Types of fretting and typical damages:
Fretting damage can take many different forms depending on the mechanics of the specific occurrence (Fig. "Load-specific wear types").
With steels (Fig. "Fretting nodes"), especially, special experience and expertise is necessary in order to correctly estimate the size of the damage-influencing fretting parameters from the fretting pattern.
The temporal progression of erosion is characteristic of the fretting process (Fig. "Fretting temporal progression")
Classifying the types of wear by their damaging loads:
This classification can be distinguished by relative movements and temporal force progression.
Wear due to sliding action:
The relative movement occurs evenly, while a constant force acts on the surface perpendicularly and tangentally. The wear process is predominantly abrasive (see “abrasive wear”).
The roll-off process creates an swelling pulsating force with complete relief of the strain at every cycle. This force acts upon the wear surface perpendicularly. A noticeable, damaging relative movement of the contact surfaces does no occur. The damage process occurs primarily through dynamic fatigue of the material accompanied by crack initiation and surface outbreaks (dimples, fatigue pittings).
The conact surfaces do not exhibit any relative macro-movement. However, the elastic deformations in the contact zones cause relative micro-movements. The direction of the swelling force is perpendicular to the contact surfaces. As opposed to rolling wear, every cycle is not accompanied by complete relief of the strain. This type of wear is distinguished by dynamic fatigue of the material and/or fretting and friction rust (see also “adhesive wear”).
Classifying the types of wear by their characteristic damage mechanisms:
Damage occurs when the contact surfaces micro-cold weld together and a new area is torn open. The fresh metal surfaces are very reactive and create oxides and/or combine with surrounding materials, such as moisture or lubricants. This is also referred to as friction oxidation. The direction of the forces involved can be at any angle relative to the contact surface. This type of wear can also occur along with impact wear.
This type of wear occurs primarily due to a unidirectional relative movement of the contact surfaces. It often occurs along with wear due to sliding action. The damage often takes the shape of a cutting action.
The damage mechanism is the same as with rolling wear.
Wear with physical effects and chemical reactions:
The creation of fresh metal surfaces and/or the high friction temperatures can result in chemical reactions and/or diffusion in the contact surfaces. These reactions can promote wear. By lowering the coefficient of friction (lubricating effect), increasing strength, and/or preventing further micro-cold welds, the products of these reactions can serve to resist wear.
Figure "Fretting nodes": Appearance forms of fretting corrosion (Ref. 6.1-3) in steels. Oils, lubricants, and glide coatings provide conditions for mixed film lubrication. Several of the effects described also occur with other combinations of metallic materials. Reactions with surrounding media, such as atmosphere, oils, and greases, play an important role. Materials that form corrosion-resistant reactive surfaces (e.g. Ti-alloys) tend to be more susceptible to scuffing, micro-grooving, crack initiation, and waviness. The materials that are rubbed off have a strong influence (Fig. "Fretting mechanisms"), since they change the appearance of the damage symptoms (e.g. being confused with tarnishing) and often lead to misdiagnoses. For example, they can promote mistaken conclusions as to the operating temperatures. It is especially important to note that the wear products can change wear conditions over time. Often, increases in the volume of reactive products from an abrasion process cause fittings and plug connections to jam or even explode. A typical example is gliding planes with MoS2 glide coatings sticking in ocean conditions (salt water). If the products of these reactions can escape from the wear area (e.g. by being washed or blown out), the stuck parts can loosen (e.g. inner rings in roller bearings).
Figure "Fretting mechanisms" (Ref. 6.1-15): H.Klingele illustrated a typical fretting process in the above diagram. At first, the oscillating relative movements of the contact surfaces create wear particles through abrasion, dynamic fatigue, and material transfer (scuffing). The local destruction of top coatings results in fine oxide particles. The oxidation of the fresh metallic surfaces is promoted by the friction heat. Fresh metal surfaces of these particles immediately react with the surrounding media, such as moisture, grease, and air. Particles, i.e. reaction products, are ground in the contact area and are further altered by chemical reactions, sintering, and compression. They are pressed into the surfaces and determine (change) the tribo-contact.
Fig. "Fretting nodes" shows typical damage symptoms of this process.
Figure "Machine parts wear rates": The amount of material worn away is given in connection with typical technical components and processes with regard the sum total of the wear types. Wear due to pronounced fretting corrosion typically results in roughly 0.01-1mm of material being worn away for every km of wear travel (see also Ref. 6.1-2). However, the wear travel of a unidirectional movement must not be confused with the total movement of an vibrating wear occurrence. With equal wear movements, the fact that the amount of material worn away increases along with the total wear travel allows comparisons to be made.
Figure "Fretting temporal progression" (Ref. 6.1-4): This diagram shows the temporal progression of the amount of material in gearing oil that was worn off by typical wear. This progression can be discerned by oil analyses, and it is important for estimating the total and remaining life spans and damage risks of engine parts that are subjected to wear.
Different types of wear show characteristic material losses over time. The curves only show the qualitative temporal wear process, whereas the absolute limit of the wear is discretionary.
Conversely, the temporal increase of the amount of worn-off particles in the oil, along with an analysis of the particles (e.g. composition and structure in SEM), can give important clues as to the wear process and the affected parts.
Sliding fatigue wear, as mentioned in the excerpts, is a wear process in which roughness spikes are worn off by fatigue processes and thus the amount of material worn off decreases over time. This type of wear process is known as “run-in wear” in toothed wheels and occur as dull regions outside of the circle of contact. In this case, the wear doesn`t result in an unallowable shortening of the part`s life span.
In other cases, run-in wear can also indicate a long-term problem.
Abrasion and especially sliding fatigue wear (see Fig. "Load-specific wear types") exhibit an accelerated damage sequence. These processes result in the formation of abrasive wear products or geometric changes (e.g. increase of play) that accelerate the wear process. This makes it difficult to catch these damages in time.
Corrosive wear is a wearing-off of material under corrosive conditions. As long as the corrosive portion is the larger one and no protective (against further corrosion) or destructive (e.g. strongly abrasive) reactive layers are created, the temporal loss of material is linear.
Adhesive Wear (see Fig. "Load-specific wear types") is given here with a linear (constant) wear rate. The wearing-off occurs through scuffing and tearing-open of small areas of the tooth flanks. This type of wear process can be expected only during overstress (breaking through the oil film) or when there is insufficient lubrication.
Particles created by rolling wear (see Fig. "Load-specific wear types") in toothed wheels appear in gearing oil quite suddenly. They can indicate the impending failure of the parts involved, meaning that taking appropriate measures in time to prevent part failure is usually not possible.
On the other hand, track fatigue in gearing roller bearings often results in an increase in the amount of worn-off particles over longer periods of time (depending on the operating loads). This makes the damage recognizable in time to prevent serious part failure.
Figure "Play increasing by fretting" (Example "Fretting damage trials for a training aircraft", Ref. 6.1-1): This diagram shows a play increase in a ball joint in a compressor guide vane adjuster. This type of play increase must be taken seriously, as even small angle changes in the profile of guide vanes can cause noticeable disruptions of the flow and/or cause the blades to oscillate. The total amount of material worn off (amount of material worn off = increase of play) can be diagrammed linearly (top diagram) and can be reproduced in trials that sufficiently recreate operating conditions. When the play increase is compared with Fig. "Machine parts wear rates", the linear nature of this process is typical for a wear occurrence with pronounced adhesion and corrosion, which would be quite plausible for the part in question.
Consistent lubrication of guide vane adjusters has shown itself to be problematic, especially in ocean atmospheres (see also Chapter 126.96.36.199).
Figure "Decreasing dynamic strength by fretting": A serious and common type of damage resulting from fretting is a loss of hardness. Fatigue cracks and fatigue fractures can originate at the damaged surface.
Fatigue cracks resulting from rolling fatigue, that cause local areas of the contact region to break out (fatigue pittings, fatigue grooves), are usually formed underneath the surface.
The sensitivity of metallic materials to losses of dynamic strength due to fretting varies greatly. Steels are considered to be fairly resistant to this, while high-tensile titanium alloys can lose up to 30% of their dynamic strength. This can mean a shortening of their life span by several decimal powers.
Figure "Influences on fretting": In tribo-systems, fretting damage is influenced by many different factors. These are primarily
The complexity of a tribo-system requires evidence from technical trials and a great deal of experience in order to ensure sufficiently safe operation. These systems are not suitable for the application of mathematical/analytical methods. Every crossing of the “experience horizon” (e.g. in surface pressure or glide coatings) is accompanied by the danger of extensive damages. What makes this type of damage especially problematic is that it usually occurs after a long period of operation (incubation time), by which time the affected parts are serially produced and in widespread use.
Influence of surface pressure in the contact area:
The surface pressure in the macro-region (e.g. dove-tail roots of compressor rotor blades) is primarily dependant on design factors (wedge angle, bearing surface, etc.) and RPM (centrifugal force, bending forces from the gas flow). Before start-up, surface pressure is next to nothing (except in jammed blade roots) and reaches its maximum at full RPM. High surface pressure can lessen or elimitate relative movement between the contact surfaces. This effect requires sufficiently high friction forces that act against the external forces. If pronounced micro-movements occur (Fig. "Decreasing dynamic strength by fretting"), the rising surface pressure is typically accompanied by increased friction forces and wear rates (Fig. "Coefficient of friction indicating stressing"), and a loss of dynamic strength. This is true especially in cases where the glide coatings have been destroyed or damaged by the high surface pressure. In these cases there is direct contact between sensitive metallic materials (e.g. titanium alloys). An increasing coefficient of friction indicates this type of process.
The large differences in published results concerning increases of dynamic strength in titanium surfaces through surface treatments (some trials showed great increases in dynamic strength whereas others registered almost no change; Fig. "Fretting at titanium alloys") are most likely due to varying surface pressure in the trials.
Figure "Fretting processes in contact zone" (Ref. 6.1-5): A major cause of fretting-inducing micro-movements is varying stiffness in contact surfaces. This causes uneven elastic deformations in the contact surfaces when subjected to dynamic loads, resulting in corresponding micro-movements. Because these zones often lie at jumps in stiffness with pronounced notch effects and additionally high shearing stress (energy, see Fig. "Fretting load parameters in dovetail"), the danger of fatigue cracks being initiated is especially high.
These occurrences should be closely examined with the aid of two examples:
When two objects rest against one another under pressure loads, pressure distribution is dependant on the geometry of the bearing surfaces (Hertz pressure).
The distribution of surface pressure (top left diagram) leads to a corresponding distribution of friction force. These conditions result in corners being slightly rounded-off. Tangental forces can be absorbed in line with the curve “p.m”. Elastic deformation of contact surfaces that are impacted only by pressure loads can create tangental forces. If these tangental forces become greater than the friction force in the edges of the contact zone, relative movement occurs. Pulsing pressure loads combined with these micro-movements can lead to friction wear. This damage process is important in fretting and fatigue failures in blade roots (bottom left diagram).
The top right diagram shows elastic deformations in a pressure surface caused by pressure from a stamp. Here, glide movment in the outer zones of the edges is caused even without the edges being rounded-off. When dynamic loads are present, they cause corresponding oscillating micro-movements and fretting. This type of stress is typical for a shrunk shaft region with sufficiently high rotary bending (bottom left diagram). The jump in stiffness around the shrink-fit`s outlet causes fretting.
Therefore, the designer must ensure that jumps in stiffness (differences in elasticity) are minimized in separable connected parts that are subject to high stress levels.
Influence of the coefficient of friction on the contact zone:
The coefficient of friction in the contact zone plays a decisive role in damage sequences (Fig. "Coefficient of friction indicating stressing"). It has a strong influence on the tension levels in the contact zone. The transient area is particularly interesting (see Fig. "Fretting processes in contact zone"). In this area, tensile stress and shear stress are often combined. The effective stress of this stress combination (also referred to as engergy, see Fig. "Fretting load parameters in dovetail") determines the dynamic fatigue. It can thus be assumed that fretting damage increases along with a rise in the coefficient of friction (Fig. "Coefficient of friction indicating stressing"), while the damage (e.g. abrasion and adhesion) and the coefficient of friction influence one another alternately (Ref. 6.1-17, Example "Increased stress caused by patchy lubricant"). This means that the coefficient of friction changes over the time the stress is present (Fig. "Coefficient of friction indicating stressing"). The change in the coefficient of friction is influenced by multiple factors (see page 16), including the materials of the parts involved as well as operational and design factors. This shows why it is extremely difficult to design trials that accurately simulate fretting under operating conditions. Any results from such trials can only be expected to be practically applicable to the specific case in question (e.g. for designing a remedy for a type of damage).
Figure "Coefficient of friction indicating stressing": The top left diagram shows the maximum main stress in a dovetail root. One can see that the main stress zones of the blade (1) and the disk slot (2) are not at the same points. The maximum stress levels in the contact area are noticeably dependent on the coefficient of friction (top right diagram). If the contact zones sieze, the coefficient of friction can be considerably higher than 1. The center diagram shows how the coefficient of friction changes, depending on the number of glide movements (i.e. usually dependent on time) in different tribo-systems as taken from a laboratory test. The surprising sudden exponential increase of the coefficient of friction in all systems after roughly the same number of movements (curves 2-7) indicates failue due to the parts siezing. This behaviour is indicative of the high surface pressure in the trials. Tribo-systems such as blade roots should be designed to prevent such high surface pressures. The flatter curve 1 seems to indicate that the glide process stabilizes itself.
It is possible to optimize wear performance by treating the contact surfaces (bottom diagram). Each of these treatments has its own advantages and disadvantages, some of which are mentioned here:
Fixing: Relatively simple, also easily done after construction (peening). A loss of effectiveness due to creeping at operating temperatures can be problematic.
Glide coatings: Easy and economical (e.g. graphite). Disadvantages are possible changes (e.g. oxidation, abrasion) followed by increases in the coefficient of friction at high operating temperatures.
Metallic coatings: Advantages: strongly adherent (e.g. galvanized copper-plating with gunmetal), avoids direct contact with the base material. Disadvantages: expensive, changes dimensions, glide behaviour (oxidation, fatigue) changes over time at high operating temperature.
Loose inserts: Advantage; very effective (e.g. copper foil). Disadvantages: can fall out, difficult to mount, must be taken into consideration with regard to dimensions.
Figure "Fretting load parameters in dovetail" (Ref. 6.1-9): This diagram shows the stress relationships in the disk slot of a dove tail connection. Similar conditions exist in blade roots (see ). The excerpts define characteristic parameters of fretting damage to the fatigue strength of a contact zone. They are suitable for use in part design, especially blade root design.
Fretting Damage Parameter “FDP” (bottom diagram):
The “FDP” is the product of the shearing stress in a contact zone and the fretting travel (double amplitude). The parameter reaches its maximum in the disk region across from end of the blade root (see Fig. "Fretting processes in contact zone" zone B). Experientially, this area is the most likely location of a dynamic crack initiating in the disk slot. After initiation, the crack progress is determined by the height of the (direct) stress (compare also Fig. "Loads in dovetail contact surfaces" for the blade root) parallel to the contact line. Because of this, a further characteristic parameter (FFDP) was defined to cover the progress of the crack until the affected part fails.
Combined fatigue-fretting damage parameter (FFDP, center diagram):
The FFDP is the product of the local direct stress parallel to the contact line, shearing stress, and fretting travel.
Because both direct stress and shear stress are present in the wear zone, the resulting comparative stress determines the dynamic fatigue (energy, see below). The comparative stress also takes into account the coefficient of friction in the shear stress. It does not include the friction energy, however, because the friction travel is not part of the formula. For this reason, the data concerning the anstrengung will not represent operating conditions as closely as parameters that take friction travel into account.
The following methods can be applied when designing parts on basis of the given parameters:
1. Evaluation of parts with or without crack initiation taken from typical engine operation and calculation of the respective parameters. The direct stress parallel to the contact line, as well as the shear stress, and be calculated to a sufficiently accurate degree, provided the coefficient of friction is also sufficiently known. The local fretting travel can be determined by microscopic analysis of the damage surface (SEM). This data combined with experience can be used to determine allowable thresholds.
2. Ascertainment of parameters in laboratory tests that sufficiently simulate operating conditions.
3. Designing the parts in accordance with the parameters known from trials and operation. Confirming that trials are sufficiently close to operating conditions by analyzing parts after subjecting them to strain that is representative, significant, and typical during actual operation.
Figure "Loads in dovetail contact surfaces" (Ref. 6.1-10): In a manner similar to Fig. "Fretting load parameters in dovetail" for the disk slot, the surface pressure caused by the centrifugal force spreads across the contact area of the blade root (top left diagram). The maximum is at point “A” (top right diagram). The curve with two pronounced maximums is caused by differences in stiffness in both the blade root and disk slot between “A” and “B”.
As has already been shown in Fig. "Coefficient of friction indicating stressing", the stress in the contact surface increases along with the coefficient of friction (top left diagram for m=0,5 and m=0). The number values are valid for an assumed centrifugally-induced even stressing of the root-neck cross-section of 100N/mm2.
If the blade undergoes flexural vibrations, the centrifugal force overlays with the flexural vibrations and the resulting load distribution in the surface of the blade root (bottom right diagram) puts a pronounced maximum of tensile stress on “A” (bottom left diagram). The tendency for dynamic fatigue fractures in this region is determined by the energy (Fig. "Fretting load parameters in dovetail"). Due to the moderate notch effect, the stress increase in the root radius (“C”) is relatively small, but must be considered when designing blade roots.
Figure "Fretting simulation in blade foot area tests": There are several different test arrangements that are used to simulate as accurately as possible the operating conditions for dynamic strength losses due to fretting in the contact surfaces of blade roots. The problems these trials can encounter are illustrated by the following two examples.
In one trial, friction blocks, in this case in bridge-form (top right diagram), are pressed with defined pressure against a flat sample that is being subjected to tensile-compressive stress. When the sample is elastically deformed, fretting takes place at the contact surfaces due to the differences in strain between the sample and the friction block (compare with Fig. "Fretting processes in contact zone"). These trials are relatively inexpensive (basic samples), the parameters are reproducible, and the transmission of force comparatively simple. Crack initiation usually happens in the contact zones. The dynamic strength tolerated by the sample indicates the damage. Deviations from the conditions in a blade root are due to the stress gradients across the cross-section (stress gradients in the blade root, consistently high stress in the sample cross-section) and the difficulty of simulating the stiffnesses in and around blade roots. Therefore, the trial did not result in the cracking process pausing or stopping, which is often observed in real situations and results in considerably longer periods of time elapsing before the blade fails. Changes of the coefficient of friction over operating time show less of an effect in trials than during normal operation, which makes selecting and optimizing the tribo-system considerably more difficult. Experientially, many tribo-systems behave similarly in this test arrangement, whereas their behavior in the engine shows pronounced differences.
Results that are closer to operating conditions than the friction-block trials can be expected from trials of samples that have the geometric shape of blade roots (bottom right diagram), and simulate flexural alternating stress as well as the tractive force intended to simulate centrifugal force. The drawbacks of these trials are the relatively expensive samples and the elaborate transmission of force.
With a similar test arrangement (bottom left diagram), the coefficient of friction can be measured continually throughout the trial. To do this, the V-forces are measured at separate disk slot overlays and set relative to the force that presses the blade root from the bottom outward. This trial makes the optimization of surface coatings and glide coatings possible.
Figure "Coefficient of friction influenced by service" (Ref. 6.1-6): The coefficient of friction between contact surfaces plays a critical role in fretting. The higher the coefficient of friction overcome by micro-movements is, the higher the stresses in the fretting zone. High shearing loads are created in the damaged contact surfaces.
The left diagram shows the principal relationship between the coefficient of friction and the normal force when deformation and adhesion are part of the friction force. From this it can be deduced that the coefficient of friction of a tribo-system reaches a minimum at a certain normal force. Under this condition the best dynamic strength should be found during fretting. For this reason, the coefficient of friction of contact surfaces of blade roots is kept as low as possible through the use of suitable glide coatings. This can be optimized by the selection of dovetail angles and the size of the contact surfaces.
Depending on the materials, the coefficient of friction is also determined by the glide speed, i.e. the size and frequency of the relative movements between the contact surfaces (right diagram). This shows that the best tribo-system (material combination; determined by the lowest possible coefficient of friction) for each particular situation should be found.
The effect of strain on fretting-induced fatigue
The load distribution in the contact zone has a strong effect on the dynamic strength. Because of this, form- and stiffness notches in fretting zones behave as though the notch factor was increased considerably.
Trials in which fretting occurred and was followed by dynamic fatigue stress do not give satisfactory results concerning the behavior of engine parts when simultaneously subjected to fretting and dynamic loads. Also, results of engine part trials can be easily falsified if the load levels and collective load, as well as the load- and rest intervals, do not adequately recreate operating conditions. If the load levels are too high, damage in the fretting zone may not develop fully and the part may fail due to a mechanical notch in a different place. Appropriately realistic time frames, on the other hand, would lead to parts failing in the fretting zones, as happens during actual operation.
The effect of wear time, oscillating frequency, and hold times on the contact zone.
The wear conditions in the contact zone can change along with increasing wear time (the time during which the wear process is active), i.e. number of movement cycles. For example, under certain wear conditions, the coefficient of friction of Ti6A14V increases from m =0.45 during the first load alternation to m =0.8 after 103 load alternations and to m =1 after 106 load alternations (Fig. "Coefficient of friction indicating stressing"). Accordingly, the wear appearance changes over time. Surface deformations, disruptions, and micro-fissures form, and in the final stages macro-fissures result in the affected part failing. The temporal influence must naturally be viewed in relation to other influences, such as temperature and material combinations (Fig. "Decreasing dynamic strength by fretting"). For example, the wear rates of certain materials can be much lower at high temperatures than at room temperature, but reach a maximum at 300°C. The same is true for influences on dynamic strength. For example, it is completely plausible that a coating does its job well in a high-pressure compressor but becomes brittle and fails when used in the lower-temperature front compressor stages (Fig. "Blade foot coatings operation behaviour"). In this case, as well, the forming of oxides with or without a “lubricating effect” could be an important factor. If dry fretting rust forms on steels (Fig. "Fretting nodes"), it can escape from the contact zone in powder form and result in temporally dependant play increases, changes of the contact surface, and load transfers to other areas and parts. If the fretting rust remains between the contact surfaces as a hard, lacquer-like film, over time this can lead to play reduction and, in extreme cases, jamming and seizing. This type of damage has been observed in thrust reverser bearings and guide vane adjusters, among others (Fig. "Wear in protective sleeves"). Should jamming occurrences impede a part's dampening, it can lead to powerful oscillations and fatigue fractures. In this case, the dynamic fatigue may not be confined to the fretting zone.
The (middle) glide speed during oscillating stress is a function of the oscillating amplitude (micro movement) and the frequency. During oscillation, the glide speed is constantly changing. With sinusoidal oscillations, the speed at the turning points is zero and reaches a maximum in the “middle” of the movement. The middle
glide speed of an oscillating movement is different from the glide speed of unilateral movements, and the two can cause very different types of damage (Fig. "Play increasing by fretting").
If movements are slow, e.g. disk expansion during engine start-up followed by a long hold time (often hours), it is usually pointless to specify the middle glide speed across the entire movement cycle. It is better to specify the speed during the gliding process. This glide speed is co-determinant of the friction heat in the glide surface and therefore has an important influence on oxide formation during the wear movements. On their part, these oxides (depending on their makeup; e.g. steel FeO, Fe2O3, Fe3O4 Fig. "Wear products influencing fretting") influence the friction conditions, especially the coefficient of friction and wear. Therefore, the speed of micro-movements is of special importance for the wear process (Fig. "Parameters affecting fretting behaviour").
The frequency factor is partially contained in the amplitude, speed, and hold time factors. Generally, if the amplitude is remains constant, a marked increase in wear and wear rate along with frequency can be expected, since the number of wear movements per time unit increases. Experientially, the amplitude at a blade root decreases as the frequency rises because the dampening friction accompanying oscillating energy that must be absorbed also increase. Also, oscillations of a higher order had smaller amplitudes than fundamental flexural modes and an oscillating shape (knot) that induces only very small movements in the blade root. This explains why, in practice, wear often decreases as the frequency rises. If the higher frequency creates more friction heat, this can also change the wear behavior, depending on the materials and possible glide coatings used.
It has been observed that high-frequency blade oscillations can lead to a pronounced drop in blade-root wear due to the vibration of the contact surfaces. The higher fixative forces against axial offset created by this vibration must also be considered.
The hold times at the turnaround points of the oscillations (stick-slip effect), as well as the intervals between oscillations, affect the force needed to “break free” at the beginning of a new wear movement (
Figure "Damage of blade feet by increasing load level" (Example "Increased stress caused by patchy lubricant", Ref. 6.1-17): Even a small load increase above those levels shown to be unproblematic in lower-output engines led to fatigue cracks forming (LCF) in the contact areas of the fan-blade roots. This necessitated a change from the original graphite coating to a CuNiIn-spray-on coating (ductile alignment of the contact surfaces, wear protection) with an additional graphite spray coating (the lower coefficient of friction results in lower shearing stress levels in the contact surface; details below).
Figure "Wear movements depending from operation": The micro-movement-induced glide speed and the surface pressure change distinctly and often during engine operation (see also Fig. "Blade foot fretting operation phases"). This also changes many other parameters than influence fretting. The parameters that determined fretting in a specific case should be researched as much as possible during damage analysis in order to obtain parameters that are as close to operating conditions as possible. These parameters can then be used in trials to confirm preventive measures or design improvements.
When the engine is powering up (Example "Increased stress caused by patchy lubricant"), “stick-slip” movements can be expected. These movements are caused by the disk expanding at the blade-roots, and can be observed in the corresponding changes in glide-speed.
If the blade oscillates at high frequency during this process, it can lead to a considerable drop in the coefficient of friction (similar to sand sliding off of a vibrating angled surface, see Fig. "Coefficient of friction during relative motion"). At high RPM the centrifugal force is usually strong enough to prevent large relative movments in the contact surfaces of the blades. However, blade and/or disk vibrations will result in micro-movements due to the difference in stiffness of the materials.
The relatively low RPM of an engine that has been shut down in flight and is being driven only by the wind (windmilling) cause only small contact pressure. Blade oscillations can result in large, high-frequency relative movements due to the low surface pressure. In this case, a localized separation and collision (hammering) of the contact surfaces can occur. This type of wear can be different from pure glide wear, depending on the materials involved, e.g. breakouts following dynamic fatigue. Windmilling thus promotes fretting in blade roots and clappers, if they are present.
Figure "Coefficient of friction during relative motion": The coefficient of friction affects the stress levels in the axial fastenings of rotor blades. In compressor blades, usually an axial component resulting from the gas force appears moving against the direction of flow (forward, see also Fig. "'Blade walking' by axial forces"). In turbine blades, the force component is in the direction of flow (rearwards).
If one considers the seemingly trivial fact that the friction force is not directionally dependent, it becomes clear why, when the contact surfaces are gliding, additional forces do not meet resistance through friction. At this point, the coefficient of friction is almost zero.
There are two typical cases, in which this situation occurs with joined parts, especially rotor blades:
If the blades and/or disk vibrate sufficiently for micro-movements to occur in the contact surfaces, the friction forces decrease considerably, similar to the action of a shaking trough (top right diagram). If this results in the blades moving axially, the fastenings are subjected to unexpectedly high stress levels. This effect is especially pronounced wehn large blades, such as fan blades, are equipped with root-contact surfaces that run diagonally outward due to the hub contour (middle diagram). In this case, if the axial centrifugal component is considerably larger than that of the gas force, the blade may offset in direction of the flow.
The same effect, only more pronounced, was observed in a vertical tesking rig (in a vacuum, i.e. no gas stress).
In this situation, the powering-up of a fan rotor with diagonal root contact surfaces resulted in a sudden axial offset of the blades up vertically while the axial play was increased several tenths of a millimeter, without any noticeable vibrations. Evidently the expansion of the compressor disk due to centrifugal force led to the blades sliding accompanied by a corresponding drop in the coefficient of friction. This process can be compared with the loss of control in runaway drive wheels (bottom right diagram). The offset occurred in the individual blades at different times, causing large imbalances. The solid fastening lugs were subjected to such high stress levels that they were plastically deformed. A reeinforcement and a fitting devoid of axial play solved the problem.
Excerpt: “….(the OEM) is to issue service bulletins to all….. operators which will revise, and in most cases alleviate, the tight inspection intervals for the engine's fan blades which have been in place since the failure of a blade on a Emirates (aircraft) in Melbourne, Australia earlier this year.
Intensive tests conducted by the engine maker have confirmed the failure was caused by differential stresses imposed on the blade root by patchy lubricant, rather than by a fundamental design flaw, as some operators had initially feared.
The fix is the use of Metco 58, a copper-nickel-indium alloy coating material, which had been introduced into the …fan blade production process last July. The failed blade was an earlier production unit which had been coated with a graphite-based paint called PL 237. Tests showed that under high thrust settings, this becomes patchy and causes stresses to build up in the blade root, resulting in cracks.
...a fleet wide inspection of all fan sets (124 aircraft in service plus spares) revealed four cracked blades….after the completion of tests the UK Civil Aviation Authority is likely to `move the inspections out to 1,200 cycles in most cases…
Under an amendment to a Federal Aviation Administration airworthiness directive issued in February, operators have been required to inspect….at intervals as often as every 200 cycles in the case of the higher thrust… (engine) versions.
Even the lower powered (engines)…required inspection every 400 cycles…
Tests proved that the phenomenom only affected the higher power engines…
The fan root requires lubricanting because, under acceleration, the energy of the fan forces the disc to dilate. The blade, which has a curved root, then sits further out in the dovetail fitting. Under deceleration, the blade root slides back as the disk returns to its original shape.
Lubrication is required to ease the friction between the blade root and the dovetail, and the cracks occurred when the original lubricant became patchy under high thrust loads. This allowed parts of the surface to stick, which caused differential stresses to build up and eventually cracks to appear.”
Comment: The affected engine type is a large modern fan engine with a large bypass ratio. The damage process is described in Fig. "Wear movements depending from operation". This case demonstrates the dependence of the critical root stresses on the selection of the right tribo-system specific to the parts involved.
Ref. 6.1-18 describes the Metco 58 coating being applied through flame spraying. However, this is followed by the aerosol application of a graphite glide coating. It is emphasized, that cracks only formed in especially high-stress engine versions. This shows that in contact surfaces of titanium-alloy compressor blades, especially, even a small increase in the stress levels can lead to serious damages.
The Impact of Corrosion and Oxidation on Fretting
Corrosive influences from the surrounding atmosphere have an effect if the contact between surfaces is not sufficiently close. With compressor blades, this usually involves a saline solution created by condensation water forming in an ocean atmosphere. Fretting oxidation in corrosion-sensitive materials such as Al alloys and steels is usually reinforced by increasing relative humidity.
If noticeable oscillations occur in the rear compressor stages, then the operating temperatures are already too high for any watery corrosive medium to be present.
A further possible damage process is the disintigration of glide coatings due to the operating temperatures and/or the friction heat and mechanical stress during the finishing process. In the compressor, temperatures at friction surfaces are usually between 200-500°C. However, “temperature spikes” in direct contact with the roughness can be much higher. This increases the oxidation tendency of freshly created metallic wear surfaces. As the air pressure (in modern engines clearly more than 30 bar) increases on the way to the compressor exit, it correspondingly increases the oxygen supply and the oxidation tendency (Fig. "Wear products influencing fretting").
Products of the wear process, especially oxides, influence wear behaviour. They can accelerate the wear process. If they act as a glide coating, however, then they create satisfactory operating conditions. In superalloys, especially cobalt-based ones (in combustion chambers), the formation of oxides results in good wear behaviour at high temperatures (Fig. "Parameters affecting fretting behaviour").
Figure "Wear products influencing fretting": Depending on the material, the type and amount of wear products can be determined by the surface pressure (top diagram). Therefore, in trials it is important to simulate the surface pressure under operating conditions.
Steels exhibit an especially pronounced oxide formation (Fe2O3 and Fe3O4 ) with varying wear behaviours. As shown in the bottom diagram, melting temperatures can be expected locally in the contact zone even at relatively low glide speeds. If the oxide layer is worn through, mechanical metal removal occurs.
The Effect of Materials on Fretting
A material is damaged in different ways during fretting.
Metal removal due to abrasive or adhesive wear, which can be reinforced by pitting due to fatigue cracks under the surface, leads to dimensional changes.
A drop in the dynamic strength can be caused by macro-crack-forming fatigue processes. The dynamic strength is characterised by the stess behaviour (Fig. "Fretting load parameters in dovetail"). Both damage types are influenced by factors resulting from the material combination:
Titanium alloys are especially sensitive to fretting fatigue (Fig. "Fretting processes in contact zone"). The loss of dynamic strength can reach 20% even without fretting. As operating temperature increases, the dynamic strength drops even more with fretting (Fig. "Parameters affecting fretting behaviour"). Al-alloys are similarly sensitive. Cr-steels and Ni-based forged alloys are less sensitive to this. Wear trials of several material combinations with and without glide coatings were published, and metallic intermediate layers (copper) with glide coatings showed the best performance. In practice, at lower temperatures short-blasted surfaces surfaces and graphite-based glide coatings have proven to be effective (Fig. "Fretting at titanium alloys"). Titanium alloys are particularly sensitive to fretting fatigue (Fig. "Fretting at titanium alloys"). At high temperatures, which would accelerate oxidation of graphite coatings, metallic spray-on coatings (bronze) seem promising when applied to blade roots.
Figure "Parameters affecting fretting behaviour": For a cobalt-based alloy of the type often used in turbine guide vanes in high-pressure stages or in combustion chambers, the fretting process is shown relative to temperature, surface pressure, wear frequency, and time factors.
A fretting maximum can be observed at 250°C (top left diagram). As operating temperature rises, fretting decreases and occurs at a rate clearly lower than that at room temperature. The reason for this is the creation of wear-reducing oxides.
At 20°C between 10-20 Hz, wear increases linearly along with increasing oscillation frequency (top center and bottom right diagrams). This behavior would be expected to accompany increased wear travel, if one assumes the wear process to be adhesive and abrasive. The considerable increase in wear rate along with surface pressure at low temperatures is a further indicator of metallic contact, wear surface, and the possible wear process. This supports the assumption that the high wear rates at low temperatures are due to a lack of lubricating oxides.
If temperatures are around 600°C ( top right diagram), the observed behavior is the opposite of that at 20°C. As the oscillation frequency increases, the wear volume decreases markedly. This indicates a different wear process, one that results in the formation of a wear-resistant layer (probably oxides).
Observations indicate that pre-oxidizing nickel alloys can reduce wear in the blade roots and disk slots to a large degree. Also, the first roughly 100 start-up/shut-down cycles are decisive for long-term wear behavior. This behavior should also be considered when conducting cyclical centrifuging trials in a testing rig.
Figure "Fretting at titanium alloys": The fretting-sensitivity of titanium alloy Ti 6Al 4V is a typical example of this group of alloys. The trials were conducted at room temperature with pressure blocks and flat samples (Fig. "Fretting simulation in blade foot area tests"). Without the influence of fretting, shot peening and abrasive-blasting result in similarly high dynamic strength levels. The use of only shot peening or only glide coatings results in the dynamic strength levels during fretting being only 30% of the levels without fretting. This is suprising, since it is usually assumed that shot peening alone is responsible for the increase in dynamic strength during fretting. For the given wear parameters (e.g. room temperature), at least, the positive effects of shot peening do not apply. Experience has shown that the greatest drop in dynamic strength can be expected from abrasive-blasted samples subjected to fretting, since the high degree of roughness prevents hardening and abrasive particles embedded in the surface can promote the damage process.
The best results were achieved with samples that were shot peened and glide-coated. The dynamic strength of these samples during fretting was about 80% of that without fretting. This behavior might be explained by the reservoir-effect of the callote-structure of the surface combined with the glide-coating layer. In this case, as well, the low coefficient of friction seems to be the deciding factor. A low coefficient of friction reduces stress and thus acts to increase dynamic strength ( ).
One can see in the bottom diagram that in temperatures up to about 350°C, increasing temperature results in a loss of dynamic strength in samples subjected to fretting. Apparently, protective oxide layers do not form in this temperature range.
It must be remembered that these specifications do not allow conclusions to be drawn about behavior in the threshold around 500°C, since this is where considerable oxidation occurs (see Fig. "Blade foot coatings operation behaviour"). Whether or not glide-coatings have beneficial effect on the dynamic strength is dependant upon surface pressure (wearing through the protective layer) and temperature (oxidation of the glide coating).
A loss of dynamic strength due to a decrease in the hardening of shot peening (creeping at high temperatures) is evidently less critical in titanium alloys. Shot-peened samples simply exhibit a slightly higher dynamic strength with and without fretting than abrasively-blasted samples without a clear hardening effect.
Figure "Blade foot coatings operation behaviour": This example shows how strongly dependant on operational factors, especially operating temperature, the behavior of a material during fretting is. A compressor rotor blade made of a high-tensile titanium alloy with a metallic wear-resistant layer behaves satisfactorily during fretting in the rear compressor region (B), whereas the uncoated but shot-peened blade fails due to dynamic fatigue in the fretting zone after a short period of operation.
The high surface pressure results in plastic deformation of the protective layer without crack formation in either the layer or the base material. Unallowable formation of oxides is not observed. Evidently, the coefficient of friction, i.e. stress, over the period of operation is so low ( ) that it does not result in dynamic fatigue. The increased operating temperature, possibly in combination with the relatively large oxygen supply (high pressure), most likely play a decisive role in the behavior of this tribo-system. Interestingly enough, the high operating temperature, relative to that typical in the rear compressor region, does not result in a large drop in dynamic strength during fretting, as could be extrapolated from Fig. "Fretting at titanium alloys".
When the same material combination is used in the forward compressor region (A), i.e. at lower temperatures, the wear-coating is shattered, cracks form, and oxides are impressed. The base material also exhibits scribing, which indicates a high coefficient of friction, i.e. high stress. The strong oxide formation despite the low operating temperature is surprising. Evidently this is a type of oxide that acts brittly at these temperatures and does not have any lubricating qualities. High coefficients of friction should result from a combination of alternating metal attrition and oxide formation.
Effect of Wear Movement on Fretting:
Various types of surface stress can occur, depending on the type of wear movement, e.g. usually hammering, gliding, rolling-off, or a combination of these movements. If the movement is a gliding one parallel to the surface, an abrasive, geometrically altering wear process can be expected, whereas hammering often results in fatigue and crack formation. Experience has shown that hammering wear can cause especially large drops in hardness. The creation of and transfer of wear particles also depends on the type of wear movement. Blade oscillations and changes in RPM cause locally different and changing surface pressures in the contact surfaces of dove-tail blade roots. If the root tilts due to low centrifugal force (e.g. during windmilling, start-up, or shut-down), it can result in a hammering movement at the edges of the disk slot, while the center of the surface experiences gliding movement (Fig. "Blade foot fretting operation phases"). At high centrifugal force, however, blade oscillations commonly result in gliding micro-movements.
During engine start-up, the elastic expansion of the disk and accompanying elastic deformations in the blade roots and disk slots primarily cause gliding movements.
If the contact surfaces seperate during the wear movement and a so-called “pump-effect” occurs, sucking surrounding media (air, water, etc.) into the gap or blowing wear products out, the conditions (corrosive, adhesive, etc.) are completely different than if the surfaces were in contact with one another. This is especially true if a ductile, sealing intermediate glide layer is present (i.e. grease or glide coating).
The influence of the amplitude value on the wear process must also be considered. If glide films such as oxide films are present on the contact surface, there is a certain minimum amplitude for destruction of the glide layer (depending on material involved and surface pressure; Fig. "Wear products influencing fretting"). If this amplitude is reached and metallic contact occurs, it results in powerful abrasive and adhesive wear (Fig. "Coefficient of friction indicating stressing"). The wear usually increases along with oscillation travel. The movement amplitude also has a pronounced effect on the form, transport, and accumulation of the wear products (oxides, metallic particles, glide coating remnants). The macroscopic wear width (width of the fretting zone in direction of the glide movement) seldomly corresponds to the oscillation travel of the wear movement. The oscillation width of the micromovement is usually much smaller and can be estimated through a microscopic inspection (SEM).
Figure "Blade foot fretting operation phases": Characteristic fretting movements can occur in many different phases of operation (Fig. "Coefficient of friction influenced by service"). These can be accompanied by low-frequency (e.g. change in RPM) and high-frequency (e.g. blade oscillations) micro-movements of varying size. Along with gliding movements, hammering movements can also occur, if blade oscillations occur under weak centrifugal force. The location of the fretting zone can also shift dramatically in this case.
Changes in the hold time can affect the wear conditions (bottom diagram). For example, if hold times are short, the wear force is clearly dependent upon their duration.
Influence of Contact Surface Geometry on Fretting:
Between technical surfaces, direct contact only exists in limited zones or islands, which only represent a fraction of the total surface. Thus, contact occurs primarily through peaks of roughness. The contact zones are determined by the geometric shape of the contact surfaces (e.g. flat, arched, spherical), imprecise measurements (Fig. "Blade foot wear symptoms"), and elastic and plastic deformations. The influence of the topography of the contact surfaces is addressed below along with manufactural factors. Analyses have shown that the volume of material that is stressed by the wear process is crucial for the loss of dynamic strength. If fretting occurs in large surface areas, i.e. large volume under stress (in a zone near to and below the contact surface), the danger from fatigue is considerably greater than in small contact surfaces (if wear factors are the same, i.e. surface pressure, etc.). A clear improvement in the ability of specific surface structures to resist fatigue during fretting can be achieved by designing the zones affected by fretting in such a way that they are seperate from the zones subjected to high dynamic stress, so that both effects do not occur simultaneously in the same area. For this reason, molded (e.g. grooved contact surfaces) have been shown to be advantageous (Fig. "Fretting fatigue influenced by surface topography"). In this way, the dynamic strength of an Al-alloy surface was doubled in comparison to a flat surface. This realization can also explain the advantageous shot-peening effect, in which the contact materials do not exhibit a clear hardening due to plastic surface treatments, but still show a increase in dynamic strength during fretting. So far, there has been no known serial application of grooved surfaces to fretting-stresses parts.
The manufacture of contact surfaces is extremely important for the behaviour of two surfaces during wear. Thus, the location of the wear point (Fig. "Blade foot wear symptoms") and the amount of surface pressure are dependent upon the manufacturing tolerances (for example, the angle of the dove tail groove and the alignment of the lay-on surfaces). Even if the required tolerances are met, within these there will be a noticeable deviation from the ideal shape (e.g. arched instead of planed contours), which influences the shape and location of the wear zones. The manufacture-dependent absolute and middle roughness, as well as the roughness profile are also important. These include the shape and preferred direction of the roughness profile relative to the glide movement. In this way, roughness can increase or decrease the wear effects as well as the dynamic fatigue. The fatigue resistance of shot-peened surfaces (Fig. "Fretting at titanium alloys") during fretting has not been completely explained. Hardening and residual stress are often credited with this effect, but on the other hand, there are strong indications that the “calotte-topography” is the determining factor, at least in tensile titanium alloys. As with molded grooves, this explanation is based on the small stressed volume at the tips and on the “depot-effect” of the calotte-shaped impressions on the wear products. An interesting observation with regard to this is that fatigue only takes effect from a critical contact surface area size onwards, depending on oscillation stress. If additional glide coatings are applied, it is probable that the calottes exhibit reservoir-effect. Trials have shown that shot-peened samples of tensile titanium alloys without fretting can have even lower dynamic strength than untreated samples. During fretting, however, the treated samples had considerably more dynamic strength than the untreated ones. This behavior can be explained by the influence of topography.
Figure "Blade foot wear symptoms": Typical, schematically depicted wear patterns in blade roots. These can occur as a result of manufacturing tolerances such as angle mismeasurements. These images can be superimposed on wear zones caused by blade oscillations (bottom diagram). The location of the zone (in a rotor blade towards the shaft or in a disk groove radially inside, see Fig. "Coefficient of friction indicating stressing") most affected by dynamic fatigue allows estimates as to the danger posed by the damage.
Figure "Fretting fatigue influenced by surface topography": Several observations as to the behavior of fretting-damaged titanium alloys, with regard to oscillation fatigue, indicate that the hardening effect of shot-peening is less effective than a calotte structure at minimizing a loss of dynamic strength (at least under certain operating conditions). This has led to fretting trials with specifically molded surfaces. In a block trial (Fig. "Coefficient of friction influenced by service"), cross-shaped impressed grooves showed a level of improvement similar to that of shot-peened surfaces (top right diagram).
There are several explanations for this effect, which may well act in combination:
6.1-1 M. Nakao, M. Ikeyama, S. Abe, “Analytical Condition Inspection and Extension of Time Between Overhaul of F3-30 Engine”, ASME-Paper 91-GT277 of the Gas Turbine and Aeroengine Congress and Exposition Orlando, Fl June 3-6, 1991.
6.1-2 H.Uetz, J.Föhl, “Erscheinungsformen von Verschleißschäden”, VDI-Reports Nr. 243, 1975, Pages 143-156.
6.1-3 A.A. Bartel, “Reibkorrosion”, VDI-Reports Nr. 243, 1975, Pages 157-170.
6.1-4 W.T. Sawyer, Office of Naval Research (Code 463), Department of the Navy, Arlington, Virginia, “Proceedings of the 6 th Meeting of the Mechanical Failure Prevention Group”, MFPG Technical Report No. 9, November 2-4, 1971.
6.1-5 R.B. Waterhouse, “Physics and Metallurgy of Fretting”, Agard-CP-161, 1974
6.1-6 J.W. Kragelski, “Reibung und Verschleiß”, VEB Verlag Technik Berlin, 1979, Pages 269-291.
6.1-7 R.K. Betts , Fa. General Electric, “Wear and Fretting Fatigue Resistant Coatings”, AFML-TR-74-18, 1973.
6.1-8 K.H. Kloos, E. Broszeit, “Verschleißschäden durch Oberflächenermüdung”, VDI-Reports Nr. 243, 1975, Pages 189-204.
6.1-9 M.J.He, C.Ruiz, “Fatigue Life of Dovetail Joints: Verification of a simple Biaxial Model”, Zeitschrift “Experimental Mechanics, June 1989, Pages 126-131.
6.1-10 H.A. Jergeus, Stal-Laval Turbin AB, “Microslip initiated Cracks in Compressor Blade Attachments”, Paper des Rimforsa Research Seminar on Material Science, 15-18 August 1977.
6.1-11 R.B. Waterhouse, “Fretting Corrosion”, Pergamon Press, 1972.
6.1-12 W.J. Harris, Rolls -Royce (1971) Ltd. Small Engine Division, “The Influence of Fretting”, AGARD-CP-161, Oct. 1974, Pages 7-1 to 7-11.
6.1-13 D.H. Buckley, NASA Lewis Research Center, “Effect of Various Material Properties on the Adhesive Stage of Fretting”,AGARD-CP-161, Oct. 1974, Pages 13-1 to 13-17.
6.1-14 J.Thiery, E.R. Spinat, SNECMA, “Comment Reduire L'Usure des Pièces non Lubrifiées dans les Turbomachines”, AGARD-CP-161, Oct. 1974, Pages 6-1 to 6-13.
6.1-15 R.L.Jonson, R.C. Bill, U.S.Army, “Fretting in Aircraft Turbine Engines”, AGARD-CP-161, Oct. 1974, Pages 5-1 to5-13.
6.1-16 L.Engel, H. Klingele, “Rasterelektronische Untersuchungen von Metallschäden”, ISBN 3-446-13416-6, Carl Hanser Verlag München Wien, Page 171
6.1-17 Guy Norris, “R-R eases Trent 800 inspections”, Zeitschrift “Flight Unternational”, 5-11 June 2001.
6.1-18 S.W. Kandebo, ”“Rolls-Royce Resolves Trent 800 cracks”, Periodical “Aviation Week & Space Technology”, June 4, 2001, Page 52.