Bearing race case depth vs. strength under pressure

[Lyrl]

My employer (a commercial heat treater) has been trying to develop a case hardening process for swivel pipes. The customer we are working with wants offer the pipes in a new material, and we've been doing test runs with parts made from their candidate material.

We developed a process for 0.020" effective case depth. They assembled the pipes together (one has an exterior bearing race and one an interior bearing race that fit together to provide the swivel functionality), added the bearing balls to the races, then filled the interiors of the pipes with water and pressurized it to 15,000 psi. Upon disassembly, they found a majority of the bearing balls had left dents in the races one or two thousandths deep. This is a failed test, presumably because of fears the bearings would seize up due to the dents. (In this application, wear is not a concern because the pipes would only be swiveled occasionally - but they need to have the ability to swivel freely throughout their life.)

The actual test requirement is that the races take an interior pressure of 22,500 psi without any molding to the bearing balls. They customer did a calculation like this: not all the ball locations dented at 15,000 psi so that's near the limit of the 0.020" case; 22,500 divided by 15,000 is 1.5; 0.020" * 1.5 + some amount to get past the near-limit + some safety factor = 0.045" new case depth requirement.

Due to the chemistry of the material, 0.045" is not achievable within the financial constraints of this project (the required carburizing cycle would be too long).

I'm not familiar with this kind of pressure test - is their math correct?

 

[tbuelna]

If your race surfaces exhibited permanent deformation of .001-.002" at some ball contacts after a single static test load cycle just 66% of the max qual test requirement, then you need to do some redesign of your swivel joint bearing. What you describe is a typical static brinell failure of a rolling element bearing race. What happens when you get even a tiny .001-.002" deformation in the bearing race surface is that as the ball travels back-and-forth over the deformation repeatedly, it quickly generates additional damage at the location.

Since you did not see uniform contact brinelling around the race surface circumference, this would seem to indicate that the ball complement was not uniformly loaded during the test. So your customer's calculations are likely not valid. As your ball complement is not sharing the load efficiently, the first thing I would suggest is doing some design optimization. Find the best combination of ball diameter, number of balls, contact angles, preload, race osculations, etc. Before you can determine what case depth and hardness profile is required for your race to prevent brinelling, you need to know the load applied at each ball in the complement at test conditions.

 

[FennLane]

How about a full complement needle roller?

 

[Lyrl]

The customer has a working design in a different material, so I think it's reasonable they're looking at heat treat rather than product dimensions as the culprit for the failed test. Also, the customer is looking at selling this as an "upgrade" material on their existing product line, so it has to be compatible with the existing product (making dimensional changes not really a possible option). This material just takes a long time to accept a reasonable amount of case; it may end up just being unsuitable for this application.

It is helpful to see that design loads are typically calculated in a more rigorous method than what the customer has currently done. We're going to do a couple more tests with this material before declaring it a dead end.

 

[Tmoose]

Do you have info about the ball size, etc?
Did the inner or outer consistently dent?

Do the pipes have to rotate while under high pressure?
I'm thinking of training "wheelS" in the form of bronze or steel bushings that keep the balls from experiencing the force of pipe expansion.

Hertz stress calculator for some geometries here.

http://www.amesweb.info/HertzianContact/HertzianContact.aspx

 

[tbuelna]

Lyrl-

Using a thicker case may or may not help. You mentioned the race "effective case depth" was .020", but was this as-carburized or after finish grinding? One problem you might have is the amount of case removed during any finish grind operation. If you start with a .020" carburized case and then grind away .015" from the surface, the remaining case will not have the hardness you require for a rolling element bearing race.

Depending on the ball diameter and race osculation, a .045" thick case may not perform any better than a .020" thick case.

It would be helpful if you could provide a sketch of the bearing installation with dimensions and how it is loaded. This would allow us to make better recommendations on how to address your problems.

 

[dgallup]

Have you measured the hardness of your case and compared it to the hardness of the current material? It seems to me you are low on hardness rather than depth but it could be both.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[Lyrl]

Tmoose - thanks for the calculator link, that is a helpful tool. For "force", should I enter the test psi as pound-force? If I do that with my best estimates of the dimensions, it shows the maximum shear stress at 0.040" depth, which might explain why we have bad test results with the shallower case depth.

Both the inner and outer dented, but the inner dents are much more visible. The dents are also each on only 1/2 of the race surface, so there may be some misalignment; I need to ask our customer how the races are aligned with each other. I will also ask if they need to be rotated while under pressure. Your suggestion about shims - would that be something placed between the flat surfaces of the two races, sized so that with no pressure it would leave a gap for rotation, but under pressure the flat surfaces and shims would take the load: no swiveling could be done under pressure, but this would protect the balls?

tbuelna - These parts are not machined after heat treat. The inner race has a maximum print diameter (the flat surfaces between the three races) of 3.97", and the outer race has a minimum print diameter of 4.03". The races are not dimensioned on the prints I have, but assuming the prints are to scale they are 3/8" diameter. I don't have any data about the balls, although based on the dents they are close in size to the races. (I count 32 dents.) The outer race has a hole to each of the three races used for loading the balls.

dgallup - Our surfaces are 56-57 HRC. We haven't been provided with samples from the current material, but based on what we know about it I'd guess their surfaces are about 60 HRC. Our core hardness is 34 HRC, which is a few HRC points higher than typical for the current material. I agree, the lower surface hardness is probably a factor in addition to the lower case depth (compared to current material).

 

[tbuelna]

Lyrl- Does your pipe swivel joint look like the one shown below? Does your joint use a single row of balls or two rows?

You mentioned internal fluid pressure was applied to the pipe system during the test where race brinelling occurred. But were any external moments applied to the swivel joint at the same time?

If the balls in your design are loaded thru a radial hole in the outer race, it is likely there is no separator used, right? Are you using the maximum number of balls (ie. a full complement) that will fit in each row?

Since there is no finish grinding performed after case hardening, there is probably a fair amount of distortion in the race surfaces. Could you provide some inspection data for the race surfaces, such as race surface profile, runout and gage diameter?

While you stated that finish grinding of the case hardened race surfaces is not practical due to cost, there are some other things you could do that would provide a more accurate race surface after case hardening which would not add much cost. One thing that might help would be to use some type of quench plug/fixture to minimize race radial distortions during quenching. Another thing that might help would be to pre-machine the race surfaces with compensation for any distortions that occur during heat treat. This would require a bit of development effort, but would cost little to implement in production.