Tribological Investigation of the Polymer- Based Lubrication System Using a Laboratory Reciprocating Bench Test

Transcription

1 PEER-REVIEWED Published in STLE Tribology Transactions (Vol. 50, No. 4/October-December 2007, pp ). Presented at the World Tribology Congress in Washington, D.C., Sept , Final manuscript approved June 27, Review led by Dong Zhu. Copyright STLE Tribological Investigation of the Polymer- Based Lubrication System Using a Laboratory Reciprocating Bench Test Simon C. Tung (STLE fellow), General Motors R&D Center, Warren, Mich., Yong Huang, Clemson University, Dept. of Mechanical Engineering and Dennis C. Karczynski, General Motors Quality Network (STLE member), Detroit, Mich. Editor s Note: Product innovation is the result of bringing to life new technology, and it s usually tied to solving a problem, a need or a desire. If we look around the world today, we see that everyone is becoming more safety-, environmentally- and costconscious. In this month s Editor s Choice paper, authors Simon Tung, Yong Huang and Dennis Karczynski have constructed and evaluated an innovative polymer-based neardry lubricant delivery system that could replace oilflooded systems while meeting these concerns. Their research provides a guideline for possible applications in manufacturing processes. Product innovation isn t easy. And in any business it s innovation that distinguishes a leader from a follower. I encourage you to read this article and be informed. Dr. Maureen Hunter TLT editor Abstract Flooded oil lubrication systems are commonly utilized for many manufacturing processes. However, there is an increasing concern with the flooded lubrication systems due to their biological and environmental impact as well as their maintenance and disposal cost. Polymer-based material, a spongelike solid lubricant allowing the oil to weep out while in use, provides a promising alternative environmentally benign lubrication solution to satisfy the lubrication requirements for some manufacturing processes. In addition, it can minimize contamination of the coolant system, reduce waste processing and disposal costs, and alleviate the in-plant safety hazard from oil leakage on floor surfaces. The goal of this study is to develop and improve the current lubrication system by applying a polymer-based near-dry lubrication system in order to replace the oil flooded transfer line lubrication systems currently in production use. Using a laboratory reciprocating bench test based on the design of experiments, this study characterized the tribological performance of the polymer-based lubrication system under a variety of operating conditions, identified the main factor that impacts the tribological performance of the polymer-based lubrication system, and developed a model to predict the wear loss of the polymer-based lubricant plate. These results provide the guidelines for possible applications of the polymer-based lubrication system used in the manufacturing process. Introduction Flooded oil lubrication systems are commonly utilized for some manufacturing processes. A primary concern with the flooded lubrication system is the biological and environmental impact. In addition to possible fumes, smoke and odors, the fluid may also cause severe reactions on the skin and to various parts of the operator s body. Maintenance due to oil degradation and ultimate disposal also cause additional cost and reduce the manufacturing efficiency. Polymer-based material, a solid lubricant that is a mixture of polymers, oil, 1 and selected additives, whose matrix is in a sponge-like form allowing the oil to weep out while in use, provides a promising alternative environmentally benign lubrication solution to satisfy the requirements of some manufacturing processes. It minimizes contamination of the coolant system, reduces waste processing and disposal costs, and alleviates the in-plant safety hazard from oil leakage on floor surfaces. This study aims to develop and improve a polymer-based near-dry lubrication system to replace the oil flooded transfer line lubrication systems currently in use. In this study, the feasibility and durability of the proposed polymer-based lubrication system was investigated using a reciprocating bench durability test based on an experimental design. The experimental observations were further analyzed statistically. A weight loss model of the polymer-based lubricant plate was also developed, and the related worn surface characteristics were determined. 40 FEBRUARY 2008 TRIBOLOGY & LUBRICATION TECHNOLOGY

2 Experimental design Experimental setup To investigate the tribological performance of the proposed polymer-based lubrication system, a polymer-based leaded bronze plate (hardness HB 32.8) as shown in Figure 1 was used in this study. The polymer-based lubricant plate was 38.1 mm by 50.8 mm (1.5 inch x 2.0 inch), and it contained twelve polymer-impregnated holes. Inside each hole was a proprietary mixture of polymers, oil and selected additives, and the oil oozed out from the hole during the service period. The non-uniform hole distribution was expected to reduce the number of holes while achieving the same lubrication performance. The 83.8 mm by mm (3.3 inch x 5.2 inch) coupon was made of P20 steel with a hardness of HRc 55. The surface roughness (Ra) of the steel coupon that slid against the polymer-based plate was µm. To evaluate the tribological performance of the polymerbased system, a Cameron Plint high frequency and wear tester was modified to serve as a test bed as shown in Figure 2. This Cameron Plint tester provided: (1.) a reciprocating contact motion, similar to what exists in a manufacturing process and/or transfer line; (2.) a range of load and sliding speed; (3.) accurate friction coefficient information; and (4.) contact potential information to represent the lubrication film thickness along the sliding interface. A photograph of the whole test setup is shown in Figure 3. To keep the polymer-based plate in the same sliding direction, two rails were manufactured to maintain the polymer-based plate moving direction during testing. The friction force between the polymer-based plate and the rails was assumed negligible compared to that between the polymer-based plate and the steel plate under the applied normal loads (100 N and 250 N). Figure 1. Photo of the polymer-based bronze plate. Figure 2. Schematic of the reciprocating bench test. Figure 3. Photo of the reciprocating bench test. Test conditions Test conditions were determined based on the typical operating conditions for a transfer line system provided by a manufacturing factory in Michigan. The operating temperature was controlled at 30ºC using a heater block as shown in Figure 2. This temperature was the same as that in the factory. For real transfer line systems, the interface temperature will be a little higher than 30ºC because of frictional heating. This temperature rise along the sliding interface is a function of load and sliding speed and was considered negligible under lubricated conditions. The applied pressure in the factory ranges from 7.5 psi to 25 psi. Simulation of this condition in this test required a load range of 100 N to 333 N for a 38.1 mm x 50.8 mm (1.5 inch x 2.0 inch) plate. Since the highest load which can be provided by the Cameron Plint machine used is 250 N (equal to psi), both 250 N (the highest load) and 100 CONTINUED ON PAGE 42 T R I B O LO G Y & LU B R I C AT I O N T E C H N O LO G Y FEBRUARY

3 PEER-REVIEWED CONTINUED FROM PAGE 41 Table 1.Two-factor, Two-level, full-factorial DOE Table 2. Investigation of performance over longer durations Table 3. Polymer-based bronze plate weight loss data Figure 4. Friction coefficient under the operating condition (2.5Hz and 250N). N were chosen as the two load conditions to be simulated. The transfer speed in the factory ranges from about 150 inch/min to 350 inch/min. Since the amplitude of the stroke of the Cameron Plint machine used was set at 15mm, the frequencies of 2.5 Hz and 5 Hz were chosen here, which are equal to average sliding speeds of 75 mm/s (177.2 inch/min) and 150 mm/s (354.3 inch/min), respectively. Design of experiments To determine the effects of operating conditions on the polymer-based plate weight loss, a two-factor, two-level, full-factorial design of experiments was performed. As shown in Table 1, the two factors were the load and the sliding speed. Each factor had two levels. Each condition was run for 15 hours. Three replications of each factor level combination were conducted, resulting in a total of 12 runs. The response variable was the polymer-based plate weight loss. The polymer-based plate weight loss was measured by a Mettler Instrument B5C 1000 scale. To further investigate the tribological performance of the polymer-based lubrication system as a function of time under severe conditions (high pressure and high sliding speed, that is, 250 N and 5 Hz), additional tests for the durations of 60, 150 and 250 hours as shown in Table 2 were run for comparison. Due to resource constraints, these longer life tests had only two replications. All testing runs were performed in a random order. The run order as well as weight loss results are shown in Table 3. Experimental Results Study of friction coefficient and contact potential The oil impregnated in the polymer-based plate oozes out during the life span of the plate under normal operating conditions. This improves the lubrication along the sliding interfaces by helping reach a steady state condition. The lubrication effect will last for a long time (several years for a transfer line system) under normal operating conditions until the impregnated oil is exhausted. During the initial break-in phase of the polymer-based plate, a mainly dry condition exists along the interface. As oil oozes out, it reduces the friction force and helps the force reach a steady state value. However, friction coefficient measurements from the Cameron Plint tester were not repeatable even under the same operating condition in this study. As a typical case, friction coefficient may range from to under the same operating condition (2.5 Hz and 250 N) as shown in Figure 4. This variation was attributed to the surface finishing variation of the samples, the poor alignment of the sliding pair, and the sensitivity of the measurement since the friction coefficients were all quite low (0.01~0.035) and the accuracy at 42 FEBRUARY 2008 TRIBOLOGY & LUBRICATION TECHNOLOGY

4 these low levels was poor. Regardless of the differences in the friction coefficient measurements, the measured friction coefficients are much lower than that of a regular steel-bronze pair, which is around The electrical contact resistance method is also beneficial in studying metallic contact and friction between lubricated sliding surfaces. During testing, the contact resistance was measured using the Cameron Plint tester and observed as a contact voltage. A voltage of 50 mv was applied across the sliding contact using a potential divider. The contact voltage was observed on an oscilloscope as a root-mean-square (R.M.S.) voltage signal. Zero volts implies that the entire voltage drop is zero across the contact, i.e., the contact resistance is effectively zero (or only a few ohms). 50 mv implies that there is a 50 mv voltage drop across the contact and a very large or infinite resistance. A voltage signal directly related to the contact resistance is obtained by monitoring the voltage across a resistor connected in parallel with the sliding contact. In this study, a 50 mv contact voltage means the full film thickness has been formed. The dimensionless contact voltage obtained by dividing the measured contact voltage by 50 mv was used to express the development of film thickness. The value of contact potential is a function of metal contact area and film thickness. 3-4 The greater the contact area, the lower the contact potential and the thicker the tribological film, the higher the contact potential. A typical contact potential result from the reproducible measurements is shown in Figure 5. The contact potential started out very low and then increased as a layer of lubrication (lubricity) was built up causing the electrical resistance and contact potential to increase. From Figure 5, it appears that a steady-state lubrication film between the steel plate and the polymer-based plate was formed after approximately 2000 minutes of testing. Once the lubrication film was established, the contact potential reached an asymptotic value (50 mv). This steady-state lubricant film prevented the surface from wear or scuffing. Figure 5. A typical contact potential measurement vs. time. Table 4. Analysis of variance for weight loss Polymer-based bronze plate weight loss Compared with a traditional lubrication system, one major advantage of the proposed polymer-based lubrication system is its long life span, which significantly reduces the maintenance required and possible downtime. 5 It is important to predict the oil oozing rate in order to estimate the polymer-based lubrication system life span. The total polymer-based bronze plate weight loss (before and after every run) was contributed by both the polymer-based lubricant and bronze plate. As seen from the following wear residue analysis, carbon (C) and oxygen (O) accounted for approximately 80% of the total weight loss, so here it is considered reasonable to relatively evaluate the oil oozing rate based on the total weight loss. Given the total weight of the oil impregnated, the life span can be approximately estimated based on the weight loss rate. The technique of analysis of variance (ANOVA, balanced design) 6 is conducted to investigate the statistically significant effects from two independent variables (load and sliding speed) and the interaction between the load and the sliding speed. The response variable is the weight loss. The output of ANOVA is shown in Table 4. In addition to the degrees of freedom (DF), the mean square (MS), and the F-distribution value, the table shows the P-value associated with each factor and interaction. The P-value can be interpreted as the probability that the factor or interaction is not significant. Therefore, a low P-value provides an indication that a given parameter is significant and the confidence associated with that assessment. It is shown from the F-distribution values and the P-values in Table 4 that, if the one-sided confidence level is set as 95% (the commonly used alpha level), both the sliding speed and the load have a significant effect on the weight loss, while there is no significant effect from the interaction of the load and the sliding speed. If the one-sided confidence level is set as 90%, all these factors have a sig- CONTINUED ON PAGE 44 TRIBOLOGY & LUBRICATION TECHNOLOGY FEBRUARY

5 PEER-REVIEWED CONTINUED FROM PAGE 45 Figure 6. Boxplot of weight loss by load. Figure 7. Boxplot of weight loss by sliding speed. nificant effect on the polymer-based plate weight loss. The boxplots of weight loss by load and weight loss by sliding speed are also shown in Figures 6 and 7, respectively. The line drawn across each box indicates the median of the data. The bottom and the top edges of the box mark the first (25th percentile) and third (75th percentile) quartiles, respectively. These boxplots suggest that heavier load or higher sliding speed produces more weight loss. This is expected because heavier loads or higher sliding speeds lead to higher friction forces and more severe frictional heating. A full bilinear model described as Equation (1) is used to model the weight loss rate of the polymer-based plate after 15 hours of testing. The resulting 3-D surface plot is shown in Figure 8 to illustrate the trends as load and/or sliding speed increase. Δw = x y xy (1) Figure 8. 3-D surface plot (the unit of weight loss is mg). Figure 9. Polymer-based plate weight loss vs. test duration where, Δw (mg) is the weight loss of the polymer-based plate after 15 hours of testing, x (N) is the load, and y (Hz) is the sliding speed (frequency). The constants in Equation (1) are determined from a least squares fit regression analysis of the data in Table 3. In terms of applied pressure and sliding velocity, the model can be expressed as follows: Δw = p v pv (2) where, Δw (mg) is the polymer-based plate weight loss of the polymer-based plate after 15 hours of testing, p (psi) is the applied pressure, and v (inch/min) is the sliding velocity. The weight loss vs. the duration of testing is shown in Figure 9. After the initial break-in period, the weight loss rate approaches a steady rate. The trend of the weight loss can be approximated by the following logarithm function, which is also shown in Figure 9. Δw = lnt (3) where, Δw (mg) is the polymer-based plate weight loss after t hours of testing. The constants in Equation (3) are determined from a least squares fit regression analysis of the data in Figure 9. SEM and EDS analysis Typical finished surfaces of the polymer-based plate and the steel plate under two different conditions are shown in Figures 10 and 11. There were residues on the tops of the plates. The color of the residues on the surfaces of these finished plates may be green or brown. The brown 44 FEBRUARY 2008 TRIBOLOGY & LUBRICATION TECHNOLOGY

6 residue occurred only with the No. 11 run. The chemical composition of the residues was analyzed by EDS (Energy Dispersive Spectroscopy). Since an EDS analysis needs a sizable sample, the chemical composition of the green residue from the No. 9 run was analyzed and is shown in Figure 12 and Table 5. It is believed that Pb and Cu are from the bronze plate, C and part of the O from the oil used in the polymer-based structure, and the remainder of the O from the environment due to the oxidation of copper. Figure 10. Photo of the finished surfaces of No. 13 test. Table 5. Chemical composition of green residue on the steel plate (No. 9 run) The brown residue which occurred only during the No. 11 run is believed to be due to chemical degradation of the polymer-based system. The detailed chemical composition of the brown residue is shown in Figure 13 and Table 6. It is believed that the green residue came from the oxidation of copper because the atom percentage of Cu and O of the green residue was much higher than that of the brown one. Since C, which is the main component of the impregnated lubricant, accounted for a larger percentage of the brown residue than that of the green residue, it is suggested that the oil structure deteriorated faster for sample No. 11 with the brown residue. Actually, based on the weight loss data, it can be seen that the No. 11 run, which led to the brown residue, lost about 25% more weight than the other two replications under the same conditions (See Table 3) Figure 11. Photo of the finished surfaces of No. 11 test (the differences in the color may not be easily distinguished when comparing the printout of Figures 10 and 11). Figure 12. Chemical composition of green residue on the steel plate (No. 9 run). Table 6. Chemical composition of brown residue on the steel plate (No. 11 run) In addition, a scanning electron microscope (SEM) was used to analyze the surface integrity of the finished surfaces of the steel plate. The worn steel plate was cleaned with acetone before the measurement. Figures 14 and 15 CONTINUED ON PAGE 48 TRIBOLOGY & LUBRICATION TECHNOLOGY FEBRUARY

7 PEER-REVIEWED CONTINUED FROM PAGE 47 Figure 13. Chemical composition of brown residue on the steel plate (No. 11 run). Figure 14. SEM picture of the finished surfaces of the No. 11 test (50X). show (see page xx) deposited particles over the steel plate from the No. 11 test. Although the results from only one test are presented here, similar results have been observed with the other tests. These deposited materials had a chemical composition similar to that observed in the wear residue discussed above. This deposited material phenomenon may be important for transfer line applications since the contamination due to the deposition from the polymer-based system may be critical to the workpiece surface integrity. No wear mark has been observed on the surface of the steel plate, which is expected since the steel plate was under the lubricated condition during testing. Table 7 shows the changes of surface roughness of the steel plate measured along the sliding direction with a Rank Taylor Hobson Talysurf Analyzer (using a 2CR Phase Corrected filter) for three different operating conditions. For all cases, the surface of the steel plate after testing is smoother than that before testing. The reason for this is that the valleys of the original machining marks may have been filled by the polymer-based residue instead of being deepened by abrasive particle(s). Similar results were found in other runs. Summary and conclusions The tribological performance of the polymer-based lubrication system has been investigated, and a weight loss model of the polymer-based plate has been developed. Based on experimental observations and surface analysis, the research results can be summarized as follows: Figure 15. SEM picture of the finished surface of the No. 11 test (500X) 1 Friction coefficients of the tested system fell in the range of 0.01~0.035 under the typical transfer line operating conditions. The low friction coefficient will help reduce wear in manufacturing processes. 2. After the lubrication film was built-up, the contact potential approached an asymptotic value (50 mv). This steady-state lubricant film prevented the surface from wear or scuffing. It was also observed that the weight loss approached an asymptotic value after about 250 hours of sliding. Therefore, it is concluded the wear rate was gradually reduced after long sliding durations. 3. Based on the output of an ANOVA table, if the onesided confidence level is set as 95%, both sliding speed and load have significant effects on the weight loss while there is no significant effect at this level from the interaction of load and sliding speed. 46 FEBRUARY 2008 TRIBOLOGY & LUBRICATION TECHNOLOGY

8 If the one-sided confidence level is set as 90%, load, sliding speed and their interaction all have significant effects on the polymer-based plate weight loss. Table 7. Surface roughness of the steel plate before and after testing 4. A weight loss model for the polymer-based plate has been proposed based on a bilinear approach. 5. Both green and brown residues were observed on the surfaces of both the steel plate and the polymer-based plate after testing. The green residue was mainly composed of C, O, and Cu, while the brown residue contained mainly C. The organic brown residue is considered to be the result of abnormal degradation of the polymer-based structure because it was observed only in one run out of A small amount of deposited residue was observed on the worn surface of the finished steel plate. This deposition phenomenon may be important for transfer line applications since possible contamination due to the polymer-based system is critical to the workpiece surface integrity. 7. A decrease in surface roughness of the steel plate was observed at the end of testing. Acknowledgments The authors wish to thank Dr. Michael L. McMillan, Dr. James A. Spearot, Angelo Quintana, Brian Taylor, Bob Kubic and Curtis Wong for their support and help in performing the experimental studies. 1. Kroschwitz, J., Ed. (2001), Polymer-Based Lubrication System, Encyclopedia Series, John Wiley and Sons, New York. References 4. Blau, P., Ed. (2003), ASM Handbook, Friction, Lubrication, and Wear, Vol. 18, ASM Handbook Series, 3rd Ed. 2. Ashby, M.F. and Jones, D.R.H. (1980), Engineering Materials: An Introduction to Their Properties and Applications, Pergamon, New York. 3. Polishuk, A.T. (1983), Automotive Chassis Lubrication, CRC Handbook of Lubrication Theory and Practice of Tribology, Volume 1, Booser, R.E., Ed., CRC Press, Boca Raton, Fla. 5. ASTM, (2002), ASTM 2003 Annual Books of ASTM Standards, Petroleum Products, Lubricants, and Fossil Fuels, Section Neter, J., Kutner, M. H., Nachtsheim, C. J. and- Wasserman,W. (1996), Applied Linear Statistical Models, 4th Ed., Irwin, Chicago, Ill. TRIBOLOGY & LUBRICATION TECHNOLOGY FEBRUARY