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

2.2Results

The TNO human finite element model proved to be difficult to use. First, the dummy cannot be initially positioned - the finite elements are fixed and validated in a certain position. Secondly, the runtime for the computer was extraordinary by MADYMO standards. It took a multi-processor machine 36 hours to run on 4 processors simultaneously. Finally, the model consistently went mathematically unstable. The meshed surfaces experienced extensive deformation causing the nodes to become chaotic and difficult to calculate further. The model could not run to completion. Figure 2 shows the human finite element simulation with theoretical center console pelvis support.

By examining figure 2, one must question the biofidelity of this model for far-side. The pelvis translation shown in the model is perhaps unrealistic. Also, the human model displayed extensive intrusion into the body, which a true human is unlikely to react in such a way with respect to the pelvis. It is noted that the model was developed and tested preliminarily for frontal impacts.

Figure 2: Human Finite Element Simulation

Conversely, the TNO human facet model showed promise when modeling a far-side crash. This model is simpler than the finite element model. It assumes rigid bodies surrounded by fixed nodes creating a faceted surface. Further, advantageous to this model is the representation of a flexible spine by modeling each vertebra with individual rigid bodies. Also, it provided flexibility in initial positioning by rotating joints for the knees, hips, shoulders, elbows, etc.

Figure 3: TNO Human Facet Model and Spine
Some modifications were made to the model in order to make the human properly react with the seat belts. It is difficult for MADYMO to mathematically calculate a facet to facet surface contact. The quantity and penetration of nodal surface into nodal surface is difficult to resolve. Therefore, additional ellipsoids were added to the human underneath the skin layer. These ellipsoids surrounded the same rigid bodies that the facets do, except are used for the belt contact surfaces instead of the facets. Ellipsoids were added for the rigid bodies making the pelvis, abdomen, shoulders, and upper arms. Besides the seat belt contacts, these ellipsoids are inconsequential to the rest of the model.

Upon placing the human facet model into the interior of the vehicle, positioning it properly, and defining the contact functions; the model duplicated the motion of the cadaver quite well. The excursion and upper body motion of the two were similar. Figure 4 shows the two synchronized at discrete time steps.

Figure 4: Human Facet Model with Cadaver Comparison
The crash pulse recorded during the cadaver test was used as the input for the MADYMO model. This pulse measured the lateral acceleration of the Holden Commodore as it was struck on the far-side. Figure 5 below shows this acceleration plot.

Figure 5: Far-side Cadaver in Holden Commodore Pulse

One difference between the two models is noted in the shoulder belt. In the MADYMO model, the shoulder belt easily falls off the occupant. In the cadaver, some light friction keeps the belt on the occupant. In the MADYMO model, it is difficult to model friction between to faceted surfaces, which may explain this.

To further examine the human facet model, a direct comparison with a Hybrid III MADYMO model was made. A pulse from an IIHS barrier test was used to excite both models simultaneously. Figure 6 below shows snap shots of the two at 0, 100, and 150 milliseconds.

The vertical line through the passenger seat estimates the maximum intrusion distance from the impact. This estimation comes from a finite element vehicle model conducted at the GWU – National Crash Analysis Center (Digges et al, AAAM, 2005). The difference between the human facet model and Hybrid III is clear when examining the occupant excursion towards the impact. The head of the human facet model traveled approximately 30cm further than the Hybrid III.

Figure 6: Human Facet MADYMO Model vs. Hybrid III MADYMO Model

2.3Conclusions

Using visual approximations, the human facet model moved similarly to a human cadaver test under the same impact configuration. Using this single test as a benchmark, comfort was gained in this model – it moved more similar to the human than any crash dummy or other computer model did. In contrast the human finite element MADYMO model seemed not to be too realistic, and it was difficult to use.

As shown in Figures 1 and 6, the MADYMO dummy models (Hybrid III, BioSID, EuroSID 1, EuroSID2, or SID2s) did not accurately reflect the motion of a human cadaver under the same impact configurations as a cadaver test.

In the future, this model should be refined and correlated to more cadaver tests. Parameters such as chest and head acceleration plots should be examined more closely and checked for correctness.

2.4References

Alonso, B., Far-side Impact Simulations with MADYMO, Report written to George Washington University – National Crash Analysis Center, October 2004.

Digges, K., Gabler, H., Mohan, P., Alonso, B. Characteristics of the Injury Environment in Far-side Crashes. Association for the Advancement of Automotive Medicine, 2005.

Fildes, B., Sparke L., Bostrom O., Pintar, F., Yoganandan N., and Morris, A. Suitability of Current Side Impact Test Dummies in Far-side Impacts, Proceedings of the 2002 International IRCOBI Conference on the Biomechanics of Impact, Munich, Germany, September 2002.