World Conference on Computational Mechanics edition:11 location:Barcelona date:20-25 July 2014 European Conference on Computational Mechanics edition:5 location:Barcelona date:20-25 July 2014 European conference on Computational Fluid Dynamics edition:6 location:Barcelona date:20-25 July 2014
Head injury is recognized as the leading cause of mortality and morbidity in children and
young adults [1, 2]. 15% of deaths of children aged between 1 and 15 years are caused by
head injuries . It is the most frequent cause of death in the paediatric population, up
to 80% of all trauma related deaths each year . These surveys stimulate the need for
research to determine the cause and to develop improved protective headgear.
Over the past decades, mechanics and characteristics of skull fracture have been studied
[5, 6, 7]. A skull fracture occurs when contact load exceeds a certain threshold of the skull.
Computational modeling of head impacts is an ecient tool to study subject specic
variations and the mechanisms of skull fracture. Yoganandan et al. postulate that the
energy absorbed by the skull before it fractures likely represents the best predictor for
skull fracture, by incorporating impact as well as structural characteristics . Sahoo et
al. uses the strain energy of the skull bone to predict the skull fracture .
By encompassing impact information as well as structural characteristics, the energy
criterion has the potential to accurately predict skull fracture . Previous results at
our research group investigated the existence of an energy failure criterion with a double
pendulum set-up in a series of intact cadaver heads [9, 10]. For frontal impacts the
existence of an energy failure level of 22-24J is suggested , for temporal dynamic
loading conditions 5-15J . Monea et al. concluded that energy criteria for impacts are
location dependent. The interest of this study is to gain insight in the local absorbed
energy of an impact by the skull.
Materials and methods
CT-scans of each subject are obtained before and after the impact experiment and seg-
mented in MimicsTM (Materialise, Leuven, Belgium) to create a mask of the skull. A
3D object is calculated from the mask and exported to 3-MaticTM (Materialise, Leuven,
Belgium) where the mean thickness of the impact site is measured. To calculate the local
bone density, Hounseld Units at the impact site were compared to Hounseld Units of
the European Spine Phantom with known bone mineral densities. A statistical relation-
ship between the absorbed energy and the thickness and density of the skull at the impact
site are investigated.
Secondly, a simplied model of the experiments is developed in Abaqus (SIMULIA) con-
sisting of an impactor with an identical geometry as the one used in the experiments and
a hollow sphere, resembling the skull. To investigate the in
uence of the skull thickness
at impact site, a variable wall thickness of the hollow sphere is used. The strain energy
density at the site of impact in the simplied model can be compared with the average
absorbed energy per volume of the impact site. Volume of the impact site is calculated as
the thickness of the skull at impact site multiplied with the circular area of the cylindrical
impactor that hits the skull during the experiments.
Spearman correlation coecients show that there is a statistically signicant positive re-
lationship between the absorbed energy and the local density of the skull at the impact
site, the coecient is 0.48. While the Spearman correlation coecient between the local
skull thickness at the impact site and the absorbed energy is 0.39, it is not statistically sig-
nicant. There exists a statistically signicant positive relationship between the fracture
force and the local skull thickness, Spearman correlation coecient = 0.56 .
Results of the Abaqus (SIMULIA) simulation for an impact velocity of 5 m=s give a
maximum strain energy density ranging from 0.15 mJ=mm3 for a wall thickness of 6 mm
to 1.19 mJ=mm3 for a wall thickness of 3 mm. The absorbed energy per volume of the
impact site for impacts of an impact velocity of 5 m=s ranges from 0.84 mJ=mm3 to 2.22
mJ=mm3. These results correlate well to the experimentally obtained values.
Discussion and future work
The statistical analysis combined with the results of the simplied nite element model
indicate the dependence of the absorbed energy on local geometrical features and the local
bone density. Future work will include the development of a subject specic nite element
model of the experimental set-up in Abaqus (SIMULIA) to investigate the in
the local geometrical characteristics on the energy criterion. Experimental results will
be compared to the outcome of the nite element head model of the subject, ultimately
leading to a fracture criterion based on strain energy density.