Pergamon J. Bmmechanics. Vol. 27, No. 3. pp. 255 2h4.

1994
Elsevier Science Lid
Printed in Great Bntain. All rights reserved
0021 9290/94 $6.00+ 00

OPTIMIZATION OF THE BEARING SURFACE GEOMETRY

OF TOTAL KNEES
SHIVANISATHASIVAMand PETER S. WALKER
Biomedical Engineering Department, University College & Middleselt School of Medicine, Royal National
Orthopaedic-Hospital, U.K.

Abstract-Various design criteria were examined in combination to find the ideal geometry for a condylar
knee replacement. The criteria were the contact stresses on the plastic, femoral-tibia1 size interchangeabil-
ity, patella lever arm, laxity and stability and the amount of bone resection required. The variables were the
radii of curvature of the femoral and tibia1 bearing surfaces in the sagittal and frontal planes. Metal toroidal
indentors were loaded onto dished surfaces of UHMWPE covering a range of radii and the contact areas
measured. Using elasticity equations, the apparent elastic modulus of UHMWPE ranged from 400 to
600 MPa for less conforming to closely conforming surfaces. Using a value of 600 MPa, contact stresses
were predicted for a complete spectrum of radii of curvature. Finite element analysis was used to determine
the stresses beneath the contact patches when different femoral-tibia1 sizes were interchanged. A computer-
graphics program was written to analyse the effects of flexion, rotation and femoral roll-back on the contact
point locations. An influential variable was the sagittal curvature of the femoral component, notably the
point of transition between the posterior curve of small radius and the distal curve of larger radius. This
affected the patella lever arm, the stability, and the bone resection. Interchangeability was primarily
dependent upon the relative frontal radii. Contact stresses and contact locations depended upon the
combination of sagittal and frontal radii. The most suitable geometrical combinations overall were
discussed. -

INTRODUCTION motion patterns, and to allow shear forces to be

Various criteria have been applied to the design of
condylar replacement knees, which has led to the
manufacture of components which satisfy some of the
requirements but with less attention to others. Knees
have been designed which exhibit normal or improved
patella lever arms, usually by suitably located dishing
of the surfaces or by cam mechanisms (Burstein, 1984).
Other designs emphasize sufficient laxity for activities
of everyday living with progressive restraint to laxities
from the neutral position (Thatcher et al., 1987). El
Nahass et al. (1991) measured knee motions in various
activities and concluded that a knee prosthesis should
allow an internal/external rotation of - 12” to + 12”
and an anterior-posterior displacement of 13 mm.
Walker et al. (1974) showed theoretically the relation-
ship between the laxities produced by applied forces
and moments and the relative curvatures of the fem-
oral and tibia1 condyles. However, Bartel et al. (1986)
showed that the relative geometry between the fem-
oral and tibia1 surfaces also affects the contact stresses
which are believed to affect the wear and deformation
of the plastic. Walker (1988) showed that the least
stresses were experienced with surfaces allowing only
fixed axis cylindrical motion, the highest with flat
tibia1 surfaces. However, a study of retrieved speci-
mens by Landy and Walker (1988) concluded that
flatter components, although prone to higher stresses,
did not necessarily exhibit high wear. Low conformity
in the sagittal plane has been included in a number of
recent designs to allow for variable positioning and
Received in jinal form 31 July 1993.
carried by the posterior cruciate ligament (Andriacchi
and Galante, 1988). In an alternative approach where
both cruciates are resected, anterior-posterior stabil-
ity as well as reduced contact stresses are achieved
using more conforming surfaces with a posterior con-
tact location (Freeman and Railton, 1988). In an effort
to achieve the closest possible match of the compo-
nents to their respective bones, most designs now
incorporate schemes to allow interchangeability be-
tween sizes. For example, a large femoral component
can be mated with a standard tibia1 component.
The aim of this study was to examine the various
design criteria in combination, as a function of the
geometry of the bearing surfaces. The criteria were the
contact stresses, interchangeability, the effect of rela-
tive femoral-tibia1 position, the patella lever arm, the
stability, and the amount of bone resected.
METHODSAND MATERIALS
Dejinition of the geometry of bearing surfaces
The geometry of the bearing surfaces was defined in
perpendicular planes, frontal and sagittal (Fig. 1). The
frontal profile of the femoral surface was formed from
two equal curves of radius FF, one for each condyle.
In the same plane the tibia1 surface had mating curves
of radius TF. In the sagittal plane the tibia1 condylar
surface had radius TS. The sagittal condylar profile of
the femoral surface was constructed from curves of
radii 20,SO and 25 mm for the posterior, distal and
anterior sections, respectively. The anterior-posterior
length of this profile was 62 mm. The posterior-distal
transition angle (PDTA) was defined as the angle at

255
256 S.SATHAWAM and P. S. WALKER

FRONTAL TIBIAL
SURFACE cant:; point

knee components

Fig. 1. Terminology in describing the frontal and sagittal plane geometries of the condylar surfaces.

which the posterior radius ended its arc and the distal twice for each combination of indentor and
radius began. The posterior section was fixed and the UHMWPE sample. The dimensions of the elliptical
anterior section was fixed horizontally, its vertical contact patches (2c and 2d) were measured to an
position and the length of the distal curve determined accuracy of f 0.04 mm by using vernier callipers.
by the PDTA. The maximum stress in the plastic is given by
(Roark and Young, 1975)
Determination of contact stresses on the tibia1
component maxcC=g, (1)
The aim was to determine how contact stress varied
with tibia1 frontal and sagittal radii when a femoral where P is the total load applied and 2c and ti are the
component with fixed geometry was loaded onto lengths of the axes of the contact patch.
a tibia1 component. Laboratory experiments were Also,
performed to produce contact patches which were
similar to those made when knee components are c&J=, (2)
compressed together. Indentors were machined from
titanium, one with perpendicular planes of radii 30 d=b3,/a, (3)
and 50 mm, representing a femoral condyle in exten- where u and p are variables which depend on the
sion, the other with radii 30 and 20 mm, a condyle in geometry of the surfaces making contact. Values were
Aexion. Toroidal dishes were machined with min- obtained by interpolating results listed in Roark and
imum thickness 6 mm from uniform blocks of Young (1975) for cases where the major and minor
UHMWPE, with radii varying from 35 to 200 mm. axes of the femoral and tibia1 surfaces are aligned (as
A metal indentor coated with a film of dye was fixed in our Instron tests).
to the loading head of an Instron testing machine. The elasticity function is
A tibia1 dish was aligned with the femoral indentor
and a compressive force of 1 kN was applied in 1 min
and released immediately. Tests were carried out
+&y+(y), (4)

where vF and VT are Poisson’s ratios of the metal
indentor and UHMWPE sample, respectively. EF and
ET are the elastic moduli of the indentor and plastic
sample.
KD is a measure of the conformity of the surfaces:
1.5
KD=(lpF)+(l/FF)+(lflS)+(l/FS)’
where tibia1 radii are negative. TF, FF, TS and FS are
defined in Fig. 1.
The value of the elastic modulus of UHMWPE
obtained from these experiments was used to deter-
mine a ‘full-field’ solution of maximum contact stres-
ses (max UJ for an entire spectrum of femoral and
tibia1 radii.
Interchangeability between sizes of the components
The main purpose of this analysis was to find the
optimum geometry for a tibia1 component which
could accommodate different sizes of femoral compo-
nents whilst maintaining low stresses. Linear, plane
strain, finite element analysis (ANSYS) was used to
model half of the frontal profiles of the knee compo-
nents. The elasticity equations above could not be
used for cases where the contact patches reached the
edges of the plastic. The femoral component was
modelled as a cylindrical, rigid surface, which is de-
fined in ANSYS as a boundary of nodes beyond which
elements cannot penetrate. The stiffness of these gap
elements was determined by applying ANSYS to
a rigid, cylindrical indentor loaded onto a large, flat,
plastic slab. The slab had elastic modulus 600 MPa
(determined from results of experimental study) and
Poisson’s ratio 0.4. Elasticity theory was considered
applicable for this unconforming case. The slab was
interfaced against the cylindrical indentor and equal
vertical forces were applied to the nodes along the
base-line of the plastic slab. The analysis was repeated
f t f t
PLASilC
ELEMENTS _,-,
Fig. 2. Using FEA to examine interchangeability
tibia1 radius is 45 mm and the femoral component is 6% larger than standard. For clarity, all the elements
Surface geometry of knees 251
with different vertical forces until the reaction forces
at the interface summed to 1000 N. The analysis was
repeated with different stiffnesses for the gap elements
until the maximum compressive stress was equal to
that predicted by elasticity theory. This stiffness was
used for the analyses of the frontal planes of the tibia1
(5) components. The arrangement of the elements de-
pended on the point of contact; an example is given in
Fig. 2. It illustrates how a 6% enlarged femoral com-
ponent was loaded onto a tibia1 component with
outer radius 45 mm. In order to determine where the
nodes should be concentrated, the contact point was
estimated by rigid body analysis. We assumed that the
interface between the rigid, femoral surface and the
plastic elements was frictionless. Variables were the
outer frontal tibia1 radius (Fig. 1) and the percentage
sizing of the femoral component. Tibia1 components
with outer frontal radii 35,40,45 and 50 mm were
analysed with femoral components:
(a) which were sized by 6% where all dimensions
varied proportionally,
b) where the radii and width were sized by 6%, but
the bearing spacing was sized by only 3%.
However, the location of the contact patches would
vary with different relative positions of the femoral
and tibia1 surfaces. A computergraphics program was
written to analyse this. The surface of the knee com-
ponents was defined by meshes of points approxim-
ately 1 mm apart. Transformation matrices were used
to position the femoral surface at any orientation. The
contact points, defined as the points on the femoral
and tibia1 surfaces in closest proximity, were deter-
mined by iterating through each femoral point using
a descent method (Walker, 1988). Contact points on
the tibia1 surface were plotted for all possible combi-
nations of flexion up to 60”, internal/external rotation
up to 10” and posterior femoral displacement up to
8 mm at increments of 12”, 2” and 2 mm, respectively.
contact point (where
the mesh was most
concentrated) initially
determined by rigid
reaction force is
into 15 elements
1 Ttfttt:“tt l
equal vertical displacements
between sizes. This example shows a case where the outer
are not shown.
258 S. SATHASIVAMand
Using this method, analyses were also run for femoral
components with large, frontal outer radii of 60 and
170 mm.
Considerations for the sagittal plane
Domed patella components and patella tracks
which were located 5 mm within the outline of the
femoral condylar surface were added to the sagittal
profiles. The tibia1 sagittal surface was defined to be
totally conforming to the femoral sagittal profile for
this part of the study. The anterior-posterior length of
the tibia1 component was 50 mm, its posterior edge
was located 3 mm in front of the posterior of the
femoral component and the top of the tibia1 compon-
ent was sliced off so that there was a 5 mm flat at the
anterior. Six femoral profiles were created with PDTA
from - 15” to lo”. The reference axis system for the
femur and the tibia was based on the transverse axis
joining the centres of the ‘spherical’ posterior femoral
condyles with the femur at 0” flexion (Garg and
Walker, 1990). The large cross marked on Fig. 3
shows this axis. The centres of the posterior radial
sections of the prosthetic sagittal profiles were located
on this cross. Flexed femoral component profiles were
positioned so that their lowest points made contact
with the deepest points on the tibia1 dishes. The angle
ALPHA described the orientation of the patella liga-
ment and the angle DEL described the orientation of
the patella component itself. These angles were iter-
ated until the lines of action of the patella ligament,
quadriceps and the reaction of the patella on the
patella
groove
surrace -
Fig. 3. Models used to analyse the sagittal plane. Top left, patella equilibrium positions and lever arms.
Bottom left, anterior stability. Right, amount of bone resection on chamfer.
P. S. WALKER
patella groove crossed at one point, indicating that
the patella component was in equilibrium. The patella
lever arm was measured in this position as the perpen-
dicular distance from the femoral-tibia1 contact point
to the patella ligament.
Also, for each value of PDTA, the heights of the
deepest point of the condylar surface to the anterior
edge of the tibia1 dish (HAC) and the gradient of the
tangent to the anterior edge of the dish (SAC) were
calculated. Finally, the profiles were superimposed on
the mid-lateral slice of the ‘average knee’ to determine
how bone resection varied with PDTA at the thickest
region of the bone. Bone resection was defined as the
perpendicular distance between a 45” slope to the
patella track of the prosthesis and a slope at the same
angle to the femoral bone. Fitting the femoral com-
ponent to the femur requires a 45” chamfer at the
distal-anterior end and the amount of bone resected
depends on the PDTA.
RESULTS
Determination of contact stresses
The results for the apparent elasticity function of
the UHMWPE (C,) plotted against the relative radius
of curvature function (K,) are shown in Fig. 4. The
apparent elastic modulus of the plastic ranged from
400 to 600 MPa between low and high conformities,
for a load of 1 kN applied at 1 kN min- r. Poisson’s
ratio of UHMWPE was assumed to be 0.4.600 MPa
was used for the elastic modulus of UHMWPE in
patella
surface
 Surface geometry of knees 259

Ce
.
1
/ 0.0024
.
0.0022 . .
* .
: . .
0.0020 l*e l
. .
.
/
0.0018 : . l
8 .
.
0. . l

0.0016

0.0014
.
.
4
i
. . .
. . .

Fig. 4. Function C, (containing modulus of elasticity of plastic) plotted against function K, (containing
relative radii of curvature).

---I
femoral and tibia1
component

-
surfaces are femoral component
unconforming with low PDTA in
flexlon or high PDTA
- tibia1 surface is

@4
femoral and tibia1
rv,,rpw,r,,r

surfaces are
unconforming to both

conforming

CONSTANTS:
femoral frontal radius = 30mm
) femoral sagittal radius = POmm 1 P
femoral cc>mponent
with hioh w I n
surface ccGZik:iob’2
shape
r
femoral component with
low PDTA in’flexion or high
PDTA - tibia1 surface is
conforming to low PDT/ 4 in
extension

Fig. 5. The effect of varying the frontal and sagittal tibia1 radii on the maximum contact stresses (applied
force loo0 N).

subsequent analyses. The elasticity equations were increased, but then reached an intermediate level
applied again when calculating the maximum contact where further reduction had little effect. A reduction
stresses produced when a femoral component in of conformity in both frontal and sagittal planes led to
flexion was paired with a spectrum of tibia1 compo- a rapid increase in contact stresses to high levels.
nents defined by varying sagittal and frontal radii
(Fig. 5). The lowest stresses occurred when the fem- Interchangeability between sizes of the components
oral and tibia1 radii were similar, in both sagittal and For a standard femoral component (frontal radius
frontal planes (the analysis considered a minimum 30 mm) loaded on a standard tibia1 component
difference in femoral-tibia1 radii of 5 mm). If the con- (frontal radius 35 mm), the Von Mises stresses within
formity in one plane was reduced, the stresses quickly the plastic increased as the outer tibia1 radius was

El427:3-B
260 S. SATHASIVAM and P. S. WALKER

lnterchanaeabilitv Analysis

1.40- Flz1.06 F2=1.06

1.30-

1.20-
Flz1.00 Fl=l.OO

l.lO-
n Fk0.94 F2=0.94
l.OO-

0.60 ,
30 35 40 45 Franf~ 55
Outer Tibia1 Radius
Fig. 6. Results of the planar finite element analysis of the frontal plane to compare maximum Von Mises in
the plastic for different designs of tibia1 component and different methods of sizing the femoral component.
The outer frontal tibia1 radius (Fig. 1) is varied. Fl =femoral radius as fraction of standard size.
F2 = femoral bearing spacing (Fig. 1) as fraction of standard size.

increased (Fig. 6). For example, there was a 19% in-

crease for an outer tibia1 radius of 50 mm. For a small
femoral component reduced by 6% from standard
(Fl = F2 =0.94), the stresses were not affected by the
outer tibia1 radius because the contact areas were on
the inner tibia1 radius. When a 6% increased femoral
component was placed on the standard tibia1 with
35 mm outer radius, the contact area was at the edge
of the plastic resulting in high Von Mises stresses
(Fig. 6, broken line). However, for a 40 mm outer
tibia1 radius, the contact points were within the tibia1
periphery and the stresses were only elevated by 15%.
If the bearing spacing of the femoral component was
only increased by 3%, however, the contact points
were within the tibia1 periphery and the stresses were
actually reduced by 9%.
Further details of these three conditions are shown
in Fig. 7. In addition to the Von Mises stress plots
(which show the location of the maximum stress to be
approximately 1.5 mm below the surface), the centres
of the contact points are shown for the combinations
of flexion, rotation and posterior displacement. The Enlaraed Femoral on Standard Tibia1
6% enlarged femoral component shows a tendency
for some of the contact points to be closer to the
posterior periphery of the plastic than for the other
cases.
Interchangeability could also be achieved with
a constant bearing spacing and large outer femoral
and tibia1 radii, resembling ‘cylindrical’ designs. In
our example, a sagittal tibia1 radius of 60 mm is as-
sumed in order to provide anterior-posterior and
rotational stability (Fig. 8). In this case, during the
rotations and displacements, the contact points move Fig 7. Contact stresses for the frontal plane and contact
to the periphery of the plastic. This is characteristic of point locations for interchanged components where the
rotating a cylindrical roller in a cylindrical trough. In outer tibia1 radius is 40 mm. Top, femoral with 6% smaller
frontal radius but with 3% smaller bearing spacing, on
contrast, when the outer femoral radius is reduced, to standard tibial. Centre, standard femoral on standard tibial.
be equal to the tibia1 sagittal radius, nearly all of the Bottom, femoral with 6% larger frontal radius but with 3%
contact points fall within the plastic boundary (Fig. 8). larger bearing spacing, on standard tibial.
 Surface geometry of knees 261

almost cvlindrical surfaces Considerations for the sagittal plane

The patella lever arms for different femoral geo-
metries differed most in early flexion (Fig. 9). For
PDTA = - 15” the lever arm was 33.6 mm, while for
PDTA= lo” the lever arm was 40.7 mm. As flexion
increased to 90” the range of values narrowed down to
3 1.7 mm for PDTA = - 15” and 33.6 mm for PDTA =
10”. Hence, the main advantage in the femoral compo-
nents with high PDTA is increased quadriceps effec-
OUTER FRONTAL
FEMORAL RADIUS=170mm
tiveness in early flexion, rather than at the higher
OUTER FRONTAL flexion angles.
TIEIAL RADIUS=200mm
Another characteristic of high PDTA was in the
increasing anterior height and anterior slope of the
tibia1 surface in the sagittal plane (HAC and SAC,
outer FF = TS Fig. 3). The anterior heights rose steadily with the
PDTA angle from 4.9 mm for PDTA = - 15” to 12.5
mm for PDTA= 10”. The anterior gradients of the
tibia1 dishes showed a similar increase. This implied
the possibility of increased inherent stability for de-
signs with high PDTA. However, Fig. 9 also shows
that during 60” flexion and beyond, the patella starts
to impinge anteriorly on the tibia1 plastic. This can be
solved by reducing the height at the front of the plastic
OUTERFRONTAL
FEMORAL RADIUS=6Omm which causes a reduction in stability.
OUTER FRONTAL
TIEIAL RADIUS=70mm
As the high PDTA required a steeper upsweep of
the femoral sagittal profile, from distal to anterior,
Fig. 8. Contact point locations for combinations of flexion, more bone resection would be needed for installation.
internal-external rotation and posterior displacement. Top,
The profile with PDTA = - 15” required only 4.9 mm
outer frontal plane radii are large (close-to-flat). Bottom,
outer frontal femoral radius is equal to the sagittal tibia1 of bone to be resected from the lateral slice, whereas
radius. that with PDTA= lo” required 12.2 mm.

32

60 90
FLEXION ANGLE (deg)
Fig. 9. Lever arms of the patella ligament over a range of Rexion angles, for femoral components with
different posteriordistal transition angle (PDTA).
262 S. SATHASIVAM and P. S. WALKER
DISCUSSION
It is generally assumed that the wear and deforma-
tion of the plastic increases with contact stress (Bartel
et al., 1986; Rose et al., 1983; Rostoker and Galante,
1979) although the evidence from implant retrievals is
not entirely clear on this point (Landy and Walker,
1988; Collier et al., 1991). A complicating factor is that
the highly conforming designs (with lower contact
stresses) show increased wear due to entrapped par-
ticles and areas of deformation due to misalignment.
In order to calculate the contact stresses for a full
range of femoral-tibia1 geometries of toroidal shape,
the elasticity equations were used. Elasticity theory,
which predicts contact stresses for two elastic bodies
when compressed together, only applies to cases
where the bodies are non-conforming and the contact
area is small. It also assumes a planar contact area.
Finally, stresses calculated using this theory only ap-
ply to infinite surfaces (Roark and Young, 1975;
Whelan and Little, 1992). These factors would indi-
cate that standard elasticity theory may be unsuitable
for determining the contact stresses (and hence predic-
ting wear properties) of condylar knee replacements.
However, for our empirical analysis, the hypothesis
was tested that, for a fixed thickness of plastic, the
apparent elasticity of the component would vary as
a function of the conformity of the surfaces. The
experiments used to obtain this function involved
a dye method, as it produced clearly marked contact
patches for all conformities. The dimensions of the
contact patches were accurately measured but it was
not possible to achieve exact repeatability, as can be
seen from the experimental results (Fig. 4), probably
due to the viscoelastic nature of the plastic. Unfortu-
nately, the variability in the results did not allow us to
prove our hypothesis, so the upper bound value for
the elastic modulus of UHWMPE was used to make
a full-field set of maximum stress calculations (Fig. 5).
It appeared that contact stress is more sensitive to
changes in the tibia1 geometry when the conformity is
high than when it is not. Therefore, once surfaces are
in low-moderate conformity, as in the sagittal plane to
allow sufficient laxity, slight changes in the tibia1 ge-
ometry in an effort to improve the stresses will have
little effect. On the other hand, if the frontal plane is
defined to be closely conforming, then varying its
geometry will result in a considerable variation in
stress.
The lowest stresses occurred when the sagittal and
frontal radii of the femoral and tibia1 surfaces
differed by 5 mm, the closest conformity considered in
our analysis (Fig. 5). However, such a geometry would
be excessively constrained, allowing much smaller ro-
tations and displacements than those which occur in
the normal knee. The stresses were calculated for
a femoral condyle with frontal radius 30 mm and
sagittal radius 20 mm. This could represent a femoral
component with low PDTA in flexion, or one with
high PDTA in both flexion or extension, as shown in
the sagittal profile diagrams (Fig. 5). Consequently,
the low PDTA component provides moderate stresses
in extension, but high stresses in flexion. In contrast,
the high PDTA component produces moderate stres-
ses throughout the entire range of flexion. This is
because the femoral and tibia1 sagittal radii are the
same for all angles of flexion. Therefore, considering
stresses alone, high PDTA seems preferable.
Total knee systems typically include five sizes or
more with each component varying about 6% be-
tween sizes. The variability of femoral and tibia1
shapes is such that different-sized components may be
needed for each component to obtain maximum
coverage of the bone surfaces. Such interchangeability
of ‘one size up or down’ was analysed. When a large
femoral component is mated with a standard tibia1
component, the contact patches may reach the pe-
riphery of the tibia1 surface and therefore the elasticity
theory is not applicable. Instead, a finite element ana-
lysis was used, employing a plane strain model in the
frontal plane. Each analysis involved a standard tibia1
component which was mated with three sizes of fem-
oral components, with frontal radii in the region of
30 mm. As might be expected, when proportionate
sizing was used for the femoral components, the small
femoral produced higher stresses because of the small-
er radii, but for the larger component, the contact
points could reach the edges of the plastic surface, an
undesirable situation (Gunsallus and Bartel, 1992).
This effect would be more pronounced for smaller and
more conforming frontal radii. Increasing the radii at
the sides of the components, on the tibia1 surface more
than the femoral, avoided the problem, but at the
expense of higher contact stresses.
If the frontal radii at the sides were increased still
further, to produce almost cylindrical surfaces, a dif-
ferent problem was introduced. If the tibia1 surface had
a sagittal radius of around 60 mm for anterior-poste-
rior stability, during internalexternal rotation, the
contact points moved to the periphery of the plastic.
For this reason, surfaces close to cylindrical need to
mate with close-to-flat tibia1 surfaces in the sagittal
plane.
It might appear that maintaining a constant bear-
ing spacing across all sizes would allow complete
interchangeability. This is depicted in Fig. 10. For
frontal radii around 30 mm, difficulties are encoun-
tered at the extremes of size. For the largest size, there
is a slight reduction in varus-valgus stability and
there is insufficient metal at the sides. For the smallest
size, the contact points are too close to the edges of the
plastic. The problem of the large size can be addressed
by increasing the radii at the outsides of the surfaces.
An alternative solution was analysed where the bear-
ing spacing was increased and decreased at half the
amount of the overall component dimensions. It was
shown that such a scheme was feasible for one size up
and down interchangeability, both for contact stresses
and for locations of the contact areas within the
plastic boundary.
 Surface geometry of knees 263

20% between sizes but

20% between sizes constant bearina spacing

Smallest Size

000”00003
Standard Size
femoral radius=30mm
bearing spacing=48mm
frontal width=76mm

m Largest Size m
Fig. 10. Representation of the frontal femoral profiles for two bearing spacing schemes. The smallest,
standard and largest components are shown for a typical knee system. Left, proportional radii, widths and
bearing spacings. Right, proportional radii and widths but constant bearing spacing; for this scheme, the
contact points for the smallest size are close to the sides.

The entire form of the femoral sagittal profile could
be conveniently altered by varying the PDTA alone,
where a negative value implies that the large distal
radius is carried round posteriorly, while a positive
value implies that the smaller posterior radius is car-
ried round to the anterior. The effect of varying the
PDTA on patella lever arm, stability and bone resec-
tion was studied. It has been shown that patients with
short patella lever arms have compensated gait and
problems with stair-climbing (Andriacchi et al., 1982),
implying that the longer the patella lever arm, the
more advantageous. Using a model where the patella
was considered to be subjected to three forces exerted
by the quadriceps muscle, patella ligament and the
reaction force between the patella groove and the
patella (Ahmed et al., 1987; Huberti et al., 1984; Reith-
meier and Plitz, 1990), it was found that cases with
high PDTA had the longest patella lever arms due the
more posterior contact points. However, this is some-
what different from normal, where the contact points
move anteriorly in extension (Ahmed and Burke,
1983) leading to femoral roll-back in early flexion
(Rovick et al., 1991). This may well have implications
to the action of the hamstrings as well as the quad-
riceps, both of which are active in late swing and early
stance during gait (Inman et al., 1981).
When considering stability, complete conformity
was assumed in order to make a comparison between
the femoral and tibia1 components for different sagit-
tal geometries, even though in reality the surfaces
would be unconforming. The assumption was made
that the steeper the tangent and the greater the height
at the anterior of the tibia1 surface, the greater the
horizontal component of the reaction force, and hence
the greater the resistance to shear forces. On this basis
low PDTA proved to be the least stable, especially, in
flexion. This was to be expected since tibia1 surfaces
with a large radius at the bottom of the dish more
readily allow posterior displacements of the tibia1
surface relative to the femoral surface. This implies
that such surfaces require preservation of the poste-
rior cruciate ligament, or intercondylar stabilising
posts as in some knee designs (e.g. Kinemax Stabiliser,
Insall-Burstein). On the other hand, surfaces with
high PDTA could conceivably be used without a pos-
terior cruciate or other means (e.g. Freeman-Samuel-
son). However, the disadvantage of high PDTA was
additional distal-anterior bone resection of around
5 mm.
This analysis was restricted to the geometrical
parameters defined, namely, femoral and tibia1 surfa-
ces composed of circular arcs in the frontal and sagit-
tal planes, and with sagittal femoral geometry de-
scribed by small posterior and anterior radii and
a larger distal radius, joining at a variable polar angle.
Factors other than those considered are important,
such as the behaviour in varus-valgus or medial-lat-
eral loading. However, within the limits of this study,
it appeared that the most favourable characteristics
occurred for a PDTA around 0 to -S’, moderate
conformity in the sagittal plane and close frontal
conformity. To avoid contact areas reaching the edges
of the plastic, the frontal radii at the sides need to be
smaller than the sagittal radii. To allow interchangea-
bility, compromises are required in contact stresses,
medial-lateral bearing spacing, contact area location
on thk tibia1 surface, and general component aesthet-
ics.
264 S. SATHA~~VAM
and P. S. WALKER
Acknowledgements-The authors wish to thank Mark
Witham for assistance with the laboratory experiments. This
work was funded by the Department of Health, Medical
Devices Directorate, U.K.
REFERENCES
Ahmed, A. M. and Burke, D. L. (1983) In oitro measurements
of static pressure distribution in synovial joints. Part 1.
Tibia] surface of the knee. J. biomech. Engng 105,216225.
Ahmed, A. M., Burke, D. L. and Hyder, A. (1987) Force
analysis of the patellar mechanism. J. orthop. Res. 5,
69-85.
Andriacchi, T. P. and Galante, J. 0. (1988) Retention of the
posterior cruciate in total knee arthroplasty. .J. Arthro-
plasry 3, s13-s19.
Andriacchi, T. P., Galante, J. 0. and Fermier, R. W. (1982)
The influence of total knee replacement during walking
and stairclimbing. J. Bone Jt Surg. 64A, 1328.
Bartel, D. L., Bicknell, V. L. and Wright, T. M. (1986) The
effect of conformity, thickness, and material on stresses in
ultra-high molecular weight components for total joint
replacement. J. Bone Jt Surg. 68A, 1041-1051.
Burstein, A. H. (1984) Biomechanics of the Knee Surgery of
the Knee (Edited by Insall, J.N.), pp. 21-39. Churchill
Livingstone, New York.
Collier, J. P., Mayor, M. B., McNamara, J. L., Suprenant,
V. A. and Jensen, R. E. (1991) Analysis of the failure of 122
polyethylene inserts from uncemented tibia1 knee compo-
nents. Clin. Orthop. 273, 232-242.
Freeman, M. A. R. and Railton, G. T. (1988) Should the
posterior cruciate ligament be retained or resected in con-
dylar nonmeniscal knee arthroplasty? The case for resec-
tion. J. Arthropiasty 3, S3S12.
Garg, A. and Walker, P. S. (1990) Prediction of total knee
motion using a three-dimensional computer-graphs
model. J. Biomechanics 23, 45-58.
Gunsallus, K. L. and Bartel, D. L. (1992) Stresses and surface
damage in PCA and total condylar polyethylene compo-
nents. Trans. Orthop. Res. Sot. 17, 329.
Huberti, H. H., Hayes, W. C., Stone, J. L. and Shybut, G. T.
(1984) Force ratios in the quadriceps tendon and ligamen-
turn patellae. J. orthop. Res. 2, 49-54.
Imnan, V. T., Ralston, H. J. and Todd, F. (1981) Human
Walking. Williams and Wilkins, Baltimore.
Landy, M. M. and Walker, P. S. (1988) Wear of ultra-
high-molecular-weight polyethylene components of 90 re-
trieved knee prostheses. J. Arthroplasty 3, s73-~85.
Nahass, B. E., Madson, M. M. and Walker, P. S. (1991)
Motion of the knee after condylar resurfacing-An in uiuo
study. J. Biomechanics 24, 1107-l 117.
Reithmeier, E. and Plitz, W. (1990) Theoretical and numer-
ical approach to optima1 positioning of the patellar sur-
face replacement in a total knee endoprosthesis. J. Bio-
mechanics 23, 883-892.
Roark, R. J. and Young, W. C. (1975) Formulas for Stress and
Strain (5th Edn.), McGraw-Hill, New York.
Rose, R. M., Goldfarb, H. V., Ellis, E. and Crugnola, A. M.
(1983) On the pressure dependence of the wear of ultra-
high molecular weight polyethylene. Wear 92, 99-l 11.
Rostoker, W. and Galante, J. 0. (1979) Contact pressure
dependence of wear rates of ultra high molecular weight
polyethylene. J. Biomed. Mater. Res. 13, 957-964.
Rovick, J. S., Reuben, J. D., Schrager, R. J. and Walker, P. S.
(1991) Relation between knee motion and ligament length
patterns. Ctin. Biomech. 6, 213-220.
Thatcher, J. C., Zhou, X. M. and Walker, P. S. (1987) In-
herent laxity in total knee prostheses. J. Arthroplasty 2,
199-207.
Van Eijden, T. M. G. J., Kouwenhoven, E., Verburg, J. and
Weijs, W. A. (1986) A mathematical model of the patel-
lofemoral joint. J. Biomechanics 19, 219-229.
Walker, P. S. (1988) Bearing surface design in total knee
replacement. Engng Med. 17, 149-156.
Walker, P. S., Wang, C.-J. and Masse, Y. (1974) Joint laxity
as a criterion for the design of condylar knee prostheses. In
Total Knee Replacement. Publ. Institution of Mechanical
Engineers, Birdcage Walk, London, UK.
Whelan, M. P. and Little, E. G. (1992) A theoretical and
photoelastic contact stress analysis relevant to ortho-
paedic knee prostheses. Strain 61-68 (May, 1992).