Dening Field-Compatible Metrics Selecting or Constructing the Test Apparatus Conducting Baseline Testing Using Established Metrics and Rening Metrics as Needed Case Studies
An Oil Pump Gear Set with Several Wear Modes Wear of Gravure Rollers on Doctor Blades Scoring of Spur Gears Wear of Plastic Parts in an Optical Disk Drive Wear of Rotary Slitter Knife Blades Erosive Wear of Piping
Peter J. Blau
Tribomaterials Investigative Systems
14.8
Conclusions
14.1 Introduction
The selection of lubricants, materials, or surface treatments for friction and wear-critical applications often involves validation or screening tests before nal decisions are made. Testing is particularly valuable during the development of new machines for which operating conditions are much different than existing designs. An example of the latter might be a new design that cannot use off-the-shelf bearings or gears because the temperatures are too high or the surrounding environments are too corrosive. The steps involved in developing tribosimulations of current or newly designed systems are, with only one significant exception, essentially the same. The exception is that for an existing friction of wear problem, there is prior experience in the response of the materials and lubricants to the operating conditions. In a new design, however, there may be no direct prior experience in the behavior of candidate materials, although there may be some relevant experience from machinery of a similar kind. There are two types of tribosimulations: computer simulations and physical simulations. The computer simulation uses a virtual mechanical assembly that consists of several components dened in terms of a set of properties, spatial relationships, and assumed rules of interaction. Known or estimated properties of the materials and/or lubricants are provided to the model, and the expected responses of the virtual tribosystem to such variables as load cycles, deections, temperature excursions, etc., are calculated. Such programs have been prepared by both academic researchers and industry engineers for tribological components like bearings, face seals, brakes, and gears. Component designers have also developed sophisticated design tools for automotive valve trains, engines, and pumping systems as well. The second type of tribosimulation is the physical simulation. Here, materials and lubricants are screened in an apparatus that is intended to provide the essential operating characteristics of the intended
application. An additional use of virtual and physical simulations is to observe how a certain material or lubricant might react in a situation when the opportunity to observe the material or lubricant is not practical. For example, bearings for use in unmanned orbiting satellites might be observed in an ultrahigh vacuum, space-like environmental chamber created in the laboratory for that purpose. This article focuses on the development and use of physical tribosimulations. It describes the process involved in developing and validating laboratory bench-scale simulations for friction and wear-critical machine components. The steps in developing a physical tribosimulation are similar in some respects to those used to set up a computer simulation. They are: 1. 2. 3. 4. 5. 6. 7. Dene the nature of the friction or wear problem to be investigated Select a scale of simulation Dene eld-compatible metrics to use to assess the results of the simulation Select or build the apparatus (the model, in the case of the computer simulation) Conduct baseline tests to establish the repeatability and characteristics of the method Analyze baseline test results using the established metrics Rene as needed to achieve an acceptable engineering condence level
TABLE 14.1
Element
Tribosystem open or closed? Macro contact and conformity Microcontact and surface nish Type of relative motion Speed of relative motion Contact load and pressure Temperature Contact environment and lubrication
In existing applications, identifying the type of wear involves examining eld-worn parts which have been protected from the environment after having been removed from the machine. Surfaces of worn parts which have stood unprotected for some time may be corroded and difcult to analyze. Other complications include removing surface deposits of degraded or heat-altered lubricants without destroying the most telling clues as to the dominant mode of wear damage. In new designs, the engineer is placed in a position of speculating what the environment of the tribosystem will be, and adjusting the simulation appropriately. Some guidance can, however, be obtained by analyzing existing systems with similar mechanical, thermal, and chemical aspects in the key areas of tribocontact. TriboSystem Analysis (TSA) [3] is a means to dene the operating conditions of a system to be simulated. It involves a systematic analysis, aspect-by-aspect, of the operating environment of the subject component(s). Table 14.1 lists key elements for analyzing a tribosystem. Often, not all of the operating parameters are known. Therefore it is doubly important to understand the characteristic wear features or other key aspects of the tribosystem that will help to dene metrics for the simulation (see Section 14.4). Understanding the properties and behavior of the currently used materials, or those used in an application which has key operating aspects in common with the component of interest, is important. Knowing what materials have and have not worked in the given application is equally valuable because it could save a great deal of time and effort. Simply asking the questions embodied in a thorough tribosystem analysis can go a long way toward solving the problem. Verifying the answers to those questions with a second opinion or a measurement can also be helpful. Sometimes people incorrectly assume that certain operating conditions exist. Tribosimulations sometimes need to be sensitive enough to distinguish between different variants of the same material. For example, there are many ways to heat-treat steels. A wear problem may occur if a component supplier changes the heat treatment, perhaps for reasons of economy, or changes the material supplier without notifying the customer (see the later example of gear scoring). The simulation in that case must be sensitive enough to detect the differences due to different heat treatments. Detecting
small variations in friction or wear using short-term laboratory tests can be challenging, but it can be facilitated through the use of statistical methods for data analysis. Therefore, multiple tests of each material or lubricant combination are needed to establish the characteristics of the laboratory test method; particularly, if the test method is new and has no historical record of performance. Using ASTM and other standardized tests offers the benet of having previous data available from tests conducted in the same manner.
FIGURE 14.1
Level 2 tribosimulations use subassemblies which are subjected to near-operating conditions. Examples include brake pads and rotor combinations on dynamometers, uid pumps in closed-loop pump test rigs, and jet engines in engine test cells. Nearly as expensive as Level 1 tests, Level 2 tests offer additional control of the externally applied test parameters. At the same time, fewer of the eld-associated effects on performance are faithfully simulated. For example, the effects of road-induced vibrations on piston motions, the vehicle-specic ow of air past brake components, and the introduction of environmental contaminants into wheel bearing grease may be omitted in subassembly tests. In dynamometer tests of break component materials it is common to apply a series of test stages in an attempt to simulate specic types of frictional phenomena, like fade effects at elevated temperatures. Even staged Level 2 tests as complex as these cannot totally simulate the full range of habits of individual drivers and driving conditions. In some cases, however, Level 2 tribosimulations can be very effective in screening materials or lubricants because the operating conditions of the system are more clearly known. For example, loopby-loop ballpoint pen testers can show how long the products will continue to write effectively and establish failure statistics for the entire pen, whose satisfactory performance depends on the ability of the point to deliver a clear, uniform line of writing uid to the paper. One tribosimulation area of particular medical interest is that of computerized hip and knee joint testing. Attention here is given to mimicking the forces, types of motion, and impact loads to which bioimplants are subjected. This subject area can make effective use of both physical and virtual component Level 2 simulations. The selection of the uids to simulate synovial uid and to correlate with clinical results is an important issue. Material swelling in situ, in the case of polymeric materials, and the role of debris particles as they interact with the soft tissues surrounding the joints, are also of interest. Level 3 tribosimulations involve test rigs designed to test specic components, like bearings and gears. For example, bearing test rigs have been successful in developing empirical design and selection guidelines for rolling element bearings of many kinds. Multiple-station rigs, automated to take data or to ascertain critical failure conditions, like excessive heat or vibration, can be run unattended, enabling the compilation of lifetime statistics and related performance data for consumer products or machine components. Level 4 tribosimulations involve test coupons of simple shapes. Examples include pin-on-disk tests, block-on-ring tests, four-ball lubricant tests, dry-sand-rubber-wheel abrasion tests, and vibratory cavitation tests. These tests are described elsewhere in this volume and in the wear testing literature [e.g., ASM (1992, 1997)]. Their usefulness is based on their ability to simulate the key contact conditions of the components of interest. For example, a cam roller follower in the engine of a certain diesel engine might be simulated by two disks turning at different speeds to impart a desired degree of slip to the
TABLE 14.2
Situation
Field component
Laboratory specimen
contact. Furthermore, the test disks could be supplied with a lubricant and heated to simulate engine conditions. The linkage between tribosimulation levels can be important to establish the validation of Level 3 and 4 tests as effective screening methods. For example, if a set of Level 4 rankings agrees with relative rankings of the same set of materials or lubricants in Level 3 tests, and the validity of Level 3 tests in a certain application has been conrmed, then the usefulness of Level 4 tests will be greatly extended.
14.6 Conducting Baseline Testing Using Established Metrics and Rening Metrics as Needed
Baseline tests with the materials and lubricants in current use, or believed to be the leading candidates for a new application, are helpful in establishing the repeatability and characteristics of the test method itself. It is worthwhile to conduct a series of replicated tests to establish the repeatability of the baseline conditions. Knowing this, the performance of other materials or lubricants can be analyzed to determine if it is statistically different from the normal scatter in the test results. One test per materials couple or set of test conditions does not provide very strong evidence on which to make engineering decisions. If wear is measured quantitatively, then the wear factors or other metrics from baseline tests can be used in the denominator of a gure of merit. For example, the lifetime of a baseline cutting tool material, expressed as the number of workpieces produced before tool replacement is required, is divided into the lifetime of another candidate tool material to give the relative lifetime of the candidate. The higher the number is above 1.0, the greater the wear benets of substitution, other factors being equal. An alternative engineering metric might be the tooling cost per unit part machined. Sometimes suitable quantitative metrics cannot be found. It may then be possible to compare the appearance of the test specimen to a worn part to establish the success of the simulation. An example is given in 14.7.1. It is likely that laboratory simulations will not be able to reproduce every aspect of the part operating conditions. Therefore, it is, in general, unreasonable to expect that precisely the same wear rates or metrics will be obtained in the laboratory and in the eld. From the standpoint of screening, however, it is very important that candidate materials rank in very much the same order in the eld and in the laboratory. Since there are many ways to measure wear in the laboratory (weight loss, wear prole, wear scar dimensions, etc.), there may be one metric that correlates better with the eld wear results than another. Therefore if one method of laboratory wear measurement does not correlate well with eld results, another may work better. The following section describes several laboratory simulations that used different kinds of metrics and testing procedures depending on the nature of the parts being simulated.
FIGURE 14.2
FIGURE 14.3
and then a composite rating for each couple was determined (the sum of the two specimen ratings). Each test was duplicated to establish the repeatability of the results, and to enhance the investigators condence in the differences between the wear ratings of different material couples. Figure 14.4 compares the wear seen on an actual part with that produced in laboratory experiments of several candidate alloys. Results from these two kinds of simulative tests, coupled with full-scale pump rig tests at a manufacturers facility, cost modeling, and alloy processing trials, were used to select the leading alloy and surface treatment for this application.
TABLE 14.3
Factor Light abrasion
Moderate abrasion Severe abrasion Light scufng Moderate scufng Scoring Pull-out Delamination Burning Severe metallic wear Microwelding Major transfer
excess coating material, and were experiencing unacceptable wear as a result. It was decided to try ionimplanting the surfaces of the rollers to improve their wear. ASTM standards G-99 (pin-on-disk test) was used to compare the implanted and unimplanted (current) materials. While the wear of the roller material was markedly improved, the wear of the doctor blade material increased to an unacceptable level. Therefore, it was decided that ion implantation would not be an acceptable solution in this case. While the pin-on-disk method was not an exact simulation of the doctor blade operating conditions, it was felt to be adequate to evaluate one potential solution for this wear problem, and to determine that alternative methods of surface engineering or materials substitution would be required.
FIGURE 14.4 (a) Wear features on a gear side face tested in a commercial producers testing facility under severe operating conditions. (bd) Different levels of wear damage observed on laboratory test specimens in the at-onat simulator test shown in Figure 14.3.
specimens. While not exhibiting the exact geometry of the application, the small, highly-loaded contact between test specimens contained enough of the essential elements of rotary slitter knife interactions to produce a useful screening test. Wear volume is used as the metric and is computed from the test materials densities and their weight losses. By examining a great deal of crossed-cylinders laboratory wear test data that would have been impractical to obtain on the production oor, it was discovered that at least one of the blades had to be composed of carbide material in order for the slitter knives to perform satisfactorily. Since more than one material combination with satisfactory wear rates was identied in the course of the testing campaign, it was possible to select the most affordable solution to the problem from among several alternatives.
14.8 Conclusions
The development of simulative friction and wear tests requires an interdisciplinary approach, beginning with a tribosystem analysis to dene the problem and to establish key metrics that can be used to test the validity of simulations. Laboratory simulations using either custom-designed apparatus or standard test methods can be successfully applied to save time and money in solving friction and wear problems. No single test method will solve all problems, and proper test selection is critical for success. Sometimes, more than one test method will be needed to establish an engineering solution; especially, if more than one form of wear or surface damage is present in the application of interest.
References
ASM Handbook (1992), Friction Lubrication and Wear Technology, 18, ASM International Materials Park, OH. ASM (1997), Source Book on Friction and Wear Testing, ASM International, Materials Park, OH. Blau, P.J. (1998), Development of bench-scale test methods for screening P/M aluminum alloys for wear resistance, in Powder Metallurgy Aluminum and Light Alloys for Automotive Applications, Jandeska, Jr., W.F. and Chernenkoff, R.A. (Eds.), Metal Powder Industries Federation, Princeton, NJ, 97. Blau, P.J. and Budinski, K.G. (1999), Use of ASTM standard wear tests for solving practical industrial wear problems, Wear, 225-229, 1159.