Fretting in Wind Power Pitch Bearings: Micro-Slip Experiments and Bearing Test Rig Design

(1)

Fretting in Wind Power Pitch Bearings:

Micro-Slip Experiments and Bearing Test Rig Design

Román de la Presilla

Master of Science Thesis TRITA-ITM-EX.2020:186 KTH Industrial Engineering and Management

Machine Design SE-100 44 STOCKHOLM

(2)

(3)

Examensarbete TRITA-ITM-EX.2020:186

Fretting i lager till vindturbinsblad:

mikroglidningsexperiment och konstruktion av en lagerprovningsrigg

Román de la Presilla Romandlp@kth.se

Godkänt

2020 – 05 – 27

Examinator

Ulf Sellgren ulfs@md.kth.se

Handledare

Sergei Glavatskih segla@kth.se

Uppdragsgivare

Axel Christiernsson International AB

Kontaktperson

Johan Leckner

johan.leckner@axelch.com

Sammanfattning

Vindkraft är idag det snabbast växande området för grön elproduktion i Europa och står med 100 000 installerade turbiner för 15% av den totala elförsörjningen. Denna otroliga utvecklingen har berott på en massiv teknologisk insats som måste fortsätta. För att nå Europakommissionens miljömål för 2050 måste expansionen av grön elproduktion och vindkraft till och med trappas upp.

Nyligen har en mer aktiv individuell reglering av rotorbladen, vilket möjliggör att bladen kan styras in- och ut ur vinden, visat sig kunna reducera lasterna på blad och andra komponenter avsevärt, vilket därmed möjliggör stora kostnadsreduceringar.

Dessa justeringar möjliggörs genom att rotorbladen ansluter hubben via ett rotorbladslager. Dessa nya lastreducerande reglerstrategier tvingar dock lagren att arbeta under högre belastning jämfört med traditionell reglering av rotorbladens lutningsvinkel. Det här sker genom mer frekvent positionering och ofta som små oscillerande rörelser, vilket leder till en högre risk för slitage på rotorbladslagren, som i sin tur kan leda till förlust av rotorbladsregleringen. När så sker kan inte längre en säker reglering av turbinen garanteras och katastrofala fel är möjliga, så som förlust av rotorblad.

Det här projektet avser att utarbeta en design för en lagerprovningsrigg som kan användas för att testa rullager med kontaktvillkor som efterliknar de som återfinns i rotorbladslagren. Ett nytt koncept,m som är baserat på en ramlös motor, presenteras. Konceptet avser att förhindra onödigt slitage hos testriggens motorlager och förbättra de dynamiska egenskaperna för en given motorkapacitet. Projektet innefattar även en studie av friktionsbeteendet hos olika smörjmedel under små upprepande tangentiella rörelser, som utförts med en befintlig testrigg på KTH.

Sökord: Fretting, Rotorbladslager, Smörjning

(4)

(5)

Master of Science Thesis TRITA-ITM-EX.2020:186

Fretting in Wind Power Pitch Bearings: Micro-Slip Experiments and Bearing Test Rig Design

Román de la Presilla Romandlp@kth.se

Approved

2020 – 05 – 27

Examiner

Ulf Sellgren ulfs@md.kth.se

Supervisor

Sergei Glavatskih segla@kth.se

Commissioner

Axel Christiernsson International AB

Contact person

Johan Leckner

johan.leckner@axelch.com

Abstract

Wind power is the fastest-growing form of green energy production in Europe, today accounting for 15% of the total power demand with 100.000 turbines installed. This tremendous development relied on a massive technological undertaking that must be continued, and even accelerated in order to meet the European Commission’s environmental goals for 2050. Currently, more active individual control of the rotor blades, turning the blade into and out of the wind, has proven its ability to reduce structural loads on the blades and other components significantly, therefore paving the road towards strong cost reductions. To allow for such adjustment, the rotor blades are connected to the rotor hub via pitch bearings. However, these new structural load reduction control strategies force the pitch bearings into a much more demanding operation condition. More frequent positioning activity and often in the form of smaller oscillating motions, when compared to traditional pitch control. This leading to an increased risk of wear damage of the pitch bearing that could fully incapacitate the blade control. At which point the safe regulation of the turbine can no longer be guaranteed and catastrophic failure, such as the loss of a rotor blade, is possible.

This project pertains to the design a bearing test rig that can be used to test rolling element bearings with contact conditions that emulate those found in pitch bearings. A novel frameless motor-driven concept is proposed. The concept is aimed towards preventing unnecessary damage of non-test bearings and improving the dynamic performance of the test rig for a given motor capacity. One further objective of the project involved using an existing KTH single contact test rig to study the friction behavior of different lubricants when minute reciprocal tangential displacements are imposed.

Keywords: Fretting, Pitch Bearing, Lubrication.

(6)

(7)

FOREWORD

I extend my most sincere gratitude to both my academic supervisor Professor Sergei Glavatskih and my industrial supervisor Dr. Johan Leckner. The advice, feedback, perspective and passion you shared with me before and during the project were priceless. I also want to express tremendous gratitude towards the professors at Machine Design Department, for diligently sharing your knowledge and experience during these last two years. I am also grateful to Professor Stefan Bjorklund and Staffan Qvarnström for generously taking the time to think out loud with me.

Additionally, I would like to thank Tomas Östberg for his support in constructing components used in the experimental set up and to Fabian Schwack for lending his uniquely relevant experience whenever it was needed. To my family, for their unrelenting support and guidance, I express profound gratitude.

Finally, Sasha. Thank you for being my center.

Román de la Presilla Stockholm, May and 2020

(8)

(9)

NOMENCLATURE

Notations

Symbol Description

A Peak to peak amplitude of imposed displacement.

Ad Imposed displacement amplitude

As Sliding amplitude

D Hertzian contact diameter

r’ Radius of slip region

r Hertzian contact radius

s Slip Ratio

Sc Test rig and contact stiffness

Q Tangential Load

W Normal Load

δ Slip Index

μ Friction Coefficient

Abbreviations

CAD Computer Aided Design

IPC Individual Pitch Control

RCFM Running Condition Fretting Map MRFM Material Response Fretting Map

PSR Partial Slip Regime

MR Mixed Regime

GSR Gross Sliding Regime

(10)

TABLE OF FIGURES

FIGURE 1.1:OFFSHORE WIND TURBINE DURING DEPLOYMENT PROCESS. ... 11

FIGURE 1.2:WIND TURBINE CUTAWAY SCHEMATIC. ... 11

FIGURE 2.1:FRETTING CORROSION OF BEARING RING AND SEAT (TRIBOLOGY-ABC.COM). ... 15

FIGURE 2.2:FRETTING DAMAGED THRUST BEARING RACEWAY. ... 17

FIGURE 2.3:NORMAL AND TANGENTIAL STRESS FIELDS FOR AN ELASTIC CONTACT WITH AND WITHOUT SLIP (EXTRACTED FROM [12]) ... 19

FIGURE 2.4SURFACE STRESS DISTRIBUTION IN AN ELASTO-PLASTIC FRETTING CONTACT [12]. ... 20

FIGURE 2.5REYNOLDS SLIP. ... 20

FIGURE 2.6:FOUR SIMPLE FRETTING MOTION MODES FOR BALL-ON-FLAT CONTACT [37]. ... 21

FIGURE 2.7:FRETTING MAP AND FRETTING LOOPS (CROSS SECTIONS). ... 22

FIGURE 2.8:RUNNING CONDITION FRETTING MAPS &MATERIAL RESPONSE FRETTING MAPS. ... 23

FIGURE 2.9:FRETTING WEAR RATE AS A FUNCTION OF DISPLACEMENT (A) AND FRETTING REGIME AS A FUNCTION OF NORMAL LOAD AND DISPLACEMENT [38]. ... 25

FIGURE 2.10:TYPICAL FRETTING LOOP DIAGRAM [48]. ... 26

FIGURE 2.11:AMPLITUDE RATIO.A=OSCILLATION AMPLITUDE;D=HERTZIAN CONTACT DIAMETER [83] ... 30

FIGURE 2.12:FRETTING WEAR IN RECIPROCAL SLIDING OF BALL ON FLAT FOR VARYING VISCOSITY AND AMPLITUDE RATIO, MARUYAMA, ET AL.[83]. ... 30

FIGURE 2.13:THICKENER LAYER DIAGRAM [95]. ... 32

FIGURE 2.14:FALEX FRETTING WEAR TESTER, AS PRESENTED IN [104]. ... 33

FIGURE 2.15SRVMACHINE AND DIAGRAM [105] ... 34

FIGURE 2.16:FRETTING TEST RIG WITH MAGNETORESTRICTIVE ACTUATION AND MULTIPLE CONTACT CONFIGURATIONS [106]. ... 35

FIGURE 2.17:FRETTING TEST RIG WITH ELECTRIC SHAKER ACTUATION THROUGH A LEVER ARM [107]. ... 35

FIGURE 2.18:FRETTING TEST RIG FOR HIGH TEMPERATURE TESTING [108]. ... 36

FIGURE 2.19:LARGE SCALE PITCH AND YAW BEARING TEST RIG [109]. ... 36

FIGURE 2.20:BEARING FRETTING TEST RIG AS SHOWN IN MARUYAMA, ET AL.[95]. ... 37

FIGURE 2.21:BEARING FRETTING TEST RIG USED BY SCHWACK IN [24&97] ... 38

FIGURE 2.22:SCHEMATIC OF BEARING FRETTING TEST RIG USED BY SCHWACK IN [24&97]... 38

FIGURE 2.23:BEARING TEST RIG USED BY STAMMLER, ET AL. IN [113]. ... 39

FIGURE 2.24:WIND TURBINE BLADE, HUB AND PITCH BEARING [110]. ... 40

FIGURE 2.25:TYPICAL PITCH BEARING.(REFERENCE TURBINE IWT164-7.5-MW– EXTRACTED FROM [97]) ... 41

FIGURE 2.26DISPLACEMENT RATIO DIAGRAM [97]. ... 41

FIGURE 2.27:FOUR-POINT CONTACT INTO ANGULAR CONTACT SIMPLIFICATION. ... 42

FIGURE 3.1:FUNCTIONAL BREAKDOWN DIAGRAM. ... 44

FIGURE 3.2:EARLY CAM CONCEPT ... 45

FIGURE 3.3:CONVENTIONAL SERVOMOTOR DIRECT DRIVE ... 45

FIGURE 3.4:FRAMELESS DIRECT DRIVE MOTOR APPROACH ... 46

FIGURE 3.5:ARRAY OF KOLLMORGEN KBMFRAMELESS MOTORS IN DIFFERENT SIZES AND ASPECT RATIOS. ... 46

FIGURE 3.6:HOLLOW SHAFT AND SINGLE SCREW AXIAL LOADING. ... 47

FIGURE 3.7:INDUCTIVE ENCODER ... 48

FIGURE 3.8BIOMECHANICS FORCE PLATE ... 48

FIGURE 3.9:MOTOR PERFORMANCE ... 50

FIGURE 3.10:MICRO SLIP TEST RIG DIAGRAM ... 51

FIGURE 4.1:BEARING TEST RIG RENDER ... 52

FIGURE 4.2:CROSS SECTION VIEW OF THE BEARING TEST RIG. ... 53

FIGURE 4.3:BEARING MOUNTING WEDGE NUT AND TAPERED BUSHING. ... 54

FIGURE 4.4:FRAMELESS MOTOR CASE WITH SHAFT CLAMPING ATTACHMENTS ... 55

FIGURE 4.5:ENCODER MOUNTING ... 55

FIGURE 4.6:SHAFT AND MOTOR COUPLING. ... 56

FIGURE 4.7:LOAD SENSOR ARRAY ... 57

FIGURE 4.8:AKDSERVO DRIVE ... 57

FIGURE 4.9:LIX100 AND PAO100(MEAN CONTACT PRESSURE =0.64GPA /AMPLITUDE=50UM /A/D=0.63) ... 59

FIGURE 4.11:PP100 AND PAO100(MEAN CONTACT PRESSURE =0.64GPA /AMPLITUDE=50UM /A/D=0.63) ... 61

FIGURE 4.12:PP10 AND PAO10(MEAN CONTACT PRESSURE =0.64GPA /AMPLITUDE=50UM /A/D=0.63). ... 62

FIGURE 4.17:LIX10,PP10,LIX100 AND PP100(MEAN CONTACT PRESSURE =1.2GPA /AMPLITUDE=50UM /A/D=0.33) ... 67

(12)

1 INTRODUCTION

This section contains an overview of the project and highlights its relevance within the larger picture of enabling necessary cost effectiveness improvement of renewable energy.

1.1 Background and Problem Definition

According to the International Renewable Energy Agency, wind power deployment will play a fundamental role in achieving the Paris Climate Targets and it can cover, if accompanied by deep electrification, more than one-third of the energy-related carbon dioxide emissions reductions that are necessary by 2050 [1]. The same report, however, also clarifies that an accelerated deployment is necessary. This increase in deployment speed is fundamentally reliant on ongoing innovations and technology enhancements towards larger-capacity turbines that will continue to improve the cost-effectiveness of this energy production method. Thus, the colossal technological undertaking that led to wind currently being the fastest-growing form of renewable energy in Europe, today representing 15% of its power demand [2], must be continued and even accelerated in order to match the EU Commission’s vision of wind being half of Europe’s power in 2050. The need for this deployment acceleration and technological advancement becomes evident if the data for new offshore/onshore installations for 2019 is evaluated: even though the yearly total installation was a historical record amount, neither offshore or onshore growth rates are currently sufficient to meet the Green Deal [3].

Alongside crucial improvements on foundation design strategies, gearbox reliability, condition monitoring and a multitude of other aspects of wind turbine design; there is a latent idea that has the potential to result in major leaps in cost reduction that will be particularly amplified by the continued progression towards larger turbines. This idea involves individual control of blade pitch angles, often referred to as IPC: Individual Pitch Control; meaning that the angle of the blade along its long axis is controlled independently of other neighboring blades, see figure 1.2. Although the seminal idea of pitch control was developed in the 1980s, during that time it was thought of as a way to limit system-wide overloads, as opposed to stall-regulated wind turbines, and all of the blades were collectively controlled [4]. Currently, IPC implementations have proven their capacity to reduce blade root fatigue loads by 15%-30%, hub and shaft loads by 20%-40% and tower top moments by 10% [5]. These results are also backed up by alternative sources [6, 7]. Load reductions of such magnitude imply that, by cleverly turning the rotor blades into/out of the wind with an individual pitch control system, not only is power generation controlled, but also the rotor blades are subjected to lower fatigue loads and can be made narrower. Therefore, allowing for material savings in the blades (which represent 22.2% of the cost of the whole system) as well as directly impacting the loads on the tower (that represents 26.3% of the total cost) [8]. Furthermore, a great number of very promising control strategies and architectures are being put forward in the literature [9]. Currently, most wind turbine manufacturers are either using or planning on using IPC in their commercial turbines [5]. This surge of interest is rooted in the fact that, as the trend towards larger-scale individual turbines continues, the cost-effectiveness gains from reducing structural loads become progressively more significant and more aggressive pitch control reductions are becoming desirable [6,7].

(13)

Figure 1.1: Offshore Wind Turbine during deployment process.

Principle Power. Artist: Dock90

Figure 1.2: Wind Turbine Cutaway Schematic.

(14)

However, in order to unleash the full potential of this technology in large scale practical applications, an age old fundamental problem must be tackled first. Modern IPC involves frequent small oscillating motions of the rotor blades and their corresponding pitch bearings, which attach the blade to the hub allowing the rotation necessary for pitch variations. These small displacements are prone to cause fretting damage in the bearing raceways; drastically and dangerously reducing the life of the bearings and entirely defeating the cost reduction purpose of IPC. The rolling elements subjected to small reciprocal displacements push the lubricant out of the contact interface, metal to metal contact occurs, a complex mixture of wear mechanisms start to take place within the contact and a drastically accelerated damage process begins. Not only are there considerable uncertainties in the current estimation of fatigue damage for small cyclic pitch movements in bearings [10], in worst case scenarios, the pitch bearing can fail even before a fatigue crack has enough time to develop significantly due to wear altering the geometry of the raceway to such an extent that the torque provided by the IPC actuator is not enough to rotate the blade and catastrophic failure ensues, for aerodynamic regulation of the wind turbine becomes non-existent.

Grease lubrication is a fundamental factor in maximizing bearing life by reducing friction and protecting metal surfaces from damage. At the moment, however, there is no clear consensus on a lubrication strategy that satisfactorily attenuates this type of fretting damage in bearings, thus enabling the application of IPC to its full potential, and which also simultaneously meets a plethora of demanding requirements of wind power pitch bearing lubrication. Furthermore, standardized testing for grease performance in terms of fretting damage is application specific, can be unrepresentative of current field tests and tends to show conflicting results [11].

Axel Christiernsson International AB, one of the most important lubricating grease producers in Europe, is currently interested in facilitating the technological advancement of the wind power sector by systematically developing greases that directly enable the implementation of IPC by attenuating fretting damage in pitch bearings. To this end, the present work pertains to the design of a test rig that can be used to induce controlled fretting conditions on lubricated rolling element bearings that possess contact conditions similar to those found in wind power pitch bearings. A further objective of the project is also to use the KTH Micro-Slip test rig, that emulates the bearing contact conditions with a singular contact, to obtain a better insight into contact dynamics when various lubricants are used in fretting inducing conditions.

An increased understanding of this small-oscillation driven bearing damage mode is necessary for the implementation of wind power IPC to its full potential, which would have a direct impact on our capacity to generate renewable energy with a significant competitive advantage. However, the benefits of an increased understanding in this area would not be limited to wind power. Industrial applications where small bearing oscillations are present, whether they are necessary or functional oscillations, as in robotics, motion control and prosthetics; or whether they are unwanted by- products of normal operation or transportation, would also benefit from an increased bearing life and reliability with respect of this damage mechanism, resulting in both sustainability and cost reduction advantages. In fact, fretting related failures are the only type of tribological failure in industry that has not decreased its incidence rates during the past decades [12]. Thus, it represents a formidable and relevant technological challenge that has been limiting engineering design options to solutions that often tend to traverse great lengths to circumnavigate this persistent fundamental issue.

(15)

1.2 Purpose and Research Questions

The main goal of this project is to further our understanding of fretting damage in pitch bearings and the influence that lubrication has on this phenomenon. To that end, two distinct main objectives and their respective primary inherent research questions are formulated as follows:

1) Design of a novel bearing test rig to evaluate fretting damage progression in rolling element bearings: This test rig has the purpose of facilitating the development of greases better suited to fretting damage inducing operating conditions.

1.1) What should be the design requirements for this machine?

1.2) How to contain the fretting phenomena so that it only occurs within the test specimen and not throughout the test machine?

1.3) How to achieve reliable operation and sensing in harsh environmental conditions like high humidity or low temperature.

1.4) What kind of smaller scale bearing better emulates the contact conditions and fretting regimes present in modern pitch bearings?

2) Use the existing KTH Micro-Slip Test Rig to obtain a better insight into fretting contact dynamics when different lubricants are employed.

2.1) What is the effect that lubricant composition has on friction within the contact in fretting conditions?

1.3 Delimitations

- The bearing fretting test rig to be designed (Objective 1) is not going to be manufactured or physically validated during the course of this work.

- The implementation of a virtual instrument associated with the test rig to be designed (Objective 1) is outside of the scope of this work.

- Deliverables from the design of the new test rig (Objective 1) are limited to a report of the design process and outcome.

(16)

- The experiments using the KTH Micro-Slip test rig (Objective 2) are limited to analysing the fretting behavior of a finite set of representative lubricants. Without the use of any additive packages, only in their friction-displacement performance and limited to smooth steel-steel contacts.

1.4 Methods

The design process of the new bearing test rig, specified as Objective 1, will follow a functional breakdown design process in which every function and sub-function of the test rig will be identified and a variety of means will be evaluated and cross-evaluated to obtain the desired functionality. It is intended that, with this approach, a modular solution better suited to be operated in cold and humid environments will be attained. All of the design process is backed up by a literature review. Pertinent calculations will be performed in Matlab. CAD needs will be addressed via Solid Edge and any subsequent necessary modelling will be handled via a suitable software package, such as ANSYS or ADAMS.

Objective 2, which pertains to the use of the KTH Micro-Slip Test Rig, will be achieved via an experimental methodology. Where a literature review is conducted to contextualize the experiment within a larger scientific body of work. Experiments are systematically designed, executed according to a set protocol and analysed and, subsequently, results and discussion are presented.

(17)

2 FRAME OF REFERENCE

A summary of the existing knowledge and former performed research on the subject of fretting is presented.

2.1 An Introduction to Fretting Damage

Fretting arises wherever small-amplitude reciprocating displacements between contacting surfaces occur in a prolonged manner. These displacements cause surface wear and deterioration of the fatigue life of the affected components. Even contacts which exhibit no apparent macroscopic relative motion, such as bolted joints and press-fits, when loaded, inevitably allow micro-sliding (so called because it is in the order of micro-meters); thus potentially giving rise to fretting phenomena. The extent of the surface damage is much greater than what the microscopic slip magnitudes would intuitively suggest [12].

Figure 2.1 shows fretting damage that occurred at the contact interface between an inner bearing raceway and the bearing seat in the shaft. In this case caused by a relatively loose fit. Small wear particles located at the interface, product of micro slip and adhesive wear, will oxidize and subsequently start a process of three body abrasion. The surface damage also facilitates crack initialization.

Figure 2.1: Fretting Corrosion of bearing ring and seat (Tribology-abc.com).

Since machines, in the great majority of cases, contain a significant number of contacts potentially subjected to repetitive small-amplitude displacements, it is perhaps not surprising that the first documented case of fretting took the form of an accidental inconvenience during early twentieth century fatigue testing. The first documentation referring to fretting damage is found as a single paragraph in Eden, et al. [13] in their 1911 paper titled “Endurance of Metals”. During the

(18)

execution of fatigue tests, it was observed that oxide debris was being generated in the steel specimen fixtures of their fatigue testing rig. The sample holders were designed to allow for shorter and easier to manufacture samples to be used in the fatigue tests. Of the corrosion issue encountered in their fatigue testing Eden, et al. wrote the following:

“ Corrosion of Test-pieces and Holders: A good deal of trouble was caused by corrosion where the test-specimen fits into its holders. The rusting at this point was so great in the case of heavily loaded tests lasting for several days that it was often difficult to remove the broken specimen from the holder. The same trouble was experienced in an earlier repeated load's testing machine at University College. This machine was of the Wohler rotating cantilever type, with round test- specimens, 1 inch diameter, fitted into conical holes in the testing-machine spindle. In this machine red rust was produced in considerable quantity. There seems to be no doubt that this corrosion is due to the varying stress between the test-specimen and its holder, and the authors believe that the rust often found between a wheel and its shaft or a key and key-way, and the corrosion of ball- bearings is usually due to a similar action and not only to the presence of moisture. After the first few trials the specimens were well oiled before being put in place, but this was only partially effective. The specimens were somewhat easier to withdraw, but rust was still produced. ”

It was however not until 1927 that Tomlinson [14], having attributed minute displacements at the contact interface as the cause of previous observations, designed test equipment in order to study fretting phenomena intentionally. Tomlinson coined the term “Fretting Corrosion”. His work also showed that the fretting damage could be caused by displacement amplitudes as minuscule as 125 nanometers.

The first investigations regarding the influence that fretting damage can have on fatigue strength where conducted in 1941 by Warlow.Davies [15]. Fretting damage was introduced into steel fatigue specimens and posterior fatigue testing revealed that fatigue strength reductions between 13% and 17% had occurred. Which was to be expected because of the surface deterioration initially imposed with fretting. Interestingly, it was later shown by McDwell [16], in 1953, that when fretting damage occurs simultaneously with fatigue loading, strength is reduced by factors of 2 to 5 and sometimes even larger. This simultaneous action, which was proved to be critically influential in fatigue life, is actually more representative of practical conditions. In 1958, Fenner and Field [17] showed that fretting significantly accelerated the crack initiation process. The first large scale systematic analysis of fretting fatigue was conducted by Nishioka and Hirakawa [18]

and this area of research still contains many open questions [19].

Fretting damage not only occurs in distributed conformal contacts and the resulting failure can be different from crack growth and subsequent fracture. Perhaps the most classic example of fretting damage in industry involves damage to automotive wheel bearings during rail-way, truck or maritime transportation. In 1937, Almen [20] found that wheel bearings were being damaged before the car even reached the customers. In fact, the damage was more pronounced in cars that were shipped to more remote destinations and was particularly similar to brinelling indentations.

It was thus termed false brinelling, for it was not a product of plastic overload of the contact, but rather micro-oscillations that occurred during shipping, as was later further demonstrated by Pittroff [21]. Figure 2.2 shows the type of damage that these small oscillations can produce on a thrust bearing raceway. This shipping issue prompted the development of greases that attenuate this phenomena and improvements in transportation conditioning and other strategies [22].

Research on this specific topic is being published even to this day [23] and fretting damage in bearings as a whole still contains many open questions. Whether the small displacements are

(19)

caused by unwanted vibrations, as in the car shipping scenario, or whether the small displacements at the contacting surfaces is inherent to the desired operation, such as wind power pitch bearings, fretting damage will occur and there are significant uncertainties and discrepancies in our currently established approaches to predict the life of the bearing [24].

Figure 2.2: Fretting damaged thrust bearing raceway.

Fretting damage induced failures are often seen in press fits, near the outer edges of the press fitted interface. Over 90% of the failures in freight car axles are reported to be initiated at the shaft-hub connection, where stress concentrations and fretting conditions are dominant [25]. Riveted joints have also been reported to have untimely failures related to fretting damage induced crack initiation [26]. Fretting damage in surgical implants, where the usual debris associated with fretting is particularly dangerous and premature failure has severe consequences, has also been the focus of many studies [27] and is currently undergoing persistent research; often related to the influence that the biological environment where the device is located has on fretting [28]. Other examples of fretting include, but are not limited to, electrical switch gears, sockets in integrated circuit boards subjected to periodic differential thermal stresses, wire rope (at the interface between the multitude of strands), electrical contacts in relays, switches and selectors, dovetail joints in turbines and numerous other examples are abundant [25].

Despite the fact that fretting phenomena has been the object of study for nearly a century, there are still ongoing efforts to standardize the field both in nomenclature and testing [29]. The terminology involved in fretting research can often lead to confusion. Fretting fatigue, fretting corrosion, false brinelling are terms that are closely related, but not the same. Fretting fatigue refers to the process of crack initialization due to surface defects resulting from fretting, and subsequent deterioration of the fatigue life (in fact, for steels there is often no endurance limit when fretting is combined with fatigue loading [12]). Fretting corrosion refers to the production, during a fretting process in an oxidative environment and with lack of lubrication, of oxide particle debris that often accelerates the wear process. Note that fretting occurs even with non-oxidizing materials [30].

(20)

False brinelling refers to brinelling-like indentations produced, not via plastic overload of the concentrated contact (as it is in real brinelling) but rather by fretting wear.

2.2 Fretting in Hertzian Contacts

Since the objective of the present work pertains to fretting in bearings, the fundamental theoretical background will be presented from the perspective of a Hertzian contact. In order to develop an understanding of fretting, the nature of microscopic slip, often referred to as micro-slip, and why it is bound to occur, must be explained as a starting point.

Micro-Slip in a Hertzian contact subjected to a tangential load

Two solids in contact, pressed together with a normal force and subsequently subjected to an increasing tangential load, will eventually begin to slide. This sliding will occur when the tangential load reaches a sufficiently large magnitude, which is widely expected. However, even at tangential force levels well below the threshold for gross macroscopic sliding, tangential micro- slip occurs as a result of the applied tangential load. This small sliding motion is an intrinsic feature of every Hertzian contact subjected to a tangential force [31,32] and is also instrumental for the occurrence of fretting damage in Hertzian contacts [12].

In a Hertzian contact interface, the normal stress has a maximum magnitude at the center of the contact patch and smoothly decays to zero at the edge of the contact. When a tangential load is applied, and it is assumed that no slip occurs throughout the entirety of the contact patch, the resulting tangential stress at the edge of the contact must rise asymptotically to infinity. A more reasonable outcome follows if it is instead assumed that, because at the boundary of the contact region the Hertzian normal pressure is zero, some peripheral slip must occur, even for minute tangential forces. Both Cattaneo [31] and Mindlin [32] independently came to the conclusion that the condition for the initialization of slip is that the tangential traction can’t exceed the product of the coefficient of friction and the normal pressure. Figure 2.3 illustrates this thought process rather clearly. It follows that if the normal load is held constant, and a tangential load is monotonically increased starting from zero, micro-slip will occur instantly at the perimeter of the contact patch and an annular area where slip is present will grow inwards, along with the increasing tangential force. This slip annular region will continue to grow inwards until the tangential load approaches the normal load times the friction coefficient and the no-slip or stick region collapses to a point (or a line), just at the onset of gross sliding.

Experimental evidence for the existence of these regions, a central region of stick and a peripheral region where slip occurs, can be found in [33, 34]. Sato [34] showed that for reciprocating sliding of a ball on flat, by virtue of varying the sliding amplitudes, different damage patterns emerge and they correlate with a growing slip annular region of damage for increasing displacement amplitudes.

(21)

Figure 2.3: Normal and tangential stress fields for an elastic contact with and without slip (Extracted from [12])

Mindlin [32] proposes that the radius of the central stick region is given by the following equation:

𝑟^′ = 𝑟 (1 − ^𝑄

𝜇𝑊)^1/3 (1)

Where “r-prime” is the radius of the stick region, “r” is the contact radius, “Q” is the tangential force, “W” is the normal load and μ is the static coefficient of friction. Note that as the normal load or the static coefficient of friction increase, the area of stick increases as well; while an increase in tangential load decreases the size of the stick region.

The model, as it is explained so far, assumes that contacting asperities form junctions that are rigid.

When loaded in shear past a critical value, slip occurs as a result of a sudden fracture of the asperity couples, without any previous elastic or plastic deformation. This simplification has led to inconsistencies with experimental results, particularly with regards of the maximum displacement amplitude that can be assimilated by the contact without gross slip, which in reality tends to be higher than the one predicted by this model [35]. This prompted the logical extension of this model to include elasto-plastic effects [36]. The extended model instead assumes that in between the previously described regions of stick and slip, lies an annular region of plastic deformation of the asperities which have not yet fractured. See figure 2.4 and compare it with figure 2.3-C, notice how in the transition of surface stress from stick to slip regions is now rounded as opposed to sharp.

(22)

Figure 2.4 Surface stress distribution in an elasto-plastic fretting contact [12].

Other mechanisms that result in Micro-slip

Although the previously discussed models are useful for developing an intuitive understanding of fretting phenomena, they are limited to a pure sliding tangential motion. Fretting can occur wherever micro-slip takes place and a tangential load in kinematic pure sliding is only one specific case of the plethora of conditions that can allow for micro-slip to occur, even exclusively within Hertzian contacts.

For example, figure 2.5 shows a ball (grey) pressed against a flat (blue). Exclusively under the influence of a normal load “W”, the ball and the flat will deform elastically. The curved contacting length, from point 1 to point 2, has reduced in length for the ball and increased in length for the flat. The counteracting strains at the mating surfaces will result in micro slip, which in this case is more specifically referred to as Reynolds Slip. Note, in figure 2.5, the stick region in the central part of the contact patch, where the contact pressure is high and thus the friction is sufficient to prevent sliding. This type of slip can occur under pure rolling and it also indicates that fretting behavior is likely to be obtained if the load “W” is repeatedly fluctuating.

Figure 2.5 Reynolds Slip.

(23)

Another example of slip that occurs in practical applications is Heathcote Slip, which occurs in rolling motion due to the counter surface being curved perpendicular to the direction of motion (as in deep groove ball bearings). In this case the slip is imposed by the variable rolling diameter in the contact. Furthermore, angular contact ball bearings and linear rolling guides also have an imposed spinning component to the motion of their rolling elements, although in the literature the main concern regarding spinning is related to frictional temperature increases, for small reciprocal displacements, this spinning constitutes another mechanism for the occurrence of fretting. It then stands to reason that the simple case of reciprocating pure sliding is a useful but significant simplification of numerous forms combined of micro-slip that occur in practical applications. A good summary of different simple fretting mechanisms in concentrated ball-on-flat contacts can be found in [37] and figure 2.6 elucidates the four principal fretting mechanisms for a ball on flat studied in [37].

Figure 2.6: Four simple fretting motion modes for ball-on-flat contact [37].

2.3 Fretting maps and fretting regimes

Fretting maps are a widely used as an approach to represent and classify experimental fretting data and are surprisingly useful in aiding in the understanding of fretting behavior. If, during a fretting experiment, displacement, friction force and time are recorded, it is then possible to produce a Fretting Log where the Z axis represents the friction force (or alternatively the friction coefficient), the Y axis represents the displacement and the X axis represents the time or alternatively the cycle number.

Figure 2.7 shows a running fretting log (C) for a lubricated contact and two sampled loops (cross sections – A & B) extracted at an initial stage and at a later stage of the 1000 cycle ball-on-flat tangential fretting test. Note that the shape of the fretting loop experiences quite a drastic shift as the number of cycles increases. This is type of progression is quite representative of the different friction behaviors that can be interpreted with this visualization approach. During the initial stage of the test (Figure 2.7-A) the fretting contact is gross sliding. As the displacement is progressively imposed, the contact and the test rig itself initially accommodate the forces by elastically deforming, micro-slip occurs. Subsequently, after the breakaway friction force magnitude is surpassed, and the micro-slip region has vanished the central sticking region, the contacting bodies proceed in gross sliding (horizontal top and bottom of the loop A). The resulting loop shape is analogous to a parallelogram. The loop is completed because the imposed motion is then repeated in the opposite direction. After the number of cycles increases, the friction force increases as well and the load required to achieve gross slip is not achieved, thus the contact is sticking throughout

(24)

the entirety of the loop and therefore the loop almost collapses into an inclined line (Figure 2.7- B).

In examining the fretting loop presented in 2.7 – A & B, note the slope “Sc”. This represents the tangential stiffness of the contact in series with the stiffness of the test rig in the direction of the imposed motion. Furthermore, when comparing the loops A with B, note that the axis are significantly changed, the stiffness “Sc” could be calculated with loop B and the result would be practically identical. Also note, that the area contained within the loops is actually the energy that is being dissipated per cycle and that the area within the collapsed loop B, where the behavior is macroscopically elastic, is related to the micro-slip occurring within the “stuck” contact.

Figure 2.7: Fretting map and fretting loops (cross sections).

Sc

A) B)

C)

(25)

By varying the normal force level and the imposed displacement amplitude, the friction and wear behavior change, and a Running Condition Fretting Map can be drawn, which offers a overall view of the different fretting regimes. Fretting maps were originally introduced by Vingsbo and Söderberg in 1988 [38]. Alongside Running Condition Fretting Maps, Material Response Fretting Maps are also remarkably illustrative. Figure 2.8 shows how varying the normal load and the displacement amplitude, ceteris paribus, changes the shape of the fretting loops, producing 3 distinct fretting regimes: Partial Sliding Regime (PSR), Mixed Regime (MR), Gross Sliding Regime (GSR). Furthermore, typical damages are associated with different regimes. It can also be seen that the data presented in figure 2.7 is in the mixed regime domain. The attentive reader might notice that wear damage seems to be associated with larger loop areas, meaning larger dissipated energy per fretting cycle. There is indeed a relationship between the dissipated energy and fretting wear but it is not as straight forward as the diagram in Figure 2.7 indicates, as there are a large number of variables that have an effect on the fretting process [39] and several considerations that will be addressed in the following section before discussing the relationship between friction work and fretting wear.

Figure 2.8: Running Condition Fretting Maps & Material Response Fretting Maps.

2.4 Fretting Wear

Fretting wear is a complex phenomenon. Although the basic processes that occur during fretting are analogous to those occurring in a sliding contact, the fact that the sliding distance is of similar magnitude or smaller that the contact size entails that ingress of environmental agents into the

(26)

contact and ejection of debris from the contact become important and difficult to manage considerations when analyzing the fretting process. Furthermore, there are a significant number of factors that have a stark influence of the development of the fretting process. Some of the main considerations are addressed in this following section:

Role of Oxidation and Debris

Although it might seem odd to address the role of oxidation and debris before discussing perhaps more evident factors, such as amplitude, frequency or contact geometry; it will become evident that most, if not all, of the other factors become important in part due to the effect that they impart on oxidation and debris ejection.

It has been shown repeatedly that oxidation processes greatly influence the behavior of fretting in metal-to-metal contacts [40]. Debris on carbon steel and other steels fretting in air is mainly finely ground and a red-brown colored α-Fe2O3. This oxide debris can be generated via to distinct mechanisms:

1. Oxide film that formed at the surfaces and it is removed, whereupon a new oxide layer grows.

2. Metallic debris particles are created and they are subsequently oxidized.

It is a possibility that both mechanisms of debris formation can occur simultaneously within different regions of the contact or that either of them alternatively dominates during the different stages of the process.

Although it is not unusual to find literature where this oxide debris production is often perceived as a wear accelerator, acting as an abrasive agent [12], there is also evidence that it sometimes acts as coherent protective debris bed that limits metal-to-metal contact and also accommodates a fraction of the imposed movement [41]. Interestingly, Colombie, et al. [42] showed that even chalk powder, arbitrarily added to the contact, was able to form a third body protective layer. They also observed that periodically stopping the fretting experiment to resume the process after the oxide debris had been removed increased wear. Iwabuchi, et al. [43] artificially placed oxide particles on a fretting interface and concluded that if a stable compact glaze film is generated, a reduction of wear is observed, otherwise it increases abrasive wear. On the other hand, it is also generally reported that oxygen and water accentuate fretting wear and surface damage and that an inert atmosphere suppresses fretting in metals [44,45]. An interesting analysis of this slightly paradoxical conundrum is conducted by Varenberg, et al. [44], where by allowing or preventing the oxide wear debris to escape from the interface, it was concluded that the role of debris depends on the dominant fretting wear mechanism: for conditions where adhesive damage will be dominant, oxide debris retention reduces wear and, alternatively, for conditions where abrasive wear will be dominant, oxide debris ejection reduces wear. Furthermore, studies where the effect of reduced air pressure on fretting of a variety of metals [45-47] show that low atmospheric pressures induce large friction coefficients and significantly accelerated adhesive damage to the fretting interfaces. Therefore, suggesting that the oxide film is of critical importance and that caution must be exercised in applications such as vacuum equipment and aerospace.

(27)

Slip Amplitude and Normal Load

In 1927 Tomlinson [14] demonstrated that relative motions, even those in the nanometer range, were the driving mechanism of fretting damage. Slipping amplitudes greatly affect the progression of fretting damage. All things remaining equal, by progressively increasing the sliding amplitude the fretting behavior will drive the contact to traverse all three regimes of fretting previously described in the Fretting Maps section, namely Partial Sliding Regime (PSR), Mixed Regime (MR), Gross Sliding Regime (GSR). Figure 2.9 shows typical damage mechanisms for each of the three regimes. The effect of normal load can also be deduced from Figure 2.9.

Figure 2.9: Fretting wear rate as a function of displacement (A) and Fretting regime as a function of Normal load and displacement [38].

However, it is not the absolute value of the sliding amplitude or the normal load that is relevant, but rather the relationship between the contact area dimensions and the displacement amplitude, often defined in the form of a Mutual Overlap Coefficient (MOC). The MOC is the ratio of the contact area to the area that will be covered by the contact patch during the displacement. Meaning that at low MOC values the original contact area is exposed, such as in reciprocating sliding wear (MOC<<0.1) and fretting contacts will show MOC values close to unity. Other alternative ratios that are fundamentally analogous are also employed. Nevertheless, most of the literature primarily

Fretting in Wind Power Pitch Bearings: Micro-Slip Experiments and Bearing Test Rig Design