UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mechanisms of wear particle formation and detachment Knowles, Gregory D. 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1994-0221.pdf [ 7.56MB ]
Metadata
JSON: 831-1.0080930.json
JSON-LD: 831-1.0080930-ld.json
RDF/XML (Pretty): 831-1.0080930-rdf.xml
RDF/JSON: 831-1.0080930-rdf.json
Turtle: 831-1.0080930-turtle.txt
N-Triples: 831-1.0080930-rdf-ntriples.txt
Original Record: 831-1.0080930-source.json
Full Text
831-1.0080930-fulltext.txt
Citation
831-1.0080930.ris

Full Text

MECHANISMS OF WEAR PARTICLE FORMATIONAND DETACHMENTByGregory D. KnowlesB.A.Sc. (Mechanical) University of TorontoA THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTERS OF APPLIED SCIENCEinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OFMECHANICAL ENGINEERINGWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril 1994© Gregory D. Knowles, 1994In presenting this thesis in partial fulfilment of the requirements for an advanced degree atthe University of British Columbia, I agree that the Library shall make it freely availablefor reference and study. I further agree that permission for extensive copying of thisthesis for scholarly purposes may be granted by the head of my department or by hisor her representatives. It is understood that copying or publication of this thesis forfinancial gain shall not be allowed without my written permission.Mechanical EngineeringThe University of British Columbia2324 Main MallVancouver, CanadaV6T 1Z4Date:pr1( 1B, 13S4.AbstractA fundamental investigation into the mechanism of wear particle formation for martensiticstainless steel has been conducted using the simple contact configuration of a hard steel ball and aflat stainless steel surface. The effects of changes in hardness and surface finish of the flat surfacewere examined for three lubrication conditions, over a wide range of test durations. Friction forceand work data were collected during testing, and wear scars were examined by surfaceprofilometry and optical and scanning electron microscopy. Results of this study demonstrate theoperation of several complex wear mechanisms and the inherently microscopic scale of wear. Afew simple models are suggested for the separate wear mechanisms; however greaterunderstanding of many of the wear mechanisms will have to be achieved before a complete andpredictive wear model can be developed.11Table of ContentsAbstract iiTable of Contents iiiList of Tables vList of Figures viAcknowledgements ix1 Introduction 12 Literature Survey 42.1 Sliding Wear Studies 42.2 Contact Mechanics 52.3 Contact of Rough Surfaces 112.4 Wear and the Near Surface Microstructure 142.5 Wear Mechanisms and Models 162.6 The Third Body Approach 213 Wear Testing Equipment and Procedures 233.1 NRC Fretting Wear Rig 233.2 Specimen Preparation 283.3 Test Matrix 323.4 Post Test Analysis 344 Results and Discussion 364.1 Surface Examinations 364.2 Cross-Sections 454.3 Mass Loss Results 491114.4 Friction Coefficient Data 554.5 Frictional Work Input 594.6 Wear Scar Geometries and Curvatures 605 Extended Analysis of Results 645.1 Wear Process in Single Asperity Contact Model Tests 645.2 Wear Particle Formation 745.3 Third Body Effects 825.4 Third Body Models 845.5 Surface Roughness and Friction Transitions 846. Conclusions 86Appendix A 88References 91Tables 95Figures 105ivList of TablesPage3.1 Dynamic test specimen specifications 953.2 Equivalent hardnesses of4lO stainless steel stationary test specimens 953.3 Typical surface roughness measurements of stationary specimen discs 953.4 Summary of open loop test conditions 963.5 Jokisch W2-OP concentrated cutting fluid components 963.6 Summary of long-duration feedback controlled test conditions 973.7 Summary of short-duration feedback controlled test conditions 984.1 Summary of open loop controlled test results 994.2 Summary of long-duration feedback controlled test results 1004.3 Summary of short-duration feedback controlled test results 1014.4 Summary of statistical test group results 1024.5 Depths of visible plastic deformation in sectioned discs 1025.1 Shakedown factors for short duration feedback controlled tests 1035.2 Estimated initial or first-cycle indentation depths 1035.3 Shakedown factors and predicted curvatures for long duration feedback 104controlled testsVList of FiguresPage2.1 Maximum yield parameter in and below the surface versus friction coefficient 1052.2 Normal stress in the direction of the tractive force (a,) along the x-axis forvarious friction levels 1052.3 Line contact shakedown map for a kinematic hardening material 1062.4 Circular contact shakedown map for a kinematic hardening material 1072.5 Distribution of contact pressure and subsurface orthogonal and principal shearstresses for a ground surface contacted by a smooth indenter 1083.1 NRC fretting wear rig 1093.2 Dynamic specimen holder 1093.3 Ideal specimen displacement and normal load signals 1103.4 Dynamic model ofNRC fretting wear rig 1103.5 Test rig control and data acquisition systems 1114.1 Water lubricated annealed disc wear scar after ten sliding cycles 1124.2 Water lubricated annealed disc wear scar after twenty sliding cycles 1134.3 Water lubricated hardened disc wear scar after ten and twenty sliding cycles 1144.4 Wear scar from unlubricated ten cycle test on an annealed disc - 1164.5 Front and rear of wear scar from unlubricated ten cycle test on an annealed disc 1184.6 Wear scar from ten cycle, 200 N loaded, unlubricated test on annealed disc 1194.7 Wear scar from twenty cycle unlubricated test on annealed disc 1214.8 Wear scar from ten cycle unlubricated test on hardened disc 1234.9 Wear scar from twenty cycle unlubricated test on hardened disc 1254.10 Fifty cycle hardened disc wear scar, from Test K 23 127vPage4.11 Unlubricated one hundred cycle hardened disc wear scar, from Test K 27 1284.12 Five hundred cycle hardened disc wear scar, from Test K 21 1294.13 Five thousand cycle hardened disc wear scar, from Test K 20 1304.14 Five thousand cycle tempered disc wear scar, from Test K 15 1314.15 Intermediate duration unlubricated wear scar on a polished disc 1324.16 Optical niicrographs of disc wear scars from short duration, cutting fluidlubricated tests 1334.17 Long duration cutting fluid lubricated wear scar on a polished disc 1344.18 Long duration water lubricated wear scar on a polished disc 1364.19 Wear scar on a sandblasted disc, long duration, cutting fluid lubricated 1384.20 Wear scar on a 60 grit SiC pressed disc, long duration, cutting fluid lubricated 1394.21 Cross-sections of a 60 grit SiC pressed disc surface ( unworn) 1404.22 Etched sections of discs from long duration cutting lubricated tests 1414.23 Etched sections of discs from long duration water lubricated tests 1424.24 Sections of sandblasted discs from long duration water lubricated tests 1434.25 Sections of 60 grit SiC pressed discs from long duration water lubricated tests- 1444.26 Sections of wear scars from long duration unlubricated tests 1464.27 Cutting fluid lubricated short duration test mass losses 1484.28 Open loop mass loss results - 40:1 cutting fluid lubrication 1494.29 Mass loss versus surface roughness for cutting fluid lubricated tests 1504.30 Mass loss for SiC pressed discs including stress relieved discs 1504.31 Mass loss versus surface roughness for water lubricated tests 151viiPage4.32 Mass loss versus disc hardness for unlubricated tests 1514.33 Linearized cummulative distribution function of statistical test seriesmass loss results, plotted with a Gaussian model 1524.34 Normal and frictional force and displacement signals from water lubricated shortduration tests 1534.35 Normal and frictional force and displacement signals from unlubricated shortduration tests 1544.36 Friction traces from cutting fluid lubricated short duration tests 1554.37 Friction coefficient plots of various roughnesses for cutting fluid lubrication 1564.38 Friction plots of 3 cutting fluid lubricated 60 grit SiC pressed discs 1574.39 Friction plot of first fifteen hundred cycles of two cutting fluid lubricatedtests on sandblasted discs 1574.40 Friction coefficient plots of various roughnesses for water lubrication 1584.41 Friction plot of first 400 cycles for 3 water lubricated tests on polished dies 1594.42 Friction coefficient plots of various disc hardnesses 1604.43 Mass loss and frictional work comparisons 1614.44 Surface maps ofwater lubricated disc scars ofvarying surface roughness 1624.45 Surface maps of unlubricated disc scars of varying disc hardness 1634.46 Longitudinal profilometer traces of polished disc scars with best fit 164radii of curvature5.1 Disc scar widths from cutting fluid lubricated tests 1655.2 Shakedown load for cicruclar contact of a 1/2” steel ball on a flat of annealedtempered, and hardened 410 stainless steel 166viiiAcknowledgementsThis study was made possible by primary support from the Natural Sciences and EngineeringResearch Council (NSERC) and secondary support from the National Research Council Canada(NRC). The author is also grateful for the guidance and encouragement provided by supervisorsDrs. P.L. Ko (NRC) and H. Vaughan (UBC). Tremendous technical support and patience wasalso contributed by Mr. Mark Robertson (NRC), along with significant contributions from Mr.R. Leung (NRC) and Mr. T. Qian (UBC). Encouragement from colleague Mrs. Jean Chen andDr. 3. Kalousek (NRC ) were also invaluable. Thank you Pak and Mark for your calm patience,and many thanks to all ofyou whom have helped me through this challenge.ixChapter 1Intro uctionThe phenomenon referred to as wear has been recognized for centuries, and oftenaccepted as inevitable. Although much research and effort have been devoted to its study,wear has proven in most situations to be an elusive process to predict, model, or evenunderstand. After decades of careful work, a clear and complete understanding of wear inmany situations has yet to be realized. However; significant progress continues to bemade, driven by a rising demand for products with greater wear resistance and longerservice lives.Wear can occur through many possible mechanisms, and often several mechanismsoccur simultaneously or sequentially in a single situation. Researchers have attempted todescribe individual wear mechanism using simple models that represent the elements andparameters thought to influence specific mechanisms. Modeling has progressed beyond thesimple experimentally-correlated wear coefficients first proposed in the early 1950’s.More recently, the development of fracture mechanics concepts has allowed wearresearchers to present models of crack propagation and wear particle formation. Advancesin computing power have also stimulated promising work on numerical and finite elementmodels of asperity contact stresses and strains; however the immense number of iterationsrequired to accurately follow the extended contact of two wearing surfaces has so farprevented the extension of computer-based approaches into completely predictive wearmodels. A fresh idea for wear modeling recently presented by French researchers unifiesthe three major elements of tribology, namely friction, lubrication and wear, into a thirdbody concept in which the relative motion of solid first bodies is accommodated in a thirdbody or interfacial zone. Although full fluid film lubrication is currently the only third body12which is well defined and understood, with further work this modeling approach maypossibly be extended to all types of third body contacts.For a wear model of any type to be functional, it must be based on an accurateunderstanding of the events involved in that particular wear process. This may seem anobvious requirement; however, the complexities of wear often make it impossible toisolate, observe and ultimately confirm the accuracy of the many assumptions made whenmodeling. One of the tallest challenges currently faced by tribologists is to determine“what actually happens” in exact terms when a pair of surfaces make contact and wear.The most basic step in many wear mechanisms is the formation of wear particles or weardebris, and many models have been devised to describe the process of wear particleformation. Thus understanding exactly how wear particles form has become an essentialrequirement for accurate wear modeling.To that end, a fundamental investigation of wear particle formation wasconducted, specifically for steel-on-steel contact under constant low-amplitude unidirectional sliding, varying the number of wear strokes or cycles, the hardness and thesurface roughness of one surface. The configuration of the test specimens consisted of asphere sliding against a flat disc, which represents a small three-dimensional asperitycontacting a larger asperity with a radius of curvature many orders of magnitude greaterthan the first. The wear process was carefully studied to determine the mechanisms ofwear particle formation, how they change as the number of wear cycles increases, and howchanges in hardness, surface finish, and the friction or lubrication conditions betweencontacting surfaces affect particle formation. Several simple models were presented todescribe the observed mechanisms of particle formation.In this study, the wear characteristics of AISI 410 stainless steel were examinedexclusively for several reasons. This martensitic grade of stainless is the least expensive ofall stainless grades, and is widely employed in many applications, ranging from steam andgas turbine blades, nuclear reactor control rod mechanisms and boiler tube supports,3valves, fasteners, shafting, machine parts to household kitchen items. While usage of themartensitic grades is exceeded by the austenitic grades, wear research of stainless steelshas been disproportionately focused on the austenitic grades. Few detailed studies havebeen performed on the martensitic stainless steels. The wide hardness range of the 410grade made it appealing choice for investigation of the effects of hardness.Results of this study indicate that despite efforts to simplify wear with a singlepoint contact, several complex wear mechanisms operate. During the first few wearstrokes, wear scar geometry is established by either a smooth continuous process ofdeformation or erratic delamination of a relatively thick surface layer. In the latter case,wear particle formation begins immediately. After the wear scar geometry is established,particle formation occurs by the fine delamination of the wear scar surface, and from thefracture of roughness features in the wear scar. Lubricated conditions promote theentrapment of wear debris, which leads to the formation of compacted layer of fine debriscovering the wear, reducing but not eliminating further wear. The mechanisms revealed inthis study have yet to be descriptively modeled. Only when each of these complexmechanisms is fully understood, can a combination of all into a predictive wear model beachieved.Chapter 2Literature SurveyOn the topic of wear there exists a veritable plethora of published information,which makes the task of surveying that literature at first appear rather challenging.However, the study of such a diverse subject has necessitated categorizing it into separateprocesses, so information on unrelated processes can be disregarded. Unfortunately,despite continuing efforts to standardize, many different wear classification systems exist,which certainly adds to the complexity of sifting out the relevant information.Reviews of the well-known theories of wear processes have been frequently madeby many knowledgeable authors. A brief review can be found by Magel [1]. For a moredetailed treatment, Blau devotes a complete chapter to the classification of wear modes inhis text, which also includes an excellent review of basic wear theory references [2]. Blauillustrates the diversity in wear classification by listing eight different wear classificationschemes, all proposed by notable tribologists. To that collection he then adds his ownscheme, consisting of eight major wear modes. Under his system, a mode can be made upof one or more wear processes, with a wear process resulting from a distinct combinationof the most basic mechanisms. The list of wear modes includes abrasive wear, cavitation,chemical and oxidative wear, erosion, impact wear, fretting, fatigue, and sliding wear. Thelast term in this list covers wear processes usually referred to by other authors as adhesivewear. Blau expresses his dislike of the term adhesive wear, since it implies that adhesion isthe dominant mechanism in a term more often used as a catch-all for types of wear notalready identified by the other terms.To repeat a review of basic wear modes and theories again would only addunnecessary length to this document, and instead the intentions of this chapter are to45provide a review of published work, similar in some aspect to this study, along with asummary of new or important findings from other relevant areas. Topics to be coveredinclude sliding wear studies of steels, classical Hertzian contact stress formulations and theimplications of surface roughness on contact stress, wear particle formation mechanismsand models, and the third body concepts of friction and wear.2.1 Sliding Wear StudiesThe majority of martensitic stainless steel wear studies reported in the literaturehave been conducted in unlubricated conditions. Schumacher tested over 30 materials inself-mated pairs with emphasis on both martensitic and austenitic stainless steels [3]. Histest configuration consisted of two 0.5 in. perpendicular cylinders loaded with a 70 Nforce (Hertzian maximum contact pressure 1.6 GPa). Slow rotation of one cylinderproduced a sliding velocity of 0.07 rn/s for 400 m sliding distance under unlubricatedconditions. Very hard alloy and tool steels showed the lowest wear, while for theaustenitic and the hardest martensitic (440C) stainless grades, the wear was an order ofmagnitude higher. Lower carbon martensitic steels fared much worse, generally wearingtwo to three orders of magnitude greater than the alloy and tool steels. Schumachersummarized the results by noting that the initial hardness of a metal can be a deceptivecriterion for its relative wear resistance, since work hardening capacity and favorableoxide film formation can have strong effects on metallic wear [3].Studies of unlubricated sliding wear of martensitic stainless steels were alsoreported by Smith [4]. His tests were performed with a 50 mm tip radius pin-on-flatarrangement in a very slow reciprocating sliding motion ( average speed 22 mm/s) in airand CO2 over a range of ambient temperatures from 20 to 300 °C. Two non-standardstainless steels, similar in chromium content but with lower carbon and higher nickelcontents than AISI 410, were tested. Loading for tests in air ranged from 10 to 90 N(Hertzian maximum contact pressure 0.21 - 0.44 GPa), and many interesting trends were6noted. Initial specific wear rates were high, similar to rates for martensitic stainless steelscalculated from the results of Schumacher [3], but were independent of load over the loadrange examined. In some instances, wear transitions were observed in the form of adecrease in wear rate between a factor of two and two orders of magnitude. No consistenttrend in the extent of the decrease in wear rate could be found with either temperature orload. The friction behavior of specimens worn in air began low ( t = 0.2), rising andfluctuating rapidly between t = 0.7 and 0.85. Then coinciding with the wear transition, thefriction level remained at a similar high level with much slower fluctuations. Wear debrishad the appearance of a gray metallic powder, and analysis by X-ray diffraction indicatedvarying amounts of c&Fe and c-Fe2O3.Differences in wear rates of the severe pretransition stage were attributed to the mode of surface deformation. The highest wearrates were obtained for surfaces deformed by galling or severe surface distortions. Lowerwear rates were associated with grooved surfaces and prow formations. No informationon or explanation of the conditions of the milder wearing post-transition surfaces wasprovided.Poggie and Wert reported results from dry sliding wear studies of tool steel flatsagainst hardened AISI 52100 balls [5]. One of the tool steels tested (AISI D2) had amicrostructure similar to martensitic stainless steels. D2 has a high chromium content,although the carbon content is closer to the hardest martensitic grades ( AISI 440C).Wear testing was conducted in ambient air at a constant sliding speed of 3 mmls, under aconstant 20 N load (1.7 GPa maximum Hertzian contact pressure ), and in areciprocating motion over a 60 mm stoke for ten thousand cycles. Wear rates were verylow, with specific wear coefficients similar in magnitude to those reported by Smith [4] forpost-transition mild wear. Poggie and Wert also reported coefficient of friction values near= 0.2 at the start of testing and rising to 0.8 after the first five to ten cycles. SEMexaminations of the wear scars showed no evidence of ploughing or abrasion. Wear debriswas reported as primarily made up of fine oxides from both the ball and tool steel,7although no analysis of the debris was performed. The authors suggest given the elasticcontact stress conditions, mild oxidative wear was the sole operative wear mechanism.This conclusion was supported by observations of only slight changes in residual stressmeasurements made on specimens before and after testing.A series ofwear tests under lubricated conditions was performed and reported byJahanmir [6]. The experiments were performed with a ball-on-cylinder configuration underlow load (9 N - Hertzian contact stress 300 MPa) and slow speed ( 5 x 10 mis). Ballswere made from hardened AISI 52100 alloy steel and the cylinders from either normalizedAISI 52100 or normalized plain carbon ATSI 1020 steel. Thus the microstructures of thecylinders were mainly ferritic, with spheroidized cementite particles in the 52100 and somepearlite in the 1020. Specimens were lubricated with different friction-reducing additivesin the base oil to vary the friction coefficient. Test durations were not reported. Wearparticle formation mechanisms were assessed by SEM examinations of the wear scars forthree friction levels. At low friction levels ( 0.05 to 0.1) wear scars displayed minimalsurface damage. Original grinding marks on the cylinder were plastically deformed andclosed up, producing a very smooth wear track. Sectioning showed no evidence ofsubsurface deformation. Wear was attributed to mild surface deformation. At intermediatefriction conditions ( i = 0.15 to 0.25 ) extensive surface damage was found. Wear scarsexhibited grooving by harder debris, and delamination of surface layers into plateletformations. Sectioning revealed subsurface plastic deformation extending about 4 p.mbelow the wear scars. Plowing and delamination were identified as wear mechanism forthe intermediate friction conditions. At high friction levels ( p. = 0.35 to 0.65 ), the wearscars showed more severe delamination of surface layers, and deeper subsurfacedeformation (to about 8 jim). Subsurface cracks were also found, along with voidsformed around carbide particles in the normalized 52100 steel. At the highest frictionlevels, a shear dimple pattern, indicative of microvoid coalescence, was found on some8wear scar platelets. Jahanmir suggests that under high friction conditions, the primarywear process is adhesive wear.2.2 Contact MechanicsIn reviewing wear research literature, some analytical work must also be included.Contact mechanics provides useful information on the near-surface stress state of slidingbodies. A brief review of relevant publications related to this area is now considered.Circular contact in the absence of friction was first analyzed over a hundred yearsago by Hertz [7]. Elastic stress fields beneath the contacting surfaces were determined byHuber [8]. Boundary conditions for sliding were established much later by Mindlin [9],who assumed the shear stress in the contact area was everywhere proportional to thenormal stress by the coefficient of friction between the sliding bodies. Stress fields beneathsliding bodies were determined by Hamilton and Goodman [10], and later simplified byHamilton [11]. The development of stress fields for line contact has followed a similarcourse, with stress equations determined by Poritsky [12]. The relationships developed bythese authors describes the elastic stress field, which can be combined with a failurecriteria to determine the location where plasticity will first occur. Using the von Misesyield parameter, Ji (the root of the second invariant of the stress deviator tensor), forthe zero friction case, the location of the maximum value of this yield parameter occurs ina circular contact below the surface at a depth of 0.5a, where a is the contact radius. Asthe friction coefficient is increased, the subsurface maximum moves closer to the surfaceand the value of the yield parameter at the surface increases. By t = 0.3, the surface valuehas exceeded the subsurface peak, and when ii = 0.5 there is no longer a subsurfacemaximum. This trend is summarized in Figure 2.1, which shows the magnitudes of surfaceand subsurface yield parameter in non-dimensional form via division by the maximumcontact pressure, p0. Figure 2.1 shows that the magnitude of the subsurface peak is aboutone third of the maximum contact pressure and increases little with friction. However the9surface value does increases dramatically with friction, to the extent that at typical frictionvalues for unlubricated sliding ( t 0.7 ) the yield parameter is 73% of the maximumcontact pressure.The location of maximum tensile stress also indicates a potential site for failure byfracture. In the absence of friction, one of the principal stresses, in the radial direction, istensile on the circumference of the contact circle. With increasing friction, this tensilestress grows at the trailing edge of the contact and become compressive at the front. Thiscondition is depicted in Figure 2.2, which shows the distribution of normal stress in thesliding or surface tangent direction for three friction levels. The large tensile peak at thetrailing edge continues to increase with the friction coefficient, and at i = 0.7 this tensilestress is greater than the maximum contact pressure by a factor of 1.32. Hamilton andGoodman [10] use stress maps to show that this system of tensile stresses is quite shallow,decaying from a maximum of 0.98 p0 at the surface (for jt = 0.5) to less than 0.2 p0 at adepth of 0.25a.General relationships for the case of the more general elliptic contact area havebeen found for both normal and tangential loading by Sackfield and Hills [13]. With theirequations, the authors demonstrated the weak effect of contact geometry, or morespecifically the ellipticity of the contact area, on the magnitude of the surface andsubsurface values of the von Mises yield parameter. Johnson has also demonstrated thisweak effect of contact geometry on contact size and pressures as well [14], noting errorsof only 5 to 8 % from circular contact equations for contact dimension, pressure, anddeflection when applied to an elliptic contact having a 3:1 ratio of major to minor axes.Results of this type have spawned a rule of thumb among contact mechanics to treatmildly elliptic contacts as circular, and contacts of ellipticity ratio of 5 or greater as linecontacts.Having considered the analysis of elastic loading, the next step is to investigatewhat happens when the limits of elastic contact are exceeded. For a frictionless circular10indentation, the development of the plastic zone from a small subsurface region into anunconfined mode of free surface flow has been modeled successfully both numerically andanalytically. Johnson provides several sections on this subject in his text [14]. For a slidingcircular contact with friction, little published work is available.Efforts have instead been focused on the simpler case of line contact of cylindricalasperities. Although the assumption of plane strain limits the deformation to twodimensions, the results of these analyses still provide useful insight on what will likelyoccur in the three dimensional case. At the core of this work is the concept of“shakedown”. This theorem was first proposed by Melan [15], and has been extended andeloquently summarized by Johnson et al. [16]. The plastic deformation that occurs afterthe elastic limit has been exceeded can lead to a return to an elastic state of loading forthree distinct reasons. First, the initial plastic deformation results in greater conformitybetween the surfaces and the stresses are attenuated. Second, protective residual stressesare established in the first few loading cycles. Finally, the material strain hardens.Johnson et al. rule out the first reason as a viable means of achieving shakedownfor a semi-infinite solid deforming in plane strain and focus on the remaining two reasons.The model chosen to represent the stress-strain behavior of the contacting asperities willdetermine the theoretical shakedown limit. A kinematic hardening model has been foundto be the most accurate for modeling the stress-strain behaviour of real materials under thecyclic loading which occurs during sliding. Figure 2.3 presents a shakedown map for linecontact showing the variation in the load factor with the friction coefficient, ii. The loadfactor is the ratio ofp0, the Hertzian line contact maximum pressure, and ke, the initialyield stress in shear of the material. The elastic limit is shown as a dashed line, and loadingbelow this level will not result in any plastic deformation. When the friction coefficient islower than 0.25, the elastic limit is exceeded first at the subsurface peak. If the load factorfalls in the region labeled “Elastic Shakedown”, below the kinematic hardening plasticshakedown limit but above the elastic limit, initial plastic deformation will induce residual11stresses or work harden the material to the extent that the stress state will become elastic,and no fhrther plastic flow will occur. If the load factor exceeds the shakedown limit, thecontact will not shake down, and plastic flow will continue with each loading cycle. Atfriction levels higher than i’ = 0.26, the shear stress is highest at the surface, and plasticflow will occur there first, unless the load factor is large enough to induce plastic flow atboth surface and subsurface peaks. A similar shakedown map for point contacts is shownin Figure 2.4.2.3 Contact of Rough SurfacesAll of the contact geometries used in the analyses considered thus far have beensmooth bodies. In reality, bodies have surface roughness, usually from the method offabrication. Recognizing its presence leads to questioning the effect of surface roughnesson contact pressure and stresses. Recently several authors have published work on theeffects of surface roughness on contact pressure, and the common theme among theirconclusions is recognition of the dramatic changes in the distributions of contact pressureand stresses that result from considerations of surface roughness. To demonstrate theimpact of surface roughness on contact mechanics, the results of a few relevant studies arepresented below.The effects of surface roughness on line contact pressure were analyzed byMerriman and Kannel [17]. The surface profiles of three different textures, peaked,Gaussian, and grooved, all flat surfaces with the same centerline average roughness, weregenerated. An iterative numerical procedure was then utilized to determine the contactpressure distribution for the three rough surfaces when pressed against a 9.4 mm radiuscylinder. Pressure distributions were compared graphically with the smooth-body Hertziandistribution under a low and a high load (lOx greater) and two magnitudes of surfaceroughness (the rougher from a 6x multiple of the first surface profiles). Many interestingconclusions were made from this work. The type of surface texture produced a12correspondingly distinct pressure distributions. The Gaussian texture displayed a band ofpressure variation superimposed on the Hertzian distribution. The peaked textureproduced much greater pressure spikes corresponding to the surface peaks, while thegrooved surface had essentially the opposite effect, with zero pressures over groovevalleys. Increasing the magnitude of the surface texture produced similar trends for thethree different textures, but with a proportional increase in pressure spikes. The mostnotable effect of increasing the load was the formation of pressure spikes on the groovedsurface at the edges of the valleys. Pressure spikes ranged from two to six times themagnitude of the Hertzian maximum contact pressure.A fWl three-dimensional analysis was performed by Yongqing and Linqing [18]using a similar numerical scheme to determine the contact area, pressure distributions, andsurface deformations for a flat rough surface pressed against a smooth one. Two roughsurface topographies were gathered by profilometer mapping of a real surface. Theanalysis produced large pressure peaks at the location of taller asperities. Magnitudes ofthe pressure peaks were 50 to 70 times the nominal average pressure (based on theapparent contact area) and, although not included in the purely elastic analysis, plasticdeformation of asperities was expected. Real contact areas were only a very small fractionof the apparent area, and for light loads the real contact area was proportional to the load.However, under higher loads the relationship between load and real contact area was farfrom linear. The authors note that the interaction of asperities had a strong effect on thecalculated pressures and deformations, especially at higher loads, and attribute thenonlinearity of the load-versus-real contact area relationship to the interactions ofasperities. Since only two surfaces, a scraped and a milled surface, were analyzed and noquantitative data was given on the roughness of those surfaces, the effect of greater orlesser roughness on the contact pressure distribution and principally the magnitude of thepressure peaks was not determined.13Numerical calculations were extended by Bailey and Sayles [19] into the contactsubsurface to determine stress distributions resulting from the contact of non-conformingrough bodies. The effects of light friction ( p. = 0.1) were also included. Starting withdigitized topographic data of a ground flat, the pressure distribution for line contact with asmooth cylinder was numerically evaluated. Using this pressure distribution a secondnumerical procedure evaluated the subsurface stress field. When included, frictionalsurface tractions were added as a constant proportion of the contact pressure. A numberof interesting trends were observed. First the numerical stress routines were validated for asmooth-body pressure distribution against classical analytical solutions. Surface profileswere then taken from actual test specimen both before and after wear testing at the sameposition. Before testing, the calculated pressure distribution for the ground surfaceexhibited wide variations and peaks of4p0, the maximum Hertzian contact pressure. Thestress field produced by this pressure distribution exhibited highly stressed zones whichreach a maximum at, or very near the surface. Magnitudes of these surface maximumswere twice the magnitude of the subsurface smooth-body maximum values. A sample ofpressure and stress distribution result appears in Figure 2.5. The authors recognize that themagnitudes of these stresses are unrealistically high and beyond the elastic limit, andplastic deformation was expected. This was confirmed by the pressure distributionobtained from a post-test surface profile, with pressure peaks slightly less than the materialhardness. Stress contour maps from the post-test pressure exhibit less intense surfacepeaks, although the authors do not mention if these post-test still exceed the elastic limit.Analytical results show that the effect of friction on the smooth-body stress field isa shifting of the subsurface stress peak towards the surface and in the direction oftraction, along with an increase in peak magnitude. The results of the numerical stresscalculations for rough bodies with light friction showed the same trend, shifting both thesubsurface peak and the near-surface stress zones closer to the surface and increasingmagnitudes.142.4 Wear and the Near Surface MicrostructureIt has noted in section 2. 1 that extensive plastic deformation is often seen in thenear surface below wear scars. Maximum deformation is observed at the surface anddecreases with depth. The shear strain endured by the material in this deformed region hasbeen estimated from 2 to 8, depending on the sliding conditions [20, 21]. Shear strains ofthis magnitude would not accumulate without the large compressive stresses imparted bythe contact pressure. Crack formation and propagation mechanisms are undoubtedlyrestricted by the compressive stresses. For most metals, deforming to such high strainsdrastically alters their microstructure. Since wear occurs as material removed from thisdeformed region, a better understanding of wear mechanisms will require some knowledgeof the near surface microstructure.Changes in near surface microstructure from wear have been studied and reportedfrequently in the literature. However, for a reader without a solid background inmetallurgy, interpreting the papers on this topic can be time consuming. It is not a simpleissue, and investigations usually involve several complex surface analysis techniques, theresults of which are not always intuitively obvious. Still, it is well worth the effort, and abrief survey of observations and conclusions from several researchers in this area aresummarized below.A very good introduction to this topic can be found in an article authored byRigney and Glaeser [21]. They review the characteristics of the surface region for a single-phase metal subject to wear, utilizing micrographed sections by other researchers. Theauthors point out a highly strained shallow region closest to the surface, and below that aregion of lesser strain, usually extending much deeper than the first region. The first highlystrained region is labeled the fragmented layer, and the second region the deformed layer.The microstructure of the deformed layer is characterized by bending over and thinning oforiginal grains in the direction of accumulated strain. The fragmented layer has a very finesubstructure, crystalline and highly textured. The term cellular is used to describe this15substructure, and is defined as a region of dimensions 1 urn that are relatively free ofdislocations, separated by highly tangled regions of dislocations which serve as cellboundaries or walls. Cell walls are both sources and sinks of dislocations which can movefrom one side of the cell to the other without changing the nature of the microstructure.Dimensions of the cells are suggested to scale close to the ratio of the shear modulus timesthe Burgess vector divided by the effective stress. References are given to observations ofthis cellular structure in situations other than wear where metals are subjected to highstress at low temperatures, including high strain tension, torsion, and compression tests,cold rolling, and at the sides of crack tips in fatigue experiments.In wear specimens, the cells are slabs that are elongated in the sliding directionwith a thickness much less than the longest dimension. The authors note that the complexdislocation tangles comprising the high energy cell boundaries are suitable regions forvoid formation and crack nucleation. Frequently observed flake-like wear debris can thenbe attributed to fracture along cell boundaries. Wear behavior exhibiting flake-like debriscan then be understood by focusing on and investigating the characteristics of the cellularmicrostructure. The authors support this approach by noting a dependence of cellularsubstructure formation on stacking fault energy, with high stacking fault materialsforming cells more readily. A correlation of low wear behavior with low stacking faultenergy has been observed. If the formation of a cellular microstructure aids thedelamination of plate-like wear debris, then the higher wear of high stacking fault materialsis explained by a greater tendency to develop a cell structure.One very pertinent issue not discussed by Rigney and Glaeser is the formation of acellular microstructure in body centered cubic metals, like ferritic steels, in which stackingfaults do not occur. In a study of wear particle formation during unlubricated sliding ofcarbon and alloy steels, Salesky et al. [22] observed both plate-like debris and a finedislocation cell structure. They observed well-defined cell structures within lum of thewear surface in TEM (Transmission Electron Microscopy) cross-sections of ferritic,16pearlitic and even martensitic steels. The dislocation cells tended to have three-dimensional plate-like sizes, and cell thicknesses were quite similar to the thicknesses ofthe flake-like wear debris. In the ferritic and pearlitic steels, larger dislocation cells wereobserved immediately beneath the wear interface, to a depth 0.2 -0.3 tim. Below that, athinner and longer dislocation cell structure was observed. The authors suspect the firstlarger cell region to have undergone dynamic recovery and consequently a large reductionin strain density. This would create a large strain gradient between the recovered cells andthe underlying heavily dislocated cells. Wear cracks were expected to initiate at this straingradient and lead to the delamination of the recovered cells into flake-like debris.2.5 Wear Mechanisms and ModelsThe study of wear mechanisms and wear particle formation is usually undertakenwith intentions of modeling those phenomena and eventually arriving at a relationship thatcan be used to predict the wear of real components. Wear models appearing in theliterature differ as widely as the wear processes which they describe, and a mode! isusually specifically associated with a particular wear process. A review of severalinteresting wear mechanism and models related to sliding wear is given below.2.5.1 Wear by DelaminationThe delamination theory of sliding wear was first proposed by Suh [23] in 1973 toexplain observations of thin flake-like wear particles. His concept stimulated a great dealof research which has resulted in changes to and fbrther updates of the theory [24]. Theformation of loose wear sheets is explained by the following sequence. In the early stagesof sliding a relatively smooth surface is generated by the fracture or deformation of initialsurface asperities. The surface traction acting at the contact points induces incrementalplastic deformation which accumulates with repeated loading. As the subsurface17deformation continues, cracks nucleate below the surface and propagate parallel to thesurface. Finally the cracks shear to the surface and a long thin wear sheet delaminates.Three mechanisms are associated with delamination theory. There is the initialasperity deformation and fracture, which is associated with the “running-in” of slidingsurfaces. Suh notes that asperity interactions will also occur in the steady-state, sinceasperities will be generated by the uneven delamination of wear sheets. However, the wearparticles formed will be much smaller than the sheet-like particles formed by thedelamination process. Hence the delamination mechanisms of crack nucleation andpropagation will control the steady-state wear process. In materials with pre-existingcracks or in which cracks readily nucleate, crack propagation will determine the wearrate. Conversely, in materials with rapid crack propagation rates, crack nucleation will bethe wear rate determining mechanism. Fracture mechanics treatments of crack nucleationand propagation have produced analytical models of these two mechanisms; howevermany of the required material properties are not currently available. This has hinderedestimates of delamination wear from first principles.A number of researchers have proposed that plate-like wear particles may begenerated by entirely different mechanisms. An excellent review of some alternateconcepts of delamination mechanisms has been published by Samuels et a!. [25]. In situexperiments run in scanning electron microscopes support these different proposals.Delamination of plate-like particles has been bbserved after only a single pass of anasperity. The authors note that while the existence of delamination is well established, thetheory of formation of the lamellar particles remains open to alternative mechanisticinterpretations.Recently, a very different explanation for the formation of flake debris duringlonger duration steady state wear has emerged. Investigations have shown that thematerial which delaminates is not merely the deformed base material but instead amechanically mixed layer of transfer material, which can include material from both sliding18surfaces as well as the sliding environment, such as oxygen [26]. A very detaileddiscussion of the processes involved in producing this layer of transfer material can befound by Rigney et al. [27]. If the hardness of the transfer material is not sufficient toembed into the base material, it will accumulate on the surface until it reaches a criticalthickness and subsequently delaminate as flake debris. Critical thickness was determinedfrom a friction model of coated systems and occurs when the friction coefficient at thetransfer/base material interface equals the friction coefficient between transfer layers. Thistheory provides an explanation for observations of flakes of remarkably uniform thicknessfor a particular sliding system. However, until additional work on the rate of growth of thetransfer layers is accomplished, further development of the concept into a predictive wearmodel is stalled.2.5.2 Wear by FatigueTypical calculations of the thickness of the wear volume removed per cycle areoften less than the thickness of one atomic layer [5]. Thus under steady-state conditions,it is highly unlikely that the complete formation of a wear particle occurs during a singleloading cycle. Particle formation more likely results after many cycles as the culminationof tiny increments of damage. It seems plausible then to consider wear as a fatigueprocess. In section 2.2 it was noted that for a point contact with friction, one of theprincipal stresses alternates between cofripression and tension. This stress fluctuation hasbeen suggested by some researchers as the cause of wear by fatigue. Recent work onfatigue and wear is considered below.One of the earliest wear models utilizing fatigue concepts was presented byHailing [28], although his analysis did not include interfacial shear stress. Thus it isapplicable only to situations where friction is very small, such as rolling. A more completeanalysis, which included the effects of tangential tractions, came later from Jam andBahadur [29, 30]. Their model determines the wear from a combination of the WOhlers19fatigue equation for the number of cycles to failure, an estimate of wear particle volumebased on contact dimensions, and a statistical treatment of asperity contacts. Predictedwear rates compared quite well to experimental results for polymeric materials.Most of the development of fatigue concepts for metals has been directed atmodeling surface pitting which occurs in mixed sliding and rolling contact of gears,bearings and cams. Two recent publications on this topic provide a good example of theextensive and sophisticated analyses that have performed. Hanson and Keer [311 start witha pre-existing microcrack and use three-dimensional fracture mechanics to predict themixed mode fatigue crack growth geometry and propagation life. They include the effectsof lubricant entrapment within the crack and compressive residual stresses. Predictedfatigue life for the first occurrence of surface pits compared favorably to experimentalobservations for a twin disk test. An important observation was also made in their report.They note that surface pitting cracks only propagate in a sliding situation on the followersurface, on which the tangential force acts in a direction opposite to the direction ofmotion of the load. In the case of pure sliding of a single asperity on a flat surface, thecontact area is stationary on the asperity, and moves in the direction of the surface tractionfor the flat (i.e. driver configuration).A more sophisticated contact fatigue model has been presented by Zhou et al. [321,which adds a crack initiation model to the crack propagation model of Hanson and Keer[311 to determine the entire contact fatigue life. Results from their model showed that atlow contact stresses, crack initiation is the dominant fraction of the total life, while athigher stresses, crack propagation is dominant. However, the fatigue lives predicted byboth models for high contact pressures near the elastic limit, or a shakedown factor of 2,are in the range of 1O to 106 cycles. Extension of these models into the realms of wearprediction would not be appropriate, since they are based on long-duration, low stresscrack propagation laws. In a real sliding wear situation, fracture would likely be expectedto occur several orders of magnitude sooner.202.5.3 Damage Accumulation Wear ModelsMost of the early work on low-cycle fatigue wear modeling has been based on theManson-Coffin fatigue relationship, which uses a strain criteria to predict the number ofcycles to failure. Hailing’s wear model [281 used this relationship, although he assumedthat all asperity interactions, both elastic and plastic, contribute to the fatigue process.This simplifying assumption, along with a few others were eliminated and improved uponby Finkin [33], and his model firmly established the concept of wear resulting fromdamage accumulation in the near surface.A more recent damage accumulation model by Kimura [34] accommodates two ofthe omissions of earlier models. First any damage inflicted on the wearing surfaces fromthe fabrication process is included. Thus transitions in the early stages of wear, which areoften influenced by initial surface conditions, can be accounted for. Second, the modelextends beyond the thickness of a wear particles and includes the damage accumulating inthe subsurface. However, it is a purely mathematical model, which treats damage inprobabilistic terms. Thus the constants used in the mathematical formulations are not well-known material parameters, but instead only have qualitative physical significance. Yet theappeal of this model is its ability to account for many complex wear behaviors, includingchanges in wear rate at start-up and during steady-state.Critics of damage accumulation models often argue that the models are too farremoved from the true wear mechanisms and processes, since they are based on statisticaltreatments of asperity interactions, in the case of Hailing or Finkin models, or on damageprobabilities, in Kimura’s model. However, recent work by Alpas et al. [20] presents apurely mechanistic approach to damage accumulation. The critical depth for maximumdamage accumulation is determined from a relationship for void growth, which ispromoted by near-surface strains but inhibited by the hydrostatic pressure of the contact.The model was able to determine the depth at which the damage rate is highest, but needsfbrther development to predict wear.212.6 The Third Body ApproachCurrently in the study of tribology, only thick film lubrication has been completelyanalyzed, while other areas, including mixed and boundary lubrication and dry frictionseem to de1,’ unification into a single general theory of tribology. However, a group ofFrench researchers has presented a convergent view of all of tribology which is based onthe concept of interface or third body flow and its load carrying capacity. This third bodyapproach is a fresh, novel, and intriguing concept, which may provide a successful path toa complete understanding and mastery of the complexities ofwear. Important elements ofthe third body approach are collected from the publications of its proponents anddiscussed in this section.Godet [35) defines interfaces or third bodies in a material sense as a zone whichexhibits a marked changed in composition from that of the rubbing specimens, and in akinematic sense as the thickness across which the difference in velocity between solids isaccommodated. The third body contact model is defined by Berthier [36]. First bodiesrefers to the two rubbing specimens. The entire interface between first bodies isconsidered the third body. The third body bulk is the central portion of the third body,with two third body screens separating the third body bulk from the two first bodies.The ubiquitous nature of third bodies is noted by Ludema [37] in his discussion ofthird bodies by acknowledging the adsorbed gases covering all surfaces in a normal earthatmosphere and the oxide layers on most metal surfaces. These layers are considerednatural third body screens by Godet [35] in his explanation of fretting and the transitionfrom a two body to three body contact. In unlubricated fretting, these natural screenskeep friction low for the first few cycles until they are eliminated. A subsequent period offirst body contact results in first body near-surface structural changes. As debris forms andis trapped in the interface, the contact is transformed from a two body to a three bodytype. When the formation and escape of debris reach consistent levels, steady state22conditions prevail. This example illustrates that although a contact may start out as a twobody, the contact transforms into a three body situation in the vast majority of cases.A key element in the third body theory is the principle of velocity accommodation,whereby the difference in velocity between the two first bodies cannot be accommodatedacross a body of zero thickness [35]. This concept is well established in fluid mechanicsas a boundary layer between a flow field and a wall. Velocity accommodation in thirdbody contacts is more complex. As identified and experimentally verified by Berthier [36],it can occur at either of the two first bodies, in the bulk third body, or in the third bodyscreens. Four modes of velocity accommodation have also been observed, including anelastic, rupture or cracking, shear, and roller formation modes. The five sights and fourmodes leads to twenty possible velocity accommodation mechanisms.In the third body approach, wear has three stages; particle formation, entrapmentin the contact, and finally elimination. This is a much wider view of wear, since the wearprocesses described above, delamination, fatigue, etc., are now seen as only particledetachment mechanisms and not wear. Adding considerations of particle entrapment andejection increases the complexity of wear. However, recognition of the importance of thetwo subsequent stages can lead to new ways to control wear. As Godet notes [35], theprotection afforded by debris beds or third bodies is beneficial in most instances, and thusan increase in debris entrapment or a reduction in ejection can cause a significant reductionin wear.Chapter 3.Wear Testing Equipment and Procedures3.1 NRC Fretting Wear RigAll of the wear tests for this study were conducted on a custom-designed rig owned bythe Institute for Machinery Research of the National Research Council of Canada. Figure 3.1provides a simplified view of the major mechanical components. A brief description of theimportant components follows below.Two electromechanical (solenoid) shakers are mounted several feet apart on a solidconcrete base and aligned with directions of action ninety degrees apart. The shakers provideforce or displacement to an annular-shaped dynamic specimen holder secured to the shakersby two shafts. The shafts connect to the dynamic specimen holder through a series of flexureswhich allow deflection of the annulus only in directions perpendicular to either shaft.Displacement of the annular specimen holder in the direction orthogonal to the two shakershafts is prevented by a circular flexure. This fiexure is attached to the annulus and the side ofthe autoclave enclosure surrounding the dynamic and stationary specimen holders. A detailedview of the holders is shown in Figure 3.2. The annulus is not easily removed, consequentlythe dynamic specimen is attached to the annulus by a secondary holder which is easilyremoved.The specimens are centrally located in a steel autoclave, which acts as a lubricantcatchment and is securely attached to the concrete base. The sides of the bowl-shapedautoclave extend higher than shown in Figure 3.1, and the two openings in the bottom forthe shaker shafts are sealed by rubber diaphragms. The stationary specimen is mounted upon asemi-circular beam which is inserted and secured through openings in the sides of the bowl.The sectioned view of this beam is shown shaded in Figure 3.2.2324Lubricating fluid is circulated by a small electric rotary pump, and passed through apaper filter before returning to the system. A nozzle positioned in close proximity to thestationary specimen directs a steady flow of lubricant across the area of contact.3.1.1 InstrumentationElectronic instrumentation associated with the test rig serves to either monitor of testconditions, or control the dynamic specimen motion and normal load, and in instances bothfunctions. Instrumentation for test condition monitoring is described below, followed by amore extensive explanation and history of the rig control systems.The tests conducted for this study were intended to be of a pure uni-directional slidingnature, and impacting, or motion in the direction of normal loading, was intentionally avoided.Displacement in the third possible direction, orthogonal to the directions of the two shakers isprevented by the restraining rod attached to the annular holder, as mentioned above. Thusdisplacement is only monitored for motion tangential to the shaker applying the normal force.Displacement monitoring is accomplished by a standard commercially produced inductivedisplacement proximitor mounted securely to the stationary semi-circular beam and aimed at asteel target attached to the annular dynamic holder. Provided the specimens are rigidly held intheir respective holders, the relative movement of the mounting hardware is transformed bythe proximitor into an electrical signal. The voltage level of this signal is linearly proportionalto the displacement.Forces acting between dynamic and stationary specimens are measured by theelectrical output of a triaxial piezo-electric load cell mounted between the stationaryspecimen and the anchored semi-circular beam. Forces imparted to the stationary specimen byits dynamic counterpart are transmitted through the load cell, then to the beam, and ultimatelyto the rig base. Resistance changes in the load cell induced by the forces acting on the cell aretranslated into a linearly-proportional electrical signal by charge amplifiers supplied andcalibrated by the load cell manufacturer. Carefully installed, the triaxial load cell is aligned25with the first axis recording forces normal to the specimen surface. The second axis is alignedto record the tangential or frictional force acting in the direction of dynamic specimendisplacement. Since the third load cell axis corresponds to the direction orthogonal to thedirections of normal force and displacement, and the rig is restrained in this direction, thesignal from the third axis is not monitored.At the start of this study, the displacement, normal and tangential (frictional) loadsignals were fed to a digital oscilloscope. The oscilloscope would display the signals, recordminimum, maximum, and average amplitude values over the duration of a test, and transmitthe statistical records and display to a printer to provide a hard copy of the test data. Thisinformation provides at the very best only a general sketch of what occurred during a test, andfast transitions in test conditions, unless observed and noted manually, are not captured.Improvements in test monitoring capabilities were required.NRC electronics personnel improved the data monitoring capabilities by developing acontinuous data acquisition system. Displacement and force signals are fed to a microcomputer equipped with several analog-to-digital processing cards, which digitally sample thesignals at a high, user-selected frequency, and store the signal levels as binary data on a largecapacity hard disc. A decoding and processing program then converts the binary test data intocorresponding displacement and load digital signals, and performs additional computations,including determining the coefficient of friction and work increment on a single cycle or multicycle basis. The package provides the ability to produce quantitative measures of testconditions over an entire test or with equal ease at any segment of the test. Test data can alsobe loaded into standard spread-sheet software packages, for further computational work orgraphing.3.1.2 Control SystemsBefore beginning a description of the systems used to control the test rig, a morecomplete description of the desired pattern of displacement and loading conditions is26warranted. As mentioned earlier, the desired condition was that of uni-directional slidingwithout impact. This motion was achieved by a sinusoidal oscillation of the dynamicspecimen, with normal loading occurring only during one direction of oscillation, and noneduring the return motion. However, to prevent the dynamic specimen from bouncing on thereturn stroke, a constant light normal load of four to six Newtons was applied. Normalloading was then applied as a truncated or half sinusoid. The phase lag of the normal load withrespect to the displacement was set so that normal loading would commence at the start of anoscillation in the wear direction, peak at the midpoint of displacement, and end as thespecimen begins its return motion. This pattern of motion and loading is illustrated in Figure3.3. It was chosen for its similarity to the loading conditions encountered by an asperity on areal surface.3.1.3 Open Loop ControlThe original control system, used by Magel [1] for his work with the NRC test rigutilized a personal computer with digital-to-analog signal generating boards to send asinusoidal signal to the tangential shaker power amplifier and a truncated half sinusoid to thenormal shaker. Sensitive rheostats placed between the signal generating boards and the shakeramplifiers were manually adjusted to select and tune the displacement and normal loadamplitudes at the beginning and throughout the test duration. MagePs tests were performed ata frequency of 30 Hertz, and through a trial-and-error basis the optimum phase lag was foundwhich maintained the desired quarter cycle lag between the displacement and the normal load.However Magel experienced difficulties finding an optimum phase lag and maintainingconsistent levels of force and displacement, especially for long test periods.Improved signal generating boards were acquired by NRC before the beginning of thisproject. Unlike the previous system, the new boards are configured to share the same timingclock and run continuously for the entire test duration. Thus improvements in maintaining thecorrect phase lag were anticipated. An improved method for determining the required phase27lag between tangential and normal input signals was also sought. The mechanics of the testrig were modeled as a single degree-of-freedom system of force generator, mass, spring, anddash pot in the tangential direction. This model is detailed in Figure 3.4. In the normaldirection, the system was represented as a stationary force generator, uncoupled to thetangential model. Model parameters were determined from available data and standardvibration test methods. The stiffness of the support springs in the shakers and the solenoid coilmass were supplied by the manufacturer, and the shaft and specimen holder masses wereestimated. Damping was first determined by the log-decrement technique in air with theshaker powered down; however, powering up the shaker added additional electromechanicaldamping, so the combined damping was subsequently determined by the frequency responsetechnique. Results showed that the system was heavily damped, not critically or over-damped(c /cnt .- 0.6), and had a natural frequency of vibration at approximately 24 Hertz. Byconducting wear tests at the natural frequency, the steady state displacement response wouldthen lag the input force by a predictable ninety degrees, and hence the tangential and normalinput signals could be generated in-phase. The substantial system damping restrains thedisplacement from excessive amplitudes commonly associated with lightly damped systems,and the system operates at maximum efficiency.The new signal generating cards were then programmed to produce a phase-lockedsinusoid and truncated half-sinusoid fed to the tangential and normal shaker power amplifiers.Level control was accomplished again with rheostats to attenuate the signals before input tothe power amplifiers. This control configuration is schematically illustrated in Figure 3.5While it vastly improved phase-lock problems; however, the open loop configuration of thesignal level control placed heavy demands on the test operator to close the control loop andtune displacement and normal load input signal levels to steady levels. Displacement controlfor higher friction tests often proved impossible to maintain at constant amplitude. A closedloop control scheme was considered the only solution.283.1.4 Closed Loop ControlUsing two high speed control cards, a feedback control system was custom designedand installed by a consultant. The system used a proportional plus derivative control algorithmto produce accurate sinusoidal displacement. Feedback control of the normal load level wasalso added a few months later. A complete description of the control scheme, includinghardware and software can be found in the Users Manual [38j.Using the feedback control system vastly simplifies the testing procedure. The desiredlevels of displacement and normal load, testing frequency and test duration are entered into aninteractive program running on the host computer, which contains the control cards. The dataacquisition computer and software would also be primed with the test duration and samplingfrequency, and both systems started simultaneously by a single keystroke. The controlprogram maintains the requested levels of displacement and normal force within tight limits,while the data acquisition system records displacement, normal and tangential force forsubsequent analysis. The role of the test operator is reduced to that of an interested observeror guardian against any extraordinary problems. With variations due to operator error,inherent in open loop control, eliminated, improvements in the repeatability ofwear testsunder identical conditions were anticipated.3.2 Specimen PreparationVery hard 12.7 millimeter diameter (1/2 inch) alloy steel balls were used exclusivelythroughout this study as the dynamic specimen. The balls were purchased from the FalexCorporation and are standard wear test balls. Relevant characteristics of the balls are listed inTable 3.1.Stationary specimens were fabricated by the NRC machine shop from SA.E 410stainless steel round bar stock into 22 mm diameter discs approximately 2.5 mm thick. Theeffects of hardness and roughness were investigated by controlling the heat treatment andsurface finish of these discs, as described in the following subsections.293.2.1 Disc Heat TreatmentsSAE 410 stainless steel is a martensitic grade with a moderate chromium content andmarginal content of other alloying elements, including manganese, silicon, nickel, carbon,phosphorous, and sulfur. Using simple heat treatments it can be hardened, which maximizesthe martensite content, to a hardness of approximately 450 Vickers, or annealed into a ferriticstructure with a minimum hardness of 150 Vickers. Hardnesses between these two extremesare achieved by hardening followed by carefi.il tempering.The round bar stock was supplied by the steel producer in a tempered state mid-waybetween the two extremes, with a hardness of approximately 300 Vickers. In this state themachineability is still quite favorable, and specimen discs were easily turned. The magneticproperty of this steel also allowed for easy machine grinding, ensuring good flatness andconstant thickness of the discs. The NRC machine shop facilities includes a small electricfurnace which is adequate to perform the simple heat treatments of full hardening orannealing; however, it lacks adequate temperature controls to carry out the controlled coolingrates necessary for consistent tempering.Full hardening was accomplished by loading de-greased specimens into a 980° Cfurnace, holding for 45 minutes, and oil quenching. Full annealing required holding at 760° Cand very slow furnace cooling, with the furnace door closed and heating elements shut down.Hardnesses were checked using a Leitz micro-hardness tester. Attempts were made to temperspecimens to intermediate hardness values, but without tight control of the cooling rate,specimens could not be tempered repeatably to a consistent hardness. Thus the three levels ofhardness used throughout this study in order of increasing hardness are the fully annealedstate, the intermediate hardness of the as-supplied tempered state, and the fully hardened state.Reference to these hardness conditions throughout this thesis will be made using the terms“annealed”, “tempered”, and “hardened”. Equivalent hardnesses in several hardness scales aresummarized in Table 3.2.30A few disc specimens were also stress relieved after the disc surfaces were roughened.Stress relieving was carried out using the NRC shop furnace at 3000 C, into which degreaseddiscs were inserted for a half hour and air cooled upon removal. A very thin oxide film didform on the disc surfaces, giving them a dull orange-grey appearance. This surface layer couldnot be avoided, unless stress relieving was performed in a more sophisticated furnace with aninert gaseous environment.3.2.2 Disc Surface Roughness PreparationA variety of techniques were employed to vary the surface roughness of the stationaryspecimen discs. Extremely flat surfaces were produced by polishing the discs with diamondgrit abrasives according to standard metallurgical techniques. Discs prepared for smoothsurface tests were polished up to and including 1 jim diamond, since the minimal reduction insurface roughness for very fine polishing beyond this finish did not warrant the additionalpreparation time. This justification was made on the basis of a dimensional analysis as follows.The wear scar sizes produced in this study were of the order of millimeters, while polishing to1 jim generates a finish with features three to four orders of magnitude smaller than the scarsizes. Fine polishing to 0.05 jim almost doubles the total preparation time, but would onlywiden the dimensional difference by another order of magnitude. Thus for the sake ofexpediency, polishing was curtailed to a 1 jim finish.Producing rougher surfaces initially required some experimentation until techniqueswere found that produced consistent results. Sandblasting of the disc surface with silica sandwas the first technique investigated, using a small re-circulating sandblasting unit located inthe NRC machine shop. Initially the results looked promising, as sand blasting is oftenreported in the literature as producing a rough surface with a Gaussian distribution of surfacefeatures. However, microscopic examination of the surfaces produced for this study revealeda significant proportion of the sandblasted surface had not encountered any impacting sand,although to the naked eye the surface appeared to have an evenly distributed roughness31everywhere. Even after repeated and extensive exposure to the sand jet, examinations under alight microscope continued to reveal unaffected areas. There was also no simple means toproduce rougher or finer surfaces, without ordering new lots of coarser or finer sand for theblaster. Consequently other techniques were investigated.An effective procedure for consistently producing a variety of surface roughnesses wasdeveloped by repeatedly pressing the disc surface against silicon carbide abrasive paper.Pressure was applied with a 25 ton hydraulic press although loading was limited to 5 tons (—45 kN) to avoid thinning or flattening of the specimen discs. To ensure the entire surface of adisc was roughened, pressing was repeated twenty times with a fresh area of abrasive paperused for each step. An extremely rough surface was produced using 60 grit abrasive paper;however, like sandblasting, even after a large number of pressings some unroughened areasremained. To circumvent this problem discs intended for 60 grit finish were first pressedagainst 120 grit paper for ten repetitions. This pre-treatment effectively roughened all areas ofthe disc and followed by pressing with 60 grit paper, it produced an extremely rough surfaceover the entire disc surface. A summary of the roughnesses parameters produced by thetechniques described above is listed in Table 3.3.Producing rough surfaces with silicon carbide abrasives as described above couldpotentially trap fragments of silicon carbide on or below the roughened surface. Thesefragments would then act as an artificial and undesirable third-body abrasives in the ensuingwear test. Consequently great care was taken to ensure the roughened surfaces were free ofsilicon carbide particles. Between each pressing, discs were cleaned by a blast of compressedair, by light scouring with a scouring pad, and by rubbing with a soft cloth. Prior to pressing, afew drops of water were applied to the surface to act as a lubricant. After roughening asurface with 60 grit paper, examination under an optical microscope revealed thatapproximately five percent of the surface was covered with silicon carbide. Most of thefragments were lodged in deep valleys and crevices. Ultrasonically cleaning in soapy water fortwo to three hours effectively removed the debris; however, as a precaution discs roughened32with 60 grit paper were optically checked for any silicon carbide fragments. Higher grit papersproduced substantially less debris on the disc surface after pressing, and lengthy periods ofultrasonic cleaning easily removed this debris.3.3 Test MatrixThe development of the test matrix was characterized by the improvements in the testrig control system.3.3.1 Open Loop Controlled TestingAn initial series of twenty-six wear tests were carried out with the open loop controlsdescribed in section 3.1.3, and the test conditions are summarized in Table 3.4. Many of thesetest were conducted using a mixture of distilled water and a commercially available cuttingfluid. The major component of this cutting fluid is refined paraffin and it acts as a veryeffective boundary lubricant. A complete list of the components is provided in Table 3.5.Diluted cutting fluid was used to reduce the friction since sliding with distilled water alonegenerally produces a coefficient of friction near 0.35, but by adding cutting fluid in a ratio of1000:1 parts water to cutting fluid the coefficient offiction is reduced to approximately 0.1.More concentrated mixtures of cutting fluid do not significantly reduce the friction coefficient.More dilute mixtures of cutting fluid produce higher friction coefficients closer to the value ofwater alone; however the.relationship between .cutting fluid concentration and the resultingcoefficient is not sufficiently consistent or predictable to successfully alter the frictioncoefficient between these two limiting values of t = 0.1 and 0.35 in a controlled andrepeatable fashion.The twenty-six open ioop tests can be arranged into four distinct groups. Eight tests inthe first group were conducted with a highly concentrated mixture of cutting fluid (40:1), avery high normal load close to the capacity of the shaker, and a large range in the number ofwear cycles. Several short duration tests were then conducted in water lubricated and33unlubricated conditions. To investigate the early stages of the wear process, a third group oftests was conducted with a moderate normal load, a 1000 1 concentration of cutting fluidlubricant, and for short test durations. These first three test groups were performed onpolished discs of the intermediate hardness. The fourth and final group of open loopcontrolled tests used polished discs in the fully hardened state and were run for a variety ofwear cycles and lubricant conditions. The intent of the fourth group of tests was to revealeffects of the increase in disc hardness.3.3.2 Feedback Controlled TestingAdditional tests were carried out with the closed loop control system, focusing onhardness and surface finish effects under a variety of lubrication conditions. Feedbackcontrolled tests can be divided into two separate sections, according to the length of the testduration, with five thousand cycles chosen as an arbitrary division between short and longduration tests. Conditions for long and short duration tests are summarized in Tables 3.6 and3.7 respectively.The long duration tests can be arranged into four distinct groups, primarilydistinguished by differences in the lubrication conditions. To investigate the effects of surfaceroughness, two groups were performed on the four surface roughness conditions, first withdilute cutting fluid as lubricant, and second with distilled water. The feedback control systemallowed smooth operation under high friction conditions, and this capability was utilized in thethird group of tests, run without any lubrication. To examine the effects of hardness, these drysliding tests included tests on the three levels of disc hardness, all with similar polishedsurface finishes. Large variations in worn mass loss results from the repeated tests in the firstthree groups prompted testing a statistically significant number of specimens under identicaltest conditions. A fourth group consisted of sixteen tests with discs of the intermediatehardness and sandblasted surface finish. They were run for fifty thousand cycles with distilled34water lubrication. A sandblasted surface finish was chosen for its minimal surface preparationtime requirements.To investigate the early stages of the wear process, three groups of short durationtests were carried out. The first group ran for fifty, five hundred and five thousand cycles foreach of the three hardness conditions, and with dilute cutting fluid lubrication, This group wastested using an early version of the feedback control program which started the test with a fivecycle ramp-up of displacement, but without a ramp-up of the normal force. When a five cycleramp-up was added to the normal load control, additional tests of extremely short durationwere performed. In the second group, tests were carried out on annealed discs for ten andtwenty wear cycles under distilled water and dry lubrication conditions. The various shortduration and lubrication conditions were then repeated in the third group except with fullyhardened discs.3.4 Post Test AnalysisIn the case of wet tests, upon removal from the rig, both test specimens wereimmediately immersed in ethanol to remove moisture from the wear scar and preventcorrosion of the wear scar or the debris surrounding the scar. Post-test analysis of the wearspecimens was then performed using a number of different techniques.Visual examinations of the worn specimens were performed with a powerful Zeissoptical microscope. Lengths and widths of disc wear scars were determined on an opticalmicroscope equipped with micrometer adjusted x-y table. Higher resolutionphotomicrographs were obtained by viewing the specimens in the NRC scanning electronmicroscope (SEM). Sectional views of the disc wear scars were obtained using standardmetallurgical techniques, with one time saving short-cut. To prevent the edge of an area ofinterest from becoming rounded during grinding and polishing, the surface is usually nickelplated prior to sectioning. The layer of hard nickel will then protect the edge under study fromrounding. Unfortunately, nickel plating is a slow process. Instead, discs were sectioned with35the low speed diamond saw, and both halves of the discs were placed together, with the wearscar on the inside. To accommodate any roughness from the scar or disc surface, a thinplastic shim was placed between disc halves. Sectioned halves were then mounted inthermoplastic, ground and polished. The second half of the disc provided reasonableprotection for the edge of the wear scar from rounding during polishing, and eliminated thelengthy process of nickel plating the disc before sectioning. The microstructure of thesectioned discs was display by etching with the appropriate acid solution.Mass lost through wear was determined by pre- and post-test weighing of thespecimens on an analytic balance. Specimens were ultrasonically cleaned in stodard solventand ethanol and desiccated to ensure the resolution of the balance was not nullified bycontaminants on the specimens.The surface roughness and the depth and shape of the wear scars was determined witha Talysurfprofilometer. The NRC instrument was also equipped with a computer controlledtable which produced 3-D surface imaging or mapping. Files from the surface mapping werealso run through curve fitting software developed by Magel [1] to determine the best-fit radiusof curvature in the sliding or longitudinal direction, and the transverse direction orperpendicular to the sliding direction.Friction coefficient values during tests and the friction work done over the testduration were obtained with the data acquisition and processing system described in section3.1.1.Chapter 4Results and DiscussionIn an ideal situation, apparatus is first developed and then used to perform thedesired series of tests. In real situations, the sequence of testing is often determined by thedevelopment of the apparatus, as was the case in this study. The matrix of test conditionsexpanded as improvements of the test rig control system were made. While thechronological order of testing is given in detail in section 3.3, the presentation anddiscussion of the results from the entire test program have been arranged along morelogical patterns of increasing test duration and by the type of results. Those types includesurface examinations, mass loss results, friction coefficient data, work input and wear scargeometry, and they follow in the sections below. All of the numerical results aresummarized in Tables 4.1 through 4.4. These tables follow the chronological pattern oftest matrix conditions given in Tables 3.4 through 3.7, with open ioop controlled testresults presented first, followed by the long duration and then shorter duration feedbackcontrolled results.4.1 Surface ExaminationsMany of the wear scars on disc specimens were examined and photographed byoptical and scanning electron microscopy to reveal the mechanisms involved in the wearprocess. Test duration, friction or lubrication conditions, and hardness condition of thedisc had a variety of effects on the appearance of the wear scar. Results of the surfaceexaminations follow below, along with discussion of the effects due to changes in testconditions. Unless noted otherwise, in all micrographs, the direction of the wear stroke isfrom left to right.36374.1.1 Very Short Duration TestsSEM micrographs of the wear scars from the group often and twenty cyclefeedback controlled tests conducted in moderate (water lubricated) and high(unlubricated) friction conditions revealed the events that occur at the start of the wearprocess and how disc hardness and friction can alter those events. Micrographs of polishedannealed discs from water lubricated tests often and twenty cycles respectively are shownin Figures 4.1 and 4.2. Smooth and continuous plastic deformation of the disc surface hasshaped the wear scar into an ellipsoidal indentation. Higher magnification of the ten cyclescar interior (Figure 4.1(b)) reveal a very smooth surface with one to five micronscratches running the length of the scar. Fracture of material from scratch or groove edgesappears either imminent or to have occurred. Micrographs of the twenty cycle scar, inFigure 4.2, displays two parallel bands of scratches within the ellipsoidal indentation. Therange in scratch width increased to ten to twenty microns, with smaller scratches withinthe larger ones. Closer inspection of the bands reveal a rough and irregular layeredtexture in most areas, making estimation of groove widths difficult. No loose debris wasvisible, although the highest magnification micrographs reveal features that suggest insome areas delamination either is imminent or has already begun. In particular, the patchindicated with black arrows in Figure 4.2 (b) appears ready to flake away, while the veryabrupt edges identified with white arrows in 4.2 (b) and (d) suggest sites from whichparticles have recently detached.When ten and twenty cycle tests were repeated on frilly hardened discs, the resultsappeared quite different. SEM micrographs of the scars on the hardened discs are shownin Figure 4.3. In the ten cycle test, a strip with relatively constant width on the discsurface has delaminated. It is this delaminated strip which gives the scar its shape andgeometry, rather than the plastic deformation into an ellipsoidal indentation seen on theannealed discs. Particles are visible around most sides of the scar. In the trailing end of thescar, a large portion of the delaminated layer can be seen fractured into three sections.38However, in the twenty cycle test, the contact surface did not delaminate. Instead, a veryfaint ellipsoidal indentation was formed. To the naked eye, the scar was almost invisible.Only a very subtle difference in the reflection from the polished disc surface made the scardiscernible. High magnification in the SEM showed the scar surface to be very smoothwith a few fine scratches and tiny pits. No visual evidence was found to explain why theten cycle hardened disc test produced a wear scar by coarse surface delamination while thetwenty cycle test did not. This question will be given further consideration when thefriction traces from these tests are presented in a subsequent section. However, it isapparent that when the initial wear scar is formed by continuous plastic deformation,increasing disc hardness severely reduces the extent of deformation.Micrographs were also taken of the wear scars from a similar set of ten and twentycycle tests run in high friction unlubricated conditions. The photos show that wear scarson both the annealed and fully hardened discs were formed by the relatively coarsedelamination process seen previously in the ten cycle water-lubricated hardened disc test.Micrographs of the annealed disc scar from the ten cycle unlubricated test are presentedin Figures 4.4 and 4.5. A limited amount of plastic deformation is visible around theperimeter of the scar, but the interior is dominated by surface layer delamination features.Instead of the ellipsoidal indentation, the scar appears as a narrow and relatively constantwidth strip which has sustained large shear strains in the wear stroke direction. Highermagnification views of the leading edge of the scar shown in Figure 4.5 (a) and (b)present smooth-topped particles or islands that were obviously part of the originalpolished surface of the scar interior. These photos indicate that in the front of the scar,much of the original scar surface has delaminated. However, at the tail of the scar, thesmooth-topped areas appear continuous and securely attached to the adjacent material.Here, the original surface has yet to be removed. Thus the entire scar area has not yet beenstripped of a complete layer. In the center of the scar, several large cracks have shearedopen. A high magnification view of the interior of one of these cracks is shown in Figure394.4 (d). The repetitive cup-like appearance indicates that fracture occurred by ductilemicro-void coalescence. A large wear particle, pointed out in Figure 4.4 (a), was observedjust below the wear scar. A closer view of the particle shown in Figure 4.4 (b) reveals itsmulti-layered structure. The few smaller particles that were attached to this larger onemost likely originated from the fractured edges of the jagged layers.Another ten cycle test run with a two-fold increase in the normal load, alsoproduced a scar formed by coarse delamination. However, the final return stroke of theball peeled back the material in the trailing edge of the scar. Micrographs of this area ofthe scar are shown in Figure 4.6. The highest magnification photo shows the top layer isapproximately 5 microns thick, with thinner layers below. The boundaries between layersappear very well defined in the peeled-back sections, with some separating into gapsbetween layers. This suggests very weak adhesion between layers.In the twenty cycle tests, complete surface delamination has occurred. SEMphotos of the annealed disc scar after twenty cycles, presented in Figure 4.7, show verylarge and thick particles about to separate from the trailing edge of the scar. The width ofthe scar has noticeably increased, but outside the scar, the plastic deformation associatedwith ellipsoidal indentation remains very minor, and development of the scar geometry isobserved to be principally governed by coarse delamination.On hardened discs, scars from the 10 and 20 cycle duration unlubricated testswere again formed by surface layer delamination. Micrographs of the ten cycle hardeneddisc scar in Figure 4.8 show a very narrow scar, increasing in width, and ending with alarge clump of material at the trailing edge. This clump was likely trapped at the front ofthe ball and ploughed the relatively constant width of the second half of the scar. Overall,the scar interior displays a layered texture similar to the annealed disc scars. However, thehardened scar appears smoother, with fewer fine striations in worn surfaces, suggestingless shear strain has occurred in the hardened disc scar. The close-up of a low spot in the40scar interior shows a much smoother fracture surface, characteristic of a brittle fractureprocess.After twenty cycles, the scar on a hardened disc, presented in Figure 4.9, appearsat first glance, quite similar to the corresponding annealed disc scar. Closer examinationshows no signs of plastic deformation around the scar edges, but a slightly more abundantquantity of finer particles on the hardened disc. Differences in scar roughness appearminimal on a larger scale, since the thickness of delaminating layers is similar. But on afiner scale, the hardened disc is smoother, again suggesting smaller accumulations of shearstrain and brittle fractures.The careftul examination of the wear scars from very short duration tests inunlubricated conditions have shown that the development of the wear scar occursexclusively by the mechanism of coarse delamination, regardless of disc hardness. Theeffects of disc hardness were limited to a difference in appearance or texture of wornareas, which suggests the operation of different fracture processes.4.1.2 Intermediate Duration TestsMicrographs of wear scars from slightly longer duration tests show that as theprocess of wear develops, wear scars from a variety of test conditions do show significantdifferences in some cases, but also remarkable similarities in others. Figures 4.10 and 4.11show fifty and one hundred cycle scars on hardened discs under low and high frictionconditions respectively. Delamination and grooving of the wear scar surface is evident inboth cases, and a proliferation of finer micron and submicron sized particles exists in bothwear scars. As the number of wear cycles is increased, a significant difference in theappearance of lubricated versus unlubricated scars emerges.In both lubricated conditions (water and dilute cutting fluid), as the number ofwear cycles is increased to the hundreds and then thousands, a compacted layer of finewear debris forms in the interior of the wear scar. Figure 4.12 shows the entire wear scar41and several close-ups from a five hundred cycle cutting fluid lubricated test on a hardeneddisc. Long slender wear particles are seen accumulating at the front of the wear scar, andfine delamination in the scar interior is apparent. Figure 4.13 presents the same conditionsbut after five thousand cycles. The patchy smooth layer seen in the interior of the scar cannow be considered as having a similar composition as the less densely compacted finewear debris surrounding the scar. The formation of a compacted debris layer severalmicrons thick in the scar interior is very obvious in micrographs of a tempered disc after afive thousand cycle cutting fluid lubricated test shown in Figure 4.14.In unlubricated conditions, surface examinations of the wear scars fromintermediate duration tests did not produce evidence of the formation of a compacteddebris layer. To the unaided eye, after dry sliding for a few thousand cycles, copiousamounts of loosely adhered fine black powdered debris were observed around the wearscar. To avoid contaminating the vacuum chamber of the SEM, this debris was gentlyblown from the disc surface. Micrographs from an intermediate duration test (D5 ) appearin Figure 4.15, and they show a large scar area filled with micron and submicron sizedparticles. A portion in the center of the scar resembles smooth topped islands. In highermagnification views of these features, it is not immediately clear if these islands are formedfrom compacted fine debris, or conversely, if the fine particles are created by thecontinued fracture of the edges of these islands. However, closer examination of what canbe seen of the valley surfaces between bits of fine debris shows the valley surfaces havethe same smooth appearance as the island tops. If the valleys are assumed to be formedfrom disc material, then the islands can also be considered disc material, the source of thefine particles, and not a compacted debris layer.Optical micrographs were taken of scars from low friction intermediate durationfeedback controlled tests on all three disc hardnesses to illustrate the differences in scarappearance brought on by changes in hardness. Figure 4.16 presents annealed, temperedand hardened disc scars after fifty and five hundred cycles, all at identical magnification.42As hardness is increased, a marked decrease in the width of the ellipsoidal indentation isapparent. At fifty cycles, scars of all three hardnesses display varying degrees of ellipsoidalindentation, with a grooved or scratched scar interior. With much higher magnification,dark wear debris could only be seen on the hardened scar. After five hundred cycles, darkwear debris is visible in and around the scars of all three hardness conditions. Tests werealso done at five thousand cycles, and although not included in Figure 4.16, scars of allthree hardnesses displayed a compacted layer of debris in the scar interior, and weresurrounded by loosely compacted debris clumps around the perimeter. So aside fromdifferences in the magnitude of indentation, disc hardness had almost no effect on the scarappearance for conditions of low friction and intermediate test durations.4.1.3 Long Duration TestsExamination of wear scars from the longer duration tests produced more evidenceof the formation, under lubricated conditions, of a thin compacted debris layer in the scarinterior. These observations are supported by reported observations of similar phenomenaby many other authors, including Rigney et al. [42], Berthier et al. [43] and Blau [44].However, to the naked eye, or observed under the optical microscope, the smoothenedbright appearance of the interior of a long duration lubricated wear scar would suggestthat the interior of the scar is metallic and consists of original disc metal. Surrounding theinterior of a typical lubricated disc scar, a collection of fine black loosely-attached debriswas always found. Carefi.il observation in the SEM of the interior and the scar perimeterrevealed that the material in these two regions was the actually the same oxidized weardebris.Due to inaccurate control of the displacement amplitude, surface examinations ofthe open loop long duration tests did not show the unique transition from scar interior tosurrounding debris very well. However, the implementation of feedback control produceda very distinct transition between the scar interior and the debris around the scar43perimeter. Mlcrographs from a long duration cutting fluid and water lubricated tests onpolished discs are shown in Figure 4.17 and 4.18 respectively. Both scars exhibit a verysmooth interior with many very fine grooves usually traversing the length of the scar. Ahard particle responsible for the fine scratches in the scar interior is shown in highermagnification in Figure 4.17. Close-ups of the trailing edge of both of these scars revealsthe fascinating transition from smooth compacted scar interior to more looselycompressed debris. Thus the visual appearance of a scar can be explained as follows.While loosely compacted, the fine oxidized debris would appear black. In the case of ahighly compressed smooth interior, this material reflects light and consequently appearsmetallic to the naked eye.The micrographs of these long duration tests also reveal that in some areas of thescar interior, the third body layer occasionally delaminates. The water lubricated scarshown in Figure 4.18 (1) has a multitude of cracks in the debris layer at the front of thescar. It is postulated that fine loose debris quickly collects in delaminated areas, until anew layer is formed again, to the point of delamination, and the process is repeated.Closer examination of the fine grooves found in the scar interiors reveals a secondmechanism by which the third body layer is removed. Slender particles form in manylocations when groove edges fracture. This mechanism was described by Suh [41] as first,the creation of ridges on the outer edges of grooves ploughed by asperities or hardparticles, and second, the subsequent flattening and fracture of those ridges by repeatedcompressive loading. However, it is also likely and indeed revealed in the micrographs thatfracture also occurs by the cutting tool-like action of hard particles as they arereciprocated across the scar.Increasing the surface roughness of the disc proved to have minimal effects of theformation of the third body layer. The wear scar on a sandblasted disc from the cuttingfluid lubricated K series of tests is shown in Figure 4.19. The sandblasted surface features,shown in an unworn area in Figure 4.19 (3), have been completely worn away in the44interior of the scar which displays the characteristic smooth but lightly grooved third bodylayer. Close-ups of the scar interior also show cracked areas ready to delaminate andslender particles formed from groove edges. The larger particle in the bottom half of themicrograph in Figure 4.19 (2) does not appear to have originated at the current location,but may have acted as a hard particle forming the groove in which it is currently located. Itis apparent from the numerous fractures across the length of the particle that it would havebeen reduced to much smaller bits had the wear test continued.When the surface roughness features were larger than the depth of wear scarindentation, as was the case for the surfaces pressed against 60 grit SiC, deep valleys inthe wear scar area were filled in with compacted debris. Figure 4.20 shows the wear scaron one of the 60 grit pressed discs from the cutting fluid lubricated K series. Near the scarperimeter, where the depth of the depression is small, filled-in valleys appear as slightlydarker regions, which accounts for approximately 10 to 20 % of the scar area. Theremainder of the scar resembles the typical compacted third body layer found on the scarsof polished discs. The outer perimeter of the scar displays clumps of fractured debrischaracteristic of cutting fluid lubricated specimens. An examination of the edge, showingthe very uneven disc surface and the loose adherence of debris clumps to it, suggests thata rough surface may hinder the build-up of debris, although obviously does not completelyprevent it. Yet aside from the filled-in valleys, no other major differences between scars onrough and smooth discs are apparent.Long duration tests under unlubricated conditions produced scars similar inappearance to the moderate duration scar shown in Figure 4.15, although the former haslarger quantities of loose powdered wear debris formed around the scars. In the interior ofthe scar, no evidence was observed of any substantial formation of a third body layer.454.2 Cross-SectionsA representative selection of discs from the feedback controlled tests weresectioned longitudinally through the wear scar, mounted, polished, etched, andphotographed in the SEM. Purposes for this analysis included examining the materialbelow the wear scar for plastic deformation, for surface and sub-surface cracks, and toconfirm that frictional heating had not re-crystallized the near-surface material. As well,differences in the microstructure of the three hardness conditions were to be identified, tohelp determine the influence of the microstructure on the wear process. Results from thecross-sections follow below, beginning with a few comments concerning the entireprocess.Obtaining good etched sections to fulfill the purposes listed above requiredconsiderable skill, and in the absence of skill, large amounts of time and effort. The firstessential requirement for favorable results was very good polishing. The secondrequirement was the correct exposure time to the acid etching solution. A mistake in eitheroperation usually meant starting the entire process back at the beginning.The time saving procedure of using the second half of the disc instead of nickelplating produced reasonably good edge retention if the amount of grinding and polishingwas kept to a minimum. However under repetitive grinding and repolishing, the plasticshim sandwiched between the two disc halves either abraded faster than or separated fromthe disc. In either case, edge retention was poor. Thus the time-saving benefits of thistechnique were off-set by a reduction in the quality of edge retention.The first result from the sectioning came from viewing slices of the rougher discspressed against silicon carbide abrasive papers. Examination of the sections before etchingrevealed that despite efforts to avoid it, silicon carbide crystals were imbedded in andbelow the surface. SEM micrographs of an unworn areas of disc surface pressed against60 grit SiC are shown in Figure 4.21. Silicon carbide appears as darker crystals lodgedintermittently in the surface. Another micrograph in Figure 4.25 shows a SiC particle46securely trapped in the wear scar surface. Although not included in any of the figures,similar particles were found in the near-surface of the sectioned 120 grit pressed disc.Thus the higher wear experienced by all balls tested against SiC pressed discs can beattributed to the abrasive action of the imbedded SiC particles.Etching was expected to reveal a variety of micro-constituents. In the annealedcondition, a matrix of ferrite with iron and chromium carbides was expected. For thetempered discs, a combination of plate martensite with some ferrite and carbides, wasanticipated. Finally, the hardened condition was expected to be entirely martensite. Inphotographs of etched 410 stainless steel listed in the Metals Handbook [45], ferrite andmartensite appear white and grain boundaries and carbides are dark. Under the opticalmicroscope, the etched sections followed this format. However, in the SEM, the colourscheme was reversed and the contrast reduced.A collection of the best SEM micrographs of sectioned discs are presented inFigures 4.22 through 4.26, in order of increased friction conditions. In Figure 4.22, thefirst two micrographs show the deformation below the surface of a polished disc run withdilute cutting fluid. The deformation is made apparent by the gradual bending of verticalmartensite platelets towards the horizontal sliding direction. Plate thickness decreases asthe bending increases. This section was over-etched, so the grain boundaries and carbideshave been deeply eroded by the acid in the etchant. The third micrograph in Figure 4.22,showing the material below the wear scar of a 120 grit pressed disc run with cutting fluid,is more lightly etched. No sub-surface deformation is noticeable, although there is a ‘v’shaped crack just below the surface. Crack edges have been widened by previous overetching of this section. Below the wear scar on a polished disc run in water, shown inFigure 4.23, the deformation was deeper and surface and sub-surface cracks were found.Figure 4.24 provides both an etched and unetched view of the near surface in the wearscar of a water lubricated sandblasted disc. The etched photo shows some indication ofsub-surface deformation. In the unetched photo, the plastic shim has separated from the47disc surface, and the specimen has been tilted to reveal some of the wear scar surface.Loosely-attached thin flakes cover some of the surface. The disc was ultrasonicallycleaned before both sectioning and mounting, so the thin flakes are not expected to be apart of the compacted particle layer covering the scar after testing. Thus the flakes areassumed to be nearly formed wear particles which have almost separated from the basematerial.Figure 4.25 contains several unetched features from the wear scar of a waterlubricated 60 grit pressed disc, along with an etched view. Figure 4.25 shows a cavitybelow the wear scar surface which likely held a relatively large silicon carbide crystal.Most embedded crystals were close to the size of the one shown in Figure 4.25. Althoughthe tip of the particle does not protrude much above the disc surface, the particle appearsembedded securely enough in the disc surface to support abrasive cutting of the muchharder ball by the particle tip. The third micrograph in Figure 4.25 shows a looselyattached thin particle in the center of the scar. Finally the etched micrograph in this Figureshow a surface crack and a group of sub-surface voids, along with deformation in the subsurface grain structure.All of the sections presented thus far were from tempered discs. Themicrostructure from this heat treatment included usually slender plates of martensite, smalldots of carbides that show up as white dots in the SEM micrographs, and small areas offerrite around the carbides. Figure 4.26 shows etched micrographs of annealed andhardened discs from the unlubricated tests. The etchant reveals a fine dispersion ofcarbides in the annealed disc, but no boundaries between the ferrite grains are visible. Thismakes it impossible to determine the depth of sub-surface deformation. However, anoblique surface crack is visible. The absence of martensitic platelets below the wear scar isreassuring. This confirms that frictional heating did not alter the microstructure, even forthe high friction dry sliding tests. The hardened section in (c) and (d) of Figure 4.26 alsoconfirms that a completely martensitic microstructure was achieved. Plastic deformation48below the scar surface is evident from the bending and thinning of martensitic platelets.The relative smoothness of the wear scar surface on the hardened disc is apparent fromcomparison with the annealed scar.Of the seven sectioned discs, plastic deformation below the wear scar surface wasvisible in four. Depths of the deformed material were estimated from the micrograph scalemarker and are summarized in Table 4.5. The depth ranges listed can only be considered arough estimate of the true depth, since the section shows only a single plane near the scarcenter where the depths is expected to be at their maximum. A common feature of thedeformation from all friction conditions is the occurrence of maximum deformation at thesurface, with a gradual decrease in deformation with depth into the sub-surface. Results inTable 4.5 suggest that the depth of deformed material increases when the frictioncoefficient is increased from values typical of cutting fluid to water-lubricated values.Unfortunately, without a section of a tempered disc from an unlubricated test, it is notcertain that the deformation depth continues to increase when the friction reaches typicaldry sliding levels. In two of the sections, sub-surface deformation was not visible, which isnot to say that it did not occur. In those micrographs with visible deformation, thedeformation shows up best when there are long slender martensitic plates below thesurface that originally were aligned close to vertical. When martensite plates are originallyaligned horizontally, deformation is much more difficult to identifj.The majority of the surface cracks and sub-surface voids observed in the sectionsrun parallel with the bent grain boundaries in the surrounding sub-surface. This suggeststhat cracks nucleate and propagate along inter-granular regions. The one obviousexception is the surface crack shown from the 60 grit SiC pressed section in Figure 4.25(d). The tip of this crack appears quite blunt, and grains deformed around the crack tipare visible. It is possible that the crack is actually the bottom of a sharp ‘v’ - shaped valleysimilar to the one appearing on the right side of Figure 4.21 (a). This would explain the49grain deformation at the tip. The sides of the valley would have been pushed together bythe near-surface deformation, creating the appearance of a crack.4.3 Mass Loss ResultsResults of measurements of the material removed by wear are presented below,starting with short duration tests, followed by open loop controlled test results ofintermediate and long duration, and ending with the extensive feedback controlled test,which are grouped by effects of changes in test conditions.4.3.1 Short Duration TestsMass loss from disc specimens from the short duration feedback controlled testswere estimated from surface maps made by profilometer. Given the level of a datumsurface around the wear scar, a computer estimate of the volume below the datum surfaceminus the volume above yielded the wear volume. Multiplied by the density of4lOstainless steel (7.8 Mg/m3 ), the wear volume estimates were translated into masseslosses, and the results appear in Table 4.3. Results from the cutting fluid lubricated tests,which cover the range of disc hardnesses are plotted in Figure 4.27. Surprisingly, the wearat five thousand cycles is higher for the harder disc conditions than the annealed. Althoughthe results in this figure indicate measurable wear only at five thousand cycles, visualexamination provided evidence ofwear debris at five hundred cycles. Thus, mass lossoccurring in the first 500 cycles is too small to be resolved by the volume-loss approach.4.3.2 Open Loop Controlled TestsSince the open loop tests included a mixture of both short and long duration tests,results from these tests are considered next, divided into four groups of common testconditions. Mass loss data for the first group of open loop tests are presented in Figure4.28 along with the results of similar tests reported by Ko et al. [39]. Test conditions for50the two sets of data were identical, except for the normal force; 600 N in the set for thisstudy, and 400 N for the tests by Ko. Between ten and one hundred thousand cycles, thetwo sets of results are quite close. Beyond one hundred thousand cycles, the 600 N setexhibits a slightly steeper slope. The slope of the 400 N set is close to one, which agreeswith the simple wear relationship proposed by Archard [2]:L Hwhere V is the wear volume, L the sliding distance, F the normal force, H the hardness orflow pressure of the softer material, and K1 the specific wear coefficient. In this case, thewear volume is in direct proportion to the mass loss, as is the sliding distance to thenumber of cycles. Thus the slope on a log-log plot of mass loss versus sliding cycles isexpected to be unity, which is reasonably close to the slope of the 400 N data. Accordingto the Archard equation, the data from the 600 N tests should lie above the 400 N data bythe log of 600/400 or 0.18 of a decade. Instead it appears the 600 N exhibits a slightlyexponential rather than linear relationship between wear and sliding distance.Mass loss results from the second group of open loop tests listed in Table 4.1 donot exhibit a clear trend. Inconsistency in the results can be attributed to the difficulties inmanually controlling the higher friction conditions of dry and water lubricated sliding.Variations in force and displacement levels of± 25 % occurred, usually at the start of thetest. In tests with high normal load and high friction there were stick-slip frictionproblems which would cause the wear scar to form off-set from the center of the disc.Often a circular indentation would be made on the edge of the wear scar from sticking ofthe ball during the first entire normal load cycle. However, one observation can be madefrom this test group with respect to the strong effect of lubrication conditions on wear.51Despite problems with restraint of the ball in test K 11, this test presented the firstindication that the mass loss is an order of magnitude greater for dry sliding compared tothe lubricated tests.The third group of tests for short durations with dilute cutting fluid lubricationdemonstrated that the mass loss in the early stages is so slight, having the same magnitudeas the repeatability of the balance, that determining mass losses from measurements madewith the analytical balance was not viable.The last group of open loop tests included both dilute cutting fluid lubricated andunlubricated tests of short and long durations, all performed on fully hardened discs. Masslosses for the short duration tests again proved too small to be determined by analyticalbalance. In the longer duration tests, the mass loss results appear at first somewhatinconsistent when compared to results of the first group of tests. The mass loss after fiftythousand cycles (Test K 24) was negligible. At one hundred thousand cycles (Test K25) the mass loss is appreciable, at 0.24 mg. If this result is compared with the mass lossof a similar duration test (K 4 ) from the first group of tempered discs, the mass loss onthe harder disc (K 25) is 60 % greater than that of the softer tempered disc (K 4).However, there are significant differences in test conditions between tests K 4 andK 25 which explain why the wear is higher on the harder disc. The test on the tempereddisc was performed with a much more concentrated cutting fluid mixture and at twice thenormal load. If effects of increasing hardness by a factor of 1.5 and halving of the load areestimated using the simple Archard relationship for wear, then mass loss of the harder discis expected to be one-third of that from the tempered disc. However, the effects of thedifference in cutting fluid concentration on wear have been shown by Ko et a!. [39] to bequite pronounced. A six-fold reduction in mass loss for the increase in cutting fluidconcentration from 1000:1 to 40:1 was observed in 400 N sixty thousand cycle tests ontempered discs. If this lubricity effect is applied to the comparison, then wear on the52harder disc is expected to be twice that on the tempered disc. Thus the actual 60% greaterwear of the harder disc is not unreasonable.4.3.3 Long Duration TestsMass loss results for the long duration feedback controlled tests appear in Tables4.2 and 4.4 are formatted into four distinct categories to show the effects of changes intest conditions. The first two, from the K and Wcategories, examined the effect ofchanges in disc surface roughness. The third, or D group, studied the effect of dischardness, and the final S group served as a statistical investigation of results from aconstant set of test conditions.4.3.4 Effect of Surface RoughnessIn the K group, dilute cutting fluid was used as lubricant, producing low frictionconditions with .t = 0.09 to 0.14 . Test conditions included a 300 N normal load,tempered discs, and a test duration of 50, 000 cycles. Figure 4.29 provides a comparisonof the mass loss ofboth disc and ball for the four surface roughness conditions producedfor this investigation. The figure shows similar results for the polished and sandblastedconditions, with 0.2 to 0.3 mg of wear occurring on the disc and negligible wear to theball. However, for the rougher silicon carbide pressed surfaces, disc wear was reduced bymore than a half while wear of the ball increased to levels similar to that occurring on thedisc. Note that the mass loss values for the extremes of surface conditions, the polishedand 60 grit pressed surfaces, are averaged results of three repeated tests, while for theintermediate conditions, the sandblasted and 120 grit pressed surfaces, they are single testresults. The ranges in mass losses in the repeated tests, 0.06 mg and 0.08 mg for thepolished and 60 grit pressed discs respectively, are a significant fraction of the averagevalues of 0.25 and 0.11 mg.53Since sectioning had yet to reveal the imbedded silicon carbide, it was postulatedthat the reduced wear of the silicon carbide pressed discs could be attributed to strainhardening of the surface of the disc during the pressing process. Repeating the 60 grit setof three tests with discs stress relieved after surface preparation (tests K 42, 43, & 44)produced on average slightly more wear on the discs but similar high wear of the ball.Average mass loss values for the 60 grit pressed K series tests with and without stressrelieving are presented in Figure 4.30 along with the 120 grit pressed test. Once thepresence of SiC particles in the discs was discovered, the increased wear of the ballsbecame easily understood. The remarkable result of this condition was the reduction inwear to the discs. Obviously, the presence of hard particles embedded in the surface of asofter material can dramatically reduce the wear of that softer surface, but with seriousconsequences on the wear of the mating surface.The second group of tests examining surface roughness effects, labeled using W,were a duplicate of the first K series except with distilled water as the lubricant,producing friction conditions of i.t = 0.3 to 0.4 All other test conditions were identical,including normal load, test duration, disc hardness, and surface conditions. Disc and ballmass loss results are compared in Figure 4.31, from which several observations can bemade. Compared with the cutting fluid lubricated results, except for the polishedcondition, disc mass losses with water lubrication were approximately twice the cuttingfluid result. Note the large range, 0.15 mg, in the three water lubricated polished disc masslosses is 75% of the average mass loss, 0.20 mg. . Mass losses of SiC pressed discs wereagain less than that of smoother sandblasted disc. Ball losses were also negligible againstpolished and sandblasted discs, but substantial against the silicon carbide pressed discs.4.3.5 Effect of Disc HardnessThe third category of tests examined the effect of changes in disc hardness on longduration wear. This group was conducted under dry sliding conditions and designated the54D group. Test conditions in this group did not match the K and Wgroups for severalreasons. The control program could not manage the large tangential loads induced by thedry friction at a 300 N normal load level, so the normal load was reduced to a manageablelevel of 120 N. Test duration was reduced from fifty thousand cycles in the lubricated teststo ten thousand for dry sliding, since the higher wear rate of dry sliding meant measurablewear occurred sooner. The effect of disc hardness on the worn mass loss is presented inFigure 4.32. It is immediately apparent that dry sliding produces wear rates of at least anorder of magnitude greater than lubricated sliding; however, the effect of changing dischardness is subtle. Doubling the hardness from the annealed condition to the temperedhardness produced only a slight reduction in the mass loss, and tripling to the fullyhardened level resulted in approximately a one third reduction. In this group of tests, theintermediate hardness condition was repeated three time, with variations in the results ofonly 8% of the average mass loss. The two extreme hardness conditions are single testresults.4.3.6 Statistical Survey at Constant Test ConditionsIn the fourth category, sixteen tests were repeated for a single set of conditions,producing a statistically significant amount of data for that single set of test conditions.Since the test conditions matched one of the Wseries tests, an additional point from the Wseries was included, bringing the total number of repeated test to seventeen. Mass lossresults varied from 0.32 to 0.63 mg, which represents a range of 0.31 mg. Fitting aGaussian or normal distribution to the data and calculating maximum likelihood modelparameters produced a mean mass loss of 0.49 mg with a standard deviation of 0.09 mg.Figure 4.33 presents a linearized probability plot of the data and the model, showing goodagreement between the data and model. Confidence intervals for population mean andstandard deviation were determined from the t and x2 distributions for a sample of 17. The5595% confidence intervals are { 0.44, 0.54 mg } for the mean and (0.067, 0.137 mg }for the standard deviation.Despite the immense improvements in the repeatability of the feedback controlsystem, the range in these results indicates that wear occurring under lubricated conditionsis inherently variable. The combination of variations in wear particle formation,entrapment, and ejection must sum to produce the significant variation observed in the testgroup. Since this statistical survey was only conducted for one value of test duration, itwould be of great interest to conduct additional surveys to determine the effect of testduration on the variability of the process. If after much longer test durations, the resultingwear proved less variable, then the initial wear occurring during development of the wearscar would beresponsible for contributing to the largest portion of the variability.Conversely, if the variation in wear expands with test duration, then the inherent variabilityof lubricated third body wear would be confirmed.4.4 Friction Coefficient DataThe data acquisition system implemented with the feedback control system enabledrecording the friction level during testing on a per cycle basis, which allows slight or short-lived friction transitions to be captured. Friction coefficient data is presented and discussedfor short and long duration tests in the sections below.4.4.1 Friction Transitions in Short Duration TestsDuring the very short duration tests for both water lubricated and unlubricatedconditions, complete force and displacement signals were collected, and they are shown inFigures 4.34 and 4.35 respectively. It should be noted that the data conversion programdoes not produce a record of the first cycle, and all of the traces in these two figures beginwith the second test cycle. In both annealed tests, the friction force stayed close to 50 N,and with the normal force at 275 N, the friction coefficient was 0.18. Thus the fine56delamination and minute scratching observed in the twenty cycle annealed test had noeffect on the friction. In the hardened disc, the change in friction level is undoubtedlyrelated to the coarse delamination of the disc surface. The friction coefficient in the tencycle hardened test reached t— 0.18 on the fifth stoke, but the tangential force continuedto increase, producing a friction coefficient of 0.55 for the ninth and tenth cycles. In thetwenty cycle hardened test, in which surface delamination did not occur, the tangentialforce never exceeded 50 N, and the friction coefficient remained near 0.18.In the absence of lubricant, the level of friction associated with the coarse surfacedelamination is even more extreme. As the plots in Figure 4.35 indicate, at the start oftesting, the friction coefficient is low. For the annealed discs, the friction coefficient staysat or below 0.3 until the eighth cycle of the ten cycle test, and until the twelfth cycle of thetwenty cycle test. Then in just three cycles, the friction coefficient increased to one. On thehardened discs, the transition occurred one or two cycles sooner, but after three cycleslevels off near 0.7.For the short duration cutting fluid lubricated tests, friction coefficient data fromthe 50, 500, and 5,000 cycle test were plotted together for the three disc hardnessconditions and shown in Figure 4.36. The majority of the tests ran with a frictioncoefficient near 0.1 and without any friction transitions at the start of testing. However,the intermediate and hardened 5,000 cycle tests did produce uncharacteristic fluctuationsnear p. = 0.3. These fluctuations approach the typical steady-state value for waterlubricated sliding. Thus they probably result not from an effect of disc hardness, but ratherfrom insufficient cutting fluid boundary lubricant in the contact area.It is apparent from the friction traces for the various short duration tests describedabove, that lubrication conditions have the primary influence on the resulting friction level.Disc hardness effects are relatively minor, except in the case where disc hardness impedesthe indentation process of wear scar formation, and scar formation occurs by coarsedelamination. In that case, a strong friction transition is observed.574.4.2 Friction Conditions in Long Duration TestsPlotting the coefficient of friction versus test duration for the variety of conditionscovered in the long duration feedback controlled tests demonstrates that for almost theentire test duration, it is the lubrication condition alone which determines the level offriction. Changes in surface finish and hardness produce only subtle transitions in thefriction coefficient occurring at the beginning of the test and lasting no longer than a fewthousand wear cycles. Lubrication conditions also have a pronounced effect on the frictiontransitions occurring with a particular combination of hardness and surface finish. Typicalplots from the first three series of feedback controlled long duration tests are examinedbelow.In the dilute cutting fluid lubricated K series, a steady-state friction level of p. =0.085 to 0.13 occurred for all four surface roughness conditions, which is displayed in thefriction plot of the whole test, as shown in Figure 4.3 7(a). Fluctuations in the steady-statelevel were very small, although there were moderate transitions at the start of the test. Thetraces of the first five thousand and fifteen hundred cycles in Figure 4.37 (b) & (c) showthe polished surface exhibits no transition at all, but immediately operates at a steady-statelevel. The two roughest surfaces, 120 and 60 grit pressed, show moderate friction peaksof p. 0.2 at the start of the test, decaying to a steady level of 0.11 after approximately400 cycles. The sandblasted surface displayed the most interesting transition, startinginitially at p. 0.2, rising to p. 0.3 in the first 200 cycles, then gradually decreasing to0.085 after about one thousand cycles. The consistency in the occurrence of thesedistinctly different behaviors proved reproducible in the repeated tests of the polished and60 grit pressed disc surfaces. Friction levels from three repeated 60 grit pressed tests,plotted in Figure 4.38 , matched closely. The sandblasted condition was also repeated intest K 41, and the friction traces, abbreviated in Figure 4.39 to the first fifteen hundredcycles, turned out quite similar. The plot shows nearly identical peaks, but a slightly fasterdecay in the repeated test. Since surface finish is the only variable changed among these58various tests, and transitions are repeatable for a given surface condition, the frictiontransitions can only be attributed to the surface finish.When tests with the same surface roughness were repeated with distilled waterlubrication, in the Wseries, a higher steady-state friction level between j.i = 0.29 and0.35 was recorded for all surface roughness conditions. Entire test duration traces, shownin Figure 4.40 (a), reveal the three rough surfaces running near i’ = 0.3 and the polishedsurface a little higher at 0.35. Focusing on the beginning of the tests shows that with waterlubrication the effect of surface roughness is completely reversed from the dilute cuttingfluid case. With water, only the polished surface produced an obvious friction transient,while the three rougher surfaces exhibit an abrupt increase to their steady-state levels.Upon close examination of the friction trace from the polished disc, two transitions areevident. The first peak to j.t 0.4 occurred in the first twenty cycles, followed by a sharpreduction to just below 0.3, and then a much longer second peak, again to almost 0.4,which gradually decayed after a thousand cycles. The repeated 60 grit tests producedtraces almost identical to the one included in Figure 4.40, but the repeated polished testshowed some variation. The friction traces for the first four hundred cycles of the polishedtests appear in Figure 4.41, exhibiting both similarities and differences among the tests. Allthree tests display a peak in the first twenty or thirty cycles, although there are significantdifferences in magnitude of the peak between tests. Only the test chosen for Figure 4.40,test W2, displays a significant second extended peak. The friction traces of the other twotests fluctuate very little after the first one hundred cycles. Thus it is difficult to concludewith confidence that the second peak on the polished trace is a true characteristic of thatsurface finish.Friction traces from the dry sliding series are the easiest to interpret. Figure 4.42gives the entire test duration and the beginning of the test, for the three hardness levels.Large fluctuations in the friction coefficient were observed for all three hardnesses, withvalues ranging between i = 0.6 and 0.8. The plot of the first hundred cycles reveals that59for all three hardnesses, friction is low, around 0.2 for the first few cycles, but then quicklyrises to the quasi-steady level. The duration of this initial low-friction condition shows nocorrelation to the disc hardness, as it was the disc of intermediate hardness whichremained there for the longest.4.5 Frictional WorkThe frictional work done on specimen discs for the feedback controlled tests areincluded in Tables 4.2, 4.3, and 4.4. As is expected, the changes in this quantity reflectchanges in friction level, applied normal force, and test duration. Since wear researchershave often sought a correlation between the frictional work and wear, it is of interest tolook at the relationship between these two quantities. Since significant mass losses wereobtained only in long duration testing, the short duration frictional work results will not bediscussed.In the dilute cutting fluid K test group, the work input ranged from about 2 to 3kJ, while in the higher-friction water lubricated Wgroup, the work input ranged only from7 to 7.5 kJ. In the unlubricated D group, the work input varied from 1.3 to 1.4 kJ. Notethat in the D group, test durations were one fifth the duration, and normal forces were40% of the level of the lubricated tests.Comparing disc wear to the input energy shows a weak connection between thetwo quantities. Mass loss-work input ratios were calculated for polished and sandblasteddiscs run in cutting fluid and water, and polished unlubricated discs. The range in theseratios are plotted in Figure 4.43 (a), which shows that the mass loss-work input ratios forthe lubricated tests are close to the same magnitude, although somewhat lower for thewater lubricated conditions. However, ratios for the unlubricated tests are an order ofmagnitude higher.Test results in Table 4-2 are arranged in groups of increasing surface roughness forthe K and W series, and increasing hardness for the D series, which facilitates reviewing60the work input column for roughness and hardness effects on the work input. In the Kseries, the SiC pressed disc tests show larger work inputs than the polished. The twosandblasted K series tests do not fit into the pattern, and instead show a range of almost 1kJ in work input, which correlates well with the disc mass losses for these two tests. In theW series, differences in work input with roughness are nominal and do not follow anypattern. In the D series, increasing disc hardness shows very slight decrease in work input;however again the differences are nominal, and less than the variation among repeatedtests.The S or statistical test group provides a large number of tests at one set ofconditions, and frictional work input results for this group, listed in Table 4.4, ranged fromabout 6.6 to 7.4 kJ, or about 11% of the average value. The range in mass losses for thistest group, at 65%, was a much higher fraction of the average value. Mass losses areplotted with corresponding work input in Figure 4.43 (b). There is a good cluster of pointslying in the center of the plot; however, the out-lying points do not follow a trend ofincreased wear with larger work input. Thus the variation in mass loss-work input data inboth Figures 4.43 (a) and (b) indicates that the correlation between work and wear is notexact in nature, but instead somewhat variable.4.6 Wear Scar Geometries and CurvaturesThe dimensions of many of the wear scars on both discs and balls are listed inTable 4.2 and 4.3, along with disc scar radii of curvature in the longitudinal and transversedirections. These measurements combined together provide an indication of the wear scargeometry and the effects that changes in test conditions had upon the scar geometry. Froma first review of the disc scar lengths and widths, several observations can be made. First,all scar lengths exceed the 2.00 mm displacement stroke of the dynamic specimen, whichindicates that the contact area at the beginning and more likely at the end of the wearstroke is large enough to significantly increase the scar length beyond the stoke length.61The second observation is derived from measuring the wear mark on the ball aftera test. The ball scar dimensions should represent the contact area when the maximum loadis applied, which will be used later for contact stress calculations. Ball scar lengths,defined as the dimension of the scar measured in the sliding direction, were only slightlysmaller than the ball scar widths. The relationship between ball scar dimensions can bequantified by considering the aspect ratio of the scar, defined as the longer contact lengthdivided by the shorter, or width over length for this group of results, and it is useful indifferentiating between the classical concepts of point and line contact. Ball scar aspectratios varied only from about 1.7 to 1.2, except in the case of some of the very short testswhere the scars formed were by coarse delamination and contact ratios as high as 2.7 wereobserved. The lubricated and polished tests generally produced the higher ball scar aspectratios near 1.7, and the lubricated rougher surfaces showing ratios of 1.4 to 1.2. In longduration dry sliding test, aspect ratios were all near the lower end of the range. With thistrend of low aspect ratio contact areas, contact conditions were much closer to theclassical circular contact, than to line contact.Another point of interest comes from comparing ball and disc scars for a giventest. The width of the scar on the ball in most instances corresponds quite closely to thewidth of the disc scar; however, several ball scars are significantly wider than theircounterpart on the disc, especially in the long duration water lubricated Wtest group. Thecompacted debris around the edges of the disc scar, which is removed during cleaning andnot included in the measurement of the scar width, must be compacted firmly enough tomark the ball. However, none of the dry sliding tests display an over-sized ball scar,supporting the physical observation of much more loosely packed wear debris around theperimeter of dry sliding scars.Focusing on disc scar dimensions alone, it is apparent that the length and width ofthe scar increases with increased surface roughness, for both the cutting fluid and waterlubricated conditions. The valleys or low points of the rougher surfaces reduce the area62available for supporting the contact loads, and thus a larger apparent contact area isrequired. The dry sliding tests produced the longest and widest scars, which is to beexpected from the higher mass and consequently volume losses. In the dry sliding tests,increasing the hardness of the disc produced only a small reduction in scar area, with a 6%reduction for doubling the hardness from annealed to the tempered, and a 17% reductionfor tripling the hardness to fully hardened. However, the aspect ratio of the dry sliding discscars was lower than those from lubricated tests. Typically, D series disc scars had anaspect ratio of 1.5, while the lubricated K and Wseries scars showed ratios of 2.0 to 2.5.This trend is expected, and can be explained as follows. Dry sliding tests produced muchlarger mass losses than the lubricated, so the volume loss of a unlubricated scar will becorrespondingly larger. The longitudinal length of the scar, made up of the stroke lengthplus half the contact widths at the beginning and end of the stoke, is held almost constant.Then for an unlubricated scar to have a larger volume loss, it must penetrate deeper intothe disc surface. Deeper penetration of the indenting ball will necessitate a wider scar.Thus for a constant wear stroke length, the aspect ratio of the disc scar should becomesmaller as the indenter wears deeper into the disc.The remarks concerning disc scar sizes can be reinforced by examining surfacemaps of the various scars. Surface maps, obtained by incremental traces of the surfaceprofilometer are shown in Figures 4.44 and 4.45 for the water lubricated Wandunlubricated D series long duration tests respectively. Unfortunately, in Figure 4.44, thevertical scales and size of the areas traced are not kept constant, so that the increase inscar area with increased surface roughness is not immediately recognizable withoutconsidering the change in map size. The scales in Figure 4.45 are identical, so that thelarger dimensions of the unlubricated disc scars are obvious. Figure 4.45 also illustratesthe decrease in the roughness of the scar interior as disc hardness is increased.Disc wear scar curvatures in the sliding and transverse direction are also listed inTable 4.2 and 4.3, and this data augments the knowledge of how the wear scar geometry63is affected by changes in test conditions. Disc scar curvature in the transverse directionwas always very close to or matched the radius of the ball for all of the various testconditions. However, comparing the longitudinal curvatures for lubricated polished discs,tests K33 and Wi, shows the increase in friction from dilute cutting fluid to waterlubricated conditions had an interesting effect on the longitudinal curvature of the scar.The best-fit radii of curvature were very close; however, as shown in Figure 4.46, thecutting fluid lubricated test produced a more asymmetric scar. Results of longitudinalcurvature measurements on the water lubricated discs indicate that increasing the surfaceroughness to the silicon carbide pressed conditions not only increased the wear scar area,but also reduced the scar curvature, or increased the radius of curvature. Dry slidingproduced scars of largest curvature, although a direct comparison between the lubricatedand dry sliding curvatures is not valid. A more applicable comparison could only be madewith a shorter duration dry sliding test with a worn mass loss similar to the lubricatedtests, since the deeply worn scars of the D series are expected to exhibit greater curvature.Chapter 5Extended Analysis of Results5.1 Wear Process in the Single Asperity Contact Model TestsFrom an overview of the entire test program, of short and long duration and forthe variety of lubrication, hardness and surface roughness conditions, a pattern ofmechanisms that make up the wear process emerges. The process starts in the first fewload cycles with the changing of the contact geometry to a more confonning shape by oneof two possible mechanisms: indentation or coarse delamination. Friction or lubricationconditions have a dominant effect on which of the two geometry transformationmechanisms will operate. Wear particle formation begins either during geometrytransformation or somewhat later, depending on the geometry transformation mechanism.After several thousand cycles, if conditions favor particle entrapment or hinder debrisejection, as in lubricated tests, a third body layer ofwear debris forms in the contact area,slowing thither wear. A detailed discussion of this complex wear process follows below.5.1.1 Initial Contact - Geometry Transformation by IndentationIf the friction is low, in the region 0.3, the geometry transformation occurs byan indentation process similar to a Brinell hardness test, in which a hemisphericalimpression is formed by a spherical indenter, but extended into an ellipsoid by thesinusoidal loading and motion of the indenter. Disc scars from cutting fluid and mostwater lubricated short duration tests exemplilj this process. Measurements of scargeometry presented in section 4.6 illustrate two important characteristics of this process.First, very short duration water lubricated tests (L 25 & 26) show that the majority ofthe deformation occurs in the first few cycles. Second, as scar geometry measurements6465from the intermediate duration cutting fluid lubricated tests illustrate, the scar geometrycontinues to expand over hundreds and even thousands of cycles, for all three hardnessconditions tested. A logarithmic plot of disc scar widths versus cycles for these tests,appearing in Figure 5.1(a), shows that the rate of increase in width is not stronglyaffected by the disc hardness. These two characteristics of large initial deformationfollowed by a more gradual expansion become important when considering an approachfor modeling this mechanism.A simple yet elegant model for the geometry of the ellipsoidal indentationdeveloped by Magel [1] predicts the longitudinal radius of curvature of the indentation.Assumptions made in the derivation of this model include line contact between ball andindentation, contact pressure equal to the softer material’s shakedown pressure, and atransverse indentation curvature equal to the ball radius. Thus the model predicts thegeometry for a steady-state condition of elastic shakedown. Once that has been reached,deformation is assumed complete. To assess the applicability of Magel’s model to the wearscars from this study requires an estimate of the contact conditions, including contactdimensions, pressures and shakedown factors.At the onset of testing, the contact will be circular. However, a plot of theshakedown load limits as a function of the friction coefficient for a ball-on-flatconfiguration for the three hardnesses, provided in Figure 5.2, shows that as the loadreaches a 300 N maximum value, the shakedown limit will be exceeded for all hardnesses,and plastic deformation, or indentation, is expected. Once transformation of the discgeometry has occurred, determination of the contact conditions is no longer so simple.Several techniques were applied, and examples of each can be found in Appendix A. Allcalculations were made at the maximum load, which is assumed to correspond tomaximum scar width. The contact conditions for this situation are considered to berepresentative of conditions over the majority of the scar area. The relative merits of theprocedures and the image of contact conditions they produce are briefly reviewed below.66The simplest technique for determining the contact conditions is based on ellipticalcontact using the dimensions of the ball scar. Unfortunately, in some of the early shortduration tests, balls were discarded and scar dimensions were not recorded, or in some ofthe shortest tests, the ball scar was not visible. For short duration tests, with minimal weardebris, the accuracy of this procedure is likely quite high. In longer duration tests, weardebris accumulating around the edges of the scar might mark the ball surface and lead toan over-estimate of the contact area. Ball scars were also not perfect ellipses, butsomewhat irregular in shape, so there is a small error in assuming a perfectly ellipticalcontact. Still, in spite of the uncertainty, this technique is likely to produce the bestestimate of the contact conditions for short duration tests.If ball scar dimensions were not available, but the scar curvatures were known,,then the contact conditions were determined from disc and ball curvatures using circularcontact equations, again for the maximum load condition. The technique producedaccurate results when disc curvatures were large, such as tests L 13 and L 5, in which thecalculated contact width matched the disc scar width quite closely. However, when thetransverse curvature of the disc scar approached the ball radius, this technique wouldpredict an unrealistic contact width much larger than the disc scar width. Since disccurvature in the transverse direction does not extend beyond the width of the disc scar, athird technique was needed.Disc scars with transverse curvatures similar in magnitude to the ball radius (6.4mm) were considered to conform closely enough to the ball that line contact, extendingthe width of the disc scar, could be assumed. Thus this technique uses a rectangularapproximation of the contact area. This is likely the case when the disc scar curvatureclosely matches the ball, but when the curvatures are not very close, results from thistechnique will contain an element of error.Calculations of contact conditions are summarized in Table 5.1, and they revealseveral interesting trends. The shortest water lubricated tests exhibit shakedown factors67greater than 5 for the annealed discs, indicating ratchetting by subsurface flow, but only3.6 for the hardened disc (test L 34), indicating a condition of elastic shakedown.Although the curvature of the hardened disc is much less conforming than the annealed (L26), the hardened disc shakes down under much higher contact pressure due to its largershear strength. Knowing the shakedown limit is about 4 for line contact and 4.7 for pointcontact over i = 0 to 0.25, shakedown factors for the longer cutting fluid lubricated testsshow that by five thousand cycles the contacts have dropped far below shakedown for allhardnesses except the annealed condition. In fact, the tempered and hardened disccontacts are actually below the elastic loading limit. Shakedown factors for the tempereddiscs (L 10, 11, & 12) start just below the shakedown limit for line contact at fifty cycles,and decrease as the test duration is extended. Clearly geometry transformation does notstop when the shakedown limit is reached. At first, wear of the disc might be suspected asthe cause of the continued change in scar geometry. However, the surface examination ofthese intermediate duration tests showed signs of only very minimal wear, certainly not ofsufficient magnitude to cause the observed changes in scar geometry. Another processmust be responsible for the gradual geometry transformation, and this issue will beconsidered in the next section.5.1.2 Indentation ModelingDuring the initial indentation process, when the friction force is small, thelongitudinal flow of material is likely small enough to be neglected. Then the formation ofthe ellipsoid can be considered as a drawn-out collection of Brinell-type sphericalindentations. Scar geometry can then be modeled with indentation relationships [46] andthe half width of the scar, a, for load, P is given by:{5.1}68where H is the material hardness. Using this relationship, the width of an indentation canbe found as a function of the applied load. If scar widths for the three hardness level aredetermined for a 300 N load and are added to Figure 5.1(a) as the first-cycle scar widths,the plot will appear as in Figure 5.1(b). This plot shows that the spherical indentationformula provides a probable estimate of initial or first cycle scar width. The depth of theinitial indentation, d, could also be determined from an indentation relationship [46]d 52irHSDwith D as the radius of the indenter, and the pile-up ratio, S = h I d, where h is the fullcrater edge-to-bottom height, and d is the depth from crater bottom to the original surfacelevel. Scar depths at the maximum load condition are estimated with this relationship andtabulated in Table 5.2. Compared to measured scar depths from Table 4.4, this functionproduces scar depths of the same order of magnitude as the actual depths. The estimatefor the annealed is very close; however, the very shallow scars on the hardened discs areover-estimated.While the functions above provide the initial indentation geometry, considerationmust also be made for the observed expansion of the scar during subsequent loadingcycles. The linear trend on the logarithmic plot of Figure 5.3 suggests using an exponentialexpression for scar width, W:W=J’ N”1 {5.3}where W1 is the initial width and m is taken from the slope of the log-log plot. Values ofthe initial width are listed in Table 5-1, and exponent, m from Figure 5.3 are 0.06, 0.11,and 0.12 for the annealed, tempered and hardened discs respectively. This empiricalrelationship describes scar width reasonably well over many thousands of cycles, asindicated by the addition of a long duration cutting fluid lubricated scar width to the69tempered data in Figure 5.1(b). However, as the test duration extends into tens andhundreds of thousands of cycles, disc wear will become the primary cause of flirtherchanges in scar shape. Wear debris is visible after five thousand cycles and mass losses aresignificant after about ten thousand cycles. Thus a limit of seven thousand cycles for thisexpression for scar width expansion by deformation is suggested.An improved model of scar expansion will require a better understanding of theprocess causing it. The material in the contact zone is both heated during the dissipation offrictional work to the lower end of the temperature range associated with creep, andhighly stressed by contact forces. The gradual expansion of scar geometry may then be theresult of low-temperature high-stress creep. Although time durations much longer thanthe time elapsed in the first several thousand cycles are usually associated with the processof creep, and the increase in temperature is likely no more than a few hundred degreesCelsius, the stress levels are high. Unfortunately, little published data is available on creepfor these unique conditions. Even if temperatures and creep rates were known,determining the strains due to creep would be a very complex process, given the cyclicfluctuations of temperature and stress and also the three-dimensional system of strains.5.1.3 Initial Contact - Geometry Transformation by Coarse DelaminationThe alternate process of geometry transformation, associated with frictionconditions oft 0.3 in the first ten cycles, is radically different from the continuous,smooth indentation process, although after many thousands of cycles the resulting wearscar appears similar. Surface examinations from two of the water lubricated and allunlubricated short duration tests showed that the initial scar geometry was predominantlyformed by delaminating or shearing of surface layers in the contact area. This process hasbeen labeled as “coarse” due to the relatively large dimensions of the delaminating sheets,as compared with a finer delamination which occurs later in the wear process. Flakes aslarge as 200 x 100 p.m were observed (test L 33, Figure 4.3 ), although particles an order70of magnitude smaller were more common. Thicknesses of the flakes varied between oneand ten microns. A wear scar formed by this process start out as a narrow gouge, butquickly expands in width. The scar eventually resembles an ellipsoidal indentation,although it has been formed mostly by material removal rather than plastic flow.For the first few cycles of all tests, friction was low, regardless of lubricationconditions. This initial low friction can be attributed to accommodation of the slidingvelocity by shearing of surface films [35], including adsorbed water vapor and oxidelayers. While the friction is low, some transformation of the contact geometry by theindentation process undoubtedly occurs. But once the surface films are exhausted, withoutadditional lubricants to replenish them, velocity accommodation occurs by sheardeformation of the first body surfaces, primarily in the disc. An increase in the frictionforce accompanies velocity accommodation in the first body, since the shear strength ofthe disc material is greater than that of the surface films. It is apparent from the SEMmicrographs of the short duration disc scars which delaminated that the near-surfacedeformation is so severe that fracture of the deformed material also occurs.The fracture which leads to delamination will be a combination of crack initiationand propagation mechanisms. Cracks can form either at or below the surface. Crackinitiation at the surface could rsult from the shallow zone of high tensile stress at the rearedge of a high-friction contact. A common feature of the short duration disc scars is astriated or stepped appearance, generally oriented in the transverse direction. Thesefeatures may originated as surface cracks. Subsequent compressive loading likely inhibitstheir propagation, while shear deformation expands or widens the mouth of the crack.The transverse cracks in the scar ofL 20 (Figure 4.4 ) must have been formed this way. Itappears that the crack tip turns parallel to the surface and may propagate in the slidingdirection.Crack initiation and propagation are also expected to occur at and along grain ordeformation cell boundaries, which are thinned and become aligned parallel to the surface71during near-surface shear deformation. The exact mechanism of crack initiation at grainboundaries is not clear; however, crack propagation along a grain boundary can beconsidered as following the path of least resistance. Delaminated debris with thicknessesmatching the typical thickness of flattened grains near the surface lends supports to thesupposition of inter-granular fracture.The rate of the fracture process is also of great interest. It is clearly abrupt andexpedient, such that extensive delamination can take place in only a few loading cycles.The swiftness of the process is best exemplified by the micrographs of test L 20 in Figure4.4 and 4.5. The friction trace for this test (Figure 4.37 (a)) shows that high frictionconditions occurred for only the last two cycles of this test. This was sufficient to split thesurface into many wide fissures and delaminate some small sections. One more highfriction load cycle likely would have delaminated much of the central area of the scar.It is apparent that the entrapment of the delaminated debris also has an importantrole in the process. In the short duration test which exhibited coarse delamination, much ofthe debris often appears to have been trapped in the contact and dragged repeatedly acrossthe scar by the indenter. Some of the delaminated particles still had a smooth polishedsurface, confirming that the velocity of the indenter had not been accommodated by slidingat the surface, but rather by subsurface shear of the disc material. The accumulation ofdebris is probably responsible for the peculiarities in the shape of both disc and ball scars.The relatively constant width of coarse-delaminated disc scars can be attributed toploughing of debris trapped at sides of the indenter. As well, the extended length ofcoarse-delaminated ball scars, often more than twice the ball scar width, likely result fromdebris entrapment at the front of the contact area. Mechanical interlocking and ploughingof the trapped debris against the striated scar surface undoubtedly adds a significant butvariable portion to the friction force. Thus entrapment of delaminated material has severeeffects on coarse delamination.72The expansion of the scar geometry may also be influenced by the retention ofdelaminated debris. If debris escapes readily, then contact pressure in the scar area willspread to the unworn sides of the scar. After breakdown of surface films, the edges willundergo coarse delamination and the width of the scar will increase. Debris entrapmentwill slow this expansion, as it helps maintain the original contact geometry for longerdurations. Eventually, an ellipsoidal-shaped wear scar will be created, similar to thatformed by indentation.So far the discussion of coarse delamination has focused on its occurrence inunlubricated conditions, although it also occurred in two water lubricated ten cycle testson hardened discs. However it did not occur in the water lubricated, twenty cycle,hardened disc test nor in similar tests on annealed discs. Several possible explanations canbe construed for this behavior. It could be attributed to differences in the longevity of thesurface films on various discs, but this is unlikely since surface preparation and treatmentwere similar for these discs. More likely, the different behaviors result from a combinationof factors. In the first few loading cycles, the observed low friction conditionsare againpresumed to result from shearing of surface films. During the first few low-friction cycles,contact geometry will be transformed towards increased conformity by the indentationmechanism, with a resulting reduction in contact pressure. The reduced contact pressuremight then be partially supported and sliding velocity accommodated by the viscous actionof a very thin layer of water. The short duration water-lubricated tests on annealed discs,which indent significantly in the first few low-friction cycles, avoid coarse delamination inthis fashion. The hardened discs do not achieve geometrical conformity as quickly. Hencea small variation in the longevity of the surface films, which may last from four to sixcycles, might decide the difference between a geometry that partially supports the contacton film of water and one that cannot and consequently suffers coarse delamination. Hencethe different behaviors of the ten and twenty cycle hardened-disc test are attributed to73slight differences in the durability of surface films and in the extent of the initialindentation.5.1.4 Coarse Delamination ModelingA complete model of coarse delamination, including both the fracture process andthe extent of its action would be beyond the extent of this study. However, a roughestimate of its extent is possible.Since sliding starts with a few low-friction cycles before coarse delaminationbegins, it is plausible that the width of the initial delaminated patch could be estimatedfrom the initial indentation width using function (5.1). Even including the effects of scarexpansion with expression {5.3} for the number of low friction cycles, this approachproduces substantially smaller widths than actual observed values, especially for thehardened disc scars.The thickness of the delaminated layer might initially be expected to be equal tothe depth of the maximum value of the von Mises yield parameter, since for moderatefriction levels, the subsurface peak is at a dimensionless depth of z/a = 0.41 and 0.06 for j.t= 0.3 and 0.4 respectively. But, for a typical contact width, a, experienced in testing,delamination thickness would be over-estimated by an order of magnitude using thisapproach. As well, for higher friction conditions of ji. 0.5, the peak stress occurs at thesurface. So rather than following the location of a peak stress, the delamination fracture isexpected to follow a fault or weakness in the material structure, that being a grainboundary. Thus coarse delamination thickness is probably determined by the grain size,and for the stainless steel tested, delamination thickness could be estimated with anaverage value of 5 rim.Based on the examination of test L 20, the complete cycle of delamination isexpected to occur in three load cycles. The relatively long lengths of the coarsedelaminated disc scars is enigmatic and difficult to predict accurately. A review of the74displacement traces for the delaminating short tests (Figure 4.36 & 4.37) confirms theaccuracy of the controller in maintaining a 2 mm motion. The lengthy disc scars mustresult from a contact area extended by debris trapped in and around the indenter. A simpleapproximation of the extended disc scar length can be made by adding the initialindentation width to the indenter’s displacement, but again this provides a veryconservative estimate, especially for the hardened discs.When all of the approximations for scar dimension are combined as a predictivemodel for the extent of coarse delamination, it provides only a very rough approximationof the area initially affected by the first few high-friction load cycles. During subsequentcycles, scar expansion occurs rapidly, primarily in width. The twenty cycle test, L 23,indicates that while the scar length remains about the same, the width doubles after anadditional ten high-friction load cycles. Without additional testing at many more shortincrements, an estimate of the expansion rate can not be made.5.2 Wear Particle FormationIf the initial phase of geometry transformation occurred by indentation, then wearparticles start to form after anywhere from 10 to 50 or even as many as 100 cycles,.Particle formation mechanisms include fine delamination of surface layers and fracture ofexposed roughness features within the wear scar. Edges of the grooves formed by hardparticles are a common roughness feature prone to fracture, along with the edges ofdelaminated areas. Once formed, wear particles that do not immediately escape thecontact zone are further fractured or fragmented into very fine submicron-sized particles.These tiny particles can accumulate in the contact zone, forming a continuous third bodylayer which slows particle formation.Alternately, if coarse delamination shapes the initial scar geometry, then wearparticle formation has already started. As the scar area expands, the scale of thedelamination becomes finer, and the fracture of roughness features also contributes to75particle formation. Each of these particle formation mechanisms is examined in detailbelow.5.2.1 Fine DelaminationThis mechanism of wear particle formation is essentially a scaled-down version ofthe coarse delamination mechanism, described earlier in section 5.1.3. Delaminating flakesare generally an order of magnitude smaller, with the largest about 40 x 20 jim. Flakethicknesses average 1 jim and seldom exceed 2 jim. Fine delamination was observed inthe short duration cutting fluid and water lubricated tests after indentation had shaped thewear scar. Once several hundred cycles of coarse delamination had shaped the wear scar inthe unlubricated tests, the scale of the delamination becomes finer. Thus fine delaminationoccurs over the entire range of friction conditions tested, with visual evidence of thismechanism found for the range of test conditions as shown in Figures 4.2, 4.10, 4.11,4.12, and 4.15.In considering the rate at which fine delamination occurs, without in-situmicroscopic observation of a wear test, it is impossible to determine if the fracture portionof this mechanism is more gradual or as equally abrupt as its larger-scale counterpart.However, it is logical to assume that if the deformation and fracture processes formingmuch larger particles require only a few cycles, then a smaller-scale version could occurwith equal rapidity. The small areas of the indentation displaying fine delamination in ashort duration water lubricated test (shown in Figure 4.2) confirm that fine delaminationcan occur in just a few cycles, but only in a reduced portion of the contact area.Also of concern and interest are the reasons for the reduced scale of thedelamination process. Initial geometry transformation quickly increases conformity andreduces the smooth-body contact pressure below shakedown and elastic limits. However,the true roughness of the first bodies will produce shallow localized high-stress zones veryclose to the surface [17], [18], [19]. These stress peaks are assumed to be of sufficient76magnitude to cause plastic deformation at the surface and a shallow gradient ofdeformation in the subsurface. Original grain boundaries are drawn out parallel to thesurface by the near-surface shear strain, and present fault lines along which fracture ofdelaminating particles is assumed to follow. Etched sections of wear scars show originalgrain boundaries near the surface to be very thin, in the order of 1 JIm, although in most ofthe sections resolution of the very near surface microstructure is poor. More exactknowledge of the microstructure in the deformed surface layer is required before theassumption of fracture along grain boundaries can be firmly established.For scars originally shaped by indentation, it is apparent from a comparison oftwenty cycle tests (Figures 4.2 vs. 4.3 (c) and (d) ) that fine delamination begins soonerand occurs more extensively in the softer annealed condition than in the hardened. Thisdifference is not considered a result of surface films providing better protection on thehardened disc, since a similar trend of more extensive debris and delamination was alsoobserved for annealed disc over the hardened in the cutting fluid-lubricated short durationtests (tests L 7 & 13 ). Wear resistance to fine delamination is a combination of resistanceagainst the deformation leading to fracture, and resistance to fracture. While the ductilityof the annealed discs ought to improve fracture resistance by blunting of crack tips, it alsowill allow larger and faster accumulation of the strains leading to fracture. Thus hardness,or resistance to plastic deformation, appears more significant in deterring finedelamination than toughness, or resistance to fracture.5.2.2 Fine Delamination ModelingThis wear mechanism operates after scar geometry has been shaped by eitherindentation or coarse delamination, generally in the period between 50 and five thousandload cycles. Beyond five thousand cycles in lubricated conditions, the formation of a thirdbody layer in the wear scar likely inhibits fine delamination, an aspect considered in a latersection. The calculations of shakedown factors for short duration tests in this 50 to 5,00077cycle period indicate that contact pressures drop below the elastic shakedown limit. Thistrend is continued in the long duration tests, as indicated by similar calculations for thelong duration feedback controlled tests, listed in Table 5.3. As proposed in the previoussection, deformation and fracture in the near surface of the scar is thought to occur atlocalized stress concentrations produced by roughness features of the wear scar. Thus inmodeling fine delamination, the scale of this mechanism is not determined by thedimensions of the contact model, but is instead linked to the inherent roughness featuresthat develop in the contact area. Those roughness features include both the wear particlestrapped in the contact and the uneven surface formed by particle delamination. Sincefracture is assumed to occur primarily along microstructural boundaries, the scale of finedelamination will be largely determined by the microstructure of the deformed near-surface.In considering the events which lead up to the fracture of a single debris flake, theprocess begins well before the potential particle is exposed to the surface. Smearing ofsubsurface grains in the sliding direction produce a long, thin-grain texture just below thesurface. Obviously, without the compressive loading under which the deformation occurs,fracture would occur much sooner. A damage-accumulation model of the induced strainand a maximum strain criteria for particle fracture presents one possible approach formodeling fine delamination. However, there are several serious difficulties to overcomefor the successful application of this type of model. First is determining the loading,including an accurate pressure and traction distribution over the contact area and updatingit over successive load cycles. Knowing the loading, the damage or strains at the surfaceand subsurface must be ascertained. Simple damage accumulation models [20] have usedstress-strain relationships from deformation theories of plasticity, but given the cyclicpattern of loading in sliding and the load path dependence of plasticity, the use ofincremental plasticity theory is much more appropriate. Both difficulties may soon beovercome with recent advances in computational power. Still a further obstacle remains,78which is the selection of the failure level, or the maximum allowable damage or strain.This value may be a property of the material, but is more likely dependent on severalfactors, such as the hydrostatic stress level, strain rate, and microstructure.Instead of using a strain criteria to predict delamination, a simpler approach wouldstart with material at the scar surface already having a highly deformed texture. During aloading cycle, the apparent contact area will have a complex distribution of contactpressure and traction determined by the interference of roughness features over theapparent contact area. In the lowest areas, there may be no contact, or if a lubricant ispresent, a light load will be transmitted by lubricant pressure and shear. Roughnessfeatures of intermediate height will carry elastic loads. At taller features, the loading willbe severe enough to cause plastic deformation, and in work-hardened areas whereplasticity is exhausted, the shear stress will reach the breaking strength of the grainboundaries until it is released during fracture.Determining the extent of fracture area is the necessary step to develop thisconceptual model into a predictive one. A quick estimate of the proportions involved canbe made using the results of unlubricated tests D 5 & 6. The five hundred cycle differencein test duration produced 0.12 mg of wear to the disc. This mass loss represents a volumeloss of 0.0308 * 10 mm3per cycle. Assuming a delamination thickness of 1 .tm, the areadelaminating in one cycle is 0.0308 mm2, which is only 0.6% of the disc scar area. Atpeak load, the maximum area under-going plastic deformation is estimated by neglectingthe contributions of elastic loading and assuming that the stress on the fracture planeparallel and 1 jim below the surface will not be attenuated significantly. Then taking theratio of the yield in shear (425 MPa) over the average friction force (75 N) gives a resultof 0.18 mm2. If the ball scar area is taken to represent the apparent contact area at peakload, 0.18 mm2 represents only 5% of this apparent contact area. These simple estimatessupport the suggestion that only a very few of the tallest roughness features fracture orencounter plastic deformation in a load cycle.79To determine the extent of the fracture area, an experimental correlation betweendelaminating area and the plastically deforming area per cycle could be used. For theexample above, the delaminating area is approximately a tenth of the plastically deformingarea. If this ratio is applied to the area needed to support the friction force at the yieldshear stress integrated over a load cycle for ten thousand cycles, with a 1 urn layerthickness, the total delaminating volume would be 0.36 mm3. This corresponds to a massof 2.8 mg, which is close to the mass losses observed in the unlubricated tests.Of course, there are severe limitation with this approach. It is not clear if thiscorrelation holds under lubricated conditions, where an unknown but likely significantportion of the friction force is derived from lubricant shear. As well, fracture of adelaminating particle does not immediately qualify it as a wear particle, since it could reattach itself or remain trapped in the contact area. It is apparent in the unlubricated tests,delaminated particles are rapidly broken up into much smaller debris, which may escapethe contact zone relatively easily.A simple model that holds for all conditions may not be possible. On the otherhand, a rigorous model would require use of micro-mechanics, since the microstructuralscale of the process invalidates the use of continuum mechanics. A completeunderstanding of the fracture process, plus consideration of the interaction of particles inthe contact after fracture would also be necessary. The complexity of such a model wouldbe substantial, with the extension of the model from a single to multi-asperity contactadding a further level of complexity.5.2.3 Fracture of Roughness FeaturesWear particles also originate from two distinct types of roughness features in thewear scar. As mentioned in section 4.1.3, hard particles trapped in the contact ploughgrooves in the wear scar. Material at the edges of these grooves fractures into long,80slender wear particles. Examples are found in the surface examinations of Figures 4.1,4.3 (d), and 4.10(a).This sort of particle formation mechanism was described over a decade ago by Suh[41], who postulated edge fracture occurred during flattening of grooves undersubsequent compressive loading. An extensive study of this mechanism alone was recentlyconducted by Xie and Williams [47], in which many characteristics of the process wererevealed. The authors found that for multiple passes of a hard wedge, run both in itsoriginal path or in increments of lateral displacement, ridge edges were fractured bymicromachining at higher wedge angles of attack or by repeated surface flow fromploughing at lower angles of attack. Specific wear rates were measured and varied almostfour orders of magnitude as wedge angle of attack was varied from 6 to 40 degrees. Theresults ofXie and Williams indicate that the particle formation rate for this mechanismwill be predominantly influenced by the angle of attack of a hard particle. Knowledge ofparticle geometry was not obtained, nor is it easy to predict, and at best it might bedescribed in probabilistic terms. However, given the large variation in wear rate with angleof attack, the uncertainty in angle of attack would translate into an even wider uncertaintyin the associated wear rate. Lateral displacement of the particle can also increase the wearrate by another order of magnitude, and in a reciprocating test, small displacements of aparticle are possible when the indenter reverses its motion or during the minimal normalloading of the return stroke.In this study, particle formation by groove edge fracture is minor and secondary tothe delamination mechanisms. It requires particles of sufficient hardness and size, alongwith the contact loading and displacement which also produce delamination. Concurrentoperation of delamination can have differing effects on the groove edge fracturemechanism. Delamination is a principal source of particles in the wear scar that may workhardened enough to plough the wear scar. Other sources of particles must also exist, asvisual evidence of extensive grooving prior to any appreciable delamination indicates ( see81Figure 4.3 (c) & (d)). Environmental contaminants, such as wear debris from earlier teststhat is not successfUlly cleaned from the test apparatus, are the most probable source ofthese first particles to plough the scar.Continuous sliding of a hard particle probably requires a surface with roughnessfeatures an order of magnitude smaller than the particle to minimize mechanicalinterlocking. Hence, the size of hard particles relative to the roughness features of thewear surface will influence grooving and consequently edge fracture. The scale ofdelamination will determine the size of both particles in the contact and surface roughnessfeatures. If particles an order of magnitude larger are not present, then ploughing and edgefracture will not occur. During fine delamination, both the particles and the roughness leftbehind are of essentially the same scale. As observed in the fine delamination ofunlubricated tests (Figure 4.15) groove edge fracture is limited. On the other hand,groove edge fracture occurs frequently during coarse delamination, which producesparticles much larger than the scar roughness.The second type of roughness feature fracture arises directly as a result ofdelamination. The edges surrounding a delamination site often contain protruding featureswhich fracture during subsequent loading. Analogous to a cantilever beam loaded by adistributed force, a protruding edge encounters a moment from the contact pressure actingon its unsupported area. The moment will be greatest at the base or attachment section,where fracture will occur if the moment-induced stresses are sufficiently large. Particlesformed are also quite small, typically no more than 2 tm in the longest dimension.Fracture of delamination edge features is also considered a secondary particleformation mechanism, and examples of this mechanism are found in Figures 4.2, 4.10,4.11, and 4.15. Particles formed by this mechanism are seldom larger than ito 2 urnalong their largest dimension, and their quantity will depend on the nature and extent ofthe delamination. The process occurs during the fine delamination of unlubricated slidingas well as on the edges of the large particles formed by coarse delamination.82Since particles produced by the fracture of both types of roughness features aretypically one or two orders of magnitude smaller than those formed by delamination, thecontribution of this mechanism to the total wear process could be neglected. Modeling ofthe groove edge fracture mechanism could be done with the relationships developed byXie and Williams [47], but would require either information or assumption of hard particleangle of attack and the geometry of the wear grooves produced.5.3 Third Body EffectsOnce indentation or coarse delamination shapes the initial scar geometry, particleformation begins to produce debris. In the presence of a lubricant and for reasons whichwill be considered later, particle entrapment in the contact zone is very high, which isequivalent to the rate of escape of particles being low. Larger particles are fragmented orpulverized by the contact pressure and translation of the indenter into very fine particles ofless than lim in size. Oxidation of these minute fragments of material likely occurs at thispoint in the process, for several reasons. Fine debris presents a large quantity of exposedsurface area on which an oxide layer can form. Growth of the oxide layer is a diffusion-controlled process, and the diffUsion distances are short in fine debris. As well, oxidationrates and diffUsion coefficients will be higher at increased temperatures from frictionalheating.By about five thousand cycles, the motion of the indenter distributes andcompresses the trapped fine debris into a third body layer over the majority of the contactarea. The layer provides a degree of protection for the disc surface, since the translation ofthe indenter can now be accommodated by shear or fracture of the third body layer.Consequently the particle formation rate from the disc is expected to decrease as a resultof the protection afforded by the third body layer.As mentioned in section 4.1.3, this layer is in a constant state of transition. Thelayer delaminates in places, and is ploughed by hard particles, with associated fracture of83groove edges. This layer is disrupted in ways that bear remarkable similarity to themechanisms which formed the original wear debris. However, as the longer duration openloop tests have shown, wear to the disc continues despite the presence of the debris layer,although it is not at all clear exactly how that wear occurs. Several possibilities exist.Surface examinations indicate that the third body layer occasionally flakes away,exposing the disc or first-body material. In those areas where the third body layer isvacant, wear could occur by the particle formation mechanisms described above, untilwear debris accumulates in the vacancy and the layer is reformed. However, a third body-vacant area would be at lower elevation than its surroundings, so it ought to encounterreduced loading and consequently less damage. More likely, the observed delamination ofthe third body layer does not occur at the interface of layer and disc surface, but rather asa result of a fracture below the disc surface, removing both material from the disc and thedebris layer. Surface and subsurface cracks observed in the sectioned discs suggests thatdelamination continues to occur in the presence of the third body layer, but at a muchreduced rate.Intuitively, one would expect a moderate flow of lubricant across the wear scarshould increase the particle removal rate over an unlubricated situation. However, theopposite effect occurs. This behavior is a result of the properties of a mixture of water andthe wear debris, as observed around the edges of wear scars. In the dry condition, thewear debris, which consists primarily of a fine powder of magnetite, or the iron oxideFe304,binds to itself very weakly. This powder exhibits a strong affinity for water,forming a dense viscous slurry when first wet. However, when more water is added, theslurry resists fbrther dissolution, and sheds the excess water. This behavior is thought toresult from a layer of water molecules adhering or hydrating on the surfaces of the finemagnetite crystals, and binding them into a slurry, that does not readily dissolve whenmore water is present. Thus the light flow of water over the wear scar keeps the fine weardebris particles together in and around the contact area. Under dry conditions, the84adhesion among oxidized particles is reduced and they can easily escape the contactregion.5.4 Third Body ModelsModels of several complex aspects of the third body layer are needed. First is amodel of how the third body layer forms, combining models of initial particle formationmechanisms with an assesment of debris retention or escape, to determine the rate atwhich initial material for layer formation becomes available. Once a partial layer hasformed, a model of particle formation in the presence of the layer is required. The finalstep would involve determining the extent or thickness of the third body, which could beaccomplished using the model of debris flake thickness proposed by Don and Rigney [261.Continued wear in the presence of the third body layer would then be a combination ofnew particle formation, layer delamination, and debris retention and escape. Since mostthird body mechanisms are currently poorly understood, complete third body modelingwill require much more additional work before a truly functional model is achieved.5.5 Surface Roughness and Friction TransitionsThe cutting fluid and water lubricated tests with various disc surface finishesproduced several unique friction transitions near the start of the tests. Those frictiontransitions can be explained within the context of the initial geometry transformationmechanisms described earlier. Several interesting comments can also be made from thepattern of these transitions.In cutting fluid, a steady-state friction level near 0.1 was observed for all surfacefinishes over the majority of the test duration. This friction level was immediately attainedby the polished surface. Initial friction peaks to a level typical of steady-state waterlubrication, but below levels associated with coarse delamination, were observed for therougher surfaces. These friction transitions suggest that geometry transformation occurs in85the rougher surfaces by the indentation mechanism. Additionally, until the initial surfaceroughness is removed from the scar area, the boundary lubricants in the cutting fluid donot operate effectively. It is also interesting to note that the size of the roughness featuresalone does not appear to determine either the magnitude or duration of the frictiontransition, since peaks for the silicon carbide pressed surfaces were lower and decayedmore rapidly than for the sandblasted surfaces. Instead the shape of the roughness featurescould have a more dominent effect, with the sharp edged features from SiC crystalsrequiring less tangential force and fewer cycles to remove than the smoother, morerounded features produced by sandblasting.With water as the lubricant, rougher discs immediately produced steady frictionlevels, and only the polished discs exhibited a distinct friction transition. The level of shortpeaks in the first thirty cycles in two of the polished tests ( shown in Figure 4.4k)indicates that the initial geometry transformation of the scar is occurring by the coarsedelamination mechanism, rather than by indentation. Thus some surface roughness maylead to a less severe mechanism of wear scar formation.Chapter 6ConclusionsTests performed over a wide variety of conditions have shown a pattern of wearmechanisms for the single asperity contact model of a hard ball rubbing a softer stainlesssteel disc. Extensive geometry transformation occurs, either by a continuous indentationmechanism or by a relatively coarse delamination mechanism, in the first few cycles of theinitial contact. In the case of scars formed by the indentation mechanism, the deformationcontinues to expand the scar size and increase its curvature for hundreds, even thousands,of cycles. Low temperature, high stress creep has been proposed as the explanation forthis extended deformation. In the case of coarse delamination, particle formation beginswith the initial geometry transformation, and evolves to a fine delamination mechanism asthe expansion of scar size slows. Particle formation in a scar formed by indentation occursby fine delamination and fracture of small roughness features within the scar. As thenumber ofwear cycles nears five thousand, under lubricated conditions, wear debristrapped in the contact are pulverized and compacted into a smooth and continuous thirdbody layer covering the scar. Wear continues at a reduced but highly variable rate, ascompared with unlubricated conditions in which a continuous layer does not form. Particleformation mechanisms under a third body layer are not clear, but are assumed to beassociated with the continuous process of layer delamination and reformation.Mass loss due to wear decreased as the disc hardness is increased; however, formartensitic stainless steel the basic mechanisms of particle formation appears largelyunaffected by hardness. Hardness did have a secondary influence on the type ofmechanism of initial scar formation, although the lubrication conditions had the primaryeffect on determining whether the scar formed by indentation or coarse delamination.8687Disc roughness was also considered to have a secondary effect on the initial scarformation mechanism. Long duration lubricated testing indicated that formation of a thirdbody layer was unaffected by disc roughness, with the valleys of any surface featuredeeper than the depth of the wear scar filling with wear debris. Despite this leveling effect,wear scar area increased with disc roughness. The shape of roughness features produceddistinct friction transitions at the start of the wear process; however, further short durationtesting with rough discs must be conducted to clarifj the exact effects of roughness onscar geometry and wear particle formation.Several of the basic wear mechanisms were empirically modeled; howeverproducing widely applicable models of the various wear mechanisms remains to beaccomplished before models of the various mechanisms can be combined into a predictivewear model. In particular, modeling of particle formation by delamination will requireworking on the microstructural scale at which delamination occurs. Particle formation bythe fracture of roughness features may be too widely varying to model and predict withany certainty. Finally, the modeling of the formation, decay and rebuilding of the thirdbody layer will be the most significant challenge facing tribologists.Appendix ACalculation Procedures for Contact ConditionsThe three techniques for calculating contact condition are presented below asexamples with typical test data, performed by Mathcad software. Thus the examplesare actual Mathcad worksheets and use Mathcad symbols, which have a few slightdifferences with ordinary mathematical symbols.A. 1 Ball Scar DimensionsTest L 32 ball scar length 0.89 mm width 0.54 mmload P :z 300 N disc shear strength k e 525 MPaContact dimension semi-axes:0.8910 0.54-10a: b:2 2Contact aspect ratioAR = 1.65bElliptical maximum contact pressure (Hertzian)p :z p0 = 1.191o Pa2•7EabShakedown factor:p0SF:= SF =2.27k iü6A.2 Disc and Ball CurvaturesThe relationships used in this procedure are from reference ( 2-16)Test L 13 ball radius 6.4 mmdisc curvatures longitudinal 1170 mm transverse 27.1 mm8889R 1 0.0064 R’ 2 1.170 2 : 0.0271load P 300 N disc shear strength k e 525 MPaElastic modulus - both steels E : 200 1O9 PaPoisson ratio v 0.3Combined modulus:/ 2E’çl-v).211E E’=l.llO PaSUM2 R 2 2 2 R 1 R’ 2 R” 2A;SUM BSUM-A A=78.552B =96.58Contact aspect ratio : 2A3AR — AR. =0.87 1B—=1.15AREquivalent radius and contact width:3PRee.4.A.B —4•E’-4Re =0.01 c =2.2710Elliptical maximum contact pressure (Hertzian):3.P 9p0:=2p0=2.77•lO Pa2mcShakedown factor:p0SF: SF=5.28k 10690A.3 Disc Scar Width Line ContactA line contact is assumed over the entire span of the disc scar width.Test L 7 Disc scar width 0.91 mm length 2.49 mmb0.9110 2.49. 102 - 2load P 300 N disc shear strength k e :z 250 MPaP’ :z2•bPredicted longitudinal curvature using Magel’s equation 4.13 (2-1)p zR p = 0.05actual measured curvature p 1= 0.0708 mEffective curvature: 1 1 -pSemi-contact width: I4P’•R’aJ it•E’ —4a=1.6410 mLine maximum contact pressure (Hertzian)P’E’ 2p0:tp0=1.28109 PaContact aspect ratioAR:za AR =2.78Shakedown factor:p0SF:= SF=5.12k eReferences[1] Magel, E. E., “Experimental and theoretical studies of the wear of heat exchangertubes”, M. A. Sc. dissertation, University ofBritish Columbia, Vancouver,Canada, 1990.[2] Blau, P. J., Friction and wear transitions ofmaterials, Noyes Publications, NewJersey, 1989, pp. 56-163.[3] Schumacher, W., “Wear compatibility of unlubricated stainless steels and othermetal couples”, in W.A. Glaseser, K.C. Ludema and S.K. Rhee (eds.), Proc. mt.Conf on Wear ofMaterials, American Society ofMechanical Engineers, NewYork, 1977, pp. 134-139.[4] Smith, A. F., “The unlubricated reciprocating sliding wear of a martensiticstainless steel in air and C02 between 20 and 300 °C”, Wear, vol. 123, 1988, pp.313-331.[5] Poggie, R. A. and Wert, 3. 3., “The influence of surface finish and strain hardeningon near surface residual stress and the friction and wear behavior of A2, D2 andCPM-1OV tool steels”, in R. G. Bayer and K. C. Ludema ( eds. ), Proc. mt. Confon Wear ofMaterials, American Society ofMechanical Engineers, New York,1991, pp. 447-502.[6] Jahanmir, S., “The relationship of tangential stress to wear particle formationmechanisms”, Wear, vol. 103, 1985, pp. 233-252.[7] Hertz, H., “On the contact of elastic solids”, J. Reine undAngewandteMathematik, vol. 92, 1882 pp. 156-171.[8] Huber, M. T., “On the theory of elastic solid contact”, Annin Phys. Lpz., vol. 14,1904, pp.153-6[9] Mlndlin, R. D., “Compliance of elastic bodies in contact”, I Appl. Mech., vol. 16,1949, pp. 259-268.[10] Hamilton, G. M. and Goodman, L. E., “The stress field created by a slidingcircular contact”, I App!. Mech., vol. 88, 1966, pp.371-6.[11] Hamilton, G. M., “Explicit equations for the stress beneath a sliding sphericalcontact”, Proc. Instn. Mech. Engrs., vol. 197C, 1982, pp. 53-59.9192[121 Poritsky, H., “Stresses and deflections of cylindrical bodies with application tocontact of gears and locomotive wheels”, J. App!. Mech., vol. 17 1950, pp. 191-201.[13] Sackfleld, A., and Hills, D. A., “Some useful results in the classical Hertz contactproblem” and “Some useful results in the tangentially loaded Hertzian contactproblem”, J. Stain Analysis, vol. 18 No. 2, 1983, pp. 101-110.[14] Johnson, K. L., Contact Mechanics, Cambridge University Press, Cambridge,1985, p. 98.[15] Melan, E., “Der spannungsgudstand eines Henky-Mises schen kontinuums beiverlandicher belastung”, Sitzungberichte der Ak. Wissenschaften Wien, Ser. 2Avol. 147, 1938, p. 73.[161 Johnson, K. L., Shercliff H. R., and Kopalinsky, E., “Shakedown of 2-dimensional asperities in sliding contact”, Cambridge University Technical Report,1989, p. B2.[17] Merriman, T. and Kannel, J, “Analyses of the role of surface roughness on contactstresses between elastic cylinders with and without soft surface coating”, J. ofTribology, vol. 111 No.2, 1989, pp. 87-94.[18] Yongqing, J. and Linqing, Z., “A full numerical solution for the elastic contact ofthree-dimensional real rough surfaces”, Wear, vol. 157, 1992, pp. 15 1-161.[19] Bailey, D. M., and Sayles, R. S., “Effect of roughness and sliding friction oncontact stresses”, I of Tribology, vol. 113, 1991, pp. 729-738.[20] Alpas, A. T., Hu, H., and Zhang, J., “Plastic deformation and damageaccumulation below the worn surfaces”, Wear, vol. 162-164, 1993, pp.188-95.[21] Rigney, D. A., and Glaeser, W. A., “The significance of near surfacemicrostructure in the wear process,” Wear, vol. 46, 1978, pp. 24 1-250.[22] Salesky, W. J., Fisher, R. M., Ritchie, R. 0., Thomas, G., “The nature and originof sliding wear debris from steels”, Proc. mt. Conf on Wear ofMaterials,American Society of Mechanical Engineers, New York, 1983, pp. 434-445.[23] Suh, N. P., “The delamination theory of wear”, Wear, vol. 25, 1973, pp. 111-124[24] Suh, N. P., “Update on the delamination theory of wear,” in D. A. Rigney (ed.),Fundamentals ofFriction and Wear ofMaterials, American Society of Metals,1981, pp. 43-71.93[251 Samuels, L. E., Doyle, E. D., Turley, D. M., “Sliding wear mechanisms”, in D. A.Rigney (ed.), Fundamentals ofFriction and Wear ofMaterials, AmericanSociety of Metals, 1981, PP. 13-41.[26] Don, J. and Rigney, D. A., “Prediction of debris flake thickness,” Wear, vol. 105,1985, pp. 63-72.[27] Rigney, D. A., Chen, L. H., Naylor, M. G. S., and Rosenfield, A. R., “Wearprocesses in sliding systems,” Wear, vol. 100, 1984, pp. 195-219.[28] HaIling, J., “A contribution to the theory of mechanical wear”, Wear, vol. 34,1975, pp. 23 9-249.[29] Jam, V. K. and Bahadur, S., “Development of a wear equation for polymer-metalsliding in terms of the fatigue and topography of the sliding surface”, Wear, vol.60, 1980, pp. 237-248.[30] Jam, V. K. and Bahadur, S., “Experimental verification of a fatigue wearequation”, Wear, vol. 79, 1982, pp. 241-253.[31] Hanson, M. T. and Keer, L. M., “An analytical life prediction model for the crackpropagation occurring in contact fatigue failure”, Tribology Transactions, vol. 35,1992, pp. 45 1-461.[32] Zhou, R. S., Cheng, H. S., and Mura, T., “Micropitting in rolling and slidingcontact under mixed Lubrication”, J of Tribology, vol. 111, 1989, pp. 605-613.[33] Finkin, E. F., “An explanation of the wear of metals”, Wear, vol. 47, 1978, pp.107-117.[34] Kimura, Y., “The role of fatigue in sliding wear”, in D. A. Rigney (ed.), -Fundamentals ofFriction and Wear ofMaterials, American Society of Metals,1981, pp. 187-219.[35] Godet, M., “Third-bodies in tribology”, Wear, vol.136 1990, Pp. 29-45.[36] Berthier, Y., “Experimental evidence for friction and wear modelling”, Wear,vol.139 1990, pp. 77-92.[37] Ludema, K. C., “Third bodies in wear models”, in D. Dowsen (ed.), WearParticles, Elsevier Science Publishers, 1992, pp. 155-160.[38] Qian, T., “Closed-loop computer control system for NRC fretting test rig - UserManual”, Version 2.0, NRC, Vancouver Canada, March 1993.94[39] Ko, P.L., Robertson, M., and Magel, E.E., ‘Wear particle formation in lubricatedsliding between a hardened sphere and a flat surface”, in D. Dowsen (ed.), WearParticles, Elsevier Science Publishers, 1992, pp 81 - 90.[40] Archard, J. F. and Hirst, W., The wear of metals under unlubricated conditions”,Proc. Royal Society (London), Vol. 236A, 1956, p. 397.[41] Suh, N. P., “Surface interactions”, in P. B. Senhoizi (ed.), Tribology Technology,Vol. I, Martinus NijhoffPublishers, the Hague, 1982, pp. 99-104.[42] Rigney, D. A., Chen, L. H., Naylor, M. G. S., and Rosenfleld, A. R., “Wearprocesses in sliding systems”, Wear, vol. 100, 1984, pp. 195-219.[43] Berthier, Y., Colombie, C., Vincent, L., and Godet, M., “Fretting wearmechanisms and their effects on fretting fatigue”, J. of Tribology, vol. 110, 1988,pp. 517-524.[44] Blau, P. J., “Friction microprobe investigation of particle layer effects in slidingfriction”, Wear, vol. 162-164, 1993, pp. 102-109.[45] Metals Handbook, 8th edition, vol. 7, American Society for Metals, 1972,pp.142-143.[46] Anderson, R. M., Adler, T. A., and Hawk, J. A., “Scale of microstructure effectson the impact resistance ofA1203”Wear, vol. 162-164, 1993, pp. 1073-1080.[47] Xie, Y., and Williams, 3. A., “The generation of worn surfaces by the repeatedinteraction of parallel grooves”, Wear, vol. 162-164, 1993, pp. 864-872. -95CharacteristicSteel AISI E - 52100Grade 25 (extra polish)Hardness 64 -66 Re (800 - 860 HV)Diameter 0.5 inch +1- 0.000 05 inchManufacturer Falex Corp., Aurora, flhinoisPart Number F-iS 19-50Table 3.1 Dynamic Test Specimen SpecificationsHeat Hardness ScaleCondition Treatment Rockwell Vickers Brinellannealed Fully Annealed 79-80 Rb 45- 155 HV 138- 147 Brtempered Hardened& 29-30 Re 295-305HV 280-289BrTemperedhardened FullyHardened 45-46Rc 445-455HV 420-429BrTable 3.2 Eqvivalent Hardnesses of 410 Stainless Steel Stationary Test SpecimensSurface Treatment Ra (itm) Xa (pm) (deg.) Rm (jtm)1 j.im polished 0.02 34 0.15 0.1sandblasted 1.3 58 8.3 12pressed with 120 grit SiC 2.7 65 14 21pressed with 60 grit SiC 5.8 120 17 45Table 3.3 Typical Surface Roughness Measurements of Stationary Specimen Discs(Measurements taken by Talysurf 5 profilometer with samplinglength of 0.8 mm and evaluation length of 5.6 mm)96TEST HARD- LOADNESS (N)K) 3ORc 600K2 3ORc 600K3 3ORc 600K4 3ORc 600KS 3ORc 600K6 3ORc 600K7 3ORc 600K8 3ORc 600ICANT CYCLES40:lcf 50k40:lcf 50k40:lcf 25k40:lcf 100k40:lcf 200k40:lcf 400k40:lcf 5k40:lcf 5000.095 poorly controlled, low dispi.0.1020.1050.1110.1090.1280.111N/AK9K10K])K 12K 13K 20K 21K 22K 23K 24K 25K 26K 27K 283ORc 6003ORc 6003ORc 6003ORc 3003ORc 52545Rc 30045 Rc 30045Rc 30045Rc 30045Rc 30045Rc 30045Rc 30045Rc 10045Rc 100thy 500distH2O 5kdry 1kdistH2O 5kno slip 5001000:lcf 5k1000:1 cf 5001000:lcf 1001000:lcf 501000:lcf 50k1000:lcf 100kdry 100dry 100dry 100Table 3.4 Summary of Open Loop Controlled Test ConditionsNote; ‘cf’ denotes cutting fluid lubrication, °k° denotes thousands of cyclesCutting Fluid ComponentsSelective refine paraffinFatty acid polydiethanol amideOleoyl-sacosidePropylene glycoleDemineralized waterHigh temperature stabilized chloroparafflnSulfonated oilsBoron acid esterothers - not disclosed% weight29.015.011.010.010.09.07.07.02.0Table 3.5 Jokisch W2-OP Concentrated Cutting Fluid Components(manufactured by Jokisch GmbH, West Germany)LUBR- NO. OF Avg. CoF COMMENTS• indent on bottom of scar- corroded, no alcohol rinse- CoF increased with # cyclesK14 3ORc 300 1000:lcf 5kK15 3ORc 300 1000:lcf 5kK16 3ORc 300 1000:lcf 2kK17 3ORc 300 1000:lcf 1kK18 3ORc 300 1000:lcf 500N/A no oscilloscope data, poor control0.20 1 impactingN/A inconsistent displacement, ball rolledN/A normal power amplifier down mid-testN/A normal load only - indentation sampleN/A no oscilloscope data0.185 irregular scar0.2 17 oxides on ball from holder0.287 OK0.296 OK0.135 displacement halved mid-test; scar with-in a scar0.264 long scratchy wear scar0.357 OK0.368 OK0.096 OK0.125 transient excessive displacementN/A totally unstable; scratchy scar0.8 16 bumpy trace; light impacting0.8 15 preload too low (2.5N); impacting97HARD- SURFACE LOAD LUBR- # CYCLESICANT * (1000s) COMMENTSK 30 30 Rc polished 300 1000:1 ofK31 3ORc sandblasted 300 1000:1 cfK32 3ORc 6ogritpres 300 1000:1 ofK33 30 Rc polished 300 1000:1 ofK34 3ORc 6ogritpres 300 1000:1 ofK35 3ORc 120 gritpre 300 1000:1 ofK36 3ORc 6ogritpres 300 1000:1 ofK37 30 Rc polished 300 1000:1 cfK41 3ORe sandblasted 300 1000:1 ofK42 3ORc 6ogritpres 300 1000:1 ofK43 3ORc 6ogritpres 300 1000:lcfK44 30 Rc 60 grit pres 300 1000:1 cfWi 30 Rc polished 300 dist. waterW2 30 Rc polished 300 dist. waterW3 30 Rc polished 300 dist. waterW4 30 Rc 120 grit pre 300 dist. waterW5 30 Rc 60 grit pres 300 dist. waterW6 30 Rc 60 grit pres 300 dist. waterW 7 30 Re sandblasted 300 dist. waterW8 30 Re 60 grit pres 300 dist. waterD i 30 Re polished 300 thyD 2 30 Re polished 120 thyD 3 30 Re polished 120 thyD 4 30 Re polished 120 dryD 5 30 Re polished 120 thyD 6 30 Re polished 120 dryD 7 45 Re polished 120 diyD8 8ORb polished 120 drySi 30 Re sandblasted 300 dist. waterS2 30 Re sandblasted 300 dist. waterS3 30 Re sandblasted 300 dist. waterS4 30 Re sandblasted 300 dist. waterS5 30 Re sandblasted 300 dist. waterS6 30 Re sandblasted 300 dist. water57 30 Re sandblasted 300 dist. water58 30 Re sandblasted 300 dist. waterS9 30 Re sandblasted 300 dist. waterSlO 30 Re sandblasted 300 dist. waterSil 30 Re sandblasted 300 dist. waterSi2 30 Re sandblasted 300 dist. waterS13 30 Re sandblasted 300 dist. waterS14 30 Re sandblasted 300 dist. water5i5 30 Re sandblasted 300 dist. waterSi6 30 Re sandblasted 300 dist. waterTable 3.6 Summary of Long-Duration Feedback Controlled Test ConditionsTEST NESS FINISH (N)50 test ran without any problems50 U5050 I,5050 II505050 U50 U50 II50 test ran without any problems50 test ran without any problems50 disc rotated 45 deg. during test50 test ran without any problems50 I,50505050 test ran without any problems10 Fn too high for control system10 test ran without any problems101015210 U10 test ran without any problems50 test ran without any problems5050 ‘I50 U -505050 H50 H50 H505050 U50 U50 U50 U50 test ran without any problems* Note : abreviation cf’ denotes cutting fluid98FEEDBACK CONTROLLED ThSTS - SHORT DURATIONL 1 30 Rc polished 300 1000:1 cfL 2 30 Rc polished 300 1000:1 cfL3 3ORc polished 300 l000:lcfL 4 45 Re polished 300 1000:1 cfL 5 45 Re polished 300 1000:1 cfL6 45Rc polished 300 1000:lcfL7 8ORb polished 300 1000:lcfL 8 80 Rb polished 300 1000:1 cfL 9 80 Rb polished 300 1000:1 cfL1O 3ORc polished 300 1000:lcfLii 3ORc polished 300 1000:lcfL12 3ORc polished 300 1000:lcfL 13 45 Re polished 300 1000:1 cfL20 80 Rb polished 100 diyL21 80 Rb polished 200 diyL22 80 Rb polished 200 thyL23 80 Rb polished 100 dzyL24 80 Rb polished 100 dryL25 80 Rb polished 300 dist. H20L26 80 Rb polished 300 dist H20L30 45 Rc polished 100 dryL31 45Rc polished 100 dryL32 45 Re polished 300 dist. H20L33 45 Re. polished 300 dist. H20L34 45 Re polished 300 dist. H20HARD- SURFACE LOADTEST NESS FiNISH (N)LUBR- NO. ofICANT * CYCLES COMMENTS50 no normal load data - signal grounded500 load cells grounded - no signal data5000 excessive displacement50 no signal data acquired500 ran without problems5000 friction increase near test end50 ran without problems500 ran without problems5000 ran without problems50 ran without problems500 ran without problems5000 friction varies widely50 ran without problems10 ran without problems0 disc scratched in setting up10 ran without problems201010 II20 ran without problems10 ran without problems20 ran without problems10 disc scratched on last cycle -10 ran without problems20 ran without problemsTable 3.7 Summary of Short-Duration Feedback Controlled Test Conditions99HARD- LOAD LUBR- # CYCLES AVERAGE MASS LOSS (mg)II LNJ ICANT (1000s’) C. of F.K8 30 Rc 600 40:1 cf 0.5 N/A (0.01) (0.06)K7 3ORc 600 40:lcf 5 0.111 0 (0.04)K3 30 Rc 600 40:1 cf 25 0.105 0.03 (0.02)K 1 30 Rc 600 40:1 cf 50 0.095 0.11 0.02K2 30 Rc 600 40:1 cf 50 0.102 0.04 (0.03)K4 3ORc 600 40:lcf 100 0.111 0.15 0.01KS 30 Re 600 40:1 ef 200 0.109 0.53 0.02K6 3ORc 600 40:1 cf 400 0.128 2.77 0.33K 13 30 Rc 525 no slip 0.5 N/A (0.03) (0.05)K 9 30 Rc 600 dry 0.5 N/A (0.04) (0.03)Ku 3ORc 600 dry 1 N/A 0.93 (0.75)K1O 3ORc 600 distH20 5 0.201 0.05 0.08K 12 30 Re 300 dist 1120 5 N/A 0.03 (0.03)K 18 30 Re 300 1000:1 cf 0.5 0.296 (0.02) (0.04)K17 3ORc 300 1000:1 cf 1 0.287 0.04 0K16 3ORc 300 1000:lcf 2 0.217 0.04 0.17K 14 30 Re 300 1000:1 cf 5 N/A 0.02 0.03K15 3ORc 300 1000:1 cf 5 0.185 0.12 (0.02)K23 45Rc 300 1000:lcf 0.05 0.368 0.13 (0.02)K22 45Rc 300 1000:lcf 0.1 0.357 (0.09) (0.06)K21 45 Rc 300 1000:1 cf 0.5 0.264 0.09 0K20 45Rc 300 1000:lcf 5 0.135 0.03 0K24 45 Re 300 1000:1 cf 50 0.096 0.01 0.05K25 4SRc 300 1000:1 cf 100 0.125 0.24 0.02K26 45Rc 300 dry 0.1 N/A (0.13) (0.02)K27 45Rc 100 dry 0.1 0.816 (0.15) (0.02)K28 45 Re 100 dry 0.1 0.815 0.17 (0.07)Table 4.1 Summary of Open Loop Controlled Test ResultsNote: abreviation “cf’ denotes cutting fluidMass losses in brackets represent mass increasesTable 4.2 Summary of Long Duration Feedback Controlled Test Results* abreviation “cf’ denotes cutting fluidNote mass loss values in brackets are mass increases10050 0.2250 0.2850 0.2150 0.2750 0.5350 0.0850 0.0750 0.1550 0.150 0.2550 0.1150 0.120.030.030.030.020.070.070.150.160.080.150.150.1324906.4 26302440215029203050292030833060290029703020Disc Disc No. of Mass Loss Disc wear scar Ball wear scar Disc curvature WorkTEST Hard- Surface Load I.ubr- Cycles Disc Ball Length Width Length Width Long Transv. Inputness Finish (N) icant* (1000’s) (mg) (mg) (mm) (mm) (mm) (mm) (mm) (mm) (3K30 30 Rc pot. 300 1000:1 cfK33 30 Re pot. 300 1000:1 ef 2.66 1.31 0.73 1.27 32K37 30 Rc pol. 300 1000:1 ci’ 2.59 1.22 0.76 1.21K31 3ORc sandbl. 300 1000:lcfK4i 30 Re sandbl. 300 1000:1 cf 2.80 1.46 1.08 1.52K35 3ORc l2ogritpr. 300 1000:lcf 3.07 1.18 1.16 1.37K32 3ORc 6ogritpr. 300 1000:lcfK34 3ORc 6ogritpr. 300 1000:lefK36 3ORc 6ogritpr. 300 1000:lcfK42 30 Re 60 grit pr. 300 1000:1 cf 3.52 1.41 1.25 1.66K43 30 Re 60 grit Pr. 300 1000:1 cf 3.22 1.42 1.24 1.45K44 3ORc 6ogritpr. 300 1000:lcf 3.12 1.40 1.15 1.54Wi 30 Re pot. 300 dist. water 50 0.12 (0.03) 2.6 1.2 0.96 1.52 28 6.4 7430W2 30 Re pot. 300 dist. water 50 0.24 (0.03) 7260W3 30 Re pot. 300 dist. water 50 0.27 (0.03) 2.77 1.27 0.96 1.30 7320W7 3ORe sandbl. 300 distwater 50 0.49 0 2.76 1.35 1.12 1.51 28 6.5 7390W4 3ORc l2ogritpr. 300 dist.water 50 0.29 0.17 3.25 1.53 1.38 1.62 53 8.2 7460W5 30 Re 60 grit pr. 300 dist. water 50 0.26 0.1 3.33 1.66 1.30 1.85 7260W6 30 Re 60 grit pr. 300 dist. water 50 0.24 0.08 7290W8 3ORc 6ogritpr. 300 disC water 50 0.26 0.1 3.30 1.62 1.21 1.70 70 5.8 7130D8 8ORb pot. 120 dry 10 3.12 0.12 3.63 2.53 2.12 2.54 15 6.6 1410D 2 30 Re pot. 120 dry 10 3.02 0.14 3.60 2.54 1.98 2.27 1330D 3 30 Re p0t. 120 dry 10 2.79 0.12 3.57 2.35 2.04 2.36 15 6.7 1380D4 3ORc pot. 120 dry 10 2.84 0.11 3.55 2.36 2.04 2.36 - 1430D 7 45 Re pot. 120 dry .10 1.95 0.16 3.43 2.24 1.80 2.24 16 7.2 1370D5 30 Re pot. 120 dry 1.5 1.41 0 3.25 2.04 1.65 2.00 211D 6 30 Re pot. 120 dry 2 1.52 0.02 3.44 2.05 1.74 2.00 270101Hard- Load Lubr- No. of Mass Loss Friction Disc scar Ball scar Disc Curvature ScarTEST ness (N) icant * Cycles Disc Ball Work Length Width Length Width Long. Transv. Depth(mg) (mg) (3) (mm) (mm) (mm) (mm) (mm) (mm) (urn)L 7 SORb 300 1000:1 cf 50 0.010 2.3 2.49 0.91 70.8 7.39 10L 8 SORb 300 1000:1 cf 500 0.009 25.8 2.49 1.02 64.6 6.82 10L 9 80 Rb 300 1000:1 cf 5000 0.032 307.3 2.53 1.20 45 6.93 13L 10 30 Re 300 1000:1 cf 50 0.001 2.6 2.35 0.61 227 9.37 1.5LII 30 Rc 300 1000:1 cf 500 0.000 32.3 2.26 0.74 147 8.01 3.4L 12 30 Re 300 1000:1 ef 5000 0.058 515.8 2.54 0.99 41.2 6.48 14L 13 45 Re 300 1000:1 cf 50 0.000 2.5 2.33 0.50 1170 27.1 0.4L5 45Re 300 1000:1 cf 500 0.000 30.7 2.31 0.67 778 42.8 0.8L 6 45 Re 300 1000:1 cf 5000 0.051 334.3 2.38 0.86 43.6 6.32 1.3L25 80 Rb 300 dist. H20 10 0.006 0.61 2.29 0.79 invisible 92.1 7.84 6L26 80 Rb 300 dist H20 20 0.001 1.43 2.26 0.85 invisible 75 7.44 6.5L32 45 Re 300 dist. H20 10 1.39 0.89 0.54L33 45 Rc 300 disC H20 10 0.005 1.13 2.52 0.45 0.8 0.42 41.6 0.87 12L34 45 Re 300 dist. H20 20 0.000 1.26 2.11 0.47 invisible 950 28 0.6L20 SORb 100 dry 10 0.005 0.48 2.46 0.40 0.75 0.28 22 0.33 11L24 80 Rb 100 day 10 0.29 2.45 0.41 0.53 0.35L23 SORb 100 day 20 1.61 2.53 0.88 1.04 0.78L22 SORb 200 dry 10 1.18 2.66 0.66 0.8 0.67L30 45 Re 100 day 10 0.012 0.57 2.97 0.44 0.62 0.39 81 0.97 10L31 45 Re 100 dry 20 1.78 2.64 0.63 0.68 0.66Table 4.3 Summary of Short Duration Feedback Controlled Test Results* abreviation “cf’ denotes cutting fluidmass losses from volume losses determined by surface mapping102HARD- SURFACE LOAD LUBR- # CYCLE MASS LOSS @ FRICTIOTEST NESS FINISH (N) ICANT (1000’s) DISC BALL WORK(mg) (mg) (J )Si 30 Rc sandblasted 300 dist. water 50 0.56 0.01 7110S2 “ “ 0.54 0.01 6930S3 “ “ ‘ “ “ 0.48 0.01 6930S4 “ “ “ “ “ 0.44 0.03 7000S5 “ “ “ “ 0.52 0.01 7231S6 “ “ 0.45 0.01 6966S7 “ “ “ 0.63 0.01 7120S8 “ “ “ “ “ 0.62 0.02 6590S9 “ “ “ “ “ 0.32 0 n/aSb I’ “ “ 0.33 0.01 7140Sli “ “ 0.47 0.01 n/a512 “ “ “ 0.38 (0.01) 7370S13 “ “ ‘ “ 0.55 0.02 6980514 “ 0.6 0.01 7060515 “ “ “ 0.46 0 7020516 30 Re sandblasted 300 dist. water 50 0.46 0 7120Table 4.4 Summary of Statistical Test Group Results@ ball mass loss values in parentheses denote a mass gainLubrication Surface Disc Sub-surfaceTest Conditions Finish Hardness Deformation Depth(pm)K 30 Cutting fluid polished 30 Re 4 - 8K 35 “ “ 120 grit pressed 30 Re not discernableW 1 water polished 30 Re 15 - 20W 7 “ sandblasted 30 Re 5 -10W 8 60 grit pressed 30 Rc 15 - 20D 7 unlubricated polsihed 45 Re 10 - 15D 8 “ polished 80 Rb not disceranableTable 4.5 Depth of Visible Plastic Deformation in Sectioned Discs103Hard- Load Lubr- No. DISC scar BALL scar Disc curvature contact shear contact shakedownTest ness (N) leant Cycles Length Width Length Width Long. Transv aspect strength pressure(mm) * (mm) (mm) (mm) (mm) (mm) ratio (MPa) (MPa)L25 80 Rb 300L26 SORb 300waler 10 2.29 i 0.79 invisible 92.1 7.84 2.3water 20 2.26 i 0.85 invisible 75 7.44 2.5250 1400 5.6 1250 1330 5.3 1L32 45 Re 300L33 45 Re 300L34 45 Rc 300waler 10 d 0.89 0.54 1.6water 10 2.52 d 0.45 0.8 0.42 41.6 0.87 1.9water 20 2.11 i 0.47 invisible 950 28 1.1525 1192 2.3 bs525 1705 3.2 bs525 1870 3.6 1L30 45Rc 100L31 45 Re 10010 2.97 d 0.44 0.62 0.39 81 0.97 1.620 2.64 d 0.63 0.68 0.66 1.0525 790 1.5 bs525 426 0.8 bs‘Table 5.1 Shakedown factors for short duration feedback controlled tests* scar formed by i - indentation or d - coarse delamination** calculation method: c - circular, from curvatures I - line contact; bs - ball scar areaVickers Pile-Up ScarCondition Hardness Ratio Depth(kgf7mm2) ( ..tm)annealed 250tempered 430factor**L 7 SORb 300 1000:lcf 50 2.49 i 0.91 70.8 7.39 2.8 250 1290 5.1 1L8 8ORb 300 1000:lcf 500 2.49 i 1.02 64.6 6.82 3.3 250 1210 4.8 1L9 SORb 300 1000:lcf 5000 2.53 i 1.20 45 6.93 4.1 250 1090 4.4 1L1O 3ORc 300 1000:tef 50 2.35 i 0.61 227 9.37 1.6 430 1620 3.8 1Lii 3ORc 300 I000:lcf 500 2.26 i 0.74 147 8.01 2.1 430 1460 3.4 1L 12 30 Re 300 1000:1 cf 5000 2.54 i 0.99 41.2 6.48 3.0 430 1200 2.8 1L 13 45 Re 300 1000:1 ef 50 2.33 i 0.50 1170 27.1 1.2 525 2770 5.3 cL5 45Rc 300 1000:lcf 500 2.31 i 0.67 778 42.8 1.1 525 2710 5.1 cL 6 45 Re 300 1000:1 ci’ 5000 2.38 i 0.86 43.6 6.32 2.5 525 1280 2.4 1L20 SORb 100L24 SORb 100L23 80ffl 100L22 SORb 200thythydrydrydrydry10 2.46 d 0.40 0.75 0.28 22 0.33 2.710 2.45 d 0.41 0.53 0.35 1.520 2.53 d 0.88 1.04 0.78 1.310 2.66 d 0.66 0.8 0.67 1.225025025025090910302357133.6 bs4.1 bs0.9 bs2.9 bshardened1.21.11.055255.13.22.8Table 5.2 Estimated initial or first-cycle indentation depths104Disc Disc Disc wear scar Ball wear scar contact shear contact shake Disc curvature Predicted PredictedTest Hard- Surf: Load Length Width Length Width aspect strength pressure -down Long. Transv curvature vs.ness Finish (N) (mm) (mm) (mm) (mm) ratio (MPa) (MPa) factor (mm) (mm) (mm) MeasuredK30 3ORc poT. 300K33 3ORc p01. 300 2.66 1.31 0.73 1.27 1.7 430 618 1.4 32 6.4 26.4 1.2K37 3ORc pot. 300 2.59 1.22 0.76 1.21 1.6 430 623 1.4K31 3ORc sandbl. 300 430K 41 30 Rc sandbl. 300 2.80 1.46 1.08 1.52 1.4 430 349 0.8K35 3ORc l2Ogr. 300 3.07 1.18 1.16 1.37 1.2 430 361 0.8K32 3ORc 6Ogr. 300 430K34 3ORc 6Ogr. 300 430K36 3ORc 6Ogr. 300 430K 42 30 Re 60 gr. 300 3.52 1.41 1.25 1.66 1.3 430 276 0.6K43 3ORc 6Ogr. 300 3.22 1.42 1.24 1.45 1.2 430 319 0.7K44 3ORc 6Ogr. 300 3.12 1.40 1.15 1.54 1.3 430 324 0.8WI 30 Rc pot. 300 2.6 1.2 0.96 1.52 1.6 430 393 0.9 28 6.4 30.0 0.9W2 3ORc p0t. 300 430W3 3ORc pot. 300 2.77 1.27 0.96 1.30 1.4 430 459 1.1W 7 30 Rc sandbt. 300 2.76 1.35 1.12 1.51 1.3 430 339 0.8 28 6.5 26.8 1.0W4 3ORc l2Ogr. 300 3.25 1.53 1.38 1.62 1.2 430 256 0.6 53 8.2 28.9 1.8W5 3ORc 6Ogr. 300 3.33 1.66 1.30 1.85 1.4 430 238 0.6W6 3ORc 6Ogr. 300 430W8 3ORc 6Ogr. 300 3.30 1.62 1.21 1.70 1.4 430 279 0.6 70 5.8 26.6 2.6D 8 80 Rb pot. 120 3.63 2.53 2.12 2.54 1.2 250 43 0.2 15 6.6 13.2 1.1D 2 30 Rc pot. 120 3.60 2.54 1.98 2.27 1.1 430 51 0.1D 3 30 Rc pol. 120 3.57 2.35 2.04 2.36 1.2 430 48 0.1 15 6.7 14.8 1.0D 4 30 Re pot. 120 3.55 2.36 2.04 2.36 1.2 430 48 0.1 -D 7 45 Re pot. 120 3.43 2.24 1.80 2.24 1.2 525 57 0.1 16 7.2 15.0 1.1D 5 30 Rc pol. 120 3.25 2.04 1.65 2.00 1.2 430 69 0.2D 6 30 Re pot. 120 3.44 2.05 1.74 2.00 1.1 430 66 0.2Table 5.3 Shakedown factors and predicted curvatures for long duration feedback controlledtest (Note: contact aspect ratio from ball scar dimensions)Figure 2.1p0030•Maximum yield parameter in and belowthe surface versus friction coefficient105IL 05IL = 025...1L0y (Ten,ion)I-0I.o(Conision)Figure 2.2 Normal stress in the direction of the tractive force (a.,,) alongthe x-axis for various friction levels07In surface0504 Below surface0•06 0700CC‘-I0-JTraction coefficient p.106Subsurface fLowSurface • Subsurface fLowRepeated plastic flow3 SurfaceElasticNf tow21ELasticLimit- Kinematic hardening shakedown limit0.0 01 02 03 01+ 05Figure 2.3 Line contact shakedown map for a kinematic hardening material107A: Subsurface flowSC: Surface + Subsurface flow14.Repeated plasticElasticflowShakedown3C—04-U8: Surface flowd0-J 2Elastic1Elastic limitKinematic hardening shakedown limit01 02 03Traction coefficient04. O5Figure 2.4 Circular contact shakedown map for a kinematic hardening material108(d)Distribution of contact pressure and subsurface orthogonal and principal shearstresses for a ground surface contacted by a smooth indenter.(a) surface pressure distribution (b) iso-view of orthogonal shear stresses(c) contour plot of orthogonal shear stress from (b)(d) effect of friction, t = 0.1, on stress distribution shown in (b)(e) iso-view of principal shear stresses (f) contour plot of stresses in (e)(g) effect of friction, i = 0.1, on stress distribution shown in (f)..0 Norma1Qd P(,.)3.02.01.0.0 SSS.O 1118.0 1677.0 nicron 2236.0 27S5O 33.NOR11cL FORCE366O N/micron(c) (0(g)Figure 2.5YFigure 3.1 NRC Fretting Wear RigxTR I AX I ALLOAD CELL109Figure 3.2 Dynamic Specimen HolderFigure 3.3 Ideal Specimen Displacement and Normal Load Signals110System Parameters:massspring stiffnessdampingfrictional dampingFSkm=O.77kgk= 17, SOON/rn0.63C.. <<C.friction viscoustime///ciscow +trtcfionmFigure 3.4 Dynamic Model of NRC Fretting Wear Rigij CD CD C,) ICOUPIJTEI.SIGNALGENER11ONAflGNUATOICOMPUTER2.DATAAQUISIS11ONSYST)1DIGITAl.OSCIII.OSCOPCISNAXER2POTIRA)W11FIRI1çN]oUV\I@Li©I_J__________0IDRIVEoiciTitosaunscor2cIIAR;A1.I[’IIFIERS112(a)(b)Figure 4.1 Water lubricated annealed disc wear scar after ten sliding cycles113(c)Figure 4.2 Water lubricated annealed disc wear scar after twenty sliding cycles114Figure 4,3 Water lubricated hardened disc wear scar after ten and twenty sliding cycles(a) & (b) ten cycle scar(a)(b)- -; ,: -‘I-—:-,r- ..‘.‘.S — - - - -‘‘/7 - - --1 -- ---> ,—-115(c)(d)Figure 4.3 Water lubricated hardened disc wear scar after ten and twenty sliding cycles(c) & (d) twenty cycle scar116(a)(b)Figure 4.4 Wear scar from unlubricated ten cycle test on an annealed disc(a) entire scar (b) escaped wear particle117(c)(d)Figure 4.4 Wear scar from unlubricated ten cycle test on an annealed disc(c) centre of scar (d) closeup of the interior of scar fissure118(a)(b)(C)Figure 4.5 Front and rear ofwear scar from unlubricated ten cycle test on an annealed disc(a) & (b) front edge of wear scar (c) rear edge119(a)(b)Figure 4.6 Wear scar from ten cycle, 200 N loaded, unlubricated test on annealed disc(a) entire scar (b) rear edge of scar120(c)(d)L ? - - - r t.:l-‘ A - - - -— •,-\_—. fj— ‘-_s._ - ---=—-—I- --. -j.--.-——--4-.-----—---— — —7---—----- - -?.-‘ -.±-VFigure 4.6 Wear scar from ten cycle, 200 N loaded, unlubricated test on annealed disc(c) & (d) close-ups of rear edge of scar(a)(b)Figure 4.7 Wear scar from twenty cycle unlubricated test on annealed disc(a) entire scar (b) front edge of scar121C-aC)‘‘CDOc‘I”10-0\—c,r-tS.S(t0-c; CDo ,-t,000 C C 0 0 CD 0-CD C12 0 CD p CD 0-0-C,, 0OQ C -I CDa0t)123(a)(b)— _fJ__ c?psz?.f :----4) .%i4;:--‘____ ______‘______-- .-——-—.1_ ______1- - ---.—Figure 4.8 Wear scar from ten cycle unlubricated test on hardened disc(a) entire scar (b) rear edge of scar124(c)(d)Figure 4.8 Wear scar from ten cycle unlubricated test On hardened disc(c) top rear edge of scar (d) close-up of scar interior125(a)(b)Figure 49 Wear scar from twenty cycle unlubricated test on hardened disc(a) entire scar (b) bottom edge mid-scar126(c)(d)Figure 4,9 Wear scar from twenty cycle unlubricated test on hardened disc(c) & (d) c1oseups of scar interior127F—.,.:,‘.- -- .q— .-t :- —-___4 - --fi .. —:....y• S--’.___-(3)(2)(1)Figure 4.10 Fifty Cycle Hardened Disc Wear Scar, from Test K 23128(2)(1) (3)Figure 4.11 Unlubricated One Hundred Cycle Hardened Disc Wear Scar, from Test K 27129(1)(2)Figure 4.12 Five Hundred Cycle Hardened Disc Wear Scar, from Test K 21130(4)(5)Figure 4,13 Five Thousand Cycle Hardened Disc Wear Scar, from Test K 20131(2)(1) (3)Figure 4.14 Five Thousand Cycle Tempered Disc Wear Scar, from Test K 15132(2)(1) (3)Figure 4.15 Intermediate duration unlubricated wear scar on a polished disc(d)ooo04(f)Figure 4J6 Optical micrographs of disc wear scars from short duration, cutting fluid lubricatedtests (a) & (d) annealed (b) & (e) tempered (c) & (f) hardened133L(a)r-all photos at50 x magnification(e)—134(b)(c)Figure 4,17 Long duration cutting fluid lubricated wear scar on a polished disc (Test K30)(a) entire scar (b) & (c) close-ups of a hard particle trapped in the scar(a)(d)(e)(f)__--- _. - _7E--—-— — .—— —. —.— 3_._4__ _(• ————-,------.. —.‘—- - ---_/ _.——— —--— .—. —• —- - - — - --L_.,-. ;_—___•_ ‘.—— — -_c•_• ———4 —-.•--s-—--• •-•— --. - - - -•‘•‘ —S. .135Figure 4.17 Long duration cutting fluid lubricated wear scar on a polished disc (Test K 30)(d) & (e) trailing edge of scar (f) front and center of scar interior136(1)Figure 4.18 Long duration water lubricated wear scar on a polished disc ( Test W 2)entire scar (1) front center of scar interior—137(2)(3)Figure 4.18 Long duration water lubricated wear scar on a polished disc (Test W 2)(2) & (3) close-ups of trailing edge(D-C.)CDQ00CD I00CD -i,—‘0CD)‘-‘ci (1 CE’ CD C 0 GQ C 0°140Figure 4.21 Crosssections of a 60 grit SiC pressed disc surface (unworn)141(a)(b)(c)Figure 422ba%he_vrZcL:a____.4..-‘-Etched sections of discs from long duration cutting lubricated tests(a) front ofwear scar (b) centre of scar - both from a polished disc withwear direction right-to-left (c) centre of scar from a 120 grit SiC pressed disc142arr— A—• a - —. * —- — .r.-- - -. — . -.--4•’—‘ .- — — t 4 ‘1 .1:. -..%; *-.-_. .,...; . -- 4* .--t.- •:a- --.-“—. .-.A .--c-;-f’fr-.4.-; •-•-‘)•t .•r-‘— .•• .—.--F ...Z6’: :. -•‘, - 2a—•1..’i ---s. —.: -. .-- •-: -•- *-p...y._... •&t’*..- e• *..ffç. — — 4 — afa -• I 4*—. es-.FK44• a.-. -•‘••,-‘,l•. _ — a” •-.—- •: j --—I,. •-1 •1• -I aI. •,* -. p. •_- -1;k’,. . - • s-: •.-‘ a-. .- -. -. -I’ —.. r-’ - — e -• - •.. .r . •• - . . • •--.-.• ••-. -j .t •-:-‘ - - --— :.•-- -._--t.i—-- : ;Figure 4.23 Etched sections of discs from long duration water lubricated tests143(a)(b)-——i? —— - /-r-,(—•1 — — .—— — —— - - —,- —a- -- -— r ——- J.__._w - — - —II ,s —. — 4 —: :- -- 9’,./‘flrFigure 4.24 Sections of sandblasted discs from long duration water lubricated tests(a) unetched (b) etched144(a)(b)Figure 4.25 Sections of 60 grit SiC pressed discs from long duration water lubricated tests(a) & (b) near centre ofwear scar wear direction right-to-left145(c)(d)Figure 4.25 Sections of 60 grit SiC pressed discs from long duration water lubricated tests(c) unetched (d) etched - both near centre of wear scar - wear directionright-to-left146(a)(b)Figure 4.26 Sections of wear scars from long duration unlubricated tests(a) & (b) annealed disc - wear direction right-to-left147(c)(d)Figure 426 Sections ofwear scars from long duration unlubricated tests(c) & (d) hardened disc - wear direction right-to-left1480.08annealed • tempered * hardenedJ_%O.06 ..EU)U) 0.04 - - .. ...0Cl)U)E 0.02 .. ...... ..0 I I IIIIII I I 11111 I I 1111110 100 1000 1E4cyclesFigure 4.27 Cutting fluid lubricated short duration test mass losses149101—S _AA0.1AFn=600NA Fn400N0.01 A0.001 — 11111111 111111(11 ii 1111111 10 100 1000Thousands of CyclesFigure 4.28 Open Loop Mass Loss Results - 40:1 Cutting Fluid Lubrication400 N data from Ko et. al.[39]1500.50.40)E‘—0.3U,U)0-J00.2Cl)0.10Figure 4.29 Mass Loss versus Surface Roughness for Cutting Fluid Lubricated Tests* Note: average of 3 tests with range in mass loss results:polished .06 mg disc, 0 mg ball 60 grit pressed .08 mg disc, .08 mg ball0.320.10Figure 4.30 Mass Loss for SiC Pressed Discs including Stress Relieved Discs,Cutting Fluid Lubricated * Note: average of 3 tests with range inmass loss for stress relieved 60 grit press .13 mg disc, .02 mg ballpolished * sandblasted 120 grit pressedSURFACE FINISH60 grit pressed *120 grit pressed 60 grit pressed * 60 gr stress rel.*SURFACE FINISH0.50.40.20.103.532.52I0.50Disc Hardness151Figure 4.32 Mass Loss versus Disc Hardness for Unlubricated Tests. * Note: average of 3tests with range in mass loss : 0.21 mg disc, 0.03 ball.0.3polished *Figure 4.31sandblasted 120 grit pressed 60 grit pressed *SURFACE FINISHMass Loss versus Surface Roughness for Water Lubricated Tests* Note: average of three tests with range in mass loss results:polished .15 mg disc, 0 mg ball; 60 grit .02 mg disc .02 mg balldiscIECl)Cl)0-JCl) 1.5Cl)Iball8ORb 3ORc* 45Rc1521 I I I I I0.8 -0.6 -yiYi0.40.2030 35 40 45 50 55 60 65Xord. Mass Loss (mg * 100)Figure 4.33 Linearized Cummulative Distribution Function of Statistical TestSeries Mass Loss Results, Plotted with a Gaussian Model (heavy line)I I I I I I153300250200150100500-50-100(b)(c)3002502001000u_So0-50-100[—Fn —Ft disp](d)Figure 4.34 Normal and frictional force and displacement signals from water lubricated shortduration tests (a) & (b) annealed discs (c) & (d) hardened discs(a) & (c) ten cycle tests (b) & (d) twenty cycle tests(a)EE0.0CEEa.0C3002502001501000- 500-50-100iJ:::::‘:::: Ea.0C3002502001501000 50-50(\PJf\J\7\7\j \/ \/ \_I \7 \I \I \7\T’J’7 ‘Ii’,r\[\p\]\\\/q ]\/\J i777I70154L—Fn —Ff ..-.-dispLJFigure 4.35 Normal and frictional force and displacement signals from unlubricated shortduration tests (a) & (b) annealed discs (c) & (d) hardened discs(a) & (c) ten cycle tests (b) & (d) twenty cycle tests•iAAAAI i,...It ......i1\100806040200 0a-20-40-60(b)LI1\(a)100806040200LI.. 0.20-40-601251007550250U-EE0.0EE0.0hWI II-25220.01501251007550::.:A:zj:::::i:z.Il U WIllIIIl1i 11112(c)11 I ll II JliII.llut44JwLiflhI IjAWM 2(d)Ca).C-)a)0C)C0o.iCa).C-)a)0C)C01550.3Ca)C)a)0C)C00.100.31 10 100 1000 1E4(a)(b)(c)00.31 10 100 1000 1E4—----1 10..,,,,,,,I I 111110001000 1E4cyclesFigure 4.36 Friction traces from cutting fluid lubricated short duration tests(a) annealed discs (b) tempered discs (c) hardened discs1560.4(a)0.3C§0.2 . ........0.1 - I.-0 I I I I I I I I0 10000 20000 30000 40000 500000.4(b)0 1000 2000 3000 4000 50000.4(c) sandblastedO.3 ..... .....0 I I I I I I I I0 250 500 750 1000 1250 1500cyclespolished— sandblasted — 120 gr. pressed —60 gr. pressedFigure 4.37 Friction coefficient plots of various disc roughnessesfor cutting fluid lubrication. (a) entire test (b) firstfive thousand cycles (c) first fifteen hundred cycles0.40.34-,C.)a)3 0.20139-0.10cyclesFigure 4.33 Friction plots of 3 cutting fluid lubricated 60 grit SiC pressed discs0.40.34-a)8 0.2C0130.10cycles15715001500Figure 4.39 Friction plot of first fifteen hundred cycles of two cutting fluid lubricatedtests on sandblasted discs0 250 500 750 1000 12500 250 500 750 1000 1250(a)(b)158— I I I I I I30000 40000 500000 10000 200000.40.30200.40.30:00.40.3C00o 0.2C00 1000 2000 3000 4000 5000(c)0.10Figure 4.40 Friction coefficient plots of various disc roughnesses for water lubrication(a) entire test (b) first five thousand cycles (c) first one hundred cycles0 20 40 60 80cycles1000—WI —W2—W3. I I I I I0 100 200 300 400cyclesFigure 4.41 Friction plot of first 400 cycles for 3 water lubricated tests on polished discs1590.6.4a)C.)a)0C.)C00.8Ca)C)a)0C-)C00.40.20(a)I0.81600.6 •‘0.4 .-.-.—.-.........- .0.2 ..._0• I I I I I0 2000 4000 6000 8000 10000cyclesannealed 8ORb — tempered 30 Rc — hardened 45 RcI(b)0 20 40 60 80Figure 4.42 Friction coefficient plots for various disc hardnesses (a) entire test (b) first one100hundred cycles-3-c,).1-’C10Cl)C’)0U)U)cuEFigure 4.43 Mass loss and frictional work comparisons (a) mass loss/work input ratios forvarious lubricant and disc surface conditions (b) mass loss versus work inputfor the statistical test ( S ) eties(a)16110I0.10.010.8=E...........I.I Icut. fi. lubr. cut. fi. lubr. water lubr. water lubr. unlubr.polished sandbl. polished sandbl. polished(b)0.6EU)U)0-JU)U)Cu0.40.2..I I I I6500 6700 6900 7100 7300Energy ( J)7500CCD CDCD H CD. Q..0 —‘CD 0 Q. J(, o° C,) ‘J-i C’) -0 ‘..-.) C-) CD cQC -e 1 CD C’) C’) CD0ON LJrJ J CDCD CD, cl ,—‘C CD CD. CD C) CD- CD C) 1 CD CD Cl)C CEE164(a)EE(b)0 0.5 1 1.5 2 2.5 3(mm)0.080.060.040.020.080.060.040.020Figure 4.46 Longitudinal profilometer traces of polished disc scars with best fitradii of curvature (a) cutting fluid lubricated (b) water lubricated0.5 1 1.5 2 2.5 3(mm)16510(a) [annealed-V tempered hardenedjI 111111 I I 1111111 I 11111110 100 1000 1E4cycles10(b) C. annealed tempered • hardenedJE0.1 I I IIIHI I I liIij I I I 1111 I I iIii( I I1 10 100 1000 1E4 1E5cyclesFigure 5.1 Disc scar widths from cutting fluid lubricated tests (a) short duration test resultsonly (b) single cycle width estimations and one long duration width added166300250 -200 -Load(N)Pa. IPt115o100 -50 -\00 01 0.2 0.3 0.4 0 0.6 0.7 0Friction CoefficientFigure 5.2 Shakedown load for circular contact of a 11211 diameter steel ball on a flat of annealed,tempered, and hardened 410 stainless steel

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0080930/manifest

Comment

Related Items