Bearing Failure in Oil Flooded Screw Compressor

[bazak]

I work for a company who packages a very specialized type of petrochemical/refining chemical process machinery. Occasionally, we will also package associated compressor packages as well if they add benefit to the proposal. An electrical engineer by school, and instrumentation and controls engineer by career, I am finding myself right in the middle of assisting with troubleshooting a bearing failure on a screw compressor package we installed overseas last year sometime. Not my typical expertise, so I thought I would offer it up to you guys.

The screw is a small oil flooded twin rotor screw compressor. This compressor is installed in tandem with a second screw compressor. The two are operated in a redundant fashion (never run at the same time). Excessive vibration led to the removal from operation and dis assembly by the compressor manufacturer. The compressor in question has approximately 4000 hrs runtime. Our customer luckily caught it before it caused any further damage.

Once we removed the casing, we found the problem almost immediately. The female driven rotor (smaller) negative thrust bearing (smallest bearing assisting with axial load toward the shaft) had significant damage. The bearing is an angular contact ball bearing (single row) from FAG and showed very strange wear that appeared like flattening on the actual balls. The attached picture shows that the ball appears to be "eaten" away. The races for the balls showed a very scratched surface similar to common flaking.

Identifying the root cause is the hope. Oil survey came back good per viscosity and metal entrainment. Process conditions seem to be adequate and there was no significant under or over load conditions present. Alignment was very likely satisfactory as I know their mechanical experience and the fact that no other bearings showed any sign of wear. I read about a phenomena called false brinelling which seems like a consideration given the proximity of the two compressors and the pattern of use. Could this lead to flaking and eventual failure?

Thanks in advance,
Brian

 

[Skogsgurra]

Is this DOL or VFD driven?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[bazak]

This compressor is started via a soft starter. I don't particularly recall the settings for that installation, but I could find more detail.

The compressor is direct coupled to the motor and has an internal slide valve that is used for capacity control. We start the compressor fully unloaded and ramp the slide valve to get to approximately 90% load during steady state.

 

[MikeHalloran]

I associate that sort of ball damage with contamination by relatively coarse particulates, or operation at near zero load, either of which can make balls 'skid', not roll.

Assuming the latter conjectural cause, I'd try starting the compressor with a little load on it, and/or ramping the load up faster.



Mike Halloran
Pembroke Pines, FL, USA

 

[ischgl99]

It looks like the bearing overheated, in the picture I see a straw/blue tint to the balls and the cage looks like it has cooked oil on it. Is the color in your picture accurate? If so, then I suspect the bearing was running hot and reduced the hardness of the balls resulting in the premature fatigue of the balls. If the steel gets above 300-350F, the material starts to loose its hardness and premature wear accelerates. Can you send it out for a hardness test? It should be a minimum of 58 Rockwell C.

 

[bazak]

MikeH-

I will discuss the load considerations during startup with the vendor. The ramp from no load to 70% load (where the slide valve controller is set to automatic) is over a time period of a few minutes. I would be surprised if that short of time period would be the cause, but it's certainly a valid concern. Thanks for the help.

Ischgl-

I have attached a picture of the bearing with a more accurate color. I am pretty confident that the bearing always had ample oil (forced lube 5.5 barg @ 55degC) and don't see much indication of overheating. Take a look and let me know what you think. Again, thank you for your opinion.

Brian

 

[ischgl99]

The color looks much better now, it does not look like it overheated, but the flaking does look like fatigue of the balls. Do you see similar damage on the races or is this only on the balls?

 

[caoimhin1]

Cool. Interesting case. I've seen similar damage to the balls on a pump once before, however there was corresponding damage on the raceways. The cause in that case was the floating bearing (deep groove ball bearing) not doing what it should, it was getting jammed in the housing. This produced enormous axial loads between the two bearings and hence the severe fatigue spalling. Check the bearing at the other end for signs of excessive axial load. Check that both housings have been machined to the correct tolerance and roundness.
You mentioned the oil viscosity came back ok. Generally recommend minimum oil viscosity of 13mm^2/s at operating temperature to ensure minimum oil film. While the oil viscosity may be ok at room temp, check how it is at operating temperature. If not ok, metal to metal contact could have caused the problem.

 

[caoimhin1]

Regarding false brinelling, it is possible but I find that engineers tend to overworry about this. Only consider if the pump would have seen excessive vibration over a reasonable period, i.e.. when idle for a long time next to other rotating machinery, e.g. if its an emergency or back-up compressor.
You would also expect to see indentations or flaking at ball spacing on the inner/outer rings with false brinelling.

 

[JJPellin]

I would interpret this as ball skidding based on several points. First, the localized damage on each ball suggests that they were sliding and not rolling. Second, this would be expected to be most prevalent on the unloaded bearing (inactive). You described this as the reverse thrust bearing. Third, your description of the "scratched" surface on the race matches almost perfectly to the description and drawing in the attached SKF bearing failure document. (Page 22, Figure 42). If this bearing is not adequately loaded to keep the balls rolling, they will skid and fail regardless of lubrication. The bearing could be changed to a 15 degree contact angle to provide more load with less thrust. It could be loaded with a spring as noted in the SKF document. It could be mounted in a pair with a second bearing and ground for slight pre-load.

Johnny Pellin

 

[MikeHalloran]

You have to wonder if a preload spring was once present, and was removed.



Mike Halloran
Pembroke Pines, FL, USA

 

[Fugeguy]

Does look like skidding damage.

Is somewhat hard to tell from the photo.

Attached is some pictures of known skidding damage.

Compare these to the bearing and let us know.

 

[caoimhin1]

The damage to the balls is inconsistent with skidding. The level of spalling is far too severe.

 

[Fugeguy]

Have you looked at the inner race yet?