Toward Aaptive Restraint System:a Simulation And Injury Analysis Study With Consideration Of Population Variation

Among vehicle occupants involved in crashes,vulnerable occupants,like obese,small female,and older,are at increased risk of death and serious injury compared with midsize,young,and male occupants.This social issue has been more and more remarkable with the increase of aging and obese populations.The midsize male(socalled 50th-percentile male)dummy is generally used as the design target when designing and optimizing occupant restraint systems.There are some obvious research limitations,such as how to evaluate the mechanical responses of these vulnerable occupants in the crash accurately,how to improve the current restraint system to broaden the protection efficiency for various occupants,etc.Adaptive restraint system(ARS)aims to enhance the protection efficiency by adjusting design parameters adaptively based on occupant characteristics and crash features.However,the research on ARS is still in conceptual level.One major limitation is lacking proper research tool which could present population variability.Crash test dummies or human body models which are widely used in the injury biomechanical field couldn't meet the ARS design requirement.Over the past decade,the University of Michigan Transportation Research Institute(UMTRI)has proposed the idea of parametric human body modeling to generation various human models which could cover the diversity of bone and body shape among population.However,it was impossible to generate various human models rapidly and automatically using data and method at that point.Moreover,the possibility to perform restraint system optimizing based on these parametric human models was severely restricted due to the complex baseline human model accompanying with low computational efficiency.Therefore,to expand the ability of parametric human models in presenting human diversity;to make a comprehensive study on injury mechanism for different occupants;to investigate the potential to optimize the restraint system and develop ARS using parametric human models,following methods and investigation were presented in this thesis.An automated and rapid human modeling method to morph a baseline finite element(FE)human model into different geometry targets was developed.A total of 100 human models with a wide range of human attributes were generated.A systematic evaluation for parametric FE models was performed based on cadaver tests and vehicle impact tests.,An investigation on human impact responses and injury analysis was then conducted using the large sample of FE human models.Finally,restraint system optimization designs for several parametric FE human models were explored preliminarily.The detail focus and innovation of this research were described in following:1.An automated and rapid human modeling method to generate FE human models was developed according to characteristics of new baseline model.It started with predicting the skeleton(ribcage,pelvic,femur and tibia)and external body shape geometries for target occupants using the statistical models developed previously,followed by assembling skeleton and external body shape using featured bony landmarks,and ended with morphing a baseline mid-size male FE human model into the target geometries body region by body region.In this study,Uniform Latin Hypercube Sampling(ULHS)method was used to select 50 men and 50 women from a set of anthropometry data.Human attributes(age,sex,BMI and stature)for these sampled 100 subjects were then used to generate 100 FE models accounting for a large skeleton and body shape variety both for men and women rapidly.The morphed models demonstrated reasonable mesh quality as the baseline model.2.A comprehensive biofidelity evaluation was performed on the morphed midsize male FE model based on a set of regional blunt impact tests and sled tests in literatures.Results showed that the morphed model maintained reasonable biofidelity with the baseline model.A series of chest impact simulations using morphed models were conducted to test the calculation robustness,and the effects of human attributes on mechanical responses were investigated as well.Statistical analysis showed that the responses from the three dummy-size models(small female,midsize male,large male models)do not cover the range/variation of the impact force and chest deflection from the sampled human models,reinforcing the need to consider population variation in estimating the occupant impact responses.3.US-NCAP frontal crash simulations and injury risk analysis were conducted with the 100 morphed models.A crash simulation and injury risk analysis method suitable for a large sample of FE human models was proposed taking a US-NCAP frontal crash case as example.The morphed human model was positioned as a driver according to a driving posture model developed previously at UMTRI.This method has strong repeatability and consistency,and it resolved many critical scientific problems in crash simulations,such as human model positioning,rapid pre-simulation setting and injury risk assessment.Results showed that human attributes exhibited significant effects on occupant kinematics and injury risks;a very strong correlation was observed between BMI and knee-thigh-hip(KTH)injury risk;occupants who have similar body size with the midsize male dummy tended to sustain lower injury risks than other occupants.4.Design Optimizations of occupant restraint system based on parametric FE human models were performed.Three different FE human models including small female,short obese female and tall obese male were selected as design targets according to injury distribution features found in previous study.Six major design parameters of restraint system were set as design variables,optimization design for these three subjects was conducted by developing response surface models separately.Results showed that the injury risk of individual subject reduced significantly by adjusting restrain system design parameters.When assigning the optimal parameters based on short obese female model to the large sample of crash simulations,it turns out that the protection efficiency for short female increased obviously,however the protection efficiency for other groups decreased.All these findings illustrate that ARS could act as an alternative to provide a better protection for entire population.The research work in this thesis provides meaningful references to investigate the effects of human variability on injury risks.It also provides a feasible technical route for large-scale crash simulations and injury analysis.The large sample of morphed parametric human models could act as indispensable tools to develop ARS.The research work is of importance in the field of improving the protection for various of occupants.