Available online at www.sciencedirect.

com

Clinical Biomechanics 23 (2008) 11481157

www.elsevier.com/locate/clinbiomech

The eect of bearing congruency, thickness and alignment

on the stresses in unicompartmental knee replacements
D.J. Simpson a, H. Gray a, D. DLima b, D.W. Murray a,c, H.S. Gill a,*
a

Nueld Department of Orthopaedic Surgery, University of Oxford, Nueld Orthopaedic Centre, Windmill Road, Oxford OX3 7LD, UK
b
Orthopaedic Research Laboratories, Shiley Centre for Orthopaedic Research and Education at Scripps Clinic,
11025 N Torrey Pines Road, 140 La Jolla, CA 92037, USA
c
Nueld Orthopaedic Centre, Oxford, OX3 7LD, UK
Received 8 February 2008; accepted 4 June 2008

Abstract
Background. Unicompartmental knee replacement oers an eective treatment for patients with single compartment knee disease and
is becoming an increasingly popular alternative to total knee replacement. An important cause of failure in a unicompartmental knee
replacement implant is polyethylene wear. Signicant contributory factors to the amount of polyethylene wear are contact stress, bearing
alignment, congruency and thickness.
Methods. Four dierent unicompartmental knee replacement implant designs (Fully-Congruent; Partially-Congruent; Non-Congruent-metal-backed; Non-Congruent-all-polyethylene) were inserted into a validated nite element model of a proximal tibia. The eect
that bearing congruency, alignment and thickness had on the polyethylene stresses during a simulated step-up activity for each design
was investigated. Additionally, contact pressures were compared to those calculated from Hertz elastic theory.
Findings. Only the Fully-Congruent bearing experienced peak von Mises and contact stresses below the lower fatigue limit for polyethylene during the step-up activity. The highest polyethylene contact stresses were observed for the Partially-Congruent and Non-Congruent-metal-backed designs, which experienced approximately three times the polyethylene lower fatigue limit. Increasing the bearing
thickness from 3.5 mm to 8.5 mm of the Non-Congruent design decreased the contact stresses in the bearing; however they did not fall
below the lower fatigue limit for polyethylene. Good agreement between nite element and Hertz contact pressures was found.
Interpretation. Fully congruent unicompartmental knee replacement bearings can be markedly thinner without approaching the material failure limit, have a greater potential to preserve bone stock and are less likely to fail mechanically.
2008 Elsevier Ltd. All rights reserved.
Keywords: Unicompartmental; Knee; Finite element; Contact stress; Polyethylene; Bearing; Step up

1. Introduction
Unicompartmental knee replacement (UKR) is becoming an increasingly popular alternative to total knee
replacement (TKR) because of its improved functional outcome, favourable long term clinical results and the benets
of minimally invasive surgical techniques (Gioe et al., 2003;
Berger et al., 2005). In particular, UKR oers a more eective solution than TKR for more active patients with single

Corresponding author.
E-mail address: richie.gill@ndos.ox.ac.uk (H.S. Gill).

0268-0033/$ - see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.clinbiomech.2008.06.001

compartment knee disease, because the mechanics of the

knee are better preserved, and more functional anatomy
is maintained (Goodfellow et al., 2006).
One of the most important causes of failure in UKR is
polyethylene wear, (Palmer et al., 1998; Ashraf et al.,
2004) which can disrupt the surface geometry of the
replaced plateau, altering joint stability and alignment
(Hernigou and Deschamps, 2004; Wright, 2005). A signicant contributory factor to polyethylene wear is the contact
stress on the polyethylene articulating surface, which in
turn is aected by bearing congruency, polyethylene thickness and contact area (Bartel et al., 1986; Engh et al., 1992;
Kuster et al., 2000).

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

Considerable dierences exist in contemporary UKR

design. Metal backed tibial components were introduced
to address the issues of polyethylene wear and subsidence
(Deshmukh and Scott, 2001). However, the use of metalbacked tibial components may mean that a thinner polyethylene insert will be required, or that more bone must
be resected.
A consequence of using thinner bearings is that higher
stresses can be generated, (Bartel et al., 1986, 1995).
Indeed, several retrieval studies have shown bearing failures related to polyethylene wear, where the original tibial
insertion thickness was less than 6 mm (Engh et al., 1992;
McAuley et al., 2001). Maximum tensile stresses in polyethylene bearings have been hypothesised to create stress elds
necessary to propagate cracks and large von Mises stresses
have been implicated in subsurface failures such as delamination (Bartel et al., 1986; Estupinan et al., 1998).
Other UKR designs use an all polyethylene tibial component to maximise the bearing thickness, or a mobile
bearing to maximise the bearing contact area. Importantly,
these UKR designs are also intended to minimise the
amount of bone that is resected, which is of particular
advantage if the UKR needs to be revised to a TKR (Johnson et al., 2007).
Experimental measurements cannot predict the stress
distributions within polyethylene components for UKRs.
This limitation can be overcome by using nite element
(FE) analysis. Previous FE analyses of UKR designs have
had some important limitations. Iesaka et al. (2001) used
FE analysis to investigate tibial component inclination on
bone stress with a simplied two-dimensional (2D) model.
A signicant disadvantage of this work was that no validation was given for the FE model. Bearing congruency in a
single plane has also been investigated with FE analysis
using 2D models (Estupinan et al., 1998; Kuster et al.,
2000; Rawlinson and Bartel, 2001) without any validation.
Other FE studies of knee replacements have used threedimensional (3D) FE models to investigate tibial tray inclination (Sawatari et al., 2005) and the eects of gait on
polyethylene stresses (Morra and Greenwald, 2003). However, all the FE models used in these 3D studies were not
validated.
In addition, the above analyses were carried out for a
small number of load cases representing discrete points of
an activity. It has been shown that the medial lateral load
split on the tibial plateau can vary considerably throughout
a motion cycle (Hurwitz et al., 1998).
The current study examined the inuence that four different UKR design geometries have on polyethylene stresses experienced throughout a functional activity using a
validated FE model. The four designs investigated were
Fully-Congruent, Partially-Congruent, Non-Congruentmetal-backed and Non-Congruent-all-polyethylene UKRs.
These implants were modelled as they represent common
contemporary UKR designs. In clinical practice, a particular implant is used following careful consideration of the
patients condition and requirements as well as the compe-

1149

tence and familiarity that a surgeon has with a particular

implant and associated procedure. Ultimately the choice
is that of the surgeon performing the procedure.
The aims were to improve understanding of the eects
that bearing congruency, thickness and alignment have
on stresses in a UKR bearing, and to give indications of
how the longevity of UKR bearings can be optimised.
2. Methods
The FE model used in this work was based on a validated whole tibia model (Gray et al., 2007). The full details
of this validation have been given elsewhere (Gray, 2007); a
brief description is provided here.
Three-dimensional geometry for the whole tibia was
derived from computed tomography (CT) scans of a cadaveric tibia (male donor, age 60 years, height 178 cm, weight
82 kg, left side). The computer aided design (CAD) geometry for the tibia was reconstructed from the CT dataset
using SliceOmatic software (v4.2 Rev-9b, TomoVision,
Magic Inc., Montreal, Canada). A mesh consisting of 10noded tetrahedral elements was created and the CT data
were used to map orthotropic material properties to individual elements (Rho et al., 1995; Viceconti, 2000; Gray,
2007).
Seventeen triaxial rosettes (Mod. KFG-3-120-D1711L3M2S, Kyowa, Tokyo, Japan) were attached (Viceconti
et al., 1992) to the prepared tibia, and it was then subjected
to nine axial loading conditions, two four-point bending
loading conditions and one torsional loading condition.
Axial loading was repeated after implantation of a medial
Oxford Unicompartmental Knee Replacement (OUKR)
(Biomet UK Ltd, Swindon, UK). The tests were performed
using a materials testing machine (MTS 858 MiniBionix,
MTS Systems Corporation, Minneapolis, Minnesota,
USA).
Measured principal strains were compared to their corresponding FE values using linear regression. The experimental results correlated well with those of the FE
analysis (axial loading; R2 = 0.97, root mean square error
(RMSE) = 8.8%; bending loading R2 = 0.96, RMSE =
9.0%) (Gray 2007).
2.1. Proximal tibia model
A reduced model measuring 75 mm from the tibial plateau was created to represent the proximal end of the tibia.
Orthotropic material properties were assigned based on the
axial distribution of Youngs modulus along the whole
tibia, and relationships established by Rho et al. (1995)
and Gray (2007). Seven material properties were assigned
to the proximal tibia model in the axial, mediolateral and
anteroposterior directions (Table 1). Maximum and minimum principal strain data were compared between the
reduced proximal model and the whole tibia model for
one axial load condition (Gray, 2007). This reduction in
model size had little eect on the principal strains at the

1150

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

Table 1
Orthotropic material properties for the proximal tibia model
Material property

Bone section properties

1

E medial-lateral (X)
(MPa)
E anterio-posterior
(Y) (MPa)
E proximal-distal (Z)
(MPa)
m(XY)
m(YZ)
m(XZ)
G(XY) (MPa)
G(YZ) (MPa)
G(XZ) (MPa)

7600

5564 5459 606

2926 2907 479

7652

5595 5489 610

2943 2923 481

13,263 9698 9514 1056 5100 5067 835

0.427
0.234
0.405
2591
3509
3566

0.427
0.234
0.405
1895
2565
2534

0.427
0.234
0.405
1859
2517
2486

0.427
0.234
0.405
206
279
276

0.427
0.234
0.405
996
1349
1333

0.427
0.234
0.405
990
1340
1324

0.427
0.234
0.405
163
220
218

locations considered. The slope of the linear regression line

comparing the FE and experimental strains was 0.999, the
intercept 0.0001 micro-strain with an R2 value of 1.000.
Mean correlation coecients of 0.989 and 0.976 was
obtained for the maximum and minimum principal strains
respectively, for the axial load case.
The proximal tibia model consisted of 282,188 10-noded
tetrahedral elements.
2.2. UKR geometry and mesh
Four UKRs were modelled:
1. Fully-Congruent a spherical femoral component articulating on a spherical, mobile bearing. The mobile bearing articulated on the metal tibial component, which was
bonded to the tibia.
2. Partially-Congruent a poly-radial femoral component
articulating on a xed concave bearing. The bearing was
xed to the metal tibial component, which was bonded
to the tibia. The conformity ratio in the sagittal plane
was 0.05 (Kuster et al., 2000).
3. Non-Congruent-metal-backed a spherical femoral
component articulating on a at bearing. The bearing
was xed to the metal tibial component, which was
bonded to the tibia.
4. Non-Congruent-all-polyethylene a spherical femoral
component articulating on a at, all-polyethylene tibial
component. The bearing (tibial component) was bonded
to the tibia.
In the above denitions the term bonded is used to
describe the kinematic constraint between two bodies
whereby no sliding is permitted between the two surfaces
(l = 1), but separation is allowed.
These models will be hereafter respectively referred to as
1.
2.
3.
4.

Fully-Congruent.
Partially-Congruent.
Non-Congruent-MB.
Non-Congruent-AP.

The same cementing technique used for the validated

implanted whole tibia model (Gray, 2007) was modelled
for each UKR. To minimise the dierences between each
FE model, the same surgical procedure was simulated for
creating each UKR FE model, (Goodfellow et al., 2006).
Each UKR was implanted in a neutral position; the femoral component, bearing and tibial component were all
aligned with the long axis of the tibia.
The same tibial tray was used for the Fully-Congruent,
Partially-Congruent and the Non-Congruent-MB UKR
models (Biomet UK Ltd, Swindon, UK). The Non-Congruent-MB and Non-Congruent-AP UKR models used
the same bearing transverse cross-sectional geometry.
Geometry from CAD models was used to construct the
FE models of the Fully-Congruent UKR (Biomet UK,
Ltd, Swindon, UK). The geometry for the Partially-Congruent UKR (Zimmer Inc, Warsaw, USA), was obtained
from 3D scanning of the components (THE SCAN TEAM,
Hertfordshire, UK). The scanned component models were
decimated into 40,000 faces and the binary stereolithography les were rendered and smoothed to accurately represent the surface topology (Solidworks 2007, Solidworks
Corporation, Concord, MA). For the Non-CongruentMB and Non-Congruent-AP UKRs, a hypothetical at
bearing was modelled with the same spherical femoral component as the Fully-Congruent UKR.
The bearing in the Fully-Congruent model had a minimum thickness of 3.5 mm, and the bearing in the Partially-Congruent model had a minimum thickness of
4.5 mm. The bearing in the Non-Congruent-MB model
had a minimum thickness of 3.5 mm, comparable to the
bearing in the Fully-Congruent model. The bearing in the
Non-Congruent-AP model had a thickness of 15.0 mm.
This was comparable to the combined thickness of the
other FE models cement layer, tibial tray and bearing; this
is also the maximum thickness used in the all-polyethylene
UKR, the St. Georg Sled (Waldemar Link, Hamburg, Germany) (Gleeson et al., 2004).
The tibial tray and femoral component in all the FE
models was modelled as linear elastic and isotropic with
the material properties for cast cobalt-chromium (E =
195,000 MPa (Lewis, 1997), m = 0.3 (Cheal et al., 1985)).
The cement was modelled as a linear elastic isotropic material (E = 1940 MPa (Lewis, 1997), m = 0.4 (Orr et al.,
2003)).
Linear material models for polyethylene have been used
in previous studies (Walker, 1988; Sathasivam and Walker,
1994); however the stresses obtained using this approximation are considerably higher than for nonlinear models,
indicating that a linear approximation is not accurate
enough for careful evaluation of polyethylene load
response (Kuster et al., 2000). A multi-linear, kinematic
hardening, elastoplastic material model (Mroz, 1967)
was implemented in ANSYS, v11.0 (Ansys Inc., Canonsburg, PA, USA) for the polyethylene. Experimental data
were entered into the polyethylene material model from
literature (Kurtz et al., 1996), with eight points dening

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

the polyethylene true stressstrain material response. The

Youngs modulus was entered as 1048 MPa.
2.3. Boundary conditions
In vivo kinematic data measured using uoroscopy, during a step-up activity, was used to determine the relative
tibial-femoral positioning for each model; this was determined as a function of knee exion angle. The data were
collected from 10 patients with an OUKR (Pandit et al.,
2008), and used to position the bearing in all the FE models. The uoroscopic measurement methodology has been
described in detail elsewhere (Pandit et al., 2005; van
Duren et al., 2007). The knees exion angles ranged from
20 to 55 for the step up activity, as this corresponded
to the exion range for which force data were available.
The load data were adapted from loads measured
in vivo using an instrumented implant during a step-up
activity (Zhao et al., 2007). This load condition was chosen
as it represents a functional high intensity activity with
large knee exion angles, and it has been shown to correlate better to outcome score parameters than walking
(Morlock et al., 2001). Separate loads were applied to the
medial and lateral compartment and these were scaled to
a subject of mass 82 kg (mass of tibia donor). The load
on the lateral compartment was applied at the nodes in
the vicinity of the central point of the compartment. The
load on the medial compartment was applied to the femoral component, on nodes located on the internal surface
geometry, on a cross-sectional line in the coronal plane.
Force vectors were applied perpendicular to the tibial tray
and the distal end of the proximal tibia model was rigidly
constrained.

1151

2.5. Analyses
For all four UKR FE models, the peak contact, von
Mises and principal stresses in the polyethylene bearing
were evaluated over the step-up activity and compared to
the upper (32 MPa) and lower (17 MPa) limits of polyethylene fatigue failure stress obtained from tensile testing
(Ries et al., 1996). Although this loading regime is dierent
from that experienced in UKR bearings, it was chosen
because the yield properties of polyethylene have been
shown to be very similar in compression and tension
(Kurtz et al., 1997).
The peak stresses in the Fully-Congruent and Non-Congruent-MB bearing were assessed as a function of bearing
thickness, for the maximum load state, which occurred at
35 of knee exion. The thickness of the Non-CongruentMB bearing was varied from 2.5 mm to 8.5 mm, in
1.0 mm increments. The thickness of the Fully-Congruent
bearing was varied from 2.5 mm to 5.5 mm, in 1.0 mm
increments.
Each femoral component was rotated about the sagittal
plane, relative to the tibial tray, (whilst maintaining the
same initial contact condition with the bearing) to represent varusvalgus mal-alignment often seen in clinical practice (Swienckowski and Pennington, 2004; Cool et al.,
2006). The femoral component was rotated up to a maximum of 20 and the stresses in the polyethylene bearing
examined.
To investigate the eect of the stiness parameter, the
contact stiness was varied from 53 to 280 N/mm, and
the results using the Non-Congruent-MB model were
examined.
2.6. Contact validation

2.4. Contact constraints

A Coulomb friction contact model (l = 0.07 (Shen and
Dumbleton, 1974)) was used to simulate the femoral component/bearing contact surfaces for all UKR models. For
the Fully-Congruent UKR model a Coulomb friction contact model was used to simulate the bearing/tibial tray contact surface. The polyethylene was bonded to the tibial tray
for the Partially-Congruent and Non-Congruent-MB models. The Non-Congruent-AP bearing was bonded to the
cement layer. The implant/cement and cement/bone interfaces were modelled as bonded contact for all four UKR
FE models. An Augmented Lagrange multiplier contact
algorithm was implemented in the contact algorithm (Simo
and Laursen, 1992). A contact stiness of 70 kN/mm was
assigned initially within the contact algorithm, and was
updated at each solution sub-step. The aect of varying
this parameter on the model output was investigated.
The FE models were solved using Ansys v11.0, using a
pre-conditioned conjugate iterative solver algorithm. The
analyses were performed on an Intel Dual CoreTM workstation (2.66 GHz, 4GB RAM), running a Suse Linux 9.2
operating system.

The results from the FE contact representation were

compared to those calculated using Hertzs theory of elastic contact (Johnson, 1987). Hertz contact theory is limited
by the assumptions and restrictions concerning an elastic
half-space, and the relative dimensions of the contacting
bodies and their resulting contact region (Johnson, 1987).
Therefore comparisons were made for a sphere-on-plane
model and for the Non-Congruent-MB FE model.
The sphere-on-plane model consisted of a sphere of
radius 20 mm in contact with a rectangular substrate
(dimensions 20  20  50 mm); this FE model had exactly
the same contact parameters as those used for all the UKR
models. This sphere-on-plane model is analogous to the
special case in classical contact mechanics of two cylinders
in contact with their axes perpendicular (Johnson, 1987). A
load of 100 N directed perpendicularly to the substrate surface acted on the sphere such that contact between it and
the substrate occurred. The sphere was modelled as linear
elastic with an elastic modulus of 195,000 MPa and Poissons ratio 0.3 (material properties of cobaltchrome).
The substrate was modelled as linear elastic, with an elastic
modulus of 1048 MPa and a Poissons ratio of 0.3 (material

1152

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

properties of polyethylene). The FE calculated contact

pressure and contact width were compared to those calculated from Hertz theory using the following relationships:

1=3

1=3
3PR
6PE2
and
p

a
0
4E
p3 R2
where a is contact width, P is normal load and p0 is maxi1v2
1v2
mum contact pressure, E1 E1 1 E2 2 (En and vn are the
respective elastic modulus and Poissons ratio of the two
bodies). R1 R11 R12 is the relative curvature (R1 and R2
are principal radius of curvature).
Following this comparison, the peak contact pressure
and contact width in the Non-Congruent-MB UKR bearing were compared to those calculated from Hertz theory
for the peak load condition (knee exion angle = 35).
For this comparison the polyethylene bearing was modelled as linear elastic (E = 1048 MPa).
3. Results
For all models the peak contact stress was observed at
the contact region between the femoral component and
the polyethylene bearing. The magnitude of this contact
stress was similar for the bearings in the Partially-Congruent, Non-Congruent-MB and Non-Congruent-AP models
(mean peak contact stress, 44.3 MPa, 48.6 MPa and
45.9 MPa, respectively). Considerably lower contact stresses were observed in the Fully-Congruent model bearing
(mean peak contact stress, 2.7 MPa). The contact stress
for the Partially-Congruent, Non-Congruent-MB and
Non-Congruent-AP bearing was concentrated in a small
area, and was more widely distributed in the Fully-Congruent bearing (Fig. 1).
Only the Fully-Congruent bearing experienced peak
contact stresses below the polyethylene lower fatigue limit
(17 MPa). All other models had peak contact stresses of
approximately three times the lower fatigue limit (Fig. 2).
For the Non-Congruent-MB, Non-Congruent-AP and
Partially-Congruent UKR models, the area on the bearing
contact surface that experienced contact stresses above the

lower fatigue limit was 28.2 mm2, 36.2 mm2 and

36.85 mm2, respectively. This related to a total contact surface area of 1038.5 mm2, 1038.5 mm2 and 1045.2 mm2,
respectively.
Peak von Mises stresses in the Fully-Congruent bearing
were considerably smaller than in the Partially-Congruent,
Non-Congruent-MB and Non-Congruent-AP bearings,
and were well below the lower fatigue limit of the polyethylene (Fig. 3). Peak von Mises stresses for the PartiallyCongruent, Non-Congruent-MB and Non-Congruent-AP
bearings were above the polyethylene lower fatigue limit
for all exion angles over the step-up activity.
Similar trends were observed for the 3rd principal stresses in the polyethylene bearing. The bearing in the FullyCongruent model experienced a considerably lower peak
compressive stress than the other bearings (Fig. 4a).
The mean peak 1st principal stress observed in the
Fully-Congruent bearing over the step-up activity was
1.9 MPa and this was well below the lower fatigue limit
for polyethylene. For the Non-Congruent-MB and NonCongruent-AP bearings the mean peak 1st principal stress
was 8.3 MPa and 5.5 MPa, respectively. The Partially-Congruent polyethylene bearing experienced a mean peak 1st
principal stress of 7.2 MPa (Fig. 4b).
3.1. Variation in bearing thickness
As the thickness of the Non-Congruent-MB bearing was
increased, the peak von Mises stress decreased (Fig. 5a).
Increasing the bearing thickness from 2.5 mm to 8.5 mm
resulted in the peak von Mises stress decreasing from
26.6 MPa to 20.7 MPa; this lower value was still above
the lower fatigue limit of polyethylene. For a bearing thickness of 3.5 mm, the peak von Mises stress occurred
2.25 mm below the bearing surface, beneath the central
contact point with the femoral component. This distance
increased to 2.83 mm below the surface for a bearing thickness of 8.5 mm.
An increase in the minimum thickness of the Fully-Congruent bearing from 3.5 to 5.5 mm resulted in the peak
contact stress decreasing from 3.21 MPa to 3.18 MPa. A

Fig. 1. Contact stress (MPa) contour plot in each polyethylene bearing for the peak load case: (a) Fully-Congruent, (b) Partially-Congruent, (c) NonCongruent-MB and (d) Non-Congruent-AP.

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

1153

5.00 MPa in the Fully-Congruent bearing. Any further

increase in mal-alignment had no additional consequence
on the peak stresses.
3.3. Contact validation and stiness

Fig. 2. Comparison of peak contact stress in each bearing over the step-up
activity.

The maximum contact pressure and contact width predicted by the sphere-on-plane nite element model was
34.01 MPa and 1.39 mm, respectively. The classical Hertz
theory predicted a peak contact pressure and width of
39.89 MPa and 1.09 mm, respectively. For the Non-Congruent-MB UKR with an elastic bearing the FE model predicted a contact pressure of 85.7 MPa, Hertz theory
predicted 85.2 MPa.
For the UKR model with a non-linear polyethylene
material model, varying the initial contact stiness between
the polyethylene and femoral component from 53 N/mm to
280 N/mm decreased the contact penetration from
0.045 mm to 0.040 mm. The contact stress increased from
45.7 MPa to 47.0 MPa. The ve fold increase in contact
stiness resulted in a 2.8% and 12.5% increase in contact
stress and contact penetration, respectively.
4. Discussion

Fig. 3. Comparison of peak von Mises stress in each bearing over the stepup activity.

decrease in thickness from 3.5 mm to 2.5 mm resulted in an

increase in peak contact stress from 3.21 MPa to 3.40 MPa
(Fig. 5b), and a small bearing deformation.
3.2. Varusvalgus alignment
A varusvalgus mal-alignment of up to 15 between the
femoral component and tibial tray had no eect on the
peak stresses in the Partially-Congruent, Non-CongruentMB and Non-Congruent-AP bearings. A 20 mal-alignment led to an increase in peak contact stress from
49.04 MPa to 52.31 MPa in the Partially-Congruent bearing. A varusvalgus mal-alignment of 10 between the
Fully-Congruent femoral component and tibial tray produced an increase in contact stress from 3.21 MPa to

Unicompartmental knee replacement oers a more eective treatment for patients with isolated single compartment knee disease as more functional anatomy is
preserved. Some contemporary UKR designs maximise
this preserved anatomy by minimising the amount of bone
stock resected. A signicant failure mechanism of UKR
implants is wear of the polyethylene bearings (Palmer
et al., 1998; Ashraf et al., 2004), which is associated with
high contact stress, small bearing thickness and mal-alignment (Bartel et al., 1986, 1995). By applying physiological
kinematic and loading conditions to a validated nite element model of an implanted tibia, it was possible to investigate the eect that bearing congruency, thickness and
alignment had on the stresses in four polyethylene bearing
designs. Furthermore, it was possible to determine which
type of UKR design had the greater potential to maximise
the amount of preserved bone stock following
implantation.
The data obtained from the nite element models in this
paper support the ndings of previous computational studies. Contact and von Mises stresses in the polyethylene
bearing for the Partially-Congruent, Non-Congruent-MB
and Non-Congruent-AP bearings were similar in magnitude to those reported elsewhere (Estupinan et al., 1998;
Kuster et al., 2000; Rawlinson and Bartel, 2001). The bearing in the Fully-Congruent design experienced much lower
contact and von Mises stresses, compared to the other
UKR designs, due to the increased congruency, and the
relationship between congruency and contact stress is well
understood (Bartel et al., 1986). We were also able to compare the results for Non-Congruent-MB with Hertz elastic
theory of contact and found a good agreement.

1154

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

Fig. 4. (a) Comparison of peak 3rd principal stress in each bearing over the step-up activity and (b) comparison of peak 1st principal stress in each bearing
over the step-up activity.

Fig. 5. (a) Variation of peak von Mises stress in the Non-Congruent-MB polyethylene bearing with bearing thickness and (b) variation of stress in the
Fully-Congruent bearing with bearing thickness.

Results obtained from the four UKR nite element

models also support the ndings of previous clinical studies. The Fully-Congruent bearing experienced very low
contact stresses, which indicates that this type of UKR
design will have favourable wear characteristics. There
has been a number of clinical studies conrming that low
wear is observed in fully congruent designs (Psychoyios
et al., 1998; Callaghan, 2001; Price et al., 2005). The considerably higher contact stresses observed in the PartiallyCongruent, Non-Congruent-MB and Non-Congruent-AP
bearings suggest that this type of bearing is more susceptible to excessive polyethylene wear, and this has been
reported previously (Bartley et al., 1994; Palmer et al.,
1998; Callaghan, 2001). However, it must be noted that

Fully-Congruent, mobile bearing designs are susceptible

to dierent failure modes, such as bearing dislocation.
The Fully-Congruent bearing is less likely to fail from
fatigue. The bearing in the Partially-Congruent, Non-Congruent-MB and Non-Congruent-AP designs experienced
peak von Mises stresses above the lower fatigue failure
stress for polyethylene throughout the functional activity
analysed. These bearing designs are therefore more susceptible to material failure, such as delamination. The mechanism for subsurface failure of polyethylene has been
described elsewhere (Bartel et al., 1995; Estupinan et al.,
1998), and retrieval analysis of UKRs has shown fatigue
failure of non-conforming polyethylene components (Kop
and Swarts, 2007).

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

Increasing the thickness of the bearing for the Non-Congruous-MB UKR resulted in a decrease in the peak von
Mises stress. Interestingly, the peak von Mises stress
remained above the polyethylene lower fatigue failure
stress limit (17 MPa) for a bearing thickness above
8.5 mm. In addition, the Non-Congruent-AP bearing,
which had a thickness of 15.0 mm, experienced peak von
Mises stresses above 17 MPa.
There have been several retrieval studies showing bearing failures where the original tibial insertion thickness
was less than 6 mm (Engh et al., 1992; Palmer et al.,
1998; McAuley et al., 2001). This has led to the assertion
by many clinicians that a bearing thickness of above
6.5 mm is safe. The results from the current study suggest
that even for a bearing thickness of 8.5 mm, the peak von
Mises stresses remain above the polyethylene lower fatigue
limit, and this aect is seen with or without metal-backing.
If a Partially-Congruent, Non-Congruent-MB or NonCongruent-AP bearing is used, a minimum bearing thickness of 8.5 mm would be recommended on the basis of
our nite element model results. However, our results suggest that there is no safe limit for the polyethylene thickness in lesser congruent designs. Where dishing occurs,
the peak bearing stresses may be alleviated because a more
congruent articulating area will be generated as a consequence of polyethylene wear. However, the concern will
be that the subsurface material on an un-worn bearing
prior to dishing would have been subjected to stresses
above the fatigue failure stress. This same subsurface material becomes the surface after dishing, and therefore has
an increased likelihood of fatigue failure. In addition, the
process of dishing will give rise to a considerable volume
of polyethylene debris and will lead to the recurrence of
varus which may increase loading.
It must be borne in mind however that bearing choice
will be balanced against the patient cohort. More elderly
and typically less active patients will transfer lower loads
to the bearing, and will subject the bearing to fewer stress
cycles over their lifetime, compared to younger patients.
Therefore the choice of a Non-Congruent UKR design
may be appropriate for a patient cohort that is less active
and elderly, but would not be appropriate for a younger,
more active population. This is probably the reason that
some Non-Congruent UKR designs have an adequate
clinical performance.
Fully-Congruent UKR bearings are preferential when
trying to minimise stresses that may lead to catastrophic
wear and can also be markedly thinner without approaching the material failure limit. A Fully-Congruent UKR
design therefore has a greater potential to preserve bone
stock during implantation. A minimum thickness of
3.5 mm was used for the Fully-Congruent bearing in the
current study. Compared to the minimum safe thickness
of 6.5 mm for a Non-Congruous-MB UKR design, a
potential 3.0 mm of bone stock can be preserved during
implantation. This bone stock saving increases to 8.5 mm
when compared to the Non-Congruent-AP bearing. Fur-

1155

thermore, the peak von Mises stresses in the Fully-Congruent bearing were still well below the lower fatigue limit of
polyethylene when a minimum thickness of 2.5 mm was
modelled.
The modelling process used in vivo measured kinematic
and load data, and great care was taken to model the tibialfemoral positioning for each UKR design. This is important for two reasons: rstly, the stresses in the bearing were
able to be assessed throughout an entire activity; secondly,
the kinematic data determined where the femoral component and polyethylene contacted with each other. If an
analysis does not use kinematic data measured in vivo, then
misleading results can be obtained. Morra and Greenwald
(2003) reported very similar contact and von Mises stresses
in a partially congruent and a fully congruent UKR bearing. In the absence of kinematic data, the partially congruent UKR may have been analysed in its most conforming
position. This would explain the low stresses found in the
partially congruent bearing.
Following implantation with a UKR, there is often
some degree of varusvalgus mal-alignment of the replaced
joint (Swienckowski and Pennington, 2004; Cool et al.,
2006). This will inevitably have an eect on the restored
kinematics of the implanted knee, but the aect of malalignment on bearing stress has not, to the authors knowledge, been reported elsewhere. The current study has
shown that the stress raising aects that were observed
for up to a 20 varusvalgus mal-alignment were extremely
small. This may have important implications for clinical
practice, because the varusvalgus mal-alignment investigated in this study does not increase the likelihood of the
bearing failing mechanically.
The tibia was modelled in this work to more accurately
assess the polyethylene stresses in each UKR design. The
polyethylene material properties used in this paper were
taken from readily available data in the literature. The
material properties of contemporary polyethylene may differ from this data; further study should be performed to
examine these eects.
Only the proximal tibia was modelled, which is a limitation of the current study. This simplication meant that an
articial modelling restraint had to be placed on the distal
tibia, so that the condition of equilibrium was not violated.
Using a proximal tibia model was justied because we were
predominantly interested in what happens in the polyethylene bearing. Using a proximal model had the added advantage of allowing us to use a much more rened mesh
without increasing computational time. The kinematic data
used in this study was obtained from only one of the UKR
designs. In reality each UKR design may have dierent
kinematics, and may have subtly dierent loading. The
loading used in this study was taken from in vivo measured
data, and is somewhat simplied in that the soft tissue
forces are not considered. The load data were captured
using an instrumented TKR, and the normal force vectors
on the medial and lateral plateau were adapted for the
UKR nite element models. As such, no lateral force was

1156

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

applied to the UKR models and this is a simplication of

the in vitro situation. However, this is the rst study to
use internally measured load data for the medial-lateral
split on a UKR model. Most previous load data has been
based on estimates derived from models that are very sensitive to errors (Zhao et al., 2007). The loading conditions
applied in this study therefore provide an insight into the
stresses experienced by dierent polyethylene bearings in
contemporary UKR designs. A high intensity load case
was used in this paper. A more complete investigation
would compare the relative activity level of a patient
cohort. However, the choice of only using a high intensity
load case is justied because bearing design should be
based upon the worst case scenario.
5. Conclusion
The results of this study show that a Fully-Congruent
UKR design has bearing stresses that are an order of magnitude lower than other contemporary UKR designs. Perhaps surprisingly, the contact and von Mises stresses in
the Partially-Congruent and Non-Congruent bearings were
very similar, and above the lower fatigue limit for polyethylene. The Fully-Congruent bearing is therefore less likely
to fail mechanically, and can be considerably thinner without approaching the material limits.
Conict of interest statement
The institute of three of the authors DJS, DWM, HSG
has received research support from commercial parties
related to the subject of this paper; one of these authors
has also received benets for personal and professional
use from commercial parties related to the subject of this
paper. The other two authors have had no commercial
interest related to the subject of this paper.
Acknowledgements
The lead author was funded by the Furlong Research
Foundation during the period this work was conducted,
other funding was provided by the Hip and Knee Research
& Development Fund, Nueld Orthopaedic Centre General Charity. The funding bodies took no role in the writing
of this manuscript.
References
Ashraf, T., Newman, J.H., Desai, V.V., Beard, D., Nevelos, J.E., 2004.
Polyethylene wear in a non-congruous unicompartmental knee
replacement: a retrieval analysis. Knee 11 (3), 177181.
Bartel, D.L., Bicknell, V.L., Wright, T.M., 1986. The eect of conformity,
thickness, and material on stresses in ultra-high molecular weight
components for total joint replacement. J. Bone Joint Surg. Am. 68 (7),
10411051.
Bartel, D.L., Rawlinson, J.J., Burstein, A.H., Ranawat, C.S., Flynn Jr.,
W.F., 1995. Stresses in polyethylene components of contemporary
total knee replacements. Clin. Orthop. Relat. Res. (317), 7682.

Bartley, R.E., Stulberg, S.D., Robb 3rd, W.J., Sweeney, H.J., 1994.
Polyethylene wear in unicompartmental knee arthroplasty. Clin.
Orthop. Relat. Res. (299), 1824.
Berger, R.A., Meneghini, R.M., Jacobs, J.J., Sheinkop, M.B., Della Valle,
C.J., Rosenberg, A.G., Galante, J.O., 2005. Results of unicompartmental knee arthroplasty at a minimum of ten years of follow-up. J.
Bone Joint Surg. Am. 87 (5), 9991006.
Callaghan, J.J., 2001. Mobile-bearing knee replacement: clinical results: a
review of the literature. Clin. Orthop. Relat. Res. (392), 221
225.
Cheal, E.J., Hayes, W.C., Lee, C.H., Snyder, B.D., Miller, J., 1985. Stress
analysis of a condylar knee tibial component: inuence of metaphyseal
shell properties and cement injection depth. J. Orthop. Res. 3 (4), 424
434.
Cool, S., Victor, J., De Baets, T., 2006. Does a minimally invasive
approach aect positioning of components in unicompartmental knee
arthroplasty? Early results with survivorship analysis. Acta Orthop.
Belg. 72 (6), 709715.
Deshmukh, R.V., Scott, R.D., 2001. Unicompartmental knee arthroplasty: long-term results. Clin. Orthop. Relat. Res. (392), 272278.
Engh, G.A., Dwyer, K.A., Hanes, C.K., 1992. Polyethylene wear of metalbacked tibial components in total and unicompartmental knee
prostheses. J. Bone Joint Surg. Br. 74 (1), 917.
Estupinan, J.A., Bartel, D.L., Wright, T.M., 1998. Residual stresses in
ultra-high molecular weight polyethylene loaded cyclically by a rigid
moving indenter in nonconforming geometries. J. Orthop. Res. 16 (1),
8088.
Gioe, T.J., Killeen, K.K., Hoeel, D.P., Bert, J.M., Comfort, T.K.,
Scheltema, K., Mehle, S., Grimm, K., 2003. Analysis of unicompartmental knee arthroplasty in a community-based implant registry. Clin.
Orthop. Relat. Res. (416), 111119.
Gleeson, R.E., Evans, R., Ackroyd, C.E., Webb, J., Newman, J.H., 2004.
Fixed or mobile bearing unicompartmental knee replacement? A
comparative cohort study. Knee 11 (5), 379384.
Goodfellow, J.W., OConnor, J., Dodd, C.A.F., Murray, D.W., 2006.
Unicompartmental Arthroplasty with the Oxford Knee. Oxford
University Press.
Gray, H.A., 2007. Finite Element Analysis of the Human Tibia. D. Phil
Thesis.
Gray, H.A., Zavatsky, A.B., Taddei, F., Cristofolini, L., Gill, H.S., 2007.
Experimental validation of a nite element model of a composite tibia.
Proc. Inst. Mech. Eng. [H] 221 (3), 315324.
Hernigou, P., Deschamps, G., 2004. Alignment inuences wear in the knee
after medial unicompartmental arthroplasty. Clin. Orthop. Relat. Res.
(423), 161165.
Hurwitz, D.E., Sumner, D.R., Andriacchi, T.P., Sugar, D.A., 1998.
Dynamic knee loads during gait predict proximal tibial bone distribution. J. Biomech. 31 (5), 423430.
Iesaka, K., Tsumura, H., Sonoda, H., Sawatari, T., Takasita, M., Torisu,
T., 2001. The eects of tibial component inclination on bone stress
after unicompartmental knee arthroplasty. J. Biomech. 35, 969974.
Johnson, S., Jones, P., Newman, J.H., 2007. The survivorship and results
of total knee replacements converted from unicompartmental knee
replacements. Knee 14 (2), 154157.
Johnson, K.L., 1987. Contact Mechanics. Cambridge University Press.
Kop, A.M., Swarts, E., 2007. Quantication of polyethylene degradation
in mobile bearing knees: a retrieval analysis of the AnteriorPosteriorGlide (APG) and Rotating Platform (RP) Low Contact Stress (LCS)
knee. Acta Orthop. 78 (3), 364370.
Kurtz, S.M., Jewett, C.W., Moalli, J.E., Edidin, A.A., 1997. An elastic
plastic material model for the true stressstrain behavior of ultra-high
molecular weight polyethylene in tension and compression. In:
Advances in Bioengineering BED34. American Society of Mechanical
Engineers, New York.
Kurtz, S.M., Rimnac, C.M., Santner, T.J., Bartel, D.L., 1996. Exponential
model for the tensile true stressstrain behavior of as-irradiated and
oxidatively degraded ultra high molecular weight polyethylene. J.
Orthop. Res. 14 (5), 755761.

D.J. Simpson et al. / Clinical Biomechanics 23 (2008) 11481157

Kuster, M.S., Horz, S., Spalinger, E., Stachowiak, G.W., Gachter, A.,
2000. The eects of conformity and load in total knee replacement.
Clin. Orthop. Relat. Res. (375), 302312.
Lewis, G., 1997. Properties of acrylic bone cement: state of the art review.
J. Biomed. Mater. Res. 38 (2), 155182.
McAuley, J.P., Engh, G.A., Ammeen, D.J., 2001. Revision of failed
unicompartmental knee arthroplasty. Clin. Orthop. Relat. Res. (392),
279282.
Morlock, M., Schneider, E., Bluhm, A., Vollmer, M., Bergmann, G.,
Muller, V., Honl, M., 2001. Duration and frequency of every day
activities in total hip patients. J. Biomech. 34 (7), 873881.
Morra, E.A., Greenwald, A.S., 2003. Eects of walking gait on ultra-high
molecular weight polyethylene damage in unicompartmental knee
systems. A nite element study. J. Bone Joint Surg. Am. 85-A (Suppl.
4), 111114.
Mroz, Z., 1967. On the description of anisotropic work hardening. J.
Mech. Phys. Solids 15, 163175.
Orr, J.F., Dunne, N.J., Quinn, J.C., 2003. Shrinkage stresses in bone
cement. Biomaterials 24 (17), 29332940.
Palmer, S.H., Morrison, P.J., Ross, A.C., 1998. Early catastrophic tibial
component wear after unicompartmental knee arthroplasty. Clin.
Orthop. Relat. Res. (350), 143148.
Pandit, H., Van Duren, B., Gallagher, J., Beard, D.J., Dodd, C.A.F., Gill,
H.S., Murray, D.W., 2008. Combined anterior cruciate reconstruction
and Oxford unicompartmental knee arthroplasty: in vivo kinematics.
The Knee 15 (2), 101106.
Pandit, H., Ward, T., Hollinghurst, D., Beard, D.J., Gill, H.S., Thomas,
N.P., Murray, D.W., 2005. Inuence of surface geometry and the campost mechanism on the kinematics of total knee replacement. J. Bone
Joint Surg. Br. 87 (7), 940945.
Price, A.J., Short, A., Kellett, C., Beard, D., Gill, H., Pandit, H., Dodd,
C.A., Murray, D.W., 2005. Ten-year in vivo wear measurement of a
fully congruent mobile bearing unicompartmental knee arthroplasty. J.
Bone Joint Surg. Br. 87 (11), 14931497.
Psychoyios, V., Crawford, R.W., OConnor, J.J., Murray, D.W., 1998.
Wear of congruent meniscal bearings in unicompartmental knee
arthroplasty: a retrieval study of 16 specimens. J. Bone Joint Surg. Br.
80 (6), 976982.

1157

Rawlinson, J.J., Bartel, D., 2001. Flat medial-lateral conformity in total

knee replacements does not minimise contact stresses. J. Biomech. 35,
2734.
Rho, J.Y., Hobatho, M.C., Ashman, R.B., 1995. Relations of mechanical
properties to density and CT numbers in human bone. Med. Eng.
Phys. 17 (5), 347355.
Ries, M.D., Weaver, K., Rose, R.M., Gunther, J., Sauer, W., Beals, N.,
1996. Fatigue strength of polyethylene after sterilization by gamma
irradiation or ethylene oxide. Clin. Orthop. Relat. Res. (333), 8795.
Sathasivam, S., Walker, P.S., 1994. Optimization of the bearing surface
geometry of total knees. J. Biomech. 27 (3), 255264.
Sawatari, T., Tsumura, H., Iesaka, K., Furushiro, Y., Torisu, T., 2005.
Three-dimensional nite element analysis of unicompartmental knee
arthroplasty the inuence of tibial component inclination. J. Orthop.
Res. 23 (3), 549554.
Shen, C., Dumbleton, J.H., 1974. The friction and wear behavior of
irradiated very high molecular weight polyethylene. Wear 30 (1), 349
364.
Simo, J.C., Laursen, T.A., 1992. An augmented lagrangian treatment of
contact problems involving friction. Comput. Struct. 42 (1), 97116.
Swienckowski, J.J., Pennington, D.W., 2004. Unicompartmental knee
arthroplasty in patients sixty years of age or younger. J. Bone Joint
Surg. Am., 131142, Suppl. 1 (Pt 2).
van Duren, B.H., Pandit, B.D.J., et al., 2007. How eective are added
constraints in improving TKR kinematics? J. Biomech. 40, S31S37.
Viceconti, M., 2000. A comparative study on dierent methods of
automatic mesh generation of human femurs. Medical Engineering
and Physics 20 (1998): 110. Med. Eng. Phys. 22 (5), 379380.
Viceconti, M., Toni, A., Giunti, A., 1992. Strain Gauge Analysis of Hard
Tissue: Factors Inuencing Measurements. Elsevier Science,
Amsterdam.
Walker, P.S., 1988. Bearing surface design in total knee replacement. Eng
Med 17 (4), 149156.
Wright, T.M., 2005. Polyethylene in knee arthroplasty: what is the future?
Clin. Orthop. Relat. Res. 440, 141148.
Zhao, D., Banks, S.A., DLima, D.D., Colwell Jr., C.W., Fregly, B.J.,
2007. In vivo medial and lateral tibial loads during dynamic and high
exion activities. J. Orthop. Res. 25 (5), 593602.