Concentric Housing to Shaft design question

[cnuk]

I am building a piece of test equipment that has five housings (approx 6" OD x 3" thick) bolted together. On each end of the bolted together housing assembly I have ball bearings which guide a shaft through the housings. I am trying to maintain the housing to shaft concentricity as tight as possible. For example, we originally had the housings with an ID of 2.002" and a shaft of 1.999". Upon assembly there is contact between the shaft and housing. This is predictable based on our tolerance stackup as each fit can be up to ~.0007" out of alignment. The answer is not just to try and tighten up the tolerances because:
a) that will be very expensive if it can be achieved. Each housing needs to be machined on each face so it is done in two operations in our NC. The fact that it is clamped/unclamped/reclamped to do second operation makes me think we can't ask for much more than the locational clearance fit we have now. Measurements of the existing parts seem to confirm this.
b) It will be very difficult to assemble if every fit is .0001" clearance on a 4" dia fit.
c) I'd like to do something that is practical on a scale larger than our test equipment. ie: production if we get there.

So, I am looking for your suggestions on how to best design and machine such an assembly. Press fitting the pieces together is an option we tossed around but I still don't know whether it will get us close enough (doesn't address the actual machining of the parts but will eliminate the error with a clearance fit). Assembling them and then boring the housing concentric would help but it is unlikely to be assembled the exact same way twice (clearance fits and stackup) so this likely will only work some times. Dowel pins are nice but sure a pain based on my experience.

I'd greatly appreciate you sharing your input and expertise.

Thank You

 

[cnuk]

CompositePro: I know that seals really don't like runout and that it's a much bigger problem that STBM (shaft to bore misalignment). The problem with my test fixture is not runout. The shaft bearing and sealing surfaces are ground in one operation. The shaft runs on bearings so runout is not an issue.

Everyone agrees that a shaft that is running out is much harder on a shaft because it exerts a higher peak load and the loading is cyclic. The four seals are really two redundant sealing systems back-to-back in order to simplify testing. Redundant sealing systems are everywhere and that's essentially what I'm testing.

The issue I started asking about is concentricity (yes I really mean concentricity) because as designed, that is the biggest problem I have with the test fixture. I am interested in the previous suggestions like: tolerance rings, dowel pins, etc which help in solving the concentricity problem. I'm interested in talking about seal runout, STBM, etc but that's not what I meant to ask about.

Thanks to all for sharing your collective knowledge.

 

[Tmoose]

I said "the presence of a fat shaft, or a slightly smaller eccentric shaft", meaning a small statically eccentric shaft.

Greg Locock wrote "The latter would put a lot more energy into the seal."

Hi Greg, were you referring to hysteretic losses from stretching the seal material?

I imagine the available energies would be heat from rubbing, and heat from hysteresis from seal lip bending and stretching.

In the limit the surface speed of the ever-so-slightly smaller diameter shaft would have ever-so-slightly lower surface speed, so I'd guess >less< heat from rubbing would be generated there.

In my static eccentricity example, As the lip bending/stretching happened during assembly, there is no real motion during operation, thus, no hysteretic energy loss.

 

[MadMango]

Can you just mount the bearings to the shaft, then shim the bearings to ensure concentricity of the entire thing?

"Art without engineering is dreaming; Engineering without art is calculating."

Have you read faq731-376 to make the best use of Eng-Tips Forums?

 

[GregLocock]

I meant that a dynamically eccentric shaft (a cam) would flex the seal continuously. I see what you mean by a statically off-centre shaft, ie one where the shaft is concentric about its rotation axis, but that rotation axis is not correctly aligned with the seal.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Tmoose]

I think 5 ball bearings in a row on an straight, round shaft would object more to housing centerlines roving concentricity misalignment than the seals would. The shaft may well be your alignment tool especially if the housings can be left loose during assembly, and finally secured with through bolts.

What is the shaft diameter? It would seem it must be mighty small to need support every 3 inches. Then I'd expect some issues of assembling the bearings with (depending on operational loading type and direction) a slight interference fit on the shaft. If the shaft or bearing ID develops a polished or worn appearance it implies the shaft-to-bearing fit is incorrect. The design has to allow assembly AND proper bearing fits.
The housing faces' perpendicularity (re: the bearing bores) must be mighty accurate or their assembled centerlines will look look this ^^^ or maybe this ^-^- or this ~-^

 

[cnuk]

I think my ASCII art is confusing. There are only two bearings in the assembly, not five. The |__| I was drawing represent the contours of the housing and not the bearings. Only where the word BRG is written are there actually bearings. Certainly five bearings on one shaft would cause significant problems if there was eccentricity between the housings and shaft. Because I only support the shaft at two points that's not my problem. I think we all agree on the relative importance of STBM vs. runout. Tmoose is describing the STBM scenario with what he referred to as a statically eccentric case. The shaft diameter is 2". The idea of using the housings for alignment has some potential however it is will complicate getting a proper seal installation.

Thanks again to all