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

6.5Simulations

The sled test configurations are defined by the following parameters: pulse shape, pulse magnitude, pulse duration, sled angle, and instrument panel angle. Three crash test cases are presented in detail to demonstrate the process used to determine the proper sled configuration. The cases to follow will concentrate on the appropriate sled angle and instrument panel angle. A discussion of the crash pulses can be found in the documentation for the complete study [Cuadrado 2008].

The available crash tests did not have a dummy located in the far-side seating position. Consequently, the motion of the far-side occupant was simulated using the MADYMO human facet model. The simulation was accomplished by applying the vehicle acceleration pulse measured at the occupant location during the actual crash. This simulation was called “Pulse” because the actual crash pulse was applied to the simulation.

Two belt configurations were simulated. These configurations were lap belt only and conventional three point lap and shoulder belt. In the baseline simulations, the far-side occupant was restrained by a lap belt only. This belt configuration was chosen to provide the most extreme upper body excursion for a restrained occupant. This configuration also represented a worst-case scenario in the event the shoulder belt had no influence on the kinematics.

The kinematics obtained from the lap belt “Pulse” simulations formed the basis for judging the suitability of a sled configuration to provide the same general motion and upper body contact locations. The sled test was simulated applying a constant direction acceleration pulse that was representative of the crash to an occupant compartment configuration that was rotated relative to the direction of travel. The simulations of the sled configurations were labelled “Sled”.

It was found that in some crashes, the rotation of the vehicle relative to the occupant was sufficient to require that the dashboard be rotated as well as the sled angle. The kinematics of the far-side occupant restrained by a lap belt was compared with the kinematics from the “Pulse” simulation. The sled and dashboard angles were adjusted until agreement was reached. After agreement of the lap belt configurations, the “Pulse” and “Sled” simulations with three point belts were compared.

6.5.1NHTSA 4660 30° Corner Impact Simulation

The 30° corner crash test (NHTSA 4660) involved two moving cars that impact in the front corner of the target car. The test was of a 1996 Toyota Avalon impacting a 1997 Honda Accord traveling at 56.6 and 56.2 km/hr, respectively. The target car was impacted at 30 degrees relative to the x-direction centreline. The center of the bullet car impacted the left front corner of the target car. Although this was a frontal crash it involved a side vector that produced significant vehicle rotation. This impact caused the right front passenger to be subjected to far-side acceleration. A right front corner crash is a common far-side crash for the driver of a vehicle impacted while making a left turn across oncoming traffic.

Using the procedure outlined in Sections 2.3 and 2.5, occupant kinematics in the “Pulse” and “Sled” configurations were compared to determine the best sled configuration to simulate the 30 degree corner crash. For this crash configuration the sled test angle was 43 degrees relative to the vehicle y-direction centreline. The vehicle rotation required a dashboard rotation of 8.4 degrees to provide accurate head contact location and timing.

Figures 4-6 show what are considered to be important times in the occupant’s motion. Figure 4 shows the initial contact between the occupant’s upper body (hand) and the vehicle’s instrument panel 60 ms into the simulation.

Figure 4. Comparison of Initial Hand Contact at t=60ms Between Crash Pulse Model (left) and Sled Test Model (right)

The most important contact in this analysis is between the instrument panel and the occupant’s head or thorax. For this case, the occupant’s head contacts the dashboard at approximately 110 ms into the simulation (Figure 5). The sled test configuration was then run with the occupant restrained in a 3-point belt and the results were shown to match that of the 3-point belt crash pulse test (Figure 6).



Figure 5. Aerial View at t=110ms between 3-Point Belt Crash Pulse Model (left) and Lap Belt Sled Test Model (right)

Figure 6. Aerial View at t=110ms between 3-Point Belt Crash Pulse Model (left) and 3-Point Belt Sled Test Model (right)

After completing the visual analysis of the occupant kinematics, the motion of the head was plotted in the coordinate system of the instrument panel. Figure 7 shows that the resultant head excursion for both the lap and 3-point belt sled tests is almost identical to crash test models.


Figure 7. NHTSA 4660 Resultant Head Excursion
Figure 8 shows the PDOF of the occupant’s head in the horizontal plane. The pulse models exhibit an overall rotation of approximately 25 degrees.

Figure 8. NHTSA 4660 Head PDOF

6.5.2Y Damage Crash Simulations

This section presents simulation results for occupant’s interaction with the vehicle interior for a crash with Y damage based on Collision Damage Classification (CDC) method of accounting [SAE J224]. This classification requires the damage to the vehicle extend over both the center and the front part of the vehicle’s side. The test condition for the Y damage case was an S-10 pickup travelling at 30 mph impacting the side of a Ford Taurus at 90 degrees. The center of impact was near the front axle. The longitudinal damage produced was over the forward 2/3 of the side and it was rated Y damage according to the CDC scale.

Using the procedure outlined in Sections 2.3 and 2.5, occupant kinematics in the “Pulse” and “Sled” configurations were compared to determine the best sled configuration to simulate the 30 degree corner crash. For this crash condition, the best sled angle was 87 degrees. The dashboard rotation was 8.4 degrees.

Figure 9 displays the contact between the occupant and center console 75 ms into the simulation. At this point the occupant begins to cross over the vehicle centerline and is in a similar position in both simulations.

Figure 9. Aerial View at t=75ms between Lap Belt Crash Pulse Model (left) and Lap Belt Sled Test Model (right)

The occupant does not contact the instrument panel in the Y damage test. For this case, the event is considered over when the occupant’s thorax began to cross the centerline of the near-side occupant’s seat at approximately 110 ms into the simulation (Figure 10). A comparison of the occupant’s position relative to the rotated instrument panel shows a similar response in each simulation. This reiterates the significance of rotating the instrument panel in high yaw crashes, as the relative position to the vehicle floorboard is considerably different in each test.

Figure 10. Aerial View at t=110ms between Lap Belt Crash Pulse Model (left) and Lap Belt Sled Test Model (right)

Figures 11 and 12 give frontal and rear views of the crash pulse model and sled test model at 110 ms into the simulation. The slight difference in the models is attributed to the rotation of the crash pulse model. This is most evident in the contact between the occupant and seat observed in (Figure 12). Overall, there is little difference between the contact location and time between each simulation.


Figure 11. Front View at t=110ms between Lap Belt Crash Pulse Model (left) and Lap Belt Sled Test Model (right)

Figure 12. Rear View at t=110ms between Lap Belt Crash Pulse Model (left) and Lap Belt Sled Test Model (right)

The Y damage simulations were run with the occupant restrained in a 3-point belt (Figures 13 and 14). The results of these simulations indicate that the sled test accurately duplicates the motion of the crash pulse model using the same configuration as the lap belt sled test. The simulations show that the occupant slips out of the shoulder belt in this crash environment.


Figure 13. Aerial View at t=110ms between 3-Point Belt Crash Pulse Model (left) and 3-Point Belt Sled Test Model (right)



Figure 14. Isometric View at t=110ms between 3-Point Belt Crash Pulse Model (left) and 3-Point Belt Sled Test Model (right)

The motion of the Y damage occupant’s head was measured in the coordinate system of the rotated instrument panel (Figure 15). The resultant head excursion for both the lap and 3-point belt sled tests is almost identical to their respective crash test models. The maximum occupant head excursion for the lap belt tests and 3-point belt test were approximately 800 mm and 650 mm, respectively. Although the shoulder belt reduced occupant excursion, the occupant still slipped out of the shoulder belt and crossed into the area that a near-side occupant would be seated.


Figure 15. Y Damage Resultant Head Excursion
The accuracy of each sled test can also be seen in the occupant’s head PDOF for each case (Figure 16). In each case the final head PDOF is similar between the sled test and crash pulse test, but the lap belted final PDOF is identical between the crash pulse and sled tests whereas the 3-point belt sled test is approximately 5 degrees different from its corresponding crash pulse test. The final head PDOF in all cases is approximately 85 degrees.


Figure 16. Y Damage Head PDOF

6.5.3SNCAP Crash Simulations

The SNCAP was chosen for the detailed study because it is based on an actual test performed by NHTSA (Test No. 3263) and it involves an angled crash vector. The test impacted a 2000 Ford Taurus with a NHTSA moving deformable barrier (MDB) travelling at 38.5 mph. The surface of the MDB impacts the vehicle side squarely but has an impact angle of 63° due to the 27° crab angle shown in the figure below. This technique is used to create a longitudinal component in the crash pulse.

Although the travel angle of the movable barrier was 63 degrees, the appropriate sled angle was found to be 87 degrees. No dashboard rotation was required.

The following figures show significant times in the SNCAP occupant’s motion. Each occupant in the lap belt simulations (Figure 17) initially contacts the center console 52 ms after impact. The relative position of the occupant within the vehicle is observed to be the same for both simulations.


Figure 17. Rear View Comparison of Initial Contact with Center Console at t=52ms Between Crash Pulse Model (left) and Sled Test Model (right)
The motion remains similar up to the point that the occupant crossed the mid-plane of the vehicle. At that point there is significant interaction between the occupant and center console and the crash is considered complete for the purposes of replicating the occupant’s motion. The results at this time are shown in Figures 18 and 19.

Figure 18. Aerial View Comparison of Occupant Interaction with Center Console at t=90ms between Lap Belt Crash Pulse Model (left) and Lap Belt Sled Test Model (right)

Figure 19. Rear View Comparison of Occupant Interaction with Center Console at t=90ms between Lap Belt Crash Pulse Model (left) and Lap Belt Sled Test Model (right)
The same sled configuration was then run with the occupant restrained in a 3-point belt and the results were compared to the 3-point belt crash pulse test. As predicted, the sled test occupant’s motion matched that of the crash test occupant and did not need adjusting with regard to the sled angle or pulse. Figures 20 and 21 indicate that the shoulder belt is ineffective in reducing significant occupant excursion for this crash environment.

Figure 20. Aerial View Comparison of Occupant Interaction with Center Console at t=90ms between 3-Point Belt Crash Pulse Model (left) and 3-Point Belt Sled Test Model (right)

Figure 21. Rear View Comparison of Occupant Interaction with Center Console at t=90ms between 3-Point Belt Crash Pulse Model (left) and 3-Point Belt Sled Test Model (right)
The motion of the occupant’s head is shown graphically in Figures 22 and 23. Comparing the results of the occupant only restrained by a lap belt shows that the head excursion is nearly identical for the crash pulse model and the sled test model until about 90 ms after impact. Likewise, the sled test occupant restrained by the 3-point belt had a similar head excursion to that of the crash test. The 3-point belt reduced occupant head excursion by approximately 120 mm, but the resultant head excursion exceeded 500 mm for each test and restraint condition.


Figure 22. SNCAP Resultant Occupant Head Excursion
The similarities between model response can also be seen in a cross plot of the occupant head excursion in the x/y plane. For this analysis the positive x-direction is in the direction that the passenger is facing when seated in the vehicle. Figure 23 shows that the motion of the occupant’s head in the horizontal plane is nearly identical between the crash test and sled test for each restraint design.

Figure 23. SNCAP Head Excursion Cross Plot

Another means of displaying the motion of the occupant’s head is the head principle direction of force (PDOF). The plot (Figure 24) shows that the motion of the occupant’s head in the sled test is linear and approximately equal to the angle of the sled (87°). However, the crash test occupant is not constant because of the acceleration being applied in both the longitudinal and lateral directions. After about 70 milliseconds the head PDOF is equal in both the sled and crash test. Of particular importance is that the head PDOF is equal when the event is considered complete at 90ms.


Figure 24. SNCAP Head PDOF