Sled Test
Belted Sled Test Validation
To prove the validation of the model, the sled testing was specifically designed to be highly controlled with as few unknowns as possible, so that the pure response of the dummy model could be measured and validated and not the restraint system.
A rigid structure was fabricated with seat belt anchorage locations taken from an average mid-size vehicle. There was no knee bolster or instrument panel.
The restraint system comprised a lap belt and a separate shoulder belt made from low strain webbing and bolted directly to the rigid seat. No slip-rings or retractors were used.
A sled acceleration pulse was developed representing a typical 30mph front impact and standard Hybrid III instrumentation was used. Seat belt load cells were used to measure seat belt forces. 
Fig. 27 Belted Sled Test Validation
Sled Test Simulation
Fig. 28 Model Time 0ms 
Fig. 30 Test Time 0ms 
Fig. 29 Model Time 100ms Fig. 31 Test Time 100ms
Belted Sled Test Results
Despite the controls put on the sled test specifications, the initial belt slack proved difficult to measure. The seat belt load cell mass caused significant belt sag under static conditions. Parametric studies were used to achieve an accurate representation of the belt shape in the model and the shape of the belt was varied until the timing of the initial increase in belt loads matched the test data.
High speed cameras recorded the test. Fig. 28 to Fig. 31 compare the kinematic response of the model and the physical test. Correlation is generally good, except that at 100 ms the hand motion is different. Parametric studies have shown that the limb motion is extremely sensitive to the exact torque settings of the friction joints, and there may be significant variation in this joint stiffness on the physical dummies depending on test set-up.
Two nominally identical sled tests were performed with the same physical dummy and there was very good repeatability between the tests. The correlation for key dummy injury criteria is shown in:
Fig. 32 Head Resultant Acceleration
Fig. 33 Chest Resultant Acceleration
Fig. 34 Pelvis Resultant Acceleration
Fig. 35 Chest Compression
Femur loads are not discussed here since there was no knee bolster used in the test, and the loads would be inertial only.
When run on a HP J21OXC workstation, this sled test model required 17.1 hours of CPU to reach 150ms, using a time step of 1.54gs.
This sled test is just the first of a series of tests designed to provide a comprehensive database for the validation of the Hybrid III dummy models. An unbelted sled test with a driver's side airbag is currently being designed and will be the next validation condition for the dummy model. 
Fig. 32 Head Resultant Acceleration
Fig. 33 Chest resultant Acceleration
Fig. 34 Pelvis Resultant Acceleration
Fig. 35 Chest Compression
