The roles of contact conformity, temperature and displacement amplitude on the lubricated fretting wear of a steel-on-steel contact
This paper investigates the effect of contact geometry, temperature and displacement amplitude on the fretting behaviour of an aero-turbo oil lubricated cylinder-on-flat contact. To be effective, the lubricant needed both to penetrate the contact and then offer protection. Lubricant penetration into the fretting contact is found to be controlled by two physical parameters, namely (i) the width of the contact that remains covered throughout the fretting test and (ii) the lubricant viscosity. The protection offered by the lubricant (assuming that it has successfully penetrated the contact) is influenced by four physical parameters, namely (i) lubricant viscosity, (ii) traverse velocity, (iii) nominal contact pressure, and (iv) chemical effects. The relationship between the three experimental parameters which were varied in the programme of work (temperature, fretting displacement and cylinder radius) and physical parameters which influence the protection offered by the lubricant film can be competing, and therefore complex wear behaviour is observed. The roles of the various parameters in controlling the wear behaviour are presented in a coherent physical framework.
1. Introduction
Fretting is defined as small amplitude oscillatory motion between bodies that are in contact. Fretting wear dominates when a contact is within the gross-slidingregime (displacement amplitudes up to around 300 µm), with fretting fatigue dominating in partial slip [1]. There is no universally accepted definition regarding the transition between fretting wear (in the gross-sliding regime) and reciprocating sliding wear; however, the damage mechanisms associated with fretting wear are distinct from those associated with sliding wear, because the magnitude of movement in fretting is very small in comparison with the size of the contact area, and, as such, debris development within the contact and its subsequent retention within or ejection from the contact greatly influences wear, both in magnitude and mechanism [2].
Lubrication is a common palliative in reducing friction and wear in a variety of contact types, and it is therefore not surprising that it has also been employed in attempts to reduce fretting wear. In addition to its role as a lubricating fluid, the lubricant will restrict oxygen access into the fretting contact that will in turn affect the development of the oxide debris bed that is commonly associated with fretting wear. Moreover, owing to the very small and reciprocal motion associated with fretting, lubricant penetration into the contact may be poor [3]; it has been argued that this poor penetration is the cause of lubricated fretting contacts having been found to exhibit higher values of the coefficient of friction (COF) than observed in non-lubricated contacts [4]. A number of reports in the literature that address lubricated fretting wear have argued that wear behaviour is influenced predominately by the ability (or otherwise) of the lubricant and oxygen to penetrate the contact [5,6].
Liquid lubrication has been reported to be the most effective form of lubricant in reducing wear in gross-sliding fretting contacts [6]. Publications by both Halliday [7] (with silicone oils) and Shima et al. [4] (using polybutene oils) describe fretting experiments, using different grades of the lubricants across a range of viscosities. Both found that the rate of fretting wear was smaller with lubricants with lower viscosity, with this effect being attributed to the lubricant being able to more effectively penetrate the fretting contact as its viscosity was reduced. Shima et al. [4] also conducted fretting wear tests (over a range of displacement amplitudes) with different lubrication states as follows: without lubricant, with a low viscosity lubricant and with a high viscosity lubricant. They found that for tests conducted with very small displacement amplitudes (in the stick–slip regime), the COF was independent of the lubrication state, and proposed that this resulted from the fact that the majority of the contact was stuck, limiting the opportunity for the lubricant to penetrate that part of the contact. As the fretting displacement increased, the COF became more dependent upon the lubrication state, because more of the contact was slipping that facilitated lubricant penetration. At small displacement amplitudes within the gross-sliding regime, the COF for tests with the high viscosity lubricant was larger than for that observed in unlubricated tests. Both research teams observed that for a fixed fretting displacement amplitude, the wear rate was independent of lubricant viscosity as lubricant viscosity was increased up to a critical value, above which the wear rate rapidly increased [4,7]. It was concluded that this effect was associated not only with reduced lubricant penetration into the contact with increasing viscosity, but also with an associated reduction in effective transport of oxygen into the contact. It was proposed that this reduced oxygen transport would then limit the development of any protective oxide bed forming in the fretting contact.
Other research papers have also highlighted the link between fretting damage and the level of oxygen penetration into a lubricated contact (rather than just focusing on lubricant penetration itself) [8,9]. Wright [8] found that the concentration of oxygen in air was around six times that in a lubricant and furthermore argued that the rate of oxygen diffusion within the lubricant would be approximately proportional to the inverse of its viscosity. Research on fretting with semi-solid lubricants such as grease has indicated that the grease very effectively limits oxygen penetration into the contact, resulting in less oxide debris formation during fretting [5,10]; in similar work, higher rates of wear were observed in fretting contacts lubricated with a grease that had a ‘higher consistency’ [11]. Wang et al. [10] argued that the efficacy of grease lubrication in reducing fretting wear is dependent on the displacement amplitude, in that under large displacement amplitudes, more shear of the grease itself results in the release of base oil, which then is able to penetrate the fretting contact more effectively.
Lubrication of fretting contacts has been observed to influence not only the wear rate and COF, but also the boundaries between different modes of fretting. Under lubrication with both oils [6] and greases [12], the contact regimes were found to shift in relation to those identified in experiments conducted under unlubricated conditions. Liu & Zhou [6] demonstrated that the transition from both partial slip to mixed fretting and from mixed fretting to gross slip in lubricated contacts occurred at higher displacement amplitudes than under equivalent conditions in unlubricated contacts, and they argued that this shift of the transitions to higher displacement amplitudes resulted from (i) the inability of the lubricant to effectively penetrate the contact and (ii) the lubricant also limiting oxygen penetration into the contact. Together, these result in the contact being more metallic in nature, and thus more adhesive; accordingly, higher applied displacements are required to force the contact into the gross-slip regime [6]. Moreover, the displacements associated with the fretting regime boundaries increased with increasing lubricant viscosity [6].
Narayanan et al. [13] have investigated the influence of temperature on lubricated fretting wear of tin-plated copper contacts; again, it was proposed that the lubricant eliminated oxygen from the contact, although in this case, this was found to lead to a reduction in wear. They also observed a higher wear rate for lubricated tests at elevated temperatures, but attributed this to thermal softening of the tin coating. Other studies by McDowell [14], Neyman [15], Neyman & Sikora [11] have also concluded that the wear of a lubricated fretting contact is primarily influenced by the ability of the lubricant to exclude oxygen from the contact, and hence the reduced tendency to create brittle oxidized wear particles which thus results in a reduction in wear rate. The same conclusion, namely that oxygen exclusion would reduce fretting wear, was also reached by Shima et al. [4], but they suggested that this reduction was only observed in cases where the lubricant had been able to penetrate the contact; as such, the effects were most strongly observed with low viscosity lubricants in contacts fretting under low loads and high displacements.
In efforts to better understand lubricant penetration, Imai et al. [16] conducted research on flat-on-flat fretting contacts and studied the ability of grooves to channel oil into the centre of the contact (the geometry of flat-on-flat contacts more effectively restricts oxygen and lubricant penetration than point or line contact geometries which are more commonly employed in experimental work). Fretting wear damage was significant on surfaces without grooves and was reduced substantially when grooves were introduced [16]. Different groove configurations were examined, and it was shown that the reduction in wear rate increased as the spacing between the grooves decreased (i.e. as the lubricant was more effectively transported into the contact) [16].
Most experimental research on fretting is conducted using simplified contact geometries (such as point contact or line contact). Wear rates derived from such experiments have then been used in the prediction of wear in more complex contacts based upon the assumption that the wear rate is independent of the contact geometry itself [17]. However, recent research has cast doubt on the validity of this assumption. Fouvry et al. [18] and Merhej & Fouvry [19] have demonstrated that in unlubricated non-conforming contacts (cylinder-on-flat and ball-on-flat, respectively), the fretting wear rate was dependent on the radius of curvature and decreased as the radius of curvature was increased. It was proposed that this decrease in wear rate was due to the reduced tendency for the debris to be ejected from more-conforming contacts, with the entrapped debris protecting the contact from further wear. More recently, Warmuth et al. [20,21] have made similar observations (again, with an unlubricated cylinder-on-flat contact geometry) and argued that the reduction in wear rate with increasing contact conformity is due to a change in wear mechanism. In a contact with low conformity (namely, a smaller cylinder radius), oxygen was able to penetrate the contact readily, facilitating the formation of oxide wear debris which then flowed out of the contact. In contrast, as the contact became more-conforming (larger cylinder radius), oxygen was more effectively excluded from the centre of the contact, resulting in metal–metal adhesion and metal transfer between the contact faces; at high fretting displacement amplitudes, this transfer was significant and the formation of substantial pits and peaks within the contact (with features measuring up to 140 µm deep) was observed. Fouvry & Merhej [22] also found that for large radius specimens (i.e. a more-conforming contact), a metallic region in the centre of the contact developed and this was also attributed to effective oxygen exclusion.
In the light of the strong influence of contact conformity on the rates and mechanisms of fretting wear in unlubricated contacts and the general understanding that rates and mechanisms of wear in lubricated fretting contacts are influenced both by lubricant penetration into the contact and the effect of the lubricant in limiting oxygen into the contact, this study examines the effect of contact conformity in lubricated fretting wear. As part of this study, the effect of temperature and displacement amplitude on the rates and mechanisms of wear across is also addressed (noting that temperature markedly influences the lubricant properties). All experiments were conducted in a static oil bath, using a fully formulated, high-performance, synthetic ester-based turbo oil which is commonly used in aeroengine applications.
2. Experimental procedure
2.1. Materials, test conditions and procedures
Fretting wear experiments were conducted on a high strength steel (specification BS S132). Heat treatment of the steel was conducted prior to machining of the specimens to size (the details of this heat treatment process can be found elsewhere [23]) and resulted in the specimens having a hardness (HV20) of 465–480 kgf mm−2. A high thermal stability aviation turbo oil (BP2197) was used as the lubricant throughout this work; BP2197 contains tricresyl phosphate (TCP) as an extreme pressure (EP) additive.
The specimen pair was assembled in a cylinder-on-flat configuration, as shown in figure 1. The flat and cylindrical specimens were ground on a linear and cylindrical grinder, respectively. The flat and cylindrical specimens had a roughness (Ra) of 0.1 to 0.3 µm and 0.4 to 0.7 µm, respectively. The flat specimen is mounted on the lower specimen mounting block (LSMB) which is stationary and the cylindrical specimen is mounted on the upper specimen mounting block (USMB). Before the specimens were brought into contact, the surrounding oil bath was filled with approximately 20 ml of oil, so that the contact was submerged with the oil level being approximately 1 mm above the contact region. The USMB was loaded through a dead weight configuration, with the normal load that results being termed P. During assembly of the fretting pair, the USMB is able to rotate around the axis of fretting; application of a small load to the pair at this stage ensures that the axis of the cylindrical specimen lies in the plane of the flat specimen, whereupon the rotation of the USMB is fixed. The main components in the rig used for the fretting experiments are illustrated in figure 2. Both the USMB and the LSMB are heated via cartridge heaters. The motion of the USMB (and hence the cylindrical specimen) is created by a force generated by an electromagnetic vibrator (EMV). The motion of the USMB (and hence the cylindrical specimen) is created by a sinusoidally varying force generated by an EMV. The displacement, Δ, of the USMB is monitored by a capacitance displacement sensor which is mounted to the LSMB and is recorded throughout the duration of the test. The amplitude of the force input is controlled to achieve a set displacement amplitude, Δ*, for the test, although (owing to compliances in the system) the displacement–time profile is not sinusoidal.
Figure 1. Crossed cylinder-on-flat specimen configuration used in fretting tests.
Figure 2. Illustration of the main components of the fretting apparatus used in this study.
The lateral force, Q, is measured (and recorded) throughout the test by a piezoelectric load cell which is connected to the quasi-stationary LSMB. To protect the force and displacement sensors, water cooled members separate them from the heated area of the rig as shown in figure 2. Both displacement and load sensors have been calibrated (both externally and in situ) in static conditions. The load and displacement signals are sampled at a rate of two hundred measurements per fretting cycle for all of the experiments.
The behaviour of the contact can be monitored throughout the test by examination of the fretting loops; an idealized gross-slip loop is plotted in figure 3. The displacement of the USMB is measured, but it must be noted that this is not the same as the slip in the contact; there is compliance in the system which physically separates the contact from the point of measurement, and hence the measured displacement amplitude, Δ*, is always slightly larger than the contact slip amplitude, δ*. The actual contact slip amplitude (δ*) can be derived by measuring the displacement at zero force, as indicated in figure 3.
Figure 3. Schematic diagram of an ideal fretting loop in the gross-sliding regime, illustrating parameters defined in the text.
The COF is defined as follows:
Table 1.
Summary of the fretting test parameters.
2.2. Estimation of wear volume and surface topography
After completion of a fretting experiment, specimens were lightly swabbed with industrial methylated spirit to remove loose debris and residual oil, thus leaving any debris that was adhered to the specimen in place. To evaluate their topography, the wear scars on both the flat and cylindrical specimens were scanned using a Bruker contour GT-I interferometer, which has a vertical resolution of approximately 0.15 nm and a lateral resolution of 4 µm. The scan area was always larger than the wear scar (figure 4) to allow the reference surface (representing the surface before wear occurred) to be defined by interpolation from the surfaces outside the wear scars (as proposed by Elleuch & Fouvry [24] and illustrated in figure 5). The volume below each reference surface was regarded as the wear volume (
Figure 4. Illustration of the area measured by profilometry with respect to the fretting wear scar.
Figure 5. Illustration of the definition of wear and transfer volumes in a fretting scar for both the flat and cylindrical specimens.
2.3. Characterization of wear scars and debris
Scanning electron microscopy (SEM) was used to characterize the nature of the wear scars, using a Philips XL30 SEM. Back scattered electron (BSE) images were used to distinguish oxide from metallic material (oxide which forms in the wear scar has a lower average atomic number, resulting in a lower brightness in BSE imaging than the steel). The identification of oxide was confirmed qualitatively by energy-dispersive X-ray spectroscopy (EDX) analysis.
2.4. Covered contact width and lubricant viscosity
In this experimental programme, the vast majority of tests were conducted across a range of contact geometries with an applied load of 250 N. Assuming a Young's modulus of 207 GPa and a Poisson's ratio of 0.28 for this steel [25], Hertzian contact mechanics for a line contact was used to predict the contact semi-width (b) and initial maximum Hertzian contact pressure (σH,max). Values of b and σH,max are presented in table 2 for the four contact geometries employed.
Table 2.
Hertzian contact semi-width, initial maximum Hertzian contact pressure and lower-bound estimate of the covered semi-widths for the contact geometries and displacement amplitudes examined (P = 250 N).
As the displacement amplitudes that are being studied are comparable in size to the elastic contact width, under certain conditions there exists a portion that remains covered throughout the duration of the test as illustrated in figure 6 and described by equation (2.3):
Figure 6. Illustration of contacts over a cycle of motion with covered width (top) and without covered width (bottom).
As such, the values of the covered semi-width, bc, throughout the test cannot be precisely determined, but can be estimated via equation (2.3) if (owing to their similarity) δ* can be assumed to be equal to Δ* and it is assumed that the contact width can be assumed to be the initial Hertzian contact width (b). With these assumptions, the estimate of bc will always be a lower bound (because the actual contact width will always be greater than the Hertzian contact width and the slip amplitude will always be less than the displacement amplitude). Estimated lower-bound values of the covered widths for the different combinations of cylinder radii and Δ* examined in this work are presented in table 2.
The literature indicates that lubricant viscosity has a significant influence on the behaviour of a lubricated fretting contact (as outlined in §1), with the lubricant viscosity itself being strongly dependent upon temperature. The viscosities of the turbo oil employed (BP 2197) at −40°C and 100°C were supplied by the manufacturer [28], and from these, the viscosity at other temperatures can be estimated [29]. The measured and estimated values of viscosity as a function of temperature are presented in table 3.
Table 3.
Estimates of the viscosity of the lubricant at the test temperatures alongside those provided by the manufacturer.
3. Experimental results
3.1. Characteristics of the fretting motion
The fretting motion is delivered via the EMV. Examples of displacement–time loops (figure 7) indicate that in each case, there is a period where the contact is essentially stationary (the small slopes of the curves in these regions are associated with system compliance), followed by a period of gross-sliding (where the slope is much larger in magnitude). For the cases where Δ* = 25 µm, the speed in the sliding period is approximately constant for the vast majority of the period of slip, with a value of approximately 5 mm s−1. When Δ* = 100 µm, the speed in sliding is slightly higher for the case when R = 6 mm (19 mm s−1) than it is when R = 160 mm (15 mm s−1).
Figure 7. Applied displacement as a function of time for lubricated contacts under ambient conditions and an applied load of 250 N: (a) R = 6 mm and Δ* = 25 µm; (b) R = 160 mm and Δ* = 25 µm; (c) R = 6 mm and Δ* = 100 µm; (d) R = 160 mm and Δ* = 100 µm. All data are taken from approximately midway through the fretting test at around 50 000 cycles.
The fretting characteristics are commonly represented by fretting loops, examples of which are presented in figure 8. Each of these loops represents behaviour very close to the end of the respective experiment (following 99 000 fretting cycles). It can be seen that the behaviour is similar with both contact conformities; variations of the force across the sliding regions of the loops can be seen which are generally attributed to the development of geometrical features associated with wear [26,30,31]. In addition, the steep sides of the loops represent the contact stiffness, with a value of approximately 38 N µm−1being observed irrespective of the lubricant temperature and contact conformity.
Figure 8. Fretting loops at 99 000 cycles for the lubricated fretting tests conducted at temperatures of ambient, 85, 150 and 230°C, with a 100 µm applied displacement amplitude: (a) R = 6 mm; (b) R = 160 mm.
3.2. Effect of contact conformity
Experiments were conducted with cylinders of different radii (6, 15, 80 and 160 mm) at ambient temperature and with a 25 µm displacement amplitude. Figure 9shows the evolution of the shape of the fretting loops with number of cycles through the tests for both the 6 mm and the 160 mm radii cylinders. It can be seen that the loops from the least-conforming contact do not change significantly throughout the test, whereas those from the most conforming contact indicate unsteady behaviour in the initial stages (with higher tractional forces) which then settles to more steady behaviour after approximately 350 000 cycles.
Figure 9. Time progression of fretting loops for lubricated fretting tests conducted at ambient temperature with a 25 µm displacement amplitude and a 250 N normal load: (a) R = 6 mm; (b) R = 160 mm.
Figure 10a is a plot of the wear and transfer rates for the experiments conducted with different contact radii. The wear and transfer rate for the least-conforming contact (6 mm cylinder radius) was very low, with rates of less than 0.1 × 10−5 mm3 N−1 m−1. The wear and transfer rates increased as the contacts became more conforming, and were around 15 times larger for the contacts with the 160 mm cylinder radius in comparison with those with the 6 mm cylinder radius. The difference between the wear and transfer rates also became larger with increasing cylinder radius, indicating that material loss from the contact was increasing.
Figure 10. (a) Plot of wear rate (
Figure 10b shows the COF over the duration of the test for the different contact conformities examined. The couples with 6 and 15 mm cylinders exhibited low values of COF with little change over the test duration; in both cases, a small initial rise was observed (with some instabilities in this region) which then subsequently fell to a steady value. In contrast to these, the tests with the 80 and 160 mm radius cylinders initially exhibited high and unstable values of COF over extended numbers of cycles. The period of instability, and the magnitude of the instability both increased as the conformity of the contact increased. In addition, the steady-state COF also increased with increasing contact conformity.
BSE images of the fretting wear scars on the flat specimens are presented in figure 11 for the experiments conducted with different contact conformities (figure 11a–d at low magnification and figure 11e–h at higher magnification). Images of the fretting wear scar from the contact with the 6 mm radius cylinder show that the scar is patchy and damage appears to be in the form of scratches in regions that were high points within the contact. For the tests with 15 mm radius cylinders, the scratches become less pronounced and areas of plasticity and material transfer can be observed in the centre of the contact. Despite the nominal contact pressure reducing with increasing cylinder radius (table 2), the wear scars from the tests with the 80 and 160 mm radius cylinders exhibit many much larger areas of plasticity and material transfer. All the damaged regions are clearly primarily metallic, because the contrast associated with oxide-based debris beds when observed in BSE imaging is absent. However, some very dark spots are observed in the damaged regions which are shown to be carbon-rich.
Figure 11. (a--h) BSE images of the top surface of the wear scars on the flat specimens at low and high magnification following experiments conducted on cylinders with radii of (a, e) 6, (b, f) 15, (c, g) 80 and (d, h) 160 mm, at ambient temperature with Δ* = 25 µm and P = 250 N.
Figure 12a–d are the corresponding profilometric contour plots of the surfaces of the fretting wear scars on the same flat specimens. The wear scars for the tests with the 6 mm radius cylinder and the 15 mm radius cylinder extend across the entire width of the specimen; however, in both cases, the damage is very shallow (typically of the same order of magnitude as the surface roughness). In contrast, damage to the surface of the flat following fretting against a cylinder with an 80 mm radius is dominated by a few large deep pits, with one as deep as 50 µm. For the wear scar in the most conforming fretting contact (160 mm radius cylinder), the pits are both deeper and wider than those observed in the contact with the 80 mm cylinder, but are again isolated and few in number; the largest pit exhibits a maximum depth of 95 µm.
Figure 12. (a--d) Topography of the surface of the flat specimen for the experiments conducted on cylinders with radii of (a) 6, (b) 15, (c) 80 and (d) 160 mm, at ambient temperature with Δ* = 25 µm and P = 250 N. The dimensions on each profile are in millimetres; the colour scale on the left is in micrometres and relates to the height of the surface.
3.3. Fretting behaviour of the least-conforming contacts in lubricated fretting
3.3.1. Effect of lubricant temperature with small fretting amplitude
Experiments were conducted with the least-conforming contact geometry (6 mm radius cylinders) with a 25 µm displacement amplitude and with a range of lubricant temperatures (ambient, 85, 150 and 230°C). Figure 13a presents the wear and transfer rates for the experiments conducted at different test temperatures. It can be seen that at ambient temperature, both the wear and the transfer rates were very low. As the temperature was increased to 150°C, the wear rate increased dramatically (to more than 10 times its value at ambient temperature), but the transfer rate changed very little. However, the wear rate for the test conducted at 230°C was almost half of that observed in the test conducted at 150°C (again, there was little change in the transfer rate).
Figure 13. (a) Plot of wear rate (
Figure 13b shows the evolution of the COF throughout the test as a function of test temperature. The COF was observed to increase with increasing test temperature up to 150°C, with the COF in the test conducted at 150°C being around twice that observed in the ambient temperature test. On further increasing the test temperature to 230°C, the COF rose rapidly to a maximum value of 0.33, and then dropped (again, relatively rapidly) to a value of 0.25.
BSE images of the fretting wear scars on the flat specimens from the experiments conducted with the least-conforming contact at different lubricant temperature are presented in figure 14 (figure 14a–d at low magnification and figure 14e–h at higher magnification). The surface of the specimen from the test conducted with lubricant at ambient temperature exhibits scratching behaviour with little sign of surface plasticity; damage appears to be occurring only at the high points within the contact, leaving other areas undisturbed. At higher lubricant temperatures (85 and 150°C), wave-like prows were observed to form on the surface; these were most pronounced in the experiment conducted with a lubricant temperature of 150°C. As the lubricant temperature was further increased to 230°C, these prow formations were no longer in evidence, with the surface exhibiting significant scratching instead.
Figure 14. (a--h) BSE images of the top surface of the wear scars on the flat specimens at low and high magnification following experiments with 6 mm cylinders conducted at (a, e) ambient, (b, f) 85, (c, g) 150 and (d, h) 230°C, with Δ* = 25 µm and P = 250 N.
Figure 15a–d are the corresponding profilometric contour plots of the surfaces of the flat specimen fretting wear scars. It can be seen that all of the wear scars extend across the full width of the contact. The wear scar from the test conducted at ambient temperature exhibits very little surface damage, it being less than 1 µm deep. Following testing at 85 and 150°C, the wear scars became wider and deeper, having a depth of around 3 and 4 µm, respectively. However, following testing at 230°C, the resulting wear scar was shallower again, having a depth of only 2 µm (although its width was similar to that of the scar developed following testing at 150°C).
Figure 15. (a--d) Topography of the surface of the flat specimen following experiments with 6 mm cylinders conducted at (a) ambient, (b) 85, (c) 150 and (d) 230°C, with Δ* = 25 µm and P = 250 N. The dimensions on each profile are in millimetres; the colour scale on the left is in micrometres and relates to the height of the surface.
3.3.2. Effect of lubricant temperature with higher fretting amplitude
To compare directly with the results presented in §3.3.1, experiments were conducted with the least-conforming contact geometry (with 6 mm radius cylinders) with lubricant temperatures of ambient, 85, 150 and 230°C and a higher displacement amplitude of 100 µm. Figure 16a is a plot of the resulting wear and transfer rates. By comparing with figure 13a, it can be seen that the rates of wear follow the same pattern (with very similar values) to those derived from tests conducted at the lower displacement amplitude of 25 µm. The transfer rate is, however, slightly lower at the higher displacement amplitude, and again remains low across the range of test temperatures examined. Figure 16b shows the development of COF over the whole duration of the tests conducted with the different lubricant temperatures with the larger (100 µm) displacement amplitude. Again, by comparing with figure 13b, similar trends can be seen, although the final values of the COF are different at the two displacement amplitudes. Notably, at both displacement amplitudes, the COF was smallest for the test conducted at ambient temperature, and largest for the test conducted at 150°C.