Analysis of Relative Motion between Femoral Head and Acetabular cup and Advances in Computation of theWear Factor for the Prosthetic Hip Joint

The amount and type of wear produced in the prosthetic hip joint depends on the type of relative motion between the femoral head and the acetabular cup. Wear particles removed from the bearing surfaces of the joint can cause adverse tissue reactions resulting in osteolysis and ultimately in loosening of the fixation of the implant. When designing a simulator for evaluation of prospective materials for artificial hip joints it is important to verify that the type of relative motion at the articulation is similar to that produced in walking, involving continually changing direction of sliding. This paper is an overview of recent research done at Helsinki University of Technology on the analysis of the relationship between relative motion and wear in the prosthetic hip joint.To analyze the relative motion, software for computing tracks, referred to as slide tracks, drawn on the counterface by marker points on the bearing surface was developed and experimentally verified. The overall relative motion of the joint was illustrated by a slide track pattern, produced by many points. The patterns resulting from walking motion and from motion produced in ten contemporary hip simulator types were compared. The slide track computations were not limited to illustrational purposes but offered a basis for computing variations of sliding distances, sliding speeds and direction of sliding during a cycle. This was done for the slide track termed the force track, drawn by the resultant contact force. In addition, the product of the instantaneous load and increment of sliding distance was numerically integrated over a cycle. This track integral of load had so far not been determined for the majority of contemporary hip simulators. The track integral can be used in determining the wear factor, making it possible to compare clinical wear rates with those produced by hip simulators. The computation of the wear factor was subsequently improved by replacing the track integral of the resultant contact force with a surface integral computed as the sum of track integrals of a large number of smaller normal forces obtained by discretizing the contact pressure distribution. The slide track software could also be utilized in the conceptual design of new simulators because it was possible to rapidly investigate the effect of changes to the motion waveform amplitudes or phases, or of omitting certain waveforms to simplify the design of a simulator.The slide track analysis showed that walking motion produced mainly open tracks on the center of contact, implying continually changing direction of sliding. This phenomenon, which is crucial for obtaining the correct wear mechanisms for acetabular cups made of polyethylene, was reproduced by simulators having abduction-adduction motion in addition to flexion-extension motion. In the force track computations involving contemporary simulators with the common femoral head size of 28 mm, the sliding distance per cycle and the force track integral per cycle ranged from 19.7 to 34.3 mm and from 17.4 to 43.5 N m, respectively. The average sliding speed ranged from 19.7 to 49.0 mm/s. The sum of track integrals computed with forces obtained by discretizing the contact pressure distribution reached a substantially higher value than the track integral obtained with the resultant contact force only. This suggests that the wear factor is actually overestimated when computed in the conventional way by dividing the wear rate with the force track integral.

Total hip joint replacement arthroplasty can provide the patient with dramatically improved quality of life by reliev- ing pain and offering increased mobility.A failed prosthesis however causes pain, reduces the ability to work and necessitates a revision operation.The most common reason for revision operations is loosening of the components (66 7o of hip revisions in Finland []).The current understanding is that loosening is primarily due to bone loss around the implant, caused by adverse tissue reactions to wear particles [2,3,4].Tiibological research of prosthetic joints deals with ftiction, wear and lubrication of the implant with the purpose of increasing its lifetime.A particular area within this field of research is the design and development of wear simula- tion equipment for pre-clinical evaluation of prostheticjoint materials.
Wear simulation is done under conditions that are in a tribological sense close to the conditions in vivo.This means that the essential features of the relative motion at the articulation, the lubrication and the loading, are taken into consid- eration when setting up the wear simulation equipment.
Monitoring of the simulation involves measuring the amount of worn material, visual and microscopic inspection of the wear surfaces and microscopic analysis of the wear debris.
The results are validated against results of in vivo radiological wear measurements, or by comparing with results of in- spection of components retrieved at revision operations or autopsy.Many studies have shown that only multidirectional motion yields realistic wear for polyethylene, which is a com- mon material for acetabular cups in prosthetic hip joints [5, 6, 7, 81.Multidirectional motion means that the direction of sliding changes many times during the motion cycle, unlike, e.g., the situation in a regular pin-on-disk wear test device, in which the relative motion is unidirectional.Since the relative motion of the articulating surfaces is known to influence the production ofwear debris [6,7,9] and since the effect ofwear partideson implantfailure due to loosening is known, analy- sis of relative motion mustbe considered an important part of the development ofwear simulation equipment.Motion analysis is preferably done during the siniulator design stage, but can be applied in the evaluadon of an existing simulator design, e.g., if the simulator does not reproduce 'the in vivo wear process.
'This piper is an overview of research activities at Helsinki University of Technology (HUT) in the field of biotribology discussing the results and implications of recent articles deal- ing with the relationship between the relative motion of the articulating surfaces and wear in the prosthetic hip joint.
Examples of other biotribology research activities at HUT indude studies on the effect of lubricant on the wear of prosthetic joint materials [0, ll] and wear of crosslinked polyethylene U2, l3l and alumina [4].The main objective of the research presented in this paper was to provide a quantitative basis for studies on the relationship between the type of relative morion and wear.The wear factor i =(Vln)llLdrc relates the wear rate to the load and the sliding distance.
There are substantial differences between patients regarding load, extent of motion and number of gait cycles per year [ 15, 16, 17, I 8].By computing the wear factof these aspects can be taken into account to quantitatively compare the wear gener- ated in a hip joint simulator with the wear produced in the human body [5] or with another wear test device.In con- trast, the conventional wear rate (expressed as mg or mm' per million cycles) is notwell suited for making such comparisons because the aspects mentioned above are not taken into account.

Methods
The cyclic relative motion of the hip joint can be illus- trated by computing tracks, known as slide tracks, made on the counterface by marker points on the surface of the femoral head or acetabular cup.When referring to the motion of the hip joint, the directions are named according to Frg. l, which is a frontal view of the right hip joint.To compute the relative motion, two coordinate systems were placed at the mutual center of the cup and the head.The reference coordinate system XyZ was fixed relative to the cup and the moving coordinate system r)z was fixed relative to the head.Initially the systems coincided with the X and x axes pointing in the medial, the Y andl axes in the posterior and the Z andz axes in the superior directions.The equatorial plane of the cup, and initially the equatorial plane of the head, was perpendicular to the coronal plane.The values of the cup angle, the head axis angle and the force angle in the coronal plane depended on the simulated case ([29], Table l).
The motion of a marker point was based on Euler rota- tions with the axes of the moving coordinate system acting as rotation axes according to the sequence determined for each simulation.See lig. 2 for an example of applying the rotation sequence FE+AA-+IER.To compute a cup track the marker point was fixed to the head, and to compute a head track the marker point was fixed to the cup.The cyclic FE, AA and IER rotation angles, i.e., the motion waveforms applied to the axes were taken fiom a biomechanical study [9] and fiom published data of diflerent simulator designs 15, 20, 21, 22, 23, 24, 25, 26, 27).The angles of all analyzed cases can be found in [28], Fig.I and [29], Frg.l.Thewaveforms applied in the analyses of the BRM simulators are shown in Frgs.3a and 3d.The slide track computations were described in detail in [28] and [30].
For visualization, the tracks were plotted on a sphere representing the articular surface (Iigs.3b,e).Flattened tracks [30] were plotted to give an overview of the whole pattern (Iigs.3c, f;.In compuring the slide tracks the number of marker points was not restricted, and the location of the points could be fieely selected.Theoretically, there are an infinite number of racks, but too many tracks on one plot would result in a tangled image.Therefore, a limited number of points were chosen to produce an illustrative track pattern. The rotation sequence was determined according to the simulator design.The first rotation changed the orientation of both the other rotation axes of the simulator relative to the acetabular cup (the reference coordinate systern).The second rotation changed the orientation of the one remaining simu- lator axis relative to the cup.The third rotation did not change the orientation of any simulator axis relative to the cup.In two-axis simulator designs, the first step in determining the rotadons was naturally omitted.The sequence for gait was FE-+AA-;IER, since the goniometer used byJohnston and Smidt [9] conformed to this sequence.
The computed slide tracks were verified experimentally.
Head tracks were engraved in the HUT:3 and HUT:BRM simulators rvith sharp pins of hardened steel embedded in the acetabular cup ([28], Flg. 4).The pins were placed very carefully to produce an accurate pattern corresponding to the marker points in the computational model.The simulators were driven for one cycle with load on to produce the tracks.
The lengths, locations and shapes ofthe engraved tracks were compared with the computed tracks.
The effect of the shape of the rotation-prevention lever of the BRM simulators on the head tracks was studied both by engraving the track grooves as explained above, and by letting a stationary pen draw tracks on the head while the simulator r.vas driven without the cup (i291, fig.5).The cup tracks of the BRM-simulator were checked by photographing the movement of head marker points with long exposure times.
To be able to determine the wear factor one must compute the integral / ldx by considering the temporal variation of the Fig. 2: Euler rotations applied to the right hip.Sequence is FE+AA+IER.In the computations, positive rotations corresponded to extension, abduction and internal rotation.The reference coordinate systemXYZwas fixed relative to the cup and the moving coor- dinate svstem rlz was fixed relative to the head.contact resultant force I along the forre track, i.e. the track drawn by I on the counterface (ng.4).In [31] this was done for eleven simulator designs and for level walking as measured by Johnston and Smidt [] by numerically integrating the product of the instantaneous load and increment of sliding distance over a cycle: ILdx=lL;Lx; The vari- ous load waveforms and peak values that were used in the computations can be found in [3 I ].The gait force track was a simplification in that the track was drawn by the resultant force in the position l2o to the verticalintheXZ-plane.While the force track thus drawn was considered adequate for wear factor computations by corresponding to the area of highest contact pressure, fluctuations in the direction of the resultant force during the gait cycle have been reported [7].The time step benveen adjacent points on the load and motion waveforms and consequently on the force track was Al; =TIN.
Additional force track computations included the sliding distance x1 =lLx1u the sliding velocity x =LxilLti and the direction of sliding relative to the cup (.The quantity ( was the direction ofthe velocity vector (and ofthe fiictional shear stress) relative to its direction at I = 0, and was indicated by the instantaneous tangent ofthe force track.The tangent was approximated by a line connecting tr,vo adjacent track points.
The sum of the changes in the direction of sliding along the force track, IIA€rl, was computed.
To take into account the variation of relative motion and contact pressure all over the contact surface, the force track integral JIdx was replaced by a surface integral computed as a sum of track integrals t32].The computation involved Fig. 4: Double-peak waveform of contact resultant force l, plotted along the force track of the HUT-BRM simulator replacing the contact pressure distribution with a large number of discrete forces and summing the integrals computed for each discrete force along their respective rrack.The dis- crete fon:es acted normal to the surface and represented the pressure on a small surface area element.Each integral was computed numerically by summing the product of the instantaneous force value and increment of sliding distance.
This approach was tested for the HUT:.BRM sirnulator in which the dynamic load had a double-peak waveform and an average value of 1.0 kN and was applied along the vertical symmetry axis of the cup (tig.5a).For this simulator, wear test results were available [33] and the relative motion compura- tions had been verified [28].Next, a brief overview of the computations is given.A detailed presentation can be found in [32].
Because in the HUT:BRM rhe load was applied along the symmetry axis of the cup, the pressure distribution could be written as the product of the peak value and a shape function of the angle 0 benveen the position vector and the Z-axis only F = f^u,f @).The wear tests had shown that during steady-state wear the area of contact was hemispherical during the whole load cycle.Therefore a pressure distribution extending to the rim of the cup and varying only in height during the cycle could be selected for the compuration, A different wear rest configuration would naturally result in a diflbrent pressure distribution as the angle between the direction of load and the symmetry axis o{'the cup significantly influences the shape of the distribution [34].In the present case the effect of the shape of the pressure distribution was studied by selecting three commonly used distributions, ellipsoidal, paraboloidal and sinusoidal [35].The hemispherical contact surface was divided into approximately equisized area elements (Fig. 5b), so that the distribution of the discrete forces would resemble the corresponding pressure distribution.
When the contact surface had been divided into surface elements Ar..,,, , tn = l, . .., NE, the discrete forces at each time step l, , i= l, ..., Nwere calculated as q,o(tt) = F(0., t;) tt,* (l) The shape functions of the ellipsoidal, paraboloidal and sinusoidal pressure clistributions were J(r -+e'?/^'?) , | -02 l; and cos 0, respectively.The corresponding values of (3) ii ia) {h} Fig. 5: (a) Schematic drawing of HUT-BRM simulator with ellipsoidal contact pressure distribution p(l) varying with time due to load l,(t).Offset lever position shown with dashed lines. (b) Ellipsoidal contact pressure distribution discretized by large number of normal forces.Temporal variation of each force along its slide track was taken into account in computing an alternative for the integral I Ldx, the sum of track integrals 1.
The pressure value p ( 0,, f,) was calculated in rhe middle of element m, which was also the point of application of q.(t,).
The slide tracks of the discrete forces were computed as described in [28].Finally, the sum of track integrals ^I was computed: -J m=I m=l i=l

Results
The gait slide tracks were mostly irregular ovals, having a characteristic thorn near the end of the gait cycle.Ti"acks with high aspect ratio and small track figures were also present (Frg.6a).The HUT:3 slide tracks were mostly elliptical, but racks with high aspect ratio and short tracks were present, so that the overall slide track pattern resembled the pattern produced by the gait simulation (ng.6b).The motion wave- forms proposed in the ISO/DIS 14242-l standard generated open slide tracks on the center of contact and an elliptical force track (Iig.6c).The BRM slide tracks included a circle, egg and tear shapes, and figures of eight (Fig. 3).In this pape4 only head tracks are shown.The cup tracks ofthe gait, HUT:BRM and HUT:3 simulations were presenred in [28], Frgs. 5, 6, 8.Both head and cup track patterns ofeight contemporary hip sirnulators were shown in [29], Egs.3a-h.
For the HUT:.3 and BRM simulators, the shapes of the individual engraved tracks and the overall engraved track pattern agreed well with the computed tracks ([28], trigs.6a, 7, 8a and 9).The lengths of the engraved force tracks of the HUT:3 and the BRM simulators were 24.2 mm and 34.6 mm, respectively for r = 14 mm.The lengths of the cor- responding computed force tracks were 23.9 mm and 34.3 mm, differences being -1.2 Vo and -0.9Vo, respectively.Borh the computed and engraved force tracks had aspect ratios of I in the BRM simulator and 3.8 in the HUT:3 simulator.
In the BRM simulato4 changing the type of rotation-pre- vention lever changed the slide track pattern.The lever with offset caused tilting and shift of rracks, compared to the zero-offset case (trig.3).The fbrce track however; remained unchanged.In [29], it was shown that a rotation-prevention lever with an offset introduced IER motion in the simulator.
turning the BRM into a three-axis device and causing the change in slide track pattern.This experimental finding was reproduced in the computations by adding the IER motion to the BRM slide track simulation done previously (t291, ng. 6).
The tangent to the slide track indicated rhe direcrion When the sum of track integrals 1 was computed for the HUT:BRM simulator the value first decreased with an in- creasing number of elements [32].Howeve4 the maximum decrease (which occurred for the ellipsoidal pressure distribution) was less than 0.14 7o when the number of elements was increased from 500 to 1000.It was therefore considered that there was no need to increase the number of elements above 1000.Wth dynamic load and 1000 discrere forces the values of 1 and the sum of element forces for the ellipsoidal, paraboloidal and sinusoidal pressure distributions were 3.49r 3.241 3.20r (units are N m when r is substituted in mm) and 1.69L, 1.53L and 1.501 respectively.The lLdx value in the resultant force case was 2.46r Nm. {a} (c)

Discussion
In the hipjoint, the direction of sliding between rhe femo- ral head and the acetabular cup changes continually during walking motion.In the slide track computations the conrinu- ally changing direction of sliding manifested itself as open track figures.Notably, in the three-axis simulators and the purely biaxial HUT:.BRM the force tracks were open shapes ([28], Ftgs.5-6, 8 and [29], Figs.3a-e,4a-e, i).Biaxial simula- tors omitting the abduction-adduction motion, had linear or high aspect ratio force tracks ([29.],Iigs.3f-h).Especially for polyethylene, multidirectional motion is a requirement for producing the amount of wear and the type of wear particles that correspond to clinical findings [7].Therefore, continually changing direction of sliding must be considered a design requirement for simulators to be used for assessing poly- ethylene wear [8].Linear.reciprocating motion has been shown to underestimate the wear rate, while exaggerating the effect ofcounterface roughness on the wear rate [9].
The slide track computations made it possible to visualize the relative motion at the articulation and to oualitativelv compare different simulator designs with each othlr and with the slide track pattern pertaining to walking motion.But the sofrware could additionally offer a basis for quantitative analy- sis of sliding distances, sliding speeds, the integral / Idx used in computing the wear factor k, and the changing direction of sliding during the motion cycle.It was also possible to pro- pose an improved method of computing the wear facto4 taking into account the variation in loading and motion on the whole contact surface of the prosthetic hip joint.
It should be noted that the computational approach pre- sented was notjust a slight modification of an earlier method [39].It was a new systematic method of determining the motion of marker points attached to either the acetabular cup or the femoral head relative to the counterface.Iirst, the Euler sequence rvas determined by investigating how a ronrion about one axis changed the orientation ofthe other rotation axes relative to the cup [28].Subsequently, the rotation angles rvere applied in accordance with the sequence.Furthermore, the method of computing the slide racks was experimentally verified both by engraving patterns of tracks on the femoral head with hardened steel pins embedded into the acetabular cup and by drawing the tracks on the head with a stationary pen while the simulator was run for one cycle [28,29].The lengths and angular extents of the computed tracks were in good agreement with the corresponding values of these experimental tracks [30].For the most popular hip wear simulato4 the BRM, an experimentally verified slide track pattern was presented [28].Additionally, it was shown that changing the shape ofthe rotation-prevention lever changed the slide track pattern due to the addition of IER motion [29].
The influence of the shape of the rotation-prevention lever can be seen in Frg. 3 and Figs.9 and 10 in [28], and Figs. 5 and 6 in [29].
Because the essential features ofwalking motion are inevi- tably captured in the gait waveforms produced by recording hip joint motion during level walking [9], these waveforms were used to compute the reference slide track pattern.However, many of the gait tracks had a thorn towards the end of the motion cycle because of the sudden changes in FE and IER.It is unlikely that a simulator would need to duplicate these sudden changes that last only for a shorl period ofthe total motion cycle, since simulators with smooth circular force tracks (BRM, CTPOD) have been shown to produce realistic polyethylene wear [9,25].If abrupt changes in directions of motion are incorporated in the design, the simulator could be plagued with annoying vibrations.To avoid this problem, the motion waveforms used in the HUT:3 simulator were ideal- izations of the gait waveforms measured by Johnston and Smidt [9].Therefore, smooth slide racks were produced while the overall track patterns for the gait and HUT:3 computations resembled each other (nS. 6).
A factor which is more likely to influence the wear rate than small differences between motion waveforms is the as- pect ratio of the force track.Under similar wear conditions (28 mm diameter polished CoCr head, serum lubricant and similar load) the wear rate produced by the BRM simulator was found to be twice that produced by the HUT:3 simulator [25].In the BRM, the aspect ratio was I (circular force track), whereas in the HUT:3, it was 3.8 (elliptical force track).
Reciprocating motion is known to result in minimal wear [6, 81.Thus, an explanation for the difference in wear rate is that the motion producing an elliptical force track is closer to re- ciprocation than the motion producing a circular force track.
Hip simulators are usually designed so that the FE motion dominates in the high pressure region of the contact.In walking, the FE motion has the largest amplitude and rvill determine the maxirnum extent of the force track figure.
In addition, the distances of the point of application of the resultant contact load to the rotation axes will emphasize the influence of FE and AA motions on the shape of the force track.The load is usually applied near the IER axis and con- sequently the distances to the FE and AA axes are much greate4 which has the effect of diminishing the role of the IER motion in shaping the force track.The AA motion deter- mines the extent of the force track transversely to the FE direction.The phase difference between the FE and AA mo- tions determines how much the force track figure opens up.For example, an elliptical force track is produced with sinusoi- dal waveforms and a phase difference of nl2 befiveen FE and AA.In the high pressure region of the three-axis simulators, the dimensions of the slide tracks in the major and transverse directions were determined mainly by FE and AA, which   resulted in roughly elliptical slide tracks.In the n'vo-axis simu- lators with the IER+FE sequence, reciprocating motion with IER rotation superimposed was produced along the force track.Open slide track figures were produced as the distance fiom the marker point to the IER axis increased, such that the largest elliptical tracks were near the equator ofthe cup ( [29], Frgs. 3f, g).In the high pressure region of the BRM simula- tors, the FE motion did not dominate over the AA motion, and therefore the tracks had aspect ratios close to one.
The slide track software made it possible to investigate the effect of changing motion waveform amplitudes or phases, the Euler sequence, or of omitting certain waveforms to sim- pli$' the design of a simulator.For example, Frg. 7 shows how omitting either AA or IER waveforms changes the gait slide track pattem.If IER is left out, the narrow part of the force track and the other tracks in the middle of the Patternwidens.
Ti.acks near the equator and the vertical column of tracks in the middle become narrower.If AA is left out, the eflect is nearly opposite.The force track and most of the tracks in the middle become narrow figures of eight, while the other tracks remain open loops.Consider designing a simulator starting fiom the gait waveforms and with the Euler sequence for gait.
In the light of Fig. 7, simplifying the three-axis walking motion to two-axis motion (to limit manufacturing costs of the design) is probably better done by omitting IER thanAA.Omitting AAwould result in a force track of high aspect ratio which may influence the amount of wear produced by the simulator.As such, nearly reciprocating motion suggested by high aspect ratio tracks is known to underestimate the polyethylene wear rate compared to the clinical wear rate of polyethylene acetabular cups.In simulators with high aspect ratio tracks on the center of contact, lower wear in this region may be compensated for by higher wear towards the equator of the cup where open tracks were produced.The track pattern of biaxial simulators without AA motion can be contrasted with the pattern of the BRM simulator in which the open tracks are located on the center ofcontact and the narrow tracks and the figures ofeight are located on the equator.
The I.Ldx integrals determined in the force track compu- tations made it possible to convert the wear rates produced by different simulators into wear factors that provide a better basis for comparison between simulators and with clinical values.As an example [31], consider the highly differing wear rates of conventional gamrna-sterilised polyethylene cups against polished 28 mm diameter CoCr heads with serum-based lubricant in the HUT:3 and AMTI sirnulators, 57 and 25 mg per one million cycles, respectively [5, 25].The corresponding wear factors were equal, which can be verified by dividing the wear rate by the density, 0.94 mg/mmn, and by the corresponding/ldxintegrals flable l): (57 mglO.gamglmm3/106/(3.10.14 Nm) = i.4'10-6 mm3/Nm and (25 mg/0.9amg/mm3/I06;/(1.33'14Nm)= 1.4'16-6 mmu/Nm.These values can be compared with the mean clinical wear factor in Charnley prostheses with conventional polyethylene cupswhichwas found to be 2.9.10-6 mm3/Nm [15].The di{Ierence between the clinical and the simulator wear factors can be explained by the considerable scratching of the stainless steel femoral heads in vivo.The range in the clinical wear factors was as large as 0.09-7.2.104 mm3/Nm.
In Thble I it can be seen that there were considerable vari- ations between the different cases regarding the length of the force track x.. Additionally, the variations persisted when x,- was divided with the cycle time T to give the mean sliding velocity along the force track.Most simulators appear to have approximately the same mean sliding velocity as that obtained for gait, except for the BRM simulators, in which it was 66-88 7o higher and one simulator in which it was as much as 2.4 times higher.If the velocity is significantly higher than in gait, there is a risk of altering the lubrication condi- tions and consequently the accuracy of the wear simulation.
The cumulativ.ihu.rg. of direction of sliding, IlaQl, can le seen as a measure of the unevenness orjerkiness of the force track.It could be anticipated that a higher value of IlABrl would indicate higherwear.Howeveq even a small loop along the track causes the value of llAQl to increase by 360" while the increase in wear remains negligible due to the small slid- ing distance.Therefore it appears unnecessary to design a simulator based on very jerky and complicated motion waveforms, i.e., a simulatorwith a very high llAQl value.The conventional way of computing lLdx along the force track does not take into account the contact pressure distribution and the variation of relative motion with location on the contact surface.It also neglects the transverse Pressure component which is perpendicular to the direction of the resultant load l.When the sum of track integrals l was com- puted by replacing the pressure distribution with a large number of discrete forces, the rvhole contact situation was better represented including components of transverse load and variation of sliding distance with location.Howeve4 the value of l was substantially higher than the value of /ldr and consequently the wear factor was lower.For example, an aver- age wear rate of 16.6 mg per one million cycles l'as obtained with the HUT:BRM simulator for gamma-irradiated conven- tional polyethylene cups against polished 28 mm diamerer CoCr heads, lubricant diluted calf serum, and test length 3 million cycles [33].In the conventional rvay of calculating A, rhe resulrwould be 16.6 mg/ (106'0.93mg/mm3.2.46.14 Nm) = 0.52.10{mm3/Nm.With an ellipsoidal conracr pressure dis- tribution and 1000 elements, the surface integral gave a value of 3.49'14 Nm=48.9Nm.Hence, *=0.36.10s mm3/Nm, which is 30 Vo lower compared with the resultant force case [32].It appears that computing the integral /Idx with a single resultant force substantially overestimates the wear factor.In fact, all wear factors published earlier for total hip prostheses are more or less overestimates, as they rely on the assumption of single point contact.
The method of replacing / tdx with 1 requires thar a reasonable estimate of the contact pressure distribution can be computed.The computation would vary from case to case because the angle between the direction of load and the symmetry axis of the cup influences the shape of the distribution [3a].The accuracy of the estimate of the distribution could be improved if additional factors such as the thickness and the material of the cup, the metal backing shell of the cup and the clearances between the head and the cup and between the shell and the cup could be taken into account.
Previous studies [39,40] have presented slide tracks traced by marker points on the head, i.e., cup tracks.In [28] and [29], head tracks were presented for the first time.The head tracks show how points on the cup slide against the counterface, and make it possible to compute the sliding distances of differential area elements of the acetabular cuo.
The sliding of an area element of the cup can be compared io the sliding of the test pin against the counterface disk in the CTPOD simulator.Recently, tests were done with a modified version of the CT'POD simulator in which the siide track of each pin had a different aspect ratio [41].These tests showed that the wear factor of conventional polyethylene rvas pro- portional to /Jnn and that for AR values exceeding 5.5 the l'ear factor decreased to unrealistically low values.Given the relationship ft=f(AR), an interesting future research topic rvould be to investigate whether these test results could be linked with the head track computations to relate the slide track pattern to the wear of the cup.A numerical measure of rvear involving the variation of load, sliding distance and track aspect ratio on the bearing surface could, in principle, be computed by summing the wear rate contributions of each surface element, as shown in (5).

+l
) ,h^6q^a, J m=], In (5), 4,,=ft,,(A\,,) is the wear factor which corresponds to the aspect ratio of the slide track of the discrete force q,,,.The applicability of this approach needs to be verified with comparative rvear tests, as the accuracy of the computation primarily depends on whether experimental wear data pro- duced directly with pin-on-disk type devices can be used for computing wear rate prcdictions for ball-in-socket.configurations.Another factor influencing the computation would be the accuracy of the estimation of the pressure distribution in the joint.6

Conclusions
This paper treated the analysis of relative motion between the femoral head and the acetabular cup in the prosthetic hip and the relation between type of motion and wear.The slide tracks produced by walking motion and by contemporary hip simulators were computed.Computations important for wear assessment included the determination of the integral /ldx along the force track and subsequently the sum of track integrals of forces obtained by discretization of the contact pressure distribution.Based on the research, the following conclusions can be drawn: L The slide track software made it possible to visualize and analyze the relative motion of points on the articular surfaces of the hip joint.In addition, it was possible to in- vestigate the effect ofchanging the Euler sequence or of omitting certain waveforms to simplify the design of a simulator.The effect of changes to the motion waveform amplitudes or phases could be investigated by only generating new waveforms, i.e., without large modifications to the software, By analyzing published data, it was possible to make a slide track overview of practically all contemporary hip simulators ([29] together with [28]).The slide track computations were experimentally verified with the HUT-BRM and the HUT-3 simulators.'fhegait tracks [28] and the tracks of other simulators [29] were com- puted according to the same principles as the verified tracks.'I-he software was used to compute quantities rele-' vant for assessing wear, such as the sliding speed or the cyclic variation of the direction of sliding [31].
2. For the BRM simulators, experimentally verified slide tracks were presented and it was shown that the shape of the rotation-prevention lever affected the slide track pat- tern ( [28], Figs.I and l0).An offset rotation-prevention lever introduced IER motion, turning the simulator into a three-axis device [29].
When the abduction-adduction motionwas omitted in the gait computations (Fig. 7b), a similar effect was seen.
Thus, the simulators can be divided into two groups: those that include and those that omit the AA-motion.Lower wear due to high aspect ratio tracks on the center of contact can perhaps be compensated for by increased wear due to open tracks elsewhere, especially if the con- tact pressure distribution extends to the equator of the cup.One can qualitatively compare simulators, e.g., by noting that under similar wear conditions, a higher force track aspect ratio implies less wear [25].However, to be able to reliably quantify the relationship between the slide track pattern and wear, comparative tests are required to be done with the same type of specimens, lubricant and loading.Published wear test reports differ fiom each other regarding the materials, the type and dilution of serum lubricant, etc. and cannot be used for finding such a relationship.4. Simulators built according to the ISO/DIS 14242-l stan- dard (assuming an FE+AA+IER rotation sequence) would produce a slide track patternwith an elliptical force track and other open slide tracks on the center ofcontact ( [29], Fig. 3e), similar to the slide tracks present in the pattern of the HUT-3 simulator.The design requirement that the direction of sliding shall be continually changing would therefore be fulfilled by simulators meeting the ISO/DIS 14242-l specifications.A similar slide track pattern was, however, produced with only the FE and AA components of the gait motion (Fig. 7a), which suggests that the simulator need not be three-axial to obtain wear conditions similar to those specified in ISO/DIS 14242-1.
The integral Jldx was evaluated for the most used con- temporary hip simulator designs by discretizing the load curve and evaluating the integral along the force track.
The integral makes it possible to compare wear test results between simulators and with clinical wear measurements in terms of the wear factor ft.
The sum of track integrals l was computed by discretizing the contact pressure distribution and summing the contributions of many normal forces distributed all over the bearing surface.For the HUT-BRM simulator the value of 1 was substantially higher than the value of lLdx, suggesting that the traditional way of cornputing the integral with a single resultant force overestimates the wear factor.
lVear tests with the slide track shape as the primary variable were recently done with a modified version of the CTPOD wear simulator [41].The u'ear factor of conventional polyethylene was found to be proportional to/JeR.Future research could elucidate the possibilities for linking this result with the slide track computations to find a quantitative relationship between the slide track pattern and the wear in the prosthetic joint.ton, J. H.: Lubrication and Wear of Ultra-high Molecular Weight Pol2etfulene in Total Joint Replncemznf^s.Trib Int, Vol.31, 1998, p. l7-33.

1
Surface integral replacing I Ld* = sum ofrrack integrals of discrete forces 4,,, L Joint contact resultant force 1n Index to surface elements and discrete forces; rn=1,...,Nn n No, of cycles N No. of time steps per cycle = no. of discrete roration angles NE No. of surface elements = no. of discrete forces p Contact pressure .P,rr"* Peak value of contact pressure Q* Discrete force corresponding to pressure load on sur- xy Length offorce track XYZ Coordinate system (fixed), attached to center point of acetabular cup x)z Coordinate system (moving), attached to center point of femoral head 0 Angle betw.egnr and Z-axis ( Direction of sliding relative to the acetabular cup AA Abduction-adduction AR fupect ratio fupect ratiq.Major dimension divided by minor dimen- .sion ofslide track : BRM CTPOD Biaxial rocking motion hip wear simulator Circularly translating pin-on-disk wear test device FE Flexion-extension Force track Track made on the counterface by the point ofjoint contact resultant force HLIT-3 Helsinki University of Technolory hip joint simulator Mark III HUT-BRM Flelsinki University of Technology BRM simulator IER Internal-external rotation In vivo Within the living organism Slide track Track made on the counterface by point on surface of femoral head or acetabular cup Track integral Line integral of normal force over its slide track 2 Introduction Fig. I : Antero-posterior view of prosthetic-joint with directions of motions shown (right hip, coronal plane)

Fig. 3 :
Fig. 3: Motion waveforms, slide tracks on femoral head and whole pattern of 4l flattened tracks of BRM hip wear simulators wirh direc- tion of rotation prevention Iever indicated.Note that diameter of circle is zrr (not 2r).Force track shown with thick line.(a)-(c)Waveforms (FE, AA) and tracks of HUT-BRM.(d)-(f; Waveforms (FE, AA, IER) and tracks of BRM simulatorwith offset le- ver.
pmax were 0.386 L / f , 0.453 L I f and 0.47 7 L/1.These values were determined by the vertical force equilibrium

Fig. 7 :
Fig. 7: Effect of omitting IER or AA motion on the gait slide track pattern drarvn on the femoral head.(a) IER motion was omitted.Sequence was FE+AA. (b) AA motion was omitted.Sequence u'as FE-+IER.Pattern shown in Fig. 6a includes all motion comPonents.
of sliding at any moment during the motion cycle.If the slide tracks were open loops or figures of eight, rhe direc- tion of sliding changed continually implying multidirectional motion.A circular or elliptical slide track meant rhar the direction of sliding (and consequently the direction of the frictional shear stress) gradually changed 360" during one cy- cle.When the aspect ratio of the track increased, the changes in the direction of sliding became more abrupt near rhe turning points of the track.Ti"acks with a very high aspecr ratio or linear tracks suggested that the motion was reciprocating.In contemporary hip simulators the length of the force track ranged from I .41r to 2.46 r.In the gait simulation the length of the force track was 1.65 r.With the common femoral head diameter of 28 mm, the force track length ranged from 19.7 to 34.3 mm per cycle, the average sliding speed from 19.7 to 49.0 mm/s and lLdx frorn 17.4 to 43.5 N m[31].A summary of the results is given in Thble I .

Table I :
Summary of force track computations for eleven simulators and gait n r to be substituted in mm: b zero-offset lever: 'oflset lever