Fitting and clamping of bearing rings

[JeanMicheling]

Hello everyone,

I'd like to discuss with you about the bearing and end cap torque for different bearings. For angular contact bearings, I have seen different methods. When I looked in the SNFA book, which is to me the best reference for angular contact bearings, they explain how to calculate the tightening torque in fonction of the preload and other parameters whereas NSK gives a unique torque value which is in fonction of the threads.Which one is the best? I'd say SNFA but I'd like to have your opinion. Moreover, SNFA says that this is a rule of thumb and the preload should be verified during operation. What is the best way to do it? NSK talked about three methods which are starting torque method, thrust static rigidity and natural frequency method but they don't give enough details. Does it really work great?

A second point is the locknut and end cap torque needed for non preloaded bearings such as deep groove or double row angular contact bearings. I haven't found something that could tell more about that. I know, I should call the bearing manufacturer, but honestly, I'd rather have the opinion of users than builders!

Thanks,

 

[diamondjim]

You almost have to rely on the manufacturer's
advice unless you know all of the specifics
of the contact angles and conformity for the
ball bearings or the specifics of the tapered
roller bearings. They in turn offer you some
guideline for the life based on the preload.
Too much preload will result in greater stiffness,
higher turning torque,but very short life.

 

[Tmoose]

Your description seems to mix 2 types of rigidly preloaded arrangements. First, standard machine tool practice of bearings clamped against themselves or pairs of precision spacer. Second, adjustable rigid preload, as common with tapered roller bearings.

Attempting to set preload by tightening end covers or nuts is danged risky at all but the very lowest speeds.

Machine Design had an article Sept 19, 2002 "How retaining forces affect spindle bearings". The tests they made sounded pretty scientific. One of their results was "excessive" OD clamping in an <> arrangement compresses the spacer and reduces preload. The conclusions were kind of controversial in my mind,
(first) since their recommended clamping forces were initially derived from such general recommendations (cover gap + 0.001 inch)
(second) their derived clamp force is wildly exceeded by recommended nut torque clamp

I forwarded that article to folks I knew in a big bearing tech department and got NO reply.

 

[JeanMicheling]

Sorry for being not clear enough, lets take that apart :

Angular contact preloaded bearing

SNFA recommends a force relative to the preload and other parameters whereas NSK gives a unique value independant from the preload. It just doesn't make sens to me. The clamping force should varie according to the preload and number of bearings. For the end cap, nsk recommands 0.0004'' to 0.001'' clearance before tightening whereas SNFA still recommands a value according to the preload. I worked with a spindle repair companie which used to set the clearance at 0.003'' to 0.005'' which is what I used to do. Tmoose, I'm not sure to understand what ''danged risky'' means (is it good or not) and what do you mean by 'cover gap + 0.001'')? Do you want to say 0.001'' clearance between housing and cover before tightening? If I understand your second point about the article, the effect on the spacer should have been worst on the inner ring since the clamping force is higher? According to NSK, excessive end cap clamping force will cause the ring to deform. I think there is a way to see this effect on vibration analysis but I'm not sure. But the real question is ''what is the best way to preload a bearing and ensuring that it's well done?'' . Another question could be ''what is the optimal end cap clamping force (or clearance) in order to avoid excessive deformation while keeping the outer ring well rested?''

Standard bearing

My question was more about a deep groove bearing than a taper roller bearing. With standard bearing, mechanics used to torque the locknut as bad as they can. I'd like to have an optimal value that will minimize the ring deformations.

 

[JJPellin]

In my experience, we use back to back pairs of flush ground angular contact thrust bearings in many applications. We tighten the shaft nut to clamp the inner rings, but do not torque that nut. We have considered it, but for flush ground bearings, I don't think it is super critical. We absoluately do not clamp the outer rings. For most applications, we specify that there is an axial clearance at the outer rings of 0.002 inch to 0.004 inch. If the outer rings are clamped in a back to back configuration, the bearing will overheat when it heats up and tries to expand. If you are buying bearings that are not ground for flush mount or if you are mounting them face to face, then you have to use a different method. For deep groove radial bearings we normally don't clamp either the outer or inner rings. The inner ring is a light interference fit to the shaft and is heated to install. The outer ring has wide open space on both sides to be sure it does not see any thrust load. I am not sure if this is what you had in mind.

 

[JeanMicheling]

JJpellin, I'm not sure to understand what you mean. The thrust bearing you talk about, are they the ones you need to preload? When you say you tighten the nut but not torque it, do you mean that you just rest the nut on the inner ring? Same question for the outer ring? If you have 0.002'' clearance before tightening the end cap bolts, I'd say that you clamp the outer ring. For your deep groove bearings, I'm surprise that you don't clamp the inner ring. You will get fretting and the ring will eventually slip from its initial place except if you have either small load or really high interference fit.

 

[JJPellin]

I should have been more specific. When I state that we use thrust bearings that are flush ground, I mean that they are ground so that then they are placed back to back with the inner rings clamped solidly together, there will be a very slight preload intended to be adequate to keep the bearing balls from skidding. We don't set the preload. It is designed into the bearing. When we tighten the shaft nut, we use a spanner wrench and hammer and rap it up tight (not just touching). When I said we don't torque that nut, I mean that we do not use a torque wrench to apply a certain amount of torque. The gap at the outer rings means that they are not clamped at all. With the end cover fully tightened, we bump the shaft for axial float and expect it to move 0.002" to 0.004". It is standard on API centrigual pumps for the radial bearing to be unrestrained. The inner ring has an interference fit and usually does not move even after many years in service. The outer ring is totally unrestrained axially so that no thrust load can be transmitted to the bearing. It is very rare for this radial bearing to move or frett even though nothing is holding it in place other than a very light interference fit between th inner race and the shaft.

 

[diamondjim]

JJPelin,
You are using "preload" and I think you
mean "interference fit". That is why the
replies are confusing. Interference
fit will reduce the clearances in the
bearings. Preload is generally defined such
that the bearings have no clearance
but actually a tight fit.

 

[diamondjim]

JJPelin,
My comments were meant for
JeanMicheling. Sorry!

 

[JeanMicheling]

Diamondjim, I wasn't talking about interference fit but preload (I really should be more clear). I was more concerned on the impact of the torque on the preload and ring distortion. So jjpellin, the 0.002 to 0.004'' clearance is the axial play you allow to the outer ring for thermal expension. This is pretty small? You shaft must be short? Quite interesting what you said on the inner ring with no clamp. I have no experience in centrigual pump. Of course, we can't assume it's feasible in every application. Also, because you have almost no axial load, it reduces the risk of slipping. On the wood industrie for instance, I have seen some issues that have been fixed by putting a locknut. But the bearing were carrying axial loads.

But what about the prelaod on angular contact bearing, I used to preload my bearings at full torque with no consideration of the preload needed. And for the end cap?
On standard bearing like deep groove, can tightening the locknut at the optimal torque can increase the bearing life? Is it negligible?

 

[diamondjim]

Jean,
You are confusing me. If the bearing is preloaded,
by definition, it is preloaded axially and radially.
Isn't this true?

 

[JeanMicheling]

It depends what you mean, with an angular contact bearing, with some type of loads, you can have a loose fit on the shaft which won't preload the bearing radially but still get an axial preload with the locknut. Of course, if you look at the balls, they are preloaded axially and radially due to the contact angle. Maybe you could point me what wasn't clear to you, I could have been wrong!

 

[diamondjim]

Jean,
It is the confusion between preload which does
deal with the bearing internal conditions and that
interference deals with the outer envelope
dimensions ie bore and od or width.

 

[JeanMicheling]

Diamond, I think I just said something wrong, if you have a loose fit on the shaft, you will for sure have a tight fit on the housing which will lead to radial preload. But I'm not sure to understand what you meant, could you develop a little more.

 

[diamondjim]

Jean,
I do not know if it is just semantics or not.

You can have a loose fit on shaft or housing
and still have a bearing preloaded or have
a bearing with clearance. The fit will affect
or reduce the clearance in a bearing. It will
increase the preload in an already preloaded
bearing, but I think you have already ruled out
using preloaded bearings for your application.

I think in some of your correspondence when you
are using "preload", you mean tight fit?

 

[JeanMicheling]

I think you got it right. I should have mentionned that preload caused by tight fit usually goes along with angular contact bearings or taper roller bearings.

 

[diamondjim]

Jean,
Tight fit is often used for the member that is
turning to assure that slippage does not occur.
Whether it affect the clearance is another issue.
Because you have a tight fit, does not mean that
you have any preload on the bearings. It will
diminish the clearance by a small amount. We
are in agreement that preload internally is not
a good thing for your application. So the bearing
manufacturer will recommend an initial clearance
that would exceed the effect that an interference
fit would cause to the internal workings of the
bearing to ensure that the bearing will have
clearance after being pressed on to the shaft
or into the housing.

 

[JeanMicheling]

I agree with you but on some cases, you have to deal with preload caused by tight fit. I got a fretting issue not so long ago, the only way I had to fix it was to increase the interference fit over the manufacturer recommandations. Interference preload is not suitable but fretting is, to me, even worst.

 

[JJPellin]

When I refered to 0.002" to 0.004" axial clearance for thermal expansion, this was not related to shaft growth. In our applications, we have a pair of thrust bearings on one end of the shaft and a radial bearing on the other end that is not restrained and can take no thrust. Any shaft growth causes the radial bearing to move axially within the housing. The axial clearance in the thrust bearing is to accomodate any thermal growth of that bearing pair. Since a typical bearing pair my only be 3 inches long, 0.002" is plenty of clearance to accomodate the differential thermal growth between the bearing and the housing. It feels as if some definitions may have reduced confusion. When I say preload, I am refering to a load on the balls of the bearing. Our radial bearings have a C3 loose internal fit with no preload. But in a horizontal shaft machine, the weitht of the rotor will load the balls on the bottom. For the angular contact thrust bearings I want a slight pre-load on the balls so that they roll rather than skid agaist the races. But I don't directly control this pre-load. I buy the bearings ground so that when I mount them with a tight nut on the shaft, they will have a very slight pre-load. An excessive interference fit on the shaft could distort the inner race and change this preload. Or an interference fit on the outer race (axially or radially) could also change the pre-load. But with a rotating shaft, it would not be normal for us to have any interference on the outer races (axially or radially). And the shaft interferences that we use are very small (typically 0.0001" to 0.0010" tight) and should have relatively little affect on pre-load at the balls. If you are installing a bearing that is not ground in the way that ours are, you very well might have to adjust the mounting to control pre-load. I would use the example of the wheel bearings in an old American car. They were tapered roller bearings. When you tightened the nut, you were loading up the rollers of the bearing. You snugged the nut up enough to put a slight load on the rollers, but not enough to cause significant drag and then you pinned the nut in place. If you were to fully tighten the nut in an appication like that, the bearing would have too much pre-load and would burn up quickly. You really need to get the information on the exact bearing that you are installing and mount it as the bearing manufacturer intended.

 

[diamondjim]

Jean,
So my assumption was wrong. You do want bearings
that have an internal preload to ensure that
fretting does not occur. I got confused by your
second point where you introduced angular contact
bearings with clearance. One of the beauties of
a nut is that you can tell by the amount of turn
of the nut and also dependant on the thread pitch
just how much axial preload you are applying to the bearings. Assuming that the contact angle is
45 degrees it then is a function of the sin of 45
degree contact angle that you are imposing on the balls.
The beauty of having preload on all of the balls in
the bearing is that more balls see part of the actual
loads imposed on the bearing.
I have not seen SNFA's book so I cannot comment on that.
How did you get a copy? I looked on the internet but
only found general information. Their literature that
I did see looks impressive. I assume you are from
Canada since the they are from France. I would think
any company supplying the aerospace industry would have
some very sophisticated equipment. The methods that
they suggest to determine the preload, can these be
done in house and at assembly?