Frictional Loss in Roller Bearing

[crunchie12268]

Hello,

I am in the process of selecting a bearing for a tension control application and came across a question I would like to bounce of someone. Say I have a cylindrical roller bearing with a radial force of 4000lbf. Assuming everything else remains the same I increase the radial force to 5000lbf do I see a decrease in, what I am going to call, a friction moment? I guess I just expected to see an increase in friction but wasn't sure if bearings are designed to do this? Thanks

 

[GregLocock]

Yes for roller bearings you get a rolling resitance proportional to the radial load.

Timken gives an equation.

http://catalog.timken.com/WebProject.asp?CodeId=7.5.9.17&BookCode=crb11flx

page 45

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[crunchie12268]

Thanks

 

[electricpete]

Also see the FAG Lubrication document
Google "1.2 Calculation of the Frictional Moment" and look at the first entry.
Or maybe this link will work (not sure):
http%3A%2F%2F

http://www.m3.tuc.gr/ANAGNWSTHRIO/S...=pAW4OokurO1xt3hMh8WlRA&bvm=bv.59568121,d.cWc

It has spread out over 2 pages a detailed discussion of
M0 [N mm] load-independent component of the frictional moment
M1 [N mm] load-dependent component of the frictional moment
also variation of these components with speed.
A 3-d figure of these components and variation with speed... also a discussion of mixed component... makes me scratch my head.

To my simple thinking the sliding that results from deformation at the contact surfaces (non-ideal rolling) would be load dependent, and the sliding that results from interaction of the rollers and the cage would be load independent. But clearly there's a lot more to it.




=====================================
(2B)+(2B)' ?

 

[boo1]

The measured function would be inversly proportional

Attached is Electricpete's linked pdf

 

[Tmoose]

I wouldn't have said bearings "are designed" to have greater frictional moment with heavier loads.
That is simply what happens.
Otherwise folks would "design" them to have NO frictional moment, if it weren't for the kickbacks from the oil and power giants.

Looking at that FAG document Note even at middlin' loads the lubricant (oil) churning etc that make up the "load independent" part of the frictional moment exceed the bearing's EHD frictional moment.
Page 17 mentions Sealed, greased bearings friction can be 3X higher or greater (after even hotter run-in) and are best determined by "FAG measuring system R27". I'm //guessing// that is an actual test device.
As far as I could see no calculation to predict FM for sealed, greased bearings is offered.

I download all SKF and especially FAG (and others) bearing documents from the 60s thru 90s because as near as I can tell they are no longer available. The online calculators are convenient but fall short in their explanations and entertainment value.

 

[tbuelna]

crunchie12268-

Oil lubed rolling element bearings, like your cylindrical roller bearing, incur two basic types of mechanical loss- frictional loss and viscous loss. With a high DN bearing, the viscous losses will predominate. With a lower DN bearing, the friction and viscous losses may be fairly similar. In your example you specifically asked about the difference in "friction moment" between two identical bearings operated under identical conditions with the exception that one had a 5000lb radial load and the other had a 4000lb radial load. But since you provided no other details of how the bearing is operated it is not possible to give you a precise answer.

With most bearing systems, what is usually of primary concern is power loss rather than friction torque. Friction torques are usually only a concern with bearing systems that oscillate or rotate very slowly. Thus, with bearing systems that constantly rotate, what matters are the combined friction and viscous losses. Based on your question about two identical bearings with one having a 25% higher radial load, the answer would be that the bearing with the higher load would have lower losses as a percentage of the power transferred through the system. Basically, the bearing with the lower load is oversized for the application.

 

[electricpete]

Based on your question about two identical bearings with one having a 25% higher radial load, the answer would be that the bearing with the higher load would have lower losses as a percentage of the power transferred through the system

You lost me on that one.
We compare effect of load for two identical bearings at same speed.
The higher loaded bearing has higher losses.
What basis do we have to infer anything about power transferred through the system (alternatively: why would we assume anything other than that the power transferred through the two systems is the same).

=====================================
(2B)+(2B)' ?

 

[tbuelna]

electripete-

No, the higher loaded bearing would not necessarily have higher losses as a percentage of the power transferred thru the system. Consider the case of a high-speed gearbox where bearing DN is high and loads are low. With such a case, bearing viscous losses can easily be many times friction losses, and viscous losses are independent of radial loads.

....What basis do we have to infer anything about power transferred through the system (alternatively: why would we assume anything other than that the power transferred through the two systems is the same)....

If we assume that the bearing radial loads from the gears are 25% higher for the same rpm, then we can assume that the power transferred is also 25% higher. In this case, the friction losses for the bearings with the higher radial loads might only be about 2% or 3% higher, while the bearing viscous losses would be similar. Yet the combined losses of the more highly loaded bearing system, as a percentage of the power transferred thru the system, would be lower.

In short, with regards to efficiency of high DN bearing systems, it is critical to optimize the pitch diameter of the bearing for a given load. If a cylindrical roller bearing has the capacity to accommodate a 25% increase in radial load for the same duty cycle, then it is substantially oversized and will not provide the best efficiency.

 

[electricpete]

If we assume that the bearing radial loads from the gears are 25% higher for the same rpm, then we can assume that the power transferred is also 25% higher.

With gears I can see the logic of your comment.

But the op said nothing about gears and your first comment said nothing about gears... so at that point it seemed kind of bizarre to me that you would invent a relationship between radial bearing load and transmitted load.

Thanks for explaining what you meant.


=====================================
(2B)+(2B)' ?

 

[drawoh]

...

As far as I could see no calculation to predict FM for sealed, greased bearings is offered.

....

My experience, admittedly with small bearings, is that bearings are sealed with rubber seals. These cause substantial friction, way in excess of the rolling resistance of the balls or rollers.

My college machine design textbook (V.M.Faires) has notes on bearing friction. Perhaps yours does! According to Faires, seals can increase bearing friction by several hundred percent. His reference is Shaw and Macks, Analysis and Lubrication of Bearings, McGraw Hill. This reference appears to be available online.

--
JHG

 

[tbuelna]

But the op said nothing about gears and your first comment said nothing about gears... so at that point it seemed kind of bizarre to me that you would invent a relationship between radial bearing load and transmitted load.

Your are correct that the OP did not specifically mention gears. There was only a vague reference to a "tension control application". I simply used the example of a gearshaft bearing set to illustrate that simple friction moment is often not the primary concern when optimizing the design of rolling element bearing systems.

Since you did not like my example of a gearshaft bearing system, how about considering the example where the bearing supports a belt drive pulley. As the power transferred by the belt drive increases, the belt "tension control" system applies a greater radial force thru the pulley bearings to prevent the belt from slipping. In this case, all other things being equal, wouldn't a 25% increase in radial force at the pulley bearings used to create tension in the belt basically equate to a 25% increase in transmitted power? Same pulley rpm, same pulley diameters, same margin of safety for friction between the pulley/belt, etc.

Section 9.2 of this NASA handbook gives a good explanation of all the factors involved in mechanical losses of rolling element bearings.

Regards,
Terry

 

[3DDave]

Terry,

Thanks for the link. Kinda funny how some will jump your case. I followed along right away.

 

[electricpete]

In this case, all other things being equal, wouldn't a 25% increase in radial force at the pulley bearings used to create tension in the belt basically equate to a 25% increase in transmitted power?

It could be an increase in transmitted power without changing position of the tensioner as you say. On the ohter hand, it could also be a change in position of the tensioner without changing transmitted power. At any rate, not very important any more.

=====================================
(2B)+(2B)' ?

 

[tbuelna]

3DDave-

Thanks for the reply. My intent was to simply get the person asking the original question to think about the situation, and hopefully respond with more specific questions. Unfortunately, I'm not very good at expressing things in writing myself. So if someone "jumps on my case" a bit over one of my posts I probably somewhat deserve it.

 

[drawoh]

...

Since you did not like my example of a gearshaft bearing system, how about considering the example where the bearing supports a belt drive pulley. As the power transferred by the belt drive increases, the belt "tension control" system applies a greater radial force thru the pulley bearings to prevent the belt from slipping. In this case, all other things being equal, wouldn't a 25% increase in radial force at the pulley bearings used to create tension in the belt basically equate to a 25% increase in transmitted power? Same pulley rpm, same pulley diameters, same margin of safety for friction between the pulley/belt, etc.

Transmitted power should not be significantly affected by the friction torque from the bearings. If it is, the bearings absolutely are not working.

Gears and pulleys apply side loads to shafts, but these are normal to the driving force. The only thing that affects the output power is the friction torque.

--
JHG

 

[3DDave]

tbuelna, I like the NASA Rolling-element Bearing report. For me the best parts were where centrigual acceleration causes the contact patches on the inner and outer races to change, plus the diagram for steel shaft fracture limitation that goes to 1E6 RPM.

drawoh, It looked to me like the point was the reverse position, that friction torque can be significantly affected by transmitted power when the load at a given RPM is proportional to transmitted power.

 

[tbuelna]

Transmitted power should not be significantly affected by the friction torque from the bearings. If it is, the bearings absolutely are not working.

Gears and pulleys apply side loads to shafts, but these are normal to the driving force. The only thing that affects the output power is the friction torque.

drawoh, your first comment about the impact of bearing friction on the mechanical efficiency of a rolling element bearing system not being significant is basically true. All other things being equal, an increase in radial load can affect friction losses slightly due to the influence of radial load on hysteresis and thickness of hydrodynamic fluid film contacts.

The OP did not really provide enough information to allow a definitive answer to the question. Unfortunately, in real life, the design of rolling element bearing systems can be quite complex, and there are often other factors that contribute to mechanical losses far more than friction. For example, with high DN roller bearings things like viscous/churning losses and skidding of the rollers in the unloaded race sector can be serious concerns.

The basic premise ("all other things being equal") behind the friction question asked in the OP ignores the reality of how rolling element bearing systems are designed. A well designed rolling element bearing system (ie. one that is fully optimized for fatigue life, efficiency, reliability, cost, weight, size, etc.) would not likely be able to accommodate a 25% increase in load without modifications. And the modifications needed to accommodate a 25% increase in radial load (typically an increase in roller complement PD) would likely increase friction losses in the bearing.

This is a very interesting topic of discussion to me. Maybe crunchie12268 can provide some additional details of the bearing system in question.

Regards,
Terry

 

[drawoh]

tbuelna,

For the record, my reference, V.M.Faires, quotes equivalent friction coefficients, which means that the friction torque in the bearings is proportional to the side load. The friction coefficients quoted are between .0011 for deep groove ball bearings, to .0045, for needle bearings. The values are based on a load that allows one million revolutions, "and are suitable only for order of magnitude approximations".

Regardless, the friction is tiny, and it is unlikely that something else is not orders of magnitude more important.

--
JHG

 

[Tmoose]

All the pinewood derby folks are convinced heavy cars go faster than light cars. opinions about front/rear weight distribution are split.
The rules DO have a max 5 oz weight limit.

The best guess I can come up with is there is something about steel/plastic/graphite bearings' coefficient of friction that lowers with increasing radial load.