Abstract:
This paper presents a validation of two public domain finite element models of the Side Impact Dummy (SID) for use in vehicle and roadside safety crashworthiness research. The models, referred to as PD-SID-V3 and SRI-SID-V2, are compared against standard bench calibration tests and an in-vehicle full-scale rigid pole crash test. The PD-SID-V3 model, developed under NHTSA sponsorship, features a more detailed pelvis and lower extremities with explicit modeling of internal structures, while the SRI-SID-V2, an improvement of the original SRI model, has a more refined head-neck model and an improved thorax. The calibration tests involve striking the dummy with a pendulum and comparing the predicted acceleration time histories and injury indices (TTI and Py) to physical test corridors.
In the thorax calibration, the SRI-SID-V2's peak responses were within the acceptable range, whereas the PD-SID-V3 predicted a higher lower spine response. Both models, however, were found to be stiffer than the physical test corridors and peaked late. In the pelvis calibration test, neither model performed well; the SRI-SID-V2 over-predicted accelerations, while the PD-SID-V3 under-predicted them, with neither model's response falling within the test corridors. For the more complex in-vehicle simulation, the SRI-SID-V2 proved more robust, predicting lower spine and pelvis accelerations that were close to the physical test data. The PD-SID-V3 simulation terminated early due to numerical problems, including negative volume error and interpenetration of foam materials, which severely limited its utility. The authors conclude that while both models require further improvement, the SRI-SID-V2 is currently the more robust and effective finite element model for crashworthiness research. They also suggest that future improvements should focus on enhancing model robustness, particularly in modeling sliding interfaces and non-linear foam material behaviors, as well as refining the geometric accuracy of the dummy components.
