Why Press Fits Fail Under Heavy Loads and How Modern Maintenance Teams Prevent Failure
Press fits seem like an ideal solution; you press two surfaces together until the friction holds it in place and there you have it. Solid, foolproof. If you looked at the press fit under a microscope though you’d realize there’s less than 30% of theoretical surface area that’s actually touching one another, the rest being void spaces. Most people, when they hear that an interference fit only has about 30% of surface area touching, are surprised that it can work at all. And while it may be fine for static applications, in dynamic load environments it breaks down.
Why friction is never the whole story
Interference fits are achieved by creating an interference (difference in diameters) between a bore and a shaft (or a bearing and a housing), and then forcing the two parts together until the elastic deformation of the part surfaces creates a clamping force at the interface.
The reason why that works is because of the surface roughness of metals relative to one another
Two supposedly flat surfaces, even if machined to a fine finish, exhibit roughness at a microscopic level. The two surfaces don’t actually touch each other at any given point; they interact at the peaks, called asperities, while the valleys remain pockets of air. These pockets of air can be seen with a powerful enough magnifier. The pocket spaces mean that the contact area between two mating parts can be significantly less than the theoretical surface area, which is why surface preparation is so important to interference fits. As long as the interference fit is static, those pockets don’t really matter. Once there’s a dynamic load applied, radial or rotational, those pockets become places for fretting corrosion to occur.
The progress of fretting corrosion
Fretting corrosion can be likened to the slow decay of a press fit. Not a sudden catastrophic failure due to overload but rather a slow breakdown over the course of months or years until it reaches the point when it needs to be repaired. Fretting corrosion is caused by the cyclic movement between the inner ring of the bearing and the shaft it rides on, or between the bearing outer ring and the housing. Movement on the level of micrometers initiates a number of deleterious effects at the asperities of the two surfaces. Micrometers of movement causes abrasion of the asperities, creating metallic dust at the interface between the two surfaces. Metallic dust at the interface, combined with oxygen, creates iron oxide (commonly known as rust) which is considerably harder than the metals at hand. These oxides then cause further abrasion on the interface, which causes more metallic dust, which causes more oxide, and soon enough, the interface is covered in abrasive dust. At this point the fretting has accelerated significantly due to the oxide dust; the increased abrasion has caused the clearance between the bearing and the outer ring to be much larger. More clearance means more leverage for the same dynamic load, which means even more oxides to be produced, which then eat away at the bearing and/or housing. The fretting has caused a self-accelerating mechanism that eats away at both surfaces in a mutually-destructive fashion. The resulting pits on the bearing seat or housing bore are the signposts for fretting corrosion that everyone sees when they open up a piece of machinery after a long period of service. This oxidation is exacerbated by the fact that the pit is a perfect pocket for storing corrosive byproducts until the next service interval. On a microscopic level, the interface looks like a red-brown pit or stain.
Fretting corrosion is especially problematic for heavy machinery where shafts deflect significantly under radial load. This causes unequal loading on the bearing, which causes the contact patch (the region of bearing where the load is carried) to vary on the bearing seat. This creates regions of the seat where the movement is allowed (higher clearances) and regions where the movement is restricted (lower clearances). Since movement is the root cause of fretting, it’s not hard to deduce that the fretting will occur in these regions of allowed movement on the bearing seat, usually localised only on one side of the seat. The amount of fretting is then dictated by the amount of load transferred through the bearing seat, which creates more oxidation. The fretting is exacerbated for misaligned bearings, causing extra fretting on the sides of the bearings due to the addition of a radial load component.
What traditional repair methodologies get wrong
When fretting corrosion has taken hold of a shaft seat until the fit can no longer be sustained, there are three primary repair strategies that are usually employed with varying degrees of success. None of them are particularly elegant nor do they address the root causes of fretting corrosion.
The first is to knurl the shaft in order to make it’s diameter larger until the fit is restored. The knurling process works by rolling a series of diamonds into the surface of the shaft, effectively displacing material until the desired diameter is reached. It is important to note that knurling only displaces material; it doesn’t add material to the shaft. Because of this limitation, the stresses introduced at the knurled regions can sometimes cause cracks at those diameters, especially if the knurling diameter isn’t large enough. Knurling is a relatively inexpensive and quick repair. It works, for a short while. The knurls rapidly abrade under the stresses of the bearings and become ineffective, leaving the repair crew in a worse position than they had been in before the repair was attempted.
The second method is to metalize the shaft seat, usually by a process called thermal spraying. Thermal spraying covers worn or damaged surfaces with a layer of metal, typically by melting the metal and spraying it on to the surface to be recoated. It works, but the process has limitations. Thermal spraying is a thermal process; heating the metal to the point where it liquefies and applies it to the surface of the shaft by force. The metallurgical properties of the steel being sprayed are changed due to the temperatures involved and there are stresses introduced by the rapid cooling of the sprayed metal when it contacts the (relatively) cold steel of the shaft; the result is flaking of the thermal sprayed layer, especially if the thermal stresses are incompatible with those of the parent metal. This incompatibility causes delamination of the thermal sprayed layer from the shaft, especially when the shaft is put under load. The repair may appear successful when the job is finished, until the machine returns to service and the thermal sprayed shaft seat fails catastrophically under the load of the bearings.
The third method to repair fretting-damaged shaft seats is to sleeve the shaft. Sleeving is, by far, the cheapest method of addressing a fretting issue, provided the shaft is not already at the required diameter. Sleeving is also quick, but it has the same issues as the other two methods in that it fails rapidly due to the mechanics of fretting corrosion. The same forces that caused the initial fretting still act on the sleeve, eating away at the surface of the new material. Sleeving only delays the problem without addressing the root cause of the fretting.
None of these three methods addresses the root problem that fretting corrosion is self-sustaining abrasion caused by the clearance between two mating parts. None of them change the fundamental mechanic problem that caused the initial fretting. They address the symptoms of a fretting problem and leave the cause untouched. All three methods are temporary, with varying amounts of time before the issue returns.
How anaerobic chemistry changes the contact model
Anaerobic adhesives cure once exposed to metal in an anaerobic environment (without oxygen). When applied to mating parts during the assembly process, they cure and fill the pockets at the asperities, creating a solid thermoset plastic. The resulting contact between the two parts is solid physical contact on nearly 100% of the mating surface, instead of the 25%-30% theoretical area in an interference fit. The cured anaerobic adhesive creates a filled joint that physically bonds the two parts together. This type of filled adhesive joint is stronger than the mechanical interference joint, where fit is dictated by the peaks and valleys of the mating parts. Filled adhesive joints are so common that they’ve found their way into the precision machinery world, where they are used as a complement to interference fits. Mechanical interference fits provide the initial location and the load-bearing characteristics, with the anaerobic adhesive filling the gaps and eliminating the relative motion between the two parts, eliminating the possibility of fretting corrosion.
The shear strength of the cured anaerobic adhesives on steel falls between 15-25 MPa, depending on the anaerobic grade, the surface finish, and the clearance between the mating parts. For most applications involving repair of damaged bearings fits in industrial machinery the shear strength of the adhesive must overcome the clearance between the bearing seat and the shaft (or housing and bearing), created by the fretting corrosion. The required clearance for such applications usually falls between 0.05-0.15 mm on the fits. Products described as Retaining Compounds are specifically designed to fill clearances and have much higher viscosities and lower shear strengths than their thin-grade counterparts. The filled grades offer much more leeway for out-of-spec fits. When talking about filled grades it’s also worth noting that the curing process takes longer due to the higher viscosity.
Calculating the shear requirements for such a repair is fairly straightforward.
Torque capacity of a repaired joint can be generally described by this equation:
T = (π/2)D²Lτ
Where T is torque, D is the diameter of the shaft, L is the length of the joint, and τ is the shear strength of the adhesive. Similarly, the axial capacity can be generally described with this equation:
F = πDLτ
After you’ve determined the torque capacity requirements for your application, it’s a simple matter of plugging in the numbers for the actual joint geometry to see if the shear strength of the adhesive is sufficient. It is good practice to apply a safety factor (usually 3:1 for applications with high shock and vibration and 2:1 for others) to the shear strength τ to account for dynamic and shock loads in the design before comparing it to the torque capacity. If the required torque capacity exceeds the torque capacity of the adhesive joint, you’ll need to look at other options, such as using a higher grade of anaerobic adhesive or lengthening the joint length. If those options are not viable then the geometry of the machine will have to be changed (machining the shaft seat) before reassembly.
The temperature has an impact on the final shear strength.
For most applications the temperature is a major consideration, as anaerobic adhesives are thixotropic; their shear strengths decrease with higher temperatures, and increase with lower temperatures. If an application requires the anaerobic adhesive to operate at 120°C, chances are the shear strength won’t be anywhere near the advertised 22MPa at 22°C. You’ll have to consult the manufacturer’s specifications to compare the target application temperature to the shear strength at that temperature.
Thermal expansion considerations in fits
Changes in temperature have an effect on the fit between a shaft and a housing. In situations where the two parts are made of dissimilar metals, or where one part is significantly hotter than the other (a common occurance when dealing with friction), the change in length due to thermal expansion has to be taken into account.
The linear expansion coefficient is important here, as it describes how much a material will expand per meter for each degress Celcius at a specific temperature. Steel, for example, has an expansion coefficient of approximately 12 µm/m per °C. Cast iron’s expansion coefficient falls around 11 µm/m per °C. The two have very similar expansion characteristics. Aluminum’s expansion coefficient is considerably higher at 23 µm/m per °C. A bearing made of steel being mounted into an aluminum housing that is subjected to high temperatures will experience lower intereference than expected due to it’s higher expansion coefficient.
A bonding adhesive provides a degree of flexibility to accommodate the changes in length by taking up the extra gap. The amount of flexibility can be calculated, but it usually doesn’t change the calculation requirements for a bonded joint. Cured adhesives can accommodate a change in gap size (and therefore change in interference) of approximately 0.001 mm for every millimeter of shaft diameter, before introducing micro-movement at the interface between the parts.
This is important, as micro-movement is the root cause of fretting corrsion. It is not a panacea for thermal calculations, however, as it is only a small amount of flexibility (on the order of thousandths of a milimeter) that prevents micro-movement. For fits where the interference is critical it is advisable to calculate thermal interference losses at the maximum operating temperature and ensure that it does not fall below the minimum interference required for the application. If a bonded joint is required to accommodate more than the flexibility of such a joint can provide, the geometry of the fit will have to be reworked.
Surface preparation is the weak link in the chain
The chemistry of anaerobic adhesives are sound; the problem is usually found in the preparation process.
Contaminants, such as oil, can reduce the shear strength of an anaerobic adhesive joint by as much as 80%. That isn’t an exaggeration; the shear strength of an anaerobic adhesive joint that is made on a clean surface has been measured and compared to the shear strength of a contaminated joint. The contaminated joint failed at only 20% of the shear strength of the clean surface joint. Always clean and degrease the mating surfaces before applying anaerobic adhesives. Degreasing rags can leave behind traces of whatever solvent was used to make them, especially if the shop is using a solvent that’s commonly found in the workshop. Always use a proper solvent with no residual hydrocarbons and allow the surface to completely dry before applying anaerobic adhesives. Inactive metals (stainless steel, anodized aluminum, zinc plate) have reduced cure rates on oxide surfaces. They require an activator to be applied before anaerobic adhesives adhere to their surface, especially on surfaces with a passive oxide layer. Failure to apply an activator results in a joint that has only 75% of the shear strength of the one made on active metals.
Surface roughness, and in particular the roughness from fretting corrosion, can play a role in the curing process for anaerobic adhesives. On one hand the roughness from fretting corrosion is an indicator that the repair is necessary. On the other hand, it provides an excellent bonding surface for the cured adhesive. Fretting damage can provide a suitable surface for making a repair with anaerobic adhesives if the surface material is ferrous. This does not mean that fretting damage should be encouraged, only that a machined surface doesn’t offer as much bonding surface area as a fretted surface. Because of this it is possible to make a repair on an existing interference fit without machining the shaft seat first.
A maintenance framework for detecting fits before they fail
Fretting damage doesn’t always progress into a catastrophic failure; with enough inspections and enough warning it is possible to catch the early signs of a failing press fit and take corrective measures before serious damage occurs. At each planned maintenance interval measure the diameter of the shaft seat and the diameter of the bearing bore with a set of micrometers. Take the measurements at 0°, 45°, 90°, and 135° around the bore. Fretting tends to appear at only one point around the bore due to the unidirectional nature of most loads, which means that the bore will start to oval with the growth in diameter occurring at the side where it experiences the load. If the ovalization on the measured bore is less than 0.05 mm, the machining of the housing and repair using a high-viscosity, gap-filling anaerobic adhesive can be postponed until the next maintenance interval. Once the ovalization reaches 0.15 mm the machining of the housing should be performed at the earliest maintenance opportunity. If the surfaces of the bore or seat are discoloured, it can be another sign that fretting damage has occurred. A healthy bearing fit typically exhibits a light bluish stain on the mating surfaces, whereas fretting damage will typically show as a red or reddish-brown stain or a powdery reddish film on the mating surfaces. This more serious sign of fretting damage should be repaired at the next maintenance interval and the shaft seat should be inspected as necessary.
The condition of the keyways should also be inspectied for damage that may have occured due to improper shaft fits. The key is designed to be the strongest element in a keyed connection and as such it is surprising to see a damaged keyway or weakened key. If the key is damaged or does not fit the keyway the keyway is itself a good indicator of a loose shaft fit. The loose shaft fit is causing the keyway to absorb the rotational impact of the movement that is happening in the loose shaft seat. The repair to the shaft seat should allow the key to survive for the life of the machinery.
The purpose of a preventative maintenance framework isn’t to prevent failures from occuring but to trend the wear rates so that corrective maintenace can be planned ahead of time. A shaft seat measured at 0.03 mm undersize at the last maintenance interval and 0.07 mm undersize at this interval shows a clear trend that can be used to plan the next corrective maintenance interval, which can prevent unplanned downtime and reduce costs.
Mechanical interference fits were a fantastic solution for the tolerances and loads of their time. Modern machinery requires higher speeds and higher loads and more time between maintenace intervals than ever before. The physics of fretting corrosion hasn’t changed but our ability to address the gaps around mechanical fits has.