Introduction: In recent years, finite element human body models (FE-HBMs or HBMs) have been developed as an evaluation tool for restraint system and safety assessment in virtual simulations. However, for HBMs to be truly useful as a tool, standardized methods need to be developed for translating the outputs from HBM simulations into predicted risk of injury. An important body region of interest for such standardization is the thorax. Although the thorax was one of the first body regions for which anatomical structures were modeled in detail in HBMs, the thorax remains one of the most frequently injured body regions in motor vehicle collisions (MVCs). Thus, there is a need to develop a framework for tuning model-specific thoracic injury risk functions which would drive towards consistent injury risk prediction across various HBMs, despite the differences in the models.
Goals of Study: The main objective of this thesis was to develop a set of guidelines and framework which standardizes the methodology for developing thoracic injury risk functions for human body models in frontal impacts and drives towards consistent injury risk prediction even across different models.
Methods of Study: The framework for developing thoracic injury risk functions for frontal impacts was developed using the THUMS v4.1 M50 model and verified by application to an alternate model, namely the GHBMC v6.0 M50. Simulations targeting test conditions used in past tests with post-mortem human surrogates (PMHS) were performed with the HBMs and further regression and optimization analyses were performed to relate thoracic measures from HBM to injuries observed in matched PMHS tests. Simulations were performed in nineteen frontal-impact loading modes derived from the literature, including hub impact tests, bar impact tests, and table-top tests with belt loading as well as sled cases. The model input conditions were adjusted based on the input conditions used for each specific PMHS test, resulting in a simulation for each individual PMHS test. In all, approximately 176 individual simulations were performed, distributed across the nineteen loading modes. Various deflection and strain-based outputs from these simulations were examined to determine which measures (or combination of measures) best predicted the rib fracture injuries observed in the PMHS tests. Further, both the models were exercised in a realistic vehicle environment subjected to 56 kph frontal collision. The subsequent injury risk prediction from the models was compared to field data describing risk from comparable collisions.
Results of Study: The study resulted in the formation of a framework and detailed guidelines on setting up simulations and developing thoracic injury risk functions for HBMs in frontal impacts. The framework was demonstrated using the THUMS v4.1 model and verified using the GHBMC v6.0 model. A total of 176 unique matched pair simulations (55 impactor cases, 115 table-top cases and 6 sled cases) from 19 different loadcases were performed for THUMS and 170 (excluding the sled cases) for GHBMC. Two types of injury risk functions, deflection based and rib strain based were developed. For the strain based IRF, the underlying rib fracture risk curve of Larsson et al. 2021 was calibrated to be used with the specific HBM in the probabilistic rib fracture risk prediction framework developed by Forman et al. 2012. The collision simulations performed in a realistic vehicle environment, showed that using the tuned strain-based chest IRFs, the THUMS and GHBMC both predicted rib fracture risks that were generally consistent with the field data for 56 km/h frontal collisions.
Impact of Thesis: The major contribution of this thesis is a set of guidelines and framework which can be used to develop thoracic injury risk functions for human body models from a consistent dataset and thus drive towards consistent injury risk prediction across various human body models. This thesis is also the first study to tune or calibrate the underlying rib fracture risk curve in the probabilistic rib fracture risk prediction framework of Forman et al. 2012 to be used with a specific HBM. The methodology developed for calibrating the fracture risk curve can be used to develop risk curves for any model. This removes the need for modifying the HBM ribcage (geometry, mesh or material properties) to get the best results in injury prediction with the strain based probabilistic rib fracture risk prediction framework.