Helmet Testing Digital Twin
Reason: Thesis contains material to be published.
until file(s) become available
COMPUTATIONAL MODELING OF A SCALABLE HUMAN BODY AND DEVELOPMENT OF A HELMET TESTING DIGITAL TWIN
Human body models (HBMs) have been present in the automotive industry for simulating automotive related injury since the turn of the century and have in recent years found a place in assessment of soldier and sports related injury prediction and assessment. This issue is the lack of models that lie outside of the 50th percentile. By a simple application of physics, it is evident that acceleration or force will affect people of varying weights differently. To this end, having the ability to scale a 50th percentile HBM to targets for weight and stature would allow for better characterization on how an impact or acceleration event will affect people of differing size, especially when ~90% of males can fall outside the 50th percentile for weight and stature and HBMs models from vendors exist in only a few variations outside the 50th percentile . Using Corvid Technologies’ 50th percentile model CAVEMAN (capable of being repositioned) as a base, scaled model from the 5th to 95th percentiles of stature and weight were generated based on ANSURII metrics, using a combination of 1D and 3D scaling transformations. These models met their stature and weight metrics when standing and weight metrics when positioned.
After creation of a framework to scale the CAVEMAN HMB, creation of a digital twin to the HIRRT Lab helmet testing model commenced. With the HIRRT Lab’s history of experimental testing of football helmets, a natural turn of events was to bring helmet performance testing into the computational space. This digital twin was a natural evolution and addition to the HIRRT Lab’s helmet testing as it would enable manipulation of helmets that would be infeasible experimentally. After calibration of the barehead using experimental data, helmeted simulation began. Angle of impact, while it was found to effect peak translational acceleration, was found to profoundly effect peak rotational acceleration. With this in mind, various angles of impact were simulated to produce curves similar to experimental results. Helmeted simulations were qualitatively dissimilar to experimental data, prompting a modification of the padding material used by the models. Following various modifications of the padding material model, these inconsistencies between simulated helmets and experimentally tested helmets persisted. These inconsistencies highlight a need for better characterization of material, such as foam, and more thorough validation of simulated helmet models. The results of the helmeted simulations are difficult to quantify, as the evaluation criteria used for the BioCore model did not include rotational acceleration, indicating a need for further research and simulation is necessary.