SMALL ELECTRIC SIGNALS CAN PRODUCE SIGNIFICANT STRESS/STRAIN IN ARTICULAR CARTILAGE
+ Li, LePing and Herzog, W.
Human Performance Lab, Faculty of Kinesiology, University of Calgary, Calgary, Alberta, Canada
Introduction
During cartilage loading, an electric potential is produced by the mobile
ions induced by the fluid pressure gradients. On the other hand,
applying electric potential/current to the tissue can produce stress and
deformation in the cartilage. This electromechanical interaction is
believed to play a significant role in articular cartilage biology and
pathology [1,2]. The objective of this study was to determine the
significance of electromechanical interactions, or to determine when a
coupled electromechanical model is required for analyzing the
mechanical behavior of cartilage. Compression and indentation tests
with different electrical boundary conditions, including prescribed
potential and current, were investigated using the finite element method.
Methods
The streaming potential φ is coupled with the fluid pressure pf by [1]
v f = −k11 ∇pf + k12∇φ , J = k 21∇ pf − k22 ∇φ (1)
f application of the potential (not shown), which is indicated by the
where v is the fluid velocity relative to the solid, and J is the current
density. This equation was incorporated into a fibril reinforced
poroelastic model [3,4] that has been found particularly suitable for
describing such fluid pressure-dependent behavior. A user-defined finite
element was implemented into ABAQUS using FORTRAN,
superimposing the standard porous element. ABAQUS uses Darcy’s
law for describing the fluid flow for low speeds, which for the simplest
case, takes the form v f = − k11∇pf . This equation had to be replaced by
Eq. (1). In the user element formulation, Eq. (1) was discretized in the
same way as Darcy’s law is discretized in ABAQUS, while including
the electrical variables. When calculating the contribution of each
individual element to the algebraic global governing equations, k11 was
assumed to be zero, since its contribution is already included in the
standard porous element. Therefore, in the sense of numerical
equations, Darcy’s law has been identically replaced by Eq. (1).
The calculation of collagen stiffness, which was previously done
through spring elements, was included in the user element. The
permeability (k11) variation on deformation was not considered due to
lack of information on the change of electrokinetic coupling (k12, k21).
For simplicity, the subchondral bone was ignored in the simulations.
For the purpose of comparison with measured data, the radius R and
thickness h of the cartilage disks were taken as 1.50 and 1.07 mm
respectively. These were the average dimensions of the bovine cartilage
explants (N=12) that were used in unconfined compression tests [2].
The finite element mesh adopted was 18×12. The Young’s modulus and
Poisson’s ratio of the proteoglycan matrix were 0.26MPa and 0.36
respectively; the tensile modulus of the collagen network was
(2000×FibrilStrain+3)MPa; k11=0.006mm4/Ns, k21=k12=−10−5mm⋅s/N,
k22=10 −6s/(mm⋅mV), and the void ratio of cartilage was 3.5. k21 and k22
were taken from [5], while k11 was chosen to fit the data.
Fig.1 Streaming potential during Fig.2 Streaming potential
ramp compression distribution along
at r = 0.8R the radius
.8 1.5
∆t = 10
Potential (mV)
1
.4
∆t = 5
0.5
∆t = 1
.0 0 supported by NSERC (Canada). We thank Garon et al. [2] for providing
0 10 20 0 0.5 1 1.5
Time ∆t (sec) Radius r (mm)
Results and Discussion
Two examples of unconfined compression are presented. The first
example (Figs.1 & 2) was aimed at testing the model. A sequence of 5-
step ramp compression/relaxation cycles was simulated. 20 microns
49th Annual Meeting of the Orthopaedic Research Society
Poster #0652
compression was applied over 10s in each step, followed by a relaxation
period of 300s. The measured potential [2] (shown with symbols; N=12)
at a given location on the articular surface is shifted relative to the
predicted potential (Fig.1; ∆t is the time increment for the last ramp). It
is not so clear whether the measured potential was zero at ∆t=0 as
required by equilibrium (zero pressure gradient). However, the observed
shifting is reasonable if indeed the experimental potential was not zero at
∆t=0, due to either a nonzero pressure gradient (not in equilibrium as
expected) or a very small external electric signal (imposed by the
electrodes used in measurement).
In the second example (theoretical only), a potential of 100mV was
applied at the articular surface in reference to the bottom of the cartilage
disk for 10s, when the axial displacement at the surface and bottom was
constrained to zero. The reaction force resultant did not decay to zero
for 140s (not shown) after the potential had been withdrawn. The
reaction force was not always compressive at all locations during
negative fluid pressure (Fig.3, drawn for the plane at half depth of the
disk). The upper part of the disk was in tension during early
deformation (Fig.4, drawn for the disk axis; that z=0 refers to the
articular surface). The fluid pressure at the disk center (r=0) was lower
than that elsewhere within 2/3 of the radius for t=5 and 10s (Fig.3),
which was not the case for ramp compression (not shown). It is
expected that the pressure and potential would be more uniform in the
radial direction if electric and mechanical loadings were applied
simultaneously. This may explain the difference in the potential
distributions obtained by measurement and computation (Fig.2, ∆t=5 &
10; more uniform in measurement), assuming that there was a small
electric signal applied to the tissue externally.
Fig.3 Pressure along the radius Fig.4 Strain along the axis
produced by produced by
electric potential electric potential
30 0
Cartilage Depth (z/h)
Fluid Pressure (kPa)
t = 10
20
t = 20 at time
10 0.5
5
10
0 t=5 20
1
-10
0 0.5 1 1.5 -6 -3 0 3 6
Radius r (mm) Axial Strain (%)
In conclusion, the stress and strain produced by electric loading of
articular cartilage can be very different from that produced by
mechanical loading. Small electric signals might produce significant
stress/strain (Figs. 3 & 4). It appears that a coupled electromechanical
model must be used in case of nonzero electrical boundary conditions,
even if the signals may be very small. On the other hand, if all
nonhomogeneous boundary conditions are mechanical/geometrical, the
electric potential typically follows the same pattern as that of the fluid
pressure (Figs. 1 & 2). Then a mechanical model may suffice (for
homogeneous material) for determining the stress/strain; and the
potential can be calculated independently from the pressure results. The
electromechanical coupling is not significant in this case.
Acknowledgments: This work was initiated at Biosyntech Inc. and
us their electronic data.
References: [1] Frank, E.H. and Grodzinsky, A.J., J. Biomech. 20, 629,
1987. [2] Garon M. et al., J. Biomech. 35, 207, 2002. [3] Li, L.P. et al.,
Clinical Biomech. 14, 673, 1999. [4] Li, L.P. et al., Biorheology 39, 89,
2002. [5] Sachs, J.R. and Grodzinsky, A.J., J. Biomech. 28, 963, 1995.