Recently I was called out to look at a rotary screw air compressor driven by a 600HP motor (2-pole / 480 volt / cast aluminum rotor). The customer stated that he had no concerns for the machine (motor or compressor), but wanted it checked out for peace of mind.

Vibration data was collected with the machine operating at full load (100%). The high-resolution spectrum taken on the motor showed numerous pole-pass sidebands spaced around 1X and harmonics indicating an open rotor (see attached).

Using a strobe light I watched that shaft speed modulate consistently making somewhat more concerned with the rotor. I’ve used this method on motors with rotor issues in the past and it has worked quit well.

I returned later and took some current data with my dataPAC and found a 21.8 dB differential between line frequency and pole-pass (see attached). Normally I would have left with confidence that a pretty severe rotor issue was present, but not in this case.

Unfortunately I have some questions as to whether the compressor could cause the things I’m seeing in the motor.
Or is the current data as clear-cut black and white some have led me to believe?

600HP_VA___MCA_Data.doc (72 Kb, 110 downloads)

Did you confirm the sideband frequency was exactly pole pass frequency? I see your pole pass sideband marked around 54hz. Was running speed 57hz = 3420 rpm?

If so, the frequency matches pole pass and the indicators taken together point strongly toward rotor problem. If not, maybe there is some load oscillations causing these sidebands and shaft oscillation.

Furthermore is there a vfd? If not, then your slip seems very high which is also an indication of rotor problem. Maybe you can check nameplate speed. If the slip is much higher than nameplate slip (speed lower than nameplate speed) without a vfd under normal voltage conditions with current below FLA, that will also tend to support rotor problem.

Now with cast aluminum rotor, that makes it a tricky question. There are no brazed connections from bar section to end ring like in fabricated rotor. It could be a porosity. Whether a cast aluminum rotor is subject to further degradation like a fabricated rotor, I don't know.

Electricpete,

The peak in the current spectrum is mislabeled. I can't get to my notes or data, but the actual speed was roughly 3540 RPM.
There is no VFD.

The load while I was collecting the current data was fluctuating from 82-86%.
The loaded condition while collecting vibration data was fluctuating 99-101%.

I failed to mention in my first post, but the fluctuation in load is what originally made me skeptical.

The customer said that the motor has always sounded bad. I've thought porosity, but that severe?

In addition... This is an Open Drip motor that we could the rotor. We started it in a dark room looking for 'fire flies' around the rotor at start up, but didn't see any. This made me doubt rotor even further, but...

Michel:

A rotor bar problem only shows an individual PPF on either side of line frequency. Other conditions that show multiple sidebands around the line frequency indicate problems on the output of the motor. In this case, I would begin to focus on driven equipment problems.

In addition, with an ~ -20 dB peak, the motor would probably run at a much lower speed. In either case, whatever is causing the problem is very severe.

Howard

I think another key to distinguishing between rotor and process oscillation will be nailing down the sideband frequency. If your are confident the sideband distance in both frequency and current is exactly pole pass frequency, then the likelihood of there being process oscillations at exactly the same frequency would seem very small.

Another important piece of information would be actual slip speed compared to expected slip speed. For example if nameplate speed is 3580, then at 80% load you would expect actual speed in the general neighborhood of 3584. If you are seeing actual speed of 3555 (which is what I computed from eyeballing your current peak at 1.5hz - twice slip speed, divide by 2 to give 0.75hz = 45 cpm slip speed), then that would be higher-slip/lower-speed than expected and would tend to confirm a rotor cause rather than process as the cause. What is the nameplate speed? Did you determine motor speed from vibration data?

Fire flies (sparks during starting) I think would more likely be associated with a fabricated rotor rather than cast rotor. To the best of my knowledge cast aluminum rotors alwasy have closed slots (they have to be closed to keep the melted aluminum in during pouring). For closed slot rotor, there is no mechanisms for externally-visible sparking.

This message has been edited. Last edited by: electricpete,

There is a compressor company coming soon for an annual PM. This customer expects them to look this machine over with 'fine tooth comb".

I’ve read that PdMA looks at the 5th and 7th harmonic of line frequency for pole-pass sidebands for confirmation of a rotor issue. I’ve expanded my current spectrum to look for PP surrounding the 5th and 7th harmonic of Line Frequency (see attached).
Do these sidebands mean anything?
Should I look for them in the future?

Would further testing from PdMA (or other motor tester) be beneficial?

600HP_Air_Compressor_Motor_Data_5th_7th_LF_Harmonics.doc (76 Kb, 49 downloads)

Electricpete,

Nameplate Speed: 3560 RPM

The high-resolution spectra shows the 1X peak anywhere from 3556-3558 RPM with load fluctuating from approximately 82-86%. I checked this speed with a strobe with the shaft speed oscillating I’m estimating that 3557 RPM is the average.
Apparently this motor is operating at a lower speed than expected…

I am fairly confident that what I am seeing is pole pass. What else could be that close anyhow?

Someone else within our company performed the ‘start up - firefly test’. I’ve heard it works, but I’ve never seen it. The cast aluminum vs. fabricated rotor design you mentioned is an excellent point.

I appreciate your input…
Thanks,

"What else could be that close anyhow?" As doc has suggested, there might be oscillation in the torque requirement of the compresssor (for example downstream pressure regulator valve oscillating). If this oscillation happnes to be coincidentally at approximately the right frequency, it would look like pole pass sidebands in both current and vibration and as oscillation of the shaft seen with strobe.

I will summarize what I see that point towards rotor as a cause of the symtpoms and things that point towards process oscillation:

Things that point towards rotor as a cause
1 - close correlation of observed sideband spacing to calculated pole pass frequency. A strong indicator.
2 - Speed lower than expected. This is also a stong indicator. I would feel even better about it if you clarify how you determined the load was 82-86% and whether the three phase currents were verified balanced.

Things that point toward process oscillation as a cause.
1 - multiple sidebands vs only one sideband/

Oscillation of the shaft seen by strobe could point toward either cause.

One thing you might check is discharge pressure and flow. Oscillation of these parameters at this frequency would tend to point in the direction of process oscillation as a cause.

Another thing would be to check variation of the sideband spacing as load varies. Rotor sidebands would move further away from center frequency when load increases.

PDMA R.I.C. test would help confirm/disprove rotor as a cause. Single-phase test is a little tougher but I think also a little more conclusive.

My best guess with the info available now is that rotor porosity is the cause of what you're seeing.

"Speed lower than expected. This is also a stong indicator. I would feel even better about it if you clarify how you determined the load was 82-86% and whether the three phase currents were verified balanced."

The load percentage was taken at the control panel display.
I do not know if the phases are balanced. There was only one CT on the 1st phase. I didn't have a large enough amp meter available to check. Which is why I only took current data on one phase.

I will see what I can learn over the phone regarding the discharge pressure and flow.

Thanks,

Michel,
MotorDoc missed the boat on multiple PP sidebands not indicating a bad rotor. You need to remember that the sidebands in the vibration spectrum (modulation), the sidebands in the current signature (modulation), and the analog meter on an MCC front panel which has needle fluctation (modulation), all are indications of the same thing, namely the inabilty of the rotor to carry the current evenly from pole to pole due to bad bars. I have numerous vibration signatures with four or five PP sidebands on either side of 1x, 2x and 3x. The only thing I ask analysts to be sure of is if the load was over 80 percent. I have seen ALL indications of a bad rotor disappear the second the load dropped below 60 percent.
My opinion- If those sidebands are within a few CPM of slip times the number of poles, then that data definitely indicates a bad rotor.

Michel,
MotorDoc missed the boat on multiple PP sidebands not indicating a bad rotor. You need to remember th

Ron:

The difference is with Electrical Signature Analysis/Current Signature Analysis versus vibration. I was not talking vibration as the original question had to do with current.

You only get 1x PPF sidebands around line frequency for broken rotor bars - that's it, no more, no less. It is not the same as vibration, for rotor bar analysis, and one of the reasons why MCSA is so powerful when detecting rotor bar problems. For instance, if you suspect a rotor bar issue with vibration, you will know definitively with MCSA.

Ummm, yes to the modulation. Not sure where you got the idea that I was saying anything different.

Howard

To all
Could someone explain to me what is the firefly test and how reliable the results are? Not being from an electrical background I am curious to this whole conversation and anything that helps verify elctrical issues to me is very much appreciated by this mechanical type person.

That means looking for sparks in airgap when the motor is started, more visible in the dark. Sparks may be normal especially for new motors. More severe sparks can be caused by rotor bar failure.

Here is an excerpt from EASA Motor Root Cause Analysis document:

“ROTOR SPARKING
There are several potential causes of rotor sparking on fabricated rotors. Some are of a nondestructive nature, and some can lead to rotor failure. (See Figure 5.) Nondestructive sparking can and probably does occur during normal motor operation. Such sparking is seldom observed due to its low intensity and/or the motor enclosure prohibits its observance. Normal operation can be defined as any condition that could subject the motor to voltage dips, load fluctuation, switching disturbances, etc. Sparking usually is not observed while running at full load. The centrifugal force at full-load speed is usually greater than the electromagnetic forces acting on the bar, due to rated load current, and tends to displace and hold the bar radially in the slot. Furthermore, the frequency within the rotor circuit is very low (equal to the slip frequency). This low frequency corresponds to a low impedance of the rotor cage circuit, essentially confining all rotor current to the cage itself. Therefore, while possible, sparking is not normally observed during operation at full load and speed. During across-the-line starting, however, the current in the rotor cage can be 5 to 7 times normal. This high current combined with the higher cage impedance, due to the frequency of the rotor current initially decaying from line frequency at standstill, will cause a voltage drop along the length of the bar in excess of 6 times the normal running value. This voltage tends to send current through the
laminations. In effect, during start-up, there are actually two parallel circuits—one through the rotor bar, and the other through the laminations. The magnetic forces created by the high current flow during start-up cause the rotor bars to vibrate at a decaying frequency, starting at line frequency, which produces a
force at twice line frequency. This tangential vibration within the confines of the rotor slot causes intermittent interruptions of the current flow between the bars and various portions of the laminations with resultant visible arcing. The rotor design and manufacturing processes include measures intended to reduce sparking. However, material and manufacturing tolerances, together with the effects of differential thermal expansion and thermal cycling, preclude any motor from “sparkless” operation. Even identical or duplicate motors can and will exhibit differing levels of spark intensity, since all component parts have tolerances and are thermally cycled during operation. The sparks observed in the air gap are actually very small particles of bar and/or core iron, heated to incandescence by current passing through the iron-bar boundary. Initial punching burrs and/or particles of bar material removed during installation can generally be expected to decrease after several starts. However, particles generated by intermittent sparking due to bar motion will not decrease during the life of the motor. The brief period of intensified sparking that can occur during starting is not detrimental to motor life. Motors with more than 20 years of operation have shown only slight
etching of the rotor bars at areas of contact with the core iron when disassembled.

Destructive sparking can occur under several circumstances, the most common being a broken bar or a defective bar-to-end ring connection. Bars usually break near where the bar connects to the end ring. Breakage is preceded by radial cracks starting either in the top or bottom of the bar. While sparking caused by fatigue failure of the rotor bar is usually greater in intensity than that previously mentioned, it is still difficult to visually detect since the majority of motor enclosures prevent “line of sight” visual observation of the air gap.”

Pete
Thank you for the info Pete is it easier to sse this happen in DC motors where you can remove some of the enclosure to see this happen and does one have to stay at a distance or is this a fairly safe procedure?

The discussion above applies for squirrel cage induction motors.

DC motors may have sparking where the brush rides on the commutator as well but for different reasons.

Pete
So being that these are squirrel cage type the frame on these are fairly open is this correct?
Secondly what is the biggest size motor of this style availible?And thanks again for the info.

Lee,
I do not believe there is a size limitation on Squirrel Cage induction motors. The largest one I worked on was 7000 HP. It drove an FD Fan for a power company. The largest motor I worked on was 23000 HP Synchronous and it was a refiner motor for a pulp mill. I worked in a motor shop for several years and have seen squirrel cage induction motors in all sizes.

I agree. We have squirrel cage inducation motors from 0.5hp up to 9000hp and they go bigger than that.

As a very rough rule, the smaller induction motors tend to be totally enclosed (TEFC or TEAO) and the large motors tend to be open (ODP or WP1 or WP2). In a totally enclosed motor, there would be no chance to see sparks. In an open motor, it is still going to be tough to see anything, but you might have a chance. The chance would be to look at the air inlet path since that always leads to the ends of the rotor. Any filters would have to be removed ahead (with review from site safety program). The path may be torturous so you still might not be able to see it. But if it's dark you might see the flashes even if you don't have a line-of-site view to see the sparks (like you can see the light from headlights of a car coming around a corner in the dark before you can see the headlights themselves). Bottom of vertical motor might be better place than top if sparks go down? Many plants have a boroscope which may give a better view, but again safety review would be required for electrical and rotating hazards. I have worked on motors at our plant almost full time the last 6 years and never yet seen externally visible sparks during starting of any of our installed plant motors.

Pete gave a very good description of the different types of motors and why you cannot see inside of some of them. And that others are not only accessible but allow for a good view of the ends of the rotors.
Something I have seen from experience and I think Ron Brook pointed it out that open rotor bars need to be checked under loaded conditions and he said 60% or greater. I agree with him and would like to share a story of a motor with an open rotor. This was done in a shop so we had more control over the way we were able to test this motor. A single phase test was performed showing no fluctuation in amps. A RIC test with PdMA was performed and showed no signs of open rotor. We then tested the motor under load on our dynomometer and used a CSI 2120 with a clamp on amprobe and plugged the numbers into a DOS prgram written by Art Crawford and his program said there were several broken rotor bars. We disassembled the motor and in several places the end ring was visibly separated from the rotor bars. This proved to us that using current analysis under significant load may be the only way to catch an open rotor. At the time we did not have PdMA's online system (EMax) and we had been using the current test with CSI equipment for years. In the Art Crawford program he instructs you to take readings around the 5th and 7th harmonic of line frequency. I am only speculating but I think this is to help determine the severity and possible number of cracked or broken bars. At the end of the program if a problem is detected, it will give you an estimate of the number of cracked or broken bars.