replacing one bearing with multiple 11

[stevepmd]

Due to increased radial loading on a shaft, we need to reselect the (rolling element) support bearings - one bearing on each shaft end. Uniformly distributed load on shaft. Larger bearings would provide the required dynamic load rating - but would require a larger dia shaft and redesigned brg housings. Can we instead replace each of the current bearings with 2 or more of the current bearings to get the required rating? How would we get uniform bearing load distribution in such an arrangement? Is there a technical reference that discusses this approach?

 

[sreid]

Can you change the type of bearing (assume you are using Ball Bearings).

 

[stevepmd]

I could change the type of bearing, but my initial concern is the manner of load distribution if one bearing is replaced by several of the same kind - can I expect the load to be shared equally? If not, how might equal load sharing be achieved? If only one of the several bearings is taking most of the load, then there is no point to this approach.

This is a statically indeterminate problem, and may only be solvable by looking at shaft deflection. Maybe some combination of flexurally-stiff (roller) and flexurally-soft (ball) bearings are needed...

 

[dvd]

If you place two (or more) bearings at each end of the shaft, you will need to align the bearings very accurately, or face the possibility of heavy induced loads. Bearing alignment always becomes an issue with this type of setup, because the forces to deflect the shaft are added to the other bearing loads.

 

[hydtools]

You would have to use matched multiple bearings to replace one.

Ted

 

[tbuelna]

stevepmd,

You can use double row bearings at each end of your shaft, but they won't load share equally. The inner bearing rows will carry most of the load. As dvd notes, shaft/housing alignment is critical, and so is shaft bending.

Double row deep groove ball bearings will give the best result for radial loads, since they use duplex races and are manufactured as matched sets.

http://www.skf.com/portal/skf/home/products?maincatalogue=1&lang=en&newlink=1_1_21

Good luck.
Terry

 

[Clyde38]

Maybe some combination of flexurally-stiff (roller) and flexurally-soft (ball) bearings are needed...

Have you considered the use of tolerance rings?

http://www.usatolerancerings.com/bearings.html

http://www.machinebuilding.net/ta/t0132.htm

[link

http://www.linkedin.com/in/clydehancock

]

[/url]

 

[stevepmd]

Thanks everyone for the input.

I think what it boils down to is that my hasty assumption of uniform load sharing among multiple bearings is correct only in the case of a perfectly rigid shaft. Elastic flexure completely disqualifies this multiple bearing approach in reality. I really should have seen the impracticality of this approach earlier.

I did a quick FEA to investigate load sharing among multiple coaxial bearings on a shaft subject to radial loading. To the extent that the FEA model represents reality, it confirmed that uniform load sharing among the bearings is not possible. The first inboard bearing reacts all of the applied load. The remaining bearings actually react toward the load and thus add to the applied load - so adding bearings makes things worse.

More thoughts on this are welcome, as are any relevant literature references.

 

[ginsoakedboy]

I don't know much about bearings, but maybe you could explore squeeze film dampers for this application.

 

[BrianE22]

Can you modify the supports so that they are "elastically matched" to the shaft? Give them a compliance to match the shaft compliance.

 

[Tmoose]

How much of the basic dynamic load are you using? more than 1/8?
What life are you shooting for?
What do you calculate for shaft deflection within the stacked bearing arrangement?

Precision machine tool spindles often use (angular contact) bearings stacked up and even spaced apart a few inches in a single "arrangment" at the work end. Their shaft diameters tend to be generously large. There are programs to calculate the relative load sharing and crazy internal motions (ball excursion) that may result from relatively modest radial loads during operation.
The motions are such that I consider ball and roller bearings similar to living organisms. GOrgeous machining on all bearing seats is required for those things, in part to achieve the speed rating, but also to survive. "Tenths' (US) count and the machine shop must know their work will be checked. Ground diameters and faces are required.

It is possible to simulate a larger shaft with end clamped spacers and get some credit for reduced bending deflection. Again, gorgeou$ machining is required.

 

[Tmoose]

"The first inboard bearing reacts all of the applied load. The remaining bearings actually react toward the load and thus add to the applied load - so adding bearings makes things worse. "

I'd check my model, and think about each bearing as a spring with stiffness << 1,000,000 lbs/inch

 

[MikeHalloran]

This is when you call your favorite bearing company's applications engineers.



Mike Halloran
Pembroke Pines, FL, USA

 

[Warpspeed]

A self aligning double row ball bearing will load share, or perhaps a self aligning roller bearing would offer much greater radial capacity than balls.

There are other aspects to this such as speed and thrust loads.

A full confession to a bearing apps engineer might be in order.

 

[see3p0]

If you have the relevant data It might be worth your while using the SKF online bearing selection application.

http://www.skf.com/portal/skf/home/products?maincatalogue=1&lang=en&newlink=1_0_2

 

[patprimmer]

I agree with Terry initially then with Mike and Warpspeed. Time to talk with a specialist applications engineer. The better bearing companies will have such a service.

Regards
Pat
See FAQ731-376 for tips on use of eng-tips by professional engineers &

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

for site rules

 

[Tmoose]

"A self aligning double row ball bearing will load share"

SKF >>Single<< row ball bearing 6210 50X90 mm - fatigue load limit 0.98 kN
larger series >>double<< row ball 2210 - fatigue limit .57

those self aligning ball bearings have teeny tiny contact ellipses on the spherical ID outer race and as a result the allowable load rating suffers terribly even with the second meat patty.

 

[dhengr]

Stevepmd:
You answered your own question about why it is difficult to replace an overloaded bearing, with two bearings to increase bearing capacity. If the new loading overloads the bearings, is the shaft now overloaded too? You certainly better check that. Most everyone seems to be talking about ball bearings, but I don’t recall you being that specific. And, you say nothing about the actual shaft dimensions, loading conditions, rpms, etc. For the same general shaft size cylindrical bearings will provide a higher radial cap’y. under steady state shaft dynamics; and spherical bearings will provide higher radial cap’y. and tolerate some shaft wobble too.

 

[stevepmd]

Shaft will not be overstressed under the new load. I was non-specific about dimensions, RPM etc., because I was looking for general feedback about the feasibility (and consequences) of retrofitting one bearing with multiple identical bearings.

Right now we have one 25mm bore double-row ball on one end, and one 25mm bore single row ball on the other. I know the sensible thing to do is just go with larger bearings - but what if there's no (radial) space for them, or can't increase the shaft dia?

There's a gear mesh involved too, so I don't want new bearings to allow more shaft flexure. So that disqualifies spherical roller (maybe tandem spherical rollers?)

Thanks, I appreciate everyone's input.

 

[dhengr]

Greater load on the same dia. shaft will cause greater shaft deflection and stresses, so you may want to take a look at your gear mesh situation. And, that usually means shorter fatigue life for the shaft too. No bearing will prevent this situation from occurring, it must either tolerate this, an increased shaft angle (slope) through the bearing, steady state or not, or you will just wear the bearing out. This was sorta the mistake of your original thinking and question. It is very difficult (almost impossible) to make two bearings treat the shaft as if it has a fixed end w.r.t. shaft deflection. And, your calcs. or FEA showed this, that is you overload the inner bearing by the amount of the uplift on the outer bearing. You must tolerate that shaft slope in the bearing and at the gear mesh, or you just grind things up. And, the problem gets more difficult if the shaft deflection is not a pretty steady state condition. Sometimes the bearing can be set in a spherical housing to match the shaft slope (elastic matching) then there is not much more movement at this interface, under steady state conditions; or spherical roller bearings tolerate this within the race, particularly good for some shaft wobble. Then in its simplest form you shim a couple bolts on your gear motor mounting base, to align it with the shaft slope.