An Experimental Study on the Scuffing Performance of High-Power Spur Gears at. Elevated Oil Temperatures THESIS

Transcription

1 An Experimental Study on the Scuffing Performance of High-Power Spur Gears at Elevated Oil Temperatures THESIS Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University By James Walter Brenneman, B.S. Graduate Program in Mechanical Engineering The Ohio State University 2013 Master's Examination Committee: Dr. Ahmet Kahraman, Advisor Dr. Brian Harper

2 Copyright by James Walter Brenneman 2013

3 ABSTRACT In this study, a number of spur gear tests were performed under high-power and high-temperature conditions representative of certain aerospace gearing applications. As the first type of tests, long cycle tests of 100 million cycles were performed at set operating speed, load, and temperature conditions. The second type of tests, load-staged scuffing tests, implemented an incrementally increased torque schedule under constant speed and oil temperature conditions. Two different gear tooth surfaces were considered in these tests: hard ground surfaces representative of rough, as machined gear surfaces and chemically polished gear surfaces that were an order of magnitude smoother than the ground surfaces. The primary failure mode of concern was scuffing of the contact surfaces due to temperature build up. The impact of surface roughness amplitudes, contact stress, and oil inlet temperature on scuffing failures were investigated. Effects of ramp up procedures for the speed and torque, as well as the introduction of a break-in test stage were also investigated to show that they are critical to the scuffing performance of gears. ii

4 DEDICATION This thesis is dedicated to my family, who provided the opportunities for me to pursue higher education and encouraged me to obtain advanced degrees. iii

5 ACKNOWLEDGEMENTS I would like to thank Neil Anderson of Pratt and Whitney for his support of this study. In addition, I would like to thank my advisor, Dr. Kahraman, for his continual support and guidance that made this study possible. I would also like to thank Dr. Harper for his detailed review of this thesis. I also could not have completed this study without first learning the test methodology and test machine details from both Garrett Olson and Nick Leque. Finally, I would like to express my appreciation to Sam Shon for this support within the laboratory. iv

6 VITA Born Canton, Ohio Internship, General Electric Internship, General Electric B.S. Mechanical Engineering, The Ohio State University 2012 Present...Graduate Research Associate, The Ohio State University FIELDS OF STUDY Major Field: Mechanical Engineering v

7 TABLE OF CONTENTS Abstract... Dedication... Acknowledgments... Vita... List of Tables... List of Figures... Nomenclature... ii iii iv v viii ix xiv Chapter 1: Introduction Background and Motivation Literature Survey Scope and Objectives Thesis Outline Chapter 2: Experimental Methodology Test Machine Inspection Procedures Test Procedure Long-Cycle Test Procedure vi

8 2.3.2 Staged Scuffing Test Procedure Test Gears Summary Chapter 3: Scuffing Test Results Introduction Results from the Long Cycle Scuffing Tests Results from the Staged Scuffing Tests Summary Chapter 4: Summary and Conclusions Thesis Summary Conclusions Recommendations for Future Work References vii

9 LIST OF TABLES Table 2.1 Load levels and normalized stresses used for long cycle tests Table 2.2 Normalized load levels and stresses used for staged scuffing tests Table 2.3 Basic design parameters of the spur gear pair used in this study [5,6] Table 2.4 Tooth profile modifications used in this study Table 3.1 Summary of all long-cycle scuffing tests performed in this study using chemically polished gear specimens Table 3.2 Summary of all staged scuffing tests performed in this study viii

10 LIST OF FIGURES Figure 2.1 High-speed gear test machines used this study [5, 6] Figure 2.2 Machine Top View Schematic showing its key components [5] Figure 2.3 Test gear box with cover removed to show gears [6] Figure 2.4 A test gear mounted on a gear CMM for its profile and lead measurements [6] Figure 2.5 Examples of measured (a) lead and (b) profile traces of a test pinion Figure 2.6 Examples of measured (a) lead and (b) profile traces of a test gear Figure 2.7 Roughness measurement locations along the surface of a gear for (a) a chemically polished specimen where spots 1 through 4 mark the locations of 1 mm traces, and 5 denotes the location of a 4 mm trace and (b) a hard ground specimen with traces in all three locations being 4 mm long Figure 2.8 Measured roughness profiles for a typical (a) hard ground specimen and (b) chemically polished specimen Figure 2.9 The loading profile of the accelerated staged scuffing tests used for ground gears ix

11 Figure 2.10 (a) Ground and (b) chemically polished test gear pairs [5] Figure 2.11 Typical images of tooth surfaces of (a) hard ground and (b) chemically polished gears Figure 3.1 Digital images of one of the teeth of the chemically polished tooth surfaces from Test 1 at load level L2 at 141 C. (a,b) an initial (0 cycles) and failed (0.11M cycles) gear tooth surface; (c,d) an initial (0 cycles) and failed (0.11M cycles) pinion tooth surface Figure 3.2 Variation of normalized pinion speed, pinion torque, contact stress and the resultant lambda ratio values during the ramping stage of the procedure used for Test Figure 3.3 Variation of normalized pinion speed, pinion torque, contact stress and the resultant lambda ratio values during the ramping stage of the modified loading procedure Figure 3.4 Digital images of the chemically polished pinion of Test 2 at load level L2 and 141 C at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles Figure 3.5 Digital images of the chemically polished pinion of Test 3 at load level L3 and 141 C at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles x

12 Figure 3.6 Digital images of one of the teeth of the chemically polished tooth surfaces from Test 4 at load level L1 at 155 C. (a,b) an initial (0 cycles) and failed (0.15M cycles) gear tooth surface; (c,d) an initial (0 cycles) and failed (0.15M cycles) pinion tooth surface Figure 3.7 Measured roughness traces of the chemically polished pinion of Test 4 at load level L1 and 155 C. (a) Initial roughness profile and (b) roughness profile at 0.16M cycles Figure 3.8 Measured roughness traces of the chemically polished gear of Test 4 at load level L1 and 155 C. (a) Initial roughness profile and (b) roughness profile at 0.16M cycles Figure 3.9 Digital images of the chemically polished pinion of Test 5 at load level L1 and 155 C at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles Figure 3.10 Digital images of the chemically polished pinion of Test 7 at load level L3 and 155 C -at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles Figure 3.11 Measured roughness traces of the chemically polished pinion of Test 7 at load level L3 and 155 C after (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100 M cycles xi

13 Figure 3.12 Measured roughness traces of the chemically polished gear of Test 7 at load level L3 and 155 C after (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100 M cycles Figure 3.13 Variation of measured tooth profiles of the chemically polished pinion during Test 7 at load level L3 at 155 C Figure 3.14 Digital images of one of the teeth of (a,b) the gear and (c,d) the pinion before and after Test 8 at 120 C Figure 3.15 Digital images of one of the teeth of (a,b) the gear and (c,d) the pinion before and after Test 9 at 141 C Figure 3.16 Measured roughness traces of the chemically polished pinion of Test 9 at 141 C (a) before and (b) after the test Figure 3.17 Measured roughness traces of the chemically polished gear of Test 9 at 141 C (a) before and (b) after the test Figure 3.18 Digital images of one of the teeth of the ground tooth surfaces from Test 10 at 120 C. (a,b) An initial (0 cycles) and failed gear tooth surface; (c,d) an initial (0 cycles) and failed pinion tooth surface Figure 3.19 Digital images of one of the teeth of the ground tooth surfaces from Test 11 at 120 C. (a,b) An initial (0 cycles) and micro-pitted gear tooth surface; (c,d) an initial (0 cycles) and micro-pitted pinion tooth surface Figure 3.20 Measured roughness traces of the ground pinion of Test 11 at 120 C (a) before and (b) after the test xii

14 Figure 3.21 Measured roughness traces of the ground gear of Test 11 at 120 C (a) before and (b) after the test Figure 3.22 Digital images of one of the teeth of the ground tooth surfaces from Test 12 at 120 C. (a,b) An initial (0 cycles) and scuffed gear tooth surface; (c,d) an initial (0 cycles) and scuffed pinion tooth surface Figure 3.23 Measured roughness traces of the ground pinion of Test 12 at 120 C (a) before and (b) after the test Figure 3.24 Measured roughness traces of the ground gear of Test 12 at 120 C (a) before and (b) after the test Figure 3.25 Measured roughness traces of the ground pinion of Test 13 at 141 C (a) before and (b) after the test Figure 3.26 Measured roughness traces of the ground gear of Test 13 at 141 C (a) before and (b) after the test Figure 3.27 Digital images of one of the teeth of the ground tooth surfaces from Test 13 at 141 C. (a,b) An initial (0 cycles) and scuffed gear tooth surface; (c,d) an initial (0 cycles) and scuffed pinion tooth surface Figure 3.28 Variation of measured tooth profiles of the hard ground pinion during Test 13 which scuffed at load level L11 at 141 C Figure 3.29 Variation of measured tooth profiles of the hard ground gear during Test 13 which scuffed at load level L11 at 141 C xiii

15 NOMENCLATURE Symbol Definition h min minimum lubricant film thickness R a arithmetic roughness amplitude R q root-mean-square roughness amplitude r p pitch radius of the pinion T p pinion torque v pl pitch-line velocity Θ λ inlet lubricant temperature lambda ratio Ω p pinion rotational speed xiv

16 CHAPTER 1 INTRODUCTION 1.1 Background and Motivation Mechanical power transmission systems used in aerospace systems such as rotorcraft gearboxes and jet engine turbofan gearboxes demand the highest achievable reliability levels. Yet these are the same systems that must endure the most demanding operating conditions, especially in terms of speed and temperature. As such, mechanical components of these systems must be designed to withstand high-speed and hightemperature operating conditions consistently during long service lives. Provided contact fatigue failure modes (micro-pitting or pitting) are eliminated through proper selections of material, lubrication method, and surface engineering techniques, the heat induced contact failures become the primary failure mode in aerospace gear systems. This type of failure is induced by heat build-up on the gear tooth surfaces when the cooling provided by the lubrication system (often in the form of jet cooling) fails to remove heat generated by the friction at the contact interfaces and by the pocketing of the oil from the gear meshes. As the surface bulk temperatures climb, the 1

17 maximum local temperatures (instantaneous temperature spikes at the contact interface plus the surface bulk temperature) reach levels that cannot be endured by the gear steel and the lubricant. At this point, the lubricant film in the contact interface breaks down and the contacting surfaces start to adhere or weld to each other [1, 2]. These failures, named as scuffing or scoring in gear literature, are often only detected by observing an increase in vibration or power demand. If not detected early, they lead to total, catastrophic failure of a gearbox [1]. While operating, gear teeth experience cyclic loading as they repeatedly make contact with the mating gear. In addition, almost all of the parameters dictating the elastohydrodynamic lubrication (EHL) characteristics of the gear contact (normal load, radii of curvature, sliding velocity, rolling velocity, the ratio of the sliding and rolling velocities, and surface roughnesses) vary in time as the contact travels along the tooth surfaces. These time-varying contact parameters define the conditions for a truly transient EHL regime in the gear contacts [3]. Furthermore, the minimum thickness of the oil film h min is often lower than the composite root-mean-square (rms) roughness amplitude 2 2 R qt of the contact surfaces ( Rqt = Rq1+ Rq2 where R qi is the rms roughness amplitude of surface i = 1, 2 ), i.e., the so-called Lambda ratio, λ= hmin Rqt, is less than one. In such conditions, typical of high-temperature (low lubricant viscosity) operating conditions, metal-to-metal contacts of the two surfaces are possible, making the lubrication regime a mixed (or boundary) EHL one [3,4]. Under such conditions, contact fatigue failures in the form of micro-pitting were reported to be the main failure mode for 2

18 high end gear steels [5, 6]. One solution to such failures is to reduce R qt by using certain surface smoothening processes such as chemical polishing [7] for aerospace power transmission applications. It is shown by the experiments of Olson [6] and Franzen [8], among others, that use of much smoother contact surfaces eliminate micro-pitting entirely since the contact is governed by a full-film EHL condition with λ> 1 (i.e. no metal-tometal contact). On the other hand, while smoothening the contact surfaces should help delay the occurrence of scuffing as well, conditions under which safe sliding versus scuffing occur are not fully understood. It has been noted that scuffing failures are more immediate, occurring at low cycle counts, especially when a break-in period is not applied for the new gear surfaces. AGMA [9] suggests running-in new gear sets at half of the load for a while before applying the full load to the gears. Scuffing occurs when material from one surface is transferred to the mating surface through instantaneous welding and tearing of the contacting surfaces [1]. The damage is more readily seen away from the pitch line (near the tip of the driving gear and in the dedendum of the driven gear) where sliding velocity and contact pressure are both high. Scuffing is considered to be a failure of the lubricating film between the surfaces. Many studies have shown that this breakdown is due to the lubricant instantaneously exceeding a critical temperature within the contact zone [2, 10]. This film breakdown leads to a rapid increase in friction and local heat generation as the tooth surfaces begin to weld together and rapidly tear apart as the gears continue to rotate [11]. The torn surfaces 3

19 are roughened by this process, which leads to additional frictional heat generation. This extra heat breaks down the lubricant film even more, causing quick spreading of the scuffing failure across the face of the gears. Once scuffing begins, the affected area grows in size until the profile and lead are altered enough to lose conjugate action. At this point, the gear pair is no longer acceptable for smooth and safe power transmission and must be removed and replaced. It has also been hypothesized that the chemicals contained in the additives mixed with the base lubricant are critical to scuffing performance. Formation of a tribological thin solid film on the contact surfaces through bonding of the additive compounds to the surfaces during break-in was attributed to the delay of scuffing in some studies. This study focuses on an experimental investigation of the scuffing performance of aerospace gears operating under high-speed and high temperature conditions. Impact of break-in, maximum contact pressure, and oil inlet temperature on scuffing performance are studied by using gears made of a high-temperature gear alloy operated with a turbine engine fluid. 1.2 Literature Survey Numerous experimental studies on contact failures of gear tooth surfaces or simpler contacts representative of gear and bearing surfaces have been published over the years. However, the volume of the work on this topic is far larger than what is published. 4

20 Due to the high costs associated with test machines, gear specimens, and long test durations, most of the experimental results collected by companies or research entities have been kept out of the public domain. As such, this section provides a review of a limited number of published studies on this topic. A number of studies focused on the effect of material type and surface roughness amplitudes on the contact performance of the gears have been performed. As the primary emphasis of these studies were on the contact fatigue (spalling or pitting and micropitting) failures, these experiments were done under rather favorable lubrication conditions to avoid scuffing failures. Among such studies, Townshend, et al [12] used a four-square gear test machine to evaluate the effect of λ on gear surface life. Gears were made of consumable electrode vacuum-melted (CVM) AISI 9310 steel and were hardened. The final test specimens had a nominal surface roughness of Rq 0.4 μm. Instead of seeking different R q values to change λ, they used different lubricants with varying viscosities. One of these lubricants that conformed to the MIL-L specification was reported to perform the best in terms of micro-pitting. Krantz et al [13] varied λ by altering the surface roughness of the test specimens instead. This study used the same test rig and gears as Townshend [12], but the gears were polished such that the surface roughness amplitudes were nearly five times lower than those of ground gear surfaces. The fatigue life of the super-finished gears was improved by nearly a factor of 4 compared to the ground surfaces studied previously. Both of these studies were 5

21 performed under high-speed conditions with unity ratio gears of about 90 mm center distance at 10,000 rpm to achieve a pitch-line velocity of 47 m/s. Another group of studies focused on low-speed tests directed towards automotive gearing applications. These tests used a FZG type test machine, standard test gears and procedure outlined by ISO standards. By changing the type of gears, it is intended to achieve a certain failure mode: macro-pitting, pitting or scuffing. A series of experimental studies performed at the Ohio State University in recent years [14-17] are of this type, except they use a different gear design than what is specified in the ISO standard to achieve pitting consistently. All of these studies were performed with a gear pair having 17:26 ratio with a pitch line velocity that is less than 10 m/s. In one of these studies [15], a scuffing test was also performed according to ISO Another group of researchers at the Ohio State University performed high-speed and high-temperature experiments using new test machines described first by Leque [5]. These test machines had the same center distance as the FZG machines (91.5 mm) and used the same basic 17:26 gear design from Refs. [14-16], but were able to operate the pinion at speeds up to 13,500 rpm to achieve pitch line velocities up to 50 m/s at elevated oil inlet temperatures up to 150 C. Rather than the typical dip lubrication method used in FZG tests, the tests gears in these test machines were jet lubricated with high pressure oil. Preliminary studies with gears made from a type of high performance aerospace steel were performed by Leque [5]. Three surface roughness classes were procured. The gears were hard ground, hard ground and chemically polished, or hard ground and super- 6

22 finished. Hard ground samples suffered severe micro-pitting as a result of surface roughness amplitudes of 0.5 μm, while any effort to reduce surface roughness through polishing or super-finishing was shown to eliminate the occurrence of micro-pitting entirely. Olson [6] used the test methodology described by Leque [5] to study the fatigue lives of two types of gear steel under high-speed and under higher temperatures that what had been studied previously with FZG gear testing. Aerospace steel gears were chemically polished to R q = 0.06 μm or super-finished to R q = 0.03 μm. SAE 4118M gears for an intended automotive application had ground surfaces with R q = 0.52 μm. This resulted in λ 0.45 for the ground gears at automotive speeds and a range of 2.9 to 7.4 for the aerospace specimens under aerospace speeds. Ground gears made of SAE 4118M were shown to exhibit pitting and micro-pitting failures, while a type of high performance aerospace gear steel with a smoother surface finish exhibited no signs of pitting or micro-pitting after extended periods of testing. Instead, certain finished gears exhibited unexpected scuffing failures on start up with the methodology of Leque [5]. All of the above studies on pitting were fully experimental. A family of modeling studies was performed by Li and Kahraman. They first developed mixed EHL models for point contact [4] and gear contact [3] problems. Later, they combined a multi-axial fatigue criterion with these EHL models to predict macro-pitting for point contact [17] and gear contact [18] problems. The point contact fatigue model was validated by using 7

23 the two-disk experiments of Li [19] for both ground and chemically polished surfaces. The gear pitting model [18] was compared to Klein s data [15] for its validation. Another group of studies by Li and Kahraman [20, 21] proposed physics-based micro-pitting models for point contacts and gears. These models captured the stress concentrations at or near the surface due to surface roughness texture using a boundary elements formulation. The point contact micro-pitting model was validated using twodisk experiments [22]. Several studies were performed on the influence of operating conditions, such as load and speed, on scuffing failures. These studies are primarily done using two-disk machines, ball-on-disk machines, or FZG gear test machines. One such study was done by Jackson et al [23] to investigate the influence of loading on scuffing failure for a roller disk machine using various amplitudes of surface roughness. ISO VG 220 lubricants were used to keep the film thickness consistent between all tests. One surface finish was achieved by grinding the disk and roller, and the other was done with a polishing method. This study utilized a run-in loading stage before the full scuffing test was performed, and all of the loading stages were run in one continuous test. This study showed that reducing the value of λ from 6 to 1.2 reduced the scuffing load by nearly 80 percent. Liou [11] used a roller disk machine to perform scuffing tests with automatic transmission fluid. A model was developed for the thermal characteristics of a lubricated contact. A second model was made for the convective heat transfer between two contacting bodies. 8

24 Li et al [2] performed a number of ball-on-disk scuffing experiments of ground specimens under various rolling velocities and sliding ratios. The normal load was increased in these tests incrementally until scuffing occurs. The surface bulk temperatures and the friction coefficient were measured. A scuffing model that combined a thermal version of the mixed EHL model of Ref. [4] with a heat transfer model of the ball was used to predict these measured parameters as well as the instantaneous (flash) temperature increases in the contact interface. The model predictions were shown to agree well with the measured bulk temperatures and friction coefficients. Furthermore, it was shown that there is a maximum temperature (bulk plus flash) at which scuffing occurs. Similarly, Enthoven et al [24] also used a ball-on-disk set-up to evaluate the critical temperature at which scuffing occurs for a steel ball sliding on a sapphire plate. Two lubricants were used: hexadecane and SHF 41, a ployalphaolefin base stock. A ball made of AISI steel was polished such that the roughness lay was either parallel or perpendicular to the direction of sliding. Temperature was monitored with the use of a noncontact method called infrared radiometry. This methodology was shown to consistently capture the evolution of surface temperature during a scuffing event. The scuffing experiments described above used equipment designed for simple, inexpensive test specimens. While these tests are useful for understanding the fundamentals of scuffing events, they do not take into consideration other effects that may be seen in an actual gear mesh. Gear scuffing tests were typically done using an FZG machine. One study by Niemann et al [25] used an FZG machine to perform 9

25 scuffing tests on spur gears with various oil types. These gears were made to have very high sliding velocities away from the pitch line since this is where scuffing is most likely to occur. Many different sets of operating conditions were explored. For the standard test, the surfaces were ground to a roughness of approximately 0.5 μm. Tests were carried out at a constant gear speed of 1,450 rpm, which is standard for FZG machines. These gears were dip lubricated with oil temperature held at 90 C. The load was set prior to rotation, and after 15 minutes of running time, the gear pair was measured and run with progressively higher loading. For the baseline lubricant, mineral oil, the gears failed at a normal gear mesh load of 1,590 N. When a lubricant designed for high pressure was used, failure was not observed, even with mesh loads as high as 15,800 N. Other tests were performed with oil temperatures held at one of three levels: 60 C, 90 C, or 130 C. This set of tests showed that lubricants without anti-scuffing additives exhibited a decrease in scuffing load with increasing temperatures, mostly due to a decrease in viscosity that results from an increase in lubricant temperature. Jiang and Barber [26] investigated the effects of oil formulation and pitch line speeds on the critical temperature at which scuffing occurs. Load stages were used similarly to Niemann [25], but speed was also used as a controlled variable. After each test, a thermocouple placed under the surface of a working flank measured the bulk temperature of the gear. A critical temperature for each scuffing test was calculated based on bulk temperature, operating conditions, and lubricant parameters. In this study, gear scuffing loads were seen to decrease with increasing speeds, and gear speed-critical temperature curves followed a parabolic shape. 10

26 1.3 Scope and Objectives While there have been many efforts to generate data for gear surface fatigue failures, there are not many studies done at extreme temperatures and high gear pitch line velocities. While Olson [6] successfully generated data points for long-cycle tests using specimens with a hard ground finish, a limited amount of data is available for chemically polished test gears. The primary goal of this study is to investigate the scuffing performance of high-speed gears operating with oil inlet temperature well beyond those in the experiments of Olson [6]. The specific objectives of this study are as follows: Apply the test methodology of Olson [6], developed originally for contact fatigue life evaluations, to investigate the long-cycle scuffing performance of gears made of a high performance aerospace steel operating under high-speed and high-temperature conditions. Investigate the scuffing performance of gear specimens during long-cycle, high-speed and high-temperature tests as a function of surface roughness amplitude (defined by hard grinding or chemical polishing), torque transmitted, and the oil inlet temperature. Develop a new staged gear scuffing test methodology to establish scuffing load limits of ground and chemically polished gears under high-speed and high-temperature conditions. 11

27 1.4 Thesis Outline Chapter 2 of this thesis describes the test machines, test methodology, and test specimens used in this study. The relevant features of the testing machines are discussed as well as key operation and maintenance details. The inspection procedures including the metrology equipment are described. The details of the long cycle testing conditions are presented along with the development of a staged scuffing test procedure. Finally, the gear specimen specifications are presented. Chapter 3 presents the long-cycle scuffing gear test results at various loads and temperatures. Results of the staged scuffing tests are also presented in this chapter. For each test, gear profile traces, surface roughness profiles, and digital images are shown to demonstrate the mechanisms leading to the failures. Finally, Chapter 4 summarizes the research described in the previous chapters of the thesis. General conclusions from the research are made and recommendations for future work are listed. 12

28 CHAPTER 2 EXPERIMENTAL METHODOLOGY 2.1 Test Machine The gear durability experiments presented in this study were performed on two identical test machines described by both Olson [6] and Leque [5]. Figure 2.1 shows the laboratory facility housing these two test machines. The machines are capable of pinion speeds up to Ω p = 13,500 rpm, which corresponds to a gear maximum pitch-line velocity of v pl = 2π Ω 60 pr p = 50 m/s where r p is the pitch radius of the pinion. Here the smaller of two gears forming the test gear pair is called the pinion while its larger mate is called the gear. In this set-up, the direction of the pinion rotation is reversible to allow for a greater number of possible testing conditions. A pinion torque of up to T p = 450 Nm was achievable in either direction. This study utilized certain torque and rotational velocity directions such that the pinion has the role of the driving member and the larger gear was the driven member. This combination of the maximum torque T p and speed 13

29 Figure 2.1 High-speed gear test machines used this study [5, 6]. 14

30 Ω p values allows the machines to operate at high power transmission conditions up to 630 kw (845 HP) at elevated temperatures, representative of aerospace gearing conditions. The machines were controlled via a computer terminal, allowing for any test condition to be applied, so long as they are programmed prior to the start of a test. One way the machine could be programmed was to run a long cycle test at a single load ( T p ), speed ( Ω p ) and oil inlet temperature ( Θ ) condition, with the machine automatically shutting down at a user-defined number of cycles for inspections. This is the programming scheme used for the long-cycle tests done in this study. In addition, the machine controller allowed for any duty cycle consisting of stages or bins of different test conditions ( Ω, T, Θ ). As the focus of this study is on scuffing limits, tests stages p p i were formed by keeping Ω p and Θ constant while increasing T p incrementally until either the maximum torque (or the corresponding maximum tooth contact stress) value was reached without failure or the gears scuffed at a certain stage i = I with Tp = TpI representing the scuffing torque. The test gears were jet lubricated with oil at pressures up to 10.3 bar (150 psi) and oil inlet temperatures up to Θ = 160 C (320 F). Nozzle manifolds were positioned both into mesh and out of mesh. Each manifold consisted of six small orifices, each at a diameter of 0.5 mm (0.02 ). The flow through each of these nozzles was controlled with a manual needle valve, and the total flow rate was monitored with a visual flow meter. 15

31 The lay-out of the test machines shown in Figure 2.2 indicates a back-to-back architecture. In this arrangement, gears of the test gear pair held by the test gearbox at one end of the machine are connected to the respective gears of a reaction gearbox on the other end of the machine. Gear pairs for the test and reaction gearboxes have the same ratio and center distance. Long, compliant shafts and a pair of elastomer couplings are used to connect the gears torsionally such that any adverse effects associated with the reaction gear box (e.g. vibrations) can be isolated from the test gear box. The reaction gearbox consisted of helical gears that were overdesigned to ensure extremely long life and to minimize the amount of maintenance that needed to be performed. The reaction gears were made with a wider face width than the test gears, resulting in lower contact pressures and root bending stresses. These gears were jet lubricated with Mobile SHC 629 gear oil to effectively lubricate and cool the gears. This oil flowed through an oil-to-water heat exchanger to keep the reaction oil at 45 C. The gear box had two shafts that would connect to the test gears, similar to a conventional back-to-back gear testing machine, such as an FZG testing machine. The rear of the gearbox had a two-piece concentric shaft. This shaft served two purposes. The first was to allow the motor to input rotational motion to the system via drive belts. The second purpose was to support a hydraulic torque actuator that allowed for the dynamic torque application between the two concentric shafts while the machine was operating. A noncontact torque-meter was also mounted on this shaft to monitor and control torque levels. The machines were designed to use a 17-tooth pinion on the left side of the test 16

32 17 Figure 2.2 Machine top view schematic showing its key components [5].

33 gearbox, and a 26-tooth gear on the right side, as viewed from the front of the machine. The center distance between the shafts was fixed at 91.5 mm, which is the same center distance used in FZG type gear testers. The test gearbox utilized two identical hardened steel shafts to support the test gears. Each shaft was supported with two oversized precision cylindrical roller bearings, with one of them used to axially locate all of the internal components. To avoid problems associated with rotating mass unbalance at high speeds, the shafts were symmetric, using two keys to engage the gears. In addition, lubrication lines to the bearings were separated from those feeding the gear mesh. This allowed for sufficient flow rates of oil at lower temperatures to be supplied to the test gearbox bearings to extend their lives. The gears were held in place axially with precision spacers and a lock nut that was threaded to the end of the shafts. A photograph of the gears loaded into the machine is shown in Figure 2.3 with the front cover removed to expose the test gear pair. In order to drive the machine, a 30 kw (40 HP) AC motor was connected to the outer portion of the concentric shaft of the reaction gearbox via a high-speed V-belt system. This size of the AC motor was sufficient since it must provide only the power required to negate frictional and other forms of power loss within the system. The motor was powered by a variable-frequency drive with speed feedback from a proximity sensor mounted near the coupling for the high-speed test shaft. For a pinion speed of Ω = 12,000 rpm, the AC motor rotates at a reasonable speed of 2,900 rpm due to the p speed increases provided by the belt drive and the reaction gearbox. 18

34 Figure 2.3 Test gear box with cover removed to show gears [6]. 19

35 The test side lubrication system utilized a 5 gallon sump that was kept full by topping off as necessary to compensate for any leakages in the hydraulic circuit of the machine. Earlier high-speed tests with these machines [5, 6] were performed at oil inlet temperatures up to Θ = 150 C. In this study, two 6 kw heaters were added to the high pressure stainless steel line that feeds the test gear meshes. These heaters allowed for tests with oil inlet temperatures up to Θ = 160 C. This also reduced the time to heat the lubricant since the added heater would achieve the desired oil inlet temperatures even as the main sump was still cool. A 11.3 liter/min (3 GPM) pump was used to deliver oil at pressures necessary for the jet lubrication of the system. Each test gear bearing was lubricated and cooled with the same test oil at 55 C. In order to filter any debris from the oil system, a 74 μm strainer was used, followed by a 25 μm oil filter. The test machines were equipped with several types of health monitoring instrumentation. PCB uni-axial accelerometers were placed on the test gearbox, as well as the reaction gearbox, to monitor the bearing vibration amplitudes to signal any bearing failures. Multiple thermocouples were devised to monitor the temperature of the lubricating oils and various bearings throughout the system. Drive motor current was monitored to immediately shut down the machine if the torque demand from the drive increased rapidly, potentially indicating a catastrophic failure of a main drivetrain component (any of the shafts, bearings, elastomer couplings, or reaction gearbox gears). Likewise, the oil levels in the test and reaction side lubrication systems were also monitored. A test was suspended automatically if any of the allowable limits set for all of these monitored parameters was exceeded. 20

36 2.2 Inspection Procedures Various gear inspections were performed for each gear pair before they were considered for a test. This ensured the gears met the specifications in terms of their geometry, especially tooth surface accuracy and surface roughness amplitudes. After the initial inspection, only qualified test specimens were used in a given test. In addition to this initial set of inspections of brand new gears, each gear was inspected periodically during various cycles of the test in order to (i) identify and document mechanisms leading to contact failures and (ii) allow suspension of failing tests at a cycle near its actual life. During an inspection, surface roughness measurements were taken at various locations on the tooth face of a single tooth. Involute and lead profile traces were also taken with a coordinate measurement machine (CMM) on four predetermined teeth of each gear. Finally, digital photographs were taken of the contact surface of a tooth on each gear. Exactly the same gear teeth were measured at each inspection to quantify the progression of any failure from one test interval to another. Tooth profile measurements were performed by using a gear CMM (Gleason M&M 225 gear coordinate measuring system), as shown in Figure 2.4. The machine was programmed to measure the lead at three different locations on four teeth. The involute profiles were measured at three different locations along the face width of these same four teeth. These measurements resulted in twelve involute profiles and twelve lead profiles for each inspection stage. Figures 2.5 and 2.6 show example sets of traces from a hard ground pinion and gear, respectively, prior to testing. 21

37 Figure 2.4 A test gear mounted on a gear CMM for its profile and lead measurements [6]. 22

38 a) b) Figure 2.5 Examples of measured (a) lead and (b) profile traces of a test pinion. 23

39 a) b) Figure 2.6 Examples of measured (a) lead and (b) profile traces of a test gear. 24

40 Another key step of the inspection procedure was measuring the surface roughness profiles of the gear contact surfaces. A Taylor-Hobson Form Talysurf 60 was used to measure the surface roughness of each specimen. Three measurements were taken at each location, and the results were averaged to obtain a representative value for each surface roughness parameter of interest. The actual measured profiles were saved with other roughness parameters calculated for future reference. For chemically polished surfaces with very low roughnesses, 1 mm long traces were taken at four locations on the tooth surface in order to get representative values for the roughness across the surface of the gears. In addition to these 1 mm traces, a larger 4 mm long trace was taken along the center of the gear face width, as this was where the highest stresses would be encountered with the use of gear micro-geometries. These locations are shown in Figure 2.7(a). For gears with hard ground surfaces, 4 mm traces were measured at three positions in the profile direction, as shown in Figure 2.7(b). Depending on how smooth the measured surfaces were, recommended low frequency cut-off values were used to filter out the waviness component of the roughness according to ISO A cut-off of 0.8 mm was used for hard ground specimens, while 0.25 mm was used for the chemically polished specimens. Due to the curvature of the gear profile, there were times when only 3.2 mm of the total 4 mm could be analyzed. The two parameters of interest in this study were the arithmetical mean roughness amplitude ( R a ) and the root-mean-square roughness ( R ). Figures 2.8(a) and (b) show typical roughness profiles along the transverse mid-plane of a hard ground pinion and a q. 25

41 a) b) Figure 2.7 Roughness measurement locations along the surface of a gear for (a) a chemically polished specimen where spots 1 through 4 mark the locations of 1 mm traces, and 5 denotes the location of a 4 mm trace and (b) a hard ground specimen with traces in all three locations being 4 mm long.. 26

42 Surface Roughness Amplitude (µm) a) Trace Distance (mm) Surface 0.1 Roughness Amplitude 0 (µm) b) Trace Distance (mm) Figure 2.8 Measured roughness profiles for a typical (a) hard ground specimen and (b) chemically polished specimen.. 27

43 hard ground and chemically polished pinion, respectively. In Figure 2.8(a), R q = 0.79 μm and R a = 0.62 μm for the ground surface while R q = μm and R a = μm for the chemically polished surface of Figure 2.8(b), indicating that the chemically polished surfaces are nearly 15 times smoother than the ground gear surfaces considered in this study. The final step in an inspection was to record digital images of the gear surfaces. The images were taken with a high resolution camera that was attached to a computer. The camera was mounted to a 1/2X objective lens. An adjustable lens allowed for the magnification to be changed so that details of interest could be better captured. A magnification of 3.5X was used for the gear, while 5X was used for the smaller pinion. Over the span of a test, the same tooth was used for the imaging inspection to see the progression of wear and other failure modes. This same tooth would be the tooth used for the surface roughness measurements. 2.3 Test Procedure Two different test procedures were used to evaluate scuffing performance of test gears in this study. Each test required a set oil temperature to be reached before ramping speed and load. After ensuring that all oil levels and line pressures were at appropriate levels, the preheating stage was started. This preheating stage continued until the mesh inlet oil temperature was within ±3 C of the set value. In order to minimize the chance of scuffing on initial startup due to high contact stresses with low speeds (i.e. extremely low 28

44 λ values), contact stresses were kept low until a large oil film thickness was established at full speed. This was accomplished by holding the torque T p at a low value while the machine increased the pinion speed Ω p to ensure the meshing surfaces did not lose contact during the speed ramping. Ramping up Ω p at a rate of 150 rpm per second and T p at a rate of 15 Nm per second, T p would reach only this low value while continued to increase to the desired set level. After this point, T p was ramped further to Ω p its set test level while Ω p was maintained at its set value. At the conclusion of a test (or a stage of a test), both T p and Ω p were reduced to zero and the machine lubrication system was switched to cooling mode to cool the machine and test specimens back to room temperature. Once the bearings were cooled to 35 C, the lubrication system was shutoff to allow for the test specimens to be removed for inspection. The pinion rotational speed was held constant at Ω p = 12,000 rpm throughout all testing conditions. The pinion was rotated in the counter-clockwise direction as viewed from the front cover of the gearbox. This direction was such that an up-mesh lubrication condition was established, which kept lubrication mechanisms consistent for all tests. The lubricant used in this study was a high performance turbine oil (BP Turbo 2380) meeting the specification of MIL-L for gas turbine engine lubricating oils. A manual needle valve was used prior to the test lubrication jet nozzles, which allowed for a range of volumetric flow rates to be achieved. A flow rate of up to 2.6, 3.0 and 3.8 l/min (0.7, 0.8, or 1.0 gallons per minute) was possible with oil inlet temperatures of 120 C, 29

45 141 C, and 155 C, respectively. The test oil sump was topped off as necessary to offset the volume lost to leakage throughout the system. The test oil was replaced approximately every 200 million cycles Long-Cycle Test Procedure The first type of test was used to investigate the long cycle performance of ground and chemically polished gear surfaces under various torque ( T p ) and oil inlet temperature ( Θ ) values. In this test, all three test parameters ( T p, Ω p, and Θ ) were set to constant values for the entire testing interval. No break-in period was used for the tests at Θ = 141 C. Meanwhile, a break-in segment of 1 million cycles (80 minutes of test at Ω p = 12,000 rpm) at 1 T 2 p was applied before the maximum temperature tests at Θ = 155 C. Five consecutive 20-million-cycle test segments constituted a successful long-cycle test to 100 million cycles, each 20M cycle segment representing 28 hours of testing (139 hour of test required for accumulating 100 million cycles). After each 20M cycle test segment, inspections specified in the previous section were performed before the specimens were placed on the machine again to run the next segment. Any gear pair that survived 100M cycles without any signs of scuffing (as well as micro-pitting or wear) was deemed a successful test and this particular test condition ( Ω, T, Θ ) was declared to represent a no-scuffing condition. During the onset of scuffing failure, several of the machine sensors exhibited spikes in their values. Specifically, high speed test gearbox bearing vibrations would increase as the damage caused by scuffing 30 p p

46 progressed. These vibrations would increase nearly by a factor of two as compared to the vibrations immediately at the end of the ramping phase. The safety limits of the machine ensured that operation beyond this level of vibration was not permitted. This allowed the machine to shut down quickly before the test machine was damaged. This also prevented the scuffing damage from progressing across the entire contact surface of the gear tooth. Three pinion torque ( T p ) levels were used in the long-cycle tests. These loads were labeled L1, L2, and L3, where L1 is the lowest load, and L3 is the highest load. Each of these loads corresponded to a maximum contact pressure. These load levels and the corresponding maximum contact stresses are listed in Table 2.1 in a normalized manner for confidentiality reasons. For each load level, long-cycle tests were performed at two oil inlet temperatures of Θ = 141 C and 155 C Staged Scuffing Test Procedure A second type of test was defined in this study to establish scuffing load limits of the gears in an accelerated manner for various amplitudes of surface roughness. The objective of this study was to establish a procedure to examine the critical contact stress for this gear design before the onset of scuffing failure. Several testing trials were used to develop the final procedure presented in this study. Normalized values of torque and contact stress are listed in Table 2.2 for all loading steps investigated. The test was completed when the gear set exhibited signs of significant micro-pitting, scuffing, or any. 31

47 Table 2.1 Load levels and normalized stresses used for long cycle tests. Load Level Normalized Pinion Normalized Pitch-line Torque Contact Stress L L L

48 Table 2.2 Normalized load levels and stresses used for staged scuffing tests. Load Level Normalized Pinion Torque Normalized Pitchline Contact Stress L L L L L L L L L L L L

49 other type of failure. At the end of the test, a final inspection was performed before the test specimens were properly stored for future reference. This staged scuffing test differs other gear scuffing tests done using ISO (FZG Scuffing Test) in several aspects. See Klein [15] for an application example of this test. The ISO scuffing test consists of 12 load stages with each stage increasing the amount of applied torque. Each stage of 21,700 cycles (15 minutes of test at 1455 rpm) is required to be performed independently with dip lubricant oil temperature set at 90 C at the beginning of each stage. In the staged scuffing test developed for this investigation, load stages are not independent as the end condition of the previous stage forms the initial condition for the next stage. The first scuffing setup used an oil inlet temperature of either 120 C or 141 C. After the preheating stage, the machine would ramp to the first loading step, defined in Table 2.2. Each load step was executed at the same fixed oil inlet temperature and speed, and every load step consisted of a 3-million-cycle interval before the next, larger load step was applied automatically. This long duration ensured that all monitored vibrations and temperatures would reach steady state values before progressing. Another version of the same scuffing test was implemented with a reduced number of cycles (0.36M cycles corresponding to a 30-minute test at 12,000 rpm) at each load stage in addition to a break-in stage of 1M cycles at the load step defined as L1. This loading schedule would still allow for all monitored variables to reach new steady state values at each load, but would prevent the progressive micro-pitting failures at higher 34

50 cycle counts. The loading schedule can be seen in Figure 2.9. This method was used solely for hard ground gear pairs, which each resulted in scuffing past the first load step. 2.4 Test Gears The gears used in this study were made from the same batch as those in Olson s [6] study. The basic design parameters of the test gear pair are given in Table 2.3. Certain lead and profile modifications were applied to the gears to keep the maximum contact stress location at the transverse mid-plane of the contacting surface while at the same time reducing the gear vibrations at higher loads [5]. The micro-geometry modifications are specified in Table 2.4. The gears were case-carburized to a case depth of 1 mm. The surfaces of the gears had a hardness of approximately 60 HRC. The surface roughness amplitudes of the ground gear tooth surfaces were within the range R q [0.55,0.90] µm. All gear specimens were machined as hard ground gears, shown in Figure 2.10(a). Then some of these hard ground gears were processed with a commercial chemical polishing procedure to (i) reduce their surface roughness amplitudes and (ii) remove any directionality of roughness lay caused by grinding marks to obtain an isotropic surface texture. Figure 2.10(b) shows a pair of chemically polished gears. The roughness amplitudes of chemically polished surfaces were measured to be within the range of R q [0.04,0.09] µm, which is an order of magnitude lower than those for the ground gears. Figure 2.11(a) and (b) show close-up images of a ground tooth and a chemically polished tooth, respectively. 35

51 3.5 3 Normalized Torque Elapsed Time (min) Figure 2.9 The loading profile of the accelerated staged scuffing tests used for ground gears.. 36

52 Table 2.3 Basic design parameters of the spur gear pair used in this study [5,6]. Parameter Pinion Gear Module (mm) 4.23 Center distance (mm) 91.5 Number of Teeth Pressure Angle (deg) Face Width (mm) Root Diameter (mm) Base Diameter (mm) Outside Diameter (mm) Circular Tooth Thickness (mm)

53 Table 2.4 Tooth profile modifications used in this study. Type Parameter Modifications Pinion E Gear Tip relief Magnitude (μm) Start roll angle (deg) Root relief Magnitude (μm) 5 5 Start roll angle (deg) End roll angle (deg) Lead crown Magnitude (μm)

54 Figure 2.10 (a) Ground and (b) chemically polished test gear pairs [5]. 39

55 (a) (b) Figure 2.11 Typical images of tooth surfaces of (a) hard ground and (b) chemically polished gears. 40

56 2.5 Summary This chapter provided relevant details in the description of the test machines, test methodology, and test specimens used in this investigation. Details of the test methodology included procedures used to inspect the gear specimens and the procedures used to accumulate cycles at selected operating conditions. These defined conditions include speed, inlet oil temperature, and torque. The failure criteria for gear specimens were also described. 41

57 CHAPTER 3 SCUFFING TEST RESULTS 3.1 Introduction This chapter presents the test results for the long cycle and staged scuffing tests. The long cycle tests each operated at one of three loading levels (L1, L2, and L3 as defined in Chapter 2) and one of two oil inlet temperatures ( Θ= 141 C or 155 C ) using BP Turbo Oil Each test was run at a constant pinion rotational speed of Ω p = 12, 000 rpm. All long cycle tests were performed with chemically polished gears. This surface finishing technique produced root-mean-square roughness amplitudes within the range R q [0.04,0.09] μm. Initial tests suffered scuffing prematurely on startup, prompting changes to the loading scheme of the test procedure. All of the 5 tests with the modified start-up conditions were suspended under the long cycle test procedure after reaching 100 million cycles of testing. Rigorous inspections were performed every 20 million cycles to monitor the surface conditions of the gear pairs. No macro-pitting (spalling) was observed for these tests, as the tooth surfaces in mesh operated under full-film EHL conditions (λ values between 1.5 and 2.1) due to the reduced roughness amplitude caused by chemical 42

58 polishing. Likewise, micro-pitting was also not observed in any of these tests, again due to the reduced roughness amplitudes. However, due to the high-speed and hightemperature conditions of the tests, scuffing was the primary failure mode for the chemically polished gears. The staged scuffing tests were performed under three different loading profiles. These tests used test gears made of high performance (high-temperature) gear steel with two different surface finishes. The first type of surface finish was a typical hard ground tooth surface yielding measured r.m.s. roughness amplitudes in the range Rq [0.55, 0.90] μm, which are 6 to 22 times rougher than the chemically polished tooth surfaces. Tooth contacts of gears with ground surfaces resulted in λ values between 0.18 and 0.22 indicating that asperity (metal-to-metal) contacts take place. The second surface finish used in staged scuffing tests was accomplished with the same chemical polishing technique used for the long cycle tests, and share similar roughnesses. 3.2 Results from the Long Cycle Scuffing Tests The results for long cycle scuffing tests are listed in chronological order in Table 3.1, with loading modifications being carried over among consecutive tests. For example, Test 6 includes both the low torque step added for Test 2, as well as the breakin added for Test 5. These tests were performed at three load levels (L1, L2, and L3) and two inlet lubricant temperatures ( Θ= 141 C or 155 C ). Out of these seven tests, Test 1 and Test 4 were suspended due to scuffing failures soon after start-up. All of these tests were run with chemically polished gear test specimens. 43

59 Table 3.1 Summary of all long-cycle scuffing tests performed in this study using chemically polished gear specimens.. Test Number Load Level Lubricant Temperature [ C] Cycles to Failure [millions] 1 ( ) L (*, ) L (*) L ( ) L (*, ) L (*) L (*) L ( ) Scuffing failure (*) Suspended without failure ( ) Intermediate torque on startup added ( ) Break-in period added 44

60 Test 1 in Table 3.1 was initiated by ramping up both the torque T p from zero to the desired load level (L1, L2, and L3) and speed Ω p from 0 to 12,000 rpm simultaneously after the inlet oil preheating stage had ended. This procedure matched what was used by Olson [6]. Test 1 was suspended at 0.11 million load cycles (less than 10 minutes of operation at 12,000 rpm) due to scuffing. This scuffing was detected by a rapid increase in test gear box vibrations, which is typical of all the scuffing failures seen in this study. Surface roughness measurements showed that the starting roughness of each specimen (R q = μm for the pinion and R q = μm for the gear) was acceptable for what was specified for the chemical polishing process. This resulted in λ = 2.03, which was similar to all other chemically polished gear sets. Digital images of the pinion and gear from Test 1 are shown in Figure 3.1. The damage that was done during the scuffing failure is seen to the left of the center of the face width. The scuffing action is seen to make deep marks that are oriented in the direction of sliding. On tests completed by Olson [6], surfaces that did not experience failure showed areas that likely represented a thin surface tribo-film that had formed on the surface and may have acted to protect the surfaces from damage. Test 1 did not show any signs of the formation of a tribo-film, since the test lasted a total time of under ten minutes. T p and The analysis of contact conditions during the initial ramp-up stage to reach the set Ω p values revealed certain undesirable conditions that might have led to this scuffing failure. The ramp-up rate of 150 rpm per seco1nd was used for Ω p, representative of the capability of the speed controller, such that it took 80 seconds for 45

61 a) b) 46 c) d) Figure 3.1 Digital images of one of the teeth of the chemically polished tooth surfaces from Test 1 at load level L2 at 141 C. (a,b) an initial (0 cycles) and failed (0.11M cycles) gear tooth surface; (c,d) an initial (0 cycles) and failed (0.11M cycles) pinion tooth surface..

62 the speed to reach 12,000 rpm. Likewise, the ramp up rate for that T p would reach its set valued long before T p was 15 Nm/sec, such Ω p did. By using a gear load distribution program, Windows LDP [27], variation of the maximum contact stress and the lambda ratio with the speed and torque during this ramp up stage were computed for this gear pair with the same lubricant at Θ = 141 C. Figure 3.2 plots normalized T p and Ω p (normalized by 12,000 rpm) and the resultant contact stress and lambda ratio values as a function of time. The contact stress values are also normalized to correspond to the loading steps described in Chapter 2. This study used an oil inlet temperature of Θ=141 C and maximum normalized torque of 2.07, i.e. load level L2. Figure 3.2 indicates that the set T p and contact stress levels are reached in about 15 seconds when Ω p is only 2,250 rpm. As a result of these transitional low-speed and high-load conditions, gears are seen to operate at lambda ratios below 1.0 during the first 25 seconds of the ramp up. The premature scuffing failure observed in Test 1 as well as those reported by Olson [6] can potentially be attributed to such adverse start-up conditions. In order to test this hypothesis, the conditions of the first set of calculations were repeated with a small change to the way the machine ramped speed and load. Instead of applying load level L2 at T p = 2.07 during the ramping of the pinion, a much smaller torque level of T p = 0.43 was used until the pinion speed reached its full value of 12,000 rpm. After the full speed was reached, a second torque ramp up would be initiated to 47

63 Lambda Ratio, Normalized Contact Stress, Normalized Torque, Normalized Pinion Speed Lambda Ratio Normalized Contact Stress Normalized Torque Normalized Pinion Speed Time(s) Figure 3.2 Variation of normalized pinion speed, pinion torque, contact stress and the resultant lambda ratio values during the ramping stage of the procedure used for Test 1..

64 bring the torque to its full set point. Figure 3.3 shows the calculated values of contact pressure and lambda ratio for this new loading schedule. It is seen in this figure that and the corresponding contact stress value are low during the low speed segment of the schedule when the lambda ratio is less than 1.0. While load level L2 corresponded to a normalized contact stress of 1.31, this initial low torque value corresponded to a normalized contact stress of T p For Tests 2 and 3 under the operating conditions listed in Table 3.1, the schedule shown in Figure 3.3 with an initial small torque value was applied. This procedure prevented the onset of scuffing after each inspection for other tests at Θ= 141 C, and these tests endured until they were each suspended at 100 million cycles. Digital images of Test 2, shown in Figure 3.4, reveal no significant changes to the contact surfaces. Very small superficial scratches can be seen on the surfaces, perhaps attributable to small sized debris particles in the lubricant. The presence of these scratches is amplified by the lighting used by the microscope. The surfaces show discoloration at the addendum and dedendum regions of the teeth, which correspond to where sliding and contact pressures are simultaneously high. These areas most likely represent the formation of a tribo-film induced by lubricant additives that act to protect the surface from scuffing and surface fatigue, similar to the coloration seen by Olson [6]. Test 3, which also reached 100 million cycles, gave similar measurements and images, except the bands of discoloration were darker that what were seen in Test 2, as shown in Figure 3.5. Test 4, as defined in Table 3.1, was the first test at an elevated temperature of 49

65 2.5 2 Lambda Ratio, Normalized Contact Stress, Normalized Torque, Normalized Pinion Speed Lambda Ratio Normalized Contact Stress Normalized Torque Normalized Pinion Speed Time(s) Figure 3.3 Variation of normalized pinion speed, pinion torque, contact stress and the resultant lambda ratio values during the ramping stage of the modified loading procedure..

66 a) b) 51 c) d) Figure 3.4 Digital images of the chemically polished pinion of Test 2 at load level L2 and 141 C at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles.

67 a) b) 52 c) d) Figure 3.5 Digital images of the chemically polished pinion of Test 3 at load level L3 and 141 C at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles.

68 Θ= 155 C. This test experienced scuffing using the same procedure that was previously successful at an oil temperature of Θ= 141 C, and was suspended at 0.15 million cycles. Figure 3.6 shows the digital images for the scuffed pinion and gear surfaces. The areas that experienced scuffing exhibit a very high local surface roughness compared to the rest of the chemically polished surface. This sudden jump in roughness is shown in Figures 3.7 and 3.8, which shows the roughness profile of the pinion and gear, respectively. The roughness profile of the gear does not show any signs of an increase in roughness, but this is due to the area of the gear that can be analyzed with the Form Talysurf 60 profilometer. The tip region of the gear surface was difficult to measure without exceeding the vertical range of the roughness probe. In order to facilitate long cycle results at this temperature, an additional break-in period was introduced prior to applying the full load. Before each testing interval of 20 million cycles at the full load, a 1 million cycle break-in period at approximately half of the full load T p was applied. This meant that a total of three load stages (first a very low load stage, as shown in Figure 3.3, to ease the contact conditions during the speed ramp up, then a half-load 1 T 2 p stage for 1 million cycles, and finally the set load T p for 20 million cycles) were programmed in the computer controller for each test. This prevented the onset of scuffing upon start-up, and allowed all three load levels to be run with oil inlet temperatures of Θ= 155 C. All tests at Θ = 155 C with this procedure (Tests 5, 6, and 7 in Table 3.1) reached 100 million cycles without signs of failure. 53

69 a) b) 54 c) d) Figure 3.6 Digital images of one of the teeth of the chemically polished tooth surfaces from Test 4 at load level L1 at 155 C. (a,b) an initial (0 cycles) and failed (0.15M cycles) gear tooth surface; (c,d) an initial (0 cycles) and failed (0.15M cycles) pinion tooth surface..

70 2 1 0M Cycles Rq = µm 0-1 Surface Roughness Amplitude (µm) a) M Cycles Rq = µm 0 b) Trace Distance (mm) Figure 3.7 Measured roughness traces of the chemically polished pinion of Test 4 at load level L1 and 155 C. (a) Initial roughness profile and (b) roughness profile at 0.16M cycles.. 55

71 0.5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) M Cycles Rq = µm 0 b) Trace Distance (mm) Figure 3.8 Measured roughness traces of the chemically polished gear of Test 4 at load level L1 and 155 C. (a) Initial roughness profile and (b) roughness profile at 0.16M cycles.. 56

72 As stresses and temperatures rose, the discoloration of the contacting surfaces at each inspection period became more apparent. Digital images for Test 5 are shown in Figure 3.9. Figure 3.10 shows images of Test 7 that was the most extreme set of long cycle conditions in this study. As would be expected, the contact surfaces showed the darkest discoloration when compared to other test results. Roughness profiles for the pinion and the gear of Test 7 are shown in Figure 3.11 and Figure 3.12, respectively. Some reduction in surface roughness amplitudes are observed here after the first 20 million cycle test segment. The R q value of the pinion decreased from μm to μm over the span of the test while the R q value decreased from μm to μm for the gear tooth surface. After these decreases over the first 20 million test interval, the roughness remained nearly constant for the remaining 80 million cycles before test suspension. This decrease in the R q values for the specimens also resulted in nearly a 20% increase in λ over the span of the test. Gear profile traces of the pinion of Test 7 measured by using a gear CMM are shown in Figure They reveal essentially no wear between test inspections, even for this most extreme temperature and load condition. 3.3 Results from the Staged Scuffing Tests The staged scuffing tests performed, as well as the results from these tests, are summarized in Table 3.2. The initial test procedure, which was used for Tests 8, 9, and 10 of Table 3.2, used an oil inlet temperature of either Θ= 120 C 57 or 141 C and started

73 a) b) 58 c) d) Figure 3.9 Digital images of the chemically polished pinion of Test 5 at load level L1 and 155 C at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles.

74 a) b) 59 c) d) Figure 3.10 Digital images of the chemically polished pinion of Test 7 at load level L3 and 155 C at (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100M cycles.

75 M Cycles Rq = µm Surface Roughness Amplitude (µm) a) b) M Cycles Rq = µm 60M Cycles Rq = µm c) M Cycles Rq = µm d) Trace Distance (mm) Figure 3.11 Measured roughness traces of the chemically polished pinion of Test 7 at load level L3 and 155 C after (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100 M cycles. 60

76 M Cycles Rq = µm Surface Roughness Amplitude (µm) a) b) M Cycles Rq = µm 60M Cycles Rq = µm c) M Cycles Rq = µm d) Trace Distance (mm) Figure 3.12 Measured roughness traces of the chemically polished gear of Test 7 at load level L3 and 155 C after (a) 0 cycles, (b) 20M cycles, (c) 60M cycles, and (d) 100 M cycles. 61

77 Involute Deviation [ µm] Cycles 20M Cycles 60M Cycles 100M Cycles Roll Angle [degrees] Figure 3.13 Variation of measured tooth profiles of the chemically polished pinion during Test 7 at load level L3 at 155 C.

78 Table 3.2 Summary of all staged scuffing tests performed in this study. Test Number Surface Finish Maximum Load Level Lubricant Temperature [ C] 8(*) Chemically Polished L (*) Chemically Polished L ( ) Hard Ground L ( ) Hard Ground L ( ) Hard Ground L ( ) Hard Ground L (*) Suspended without failure ( ) Scuffing failure ( ) Micro-pitting failure 63

79 at load level L6 (corresponding to the normalized pinion torque value of T p = 1.96 ), as defined in Table 2.2 in Chapter 2. The speed and torque values were initially increased simultaneously to the conditions of the first load stage after the lubricant preheating stage had finished, as was done for Test 1 in the long cycle test results. After every 3 million cycles of testing at a given load level, the torque T p would be increased to the next specified value as shown in Figure 2.9. If a test reached the end of the L12 load step (corresponding to T p = 3.03) without scuffing, then is was suspended without failure. Tests 8 and 9 were both done with chemically polished specimens. The only difference between these two tests was the oil inlet temperature. Test 8 used Θ= 120 C, and Test 9 used Θ= 141 C. Figure 3.14 shows tooth surface images of the gears for Test 8, and Figure 3.15 shows the same for Test 9. These tests did not experience any form of failure, even at the maximum stress at the maximum achievable load level L12 in Table 2.2 in Chapter 2. The roughness amplitudes of the tooth contact surfaces during these tests were observed to decrease to values that were typical during the long cycle tests. Figures 3.16 and 3.17 show the profile differences between the pinion and gear, respectively, for Test 9. Due to the reduction in roughness amplitude, Test 9, for example, experienced an increase in lambda ratio of 25% between the first and last loading levels. No signs of micro-pitting were detectable after the loading profile for either test, which corresponded to a total of 21 million cycles at full speed. This agrees with Olson [6] that reducing surface roughness amplitudes through chemical polishing eliminates the possibility of formation of micro-pits. 64

80 a) b) 65 c) d) Figure 3.14 Digital images of one of the teeth of (a,b) the gear and (c,d) the pinion before and after Test 8 at 120 C.

81 a) b) 66 c) d) Figure 3.15 Digital images of one of the teeth of (a,b) the gear and (c,d) the pinion before and after Test 9 at 141 C

82 0.5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L12 Rq = µm 0 b) Trace Distance (mm) Figure 3.16 Measured roughness traces of the chemically polished pinion of Test 9 at 141 C (a) before and (b) after the test. 67

83 0.5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L12 Rq = µm 0 b) Trace Distance (mm) Figure 3.17 Measured roughness traces of the chemically polished gear of Test 9 at 141 C (a) before and (b) after the test. 68

84 Test 10 was performed at Θ= 120 C by using hard ground set of gears having significantly higher roughness amplitudes. Digital images of a selected tooth before and after testing are shown in Figure The images before testing show the marks along the face width of the tooth caused by the grinding operation. This differs from gears that have been chemically polished, as those specimens show no signs of grinding marks as well as no directionality to surface roughness (i.e. roughness profiles are isotrophic). The lambda ratio during the first load step was 0.19, nearly an order of magnitude lower than those for the chemically polished tests. This test scuffed on start-up, suggesting that the first load step was too high for a hard ground surface without any break-in stage. No test was completed at an elevated oil inlet temperature of likely at the more extreme conditions. Θ= 141 C, as scuffing would be In order to prevent scuffing on initial start-up, the first load step for Test 11 was reduced to L2 (corresponding tot p = 1.15 ) as defined in Chapter 2 for the staged scuffing tests. This test did not fail due to scuffing. Instead, micro-pitting was the failure that caused the test to be suspended. After 10.3 million cumulative cycles, the recorded vibration levels exceeded preset allowable limits to suspend the test. With a 3 million cycle interval between load steps, the final result is similar to what was seen by Olson [6] with SAE 4118M gear steel before macro-pits eventually formed on the surfaces. Figure 3.19 shows images of the gear and pinion, and heavy micro-pitting is evident in the dedendum of the pinion after 10.3 million cycles. In contrast to the tests completed with chemically polished gear sets, the roughness profiles for hard ground specimens typically 69

85 a) b) 70 c) d) Figure 3.18 Digital images of one of the teeth of the ground tooth surfaces from Test 10 at 120 C. (a,b) An initial (0 cycles) and failed gear tooth surface; (c,d) an initial (0 cycles) and failed pinion tooth surface.

86 a) b) 71 c) d) Figure 3.19 Digital images of one of the teeth of the ground tooth surfaces from Test 11 at 120 C. (a,b) An initial (0 cycles) and micro-pitted gear tooth surface; (c,d) an initial (0 cycles) and micro-pitted pinion tooth surface.

87 showed that the roughness peaks were rounded off during the test, and the roughness profile for Test 11 clearly show this in Figures 3.20 and 3.21 for the pinion and gear, respectively. This test could not be run past load level L5 ( T p = 1.75 ), as the micropitting had greatly changed the involute profile of the specimens over 10.3 million cycles. This caused uncontrollable vibrations in the test gear box that would damage the test machine if it were allowed to continue at high speeds. The test procedure was changed one final time for Tests 12 and 13. The testing interval was reduced to 30 minutes, which corresponded to 0.36 million cycles at 12,000 rpm for each load step. This time period allowed enough time for surface and bulk temperatures to stabilize while the number of cycles accumulated were too low for significant progression of micro-pitting like what was seen on Test 11 to take place. A short break-in period was also added at load level L1 ( T p = 1.00 ). This break-in was defined to last for one million cycles before moving to the 360,000 cycle intervals for load levels at L2 and above. Any test that endured all the way to the end of load level L12 in this schedule corresponded to a total of 4.96 million cycles which is about 5 million cycles lower than the number of cycles required to form micro-pits in Test 11. Both of these tests with hard ground specimens produced scuffing failures. Tooth surface images of gears of Test 12 that was performed at Θ= 120 C are presented in Figure The initial λ value for this test was 0.21, which was similar to each hard ground test that was done previously. This test scuffed at load level L5 ( T p = 1.75 ), which was the same load at which Test 11 was suspended due to micro- 72

88 5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L5 Rq = µm 0 b) Trace Distance (mm) Figure 3.20 Measured roughness traces of the ground pinion of Test 11 at 120 C (a) before and (b) after the test. 73

89 5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L6 Rq = µm 0 b) Trace Distance (mm) Figure 3.21 Measured roughness traces of the ground gear of Test 11 at 120 C (a) before and (b) after the test. 74

90 a) b) 75 c) d) Figure 3.22 Digital images of one of the teeth of the ground tooth surfaces from Test 12 at 120 C. (a,b) An initial (0 cycles) and scuffed gear tooth surface; (c,d) an initial (0 cycles) and scuffed pinion tooth surface.

91 pitting. The measured roughness profiles of the pinion and gear of this test are shown in Figures 3.23 and 3.24, respectively. While the pinion roughness profiles were typical of hard ground surfaces, gear profiles exhibited additional waviness, even with the same low frequency cut-off used in every other hard ground roughness measurement. Test 13 was performed at Θ = 141 C. The roughness profiles, as shown in Figures 3.25 and 3.26 for the pinion and gear, respectively, show that this gear set had the lowest root-mean-square roughness of all the hard ground tests. However, due to the higher oil inlet temperatures, λ was still very low at This test resulted in scuffing at the start of the L11 load step ( T p = 2.85 ). In Figure 3.27, images of the failed teeth show that the scuffing damage spread across nearly the entire face width of the pinion. The measured involute tooth profiles of the pinion and gear of Test 13 are shown in Figures 3.28 and 3.29, respectively. No significant wear or changes to the profiles are evident in these profiles, which was the most extreme test run for the hard ground gear sets. The only deviations from the starting profile are located where the scuffing occurred. Material in this test was removed as the surfaces repeatedly welded together and subsequently were torn apart. Material removal occurred in the dedendum of the pinion and addendum of the gear. The pinion damage is shown in Figure 3.28 by a dip in the involute profile near a roll angle of 16. The gear is damaged at the start of the tip relief modification, and it can be seen near a roll angle of 30 in Figure

92 5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L5 Rq = µm 0 b) Trace Distance (mm) Figure 3.23 Measured roughness traces of the ground pinion of Test 12 at 120 C (a) before and (b) after the test. 77

93 5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L5 Rq = µm 0 b) Trace Distance (mm) Figure 3.24 Measured roughness traces of the ground gear of Test 12 at 120 C (a) before and (b) after the test. 78

94 5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L11 Rq = µm 0 b) Trace Distance (mm) Figure 3.25 Measured roughness traces of the ground pinion of Test 13 at 141 C (a) before and (b) after the test. 79

95 5 0M Cycles Rq = µm 0 Surface Roughness Amplitude (µm) a) Load Level L11 Rq = µm 0 b) Trace Distance (mm) Figure 3.26 Measured roughness traces of the ground gear of Test 13 at 141 C (a) before and (b) after the test. 80

96 a) b) 81 c) d) Figure 3.27 Digital images of one of the teeth of the ground tooth surfaces from Test 13 at 141 C. (a,b) An initial (0 cycles) and scuffed gear tooth surface; (c,d) an initial (0 cycles) and scuffed pinion tooth surface.