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

5.2Methodology

The model work commenced with a previous MADYMO model designed primarily for frontal impacts. It contained a Hybrid III crash test dummy seated inside a full vehicle interior. From this, any vehicle acceleration could be applied linearly in the x, y, or z-axis directions, concurrently with rotational accelerations around the three axes. Contact interactions and force deflection functions between the dummy and vehicle interior were previously defined in the model. The first far-side simulations began by applying y-axis acceleration pulses for a side impact test to this same model.

Upon deleting the Hybrid III dummy from the model, the remaining vehicle interior was used as the basis of future simulations with other dummy models. The original model was written in MADYMO Version 5.4 and converted to MADYMO 6.0.1 in XML computer language. All further modeling conducted in MADYMO 6.0.1 table format.

5.2.1Vehicle Interiors

The vehicle interior, loosely based on 1999 Ford Taurus geometry, used mostly ellipsoids to represent interior surfaces. These ellipsoids connected to each other through rigid joints. Computationally, ellipsoids are the easiest for MADYMO to process, as opposed to faceted surfaces, for increased model performance. Planes were used in some locations such as floorboard, dashboard, and roof. Although the seat appears to be constructed of ellipsoids, actually a plane existed underneath the ellipsoids for seat bottom and another for seat back contact interactions, instead of making contact with the seat ellipsoids. The vehicle configuration is pictured in figure 1.

This interior model contained a number of supplemental safety devices found in modern vehicles. A 3-point belt restraint system surrounded the driver. This belt had the option of having belt force limiter and pyrotechnic pretensioner, both of which could be turned on or off inside the input file code. An airbag could also be placed in the steering column of the model, however, the airbag does not deploy in side-impacts, and not used in these simulations.

The interior was broken into two portions - the passenger-side and driver-side. The driver-side included driver seat, instrument panel, and steering column. The passenger-side contained the passenger seat, center console, and armrest. The seatbelt system moved during the impact and was thus not a part of either driver-side or passenger-side rigid systems. It was instead included in the inertial space.

To better organize the code of the original model, each ellipsoid and plane of the interior had a characteristic function associated to it, an advantage of MAYDMO Version 6. These functions were pulled from a library of function in the original Hybrid III model. The mathematical purpose of these functions was for when another ellipsoid penetrated this interior representing ellipsoid, the function would determine the amount of force to exert back by measuring the distance penetrated. This function was combined with the characteristic function of the penetrating ellipsoid to determine a final force-deflection function.

TNO, the suppliers of MADYMO code, provide several computer models of crash test dummies. Of the many supplied, the one’s chosen for this study were – BioSID, EuroSID I, EuroSID II, SID2s, and human finite element model. Also, of interest would be the WorldSID model, however it does not exist. All side-impact dummies (SID’s) used were right-hand impact oriented.

5.2.2Dummies and Initial Setup

After inserting each new dummy into the vehicle interior, two processes needed to be followed to properly setup the model – dummy rest positioning and initial belt positioning.

After inserting the dummy into the model and properly defining all environmental contact functions, rough dummy positioning was visually achieved through EASi-Crash software. To further specify the correct dummy position, gravity (-9.81m/s² linear acceleration in z-axis) was applied to each rigid body of the particular dummies. No other accelerations or inputs were specified. The model was run for a long duration of time, allowing the dummy to fall to resting position and reach equilibrium with the interior. The final joint positions at equilibrium were taken from the appropriate output file and placed in the model.

To position the belts, each end of the lap or shoulder belt was gently accelerated towards its anchor. The dummy was locked in place and made rigid during this process. The process gently contoured the belt around the respective dummy. For each cycle, the belt nodal positions at the desired point in time were copied into the program and used as the initial positions for the next simulation.

Figure 1: Vehicle Interior

5.2.2.1BioSID

The BioSID crash dummy, based upon a 50^th percentile adult male, contains legs, head, and neck similar to Hybrid III. The BioSID rigid spine supports 6 deformable ribs for near-side impacts. The ribs mount on the left-side for left impacts and right-side for right impacts. This particular model used a right-side dummy to study far-side impacts to the driver.

The BioSID MADYMO model, developed at the time of the old Hybrid III non-segmented neck, needed updating with the current more biofidelic segmented neck. The updated neck is modeled and contained in the Hybrid III MADYMO model, however, not found in the BioSID MADYMO model. For purposes of far-side simulations, dummy neck responses were important to study, therefore necessary to update. The released BioSID MADYMO model was designed and validated against a dummy with the old non-segmented neck.

Updating the BioSID with the new neck created a non-validated model, however, assumed to be suitable. The new used the exact same underlying code and properties for the Hybrid III segmented neck. This was copied and inserted to the BioSID model between the same two rigid bodies as the old neck, as seen below. The heads are identical in BioSID and Hybrid III and both use a rigid plate between the shoulders for a base.


As shown, the new neck added extra bodies and joints between the head and shoulder, but maintained the same distance between start and end points.

5.2.2.2EuroSID I & II

The EuroSID I and II MADYMO models appear identically, however, differences exist in a couple joint properties and configuration, particularly in the shoulders and chest areas.

Due to this unique application of EuroSID models for far-side instead of near-side for which it was designed, a challenge arose with the seatbelt interactions. Typically for near-side impacts, the shoulder seatbelt gets trapped in the shoulder neck region while the dummy contacts the door. For far-side, the upper torso of the human/dummy moves laterally toward the passenger seat, sliding the shoulder belt off the outside shoulder. In the EuroSID models, a valley exists between the collar bone ellipsoid and shoulder ellipsoid. The seatbelt dropped into this valley on initial simulations and unnaturally restrained the driver.

To correct the grabbing problem, the collar bone ellipsoid was extended laterally for both the EuroSID I and II models to overlap the shoulder ellipsoid and allow the shoulder belt to slide of the shoulder during occupant excursion to the far-side. Figure 2 shows the original ellipsoid configuration and the modified EuroSID.

5.2.2.3SID2s

The SID2s, based on 12 to 14 year-old adolescents, was the only petite sized occupant model evaluated for far-side impacts. The same procedure was followed as with the other models – gravity used to pull the model into position and seatbelts pulled snug over the pelvis and chest.

The resulting initial simulations displayed some challenges with the seatbelt and torso interactions. The model, again designed for near-side impacts, contains 6 thin ribs for the upper body. When the shoulder belt moved to restrain the ribs, the nodes of the belts fell between the ellipsoids and the shoulder belt acted transparent to the chest. To alleviate the situation, the faceted mesh for the shoulder belt was made twice as fine, thereby created more nodes closer together and more contact points for the belt to ellipsoid.

5.2.2.4Human Finite Element Model

The human finite element model released by TNO Automotive is the most detailed and sophisticated of any of the preceding models. Based upon cadaver data, the model uses faceted bone models and layers of finite elements for deformable tissue. The model contains true skeletal and organ geometry. It visually appears as a human rather than dummy.

The model begins with the cadaver in a seated position with arms extended forward. Unfortunately, the model is limited against positioning the human out of this initial setup. Therefore, gravity was not used to position the dummy.

Due to the complexity to the model, run time was an important consideration. To simplify the model, all environmental surfaces were removed, except the planes for the seat and ellipsoid for center console.

The model was computed using 4 processors simultaneously to improve run-time.

5.2.2.5Human Faceted Model

A more simplified cadaver experimental model is also released by TNO. This uses rigid bodies and faceted surfaces. Using the pre-defined contact groups, the dummy was placed in the vehicle interior, and positioned. The belts were contoured around. Due to difficulties with two faceted surfaces contacting one another, as with the belts to the faceted human model, ellipsoids were placed underneath the skin facets to improve the belt contact properties.

5.2.3Crash Pulses

For these far-side simulations, the acceleration pulse was taken from a US New Car Assessment Program (NCAP) NHTSA 214 test of a 2000 Ford Taurus SE 4-door. The x and y plots are displayed below in figure 3 (the plot of large amplitude is y-axis acceleration). The vehicle was struck on the near-side, and the y-component multiplied by -1 to make it far-side for simulations. This case was almost a true 90 degree impact, only a small acceleration is seen for the x-axis.

Additionally to understand far-side impacts, the pulse was rotated 45 degrees towards the front and repeated for all simulations. The rotation was accomplished by multiplying the appropriate functions by cosine and sine of 45 degrees and adding together.

5.2.4Reverse Seatbelts

A possible countermeasure for far-side impacts is the use of a reverse 3-point seatbelt. A traditional seatbelt starts at the driver’s outside shoulder, latches at the inside waist, and ends at the outside waist. A reverse belt is just the opposite – inside shoulder to outside waist to inside waist. The rationale behind this setup is to utilize the inside shoulder belt to prevent occupant excursion. Injuries occurring from far-side impacts stem from the occupant sliding out of the shoulder belt and moving to the passenger seat. The reverse belt will capture the shoulder neck region and possibly keep the occupant in its seat.

Modeling the reverse belt involved reflecting all seatbelt anchors and the retractor mechanism about the centerline of the seat. The y-axis coordinates of the nodes for the lap and shoulder belts were also reflected about the centerline of the seat. However, by doing this, the normal for each facet reversed and needed to be reflected back towards the occupant.

5.2.5Airbags

In addition to the reverse belt, some further countermeasures included inboard chest and shoulder airbags. Ellipsoids were used to roughly simulate these. Figure 4 below shows the location and size of these ellipsoids. A soft function with high hysteresis was chosen to represent the airbag for simplicity.

5.2.6Test Matrix Summary

Six dummy models in total were analyzed, they were,

1. 
   Hybrid III
   
2. 
   BioSID
   
3. 
   EuroSID I
   
4. 
   EuroSID II
   
5. 
   SID2s
   

With these models, five vehicle configurations were simulated –

1. 
   Baseline (just vehicle interior)
   
2. 
   Reverse Belts
   
3. 
   Base with chest airbag
   
4. 
   Base with shoulder airbag
   
5. 
   Base with chest and shoulder airbag
   

All five configurations were repeated again for the six dummies, but with a 45degree crash pulse. This totaled 50 simulations for the dummies, excluding human models.