Tribological behavior of aluminum alloys in a vibratory finishing process

Tribological behavior of aluminum alloys in a vibratory finishing process

Wear 255 (2003) 1369–1379 Tribological behavior of aluminum alloys in a vibratory finishing process Mohammad Reza Baghbanan a , Akihiro Yabuki b , Ro...

502KB Sizes 2 Downloads 51 Views

Wear 255 (2003) 1369–1379

Tribological behavior of aluminum alloys in a vibratory finishing process Mohammad Reza Baghbanan a , Akihiro Yabuki b , Roland S. Timsit c , Jan K. Spelt a,∗ b

a Department of Mechanical Engineering & Industrial Engineering, University of Toronto, 5 King’s College Road, Toronto, Ont., Canada M5S 3G8 Chemistry and Chemical Engineering, Graduate School of Engineering, Hiroshima University, 1-4-1 Kagamiyama, Higashi-Hiroshima 739-8527, Japan c Timron Scientific Consulting, 467 Woburn Avenue, Toronto, Ont., Canada M5M 1L6

Abstract The contact forces and tribological behavior of aluminum alloys (AA6061-T6 and AA1100-O) in a vibratory finishing process have been investigated. A new surface force sensor was used to measure simultaneously the normal and tangential forces produced by the ceramic media in a tub vibratory finishing machine. These forces were then correlated with the resulting changes in surface roughness, hardness and wear rate of aluminum disks attached to a cylindrical carrier, as well as with the residual curvature of the finished metal sheets. The principal variables were the degree of lubrication, finishing duration, and aluminum alloy. Further observations of media motion obtained using a miniature video camera and SEM micrographs of impact craters helped determine the contact mechanics of the finishing media and the wear mechanisms of AA1100-O and AA6061-T6 alloys. Comparisons are made throughout with similar data obtained previously with a smaller, less energetic bowl finisher. Experimental results revealed that, in the dry finishing condition, media contact tends to the normal direction, with increased media sliding occurring in the water-sprayed condition. Hardness and roughness changes of the workpiece were strongly dependant on the lubrication condition. Furthermore, in the dry condition, mass gain was observed as media debris became embedded in the workpiece surface, while in the water-sprayed condition this effect was prevented by the washing action of the water. The residual curvature of the disks increased with finishing duration until it reached a steady value. Residual curvature was also measured on Almen strips fastened to the ends of a second carrier, and differences between wet and dry finishing are discussed. © 2003 Elsevier Science B.V. All rights reserved. Keywords: Vibratory finishing; Vibrating bed; Mass finishing; Contact force; Friction coefficient; Impact; Hardness; Wear

1. Introduction Vibratory finishing has been used widely to improve surface appearance and wear resistance, to polish and increase hardness, and to clean and dry surfaces. The process has been applied to metal, ceramic, and plastic parts, employing a wide variety of media shapes and materials. The finishing media can be used dry, however, it is common to wet the media with water-based lubricant. Process parameters include the frequency and amplitude of the vibration, the amount and type of lubricant, and the size, shape and properties of the media. A vibratory finishing system usually consists of a springmounted open chamber containing granular media to which a vibratory motion generator is attached. The motion generator normally consists of one or two rotating shafts with eccentric weights. Adjusting the eccentric weight or the speed of the drive system controls the amplitude and frequency of the finisher. As a result, the media become fluidized and de∗ Corresponding author. Tel.: +1-416-978-5435; fax: +1-416-978-7753. E-mail address: [email protected] (J.K. Spelt).

velop complex flow fields within the chamber [1]. The parts to be finished are entrained by the flowing media and experience a slower relative velocity. The media interact with the part surface through a combination of normal impacts and sliding. Depending on the parameters of the process, this can produce a wide range of contact conditions encompassing varying degrees of rubbing, burnishing, ploughing, cutting and three-body abrasive wear. Sofronas and Taraman [2] examined various finishing conditions to establish an empirical model of the vibratory finishing process as applied to the removal of surface material at edges. Hashimoto [3] conducted experiments to determine the optimum finishing time for parts having different initial surface roughness. Wang et al. [4] measured the normal contact forces of media in a bowl-type vibratory finisher and compared them with the resulting changes in surface roughness and hardness of two aluminum alloys, AA1100-O and AA6061-T6. The principal variables were the media size, degree of lubrication and duration of the vibratory finishing. The changes in hardness and roughness were found to depend mostly on the lubrication condition and the size of the media, since these influenced the interaction between the

0043-1648/03/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0043-1648(03)00124-8

1370

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

media and the workpiece, and hence the extent of plastic surface deformation per impact. However, the impact force parameters such as the average force, maximum force, and impulse, did not vary appreciably among the three sizes of media for dry and water-wet conditions. Thus, the differences observed in hardness and roughness were apparently due to smaller-scale differences in the impact contact conditions. Videotaped observations showed that the media were loosely packed as they flowed past the workpiece, with relatively large gaps in the packing near the workpiece surface.

Furthermore, impact craters reflected the dominance of normal impact; however, water-wet conditions showed evidence of sliding. More recently, Yabuki et al. [5] measured both the normal and tangential contact forces in the same bowl-type vibratory finishing machine. Together with a video system, it was established that collisions between the finishing media and the test surface occurred in three different modes. The ratio of the normal and tangential forces was compared with the measured friction coefficient under dry and water-wet

Fig. 1. (a) Tub vibratory finisher showing motion path of cylinder. (b) Vertical and horizontal displacements of the tub finisher.

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

1371

conditions, confirming that media sliding occurred under water-wet conditions. The present study [6] used the surface force sensor of Yabuki et al. [5] to measure simultaneously the normal and tangential forces produced by the media in a larger, more energetic tub vibratory finishing (VF) machine. These forces were then correlated with the resulting changes in surface roughness, hardness and mass loss, as well as with the residual curvature of the aluminum workpieces. The principal variables were the degree of lubrication, finishing duration, and aluminum alloy. Further observations of media motion obtained using a micro-video camera and SEM micrographs of impact craters shed further light on the wear processes occurring in the vibratory finishing.

2. Experiments The tub finisher had a U-shaped cross-section with a urethane chamber liner. Inner dimensions were 40 cm (width) × 50 cm (length) × 40 cm (depth) (model 2016, Burr Bench, Brandon Industries, TX, USA) (Fig. 1a). The tub was attached to an electric motor that rotated a shaft with an eccentric weight at 47 Hz. Two single-axis accelerometers (model 3010D, Dytran Instruments Inc., USA) attached to the sides of the tub gave the vertical (z-axis) and horizontal (x-axis) vibration amplitude and frequency. The tub displacement followed a narrow elliptical path that did not change with time (Fig. 1b). An approximately spherical ceramic media (Abrasive Finishing Inc., Chelsea, MI, USA) with an average diameter of approximately 8 mm were used. The roughness of the media decreased significantly during approximately the first 8 h of use. This wear was also accompanied by an increase in kurtosis reflecting a narrowing of the profile peak height distribution and a decrease in the skewness as the surface became more rounded. The 30 h data in Table 1 represent the steady-state condition of the media in the experiments. EDAX-SEM elemental analysis showed that the media was composed mainly of Al, Si, O, Na, and Mg. For the lubricated finishing condition, the media were washed first with water and transferred to the tub. A conTable 1 Average roughness, kurtosis and skewness of the fresh and used ceramic media Media condition

Ra (␮m)

Rku (␮m)

Rsk (␮m)

Diameter (mm)

As received Used for 0.5 h Used for 8 h Used for 30 h

6.6 ± 1.5 4.0 ± 0.9 – 3.0 ± 1.1

2.5 ± 0.7 3.0 ± 0.8 – 4.1 ± 1.7

−0.1 ± 0.5 −0.2 ± 0.5 – −0.8 ± 0.7

8.4 ± 0.5 – 8.1 ± 0.5 7.9 ± 0.7

At 30 h, media had been used 22 h in dry and 8 h in wet conditions. Results of three stylus measurements with scan length of 0.8 mm on each media for 30 media. Diameter based on three measurements on each of 40 media (±1S.D.).

Fig. 2. Cylindrical carrier showing only one threaded clamping ring (detached at top). All dimensions in mm.

tinuous 1.5 l/min water spray was used to keep media wet during the experiments. A cylindrical carrier (Fig. 2) with a density of 0.75 g/cm3 was used to carry two aluminum specimen disks, one clamped at each end using a threaded ring that screwed to the wall of the cylinder. Direct contact and rubbing between the specimens and the clamping ring was prevented using a thin acetate gasket. This arrangement facilitated the measurement of mass change as a function of finishing time. The cylindrical carrier moved in an elliptical path in the tub with its axis remaining in the direction of the media motion (Fig. 1a). The carrier submerged into the media near wall ‘A’ and rose to the free surface near wall ‘B’ of the tub, directly above the rotating shaft. In the dry condition, the period of this cyclic motion was 19 ± 4 s while in the water-wet condition it was 25 ± 2 s (±1S.D., N = 30). This slowing of the carrier entrainment speed resulted from the decrease in the friction coefficient and the resulting reduction in the energy transfer between the tub walls and the media. The specimens were 0.8 mm thick aluminum sheets (AA1100-O and AA6061-T6), 67 mm in diameter with yield strengths of 34.5 and 276 MPa, respectively. They were washed with water and ethanol before and after each experiment using an ultrasonic cleaning bath (model 1510R-MT, Branson Ultrasonic Corporation, CT, USA) for 10–15 min. The specimens were rinsed in acetone, dried and stored in a desiccator prior to mass measurement using an analytical balance (model AP-250D, Ohaus, USA; ±0.01 mg). Surface roughness of the aluminum specimens and media was measured using a stylus profilometer (model Surtronic 3+, Rank Taylor Hobson Ltd., UK). A scan length of 4 mm

1372

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

was used for the aluminum. A reference specimen with the surface roughness of Ra = 5.85 ± 0.01 ␮m was used to calibrate the profilometer before each experiment. The hardness of the aluminum specimens was measured using a microhardness tester (Micromet 2100 series, Buehler Ltd., USA) with a square-based Vickers pyramidal diamond indenter and a load of 1–10 gf (0.01–0.10 N) at three points 120◦ apart and about 1 cm from the center. The diameter of the indentation was measured usually by using the scale of the hardness tester, but for lower loads (e.g. 1 gf), scanning electron microscope (SEM) images were used to determine the size of the indentation. Hardness depth profiles were also measured on polished sections normal to the surface and on 5.7◦ polished taper sections through the surface (giving a 10:1 expansion of the surface region). A small CCD color camera (IK-SM40A, Toshiba, 410,000 pixel resolution, 42 mm × 7 mm diameter) and a fiber optic light (Subtechnique Inc., USA) inside the cylindrical carrier

were used to study the global and relative motion of the media and part in the tub finisher. A multi-axial contact force sensor consisting of one normal and three surface shear force sensors was mounted in one of the cylindrical carriers as described in [5]. The 3 mm diameter sensing disk was located in the center of one of the end faces of the cylinder and was flush with this surface. The sensor for the normal force was calibrated by applying a series of known weights, while the tangential force sensors were calibrated by loading them laterally with a spring scale. The residual curvature of the aluminum specimens after finishing, due to plastic deformation, was measured using a dial gauge on a flat table at 5 different points (i.e. at the center and at four points 90◦ apart, 20 mm from the perimeter). Residual curvature was also measured on Almen strips (76 mm×20 mm×0.8 mm) fastened to the ends of a slightly larger cylindrical carrier (80 mm long × 82 mm diameter),

Fig. 3. Normal contact forces measured in the tub finisher.

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

with screws following the procedures of [7]. In addition to the two aluminum alloys, copper was also used (C10100 annealed at 600 ◦ C for 18 min to HV = 45).

3. Results and discussion The pattern of the normal and shear contact forces measured in the tub was similar to that in the bowl-type finisher of Wang et al. [4] and Yabuki et al. [5]. Fig. 3 shows an example of the normal contact force bursts that resulted from individual media striking and rolling over the 3 mm diameter sensor disk. The high frequency component of each burst shown in the bottom graph is due to the roughness of the media as it rolls [5]. The bursts are approximately 10 ms in duration, corresponding to half of the period of vibration of the tub (i.e. 21 ms at 47 Hz) during which the media accelerate toward the sensor. Fig. 4 shows the cumulative frequency of occurrence of normal impacts as a function of the normal force magnitude in the tub finisher, i.e. the number of impact events per second having a maximum normal force greater than or equal to the corresponding Fn . For example, in the tub finisher we expect the 3 mm diameter sensor to register at least one impact event per second with a maximum force ≥1.9 N in the dry condition and ≥1.3 N in the wet condition. The comparable figures for the bowl finisher were: ≥0.9 N in the dry condition and ≥0.7 N in the wet condition [5]. Thus, Fig. 4 shows that there was a wider range of impact force magnitudes in the tub, with an appreciable number of impacts being much larger than seen in the bowl. In both the bowl and tub, dry impacts tended to have a higher maximum force, and there were more impacts per unit time in the dry condition. This is attributable to two factors: (i) water-wetting decreases the coefficient of friction in the finisher and hence the efficiency of energy transfer from

Fig. 4. Cumulative frequency of occurrence of maximum normal force (Fn ) in the tub under wet and dry conditions, i.e. the number of impact events per second having a maximum force greater than or equal to the corresponding Fn .

1373

the wall to the media is reduced; and (ii) a water film on the media promotes adhesion and acts to dissipate interparticle impact energy. A further manifestation of this, as discussed below, was the slower workpiece entrainment speed in the finisher when wet. It was to be expected that the impact forces in the tub would be larger and more frequent than those in the bowl finisher [4,5] since both the frequency and amplitude were greater in the tub. Taking the resultant tub amplitude as 2.6 mm and that of the bowl as 1 mm [4,5], the maximum wall velocities were 0.77 and 0.18 m/s, respectively. The kinetic energy of the media in the tub finisher was thus approximately 18 times that in the bowl. The relationships between the maximum normal and corresponding resultant surface shear force under dry and water-wet conditions in the tub are shown in Fig. 5. The dashed lines are maximum and minimum dynamic friction coefficients, obtained in five repeat measurements in each condition. Thus, data points between these dashed lines probably represent a sliding condition during media impact. Under dry conditions (Fig. 5a), it is seen that most media do not slide on the aluminum sensor surface since most data were below the lines representing the range of the friction coefficient. Furthermore, evidence of sliding was not

Fig. 5. Resultant surface shear force and associated maximum normal force in an impact event (burst or single impact peak) in the tub. Data for all impact events recorded in six experiments of 20 s each: (a) dry media; (b) water-wet media. Lines show measured ranges of dynamic coefficients of friction.

1374

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

observed in scanning electron micrographs of dry-finished surfaces. Under water-wet conditions, however (Fig. 5b), it was confirmed that sliding of media did occur since more data points were in the range of the measured friction coefficient. Similar results were observed in the bowl finisher [5]. Fig. 6 shows typical SEM images for AA1100-O after 150 min wet finishing. The higher magnification images show two individual impact craters representing situations with and without sliding. As the force measurements indicated (Fig. 5), the wet condition decreased the coefficient of friction, causing a larger amount of sliding; however, normal impact forces remained dominant. The small video camera was used to measure the relative velocity of the media flowing past the end face of the moving cylindrical carrier. In the dry and wet conditions it was 0.8 ± 0.4 and 0.6 ± 0.4 cm/s, respectively (±1S.D., N = 10). This decrease in relative velocity is consistent with the reduction in the friction coefficients between the media themselves and between the media and the aluminum in the presence of water. This also resulted in a decrease in the absolute velocity of the media and the cylindrical carrier; for

example, at the free surface of the tub, the media absolute velocity was 5.2 ± 1.7 and 3.8 ± 1.0 cm/s in the dry and wet conditions, respectively. In comparison, the absolute velocity of the media at the free surface of the bowl was 3.1 ± 1.7 and 3.9±1.0 cm/s in the dry and wet conditions, respectively. Figs. 7–10 show the hardness, mass loss, curvature and average roughness as a function of finishing time for the two aluminum alloys in the dry and water-wet conditions. The Vickers hardness (HV; 1 Vickers hardness number equals 9.81 MPa) generally increased as finishing duration increased, and the average hardness after approximately 60 min was comparable in the dry and wet conditions, i.e. HV ≈ 60 for AA1100-O dry and wet, HV ≈ 120 for AA6061-T6 dry and wet. Thus, the greater and more numerous impacts observed in the dry state did not result in a markedly harder surface. As explained below, this may have been an artifact of changes in the impact conditions caused by the presence of water behind the aluminum disks. In the dry condition some local maxima were observed for both alloys, although this was more evident with AA1100-O (Fig. 7a) than with AA6061-T6 (Fig. 9a). A possible

Fig. 6. SEM micrographs of AA100-O after 150 min wet finishing: (a) normal impact crater; (b) impact crater showing evidence of sliding. Boxes indicate regions shown at higher magnification.

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

Fig. 7. Changes in AA1100-O as a function of finishing time in the dry condition: (a) Vickers hardness with 10 gf load; (b) normalized mass; (c) curvature; (d) average roughness. Error bars represent ±1S.D. for three measurements on each sample.

Fig. 8. Changes in AA1100-O as a function of finishing time in the wet condition: (a) Vickers hardness with 10 gf load; (b) normalized mass; (c) curvature; (d) average roughness. Error bars represent ±1S.D. for three measurements on each sample.

1375

Fig. 9. Changes in AA6061-T6 as a function of finishing time in the dry condition: (a) Vickers hardness with 10 gf load; (b) normalized mass; (c) curvature; (d) average roughness. Error bars represent ±1S.D. for three measurements on each sample.

Fig. 10. Changes in AA6061-T6 as a function of finishing time in the wet condition: (a) Vickers hardness with 10 gf load; (b) normalized mass; (c) curvature; (d) average roughness. Error bars represent ±1S.D. for three measurements on each sample.

1376

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

explanation is the repeated breakdown, removal and recreation of a hardened surface layer that was substantially comprised of ceramic wear debris from the media. This wear debris was also responsible for the dry state hardness measurements having larger uncertainty (error bars) than did the wet state data. Fig. 11 shows the debris layer that covered roughly 50% of the AA1100-O surface and 35% of AA6061-T6 surface

after dry finishing. The origin of most of the debris was confirmed by X-ray photoelectron spectroscopy to be the ceramic media. In some cases, debris became deeply embedded in the aluminum (Fig. 11c). These relationships between hardness and finishing time were broadly similar to those reported with the same alloys and media in the smaller, less energetic (smaller amplitude and frequency) bowl finisher [4]. In [4], however, the disks

Fig. 11. (a) Cross-section of ceramic debris in the surface of AA1100-O finished for 90 min in the dry condition. (b) Oblique SEM micrograph of AA6061-T6 finished for 90 min in the dry condition, 80◦ tilting angle. The white regions are due to electrostatic charging of the ceramic debris during image acquisition. An impact crater is shown in the higher magnification image. (c) SEM micrograph of cross-sections of embedded ceramic particle in the surface of AA1100-O finished for 150 min in the wet condition.

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

were adhesively bonded to the cylindrical carrier rather than being clamped, and perhaps as a result dry finishing produced a greater change in hardness than did wet finishing. In the present wet experiments, water could enter the small gap behind the clamped disk thereby increasing the effective stiffness of the disk as compared with the dry case where a thin air gap developed as a result of plastic deformation. Consequently, collisions with media of a given velocity may have produced greater impact forces and strain hardening. This may explain why the present wet and dry hardness values were closer than those in [4]. The depth of the work-hardened surface layer was investigated by measuring microhardness as a function of distance from the surface on polished cross-sections. Fig. 12

Fig. 12. Microhardness as a function of depth beneath the surface for: (a) polished taper-section of AA1100-O finished for 30 min in the dry condition (indenter load of 5 gf); (b) polished cross-section of AA1100-O finished for 150 min in the wet condition (indenter load of 1 gf); (c) polished cross-section AA6061-T6 finished for 150 min in the wet condition (indenter load of 1 gf). In all cases, error bars represent ±1S.D. for three measurements at each distance from the surface. Surface points taken normal to finished unpolished surface.

1377

shows that HV decreased from the surface into the bulk aluminum over a distance of approximately 20 ␮m. The data points corresponding to a depth of 0 ␮m were obtained from the microhardness measurements on the unpolished surfaces. The influence of the ceramic wear debris in dry finishing is clearly evident in the increasing normalized mass change recorded with the two alloys (Figs. 7b and 9b). The oscillating behavior may also be due to the periodic removal of the brittle surface layer of compacted wear debris. For both alloys, the initial surface layer broke off after approximately 120 min in the finisher causing a marked decrease in the mass. It was confirmed that the mass gain was not peculiar to aluminum since a copper specimen which was also finished for 35 min in the dry condition showed a relative mass gain almost identical to that of AA6061-T6 (copper specimen: HV = 90, initial; HV = 95, after finishing 35 min). Under wet conditions, very little change in mass was seen with either alloy reflecting the relatively smooth surface of the ceramic media. The plastic deformation that produced the strain hardened surfaces also generated a residual curvature in each of the aluminum alloy disks. Figs. 7c–10c show that the residual curvature increased during the first 90 min, becoming relatively steady thereafter. Interestingly, for AA1100-O the average steady-state curvature (after 90 min) was greater after wet finishing than after dry finishing, e.g. AA1100-O: dry, K ≈ 1.07 m−1 ; wet, K ≈ 1.90 m−1 . This trend was not evident with the harder AA6061-T6: dry, K ≈ 0.68 m−1 ; wet, K ≈ 0.69 m−1 . As with hardness, this trend for AA1100-O seems contradictory to the normal force data of Fig. 4 that showed the dry condition as corresponding to larger and more frequent impacts. It is likely that the explanation is the same; a thin water film behind the wet finished disks increased the stiffness and the impact forces. Figs. 7d–10d show the average roughness as a function of finishing time for the two alloys under dry and wet conditions. The changes in Ra were very small in the wet condition, consistent with the correspondingly small mass changes. In the dry condition, Ra increased significantly due to the embedded ceramic wear debris from the media. The large decrease in Ra seen after 90 min with dry-finished AA6061-T6 (Fig. 9d) may be related to the relatively large decrease in mass that occurred after approximately the same finishing time, i.e. a smoother surface remained after removal of surface debris. Fig. 13 illustrates the role of the boundary condition on Almen strip curvature for AA1100-O under three conditions: dry, wet, and with a grease film applied to the back of the strip prior to the experiment. The presence of a grease film increased the curvature in dry finishing to almost the same level as that in wet finishing. This confirmed that filling the air gap behind the sheet with water or grease increased the impact stiffness and forces, generating a greater curvature.

1378

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

Fig. 13. Almen strip curvature vs. finishing time for AA1100-O in the tub finisher under four conditions: (TD) tub dry; (TDG) tub dry with a grease film behind the strip; (TW) tub wet; (TWG) tub wet with grease film. Error bars represent ±1S.D. for three measurements of midpoint strip deflection.

Figs. 14 and 15 show the Almen strip curvatures measured in the tub and bowl finishers, respectively, and illustrate the effect of the greater impact energy in the tub. As seen in Figs. 9 and 10, the relatively hard AA6061-T6 did not show a significant difference between the wet and dry cases. In contrast, AA1100-O and Cu10100 show both a greater curvature for a given time and are more sensitive to the change in impact stiffness in the wet condition, i.e. wet curvatures are greater than dry curvatures. Analogous observations have been reported for shot peening [8], in which the method of

Fig. 14. Almen strip curvature vs. finishing time for AA1100-O, AA6061-T6 and Cu10100 in the tub finisher under two conditions: (TD) tub dry; (TW) tub wet. Error bars represent ±1S.D. for three measurements of midpoint strip deflection.

Fig. 15. Almen strip curvature vs. finishing time for AA1100-O, AA6061-T6 and Cu10100 in the bowl finisher under two conditions: (BD) bowl dry; (BW) bowl wet. Error bars represent ±1S.D. for three measurements of midpoint strip deflection.

strip fixation affected the curvature and strip deflection occurred prior to the removal of the clamping screws.

4. Conclusions The nature of the normal and shear forces, and the changes in surface properties in a tub vibratory finisher were similar to those in a smaller bowl finisher that had a much smaller amplitude and frequency. This suggests that vibratory finishing data and observations can be generalized and are not narrowly specific to a particular machine, frequency or amplitude. Normal impact forces were much larger than the corresponding shear forces, and relatively few media impacts result in sliding. This suggests a simplification that may be useful in future models of vibratory finishing. For the aluminum alloys AA1100-O and AA6061-T6, dry vibratory finishing produces a surface that is comprised largely of ceramic wear debris from the media, this being particularly true for the former, softer alloy. Residual curvature created by the plastic deformation of the surface provides a convenient means of characterizing the energy imparted to the finished specimens by the vibratory finishing system. This is similar to the way in which the Almen test is used to characterize the effects of shot peening systems [7]. As expected, the increase in curvature with finishing time correlated with the increase in hardness. The curvature tended to reach a steady-state value after approximately 90 min of finishing, and was greater in the more energetic finisher. Using the standard Almen test procedure of clamping the strips to the carrier in the wet condition led to a change in the impact stiffness of the sheet, apparently as a result of water seepage behind the strip. If mass change measurements are not required, it may be

M.R. Baghbanan et al. / Wear 255 (2003) 1369–1379

preferable to bond the Almen strips to the carrier using a hot-melt adhesive [4].

Acknowledgements The authors gratefully acknowledge the support of the Natural Sciences and Engineering Research Council of Canada. References [1] A.P. Fraas, Design of machines for driving complex-mode vibrationfluidized beds, in: Proceedings of International Mechanical Engineering Congress & Exhibition, American Society of Mechanical Engineers, Atlanta, GA, 17–22 November 1996.

1379

[2] A. Sofronas, S. Taraman, Model development and optimization of vibratory finishing process, Int. J. Prod. Res. 17 (1979) 23– 31. [3] F. Hashimoto, Modeling and Optimization of Vibratory Finishing Process, CIRP Ann. Manuf. Technol. 45 (1) (1996) 303–306. [4] S. Wang, R.S. Timsit, J.K. Spelt, Experimental investigation of vibratory finishing of aluminum, Wear 243 (2000) 147–156. [5] A. Yabuki, M.R. Baghbanan, J.K. Spelt, Contact forces and mechanisms in a vibratory finisher, Wear 252 (2002) 635–643. [6] M.R. Baghbanan, Contact forces and surface characterization of aluminum alloys in a vibratory surface finishing process, M.A.Sc. thesis, University of Toronto, 2002. [7] Procedures for using standard shot peening test strip—SAE J443 Jan84 (SAE recommended practice), Society of Automotive Engineers, Detroit, MI, 1984. [8] W. Cao, R. Fathallah, L. Castex, Correlation of Almen arc height with residual stresses in shot peening process, Mater. Sci. Technol. 11 (1995) 967–973.