THE ANNALS OF UNIVERSITY “DUNĂREA DE JOS “ OF GALAŢI 7

FASCICLE VIII, 2006 (XII), ISSN 1221-4590

TRIBOLOGY

FAILURE OF A TOTAL KNEE PROSTHESIS TIBIAL TRAY

DUE TO THE FATIGUE WEAR
Lucian CAPITANU, Aron IAROVICI, Justin ONIŞORU
Institute of Solid Mechanics, Bucharest, Romanian Academy, Romania
luciancapitanu@yahoo.com

ABSTRACT
Some studies report the failure of a tibial insert of knee prostheses due to the
subsurface cracking and delamination wear [1]. The present study starts by
analysing retrieved total knee prosthesis revealing the loss of a functionality of the
prosthesis due to the fact that a larger part of the polyethylene tibial insert was
dislocated. Some qualitatively remarks on the depreciative phenomenon are done
based on optical and microscopic analysis of the affected areas of the polyethylene
component. The failure mechanism was postulated as a combination of the high level
of loading during normal activities and a non-conforming contact mechanism
between the metallic femoral condyles and the tibial insert. The question that arises
is: Could be this phenomenon evaluated quantitatively a priori, e.g. could be the
failure due to delamination wear predicted? In order to do that some finite element
simulations were performed to dynamically determinate the area of the contact and
the contact pressure for three different activities – normal walking, stair ascending
and stairs descending – that could be considered as defining the level of the activity
of the patient. The finite element model uses a modern contact mechanism, including
friction, between the metallic femoral condyles (considered rigid) and the tibial
polyethylene insert. The results reveal the disadvantages induced by the non-
conformity of the contact, it means the fact that the loads are transferred by relatively
small areas of higher stresses. A special summation technique was adopted in order
to consider the cumulative effect of all activities, based on the frequency of each
activity and on the magnitude of the determined damage function. The computed
surface of cumulative maximum damage traversing the layers of the polyethylene
insert could be identified as the cracking surface in the retrieved prosthesis. The
conclusion is that a partial answer to the question enunciated before could be
achieved by the method presented here, that is able to determinate the path of the
subsequent failure of the polyethylene inserts even that for an accurate and complete
answer some extensions of the simulations needed to be done.

KEYWORDS: Total knee prosthesis, wear, experimental methods.

1. INTRODUCTION Examining an early retrieved prosthesis, Ries et

Recent clinical studies (Blunn et al. [1],
Wasielevski et al. [2]) identify the massive wear of
the polyethylene tibial insert due to the mechanics of
non-conforming contact as the main cause to
prosthetic failure. In [3] Knight et al. report that from
209 cases of primary TKR, 18 prostheses failed, all
tibial inserts showing wear pits and delamination. The
average time to failure was around 80 months, close
to that reported by Heck et al. in [4] (i.e. 72 months),
but other studies (Engh et al. [5], Jones et al. [6],
Kilgus et al. [7], Mintz et al. [8]) show that gross
delamination could occur even earlier.
al. [9] noticed gross wear pits and delamination of the
tibial insert, accompanied by an increase of
percentage crystallinity at a plane coincident with the
plane of delamination and the presence of subsurface
oxidation peak typically observed in UHMWPE
inserts, gamma-irradiated in air and aged. Still
controversial, the idea that increasing the thickness of
tibial inserts will reduce wear was rather false.
Indeed, it is the contact non-conformity which
conducts to high contact pressure localized on small
areas and subsequently to the fatigue wear and
delamination of polyethylene part [3, 4].
8 THE ANNALS OF UNIVERSITY “DUNĂREA DE JOS “ OF GALAŢI
FASCICLE VIII, 2006 (XII), ISSN 1221-4590
TRIBOLOGY

a) b)

c) d)

Fig. 1. Pictures of retrieved TKP after 21 years operation: (a) CoCr alloy femoral-component; (b) UHMWPE
insert and 16 years operation; (c) CoCr alloy femoral-component; (d) UHMWPE insert [15].

insert in figure 3a is from an early retrieved tibial part

Fig. 2. Optical image of a UHMWPE insert estimated
by using a shape tracer. Lines indicate surface
profiles. (+) marks indicate the maximum
deformation point at each area. [15].
Analysing two retrieved TKP (after 21 and 16
years from implantation), Oonishi et al. [15] noticed
the presence of parallel scratches (oriented antero-
posterior) on femoral component (fig. 1), and large
delamination of polyethylene tibial insert (fig. 2).
Our study, based on examination of two
retrievals (see fig. 3 and fig. 4) identifies also some
characteristic marks of fatigue wear phenomena. The
(the revision was due to the loosening of the implant)
and one could observe the wear pits of the contact
surface, that are specific for adhesive wear, and also
delamination of the peripheral area of the medial part
as effect of fatigue wear.
The insert in figure 3b was from a total knee
prosthesis retrieved at 67 months from implantation.
In this case, the fatigue wear has more catastrophic
effects – massive loss of polyethylene in the medial
part, and severe delamination in the lateral part.
From the qualitative evaluation of the two
retrieved implants we could conclude that the fatigue
wear debuted with some wear pits of the contact
surface and delamination (especially in the peripheral
medial part of the tibial insert) but, in time, severe
damages occur (gross delaminations, massive losses
of polyethylene).
2. METHODOLOGY AND FE MODEL
The methodology for predicting the fatigue wear
phenomena combines FE analyses of active loading
cycles of relevant routine activities with a summation
technique that is based on computation of a
cumulative estimator of damage.
 THE ANNALS OF UNIVERSITY “DUNĂREA DE JOS “ OF GALAŢI 9
FASCICLE VIII, 2006 (XII), ISSN 1221-4590
TRIBOLOGY

a) b)
Fig. 3. Two retrieved polyethylene inserts showing (a) pitting and (b) severe delamination [6].

Fig. 4. Pictures of fatigue wear of a total hip prostheses components.

A FE model of the artificial joint contact was
used (fig. 5) for all analyses. The model includes one
femoral condyle and one half (the medial one) of the
polyethylene insert, and of the metallic tibial tray. For
deformable parts (the tibial parts) solid brick elements
with 8 nodes and 3 DOF’s per node (all three
translations) are used.
Fig. 5. Finite element model used in analyses [16].
The femoral condyle having a thoroidal shape
with a radius of 22 mm in the sagital plane (the
flexion plane), and a radius of 30 mm in the
transversal plane, was considered rigid.
Elastic linear constitutive laws are considered
for metallic as for the plastic parts. The materials
properties are listed in Table 1. The routine activities
considered relevant for this study are the active cycles
of the normal walking, stair descending, and stair
ascending activities.
The kinematics of all three movements are
illustrated in figure 6 – in (a), (b), and (c) the
segmental movement of the lower member are
sketched (Bergmann et al. [10]), in (d) the
compressive force of contact are listed comparative
for all activities (Taylor and Walker, [11]), in (e) and
(f) two determinants of the cinematic conditions (the
flexion angle and the internal-external rotation of
tibia) are plotted (Bergmann et al. [10]).
The compressive force is applied to the tibial
tray (as distributed pressure on the lower basis of it).
The two rotations (flexion and internal-external
rotation) are dynamically constraining the femoral
part. We considered that the actual level of flexion
(not higher than 70 degrees) does not induce an
antero-posterior translation (usually appearing when
the ligaments are overstrained) as for extended
flexions.
10 THE ANNALS OF UNIVERSITY “DUNĂREA DE JOS “ OF GALAŢI
FASCICLE VIII, 2006 (XII), ISSN 1221-4590
TRIBOLOGY

300
normal walking
stairs descending
250 stairs ascending

Joint contact force [%BW]

200

150

100

50

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized time [s/s]
a) d)
70

60 normal walking
stairs descending
stairs ascending
50
Flexion angle [deg]

40

30

20

10

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized time [s/s]
b) e)

25

normal walking
20
stairs descending
stairs ascending
Internal-external rotation angle [deg]

15

10

-5

-10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized time [s/s]
c) f)

Fig. 6. Dynamic conditions during all activities: kinematics of lower member [10] (a – normal walking, b – stairs
descending, c – stairs ascending) and loading [11] and kinematics [10] of the joint surfaces (d – contact force, e –
flexion angle, f – internal-external rotation angle).

Table 1. The mechanical properties of materials used in FE analyses

Material Young Modulus E [GPa] Poisson ratio ν Remarks
Co-Cr alloy 200 0.3 from ISO 5832-4
UHMWPE 1.06 0.36 from Lewis [12]
 THE ANNALS OF UNIVERSITY “DUNĂREA DE JOS “ OF GALAŢI
FASCICLE VIII, 2006 (XII), ISSN 1221-4590
TRIBOLOGY
The contact mechanism includes rolling and
sliding of the two joint surfaces – the femoral
condyliar thoroidal surface and the planar surface of
tibial insert; the friction obeys the Coulomb law with
a friction coefficient of 0.12 (Villa et al. [13]).
For every activity a damage estimator could be
computed from the variation of the shear maximum
principal stress in every element (Sathasivam and
Walker [14]):
n
1 frictional sliding) it results that the maximum shear
D kf = ∑2 τ
i =1
i +1 − τ i ⋅ ( τ i +1 + τ i
where:
D kf - damage function for activity k ,
τ i - shear maximum stress on elements for time ti .
The effect of all activities could be cumulated
by evaluating a weighted sum:
Dtot
f = wk D kf∑ (2)
where the weights are depending on the frequency of
the activity.
3. RESULTS AND DISCUSSION
Performing the dynamic analyses of the contact
between the femoral metallic condyles and the poly-
ethylene tibial insert (in the above-described condi-
tions) one could determinate the characteristics of the
contact mechanism in the artificial knee. For example,
the trajectories of the contact spot for all three
activities are plotted on the medial part of the tibial
insert (fig. 7).
Examining the sketch one could see that all
activities do not involve extension of the knee – it
means that only the median and posterior parts of
insert are used. For normal walking the trajectory is a
closed curve (like a hysteretic loop).
The stair ascending and stair descending
trajectories are quite similar (even opposite as
direction), both of them having a large internal-
external rotation (for stair ascending loading at the
beginning of the cycle, for stair descending at the end
of it). One could see that the normal walking loading
cycle will affect the median area of the medial part of
the insert, the stair ascending and descending
activities having their maxima located in the posterior
parts of the insert.
As expected, due to the geometry of the contact
(the joint surfaces are a toroidal surface for femoral
condyle and a plane surface for tibial insert) it is clear
that the contact spot will have an elliptical form with
the major axis oriented transversally to the saggital
plane of the femoral condyle. The maximum contact
pressure is a good estimate of the intensity of the
loading. The variation in time of this contact para-
meter will follow the variation of loading (see fig. 8).
One could notice for all activities an initial step
maxima (corresponding to the initiation of the
movement – the contact between the foot and the
11
ground) followed by even a higher value but
distributed on a large domain of time (corresponding
to the weight bearing stance).
The contact pressure is very important as
estimator of loading magnitude, but even more
important for estimating the fatigue wear will be the
maximum shear stress, as being the parameter used in
the damage estimator stated by formula (1). From the
contact mechanism (which involves both rolling and
) (1) stress will be located under the contact surface at a
distance depending of the aspect ratio of the contact
elliptic spot.
Using the cumulative estimator defined in for-
mula (2) the areas were the damage is likely to occurs
could be identified from the cumulative distributions
plotted in figure 9 (the L-shaped region of different
colour). One could see that the most affected areas are
the medial and posterior parts of the tibial insert.
One could also notice that the maximum dama-
ge will occur in the subsurface of the insert (where the
shear stresses from normal walking are maximal),
which is usual the starting plane of severe delamina-
tions. The magnitude of the damage estimator
(~477MPa2 at the surface and ~535 MPa2 in the
subsurface) is in concordance with the results of
Sathasivam and Walker [14] which obtained a
damage score of ~230 MPa2 for a constant loading of
1,000 N (approx. 120% BW) which represent one half
of the maximum loading from normal walking or
from stair ascending or stair descending activities.
4. CONCLUSIONS
A method combining the Finite Element Method
for evaluating the contact mechanism in an artificial
knee joint with a summation technique based on a
damage estimator was successfully used in order to
predict the areas were the damage is likely to occur.
Based on the above-described study, one can
formulate several conclusions.
First, the fatigue wear was identified as the main
phenomenon responsible for the polyethylene tibial
insert failure. The clinical studies show that even
those tibial parts retrieved for other reasons (implant
loosening, misalignment etc.) presented markers of
fatigue wear (cracks in the subsurface, wear pits,
delaminations, loss of large parts of polyethylene).
The cumulative nature of the wear fatigue
phenomena requires a qualitative and quantitative
evaluation of the transfer of loadings through the joint
and a summation technique for encountering the
variety of the human activities. The first part was
assessed by intermediate of some dynamic finite
element analyses of the joint contact mechanism. For
the second part, we assumed that normal walking,
stair ascending and stair descending are the regular
activities dominant for the phenomena.
12 THE ANNALS OF UNIVERSITY “DUNĂREA DE JOS “ OF GALAŢI
FASCICLE VIII, 2006 (XII), ISSN 1221-4590
TRIBOLOGY

Posterior
stair
ascending

stair
descending

Medial

Lateral
normal
walking

Anterior
Fig. 7. The contact paths for all three cases considered.

100

90

80
Contact pressure [MPa]

70

60

50

40 normal walking
stairs descending
stairs ascending
30

20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized time [s/s]

Fig. 8. The contact pressure variation for all three cases.

a) b)
Fig. 9. The damage score map (a) at the contact surface, (b) in the subsurface.
 THE ANNALS OF UNIVERSITY “DUNĂREA DE JOS “ OF GALAŢI
FASCICLE VIII, 2006 (XII), ISSN 1221-4590
TRIBOLOGY
Thus, for all three activities the finite element
analyses are performed and a damage score was
computed. Finally, a cumulative damage score (that
accounts for all three activities) was determined and
the areas were the fatigue wear is likely to occur, are
identified.
From the distribution of the cumulative damage
score (plotted in figure 9) one could see that the areas
that could be damaged are located medially in the
median and posterior regions of the tibial insert. Even
the stair ascending and descending are activities that
generate higher contact pressure, due to the increased
frequency, the normal walking is still dominant.
A closer look to the distribution of damage
score reveals that the maximum damage is likely to
occur not at the contact surface but, rather, in the
subsurface. The presence of maximum values in an
area were other studies (see Ries et al. [9]) identifies
an increase of the percentage cristallinity and the
presence of subsurface oxidation peak induced by the
gamma-irradiation and ageing of the prosthesis could
be an explanation of the early delamination of the
polyethylene inserts.
We intend to extend the study in two directions:
firstly, the contact mechanism would be improved by
considering the polyethylene viscoelastoplasticity and
a better friction mechanism (accounting for the
dependency of the friction coefficient to the contact
pressure and for the velocity of the relative movement
between the joint surfaces); secondary, the summation
technique would be extended to a large area of
activities and the influence of the frequency of the
loadings would be better estimated.
REFERENCES
1. Blunn G.W., Joshi A.B., Minns R.J. et all, 1997, Wear in
retrieved condylar knee arthroplasties, Journal of Arthroplasty,
Vol. 12, pp. 281-290.
2. Wasielewski R.C., Galante J.O., Leighty R.M., Natarajan
R.N., Rosenberg A.G., 1994, Wear patterns on retrieved
13
polyethylene tibial inserts and their relationship to technical
considerations during total knee arthroplasty, Clinical
Orthopaedics, Vol. 299, pp. 31–43.
3. Knight J.L., Gorai P.A., Atwater R.D., Grothaus L., 1995,
Tibial Polyethylene Failure After Primary Porous-coated Anatomic
Total Knee Arthroplasty, J. of Arthroplasty, Vol. 10, pp. 748-757.
4. Heck D.A., Clingman J.K., Kettelkamp D.G., 1992, Gross
polyethylene failure in total knee arthroplasty, Orthopedics, Vol.
15, pp. 23ff.
5. Engh G.A., Dwyer K.A., Hanes C.K., 1992, Polyethlyene wear
of metal-backed tibial components in total and unicompartmental
knee prostheses, J Bone Joint Surg, Vol. 74B, pp. 9-17.
6. Jones S.M.G., Pinder I.M., Moran C.G., Malcolm A.J., 1992,
Polyethylene wear in uncemented knee replacements, J Bone Joint
Surg, Vol. 74B, pp. 18-22.
7. Kilgus D.J., Moreland J.R., Finerman G.A. et al, 1991,
Catastrophic wear of tibial polyethylene inserts, Clin Orthop, Vol.
273, pp. 223-231.
8. Mintz L., Tsao A.K., McCrae C.R. et al, 1991, The
arthroscopic evaluation and characteristics of severe polyethylene
wear in total knee arthroplasty, Clin Orthop, Vol. 273, pp. 215-222.
9. Ries M.D., Bellare A., Livingston B.J., Cohen R.E., Spector,
M., 1996, Early Delamination of a Hylamer-M Tibial Insert,
Journal of Arthroplasty, Vol. 11, pp. 974-976.
10. Bergmann G., Deuretzbacher G., Heller M., Graichen F.,
Rohlmann A., Strauss J., Duda G.N., 2001, Hip Contact Forces
and Gait Patterns from Routine Activities, Journal of
Biomechanics, Vol. 34, pp. 859-871.
11. Taylor S.J.G., Walker P.S., 2001, Forces and moments
telemetered from two distal femoral replacements during various
activities, Journal of Biomechanics, Vol. 34, pp. 839-848.
12. Lewis G., 2001, Properties of cross-linked ultra-high-
molecular-weight polyethylene, Biomaterials, Vol. 22, pp. 371-401.
13. Villa T., Migliavacca F., Gastaldi D., Colombo M.,
Pietrabissa R., 2004, Contact stresses and fatigue life in a knee
prosthesis: comparison between in vitro measurements and compu-
tational simulation, Journal of Biomechanics, Vol. 37, pp. 45-53.
14. Sathasivam, S., Walker, P.S., 1999, The conflicting
requirements of laxity and conformity in total knee replacement,
Journal of Biomechanics, Vol. 32, pp. 239-247.
15. Oonishi H., Kim S.C., Kyomoto M., Iwamoto M., Ueno M.,
2006, PE wear in ceramic/PE bearing surface in total knee
arthroplasty; Clinical experiences of more than 24 years,
Bioceramics and Alternative Bearings in Joint Arthroplasty, 11-th
BIOLOX Symposium Proceedings, pp.101-110.
16.Onişoru J., Capitanu L., Iarovici A., 2006, Failure of a tibial
insert of a total knee prosthesis due to fatigue wear, Buletinul
Institutului Politehnic din Iasi, Tomul LII (LVI), Fasc. 6A, ACME
2006/1, pp.333-340.