Axial Clearance for Locating Ball Bearing

[Jieve]

I have a parallel shaft application where a motor drives two shafts via spur gears, one on each side of the motor. On each shaft is mounted a 22cm diameter wheel to which a weight can be attached to generate an imbalance – this is going to be used in an academic environment to demonstrate vibration in machines. Think of it like one of those vibrating filter screens used to separate different size pieces.

The bearing housings are simple custom designed split housings using SKF 6004 deep groove ball bearings. After discussing the application with SKF’s application engineering manager, we came to a setup that properly preloads the bearings and correct fits. Both outer rings on a single shaft should be interference fit and the inner rings clearance fit. The outer ring of the non-locating bearing butts up against a housing shoulder and the side (of the outer ring) facing the other housing is free to move axially. The inner ring of the non-locating bearing is axially preloaded using a Belleville spring which is tensioned against a spacer ring on the shaft and rotates with it. My design had housing shoulders on both sides of the locating bearing to locate the outer ring axially. When I asked what the axial fit should be, the SKF engineer told me that the outer ring of the locating bearing does not need to be located axially due to the interference fit. Does this sound right? I always thought an interference fit wasn’t enough to locate a bearing axially.

I was told by SKF that both bearing caps and housings need to be tightened to specified torque and bored together. While discussing the bearing housings with the machinists, they recommended that I use retaining rings on both sides of the locating bearing and as the housing shoulder for the non-locating bearing so that they could bore straight through the part (cap & housing) and keep the proper bore tolerances. It would certainly make machining more doable. Does the retaining ring setup sound like a reasonable solution? Anything I should be vary of here?

Thanks for any input.

 

[JJPellin]

I am not sure I understand why you would have an interference on the outer ring with a rotating shaft. But, I will assume that this is correct for your arrangement. I like the idea of using retaining rings to restrain the outer race of the bearings. If this was a machine designed to be well aligned and well balanced, SKF would probably be correct. The interference fit would probably be sufficient with no shoulders or retaining rings necessary. But, you have a machine that is designed to vibrate. That vibration could result in unanticipated movement of those outer races over time.

The vast majority of our applications use deep groove radial ball bearings on rotating shafts. We use an interference fit on the inner race and a slip fit on the outer race. Since the inner race is an interference, it has a shoulder on one side just to allow the mechanic to properly position it. The other side of the inner race is not retained. This would be comparable to your tight fit outer race. But, we go to great trouble to make sure that our rotating assembles are well balanced and well aligned so that vibration is minimal.

Just make sure that you are not restrained axially in such a way that you impose significant thrust loads on these bearings. They are designed for radial loads and are not suited to hold much thrust. You describe the non-locating bearing running against a housing step. If the shaft grows when it heats up, will this impose a thrust load on both of the bearings as the outer race of the non-locating bearing attempts to move in the direction of the housing step? This may be accommodated by the arrangement of the slip fit mounting of the inner races on the shaft. But, without a diagram, I am having a hard time picturing this configuration.


Johnny Pellin

 

[Jieve]

JJPellin,

Thanks so much for the response. I was told by SKF that based on the nature of the machine (vibration inducing) that the outer races would need an interference fit and the inner races loose, I'm assuming due to the fact that the load direction is constantly changing direction. It sounds like you're picturing the set up correctly, as soon as I'm back at work tomorrow morning I'll post a pic or two.

Thanks!

 

[Tmoose]

Is the "centrifugal force" greater than the rotor weight?

 

[Jieve]

Yes Tmoose, the rotor force due to gravity on each bearing is around 25N and the imbalance force at max speed is around 150N per bearing. The approx net force alternates between 125N up and 175N down per bearing each revolution. I needed the axial preload from the Belleville spring to bring the setup to the min static load necessary for the bearings.
Will post pics in a couple hours.

 

[Jieve]

Here are pics of the bearing setups.

 

[caoimhin1]

Rotating load, so they are correct to specify an interference fit on the OD, similar to a vibrating screen application. I normally never specify axially locating a bearing by fit alone. It might be a pain to accommodate, but adding a circlip can give some peace of mind against the bearing "walking" off the shaft or housing.
Btw, interested to know how much support you got from the SKF engineers. Normally bearing company engineers run a mile from academic applications...too much work for zero sales value.

 

[Jieve]

Thanks for the response, in this case the preload from the disc spring should keep the bearings from walking off the shaft. I will go with the retaining Ring idea to keep them axially located though, so the bearing doesnt walk in the housing.

Actually in my case they were surprisingly helpful, the applications engineering manager called me and worked with me for at least 30 mins On the phone or longer, answering all my questions and working through alternatives. Then we even talked again the next day after we discussed possible changes. I was actually really impressed.

So what exactly is the problem with a rotating load that requires an outer interference Fit? Does the increased friction moving around the bearing cause the outer ring to rotate? Manuals always Say mount it like this, but Don't explain why.

Also, how much clearance would U say should be between the retaining rings so it doesnt walk too much?

Thanks again for the answer.

 

[JJPellin]

The goal is to build it in such a way that it is practical to assemble and disassemble. You need to be able to control the loads on the bearings such that they stay within the capability of the bearings. Using an interference fit on both the inner outer races makes it difficult to satisfy these two requirements. But, if one of the fits is a slip fit, you need to avoid significant movement in that joint. If two metal parts move relative to one another, there will be wear, fretting corrosion and damage.

For a typical installation with a horizontal shaft that rotates. The primary load is gravity. This is in a fixed direction. So, if you made the inner race a slip fit, there will be movement between the shaft and the inner race with each rotation of the shaft. With every revolution, the race will tend to move to the opposite side of the shaft. This can add up to millions of cycles quickly.

For the outer race, on the other hand, the load is in a constant direction, so the race will tend to move to one side of the housing (down) and stay there. You have much less movement and much less chance of fretting or damage.

Your application is unusual in the fact that is has a large imbalance. So, rather than gravity dominating the dynamics of the loads, it is the imbalance. The imbalance rotates with the shaft. So, relative to the inner race, the direction to the load is constant. And relative to the outer race, the load is moving. So you reverse the fits and go with an interfere on the housing and a slip fit on the shaft. This is common in applications where you have a stationary shaft and a rotating housing.

I am not sure what the proper value for axial clearance at the retaining rings should be. Based on the rest of the machine, how much movement can be tolerated? For a pump shaft in an API pump, we normally leave 0.002” to 0.004” axial clearance in the locating bearing. But, I suspect you would want more just to make it more practical to manufacture. I doubt if you would need to stay any tighter than about 0.030” axial clearance at the retaining rings.

As I noted you need to be careful about putting too much thrust load on your radial bearings. I assume you have designed the Belleville washer to provide less axial force than the bearings can take.

It is not obvious to me how you will assembly this machine. If the bearings are chilled or the housings are heated to allow the interference of the outer races, then how do you install the shaft? We generally don’t like to have to put everything together at the same time with some parts heated or cooled. Make sure you have a good plan for getting all this together. If you try to press the outer races into the housings, be careful that you do not transfer the loads from the press through the balls or you may have a premature failure.


Johnny Pellin

 

[Tmoose]

"So what exactly is the problem with a rotating load that requires an outer interference Fit? Does the increased friction moving around the bearing cause the outer ring to rotate? Manuals always Say mount it like this, but Don't explain why. "

Like Johnny Pellin said.

Here are examples of the same mechanism that make inner ring interference necessary in most horizontal shaft applications.

http://nikkipilgrim.files.wordpress.com/2011/12/large_hula_hoop_models.jpg

http://www.kxcad.net/RecurDyn/Recur...ar/Geometries_Entities/Spur_Internal/img4.gif

 

[Jieve]

JJPellin and TMoose,

Thanks so much for the detailed answer and explanations.
With regard to assembly, since the inner rings are clearance fitted, the installation on the shaft is relatively simple. The inner race is confined axially on the locating bearing side by a shaft shoulder on one side and a sleeve on the other. The inner race of the non-locating bearing is confined axially in one direction only, against a shaft shoulder. The outer races are interference fitted, however the bearing housings are 2-piece pillow block type. The shaft can be assembled and laid into the bearing housings, then the caps are screwed on. Assembly in this configuration is relatively simple. The retaining rings will hold the outer rings of the bearings in place axially. Again, the locating bearing will have retaining rings on each side, and the non-locating only on the outer side (opposite of the shaft shoulder). Since the load goes through zero during rotation and the bearings require a minimum static load the axial preload is necessary. With the Belleville spring the resulting axial load is pretty low, far below the max allowable, but keeps the min static load requirement satisfied.

Since all 4 housings (2 shafts) are screwed down to a plate with standard H13 hole tolerances, there is a lot of allowance for alignment of the parts. This “slop” is also necessary to preload the Belleville washer; dowel holes will be drilled for location of the bearing housings to the base plate once the housings are fully assembled and properly aligned. The more I was thinking about the assembly, the outer ring of the locating bearing will be loaded up against the retaining ring due to the need to preload the Belleville washer. Therefore there should be no tendency for the bearing to wander axially in the opposite direction, even under a vibrating load, so maybe that’s what SKF meant when they said the axial clearance didn’t matter so much.

I had read in a different post that parallel shaft applications generally either require extremely tight tolerances or some slop in one of the shaft setups to accommodate for alignment. This seems to be fitting in my case, but can anyone elaborate on this statement with examples? Just trying to get a firm understand of different design applications using ball bearings.

Again thanks so much for taking the time to answer.