- Email: [email protected]
Fretting maps and fretting behavior of some F.C.C. metal alloys
Wear, 138 (1990)
153
153 - 167
FRETTING MAPS AND FRETTING METAL ALLOYS*
BEHAVIOR
OF SOME F.C.C.
OLOF VINGSBO+, MIKAEL ODFALK and NING-E SHEN Department (U.S.A.)
of Mechanical Engineering,
University of Houston, Houston, TX 77004
Summary
The aim of the research was to achieve a systematic knowledge of fretting contact and damage over a wide range of testing conditions, with emphasis on the relations between different testing parameters. In particular, the effect of material parameters on the fretting characteristics was to be studied with the aid of fretting maps. The work has concentrated on metal specimens. Three alloys were chosen to represent different hardness levels for essentially the same crystallographic structure: pure copper, a solution-hardened Cu-Si alloy and an austenitic stainless steel. Fretting testing was performed in a controlled atmosphere of dry air at a pressure of 500 Torr. Critical amplitudes of the tangential force and displacement for transitions between different fretting regimes were found and used to produce fretting maps. In addition, microstructural studies were performed with the aid of light and scanning electron microscopies to clarify the fretting mechanisms occurring. It was found that, in the present materials, there is a gradual transition from regime I (low damage) into regime II (mixed stick-slip and fretting fatigue), rather than a well-defined boundary. The boundary between regimes II and III, however, was established and plotted in fretting maps. Fretting maps are compared and discussed together with the influence of the hardness for the three materials. Characteristic fretting scar structures are shown and are related to mechanisms of fretting damage for each regime in all three materials.
1. Contact conditions in fretting Fretting is defined as “low amplitude oscillatory Depending on experimental conditions, surfaces”.
sliding between fretting contact
tribo may
*Paper presented at the International Conference on Wear of Materials, Denver, CO, U.S.A., April 8 - 14,1989. ‘Also at Uppsala University, School of Engineering, Materials Science, Box 534, S-751 21 Uppsala, Sweden. 0043-1648/90/$3.50
@ Elsevier Sequoia/Printed
in The Netherlands
154 Contact Circle
ElaStlC
Solid Sphere \
Solid
Block Slip
Within
---f--z? a N Annular
FTArea
: 8’ < r C a
Fig. 1. Elastic contact conditions of fretting for spheric symmetry: N, applied normal load; FT, applied tangential force; r, distance from the center of the contact circle; a, radius of the contact circle; a’, radius of the sticking circle.
cause damage of one or more of three types: fretting fatigue, fretting wear and fretting corrosion. In its most basic form, a fretting tribosystem is characterized by contact between convex surface elements, as first suggested by Mindlin in 1949 [ 11, and later discussed by other researchers [ 2 - 41. Mindlin’s model is shown for circular geometry in Fig. 1. The contact area is divided into two zones: a central area with a high contact pressure, where the friction may be high enough to prevent sliding (the sticking zone), and a surrounding zone of lower pressure and friction in which sliding will occur (the slip annulus). These conditions of mixed sticking and slip are characteristic of fretting at low displacement amplitudes. The “stick” zone will shrink with increasing applied tangential force (and displacement) until eventually slip occurs over the whole contact area (this is the “point of incipient slip”). The boundary region between sticking and slip is characterized by sharp stress peaks (tensile, shear or compressive, depending on its location with respect to the sliding direction) which oscillate over the contact area during the tangential force oscillation and promote crack nucleation and growth. Fretting studies generally suffer from the fact that the number of active parameters and variables, representing testing as well as material and atmosphere conditions, is very large, which makes interrelations between the different parameters difficult to distinguish and understand. This is the background of the development of a new type of testing equipment at the University of Houston which allows the displacement amplitude to be varied independently within large intervals of vibrational frequency and applied loads in controlled atmospheres. This unique facility has improved the experimental possibilities dramatically and has led to the development of a new type of representation of the fretting process in so-called fretting maps. A fretting map (see the schematic examples in Fig. 2) is a diagram, showing in two variables the different fretting damage regimes
155
Fig. 2. Schematic examples of fretting maps. Regimes of low damage (I), fretting fatigue (II) and fretting wear (III). N, normal load; D, displacement amplitude; f, frequency; T, tangential force amplitude.
representing low damage (regime I), stick-slip and fretting fatigue (regime II), gross slip and fretting wear (regime III). The key elements of a fretting map are the boundaries between the regimes. The influence of testing and material parameters on the shape and location of the regime boundaries is an issue of central interest for future fretting research. Once the relevant fretting maps for a given tribosystem are established, it is, in principle, possible to control the fretting process by choosing the active parameters so as to promote or suppress a certain type of damage. A special study of fretting maps has been published separately [ 51.
2. Experimental
details
2.1. Testing equipment The testing equipment is designed for a crossed-cylinders specimen configuration. One vertical, longitudinally vibrating cylindrical specimen is clamped between two horizontal, stationary half-cylinders. The normal load is applied to the stationary specimens with the aid of dead weights acting on a lever system. This configuration offers several advantages compared with other possible specimen and loading arrangements such as sphereon-flat for example. The most important advantages are the ease in (i) specimen manufacturing and polishing, (ii) application of normal as well as tangential loads, and (iii) keeping alignment under normal load (doublesided clamping). The fretting motion is generated by a sine generator, feeding an electromagnetic exciter-vibrator with independently variable frequency and vertical displacement, directly connected to the central specimen. The signal from an accelerometer mounted in the vibrating specimen holder is used in a feed-back compressor loop to monitor the output signal from the sine generator in order to control the fretting amplitude independently of variations in friction. The specimen displacement is obtained by integration of the accelerometer signal, and is continuously recorded on a digital oscilloscope together with the tangential force measured by a transducer.
156
Fore. I ! Tranrducor
Vfbratk?Q ; Specimen ,
Pyrex Boll Jar
,Ultrrronlc
Horn
I
S*mt-Cyllndric41 f Stattorury I Sp*ctmenr L -----_-_I --_-T : Spoclmonr, Flow
M*t*r
Fig. 3. Schematic
The imposed vibrating absolute vibrator) The main
Collar wlth Food-through.
PIJrQe/
/
Nindl*
V8cuun
diagram of the testing equipment.
elastic motion of the stationary specimens, because of the forces by the vibrating specimen, is less than 10% of the motion of the specimen. Thus, the reIative displacement is close to the recorded translation, The fretting equipment (except for the electromagnetic is enclosed in a bell jar and has a system for atmospheric control. layout is shown in Fig. 3.
2.2. Materials A simple investigation of the has been made by comparing those of hardness but essentially the same was varied in steps by studying three
influence of the material parameters for metallic alloys of various degrees internal microstructure. The hardness different f.c.c. structures: pure copper,
TABLE 1 Data on the materials tested AllOY
Copper Cu-Si Stainless steel
Twe
UNS Cl 10 UNS C655 AISI 304
Composition
Hardness
(wt.%)
WV 1(-k)
99.9 Cu 97 Cu; 3 Si 18 Cr; 8 Ni; 0.08 C; 1 Si; 2 Mn
80 2 5 I.80 rt 15 235 k 15
157
a solution-hardened Cu-Si alloy and an austenitic stainless steel. The composition and other data of the materials are given in Table 1. The microhardness measured at room temperature and with a load of 10 g varies between 80 and 235 HV for roughly the same grain size. The grain size of the copper alloys is 65 I.trn and that of the stainless steel 15 pm. 2.3. Fretting testing Before fretting testing, the machined specimens were ground and polished to a shiny surface. Two normal force levels were chosen with applied dead loads of 3.4 and 11.4 N. In principle, the fretting frequency of the present test rig is variable independent of the applied forces and the displacement amplitude. In practice, however, the power exerted on the specimens, is limited by that available through the power supply of the electromagnetic vibrator. For a given normal force the maximum displacement ~plitude decreases towards zero with increasing frequency. Thus, there is a maximum frequency for a non-vanishing displacement, which is set by the power supply rather than by the sine generator or the inertia of the vibrator. In the present study a frequency limit of 1500 Hz was chosen. 2.4. Recording fretting maps Identifying the regime boundaries which represent critical values of the testing variables for the transition from one regime to another, is a key issue in fretting research. The experimental procedure is not a trivial matter and in fact, a substantial part of the present work had to be devoted to developing and evaluating several techniques. At present it seems that different methods may be optimal for different materials and atmospheres. The final method established is described as follows. Two signals corresponding to displacement d and tangential force FT were simultaneously recorded us, time t on the screen of an oscilloscope. Each experiment at preset, constant normal load N and frequency f was started with a low vibrator power and tangential force, generating sinusoidal FT(t) and d(t) curves (Fig. 4(a)) with amplitudes T and D. The power was gradually increased searching for the first distortions ~co~espond~g to the critical amplitudes Tl and Di) to appear. In the present materials, however, it turned out that generally there was a gradual transition from regime I to regime II (see Fig. 4(b)) rather than a well-defined boundary (cf. the
Fig. 4. Characteristic examples of recorded FT(~) and d(r) curves. Material, Cu-Si; f= 800 Hz; N = 11.4 N. (a) d < D,;(b) Dl< d < D2; (c)d > Dz.
158
elastic limit in a stress-strain curve), and no Tl and D, amplitude values were recorded. During the continued increase of the vibrator power, both D and T increased until the critical amplitude T2 was reached. For a still higher applied power, however, the tangential force amplitude dropped slightly while the displacement amplitude continued to increase (Fig. 4(c)). This is in fact exactly what is expected for the transition from regime II (mixed stick-slip with partial static coefficient of friction) to regime III (gross slip with kinetic coefficient of fretting). However, it implies some practical difficulties in recording T, and D, because once the drop in T has occurred it is no longer possible to go back and find the original T2 and corresponding D, values. This can be explained by the deterioration of the fretting surfaces after the point of incipient slip. Therefore, the following procedure was adopted. The vibrator power was gradually increased while monitoring FT and a reading was taken of the maximum amplitude value before the drop, i.e. T = T2. The fretting contact was then repositi6ned on a fresh specimen surface and the procedure repeated with one important change, this being that the increase in power was stopped just before the previously recorded T2 was reached, i.e. at T < T2. The corresponding D value was recorded and defined as the critical displacement amplitude D,. This method will underestimate D2 slightly but it is reproducible and can be used in comparisons. From the accumulated data, all types of fretting maps, i.e. D(f), T(N), by pairwise combinations of the parameters. f(N) etc., can be constructed For ease of comparison, the present report is mostly based on D(f) and T(f) maps. Each datum point of the fretting maps corresponds to 3 or 4 runs and the scatter is of the order of ?lO%. It should be pointed out that the present work does not aim at precision in numerical results, but emphasizes exploring the potential of the fretting map representation. 2.5. Atmosphere control In the fretting map experiments covered in this paper, the atmosphere was dry air at a pressure of 500 Torr with the relative humidity less than 10%. Before each test run the specimen chamber (bell jar) was evacuated by a rotary pump to less than 5 X lo-* Torr. Air was then leaked into the chamber through a filter. 2.6. Scar morphology studies To clarify the active fretting mechanisms, the fretting scars characteristic of the different regimes were studied systematically. Separate test series were run with the parameters chosen to correspond to the conditions within each regime. The scar sizes were measured and the scar characteristics were studied with the aid of light and scanning electron microscopy.
159
3. Results and discussion 3.1. Fretting maps 3.1.1. Critical tangential force amplitude Figures 5(a) and 5(b) compare the frequency dependence of Tz for all materials at each clamping load. It can be seen that T2 increases with frequency indicating a strain rate hardening effect which is strongest in the Cu-Si at the higher load (Fig. 5(b)). A general observation is that Tz increases with increasing load at all frequencies in all three materials. This reflects the obvious fact that it takes a higher tangential force to achieve gross slip under a higher normal (clamping) load. Comparing the different materials indicates that, for frequences above 10 Hz, this effect is stronger in softer materials.
;J--yqg---q 100
10'
102
IO3
f 0-W
(a)
100
IO’
102
f (Hz)
(b)
Fig. 5. Z’(n fretting maps for all three materials: N= 3.4 N; (b) N = 11.4 N.
103 X, Cu; +, Cu-Si; 0, stainless steel. (a)
3.1.2. Fretting friction coefficient The ratio of the oscillating tangential force to the constant normal load represents an undetermined friction coefficient characteristic of the transient mechanisms during the application of external sliding. Its amplitude value nf = T,/N corresponds to the fully developed static friction and represents a “fretting coefficient of friction” which is plotted as a function of frequency in Fig. 6. It is obvious from the definition that pf varies with Tz and the same type of increase with frequency as in Fig. 5 is seen in Fig. 6. The differences between the present materials are small, particularly at the lower load (Fig. 5(a)). This is to be expected because
100
(a)
10'
102
f (Hz)
103
100
@I
10'
102
103
f (Hz)
Fig. 6. Fretting friction coefficient pf as a function of frequency f for all three materials: X, Cu; +, Cu-Si; 0, stainless steel. (a) N = 3.4 N;(b) N = 11.4 N.
160
they are all metals with the same surface finish. (One exception is the low pf value of the steel, shown for the high frequencies in Fig. 6(b).) As in unidirectional sliding, variations with normal load are small. The fact that the static coefficient of friction is higher than the kinetic coefficient is of no great significance for the wear behavior in unidirectional sliding because it is a transient phenomenon that occurs only once, i.e. during the initiation of sliding. In fretting contacts, however, the friction peak pf = T,/N induces both tensile, shear and compressive stress maxima around the contact area every cycle. In fact this is one of the main reasons why fretting fatigue is a severely deteriorating process in materials even at apparently harmless, low amplitude oscillations. 3.1.3. Critical displacement amplitude D2(f) maps for both levels of N are reproduced
in Fig. 7. It might be expected that the displacement would drop with increasing normal load; however, Fig. 7 shows that D, like T, increases with N in all three materials. This is because both T and D are controlled by the applied power of the vibrator irrespective of N. 5
10
4 +
$3 -cu2 n
x tf 1
0 i 100 (a)
+ xi P
q
10'
2 f (tiJO
103
0 100 @I
10'
102
103
f (Hz)
Fig. 7. D(fl fretting maps for the three materials: N = 3.4 N; (b) N = 11.4 N.
X,
Cu; +, Cu-Si; 0, stainless steel. (a)
The frequency dependence in the steel is an example of two mechanisms competing, these being often observed to affect the strain hardening properties during plastic deformation. As shown in Fig. 7(a) the displacement amplitude decreases with increasing frequency in the low frequency interval for the steel. This indicates a strain-rate hardening mechanism. As the displacement is connected with local plastic deformation in the contact zone, particularly under mixed stick-slip conditions, D should be affected by frequency-dependent strain hardening in the same way as suggested for the increase in T by an increase in frequency as in Fig. 5(b). For higher frequencies, however, the temperature increase associated with increased frequency implies a softening that dominates over the corresponding strainrate hardening. This type of hardening-softening balance generates the U-shaped D(f) curves of the stainless steel at the lower load (Fig. 7(a)) and of the Cu-Si at the higher load (Fig. 7(b)). The more complex waveform of the remaining curves of Fig. 7 could, in principle, be ascribed
161
a repeated hardening or softening dominance in successive tervals. However, that seems to be rather an overexploitation unless a reason for the repeated shifts can be found.
frequency inof the idea,
3.1.4. Influence of material parameters The influence of material parameters can be studied by plotting critical D and T values (for constant frequency) us. hardness as shown for example in Fig. 8. The critical displacement decreases clearly with hardness for both normal load levels and all frequencies, as shown for the higher load in Fig. 8(a). (The influence of the frequency, however, is not well described by this type of graph.) Tz decreases slightly with hardness for the higher normal load (Fig. 8(b)), but for the lower load the situation varies differently for different frequencies.
0 (a)
100 Hardness
200 (HVlOp)
300
0
100 Hardness
200 (HV 10~)
300
@I
Fig. 8. Critical amplitudes plotted us. hardness for different frequencies. (The three materials are represented by three hardness values along the H axis (cf. Table l).) N = 11.4 N; X, 10 Hz; +, 50 Hz; 0, 300 Hz;A, 800 Hz. (a) Dz; (b) T2.
3.2. Fretting mechanisms The crossed-cylinder specimen configuration gives a near-circular contact area, which can be directly related to the fretting model described in Fig. 1 and creates four scars per test. The results of the structural studies are presented below with representative scanning electron micrographs from one of the two scars of the stationary specimens. All the micrographs correspond’ to the same testing conditions: the higher normal load and a frequency of 800 Hz. The duration of the tests varied somewhat but was always of the order of 5 min. A general summary of the characteristic fretting scar features is most conveniently related to the regimes of the relevant fretting map. The strict subdivision into three regimes (corrosion effects are not emphasized in this work) has been slightly modified because, in the present materials, the transition from regime I to regime II was found to be gradual rather than representing a sharp boundary. 3.2.1. Stainless steel In regime I (Fig. 9(a)) the stainless steel exhibited no visible damage and a thin slip annulus with mainly
a sticking circle with non-resolved (bright)
162
(b)
Cd)
Cc) Fig. 9. Fretting scar on stainless II; (d) regime III.
steel:
(a) regime
I; (b) transition
conditions;
(c) regime
163
oxide debris. Under transition conditions the sticking circle begins to show deformed asperity contacts. The slip annulus has an inner contour of oxide debris and pore-like microcracks surrounded by material plastically smeared out along the periphery of the contact area (Fig. 9(b)). In regime II the sticking circle has suffered considerable plastic deformation. Large cracks are nucleated along the boundary between the sticking circle and the slip annulus which is also covered by bright oxide debris (Fig. 9(c)). The slip annulus appears dark in the scanning electron microscope. The gross slip of regime III (Fig. 9(d)) is characterized by uniform slip over the whole contact circle and no cracks are seen. Fine oxide debris is scattered in the depressions of the wear pattern. 3.2.2. Cu-Si In regime I the slip annulus is as thin as that in the stainless steel and also has some oxide debris. However, a high density of microcracks has been nucleated along the annulus in the Cu-Si alloy as shown by Fig. 10(a). The sticking circle is virtually undamaged. (The grooves, visible within the sticking circle, are about 10 times longer than the displacement 20 = 6 pm. They do not represent fretting slip but may have been created when the specimens were dismounted.) Under transition conditions the slip annulus is wider, covered with oxide, and severely cracked (Fig. 10(b)). In regime II the slip annulus is heavily deformed and contains a more or less continuous crack loop which practically separates the sticking circle from the rest of the scar (Fig. 10(c)). Regime III is characterized by a pattern of parallel, closely spaced cracks normal to the sliding direction (Fig. 10(d)). No traces of sliding under gross slip were observed. (The crack pattern of Fig. 10(d) does not correspond to the type of scar that can be expected in regime III. A similar structure has been reported by Kennedy et al. for rotational fretting [ 61.) 3.2.3. Pure copper In the pure copper the scar was about 1.5 times the size of the scars in Cu-Si and stainless steel for the same testing parameters. The sticking circle is nearly undamaged in regime I, whereas there is a well-developed slip annulus showing extensive plastic deformation and possibly some adhesive wear (Fig. 11(a)). Under transition conditions the sticking circle displays a few sheared adhesive contacts (Fig. 11(b)). The slip annulus is greatly deformed and has the appearance of a ductile fracture surface. It is possible that frictional heating has caused adhesion or seizure to the mating surface during fretting and that this contact was fractured during the dismounting of the specimens. In regime II (Fig. 11(c)) both the slip annulus and the sticking circle have suffered heavy plastic deformation and are completely separated by a crack around the stick-slip boundary. The dimpled surface of the sticking circle indicates that large-scale seizure had occurred and that this was broken up when the specimens were separated during dismounting. Figure 11(d) shows the plastic deformation
164
(a)
(cl Fig. 10. Fretting (d) regime III.
Cd) scar
on Cu-Si:
(a) relgime I; (b)
transition
conditions;
(c)
regim ke II:
165
(a)
(b)
(d)
(cl Fig. 11. Fretting (d) regime III.
scar on copper:
(a) regime I; (b) transition conditions;
(c) regime II;
166
which is characteristic of gross slip in regime III. Material from the scar area is smeared out over the surrounding surface in the sliding direction. 4. Summary and conclusions For increasing applied vibrator power both tangential force and displacement amplitudes T and D increase up to the values T2 and D2 so defining the fretting map boundary between regimes II and III. For higher power supplies, D continues to increase, while T stays constant or falls somewhat. This corresponds to the transition from static to kinetic friction, which is characteristic of incipient gross slip in fretting, and is in agreement with the occurrence of a frictional hysteresis, as described earlier by Vingsbo and Soderberg [5]. This observation is valid for all three materials in the whole frequency interval investigated in the present study. T2 increases with frequency, indicating strain rate hardening, while Dz may be affected by both strain-rate hardening and thermal softening for increased frequency. An increase in normal load implies increased T2 as well as D2 values due to the present external control of FT and d via the power supply. It must be pointed out, however, that this refers to critical values corresponding to the regime boundary. If, instead, the normal force were increased for a constant T, the corresponding D would drop. In agreement with Mindlin’s model it was possible to identify by scar morphology studies concentric zones of sticking and slip in all three materials with gross slip for displacement amplitudes above a critical value. As for differences between the three materials, DZ always decreased with increasing hardness while variations in T2 were different for different frequencies. The effect of hardness on scar morphology was mainly determined by differences in resistance to plastic deformation and crack formation for the three materials. The copper always displayed the strongest plastic deformation as well as the most pronounced crack formation. In both the copper and Cu-Si specimens the sticking zone was separated from the slip annulus by a complete crack loop. The strongest influence of the hardness was revealed in regime III. The stainless steel was characterized by gross slip and sliding fretting wear. All traces of previous cracks, if any, were removed. In the Cu-Si, the whole contact disc inside a thin deformed-rim annulus was surprisingly covered with a pattern of closely spaced, transverse fatigue cracks which would have been more expected in regime II. The softest of the materials, the pure copper, displayed a gaulling type scar indicating that the frictional welding had taken place over the entire contact area before conditions of gross slip were reached. Acknowledgment This work was supported by the Tribology Science Foundation under Grant MSM-8516963.
Program
of the National
167
References 1 R. D. Mindlin, Compliance of elastic bodies in contact, J. Appl. Me&., 16 (1949) 259. 2 K. L. Johnson, Surface interaction between elastically loaded bodies under tangential forces, Proc. R. Sot. London, Ser. A, 230 (1955) 531. 3 G. M. Hamilton and L. E. Goodman, The stress field created by a circular sliding contact, J. Appl. Me&., 33 (1966) 371. 4 J. J. O’Connor, The role of elastic stress analysis in the interpretation of fretting fatigue failure. In R. B. Waterhouse (ed.), Fretting Fatigue, Elsevier, Barking, 1981, pp. 23 - 66. 5 0. Vingsho and S. Soderberg, On fretting maps, Wear, 126 (1988) 131. 6 P. J. Kennedy, S. J. Calabrese and M. B. Peterson, Microdamage in sliding contacts, Wear, 121 (1988) 223.
153
153 - 167
FRETTING MAPS AND FRETTING METAL ALLOYS*
BEHAVIOR
OF SOME F.C.C.
OLOF VINGSBO+, MIKAEL ODFALK and NING-E SHEN Department (U.S.A.)
of Mechanical Engineering,
University of Houston, Houston, TX 77004
Summary
The aim of the research was to achieve a systematic knowledge of fretting contact and damage over a wide range of testing conditions, with emphasis on the relations between different testing parameters. In particular, the effect of material parameters on the fretting characteristics was to be studied with the aid of fretting maps. The work has concentrated on metal specimens. Three alloys were chosen to represent different hardness levels for essentially the same crystallographic structure: pure copper, a solution-hardened Cu-Si alloy and an austenitic stainless steel. Fretting testing was performed in a controlled atmosphere of dry air at a pressure of 500 Torr. Critical amplitudes of the tangential force and displacement for transitions between different fretting regimes were found and used to produce fretting maps. In addition, microstructural studies were performed with the aid of light and scanning electron microscopies to clarify the fretting mechanisms occurring. It was found that, in the present materials, there is a gradual transition from regime I (low damage) into regime II (mixed stick-slip and fretting fatigue), rather than a well-defined boundary. The boundary between regimes II and III, however, was established and plotted in fretting maps. Fretting maps are compared and discussed together with the influence of the hardness for the three materials. Characteristic fretting scar structures are shown and are related to mechanisms of fretting damage for each regime in all three materials.
1. Contact conditions in fretting Fretting is defined as “low amplitude oscillatory Depending on experimental conditions, surfaces”.
sliding between fretting contact
tribo may
*Paper presented at the International Conference on Wear of Materials, Denver, CO, U.S.A., April 8 - 14,1989. ‘Also at Uppsala University, School of Engineering, Materials Science, Box 534, S-751 21 Uppsala, Sweden. 0043-1648/90/$3.50
@ Elsevier Sequoia/Printed
in The Netherlands
154 Contact Circle
ElaStlC
Solid Sphere \
Solid
Block Slip
Within
---f--z? a N Annular
FTArea
: 8’ < r C a
Fig. 1. Elastic contact conditions of fretting for spheric symmetry: N, applied normal load; FT, applied tangential force; r, distance from the center of the contact circle; a, radius of the contact circle; a’, radius of the sticking circle.
cause damage of one or more of three types: fretting fatigue, fretting wear and fretting corrosion. In its most basic form, a fretting tribosystem is characterized by contact between convex surface elements, as first suggested by Mindlin in 1949 [ 11, and later discussed by other researchers [ 2 - 41. Mindlin’s model is shown for circular geometry in Fig. 1. The contact area is divided into two zones: a central area with a high contact pressure, where the friction may be high enough to prevent sliding (the sticking zone), and a surrounding zone of lower pressure and friction in which sliding will occur (the slip annulus). These conditions of mixed sticking and slip are characteristic of fretting at low displacement amplitudes. The “stick” zone will shrink with increasing applied tangential force (and displacement) until eventually slip occurs over the whole contact area (this is the “point of incipient slip”). The boundary region between sticking and slip is characterized by sharp stress peaks (tensile, shear or compressive, depending on its location with respect to the sliding direction) which oscillate over the contact area during the tangential force oscillation and promote crack nucleation and growth. Fretting studies generally suffer from the fact that the number of active parameters and variables, representing testing as well as material and atmosphere conditions, is very large, which makes interrelations between the different parameters difficult to distinguish and understand. This is the background of the development of a new type of testing equipment at the University of Houston which allows the displacement amplitude to be varied independently within large intervals of vibrational frequency and applied loads in controlled atmospheres. This unique facility has improved the experimental possibilities dramatically and has led to the development of a new type of representation of the fretting process in so-called fretting maps. A fretting map (see the schematic examples in Fig. 2) is a diagram, showing in two variables the different fretting damage regimes
155
Fig. 2. Schematic examples of fretting maps. Regimes of low damage (I), fretting fatigue (II) and fretting wear (III). N, normal load; D, displacement amplitude; f, frequency; T, tangential force amplitude.
representing low damage (regime I), stick-slip and fretting fatigue (regime II), gross slip and fretting wear (regime III). The key elements of a fretting map are the boundaries between the regimes. The influence of testing and material parameters on the shape and location of the regime boundaries is an issue of central interest for future fretting research. Once the relevant fretting maps for a given tribosystem are established, it is, in principle, possible to control the fretting process by choosing the active parameters so as to promote or suppress a certain type of damage. A special study of fretting maps has been published separately [ 51.
2. Experimental
details
2.1. Testing equipment The testing equipment is designed for a crossed-cylinders specimen configuration. One vertical, longitudinally vibrating cylindrical specimen is clamped between two horizontal, stationary half-cylinders. The normal load is applied to the stationary specimens with the aid of dead weights acting on a lever system. This configuration offers several advantages compared with other possible specimen and loading arrangements such as sphereon-flat for example. The most important advantages are the ease in (i) specimen manufacturing and polishing, (ii) application of normal as well as tangential loads, and (iii) keeping alignment under normal load (doublesided clamping). The fretting motion is generated by a sine generator, feeding an electromagnetic exciter-vibrator with independently variable frequency and vertical displacement, directly connected to the central specimen. The signal from an accelerometer mounted in the vibrating specimen holder is used in a feed-back compressor loop to monitor the output signal from the sine generator in order to control the fretting amplitude independently of variations in friction. The specimen displacement is obtained by integration of the accelerometer signal, and is continuously recorded on a digital oscilloscope together with the tangential force measured by a transducer.
156
Fore. I ! Tranrducor
Vfbratk?Q ; Specimen ,
Pyrex Boll Jar
,Ultrrronlc
Horn
I
S*mt-Cyllndric41 f Stattorury I Sp*ctmenr L -----_-_I --_-T : Spoclmonr, Flow
M*t*r
Fig. 3. Schematic
The imposed vibrating absolute vibrator) The main
Collar wlth Food-through.
PIJrQe/
/
Nindl*
V8cuun
diagram of the testing equipment.
elastic motion of the stationary specimens, because of the forces by the vibrating specimen, is less than 10% of the motion of the specimen. Thus, the reIative displacement is close to the recorded translation, The fretting equipment (except for the electromagnetic is enclosed in a bell jar and has a system for atmospheric control. layout is shown in Fig. 3.
2.2. Materials A simple investigation of the has been made by comparing those of hardness but essentially the same was varied in steps by studying three
influence of the material parameters for metallic alloys of various degrees internal microstructure. The hardness different f.c.c. structures: pure copper,
TABLE 1 Data on the materials tested AllOY
Copper Cu-Si Stainless steel
Twe
UNS Cl 10 UNS C655 AISI 304
Composition
Hardness
(wt.%)
WV 1(-k)
99.9 Cu 97 Cu; 3 Si 18 Cr; 8 Ni; 0.08 C; 1 Si; 2 Mn
80 2 5 I.80 rt 15 235 k 15
157
a solution-hardened Cu-Si alloy and an austenitic stainless steel. The composition and other data of the materials are given in Table 1. The microhardness measured at room temperature and with a load of 10 g varies between 80 and 235 HV for roughly the same grain size. The grain size of the copper alloys is 65 I.trn and that of the stainless steel 15 pm. 2.3. Fretting testing Before fretting testing, the machined specimens were ground and polished to a shiny surface. Two normal force levels were chosen with applied dead loads of 3.4 and 11.4 N. In principle, the fretting frequency of the present test rig is variable independent of the applied forces and the displacement amplitude. In practice, however, the power exerted on the specimens, is limited by that available through the power supply of the electromagnetic vibrator. For a given normal force the maximum displacement ~plitude decreases towards zero with increasing frequency. Thus, there is a maximum frequency for a non-vanishing displacement, which is set by the power supply rather than by the sine generator or the inertia of the vibrator. In the present study a frequency limit of 1500 Hz was chosen. 2.4. Recording fretting maps Identifying the regime boundaries which represent critical values of the testing variables for the transition from one regime to another, is a key issue in fretting research. The experimental procedure is not a trivial matter and in fact, a substantial part of the present work had to be devoted to developing and evaluating several techniques. At present it seems that different methods may be optimal for different materials and atmospheres. The final method established is described as follows. Two signals corresponding to displacement d and tangential force FT were simultaneously recorded us, time t on the screen of an oscilloscope. Each experiment at preset, constant normal load N and frequency f was started with a low vibrator power and tangential force, generating sinusoidal FT(t) and d(t) curves (Fig. 4(a)) with amplitudes T and D. The power was gradually increased searching for the first distortions ~co~espond~g to the critical amplitudes Tl and Di) to appear. In the present materials, however, it turned out that generally there was a gradual transition from regime I to regime II (see Fig. 4(b)) rather than a well-defined boundary (cf. the
Fig. 4. Characteristic examples of recorded FT(~) and d(r) curves. Material, Cu-Si; f= 800 Hz; N = 11.4 N. (a) d < D,;(b) Dl< d < D2; (c)d > Dz.
158
elastic limit in a stress-strain curve), and no Tl and D, amplitude values were recorded. During the continued increase of the vibrator power, both D and T increased until the critical amplitude T2 was reached. For a still higher applied power, however, the tangential force amplitude dropped slightly while the displacement amplitude continued to increase (Fig. 4(c)). This is in fact exactly what is expected for the transition from regime II (mixed stick-slip with partial static coefficient of friction) to regime III (gross slip with kinetic coefficient of fretting). However, it implies some practical difficulties in recording T, and D, because once the drop in T has occurred it is no longer possible to go back and find the original T2 and corresponding D, values. This can be explained by the deterioration of the fretting surfaces after the point of incipient slip. Therefore, the following procedure was adopted. The vibrator power was gradually increased while monitoring FT and a reading was taken of the maximum amplitude value before the drop, i.e. T = T2. The fretting contact was then repositi6ned on a fresh specimen surface and the procedure repeated with one important change, this being that the increase in power was stopped just before the previously recorded T2 was reached, i.e. at T < T2. The corresponding D value was recorded and defined as the critical displacement amplitude D,. This method will underestimate D2 slightly but it is reproducible and can be used in comparisons. From the accumulated data, all types of fretting maps, i.e. D(f), T(N), by pairwise combinations of the parameters. f(N) etc., can be constructed For ease of comparison, the present report is mostly based on D(f) and T(f) maps. Each datum point of the fretting maps corresponds to 3 or 4 runs and the scatter is of the order of ?lO%. It should be pointed out that the present work does not aim at precision in numerical results, but emphasizes exploring the potential of the fretting map representation. 2.5. Atmosphere control In the fretting map experiments covered in this paper, the atmosphere was dry air at a pressure of 500 Torr with the relative humidity less than 10%. Before each test run the specimen chamber (bell jar) was evacuated by a rotary pump to less than 5 X lo-* Torr. Air was then leaked into the chamber through a filter. 2.6. Scar morphology studies To clarify the active fretting mechanisms, the fretting scars characteristic of the different regimes were studied systematically. Separate test series were run with the parameters chosen to correspond to the conditions within each regime. The scar sizes were measured and the scar characteristics were studied with the aid of light and scanning electron microscopy.
159
3. Results and discussion 3.1. Fretting maps 3.1.1. Critical tangential force amplitude Figures 5(a) and 5(b) compare the frequency dependence of Tz for all materials at each clamping load. It can be seen that T2 increases with frequency indicating a strain rate hardening effect which is strongest in the Cu-Si at the higher load (Fig. 5(b)). A general observation is that Tz increases with increasing load at all frequencies in all three materials. This reflects the obvious fact that it takes a higher tangential force to achieve gross slip under a higher normal (clamping) load. Comparing the different materials indicates that, for frequences above 10 Hz, this effect is stronger in softer materials.
;J--yqg---q 100
10'
102
IO3
f 0-W
(a)
100
IO’
102
f (Hz)
(b)
Fig. 5. Z’(n fretting maps for all three materials: N= 3.4 N; (b) N = 11.4 N.
103 X, Cu; +, Cu-Si; 0, stainless steel. (a)
3.1.2. Fretting friction coefficient The ratio of the oscillating tangential force to the constant normal load represents an undetermined friction coefficient characteristic of the transient mechanisms during the application of external sliding. Its amplitude value nf = T,/N corresponds to the fully developed static friction and represents a “fretting coefficient of friction” which is plotted as a function of frequency in Fig. 6. It is obvious from the definition that pf varies with Tz and the same type of increase with frequency as in Fig. 5 is seen in Fig. 6. The differences between the present materials are small, particularly at the lower load (Fig. 5(a)). This is to be expected because
100
(a)
10'
102
f (Hz)
103
100
@I
10'
102
103
f (Hz)
Fig. 6. Fretting friction coefficient pf as a function of frequency f for all three materials: X, Cu; +, Cu-Si; 0, stainless steel. (a) N = 3.4 N;(b) N = 11.4 N.
160
they are all metals with the same surface finish. (One exception is the low pf value of the steel, shown for the high frequencies in Fig. 6(b).) As in unidirectional sliding, variations with normal load are small. The fact that the static coefficient of friction is higher than the kinetic coefficient is of no great significance for the wear behavior in unidirectional sliding because it is a transient phenomenon that occurs only once, i.e. during the initiation of sliding. In fretting contacts, however, the friction peak pf = T,/N induces both tensile, shear and compressive stress maxima around the contact area every cycle. In fact this is one of the main reasons why fretting fatigue is a severely deteriorating process in materials even at apparently harmless, low amplitude oscillations. 3.1.3. Critical displacement amplitude D2(f) maps for both levels of N are reproduced
in Fig. 7. It might be expected that the displacement would drop with increasing normal load; however, Fig. 7 shows that D, like T, increases with N in all three materials. This is because both T and D are controlled by the applied power of the vibrator irrespective of N. 5
10
4 +
$3 -cu2 n
x tf 1
0 i 100 (a)
+ xi P
q
10'
2 f (tiJO
103
0 100 @I
10'
102
103
f (Hz)
Fig. 7. D(fl fretting maps for the three materials: N = 3.4 N; (b) N = 11.4 N.
X,
Cu; +, Cu-Si; 0, stainless steel. (a)
The frequency dependence in the steel is an example of two mechanisms competing, these being often observed to affect the strain hardening properties during plastic deformation. As shown in Fig. 7(a) the displacement amplitude decreases with increasing frequency in the low frequency interval for the steel. This indicates a strain-rate hardening mechanism. As the displacement is connected with local plastic deformation in the contact zone, particularly under mixed stick-slip conditions, D should be affected by frequency-dependent strain hardening in the same way as suggested for the increase in T by an increase in frequency as in Fig. 5(b). For higher frequencies, however, the temperature increase associated with increased frequency implies a softening that dominates over the corresponding strainrate hardening. This type of hardening-softening balance generates the U-shaped D(f) curves of the stainless steel at the lower load (Fig. 7(a)) and of the Cu-Si at the higher load (Fig. 7(b)). The more complex waveform of the remaining curves of Fig. 7 could, in principle, be ascribed
161
a repeated hardening or softening dominance in successive tervals. However, that seems to be rather an overexploitation unless a reason for the repeated shifts can be found.
frequency inof the idea,
3.1.4. Influence of material parameters The influence of material parameters can be studied by plotting critical D and T values (for constant frequency) us. hardness as shown for example in Fig. 8. The critical displacement decreases clearly with hardness for both normal load levels and all frequencies, as shown for the higher load in Fig. 8(a). (The influence of the frequency, however, is not well described by this type of graph.) Tz decreases slightly with hardness for the higher normal load (Fig. 8(b)), but for the lower load the situation varies differently for different frequencies.
0 (a)
100 Hardness
200 (HVlOp)
300
0
100 Hardness
200 (HV 10~)
300
@I
Fig. 8. Critical amplitudes plotted us. hardness for different frequencies. (The three materials are represented by three hardness values along the H axis (cf. Table l).) N = 11.4 N; X, 10 Hz; +, 50 Hz; 0, 300 Hz;A, 800 Hz. (a) Dz; (b) T2.
3.2. Fretting mechanisms The crossed-cylinder specimen configuration gives a near-circular contact area, which can be directly related to the fretting model described in Fig. 1 and creates four scars per test. The results of the structural studies are presented below with representative scanning electron micrographs from one of the two scars of the stationary specimens. All the micrographs correspond’ to the same testing conditions: the higher normal load and a frequency of 800 Hz. The duration of the tests varied somewhat but was always of the order of 5 min. A general summary of the characteristic fretting scar features is most conveniently related to the regimes of the relevant fretting map. The strict subdivision into three regimes (corrosion effects are not emphasized in this work) has been slightly modified because, in the present materials, the transition from regime I to regime II was found to be gradual rather than representing a sharp boundary. 3.2.1. Stainless steel In regime I (Fig. 9(a)) the stainless steel exhibited no visible damage and a thin slip annulus with mainly
a sticking circle with non-resolved (bright)
162
(b)
Cd)
Cc) Fig. 9. Fretting scar on stainless II; (d) regime III.
steel:
(a) regime
I; (b) transition
conditions;
(c) regime
163
oxide debris. Under transition conditions the sticking circle begins to show deformed asperity contacts. The slip annulus has an inner contour of oxide debris and pore-like microcracks surrounded by material plastically smeared out along the periphery of the contact area (Fig. 9(b)). In regime II the sticking circle has suffered considerable plastic deformation. Large cracks are nucleated along the boundary between the sticking circle and the slip annulus which is also covered by bright oxide debris (Fig. 9(c)). The slip annulus appears dark in the scanning electron microscope. The gross slip of regime III (Fig. 9(d)) is characterized by uniform slip over the whole contact circle and no cracks are seen. Fine oxide debris is scattered in the depressions of the wear pattern. 3.2.2. Cu-Si In regime I the slip annulus is as thin as that in the stainless steel and also has some oxide debris. However, a high density of microcracks has been nucleated along the annulus in the Cu-Si alloy as shown by Fig. 10(a). The sticking circle is virtually undamaged. (The grooves, visible within the sticking circle, are about 10 times longer than the displacement 20 = 6 pm. They do not represent fretting slip but may have been created when the specimens were dismounted.) Under transition conditions the slip annulus is wider, covered with oxide, and severely cracked (Fig. 10(b)). In regime II the slip annulus is heavily deformed and contains a more or less continuous crack loop which practically separates the sticking circle from the rest of the scar (Fig. 10(c)). Regime III is characterized by a pattern of parallel, closely spaced cracks normal to the sliding direction (Fig. 10(d)). No traces of sliding under gross slip were observed. (The crack pattern of Fig. 10(d) does not correspond to the type of scar that can be expected in regime III. A similar structure has been reported by Kennedy et al. for rotational fretting [ 61.) 3.2.3. Pure copper In the pure copper the scar was about 1.5 times the size of the scars in Cu-Si and stainless steel for the same testing parameters. The sticking circle is nearly undamaged in regime I, whereas there is a well-developed slip annulus showing extensive plastic deformation and possibly some adhesive wear (Fig. 11(a)). Under transition conditions the sticking circle displays a few sheared adhesive contacts (Fig. 11(b)). The slip annulus is greatly deformed and has the appearance of a ductile fracture surface. It is possible that frictional heating has caused adhesion or seizure to the mating surface during fretting and that this contact was fractured during the dismounting of the specimens. In regime II (Fig. 11(c)) both the slip annulus and the sticking circle have suffered heavy plastic deformation and are completely separated by a crack around the stick-slip boundary. The dimpled surface of the sticking circle indicates that large-scale seizure had occurred and that this was broken up when the specimens were separated during dismounting. Figure 11(d) shows the plastic deformation
164
(a)
(cl Fig. 10. Fretting (d) regime III.
Cd) scar
on Cu-Si:
(a) relgime I; (b)
transition
conditions;
(c)
regim ke II:
165
(a)
(b)
(d)
(cl Fig. 11. Fretting (d) regime III.
scar on copper:
(a) regime I; (b) transition conditions;
(c) regime II;
166
which is characteristic of gross slip in regime III. Material from the scar area is smeared out over the surrounding surface in the sliding direction. 4. Summary and conclusions For increasing applied vibrator power both tangential force and displacement amplitudes T and D increase up to the values T2 and D2 so defining the fretting map boundary between regimes II and III. For higher power supplies, D continues to increase, while T stays constant or falls somewhat. This corresponds to the transition from static to kinetic friction, which is characteristic of incipient gross slip in fretting, and is in agreement with the occurrence of a frictional hysteresis, as described earlier by Vingsbo and Soderberg [5]. This observation is valid for all three materials in the whole frequency interval investigated in the present study. T2 increases with frequency, indicating strain rate hardening, while Dz may be affected by both strain-rate hardening and thermal softening for increased frequency. An increase in normal load implies increased T2 as well as D2 values due to the present external control of FT and d via the power supply. It must be pointed out, however, that this refers to critical values corresponding to the regime boundary. If, instead, the normal force were increased for a constant T, the corresponding D would drop. In agreement with Mindlin’s model it was possible to identify by scar morphology studies concentric zones of sticking and slip in all three materials with gross slip for displacement amplitudes above a critical value. As for differences between the three materials, DZ always decreased with increasing hardness while variations in T2 were different for different frequencies. The effect of hardness on scar morphology was mainly determined by differences in resistance to plastic deformation and crack formation for the three materials. The copper always displayed the strongest plastic deformation as well as the most pronounced crack formation. In both the copper and Cu-Si specimens the sticking zone was separated from the slip annulus by a complete crack loop. The strongest influence of the hardness was revealed in regime III. The stainless steel was characterized by gross slip and sliding fretting wear. All traces of previous cracks, if any, were removed. In the Cu-Si, the whole contact disc inside a thin deformed-rim annulus was surprisingly covered with a pattern of closely spaced, transverse fatigue cracks which would have been more expected in regime II. The softest of the materials, the pure copper, displayed a gaulling type scar indicating that the frictional welding had taken place over the entire contact area before conditions of gross slip were reached. Acknowledgment This work was supported by the Tribology Science Foundation under Grant MSM-8516963.
Program
of the National
167
References 1 R. D. Mindlin, Compliance of elastic bodies in contact, J. Appl. Me&., 16 (1949) 259. 2 K. L. Johnson, Surface interaction between elastically loaded bodies under tangential forces, Proc. R. Sot. London, Ser. A, 230 (1955) 531. 3 G. M. Hamilton and L. E. Goodman, The stress field created by a circular sliding contact, J. Appl. Me&., 33 (1966) 371. 4 J. J. O’Connor, The role of elastic stress analysis in the interpretation of fretting fatigue failure. In R. B. Waterhouse (ed.), Fretting Fatigue, Elsevier, Barking, 1981, pp. 23 - 66. 5 0. Vingsho and S. Soderberg, On fretting maps, Wear, 126 (1988) 131. 6 P. J. Kennedy, S. J. Calabrese and M. B. Peterson, Microdamage in sliding contacts, Wear, 121 (1988) 223.