Submarining is qualitatively defined as the mechanism in which the lap belt, initially positioned superficial to the anterior superior iliac spines (ASIS), fails to engage the bony pelvis during occupant forward excursion, and translates superior and posterior relative to the ASIS, loading the abdominal soft tissue. While submarining increases the likelihood of abdominal injuries due to direct lap belt loading, the resulting effects on occupant kinematics and restraint interaction increase the risk of injury to nearly every body region. A multitude of intrinsic and extrinsic factors have been hypothesized to affect lap belt-pelvis interaction and submarining occurrence in the literature, and modifications to federal safety regulations were made to improve pelvis restraint based on this research. However, further research is required to investigate how these parameters affect submarining using a restraint system equipped with modern technologies and to provide updates for safety regulations. Additionally, recent advancements in automated driving system (ADS) technologies introduce new challenges to restraint design that also need to be addressed.

The goal of this dissertation was to evaluate how parameters pertaining to the vehicle environment (extrinsic) and occupant (intrinsic) affect lap belt-pelvis interaction and submarining occurrence through experiments using post-mortem human surrogates (PMHS) and simulations using computational human body models (HBMs).

First, a comprehensive literature review was performed to identify the intrinsic and extrinsic factors that have been hypothesized to affect lap belt-pelvis interaction, particularly submarining. Of these factors, investigating the effects of a reclined torso angle and the angle of the lap belt relative to the pelvis required additional experimental research and subsequent HBM validation. This finding informed the design of the experimental and computational studies of this dissertation.

As no experimental data existed to understand how a reclined seating posture affected lap belt-pelvis interaction and submarining occurrence, an innovative methodology was developed to investigate this through experimental sled tests using PMHS. One subject that exhibited a relatively large lap belt-pelvis angle (indicating a shallow fore-aft lap belt angle) submarined. This prompted the need for further experimental study to identify a submarining threshold through variation of this parameter. The GHBMC HBM was shown to exhibit better biofidelity over the THUMS in the reclined sled test condition, with more biofidelic lumbar spine compliance and lap belt-flesh-pelvis interaction.

Lap belt-pelvis angle was further investigated in belt pull experiments through systematic variation of lap belt and torso angle, which identified a submarining threshold. Additionally, this parameter was found to affect the placement of the lap belt relative to the pelvis. The GHBMC HBM ranked superior relative to the THUMS HBM, as the THUMS showed a downward lap belt migration relative to the pelvis at a steeper fore-aft lap belt angle, which was not seen in the PMHS.

This dissertation’s final parametric study, in which the effects of extrinsic and intrinsic factors on lap belt-pelvis interaction and submarining would be quantified, was setup using the GHBMC with several varied intrinsic and extrinsic factors. These simulations were sampled to approximate the design space for a subset of 480 simulations and were used to develop, train, and test a Neural Network (NN) metamodel which predicted submarining occurrence and distance, pelvis kinematic and lap belt kinetic outputs based on these varied parameters.

From the NN metamodel’s predictions, fore-aft lap belt angle and recline angle were identified as the dominating factors that affected submarining occurrence. A shallow lap belt angle, in combination with a reclined torso angle, was linked to a higher likelihood of submarining. Submarining risk was decreased for a steeper lap belt angle in both postures, however in the reclined posture this risk was only substantially mitigated at the steepest fore-aft lap belt angle (87°). Additionally, the range of permitted lap belt angles by FMVSS 210 was shown to be insufficient in mitigating submarining risk in a reclined posture with a modern restraint system (equipped with dual lap belt pretensioners). For a reclined seating posture, the lap belt anchorages must be positioned further forward relative to the occupant to reduce submarining risk. Potential trade-offs resulting from moving the lap belt anchorages further forward included increased pelvis forward displacement (from 63° to 75° in upright and from 75° to 87° in recline) and increased lap belt tension (from 46° to 75° for both postures).

This dissertation provided the automotive safety community with a wealth of data to inform restraint system design for current and future vehicles. Specifically, this improves the automotive safety field’s understanding of the fundamental characteristics that influence lap belt-pelvis interaction and submarining. This research also identified limitations of current safety standards for a restraint system equipped with modern technologies, and for a reclined seating posture. Finally, the quantification identified consistencies and differences across different torso angles, providing guidance on potential vehicle environment modifications that need to be made to mitigate submarining in a reclined posture versus an upright posture.