Ž .Wear 237 2000 163–175www.elsevier.comrlocaterwear

Scuffing behavior of 390 aluminum against steel under starvedlubrication conditions

Hyung Yoon ), Todor Sheiretov 1, Cris CusanoAir Conditioning and Refrigeration Center, Department of Mechanical and Industrial Engineering, UniÕersity of Illinois at Urbana-Champaign,

Urbana, IL, USA

Received 19 January 1999; received in revised form 23 September 1999; accepted 23 September 1999

Abstract

The scuffing characteristics of 390-T6 aluminum–silicon alloy sliding against 52100 steel are examined under starved lubricationconditions. The evaluation is based on a shoe-on-disc geometry which simulates the platershoe contacts in an automotive air conditioning

Žswashplate compressor. Four shoe geometries flat shoe, crowned shoe, crowned shoe with a dimple at the center and crowned shoe with. Ž .a groove are tested against 390-T6 aluminum discs. All tests are conducted in a high-pressure tribometer HPT under R134a

Ž . Ž .tetrafluoroethane environment with a base polyalkylene glycol PAG lubricant. The effects of degree of lubricant starvation, slidingvelocity, geometry of contact, surface topography, and a tin coating on scuffing are evaluated. It is hypothesized that scuffing of 390-T6Al alloy under starved lubrication conditions is due to macroscopic adhesions leading to plastic shearing of the bulk material. q 2000Elsevier Science S.A. All rights reserved.

Keywords: Scuffing; Aluminum; Starved lubrication; Swashplate compressor

1. Introduction

Swashplate compressors are widely used for automotiveair conditioning systems. The swashplate drive mechanismfeatures an inclined plate which is rigidly attached to therotating shaft. The unidirectional rotation of the shaft istransformed to simple reciprocal motion of the pistonthrough the swashplatershoe contact. A schematic of atypical automotive air conditioning swashplate compressoris given in Fig. 1. The shoesrplate contacts in automotiveswashplate compressors are lubricated by a lubricantrre-frigerant mist. Therefore, it is not unusual for these con-tacts to experience starved lubrication conditions. With the

Ž . Žphaseout of R12 dichlorodifluoromethane , R134a tetra-.fluoroethane is the current refrigerant of choice for swash-

plate compressors. Mineral oils, which were used withR12, are not miscible with R134a. A family of misciblelubricants used with R134a, and presently used in swash-

) Corresponding author. Caterpillar, Technical Center, Building E,P.O. Box 1875, Peoria, IL 61656-1875, USA. Tel.: q1-309-578-2700;fax: q1-309-578-2953; e-mail: yoon hyung k@cat.com– –

1 Presently at Schlumberger Well Services, Houston, TX, USA.

Ž .plate compressors, are polyalkylene glycols PAG . Thisnew lubricantrrefrigerant mixture poses new problems,both for its thermodynamic properties as well as its tribo-logical properties. R134a lacks the lubricative properties ofR12. This lack of lubricative properties is thought to beone of the major reasons for the observed increased scuff-ing failures in some automotive swashplate compressorsusing the PAGrR134a mixture. There is an increasinginterest in the air conditioning and refrigeration industry ingeneral, and the automotive industry in particular, for abetter understanding of these failures.

A common materials pair presently used for the criticalswashplatershoe contact is a 390-T6 aluminum plate slid-ing against a 52100 steel shoe. In the past, most scuffingstudies have focused on steel components such as gears,

w xcams and followers, and piston ringrcylinder pairs 1,2 .Tribological studies of aluminum alloys in the literature

w xhave focused mainly on friction and wear 3–8 . Recently,w xBarber et al. 9 have conducted tests with 390 Al–Si

alloys to determine the effects of silicon particle size andetching on scuff and wear resistance. It was shown that theamount of near-surface silicon and the oxide films formedduring the rubbing process are important conditions for

0043-1648r00r$ - see front matter q 2000 Elsevier Science S.A. All rights reserved.Ž .PII: S0043-1648 99 00321-X

( )H. Yoon et al.rWear 237 2000 163–175164

Fig. 1. Schematic of an automotive air conditioning swashplate compressor.

improving the scuffing resistance of these alloys. Somiw xReddy et al. 10 have investigated the wear and scuffing

behavior of binary Al–Si alloys, containing up to 23% Si,sliding against a hard steel counterface under dry slidingconditions. The transitions from mild to severe wear andsevere wear to scuffing are related to the destabilisation ofthe protective layers caused by subsurface flow. In their

w xlater paper 11 , it is hypothesized that scuffing occurs ifthe shear stress at a critical depth under the surface ex-ceeds the temperature-dependent shear strength of the ma-

w xterial at this depth. In our previous studies 12,13 , thescuffing behavior of aluminumrsteel contacts was exam-ined under dry sliding conditions. Based on the experimen-tal observations, it was hypothesized that the accumulationof plastic deformation and fatigue damage in the subsur-face is responsible for scuffing.

The present research focuses on the scuffing behaviorof a 390-T6 aluminum disc sliding against a 52100 steelshoe. The contact is lubricated by a PAGrR134a mistwhich results in starved lubrication conditions.

2. Experimental setup

( )2.1. High-pressure tribometer HPT

The amount of refrigerant dissolved in the lubricant isaffected by the environmental pressure and temperature. Ifthe lubricant and refrigerant are miscible, increasing thepressure and decreasing the temperature will increase theamount of refrigerant which will saturate into the lubricant.Therefore, in this study, an HPT was used to provideaccurate control of the environmental pressure and temper-ature, and thus approximately simulate the environmentalconditions found in refrigerant compressors. A schematic

of the pressure chamber is shown in Fig. 2. In the HPT, theenvironmental pressure and temperature can be controlledup to 1.72 MPa and 1208C, respectively. The desiredtemperature of the contact is obtained by an externalrecirculating unit which is capable of controlling the tem-perature from y108C to 1308C. The lower specimen issecured in place by a specimen holder and the upperspecimen is attached to the rotating spindle. The remov-able cup, which contains a spraying nozzle and shoespecimen, is bolted to a force transducer module as shownin Fig. 2. The transducer is outfitted with an intricate arrayof strain gages which are used to measure the appliednormal load and the frictional torque during a test. Acontact load of up to 4.45 kN can be applied to thespecimens.

The HPT is equipped with apparatus for electric contactresistance measurement and a computer control of theaxial load and the angular velocity of the spindle. Theelectric contact resistance between the test specimens pro-vides indirect information about the regime of lubrication,the formation of protective surface films, and the extent ofmetal-to-metal contact. A schematic of the contact resis-tance measuring setup is given in Fig. 3. The measurementrange of the circuit used is 10y6 –10q4

V. This sensitivitywas achieved by the development of a special four-termi-nal measurement circuit, methods for noise suppression

w xand data processing software 14 . Computer control of theaxial load and angular velocity of the spindle is achievedthrough control boards and a set of solid state relays. Thecorresponding software, which was also developed, allowsfor complicated loading and sliding velocity histories.

2.2. Geometry of contact

Ž .In this study, a shoe-on-disc geometry Fig. 4 is usedto simulate the platershoe contacts in a swashplate com-

( )H. Yoon et al.rWear 237 2000 163–175 165

Fig. 2. Schematic of the HPT pressure chamber and a lubricantrrefrigerant delivery system.

Fig. 3. Schematic of the contact resistance measuring setup.

( )H. Yoon et al.rWear 237 2000 163–175166

Fig. 4. Geometry of contact.

Ž .pressor Fig. 1 . During a test, the shoe is supported in asocket. The shoersocket geometry is such as to producecontact in a circumferential band, giving the shoe freedom

Ž .to align with the disc Fig. 4 . Four different shoe geome-Žtries flat shoe, crowned shoe, crowned shoe with a dimple

.at the center and crowned shoe with a groove were tested.The crowned shoes are the actual shoes presently used inswashplate compressors. The crown height of these shoesvaries between 2 and 10 mm. A picture showing the faces

of the shoes tested is given in Fig. 5. Typical surfaceprofiles of the various shoe geometries are given in Fig. 6.Shoes, which have a crown height between 4 and 7 mm,were tested in this study. The purpose of crowning is tohelp the platershoe contacts generate hydrodynamic filmsduring operation, since a perfectly smooth and flat shoe,which is supported at its geometric center, theoreticallycannot generate these films. The custom-made flat shoeswere tested to examine the effects of the crown height on

Fig. 5. The top views of the various shoe geometries tested.

( )H. Yoon et al.rWear 237 2000 163–175 167

Ž . Ž . Ž . Ž .Fig. 6. Typical surface profiles of various shoe geometries: a flat shoe; b crowned shoe; c crowned shoe with a dimple; d crowned shoe with agroove.

scuffing under starved lubrication conditions. The size andapparent areas of contacts for these shoes are given inTable 1.

2.3. Materials tested

All shoes tested were made of 52100 steel with anaverage hardness of 62 HRC and average surface rough-ness of 0.03 mm R . All tests were conducted with 390-T6a

aluminum discs with an average hardness of 72 HRB.Unless otherwise noted, the average surface roughness ofthe discs was 0.04 mm R . The 390-T6 Alr52100 steela

contact pair is the actual material combination used in theplatershoe contact of many swashplate compressors. Thechemical composition of the 390-T6 aluminum alloy testedis given in Table 2.

Table 1Geometric dimensions of various shoes tested

Shoe Diameter Area of dimple ApparentŽ .geometries mm or groove area of

2Ž .mm contact2Ž .mm

Crowned shoe 10.0 – 78.5Crowned shoe 9.6 5.5 66.9with a dimpleCrowned shoe 10.0 18.1 60.4with a grooveFlat shoe 6.35 – 31.7

2.4. Test conditions and lubricant

Starved lubrication conditions are obtained by applyingvarious amounts of lubricant on the surface of the upper

Ž .specimen disc through the lubricant delivery systemshown in Fig. 2. A pre-determined weight proportion of amiscible lubricant is added to the liquid refrigerant in thepressure vessel. The desired amount of lubricantrre-frigerant mixture is fed to the contact regions through anozzle by heating the pressure vessel. Many preliminarytests were conducted to determine a lubricantrrefrigerantpressure which gave an approximate constant feeding rateduring the test. For all tests conducted, a constant pressureof 1.2 MPa was maintained by controlling the temperatureof the pressure vessel. The separation between the top of

Ž .the nozzle and the upper specimen disc is such as tomake the diameter of the base of the spray cone approxi-mately 0.5 cm. The impact of the spray on the disc spreadsthe lubricant to wet the whole contact area. The average

Ž .lubricant supply rate LSR is determined by weighing themixture before and after a test, determining the amount oflubricant used and dividing by the test duration. The

Table 2Chemical composition of 390-T6 Al tested

Ž .Material Alloying elements % by weight

Al Si Fe Cu Mn Mg Zn Ti

390 Al Balance 16–18.5 1.0 3.0–4.0 0.5 0.4–1.0 1.0 0.25

( )H. Yoon et al.rWear 237 2000 163–175168

Table 3Lubricant data

2Ž . Ž .Designation Lubricant type Family Additives Density grml Viscosity mm rs

at 408C at 1008C

PAG Polyalkylene glycol Uncapped None 0.99 49.5 9.8

carrier gas used for the lubricant is the refrigerant understudy. All tests were conducted under R134a environmentwith an uncapped PAG. Data for the PAG lubricant usedare given in Table 3.

The environmental pressure and temperature affect notonly the amount of refrigerant dissolved in the lubricantbut also the surface properties by chemical reactions ofactive species with the metal surfaces. In order to approxi-mately simulate the environmental conditions existing inswashplate compressors, the test chamber is initially purgedto a vacuum of 0.2 Torr. After purging, the test is initiatedby spraying the PAGrR134a mixture on the disc. Therefrigerant pressure in the chamber was kept constantduring the test by means of a pressure-release valve. Alltests were conducted at an environmental pressure of 0.17MPa and temperature of 1218C. These conditions are closeto the nominal conditions found at the platershoe contactof a swashplate compressor. For all tests conducted, a222.5 N load step and a step duration of 15 s were used.For a given constant sliding velocity, the load was in-creased stepwise until scuffing occurred. A step durationof 15 s was used since, under starved lubrication condi-tions, a steady state temperature is reached after approxi-mately 10 s. It should be noted that scuffing is a loadinghistory-dependent phenomenon. Therefore, different re-sults might be obtained if a different loading history wasused. However, it is believed that the comparative studypresented in this study is still valid.

3. Results and discussion

The major difficulty in conducting experimental studiesof scuffing is the reliable reproduction and identification ofthe scuffing condition. In this study, scuffing was identi-fied by the sharp transitions of the contact resistance andcoefficient of friction measurements. Typical test data aregiven in Fig. 7.

3.1. Effect of degree of starÕation and sliding Õelocity

The contact pressure at which scuffing occurs is afunction of the degree of lubricant starvation and slidingvelocity. Fig. 8 shows scuffing data for crowned shoeswith a dimple. The contact pressures given are calculatedbased on an area which excludes the area of the dimple. Asexpected, for every sliding velocity used, the scuffing

pressure increases as the LSR increases. For a given LSR,the scuffing pressure decreases as the sliding velocityincreases. This phenomenon can be attributed to thermaleffects. More heat is generated at the interface with highersliding velocities, causing a breakdown of the protectivesurface films and subsequent scuffing failure. The results

Ž .given in Fig. 8 are plotted in Fig. 9 as pressure P vs.Ž . Žvelocity V , for various LSRs. Three LSRs 20, 30, 40.mgrmin , from which three pressure data points can be

obtained in the curve fits given in Fig. 8, are chosen forthis plot. This curve is commonly referred to as PV curvefor evaluating the scuffing resistance for a particular lubri-cant-bearing material system. It should be noted that thedata points shown in Fig. 9 fall closely on lines given byPVsconstant. The value of this constant increases as thedegree of lubricant starvation decreases.

3.2. Effect of the contact geometry

As previously indicated, a perfectly smooth and flatshoe, which is supported at its geometric center, theoreti-cally cannot generate hydrodynamic films and thus, canonly operate in the boundary lubrication regime. The steelshoes presently used in swashplate compressors have asmall crown. Some shoes also have a dimple or groove.However, the effects of shoe geometry on scuffing are notwell-known. Scuffing data for 390-T6 Al have been ob-tained with flat shoes, crowned shoes, crowned shoes witha dimple and crowned shoes with a groove. Scuffingresults obtained with flat shoes are compared with thoseobtained with crowned shoes in Fig. 10. The data show

Ž .that, for severely starved conditions LSR-30 mgrmin ,the scuffing pressure obtained with flat shoes is higherthan that obtained with crowned shoes. However, as theLSR increases, the scuffing pressure obtained with crownedshoes is much higher than that obtained with flat shoes.This indicates that scuffing is a function of crown heightand the degree of starvation. When the contact is severelystarved, the crowning is detrimental because, for a givenaverage pressure, the peak contact pressure and tempera-ture for the crowned shoe are higher than those for a flatshoe.

Scuffing results for 390-T6 Al obtained with variouscrowned shoes are shown in Fig. 11. The major role of adimple or groove in a shoe is to trap debris and lubricantduring operation. However, when the contact is starved, adimple or groove seems to be detrimental. It is possible

( )H. Yoon et al.rWear 237 2000 163–175 169

Fig. 7. Typical test data.

that, for the LSRs used, the amount of lubricant supplied atthe contact is not enough for a dimple or groove to beeffective as a lubricant reservoir. Also, for a given averagepressure distribution for the three shoes, the discontinuousgeometries of the crowned shoes with a dimple or grooveŽ .Fig. 6 resulted in higher peak pressures and temperatures.

Ž .Due to the limiting load capacity 4.45 kN of the HPT,scuffing data at higher LSRs could not be obtained. Itshould also be noted that, in swashplate compressors, theshoesrplate contacts experience a squeezing action. Whateffects, if any, the dimple or groove has on scuffing withthe addition of this squeezing is not known.

Fig. 8. Scuffing results for 390-T6 Al for various lubricant supply rates and velocities. Shoe geometry: crowned shoes with a dimple.

( )H. Yoon et al.rWear 237 2000 163–175170

Fig. 9. Scuffing PV for various lubricant supply rates. Solid lines indicate PVsconstant. Shoe geometry: crowned shoes with a dimple.

3.3. Effect of surface topography

Surface topography is known to play an important rolein friction, wear and scuffing behavior of lubricated con-

w xtacts. Ludema 15 , in a review of scuffing, emphasizes theimportance of the influence of surface asperities on scuff-ing. Surface roughness affects the magnitudes of the realarea of contact, and the local contact pressures and temper-atures. It also plays a major role in the establishment ofhydrodynamic or elastohydrodynamic films. To examinethe effects of surface roughness on scuffing, tests wereconducted using 390-T6 aluminum discs with three differ-ent surface roughness values. The average surface rough-

nesses tested were 0.03, 0.49 and 1.50 mm R . The 0.03a

mm R surfaces were obtained by superfinishing, whilea

the 0.49 and 1.5 mm R surfaces were produced bya

conventional grinding and abrasion with 60 grit Al O2 3

abrasive paper, respectively. These discs were testedagainst crowned shoes with a dimple. The scuffing pres-sures as a function of sliding velocity for these specimensare given in Fig. 12. As expected, the results show that thescuffing resistance increases as surface finish improves Asillustrated in Fig. 13a, the 0.03 mm R surface gives aa

much lower coefficient of friction preceding scuffing fail-ure than the other surfaces. It is also shown in Fig. 13bthat the friction force obtained with 0.03 mm R surface isa

Fig. 10. Scuffing results for 390-T6 Al obtained with flat and crowned shoes. Sliding velocity: 1.86 mrs.

( )H. Yoon et al.rWear 237 2000 163–175 171

Fig. 11. Scuffing results for 390-T6 Al obtained with various crowned shoes. Sliding velocity: 1.86 mrs.

significantly lower than those of the other surfaces. Thus,much less heat is generated at the interface with the 0.03mm R surface, resulting in improved scuffing perfor-a

mance.The effects of surface roughness on scuffing were also

examined as a function of LSRs with 0.03 and 1.50 mmR specimens. As shown in Fig. 14, for the sliding velocitya

used, surface roughness effects are more significant as theLSR increases, i.e., scuffing of the smoother specimen isaffected to a greater degree than that of the rougherspecimen as the LSR increases. For the severely starved

Ž .contact LSRs25 mgrmin , the scuffing pressures areabout the same for both roughnesses. For a given nominalcontact pressure, the local asperity contact pressures are

Fig. 12. Effect of surface roughness on scuffing pressure as a function of sliding velocity. Lubricant supply rate: 40 mgrmin; shoe geometry: crownedshoes with a dimple.

( )H. Yoon et al.rWear 237 2000 163–175172

Fig. 13. Variation of coefficient of friction and friction force for differentsurface roughnesses. Lubricant supply rate: 40 mgrmin; sliding velocity:

Ž . Ž .1.86 mrs. a Coefficient of friction; b friction force.

higher for the rougher specimens since their real area ofcontact is smaller. It is believed that, when the contact isseverely starved, more protective surface films are formedwith the rougher specimens due to more asperity interac-tions, resulting in approximately the same scuffing resis-tance for both surfaces.

Skewness is another important parameter characterizingthe behavior of sliding surface. This parameter gives somemeasure of the degree of asymmetry of the asperity heightdistribution. If the surfaces have deeper valleys than peaks,the skewness will be some negative value. It is known that,in some tribological applications, surfaces with a negativeskewness are beneficial under mixed or boundary lubrica-tion conditions. In order to evaluate the effect of skewness

Žon scuffing, discs having two different skewness y0.34. Žand y1.26 but the same surface roughness value 0.49

.mm R were tested. Scuffing data were obtained ata

various sliding velocities, as shown in Fig. 15. The resultsshow that these different skewness do not seem to improve

the scuffing resistance. It is believed that the effect ofskewness would be more apparent if a significant skew-ness difference was present on the harder counterfacematerial.

3.4. Effect of a tin coating

Scuffing failure can be prevented or minimized byusing some form of surface coatings or surface treatmentson the sliding contacts. Tin coatings have been widelyused in many automotive sliding components such aspiston rings and swashplates in air conditioning compres-sors. Together with surface roughness, coatings can affectlocal pressures and temperatures which, in turn, affectscuffing. In order to examine the effect of a tin coating onscuffing under starved lubrication conditions, scuffing data

Žwere obtained with tin-coated 390 aluminum discs 1.12. Žmm R and are compared to the uncoated discs 0.49 mma

.R in Fig. 16. In the present study, a 2–3 mm thick tina

coating obtained by a chemical conversion process wastested. Even though the initial surface roughness of thetin-coated discs is much higher than that of the uncoateddiscs, for the 40 mgrmin LSR, the scuffing pressure of thetin-coated discs is about three times higher than that of theuncoated discs. Scuffing data for the coated discs, at a

Žmuch more severe starved lubrication condition 15.mgrmin , are also given in Fig. 16. These data show that,

even at this severely starved lubrication condition, thecoated discs provide better scuffing resistance than theuncoated discs. This behavior is attributed to the low shearstrength characteristics of the tin material. Tin tends todeform quickly and increases the conformity of the slidingsurfaces, resulting in lower local pressures and tempera-tures. It is also noted that the tin coating aids initialrunning-in by covering surface irregularities and acts as a

w xtemporary lubricant 16 .For scuffed tin-coated aluminum discs, some of the

aluminum substrates are exposed due to the wear of thecoating. As indirectly illustrated in Fig. 17, the coated discexperiences a severe wear regime before scuffing occurs,as indicated by the increased friction coefficient. Increasednoise and vibrations are also typical in this regime. Itshould also be noted that the contact resistance for thecoated discs is relatively low from the beginning of the test

Ž .compared to that of the uncoated discs Fig. 7 . Thisindicates that the coated discs have more surface interac-tions due to their high initial surface roughness and lowhardness. However, their average friction coefficient islower than that of the uncoated specimens, which is againdue to their low shear strength characteristics. From visualobservations, scuffing of the tin-coated specimens seems tobe caused by the removal of the coating, which leads to theformation of macroscopic adhesion between the exposedaluminum substrate and the steel counterface. Ni and

w xCheng 17 also indicated that the aluminum–tin bearingsfail because of the rapid depletion of tin at the contact area

( )H. Yoon et al.rWear 237 2000 163–175 173

Fig. 14. Effect of surface roughness on scuffing as a function of lubricant supply rate: Vs2.79 mrs; shoe geometry: crowned shoes with a dimple.

Fig. 15. Effect of skewness on scuffing pressure as a function of sliding velocity. Lubricant supply rate: 40 mgrmin; average surface roughness: 0.49 mmR . Shoe geometry: crowned shoes with a dimple.a

Fig. 16. Effect of a tin coating on scuffing pressure as a function of sliding velocity. Shoe geometry: crowned shoes with a dimple.

( )H. Yoon et al.rWear 237 2000 163–175174

Fig. 17. Typical scuffing test data for a 390 tin-coated aluminum disc.Lubricant supply rate: 40 mgrmin; sliding velocity: 3.72 mrs.

due to high temperature. They considered the meltingtemperature of tin as a scuffing criterion for aluminum–tinbearings.

3.5. Examination of scuffed surfaces

In order to better understand the scuffing process understarved lubrication conditions, the scuffed surfaces of theuncoated discs were examined using a scanning electron

Ž .microscope SEM . A typical scuffed surface of 390-T6 Aldisc and aluminum material transfer on the counterfacesteel shoe are shown in Fig. 18. Fig. 18a shows smallgrooves caused by plastic shearing in the scuffed region.When scuffing occurs, the macroscopic adhesions formedat the sliding interface cause plastic shearing of the bulkaluminum material, resulting in the formation of trans-ferred aluminum patches on the counterface as shown in

Ž . Ž .Fig. 19. a Delamination on the scuffed surface and b it’s correspond-ing surface profile.

Fig. 18b. Also, due to gross plastic deformation during thescuffing process, local delamination on the scuffed surface

Ž . Ž .Fig. 18. a Typical scuffed 390 aluminum surface and b material transfer on the steel shoe.

( )H. Yoon et al.rWear 237 2000 163–175 175

Ž .is often observed Fig. 19a . This delamination causesdeep grooves as shown by the surface profile given in Fig.19b. It is hypothesized that scuffing of 390-T6 Al alloyunder starved lubrication conditions is due to the macro-scopic adhesions leading to plastic shearing of the bulkmaterial. Based on the observation of the scuffed surfaces,the sequence of the scuffing process seems to involve thefollowing events:Ž .a Local breakdown of the lubricant and adsorbedfilms;Ž .b Breakdown of the protective oxide films due toplastic deformation of asperities;Ž .c Local exposure of the bare metal and the formationof microscopic adhesions; andŽ .d Formation of macroscopic adhesions causing plasticshearing of the bulk material.

4. Summary of the results

Scuffing data for a 390-T6 aluminum disc sliding againsta 52100 steel shoe are obtained using the HPT. Thecontact is lubricated with a PAGrR134a mixture understarved lubrication conditions. The effects of degree oflubricant starvation, sliding velocity, geometry of contact,surface roughness and skewness, and a tin coating onscuffing are evaluated. The experimental results can besummarized as follows.

Ž .1 For a given environmental temperature and specifiedcontact geometry, the scuffing pressure increases as theLSR increases and decreases as the sliding velocity in-creases.

Ž .2 PVsconstant relationship seems to characterize thescuffing behavior of 390-T6 Al plater52100 steel shoecontacts. The value of this constant increases as the degreeof lubricant starvation decreases.

Ž .3 Scuffing is a function of crown height and thedegree of lubricant starvation. Crowning is detrimentalwhen the contact is severely starved.

Ž .4 When the contact is severely starved, a dimple orgroove is also detrimental.

Ž .5 For the whole velocity range tested, the scuffingpressure of 390-T6 Al discs increases with improved sur-face finish; however, a skewness change from y0.34 toy1.26 for discs with the same surface roughness does notseem to improve the scuffing resistance.

Ž .6 For a given LSR, the scuffing pressure of tin-coated390-T6 Al discs is about three times higher than that of theuncoated discs.

Ž .7 It is hypothesized that scuffing of 390-T6 Al alloyunder starved lubrication conditions is caused by macro-scopic adhesions leading to plastic shearing of the bulkmaterial.

Acknowledgements

The authors are grateful for the financial support by theAir Conditioning and Refrigeration Center, Mechanicaland Industrial Engineering at the University of Illinois atUrbana-Champaign. We would like to thank the FordMotor for providing the tin-coated and uncoated aluminumdiscs. Crowned shoes tested in this study were provided byFord and by Delphi Harrison Thermal Systems, a Divisionof General Motor.

References

w x Ž .1 A. Dyson, Scuffing — a review, Tribology International 8 4Ž .1975 77–87, Part 1.

w x Ž .2 A. Dyson, Scuffing — a review, Tribology International 8 6Ž .1975 117–125, Part 2.

w x Ž .3 A.D. Sarkar, Wear of aluminum–silicon alloys, Wear 31 1975331–343.

w x4 H. Torabian, J.P. Pathak, S.N. Tiwari, Wear characteristics of Al–SiŽ .alloys, Wear 171 1994 49–58.

w x5 A.H. Hanna, F. Shehata, Friction and wear of Al–Si alloys, Lubrica-Ž .tion Engineering 49 1992 473–476.

w x6 J. Ferrante, W. A. Brainard, Wear of aluminum and hypoeutecticaluminum–silicon alloys in boundary lubricated pin-on-disk sliding,

Ž .NASA Technical Paper 1442 1979 .w x7 K. Okabayashi, M. Kawamoto, Influences of silicon content and

grain size of primary silicon crystals on the wear of high siliconaluminum alloys, Bulletin of University of Osaka Prefecture 19Ž .1968 199–216.

w x8 M.K. Jasim, E.S. Dwarakadasa, Wear in Al–Si alloys under dryŽ .sliding conditions, Wear 119 1987 119–130.

w x9 G.C. Barber, J.J. Matthews, S. Jaffry, Wear and scuff resistance ofŽ .aluminum 390, Lubrication Engineering 47 1991 423–430.

w x10 A. Somi Reddy, B.N. Pramila Bai, K.S.S. Murthy, S.K. Biswas,Ž .Wear and seizure of binary Al–Si alloys, Wear 171 1994 115–127.

w x11 A. Somi Reddy, B.N. Pramila Bai, K.S.S. Murthy, S.K. Biswas,Mechanism of seizure of aluminum–silicon alloys dry sliding against

Ž .steel, Wear 181–183 1995 658–667.w x12 T.K. Sheiretov, H.K. Yoon, C. Cusano, Scuffing under dry sliding

Ž .conditions: I. Experimental studies, Tribology Transactions 41 4Ž .1998 435–446.

w x13 T.K. Sheiretov, H.K. Yoon, C. Cusano, Scuffing under dry slidingŽ .conditions: II. Theoretical studies, Tribology Transactions 41 4

Ž .1998 447–458.w x14 T.K. Sheiretov, Scuffing of aluminumrsteel contacts under dry

sliding conditions, PhD Thesis, Department of Mechanical andIndustrial Engineering, University of Illinois at Urbana-Champaign,1997.

w x15 K.C. Ludema, A review of scuffing and running-in of lubricatedŽ .surfaces, with asperities and oxides in perspective, Wear 100 1984

315–331.w x16 W.L. Brown, Development of piston-ring coatings, SAE Technical

Ž .Paper 690754 1969 .w x17 X. Ni, H.S. Cheng, Seizure failure of copper–lead with overlay and

aluminum–tin connecting rod bearings, Tribology Transactions 39Ž . Ž .1 1996 194–200.