Objective: Previous studies have reported disparity in injuries
between male and female drivers in the risk of certain types of injuries in
frontal crashes that may be due to a myriad of sex-related differences,
including body size, shape, anatomy, or sitting posture. The objectives of this
study are 1) to use mesh-morphing methods to generate a diverse set of human
body models (HBMs) representing a wide range of body sizes and shapes for both
sexes, 2) conduct population-based frontal crash simulations, and 3) explore
adaptive restraint design strategies that may lead to enhanced safety for the
whole population while mitigating potential differences in injury risks between
male and female drivers
Method: A total of 200 HBMs with a wide range of body sizes and
shapes were generated by morphing the THUMS v4.1 midsize male model into
geometries predicted by the statistical human geometry models. Ten male and ten
female HBMs were selected for population-based simulations. An existing
automated simulation framework was leveraged to rapidly set up crash simulations
with the morphed HBMs and previously-validated driver compartment and restraint
models. A total of 1,000 frontal crash simulations were performed under varied
restraint designs and crash severities. A surrogate model was developed based on
the simulation data using a Gaussian Process (GP) method. Two design
optimization schemes were used to flexibly adjust design parameters based on
subject variables to minimize population injury risks while minimizing
differences in injury risk between male and female HBMs.
Key Results: The simulations indicated that the joint injury
probability (Pjoint) is more sensitive to the seatbelt and driver airbag
variables at 35 mph, while the variability is greatly reduced at 25 mph for all
design variables. The optimal adaptive design strategy from these models
suggested a higher seat belt load limit, higher airbag inflation pressure,
smaller airbag venting, and higher steering column force for occupants with
higher body mass index (BMI). The adaptive design reduced the population Pjoint
by 19.6%, 31.8% and 38.8% from the baseline design when Delta-V equals to 25, 30
and 35 mph, respectively. For high speed crashes (Delta-V = 35 mph), the
proposed adaptive design reduced the average Pjoint differences between men and
women from 24.02% to 2.84% compared to the baseline design. Surprisingly, a
restraint strategy constrained to sex-based balance is able to maintain similar
injury risks between male and female drivers.
Major Conclusion: This study is the first to integrate finite
element crash simulations with adaptive restraint design optimization to
potentially reduce population injury risks and safety balance between male and
female occupants. Gaussian process was shown to be an effective surrogate to FE
simulations.