the baseline vehicle crash pulse. The sled angle and pulse were then adjusted on a trial and error basis until the timing and contact locations of the sled test occupant matched that of the crash pulse model.

Results from sled test simulations revealed that the angle between the occupant’s torso and the instrument panel significantly differed between the crash pulse model and sled test model (see Fig 2). This difference is only noted in crashes where vehicle yaw was a factor. Therefore, the difference was attributed to the rotation of the vehicle relative to the occupant.

Figure 2. Comparison of 30° Corner Impact Crash Pulse Test (left) to Initial Sled Test (right)

In order to deal with crashes in which the vehicle rotates relative to the occupant, the dashboard on the sled was rotated. This technique created a sled buck with the seat and floorboard in the coordinate system of the crash tests’ initial position and an instrument panel in the coordinate system of the crash tests’ final position. This sled modification allowed for the vehicle yaw to be considered (see Figure 3).


Figure 3. Comparison of 30° Corner Impact Crash Pulse Test (left) to Initial Sled Test (right) with Rotated Instrument Panel