4

Invited Papex for the World Tribology Congress Institution of Mechanical Engineers London, England - September 1997

Design and Validation of LaboratoryScale Simulations for Selecting Tribomaterials and Surface Treatments

Peter J. Blau, B.S., M.S., Ph.D., FASM Metals and Ceramics Division Oak Ridge National Laboratory P. 0. Box 2008 Oak Ridge, TN 37831-6063 USA

e-mail: blaupj@ornl.gov (423) 574-5377; fax: (423) 574-6918

ABSTRACT

Engineering approaches to solving tribology problems commonly involve friction, lubrication, or wear testing, either in the field or in a laboratory setting. Since wear and friction are properties of the materials in the larger context of the tribosystern, the selection of appropriate laboratory tribotesting procedures becomes critically important. Laboratory simulations must exhibit certain key characteristics of the application in order for the test results to be relevant, but they may not have to mimic all operating conditions. The current paper illustrates a step-by-step method to develop laboratory-scale friction and wear simulations based on a tribosystem analysis. Quantitative or qualitative metrics are established and used to validate the effectiveness of the tribosimulation. Sometimes standardized test methods can be used, but frequently a new type of test method or procedure must be developed. There are four factors to be addressed in designing effective simulations: (1) contact macrogeometry and the characteristics of relative motion, (2) pressure - velocity relationships, (3) thermal and chemical environment (including type of lubrication), and (4) the role of third-bodies. In addition, there are two typical choices of testing philosophy: (1) the worst-case scenario and, (2) the nominal-operations scenario. Examples of the development and use of simulative friction and wear tests are used to illustrate major points.

.

Research sponsored by the U.S. Department of Energy, Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Transportation Technologies, as pari of the High Temperature Materials Laboratory User Program, under contract DE-AC05-%OR22464 with Lockheed Martin Energy Research Corp.

1

T h e submitted manuscript has been authored by a contractor Of the U.S. Government under contract NO. DE- AC05-960R22464. AccordinQlV. the U.S. Government retains a nonexclusive. royalty-free license to publish or reproduce the published form of this contribution. or allow OthefS to do 50, tor U.S. Government purposed.’

1

1.0 Introduction

Laboratory-scale friction and wear tests often play an important role in tribomaterials and lubricant selection. Since wear and friction are characteristics of the materials in the larger context of the tribosystem, the selection of the proper testing methodology is vitally important when testing is used to screen materials for specific applications. Laboratory-scale simulations must thenfore exhibit key characteristics of the intended application in order to be useful. It is possible, and even probable, that an application which must meet several performance specifications will requirt more than one type of screening test. Thus, a "balance of properties," which could include not only tribological performance but other factors, such as fatigue and corrosion resistance, may be required in d e r to complete the material evaluation.

History fails to record when engineers first began using laboratory simulations to solve friction and wear problems, although it is probable that laboratory-scale friction experiments predated those of wear.' It is reasonable to assume, however, that the first experiments in tribomaterials selection were performed under field conditions, often with disastrous results, both in terms of the machine's response and even perhaps, the fate of the test engineer. Later, three factors -- the ability to explore alternatives in private before drawing conclusions, the avoidance of the risks associated with full-scale tests, and the convenience of making measurements on a controllable, model system -- began to make laboratory-scale testing appealing. As engineers withdrew into their testing laboratories they faced the challenge of connecting their results convincingly to the behavior of actual machines in the field. This central issue in material and lubricant screening stil l exists today.

Those charged with selecting materials or lubricants for tribological applications undoubtedly would prefer simply to use handbooks, simple formulae, tables of data, or computer databases. While a number of comprehensive tribology handbooks have been ~ublished*g,~J, these cannot address every possible friction or wear situation, and experience shows that it is prudent not to extrapolate friction and wear data much beyond the conditions under which they were obtained. Despite the recent growth and sophistication of computer programmes for design and materials selection, physical properties data, like friction coefficients and wear rates, must still be found largely by experiment.

A principal question in tribosimulation is: With what level of exactitude must the laboratory test method duplicate the operating conditions of the machine in order to produce useful information for selecting materials and/or lubricants? Sometimes, useN design information can be obtained from relatively simple testing geometries, like the pin-ondisk or block-on-ring. Sometimes, however, such simple systems inadequately simulate the tribological behavior of the component under its normal conditions of use.

2

Godet, et al.6 once observed: "[A] simulation can be compared to a black box, entry wnditions are fed to it and results are transposed to the application With a minimum amount of interpretation. The difficulty resides in carefully designing the black box." The design and validation of laboratorj -scale simulations for selecting tribomaterials are therefore issues of major consequence in mechanical systems engineering. This paper discusses methods to characterize tribosystems and to develop and validate simulative tests.

2.0 Factors in the Development of Tribosimulations

The five basic steps in developing an effective tribosimulation are: (1) reaching a clear understanding of the tribosystem, (2) developing metrics by which the problem can be quantified and its solution tested, (3) selecting the necessary conditions for the simulation, (4) selecting or developing a testing procedure to reproduce those necessary conditions, and (5 ) validating the simulation.

2.1 Understanding the Nature of the Tribosystem

The first step in the design of a mbosimulation involves identifying the primary causes for concern about the tribocomponent's behaviour. Concern may have arisen during the design review of a new, yet untested machine, or by a documented series of functional problems in the field. Perhaps the problem is merely a perceived one based on surface appearance and not with a measured loss of function. Understanding the origins of concern over the problem can better focus the initial approach to its solution.

Several formal approaches have been developed to define the attributes of tribosystems.7~* These offer varying degrees of detail and are somewhat different in structure. In the present discussion, we shall consider sixteen elements of tribosystem analysis. "lese are listed in Table 1 and can be subdivided into five sets. Elements 1 and 2 concern the macrogeometry of the contacting surfaces. Elements 3-5 affect the manner in which forces are transmitted in the vicinity of the tribowntacts. Elements 6 9 concern the interfacial environment. Temperature includes both frictional heating effects and the temperature of the surroundings. Elements 10 and 11 are associated with the contacting materials. Element 10 includes the composition, microstructure, processing conditions, and surface finishing of the tribomaterials. Element 11 has implications for such aspects as running-in, the operable lubrication regime, and the cost of fabricating the components. Elements 12-16 better define the problem in terms of how the failure is determined and what kinds of prior observations have been made. They also address the issue of whether additional requirements on the materials, beyond tribological concerns, must be met.

3

Despite their separate listing in Table 1, the elements of tribosystem analysis are not independent of one another. For example, the contact geometry, size, and applied load m needed to ascertain the contact pressure. The lubrication regime and its stability are determined by several of the elements. Thus, interactions of the relevant tribosystem elements must be considered.

Table 1. .The Elements of Tribosystem Analysis

16 I Failure history and related obsetvations I Information for mbosystem analysis is gathered from a variety of sources: operating records,

specification drawings, verbal reports, and interviews. It is not uncommon, however, to discover that no one really knows the precise operating conditions of the subject component(s). Depending on the type of component, even such things as contact pressure or normal force on the surface may not be accurately known. Perhaps there is little or no information on exactly how the component was used or how it was maintained. The latter is commonplace for builders of processing equipment who are faced with customer complaints about the wear of their equipment, but are unable to extract exact information from chose customers on the relevant conditions of use. Certainly, it is desirable but not usually possible, to instrument and monitor operating machines or

4

to develop detailed computer simulations. Desirable as it might be to begin with all the i n f o d o n contained in Table 1, one is much more likely to be faced with the challenge of developing a tribosimulation for an incompletelydefined tribosystem

One should next identify which types of surface phenomena are of primary importance. For example, wear or surface damage of several types can be p e n t at Merent areas of the same part, and some of these may not be critical to part performance. The author has had occasion to examine the wear of metal-coated extrusion machine screws and to design an appropriate tribosimulation. Such screws turn slowly (e.g., tens of revolutions per minute), forcing hot, molten plastic through the bore of the extrusion machine barrel and through an orifice at the outlet end. On a single screw, it is common to see a variety of wear types at different locations: polishing abrasion, scoring, erosion, and scuffing as well as pitting due to corrosive chemicals in the plastic. Possible causes of some of the wear damage include abrasion by mineral Niers in the plastic or by wear debris particles produced on other components within the machine which worked their way downstream.9 In the present case, it was necessary to select the most performance-reducing form of wear on which to base the design of a meaningful laboratory-scale simulation, and to design the simulator for maximum flexibility in order to test a variety of potentially hannful conditions.

More than one type of screening test will probably be needed to assess a material's resistance to several wear modes. For example, Czichos, et al.10 demonstrated that four surface treatxknts for steel ranked in opposite order in their resistance to sliding and abrasive wear. Thus, confining screening tests to only one of the operative wear types would be inadvisable and potentially disastrous.

2.2 The Development of Metria

The next step is to develop one or more specific measures of friction, wear, or surface damage (Le., "metrics"). These will be used for judging the success of the simulation. Metria are such quantities as wear volume, friction coefficient, and the change in diametral bearing clearance per unit of machine operating time. Sometimes it is not possible to obtain quantitative metrics.. Instead, the success of the simulation could be judged by examining field components and comparing their surface features to those produced in laboratory tests. In that case, the metric is a visual inspection rather than a dimensional measurement. Other, more indirect measures include comparing the type of wear debris produced or, less commonly, comparing the subsurface microstructures which are induced by the surface contact history.ll Fischer12 has stated that it is essential to clearly identify the acting wear processes in order to select materials for an application. In sliding contact situations, it is helpful to examine both members of the contacting pair to confirm the identification of the wear modes.

5

Examples of both quantitative and qualitative metrics are given in Table 2. Quantitative memcs are preferable to qualitative metrics because the d e p of material suitability can be more precisely specified; however, the ability to define quantitative metries is related to the wear mode and to the uniformity of contact. For example, galling is a severe and crippling form of surface damage, yet it is difficult to measure quantitatively.13 Engineers have developed standard test methods14 and developed a "critical load for galling" criterion. This is an interesting metric because it combines quantitative measurements (critical load) and qualitative observations (whether or not galling has occurred at a given load). Similarly, scuffing is of great concern to manufacturers of internal combustion engines, yet it is difficult to measure since it tends to be localized and non-uniform. A standardized set of photographs or photomicrographs is an alternative for qualitatively assessing or ranking the severity of wear.

Table 2. Examples of Metrics for Use in Tribosimulation

Application wear of a hard coating on a antact surface

$ding wear of a lubricated ournal bearing

rolling element bearing

solid lubricant coating on a sliding surface

polymer bushing material

Quantitative Metrics Near rate of the coating; critical pressure or time for wear- through; percent of original coating remaining after specific exposure time; change in friction force; time until the acoustic emission increases war rate: change in clearance; temperature rise in the bearing; concentration of debris in the lubricant number of cycles to failure; c & i i bad to failure (step test); maximum ON number

average (and variation) of fridiion force versus time; remaining bebw a critical vabe of friction coefficient; time to bss of fundion; maximum temperature for acceptable life or friction coefficient wear rate; friction coefficient; PV-yimit; maximum operating temperature; surface rouqhness after exposure

QUalit8tiVe Metrics presence of wear-through; transfer of material to the counterface; type of microfractures in the coating; evidence for delamination

presence or absence of 'm'ld" or 'severe" wear; appearance of wear debris in the lubricant; noise or vibration excessive noise or vibration; presence of pits or spalis; comparison to photographs of various types of surface damage severity appearance of the coated surface; tendency to transfer to the counterface; manner of bss during wear

noise; surface appearance and presence or absence of deformation

6

The units which are selected to measure wear or surface damage should be meaningful in the context of the problem :it hand. For example, it might be possible to obtain a wear rate for mechanical face seal miterials in units of mm3/N-m; however, that parameter may be less meaningful to the manufacturer or user of the component than the leak rate in liters per hour. Many units for wear have been 1sed15 and each has its proper place.

Studies of the wear of railroad rails provide an example of the selection of the appropriate metrics. Data on the wear of railroad rails have been collected for many years, indicating that curved track sections experience the most severe wear. Clayton16 concluded that fundamental models for wear were not useful for solving the problem, and that laboratory testing was needed. The three conditions to be fulfilled before the laboratory data would be used to predict rail wear behavior were: (1) the wear mechanisms of the service components had to be idendfied as well as possible, (2) the laboratory test had to generate surface damage representative of the field components, and (3) the relative performance of the materials had to agree in laboratory and field results. Clayton stated: "Laboratory testing with different test machines over the last fifty years shows that even small changes in the operating environment can alter the wear mechanism, often with large adjustments in wear rate." Data obtained from an Amsler machine (roller-on-roller), designed in the late 1920s, agreed best with full-scale test track results (see Table 3). Three different wear regimes were identified: I. was a combination of oxidative and metallic flake formation, II was mainly metallic wear, and III involved a break-in period in which large debris fragments became embedded and caused abrasion of the opposing surfaces. Clayton's approach clearly exemplifies the process of identifying the primary cause(s) of concern, characterizing the material response, and developing useful metrics.

Table 3. Comparison of Laboratory Wear Rankings with

Test Track Results for Four Steels* (Laboratory tests: 1220 MPa contact pressure, 35% slide/roll ratio, no lubricant)

The foregoing example described a comprehensive study involving extensive data collection over a period of years. Such a data-rich situation is probably more the exception than the rule. In fact, many wear problems are identified with a relatively small amount of anecdotal or incomplete

7

information. Some of the information might be first-hand, through direct observations of wear or friction behaviour, but in other cases, awareness of a tribological problem may come from second- hand sources such as a customer complaint or the report of a wear failure by people who are not trained in tribology or failure analysis. Therefore, a method like tribosystem analysis is needed to better define the problem before meaics can be established.

2.3 Selection of the Test Methodology

The fact that friction and wear are aibosystem properties, rather than exclusively materials properties, makes the selection of testing methodology important. When friction researchers test different material combinations on the same mbometer, and using that data alone, develop models to predict frictional behavior, they implicitly ignore that point. Figure 1 shows that the friction coefficients for AIS1 52100 steel sliding on Ni3Al alloy varied by a factor of nearly eight on three different testing machines, and that even within the same testing machine, the fiction coefficient was loaddependent.17 These differences were due to the changing relative role of ambient surface films, third bodies, and the ploughing component of fiction. Varying the test machine stiffness characteristics while keeping the materials pairing constant has been used by mechanical engineers to evaluate such things as the effects of system stiffness and vibrations on friction and wear.'*

DIN standard 50 322,'g describes various levels of simulation, ranging from field trials to simple laboratory tests. Likewise, the test methods used for the simulation can be systematically characterized. For example, ASTM has produced a standard guide for the presentation of sliding wear data*O. Seven categories of entries comprise that guide. They are test identification, test type, test conditions, material definitions, specimen identifications, test results, and documentation. Despite its detail, it does not address all the issues of tribosystem analysis.

As the test method approaches the real operating conditions more closely, running the tests

usually becomes more complicated and expensive, and the results may become harder to interpret. However, even when one uses simple tests, there may be difficulties in selecting the appropriate one. For example, Budinski21 compand four different methods for measuring the sliding friction of polyester films against stainless steel in order to obtain infomation for their manufacture and handling. The first method was a sled test in which a small flat steel specimen was pulled on the film. The second was a pin-on-disk test using the film as the disk material. The third was a capstan test in which the film was slid over a steel cylinder (capstan). The fourth was a flat-on- drum test in which a specimen of the Nm was pressed tangentially against a rotating stainless steel drum. Friction force traces were recorded for each testing system and exhibited differing degrees of fluctuations and average friction coefficients. Fig. 2 summarizes Budinski's results, showing the average and f 2 standard deviations from the average value for each test method. The pin-on-

8

disk test gave the least variability in results, but it also damaged the films during the test, making it unsuitable. The drum-type test also tended to damage the films. Budinski found that the various tests may each be useful under certain circumstances of contact pressure and sliding speed, but no one test was completely satisfactory in all respects.

A first-order simulation is one used primarily for preliminary material or lubricant screening. It should contain only the most basic elements of simulation, yet provide a simple and straightforward means to obtain quantitative, meaningful data. Sometimes, a first-order simulation is sufficient to solve the problem. Other times, more exhaustive trials of the leading materials or lubricants in closer-to-actual conditions are needed. The following four factors should be taken into account when designing a successful first-order laboratory simulation:

1. Contact macrogeometry and the characteristics of relative motion 2. The pressure - velocity relationships 3. Thermal and chemical factors (including type of lubrication) 4. The participation of third-bodies

Inattention to any of these factors could seriously reduce the effectiveness of the simulation. Factors that go beyond those of first-order simulations include matching the contact vibrational characteristics, contact stiffness, heat flow paths, and the thermal-mechanical interactions with the surroundings that occur under field conditions.

In sliding contacts, test geometry selection should consider whether the contact is conformal, such as a brake shoe on a drum, or non-conformal, such as a sphere on a flat surface. Contact conformity affects the distribution of lubricant in the interface, heat flow, and die movements of third-bodies. The speed constancy and reversibility of direction of motion has an effect on at least three related mbosystem elements: the regime of lubrication in the contact, the texturing of the surface and subsurface of the material, and the influence of third-bodies. Contact geometry and motion characteristics also af€ect heat retention and dissipation in the vicinity of the contact. It should not be surprising, therefore, to observe differences between unidirectional and reciprocating sliding friction and wear even for the same material couple and macrogeometry (e.g., see Table 4). In other forms of wear, such as erosive wear, relative motion considerations involve such things as the angle of impingement, the particle shape and rebound characteristics, particle flux screening, and the particle velocity distribution.

Simulators have been commonly used for friction and wear tests of internal combustion engine part~.22,=,2~ Piston ring-groove wear is one example in which the type of motion is complex and affects the localized wear processes. Wear of the uppermost piston ring groove in internal combustion engines can result in excessive crevice volumes which trap and expel unburned

9

fuel, thereby increasing engine hydrocarbon emissions. The motions of the piston ring in its groove are complex.s,26 There is radial fretting in-and-out of the groove, impact on the top and bottom groove surfaces, .ilting (rocking), and circumferential ring rotation (Fig. 3). The question becomes one of decidin,; which motions or combinations thereof are responsible for the most insidious forms of ring groove wear, and how these motions can best be incorporated into a tribosimulation. Bialo, et al.2' used low-amplitude sliding contact in kerosene at room temperature in their studies. WackerF8 on the other hand, took a more comprehensive approach, trying to simulate as many aspects of the ring motion as possible. Different engines behave differently in terms of the prevalence of such features as ring-groove bell-mouthing or lip-recession, etc., so that the type of tribosimulation should be tailored to replicating the wear mode which occurs within the specific engine of interest. Whether it is possible to reduce the complexity of the motion and yet still produce the proper form of wear must be determined in preliminary, "scoping" trials with the proposed test apparatus.

Table 4. Effects of Sliding Motion on the Friction and Wear of SiIicon Nitride

(Type NBD 100 silicon nitride ball and polished flat, 0.925 mm diameter, 10 N normal force, average speed 0.2 m/s, average data for 2 tests per condition)

Contact pressure and sliding velocity are two of the most critical aspects of simulation design. They are treated together because their affects on wear are often interrelated. If a given pressure p is applied to a contact which is undergoing relative motion with velocity v, then the energy input rate to the surface, and the temperature rise of the contact, are proportional topv. Energy inputs and outputs to tribosy stems were discussed systematically by Czichos.29 Frictional energy can produce heat, sound, or fracture with the subsequent creation of new surfaces (debris). Thus, it makes sense to consider matching the contact pressure, the relative velocity, and consequently the pv product, of the application. In fact, the pv product is often specified when selecting polymeric bearing materials.30 and engineering design diagrams which plot the locus of the pv limit across various pressure and velocity ranges have been developed for material selection.

10

The temperature and surrounding environment of the tribological contact are sometimes important elements in achieving successful simulations. For example, whether the t e m p m of the contacting surfaces is above or below the boiling point of water can affect the adsorption of moisture, a key factor in both friction and wear of many materials, including metals and ceramics. In addition to its effects on water and other adsorbates, temperature can affect the shear strength and tribochemistry of the materials at the contact surface.

Wear and corrosion have been shown to have synergistic effects. In fact, ASTM test methods have been developed specifically to address this point.31 Grobler and Mosten?* for example, conducted abrasive wear tests with corrosive solutions to help Screen and develop new alloys for the mining industry.

The type and characteristics of third-bodies can have an important influence on the success of a simulation. For example, the use of carefully-characterized used engine oil has been shown to produce a better simulation of the wear of candidate ring and liner materials than the use of fresh oil. In fact, many practical lubricated wear problems are the result of the contamination of oils and other fluids by third bodies. For example, studies of the abrasive particle contamination of engine oils of varying viscosities showed that the abrasive wear rates of piston rings depended on the size of the abrasive, reaching a maximum at 20 pm within the particle size range of 5-65 pm.33

The characteristics of third bodies are obviously important when developing abrasive and erosive wear simulations. Not only size, but the shapes of particles can affect abrasive wear rate, as demonstrated by SwansonM in three-body abrasion tests of steels with both sharp and rounded silica grains. Mechanical properties of particles are another aspect of third-body effects. A recent paper by Shipway and Hutchings35 clearly established the importance of the relative hardness of the erodent, with respect to the target material, on erosion rate.

2.4 Selecting or Developing a Testing System

After analyzing the tribosystem, developing a set of metrics, and isolating the testing conditions, the testing machine is selected. V ~ i t i k ~ ~ has developed one approach for matching the application characteristics to those of the testing machine. The Tribological Aspect Number (TAN Code) consists of four elements, each of which has several choices: (1) velocity characteristics (4 choices), (2) contact area Characteristics (8 choices), (3) contact pressure characteristics (3 choices), and (4) entry angle characteristics (10 choices). While these help establish the mechanical characteristics, they do not explicitly address the thermal and chemical aspects of the simulation.

Ideally, an existing testing machine or established method can be used, but the specific details of the problem at hand may require modifications to an existing method or that a new

11

machine or method be developed. Perhaps, an existing system can be modified to encompass all key elements of the simulation.

Under the auspices of the Versailles Project on Advanced Materials and Standards, a compilation of international friction and wear testing standards was prepared.37 It contains a listing and index to hundreds of friction, lubrication, and wear test methods from seven participating countries. ASTM alone had 115 standards, practices, and guidelines listed. The advantages of using published standards wherever possible are that the repeatability and reproducibility of the methodology has generally been verified by extensive testing, and data are available with which to compare your results. A few of the published standards were developed on the basis of extensive international interlaboratory testing programs, providing a degree of confidence in the precision and accuracy of the data which would be nearly impossible to obtain at a single laboratory.38 Some ASTM published standards involve simple geometries, such as the ASTM G-77 block-on-ring test or the ASTM G-99 pin-on-disk test. Others are designed for specific applications such as ASTM D-821 "Methods for Evaluating Degree of Abrasion, Erosion, or a Combination of Both in Road Service Tests of Traffic Paint."

U.S. Military Specifications also contain standard procedures for wear testing with very specific metrics. For example, MIL-R-46762, as described by Wood and Taylor?9 specifies wear criteria for road tires as follows: "The average height of the rubber chevrons or pads at the end of the road test shall not be less than 50 percent of their original height. Not more than 30 percent of the number of shoes or pads shall lose more than 3 square inches of rubber bearing area, and the original total bearing area of the rubber chevrons or pads shall be reduced by not more than 7 percent." Standard test methods can also be modified to more closely approach the actual component conditions. For example, Perez et a1.N used a modified four-ball wear test method to obtain componentcorrelated data on the wear of steels in various hydraulic fluids for pumps. If used frequently and successfully, such modified procedures become de facto standards themselves.

If no standard test method or suitable testing machine can be found, one confronts the task of designing and constructing a new wear or friction simulator. The disadvantages of this approach are (1) there are no pre-existing data with which to directly compare results, and (2) any new testing machine needs to be characterized to determine the peculiaxities of its behavior and the repeatability of its results.

The importance of understanding the operating characteristics of a customdesigned testing machine, even one as simple as a pin-ondisk tribometer, is illustrated in Fig. 4. In that case, a 9.525 mm diameter silicon nitride ball was slid on a Ti hard-coated tungsten carbide disk under a normal force of 5.0 N and a sliding speed of 0.5 m/s in room temperature air. The friction force trace in Fig. 4 (a) was obtained using a ball holder whose design did not fix its mounting shaft

12

tightly enough, resulting in a Certain amount of wobble and a continually-rising trend in friction. When replaced with a pin holder of alternate design, the friction trace for the same two specimen materials exhibited a Merent shape and reached steady-state with a smaller amplitude of variation in the friction force (Fig. w)). The mechanical design of wear testing machines goes beyond the scope of this papeG however, it is advisable to allow sufficient flexibility in the design so that a range of operating conditions, containing within them those required by the current work, is provided.

Just as small changes in the design of testing machines can produce significantly different test results, so can the specimen mounting procedure. In line contact tests, such as the block-on- ring test, small specimen alignment errors can significantly affect the repeatability of tests, the amount of wear, and the frictional running-in characteristics. The importance of assuring that the specimen is properly aligned is illustrated fur aluminum bronze-on-steel block-on-ring tests in Fig. 5, adapted from Blau and Whitenton.41 As the tilt of the block was intentionally increased, the running-in period shortens and the total wear recorded increases. For a block tilt of only 0.84 degrees, the total wear of the block specimen was 2.6 times higher than when it was properly aligned. Note that the steady-state wear rates (slopes) of all tests, after running-in, were quite similar indicating that once running-in had occurred, wear progressed approximately linearly.

There are two popular approaches for selecting the operating conditions: (1) the worst-case (or "factor of safety") scenario and, (2) the nominal operations scenario. In the former approach, one tries to determine the most Severe conditions under which the subject component must operate and conduct tests under conditions which are equal to or greater than those. An even more conservative approach is to apply an engineering factor of safety which increases the severity of conditions still further. If the materials pass such a test repeatedly, they are expected to perfm without pmblems in the actual, albeit less severe environment of the application.

One drawback of the worst-case approach is that the materials may tend to wear in a way that is uncharacteristic of the application, making it impossible to compare worn features to the type of surface damage observed on nody-operating components. The same consideration is true for so-called 'accelerated' testing strategies which apply extreme o p t i n g conditions. A more straightfoward approach is to operate the simulation at the middle or at the upper margin of the range of normal operating conditions (i.e., the "nominal operations scenario"). In this way, the metrics which were established based on the actual component characteristics are mre straightfornard to compare.

Accelerated tests, as noted above, must be used with due caution, but they can be quite informative if designed and applied properly. Furrest et al.a used the established relationship between the grit size of an abrasive and the wear rate to estimate the wear of magnetic reuxding head materials in a video tape system. By using larger grit sizes, wear was accelerated by up to a

13

factor of four, leading to faster screening of candidate materials. The effects of abrasive particle shape changes could likewise be studied.

As a final point, there are statistical tools available to help design sets of simulative experiments in which specific combinations of operating variables can be investigated. Such tools can identQ which variables and their combinations seem to have the most influence on material performance. In particular, the methodology developed by T a g ~ c h i ~ ~ in the 1980's is receiving widespread attention. This type of approach has been used by Mok and G ~ r m a n ~ to investigate the effects of five factors, at two levels, on the wear of rubber materials for mud pumps. The dedicated simulator they constructed was based on an oscillating design and allowed for the introduction of flowing mud at the desired temperature. Comprehensive analysis of the data was coupled with observations of wear surface features not only to identify significant wear-influencing variables, like load and sand concentration, but also to better understand the material responses which accompanied these phenomena.

2.5 Validating the Simulation

If the memcs have been well defined and the simulation developed with them in mind, the process of validating a simulation becomes a relatively straightforward one; namely, comparing, item by item, the results of the laboratory test with the attributes of the actual components. To illustrate this approach, consider the hypothetical example given in Table 5. A plain bearing might experience a certain amount of wear in 100 hours of operation, and a block-on-ring test might be selected as a laboratory simulation. Both quantitative and qualitative metrics are listed in the table. The quantitative metrics for wear do not agree well in this example, but most of the observed features on the surfaces of parts removed from service have been reproduced in the simulation. Mixed results like these may not allow one to clearly establish the success of the simulation unless one or two key metrics in the list are seen to be more important than the others, and are used as the primary measure of simulation success.

Work by Zhou et alp5 provides another example of validating a simulation by the observation of surface features. A single-wire, bench-scale laboratory fretting test was developed to simulate the contact conditions a wire in a many-strand cable would experience. This test produced the same form of surface features as observed on actual cables: similar fretting zone size, similar delamination features, similar crack nucleation modes, and number of cycles to nucleate cracks.

The relative ranking method is a more convincing validation than confirmation by the observation of characteristic surface features alone. In this approach, one determines to what extent a Series of materials ranks in performance in the same order in field and laboratory trials.

14

The earlier example of p a i l wear illustrates that approach flabk 3). Ihe numerical values, such as wear rates or threshold mditions for the onset of surface damage, associated with the rankings may differ in magnitude: between field trials and Iaboratory tests, but at least the order of merit of various material combinations or lubricants is similar. Of course, it is not possible to use the relative ranking method when only one materials couple has been used in practice.

Table 5. Validation Table for a Hypothetical Tribosimulation

(after 100 hours of exposure)

In summary, the following attributes, listed in descending order or desirability, can be used to validate a simulation:

1) Numerical wear rates, characteristics of surface damage, and friction coefficients

2) The relative ranking and the magnitude of the differences between several candidate

3) The relative ranking (but not necessarily the magnitude of the difference) of several

4) The wear type, wear particle characteristics, or surface damage features have nearly-

produced in the simulation correspond directly with those observed in the application

materials is the same in the test and the application

candidate materids is the same in the test and the application

identical characteristics in the simulation and in the application.

15

3.0 Simulation Hierarchies and Supporting Tests

The value of laboratory-scale tribosimulations depends on their ability to provide necessity data for supporting decisions on the selection of materials, surface treatments, and lubricants. It may be cost-effective to Screen a large number of candidate materials using a simple, first-order simulation and then conduct more realistic tests on a smaller number of leading candidates under conditions which more accurately approach the field conditions. As noted earlier, German DIN standard 50 322 lists six levels of testing which vary in simplicity and correspondence to actual component conditions.

Sometimes, conducting simulative weat and friction tests alone are not sufficient to address the problem at hand. For example, Santner and Meier zu Kock& published a recent article on the subject of tribosimulation in which three examples were provided to illustrate their points. One in particular involved graphite-filled PTFE slider bearings for a tumble drier. There was a question regarding whether the supplier had changed anything when bearings began to perfom poorly. Simple pin-on-disk tests were run using pins of the two materials. Results showed differences between the behavior of the two suspect lots of polymer composites both in terms of fiction coefficients and in terms of the wear surfaces of the disk specimens (steel cut from the tumbler drums). The composite performing well in sewice had frne wear marks and transfer of the F"FE while the poorly-performing lot showed abrasion marks on the disks similar to those observed on the drum surfaces. absorption tests were needed in addition to the tribotests to better identify the source of the problem. It was found that there was no clear difference in chemical compositions of "good" and "bad" materials, and that major evidence for poor performance seemed to be a difference in the adhesion between the graphite filler and the matrix. While the% was no compositional basii for screening out bad materials, a simple friction test could be used by the tumble drier manufacturer prior to installing the bearing materials. Using a combination of test methods may therefore pave

Chemical analysis, X-ray diffraction, differential thermal analysis, and water

the way to an answer if evidence from any single test method is not in itself sufficient.

4.0 Summary

The solution of friction and wear problems by simulative testing involves five steps: (1) developing a clear understanding of the tribosystem involved, (2) developing metrics by which the problem can be quantified, (3) selecting the necessary conditions for the simulation, (4) selecting or developing a testing system and pmedure to jwvide those necessary conditions, and (5) validating the simulation based on the metfics established earlier. Tribosystem analysis is the first step in defining the problem. It is important, at minimum, to simulate the macrogeometry of

16

contact, the type of motion, the contact pressure and velocity, the thermal-chemical environment (temperature, lubricant, and gaseous environment), and the possible influence of third-bodies, but it should not be necessary to simulate every aspect of the application. Validation of the tribosimulation can be based on any of several criteria: friction coefficient values or wear rates, relative rankings of specific material combinations, the observation of key surface features, of the wear debris characteristics. It may be necessary to make adjustments in the testing procedure to satisfy the established metrics. The worst-case scenario and the normal operating condition scenario are options for developing friction and wear simulations. More than one type of screening test may be required to evaluate materials or surface treatments depending on b c e requirements and the required balance of properties. Other types of materials analysis, in addition to tribotesting alone, may be needed to obtain a satisfactory solution for some types of friction and wear problems.

Acknowledgments

The author would like to acknowledge the support of the U.S. Department of Energy, Assistant Secretary for Energy Efficiency and Renewable Energy, office of Transportation Technologies, High Temperature Materials Laboratory User Program, under contract DE-AC05-960R22464 with Lockheed Martin Energy Research Ccxp.

DISCLAIMER

This report was prepared as an amount of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsi- bility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Refer- ence herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, mom- mendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

-~

17

1. The influence of the tribosystem is apparent when measuring the friction coefficient of the same two materials in three different testing machines.

2. The range and average values for the sliding friction coefficient of polyester films against stainless steel varied significantly among four testing machines.

3. Several types of motions are present for a piston ring in its groove. The appropriate motion needed for the simulation is determined by examining piston cross-sections and the contact surfaces of used rings and ring grooves.

4. Friction versus time behavior for silicon nitride sliding on a T i hardcoated tungsten carbide disk under a normal force of 5.0 N and a sliding speed of 0.5 1 4 s in room temperature air. (a) effects of slight rocking of the ball specimen holder, (b) same experiment with a rigidly-held ball specimen.

5. Suppression of the wear-in perid and the resultant increases in cumulative wear amount resulting from intentional misalignment of the block specimen in block-on-ring tests.

18

References

Dowson, D., m o r y of Trib&gy, Longman, 1979. ASM Handbook, Frictkon, Lubrication and Wear Technology, Vol. 18, ASM International,

-

Materials Park, Ohio, 1992.

(1984), III, CRC Press, Boca Raton, Florida,.

New York, 1980.

3 Booser, R., ed., Handbook of Lubrication: Theory and Practice of Tribology, Vol. I (1983), 11

Peterson, M.B. and Winer, W. O., ed., Wear Control Handbook, Amer. Sa. of Mech. Engr.,

5 Bhushan, B. and Gupta, B. K., Handbook of Tribology, McGraw-Hill Pub., New York, 1991. 6 Godet, M., Berthe, D., Dalmey, G., Flamand, L., Floquet, A., Gadallah, N. and Play, D,

"Tribo-Testing," in Tribological Technology, Vol. 11, ed. P. B. Senholzi, Martinus Nijhoff Pub., The Hague, 1982, pp. 525-609.

Wear Technology, Vol. 18, ASM International, Materials Park, Ohio, 1992, pp. 473-479.

Field of Wear," Deutsches Institut fur Normung e.v., Berlin, Gexmany, 1979. Blau, P. J., "Tribocommunication," in

Park Ridge, NJ, pp. 418-430.

. . . . Czichos, H., "Basic Tribological Parameters," in ASM Handbook,

8 DIN standard 50320, "Wear, Terms; System Analysis of Wear Processes; Classification of the

. . ' , Noyes Pub., . . Wear Tr- of

lo Czichos, H., Trib0l~g.y: A S- Elsevier Pub., Lausanne, 1978, pp. 322-325. 11 Xu, L., Clough, S., Howard, P. and St. John, D. "Laboratory assessment of the effect of

white layers on wear resistance of digger teeth,"Weur, Vol. 181-183, 1995, pp. 112-117. 12 Fischer, A., "Well-founded selection of materials for improved wear resistance," Wear, Vol.

l3 Peterson, M. B., Bhansali, K. J., Whitenton, E. P. and Ives, L. K.,"Galling Wear of Metals,"

14 ASTM G-98-9 1, Standard Test Method for Galling Resistance of Materials, ASTM Ann. Book

15 Blau, P. J., "The Units of Wear -- Revisited," Lubrication Engineering, Vol. 45 (10) pp. 609-

16 Clayton, P. "Predicting the wear of rails on curves from laboratory data," Wear, Vol. 181-183,

l7 Blau, P. J., "Scale Effects in Friction: An experimental study," in Fundamentals of Friction,

194, pp. 238-245, 1996.

hoc. Wear of Materials, ASME, New York, 1985, pp. 293-301 .

of Standards, Vol. 03.02, pp. 382-384.

614.

1995, pp. 11-19.

ed. I. L. Singer and H. M. Pollock, Kluwer Pub., Dordrecht, 1992,523-534.

19

l8 Aronov, V., DSouza, A. F., Kalpakjian, and Shanef, I., "Experimental investigation of the effect of system regidity on wear and friction-induced vibrations," J. of Lubr. Tech., Vol. 105, pp. 206-21 1.

l9 DM std 50 322, .... Deutsches Institut fur Normung c.v., Berlin, Germany, ASTM G-118-93, "Standard Guide for Recommended Data Format of Sliding Wear Test Data

Suitable for Databases," ASTM Book of Standards, Vol. 03.02, West Conshohocken, Pennsylvania, 1996, pp. 489-492.

ASME, New York, 1989, pp. 459-468.

Combustion Engine Materials Tribology Problems," in Tribolow: Wear Test Selectionfqt -, ed. Ruff, A. W. and Bayer, R. G., ASTM Spec. Tech. Pub. 1199, West Conshohocken, Pennesylvania, 1993, pp. 133-148.

Patterson, D. J., Hill, S. H., and Tung, S. C, "Bench Wear Testing of Engine Power Cylinder Components," Lubr. Engrg., February, 1993, pp. 89-95.

Malatesta, M. J., Barber, G. C., Larson, J. M., and Narsimhan, S. L., "Development of a Laboratory Bench Test to Simulate Seat Wear of Engine Poppet Valves," Tnbobgy Trans.,

21 dinsk ski, K. G., "Friction of plastic films," pro^. International C O ~ L on wear of Materials,

Blau, P., "A Retrospective Survey of the Use of Laboratory Tests to Simulate Internal

Vol. 36 (4), 1993, pp. 627-632. 25 Ma, M-T., Smith, E. H., and Shenington, I., "A three-dimensional analysis of piston ring

26 Ma, M-T., Smith, E. H., and Sherrington, I., "A three-dimensional analysis of piston ring lubrication - Part 1: modeling," J. of Engineering Tribology, Vol. 209, 1995 (Jl), pp. 1-14.

lubrication - Part 2: sensitivity analysis," 1. of Engineering Tribology, Vol. 209,1995 (Jl), pp. 15-28.

2' Bialo, D., Duszczyk, J., de Gee, A. W. J., van Heijningen, G. J. J., and Korevaar, B. M., "Friction and Wear Behaviour of Cast and Sintered AI-Si Alloys Under Conditions of Oscillating Contact.," Wear, Vol. 141,1991, pp. 291-309.

Wacker, G., "The Use of a Testing Machine for Simulating Piston Ring Groove Wear," in ts of W u ed. Hornbogen, E. and Zum-Gahr, K. H., Deutsche Gesell.

fur Metalkunde, 1981, pp. 247-267. 29 Czichos, H., Ref. 10, pp. 33-42.

Dixon Bearing Manual, Furon Advanced Polymers Division, Bristol, Rhode Island, 1993,23 PP.

31 ASTM G-119-93, "Standard Guide for Determining Synergism Between Wear and Corrosion," Annual Book of Standards, Vol. 03.02, ASTM, West Conshohocken, Pennsylvania, 1996, pp. 493-498.

32 Gmblcr, P. E. and Mostert, R J., "Experience in the Laboratory and Commercial Development of Abrasion-Corrosion Resistant Steels for the Mining Industry," Roc. Intern. Conf. on Wear of Materials, ASME, New York, 1989, pp. 289-295.

33 ASM Metals Handbook, "Service Characteristics of Carbon and Alloy Steels: Wear Resistance", ASM International, Materials Park, Ohio, 1985, p. 465.

34 Swanson, P. A. and Klann, R. W., "Abrasive Wear Studies Using the Wet Sand and Dry Sand Rubber Wheel Tests," Proc. Intern. Conf. on Wear of Materials, ASME, New York, pp. 379-381, 1981.

35 Shipway, P. H. and Hutchings, I., "The role of particle properties in the erosion of brittle

36 Voitik, R., 'The Tribological Aspect Number," in Wear Test Test Selection for Design and materials," Wear, Vol. 193, pp. 105-113, 1996.

Application," ed. A. W. Ruff and R. G. Bayer, ASTM, Spec. Tech. Pub. 1199, Philadelphia, 1993, pp.

37 Blau, P. J. ed., A Compilation of International Standards for Friction and Wear Testing of Materials, Versailles Agreement on Advanced Materials and Standards (VAMAS), Technical Working Area 1, report No. 14, National Institute of Standards and Technology, Gaithersburg, MD, 1993,49 pp.

Methods, conducted a series of pin-on-disk tests on steel and ceramics during the 1980's. These resulted in several new ASTM and DIN standards.

39 Wood, F. W. and Taylor, I. W., "A Review of DOD Wear and Friction Tests for Rubber Products," in wear & Fnchon of EmomerS, ed. Denton, R. and Keshavan, M. K., ASTM Special Tech. Pub. 1145, Amer. SOC. for Testing and Materials, Philadelphia, 1992,

Perez, J. M., Hausen, R. C., and Klaus, E. E., "Comparative Evaluation of Several Hydraulic Fluids in Operational Equipment, A Full-S&e Pump Stand Test and the Four-Ball Wear Tester. Part n. Phosphate Esters, Glycols, and Mineral Oils," Lubrication Engr., Vol. 46

41 Blau, P. J., and Whitenton, E. P., "Effect of Flat-on-Ring Alignment on the Sliding Friction

uI Versailles Project on Advanced Materials and Standards, Technical Working Area 1, Wear Test

. .

pp. 12-29.

(4), pp. 249-255, 1990.

Break-in Curves for Aluminum Bronze on 52100 steel," Wear, Vol. 94, 1984, p. 201. Forrest, D., Matsuotka, K,, Tse, M.-K, and Rabinowicz, E., '*Accelerated wear testing using

the grit size effect,'' Wear, Vol. 162-164, 1993, pp. 126-131. 43 Taguchi, G., "System of experimental design, Krans, New York, 1987.

21

~

4 Mok, S, H. and Gonnan, I). G., "Using Taguchi experimental design to investigate o p t i n g variables that significantly affect wear in mud pumps," J. of Engineenkg Trib., VoI. 209, 1995, pp. 29-40.

4s Zhou, 2. R., Goudreau, S., Fiset, M. and cardou, A.,"Single wire fretting tests for electrical conductor bending fatigue evaluation," Wear, Vol. 181-183,1995, pp. 537-543.

4 Santnex, E. and Meier zu Kocker, G."Utility and limitations of mbosimulation for quality control and material preselection," Wear, Vol. 181-183, 1995, pp. 350-359.

22

i

V i

0 QD CO P r 0 0 q o 0 . . .

C'

r 0

0 0 0

.

. I

m

0 to v) d (3 T-

0 0 0 0 0 0 0 0 0 . .

circumferential rotatlon 1 (possibly intermittent)

fretting In the plane r f thn nrnnva . ... " =.ww...

t impact on the

upper and lower lands rocking (twisting) motion 7 -

some Iocatlons for ring-groove wear and surface damage

4-

(a) LOOSE BALL

. . . . . . . . . . .

SLIDING TIME w 608

@) TIGHTLY-HELD BALL SPECIMEN HOLDER

SLIDING TIME

cc E E 2 0

0 w u)

m 0 K 0 U U w

Y 0 0 A

Y

F

3,

3

a

CDA 288 AI-bronze block on 52100 ring 10 N, 0.2 m/s, argon after Blau and Whitenton (1 984)

f .

slope

0 0

# #

0 0

P 0.280 0 4 e

# 0

0

2 5 0 I

2 0 0 -

1 5 0 -

100 -

, 1 1 1

0 1 0 0 2 0 0 3 0 0 400 5 0 0 6 0 0 7 0 0 SLIDING DISTANCE (m)