Misalignment of high speed shaft due to locking nut 3

[Tom Jones121]

I haven't got as far as vibration yet so I'm not sure if this is the right place?
I have a shaft which clamps a series of rotating discs. All are balanced and true until I tighten the clamping nut (M10 x 1 LH) then the shaft shows some runout. We've improved the situation with a sleeve nut but if anybody has any better ideas I'd be interested. I hear some machine spindles have a complex grub screw arrangement but I'm not sure it's possible on a 10mm shaft.

 

[electricpete]

We have to wonder why tightening the clamp nut results in shaft runout... presumably the clamped disk surfaces are not truly perpendicular to the shaft prior to clamping, so clamping creates stresses.

fwiw I would think the solutions involve improving the geometry of those disks.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

also could be the shoulder.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

I should mention: grub screw or set-screw can cause a problem for high-speed machinery since it can tend to push the mounted disk off-center. While this offset can often be balanced out, it is usually preferable to use interference fit to establish centering of the disk on the shaft. Set screws tend to be used for mounting onto small and low speed rotors, but not large high speed machines.

Of course, we know very little about your machine. I’m sure with more details others might chime in more specific and useful comments.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Tmoose]

Like Pete said, I think the faces and geometry of the spacers and rotors need to be gorgeous and well controlled. Not the kind of measurement to be made with dial verniers.

Machine tool spindles typically have tolerances < 0.0001 inch ( " a tenth ") on shoulder runout, spacer face parallelism, etc. That is necessary for the running accuracy and preload control. And locknuts with faces ground presumably perpendicular and true to thread pitch features.

Every now and then things will still "kick" a little when the nut is tightened. Some manuals describe a process of "easing down" the nut face by selective scraping or controlled grinding to keep the assembly true.

If I was really in a bind washers with Spherical mating faces are available, and offer the least influence when tightened.
Extreme example -

http://www.medfordtools.com/metalworking/flexbase.jpg

Some of the spindles we made for GM had 4 screws positioned radially to allow tweaking the centering of the piloted flange mount tooling a little better than 2 tenths runout and a few tenths clearance could.

 

[JJPellin]

Extreme precision in the perpendicularity of the faces of the individual part is the first option. Depending on how many parts you are clamping in place with that nut, the tolerance stack-up can make it very, very difficult to maintain acceptable run-out. This used to be a problem with long multi-stage pumps where all impellers and spacers were clamped in place with a single nut. There were a few solutions developed for the pumps. Some options that come to mind might include:

• Individually locate each part with its own shaft step, split ring or snap ring. Design each part for an interference fit to the shaft. This can be difficult to assemble and disassemble.
• Drive each part with a polygon fit to the shaft to eliminate the need for a drive key. This can be very expensive to machine.
• Clock the parts using a trial and error process of rotating individual parts one at a time and finding the orientation of that part that produces the lowest run-out result. This can take a long time and a lot of work.
• Change from a standard nut to a “Super-Nut” configuration with axial set-screws against a hardened washer to provide the clamping force. This allows for some adjustment capability to minimize the run-out.

None of these are probably a very good choice because of the small size of your application. But without more details, it is hard to be sure.


Johnny Pellin

 

[Tom Jones121]

Thanks for all the comments.
The device is a axial flow compressor with 5 stages. We've got it much better by using a sleeve nut and hardened washers so we can live with it for now but it's not really a production solution. We have everything clocked up to less than 2/10 thou and balanced.
I'd be interested if you can point me to any pump info Johnny as this is most like our application. We haven't tried spherical washers yet but it's a possibility.

 

[Tmoose]

Are all the components in one stack, with one nut?
The nut all by itself can be the problem. The improvement with the sleeve nut suggests it is a big part.

Are the "hardened" washers faces also ground ?

Depending on how snug the spacers fit on the shaft, parallel faces on impellers and spacers may not be enough.

I'd be inclined to do things like probe for gaps between faces with the nut just barely tight, or the shaft stacked vertical with no nut. Also to remove any drive keys, and paint the mating faces with dykem or magic marker, then wring the pieces together by hand to test for even contact.

Also mark axially "high" spots to be sure they are not all in a line when assembled

 

[drawoh]

SplashWomble,

How do you know your discs are balanced before you apply the nut?

I see two possibilities.

[ol]
[li]You have clearance in the mounts for the discs, and they are able to move when you tighten your nut. Again, I wonder how you know everything was balanced. You might look at your mounting clearances and tolerances.[/li]
[li]There is some sort of elastic displacement going on. An M10 nut will exert something over 4000lb force, and you clamp it with something over 20lb.ft or 30N.m torque. If there is something flexible in there, it is going to flex. You need to increase diameters to create a stiffer joint. Commercial, stamped washers probably do not accurately exert the same force all around. Consider accurate, machined washers. Consider specifying a very low clamping torque and lubricating threadlocker.[/li]
[/ol]

JHG

 

[JJPellin]

I am not familiar with spherical washers. I work mainly with multi-stage pumps. The mountings they use eliminate the stack up issue that you are describing. I would suggest researching the following:

Sulzer pumps Model MSD uses individual shaft steps with interference fits.
Conhagen (Houston) made a multistage pump for us that uses polygon fits and split rings to locate and drive each impeller.
Ruhrpumpen Pumps has a shop in Tulsa (the old Byron Jackson shop) that makes vertical multistage pumps using snap rings to locate the individual impellers.

Multistage compressors made by Elliott or Dresser Rand often have individually located impellers with shaft steps. It is common to leave thermal gaps between the stages but no axial contact between stages.

I like to retain impellers from movement in the direction of normal thrust using shaft steps or split rings. Depending on the design, the impellers can be retained from moving opposite of normal thrust using snap rings, thermal gaps or nothing. Any major API pump manufacturer would routinely use these techniques.


Johnny Pellin

 

[Tom Jones121]

We balance each disc on a separate shaft prior to assembly. All individual parts are ground as well including any washers.
We torque the nut to 10ft.lbs. It's a very fine LH thread so it isn't going to come off but it needs to provide enough friction force as the discs are not keyed.
I'll investigate the stepped shaft etc. but we're hoping to be making this in high volume so need a robust, low cost solution.