Methods for assessing submarining occurrence in PMHS frontal sled tests: Exploring potential indicators.

Objective: Assessment of submarining occurrence in PMHS (Post-Mortem Human Subject) testing can be challenging, particularly for obese PMHS. This study investigates varied kinetic and kinematic response parameters as potential indicators of submarining. Data from 36 whole-body PMHS frontal sled tests conducted under varying boundary conditions were analyzed, incorporating three spring-controlled seat configurations, two extreme anthropometric profiles, two crash pulses, and two seatback angles.

Methods: Submarining assessment was based on seven response parameters: (1) Visual inspection of video, (2) Lap belt load trace morphology, (3) Pelvis rotational displacement, (4) Iliac bone fracture, (5) Torso-knee angle, (6) Hip-forward displacement, and (7) Anti-submarining plate deflection. High-speed video evaluations were independently performed by two authors and an expert from injury biomechanics field, with inter-observer agreement quantified using the Kappa statistic. A logistic regression model was developed to estimate submarining probability based on alternative indicators, using cases where inter-observer agreement confirmed submarining or its absence. A bootstrap technique with stepwise covariate selection identified the most significant indicators for the final model.

Results: Among the 36 tests, the final inter-observer agreement for cases with definitive conclusions was 98.5%, with a Kappa score of 0.95. Based on these analyses, 33 out of 36 cases were used for training the statistical model. The stepwise logistic regression model identified lap belt load trace morphology, pelvis rotational displacement, and torso-knee angle as the most significant indicators, forming the basis of the final submarining predictive equation. When final model was applied to all cases, the tests involving a soft spring seat exhibited the highest submarining probability (33%), while a reclined (45°) seatback position accounted for 21% of submarining cases. Among subject profiles, small females showed the highest likelihood of submarining (23%), whereas obese subjects exhibited a comparatively lower likelihood (11%).

Conclusion: This study explored varied kinetic and kinematics indicators for assessing submarining occurrence in frontal PMHS sled tests and developed statistical regression model to predict the occurrence of submarining. Softer spring seat, small female anthropometry, and reclined configurations had higher probability of submarining. These findings may help guide future biomechanics research by highlighting relevant indicators and extrinsic factors that influence submarining. For the modeling community, the probabilistic framework and identified predictors offer inputs for refining computational human body models and enhancing the fidelity of injury prediction in diverse occupant scenarios.