More details are needed. I assume it is a deep groove radial ball bearing. I assume it is grease lubricated. I assume it has appropriate axial clearance so it cannot take a thrust load. Please verify these assumptions. Is it a hollow shaft or solid shaft design? What type of coupling is used to couple to the pump. I assume it is a solid shaft and it probably uses an axially stacked rigid coupling with a split ring and key on the motor shaft and a threaded nut on the pump shaft to adjust lift. The pump probably has no bearing housing. Have you verified that alignment between the motor shaft and the seal chamber radially and axially?

If my assumptions are correct, I would expect one of the following causes:

The motor shop made a mistake and installed the spacer ring on the lower bearing that would be used if that was the fixed bearing. On subsequent failures, they keep putting it back the way the found it rather than going back to the original design. When the shaft heats up, it puts a thrust load on the lower radial bearing and it fails.

The motor shop is over-greasing that bearing. Some mechanics think you need to pump that bearing housing completely full of grease until it is purging out on the far side. That is too much grease.

There is a severe misalignment between the motor and the first bushing under the mechanical seal. This could be a rabbet fit or it could be made to be aligned and doweled. How did you verify the alignment? Did you check for concentricity and angularity?

The coupling has some defect causing it to misalign the two shafts, putting extreme loads on the lower motor bearing.

Johnny Pellin

Yes, this is a solid shaft motor with a grease lubricated deep groove radial ball bearing. I don't believe the bearing is fixed vertically. There is nothing constraining it from moving up or down as the motor heats up. On the ODE, there are two thrust bearing for downward thrust and one for upward thrust.

Yes, this is a standard rigid spacer coupling and there is no bearing box on the pump. This is a spigot fit pump and motor

We initially found misalignment between the motor and pump by sweeping the stuffing box bore from the motor shaft.

We sent everything out to be re-machined. When it got back, we re-swept the motor shaft to the stuffing box bore and it was within 0.0015".

While it was out, we also had the coupling checked along with the rest of the pump.

I thought this alignment correction would fix the problem but the lower bearing failed again after about 4 hours of run time. There isn't a whole lot left of the bearing to analyze. The motor repair shop is suggesting a shaft grounding ring to eliminate the possibility of arcing. They made this recommendation based on the fact that it's a 2300V motor.

Also, if there was still severe misalignment wouldn't we have seen that in the vibration spectrum at startup?
Attached is the spectrum we pulled. There are peaks at 1x and 2x but they are pretty low amplitude.

Thanks for the response!

• https://files.engineering.com/getfile.aspx?folder=158bacdb-e3e7-4037-840d-e

I would not expect a clear indication of misalignment in the vibration spectra. This is a rigid coupling and the pump uses product lubricated bushings. It will not show up the way it might with a flexible coupling. Next time sweep the seal chamber face as well as the bore to check for angularity.

I don't understand the need for that triple thrust bearing with only 9 stages and a light service (LPG). But, that is not where the failures are occurring.

So, we are left with an assembly error limiting the axial float of that radial bearing or over-greasing. I have had motor shops machine housings incorrectly or install spacer rings in the wrong locations causing thrust to be imposed on the radial bearing. Review the drawings for that bearing housing with the motor shop and have them verify that they are building to the correct design.

Johnny Pellin

Thanks for the reply, do you agree with their recommendation for a shaft grounding system?

I know they're common with VFDs but I haven't seen them anywhere else. Their recommendation is based on the fact that this is a 2300v motor that has now been repaired a few times.

I don't know about shaft currents in VFD motors. Why would the arcing only occur at the lower bearing? I would not expect arcing to cause a failure this quickly. I would expect a gradual increase in vibration as frosting or fluting occurred. I would not expect complete destruction in 4 hours. You can check for shaft voltage with a probe while running. I found this recommendation in an on-line article: "Shaft voltage can be measured using a digital oscilloscope—with a bandwidth of 100 megahertz (MHz) or higher—by touching a conductive microfiber probe tip to the spinning shaft of a motor."

Johnny Pellin

Got a model number and construction/parts diagram showing the internal arrangement of both motor bearings ?

Like others have said, I'm suspicious that the poor DE (ball) bearing had become the thrust bearing due to assembly error, or (long shot) installation error when setting the pump impellar height.

Can the shop test run the next mounted vertical for 5 hours?

I'd dykem or magic marker the shaft and scribe some lines on the DE shaft showing the axial shaft location when fully seated on the thrust bearing, and also when the shaft is raised off the thrust bearing .03" and .06".

I'd like to see nice detailed pictures of the failed bearings.
Things like:
- Condition of grease ( from like new, to charcoal ).
- the cage/retainer, even if in pieces, showing the pocket wear and any fracture surfaces.
- the race and outer surfaces of the inner and outer rings.

I'd continuously monitor housing temperature as close to the problematic bearing as I can get right from start up.

The provided velocity spectrum with an fmaz of 24,000 cpm is not looking in the right place to detect lubrication and bearing loading issues.
Does your analyzer's software offer some condition monitoring capabilities ? I'm thinking "parameter sets" described as "high frequency" blah-blah-blah.
At the next start up I'd measure vibration as close to the problematic bearing as I can get, using Spike Energy or some other demodulated technology.
Even an "overall" reading in acceleration with an fmax of 300,000 cpm or so will likely climb FAST when the bearing's life is becoming unbearable.

Is the vibration at the NDE/top bearing similar?

Here is a simple test that may be of use:
After motor-pump assembly rotate/bar shaft by hand with strap wrench, as necessary
Run pump for about 0.5 hour and shutdown
Rotate shaft by hand and compare compare required force or rotation resistance

This might indicate there is differential thermal expansion of motor rotor and case that applies excessive thrust load on DE bearing.

Consider installing a temporary vibration monitor capable of high frequency and demodulated measurements. 4-hour failure is extremely short!

Walt

You have got a lot of good advice.

Quote:

Got a model number and construction/parts diagram showing the internal arrangement of both motor bearings ?

Like others have said, I'm suspicious that the poor DE (ball) bearing had become the thrust bearing due to assembly error

.
That would be a big suspicion of mine based on similar experience.

I have two different designs of vertical motors where the upper bearing takes downthrust and lower bearing takes any momentary upthrust (by design) where we have seen rapid lower bearing failure (much more rapid progression of bearing defect after detection by vibration than we're used to, although it was not soon after installation) and subsequent inspection of those removed bearings showed upthrust. Subsequent investigation revealed the stator frame expanded thermally more than rotor and so the endplay went to zero (and below) when the motor was heated resulting in axial upthrust on lower bearing. This won't happen during an unloaded shop run because the stator doesn't get as hot. The problem was resolved by increasing the endplay, and in the case of one of the motor designs where we have continuous temperature monitoring, when we adjusted the remaining motors as corrective action, we could see the lower bearing housing temperature immediately decreased after we increased the endplay. Also the very-low-level bearing defect frequencies associated with the lower bearing shifted ever so slightly indicating lowering contact angle.

Yours seems different in that the upper bearing configuration is supposed to take both downthust and upthrust. Somehow I think that's not working the way it's supposed to. Do measurements to figure out how far up the assembled lower bearing can move in the housing before it contacts the top of the housing and can't move any further. That distance needs to be a comfortable margin above the endplay established by the upper bearings. Also verify some room for movement in the downward direction in the housing (in case rotor expands more than stator, different than our motors). Also verify the lower bearing fit (should not be too tight in the housing) and the condition of the lower bearing housing (damaged housing might create friction preenting lower bearing from sliding in the housing to accomodate thermal changes).

=====================================
(2B)+(2B)' ?

Thanks Pete, that's really helpful info. We are running an unloaded test tomorrow. The motor shop wants to rule out any bizarre shaft current issues since it's a 2300v motor.

I'll have them re-evaluate the bearing stack up and clearances too.

Thanks

While running the motor unloaded, we didn't see anything unusual in regards to shaft current. The lower bearing is running about 117F unloaded (that doesn't really tell us much)

I talked with the motor shop and before the last failure, they gave the lower bearing over 0.200" to grow up and down. They wanted to be absolutely sure the bearing didn't grow into a stationary part. The upper 3 angular contact ball bearings are fixed in place so there is a few thousands at most worth of endplay.

Anyone have a suggestion on what to try next? I wish we could load the motor in the shop but they don't have the capability.

We could quadruple check the alignment in the field but that would involve shipping the motor back and forth again.



Thanks

How is the shop greasing that lower bearing? If they are pumping the housing completely full of grease until it vents on the opposite side, the bearing will tend to overheat. How long was the motor solo run? Are they adding the grounding brush during this repair?

Johnny Pellin

They're definitely not over greasing it. They said they fill the bearing up to about 1/2 capacity.

The motor is still running unloaded and I'm going to keep running it all day to see if something goes squirrelly.

We didn't add the grounding brushes because we wanted to measure the current first. I'm not sure if this shaft current is load dependent but we didn't see anything unusual when it was measured unloaded.

That's a head scratcher.

If you're still at the shop, ask them if they have the lower bearing shaft fit and lower bearing housing fit measurements recorded? Also is the bearing a C3 internal clearance? These are just good things to double check for a variety of reasons. If you have an incorrect shaft diameter (too large, reduces bearing internal clearance) or housing ID (too small, restricts bearing from moving to accomodate thermal changes), that condition may have persisted through all three bearings.




=====================================
(2B)+(2B)' ?

I finally got some info out of GE regarding this motor. They said this particular lower bearing design has been a major problem. For some reason the bearing doesn't slide is the bearing cavity as designed. When the shaft heats up and tries to push the bearing down, it gets stuck and overloads.

GE's recommended fix is to replace the lower ball bearing with a roller bearing. Not sure why it took this long for this information to come to light but I'm just glad it's a known issue.

We have a newer model of a similar style motor and it has a roller bearing instead of a ball bearing on the DE. GE must have know the error in this design and fixed it in their newer model.

Ball and roller bearings require a minimum radial load to prevent ball/roller skidding.
Ball bearings can be axially loaded with a light preload spring to stabilze them.

Motor with horizontal shaft that are direct couple to whatever it it s they are driving apply modest radial loads to their bearings. These motors usually have ball bearings at both ends.
Unlike belt drove motors whose belt loads on the DE can be severe, so often have roller bearing at the DE.

There are tales of motors with roller bearings used direct coupled, and suffer from roller bearing ailments from noise to short lives due to cascading race damage from skidding.

Your DE bearing has really low radial loads, so I wonder if the roller bearing motors don't have issues of their own.

I think a better "solution" is to put the ball bearing in a long sleeve with OD clearance to take the Length/Diameter ratio from jam prone to jam resistant.
Efforts at polishing the bearing OD and radiused corners, the housing bore, fancy lubes, and a little more clearance might be helpful, but I'm guessing GE may have already pursued that.

Thanks for the feedback ME1289.

I agree with Tmoose. Cylindrical bearing without any radial load is susceptible to skidding.

The vast majority of vertical motors have downthrust bearing on top and the lower bearing is a deep groove that is supposed to slide in the housing. (two variants: upthrust limit may be provided either at upper lower bearing, but either way lower bearing has to be able to slide in the housing).
The standard bearing fit measurements are supposed to allow sliding.
What special about this motor design that a standard deep groove bearing (*) isn't sliding in the lower housing like all the others (**) ?

Call me skeptical. This sounds like a story that would come from a repair shop, not an OEM (repair shops say it's the design not the repair workmanship, OEMs say it's the repair workmanship not the design). So I'll admit it has more credibility coming from the OEM but I'm still skeptical. Did this OEM have a nice long talk with your repair shop before he talked to you? Does he have a business relationship with the repair shop? Does the OEM stand to make money off of a redesign?

Going back to lower bearing fit measurements.... were these ever verified? Three measurements 60 degrees apart. Is the housing bore smooth?

(*) I assume the lower bearing is a standard deep groove 62xx or 63xx?
(**) I guess I can see that if the housing were made with an oddball material then it might have a low temperature coefficient of expansion or be susceptible to some kind of galling in combination with bearing material but I've never heard anything like that on an electric motor and it seems like a real longshot to me.

=====================================
(2B)+(2B)' ?

"I finally got some info out of GE regarding this motor. They said this particular lower bearing design has been a major problem. For some reason the bearing doesn't slide is the bearing cavity as designed. When the shaft heats up and tries to push the bearing down, it gets stuck and overloads."

The hand rotation test that I suggested would have detected/verified this overload condition. Did you ever try it?

Walt

By now,there should be witness marks on the DE bearing bore to tell you where the bearing was actually riding.

Mike Halloran
Corinth, NY, USA