Tapered roller bearing vs Drawn cup needle + needle thrust bearing

[sensij]

The application is supporting a roller in a miniature forming mill. The speeds and loads are relatively low, 90 RPM and approximately 50 - 200 lb of radial load. The only real axial load comes from mounting preload, the temperature expansion should be insignificant at these sizes and speeds. The rolls are not driven, so at this low speed vibration should be low. Shock loads should also be minimal, although it is possible to create them by mis-setting that roll positions.

The existing design has the roll shaft supported on each end by a full complement drawn cup needle bearing (B-1212), with a needle thrust bearing (TRA-1220) between the working portion of the roll and the needle bearing housing. Lithium grease is used for lubrication, applied a couple times per week. The working portion of the roll is only 1" wide with a 2" diameter, so there should be very little axial deflection under load. The shaft and bearing housing are both hardened tool steel.

Roll position is critical, axial position needs be held to less than 0.001". Radial position under load should also be held better than .001", but is not as critical as axial position. There are top and bottom rolls in a single stand that need to align to this tolerance, and several roll stands in series that also need to align. The housings and ways are ground to 0.0001" tolerances.

In practice, the existing bearing setup does not maintain the position tolerance required, because the needles wear quickly. After a few months of operation (typically, 90-100 hours / week), there is at least a needle diameter's worth of space in the full complement. The drawn cup is pressed into the housing, and has a loose fit on the shaft. More frequent bearing replacements is one option, but I'm hoping for a better option.

As a band-aid, I have started using drawn cups with cage-mounted needles (B-1212). I'm hoping that removing the needle on needle contact will help slow down the needle wear, but it is too soon to know if it is helping yet.

I am considering biting the bullet and doing a housing redesign using tapered roller bearings (A2047 / A2126) instead of the needle cup and needle thrust assembly. Timkin's lifetime calculator rates these as having > 40 million hour life. I understand these need to be preloaded to avoid sliding, but I do not have enough experience to know how to do the preload in a way that maintains positional tolerance through a reasonable life of the bearing. Also, I think I understand from the catalog that if the axial load exceeds e (41%) of the radial load, sliding could also become a problem, but I'm not as sure about that.

Any thoughts out there on whether TRB's are right for this? Having the new housings ground to the tight tolerances required is not cheap, so I'm looking for confidence that this is a path that will yield better results.

Any ideas on why the drawn cup needles wear out so quickly? The load rating suggests they should be fine. Misalignment between the inner and outer housings is the first explanation that comes to my mind, but maybe there is something else I should be looking at?

Thanks,
Jason

 

[Tmoose]

Looks like the basic dynamic load rating of the full complement B-1212 is 4900 lbs.

http://www.koyousa.com/brochures/pdfs/catb2016e-4.pdf

FAG would have said that it it was a ball or roller bearing with all race hardness lube and cleanliness parameters met the endurance life would infinite at 4900/8= ~ 600 lbs.
The normal HK dyn load is a mere 1500 lbs.

Have you consulted the manufacturer?

First cut suggests the speed is too low to achieve EHD lubrication. That is a big problem for decent life.

I think I'd be looking at using grease with a real high base oil viscosity and extreme pressure and anti-wear additives. The required lube can be calculated or otherwise determined based on rpm and bearing size. See page A-35 in the Koyo link above.
Also The hardened shaft needs to be over RCH 55, the right finish (smooth with no grinding or heat-treat checks), good geometry in regards roundness, coaxiality end-to-end, sufficient case depth ( calculated from load and bearing geometry), the shaft stiffness must be sufficient to maintain good coaxiality, and sealing must be good to prevent contamination.

 

[MikeHalloran]

Your data has to be bogus.
A mere 200 lb of radial load at a roll bearing is not credible even for a penny roller. The short needle life tends to corroborate that assertion.
Where did you get that number?

Is there a load limiting mechanism in the mill, like a big spring in the nip jacks?
Even if there is, consider the possibility that your users are cranking down harder than you thought, or designed for.

I wouldn't attempt a design change until I had more confidence in the load numbers, so that the predicted needle bearing life comes close to the life you have measured.



Mike Halloran
Pembroke Pines, FL, USA

 

[sensij1]

The load estimates come from load cells that I've placed at the end of the jack screw and monitored while the mill is running. Yes, the mill is capable of delivering more force than the estimates I've provided, but we are working with strip that is only 0.002" - 0.010" thick, it doesn't take much force to bend it. If the strip were to wad up it would create a nice shock load, but this rarely happens. From a theoretical perspective, let's say the contact area is 0.1" x 0.045"... that is about 3 degrees of contact from the roll. 200 lb would be 44 ksi, near the yield strength of 304. If the forces were much more than that, we'd actually be thinning the strip. I don't know the exact contact area, but I think the orders of magnitude here are correct, or not off enough to explain why 1000000 hours of rated life turns in to 1000's.

I will spend some more time with the load cell and try to confirm the numbers again, but in the meantime, here is a picture of the worn cup needle bearing. This one had been in service since August, and was pulled out today.

 

[MikeHalloran]

Thanks for the excellent photo.
I see some indications that catalyze questions.
Understand that I may be misinterpreting photographic artifacts, of course.

Do I see fine particulate material on the needles?
Or is that spalling?
Or just grease?

The near ends of all the needles appear unusually bright, right at the intersection between their OD and the end face.
Are they perhaps riding on a radius or shoulder at the proximal end of the shaft's bearing surface?

Are the needles blued relative to new ones,
or is that just shadow?



Mike Halloran
Pembroke Pines, FL, USA

 

[sensij]

There is a little bit of particulate mixed in with the grease on this one, because I recovered it from a trash can full of sawdust after the bearing change today. I made some effort to clean it up for the picture, but it no longer represents the as-pulled condition. We have another bearing change scheduled for next week, so I'll try to get a more representative sample then.

I think the bluing is just shadowing, but I'll look closer at it tomorrow. I've taken pictures from both ends to try to help distinguish shadow and perspective effects.

The brightness at the ends is a interesting observation. I'll look through some more of these and see if that is consistent, and confirm that it is not present on the bearings out of the box. There is still a thrust bearing + washer between the end of this cup bearing and the shaft's corner radius, so the needles can't be hitting it. The thrust bearing is riding on the cup needle bearing's housing, but the cup needle bearing is set back slightly so there is no direct contact between the two bearings.

Also, I didn't notice it until looking more closely at this sample, but the wear isn't uniform down the length of the needle. The stamped end of the needles is worn slightly more, making the gap that has formed sort of trapezoidal. I might be able to get a pin gage in there and see what the difference in wear is between the ends. I am sure that the bearings are oriented consistently so that the stamped face is always facing in or out, but need to double check on what the orientation actually is.

Thank you for the comments. Based on TMoose's suggestion, I've started having a conversation with an engineer at Timken about this as well. They don't make needle bearings anymore, but he should be able to help me assess the suitability of the tapered rollers. However, If I'm missing a wear mechanism that these bearings are exposing, the inputs I'm giving him might as well be bogus.

 

[sensij1]

Attached are a couple more pictures. I went through the last set that was pulled out and found one so worn that the

 

[sensij1]

Sorry about this mispost, I misunderstood how attachments work.

Anyway, the outboard side of the same bearing pictured about is shown here.
Outboard = away from the work roll = round side of the drawn cup.
Inboard = facing the work roll = stamped side of the drawn cup.

You can see a pattern of burnish marks that didn't show up on the inboard side. On this bearing, I could get a 0.0525" pin into the gap between rollers on the inboard side, and a .0320" pin in the gap between rollers on the outboard side, so the wear is definitely not uniform across the roller.

Another bearing that had been pulled out of the same mill was so badly worn that the needles fell out of the shell. A few of them are pictured. You can see about 20% of the length on the left side is burnished differently than the rest. The diameter of the burnished end is around 0.0932", the middle of the roller is 0.09405", and the opposite end is 0.0941". I don't know what the original needle diameter was, but 27 of them are used to make the full complement, so a line-to-line maximum diameter is 0.0985".

Is it correct to conclude that this wear pattern would result from the shaft axis not being parallel to the bearing axis? I've measured a couple of shafts and am not seeing any taper in the diameter that could also explain it.

 

[caoimhin1]

900N radial load, 90rpm? Why not just use a mounted ball bearing unit? Takes care of alignment, integrated sealing, easier to install than tapers and no housing design necessary. h8 tolerance on the shaft.