Element Selection in Finite Element Analysis for Simulations of the Foot and Footwear

1
Tadepalli, S C; 2Erdemir, A; +1Cavanagh, P R
+1University of Washington, Seattle, WA, 2 CoBi Core & Dept. of Biomedical Engineering, Cleveland Clinic, Cleveland, OH
cavanagh@u.washington.edu

INTRODUCTION:
Accurate prediction of stresses at the foot-ground or foot-shoe
interface can provide insight into the biomechanics of foot pathology
especially in people with diabetes and peripheral neuropathy [1]. Finite
Element (FE) models of the foot comprised entirely of tetrahedral or
hexahedral elements have been previously developed [2-8]. Our
previous work has demonstrated that poor element selection may hinder
accurate representation of normal and shear stresses under the foot [9].
This is particularly true in the setting of nonlinear tissue response, tissue
incompressibility, contact with friction, and the use of conventional
linear tetrahedral elements. Hexahedral elements are generally preferred
over tetrahedral elements because of their superior performance in terms
of accuracy and convergence [10]. However, while tetrahedral meshing
can be easily automated, hexahedral meshing commonly requires user
intervention and is sometimes unrealistic due to the geometric
complexity of the foot. Recent advances in finite element formulations
have led to the development of higher order enhanced tetrahedral
elements which may serve as adequate substitute for hexahedral
elements especially under large deformations. Hence the objective of the
present study is to extend our previous work [9,11] by evaluating
various types of meshes with different element types that can be used to
model interaction of a bone-soft tissue construct in contact with a rigid
floor and deformable footwear materials under physiologically realistic
compressive and shear loading conditions.
were 0.66 and 2.45 times that of hexahedral elements in barefoot
simulations and 0.7 and 2.2 times that of hexahedral elements in
footwear simulations. All models using quadratic tetrahedral elements
(C3D10MH) had numerical difficulties in deformable-deformable
contact conditions that hindered convergence to a solution.
DISCUSSION:
Our results show that for foot and footwear simulations FE models
using enhanced quadratic tetrahedral elements can perform well under
both compression and combined shear and compressive loading. Linear
tetrahedral elements do not provide a satisfactory simulation compared
to hexahedral or enhanced quadratic tetrahedral elements. However, the
average computational time required by a model using enhanced
quadratic tetrahedral elements is much higher compared to that of an
equivalent hexahedral mesh [11]. Given the large amounts of time
required for meshing complex anatomical structures using hexahedral
elements, the use of enhanced quadratic tetrahedral elements (C3D10I)
appears to be a feasible alternative even in the setting of large-
deformation, deformable-deformable contact, nonlinear, and
incompressible tissue properties. It should be noted that other FE
solvers, or different versions of the same solver, may have varying
formulations to accommodate material and geometric nonlinearities and
the results of a test problem similar to that presented here should be
explored on a solver by solver basis.

METHODS:
A previously developed FE model of a bone-soft tissue construct
representative of heelpad geometry was used [9, 11]. All simulations
were performed using ABAQUS 6.10. The model consisted of bone, soft
tissue, floor, and insole components. The bone and floor were modeled
as rigid bodies while soft tissue was modeled as an incompressible
hyperelastic Ogden material [4]. The insole, representative of Microcel
Puff, a commonly used footwear material, was modeled as a nonlinear
hyperfoam [5]. The bone and floor were meshed using 2D rigid shell
elements (triangular and quadrilateral); the insole was meshed with
hexahedral elements (C3D8H) and the soft tissue was meshed using one
of the following elements, all of which accommodated
incompressibility: hexahedral (C3D8H), linear tetrahedral (C3D4H),
quadratic tetrahedral (C3D10MH) and enhanced quadratic tetrahedral
(C3D10I). A mesh convergence study ensured the appropriate and
equivalent element sizes across models to ensure a meaningful
comparison. Tied contact was defined between bone and soft tissue to
prevent any relative motion. Frictional contact (μ=0.3) was defined
between the soft tissue and insole as well as between insole and rigid
floor (or tissue and floor in barefoot simulations). The floor was
completely fixed in all degrees of freedom while bone and soft tissue
were allowed to move in the vertical direction (the direction of the
applied load) and also horizontally along the direction of the applied
shear force. Compression only loads of 300 N simulated standing; to Figure 1: Contact pressure distributions simulating barefoot (left) and a shod
simulate nearly heel strike like condition, a compressive load of 700N conditions (right) predicted via hexahedral elements (a, b); linear tetrahedral
was applied to the bone in addition to a shear force of 100N. Simulations elements (c, d); enhanced quadratic tetrahedral elements (e, f). Panels 1a-d are
compression-only loading. Panels 1e and 1f are combined compression and shear
were conducted to assess the influence of element type on the contact loading. Note the patches of elevated predicted pressures in simulations using
pressure and contact shear stress predictions (between the soft tissue and linear tetrahedral elements under compression loading.
floor or insole) and on solution time. The simulations were performed on REFERENCES
a 16 processor computer with 64 GB RAM [12]. 1. Owings TM et al. Diabet Med. 2009; 26(11):1141-1146; 2. Cheung
JTM et al. J Biomech. 2005; 38(5):1045-1054; 3. Erdemir A et al. J
RESULTS: Biomech. 2005; 38(9):1798-1806; 4. Erdemir A et al. J Biomech. 2006;
Figure 1 illustrates exemplar contact pressure distributions predicted 39(7):1279-1286; 5. Goske S et al. J Biomech. 2006; 39(13):2363-2370;
by models with different loading conditions and with different element 6. Budhabhatti SP et al. J Biomech Eng. 2007; 129(5):750-756. 7. Chen
types simulating the stress within the soft tissue. During compression- WP et al. Clin Biomech. 2003; 18(6):s17-24; 8. Gefen A. Med Eng
only loading on both rigid and deformable surfaces, linear tetrahedral Phys. 2003; 25(6):491-499; 9. Tadepalli SC, et al. ASME SBC, 2010;
elements generated extremely noisy pressure distributions which 10. Benzley S, et al. Proceedings of the Fourth IMR, 1995; 11. Tadepalli
overestimated peak pressure compared to the hexahedral elements (Fig SC et al. ASB, 2010; 12. ABAQUS 6.10 [http://www.simulia.com]
1a and b) because of many isolated patches with predicted high
pressures (Figures 1c and d). However, the enhanced quadratic ACKNOWLEDGEMENTS:
tetrahedral elements performed well under both compressive and This study was supported by NIH 7 R01 HD037433-08. The authors
combined compressive and shear loading (Fig 1e and f). The simulation wish to thank Subham Sett of SIMULIA for providing assistance with
times for the model comprising linear and quadratic tetrahedral elements the finite element analysis.

Poster No. 1391 • ORS 2011 Annual Meeting