Back in 2003 we struggled to analyse a double-helical speed increaser gearbox (driven by 3600rpm horizontal sleeve bearing motor and driving a centrifugal pump ~ 7,500rpm) that had shown impacting in time waveform at motor speed of 30g's pk/pk. I had posted a detailed description of the machine and our troubleshooting back on the old board. I searched Oli's board but couldn't find it (does anyone have a link?,,,,I'm not eager to recreate the whole thing).

Anyway, here is a TWF and spectrum that we saw in 2003.

After this, we inspected the sleeve bearings in the gearbox and saw nothing. We inspected the coupling and saw nothing. Alignment was checked sat. We inspected the gear and saw nothing, but we figured there must be some problem with it that could be found by more expert tests at the repair shop. We replaced that gears and the vibration went away immediately.

Now in 2006 those removed gears have finally been inspected at the repair shop. The report from the shop is all gear measurements are normal and there are absolutely no defects or anomalies on the gears.

The one and only defect reported - an oversized keyway on the bull-gear input shaft. Now this operates at the motor speed which matches our TWF impacting at motor speed. But I would not have have thought a loose keyway would cause impacting during steady state operation. What do you guys think?

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

ccp_1a_gearbox3.ppt (560 Kb, 69 downloads)

We saw a very similar problem on a gearbox a couple of years ago. In this case it was a small worm box driving a roll in a paper machine. There was impacting at once per rev on the output shaft. We eventually diagnosed a possible broken tooth on the worm wheel. After disassembly, the keyway on the output shaft where the coupling was mounted was found with significant wear. No other problem or damage was found.

So I guess my simple answer would be "yes".

Attached is the waveform taken while the damaged box was still operating.

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

keyway.doc (32 Kb, 37 downloads)

Pete,

I have not experienced anything like that but if it was because of a loose key, wouldn't this impacting be due to presence of torcional vibration which existed then and possibly exists now?

Taking el. current spectrum may indicate its presence.

If so, it is still hard to explain as to why impacting is at 1x?

David

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

Very interesting Ian.

Good points David. Attached is current signature analysis a year after repair. We have low-magnitude pole pass sidebands and 1x upper sideband above FL.

CCP1A_CSA.ppt (191 Kb, 23 downloads)

Where is the low SB? Could be you have a HP filter in place in the low range cutting out the low 1x SB? 1x modulation of line frequency (60 HZ) could be possibly caused by a faulty bull gear in the gear box.

quote:

If so, it is still hard to explain as to why impacting is at 1x?

A loose key (or worn/oversize keyway) is just another type of looseness, except that the looseness is "angular", instead of radial or axial. 1x impacting is a fairly predictable outcome of looseness, so it seems to me that a 1x impacting in this case is not unreasonable.

Of course, TIOMOAICBTW...

I also have a hard time visualizing why a loose key might cause impacting at 1x.

If everything is in steady state and the key is transmitting torque, then it seems like the key should be pushed to one side of the oversized keyway and be held there by the torque. Then no relative movement of key within the keyway and no impacting. But torque oscillations if present would complicate the picture a little...although I'm still not sure if that would cause the key to lose contact unless the torque transmitted through that key actually oscillated below zero due to the dynamics of the torsional mass spring system. I have to think about that some more. I have no reason to suspect torque oscillations since it's not a reciprocating driver or load. And if torque osicllations existed, why would they be at 1x speed? (although it's interesting about that current sideband... I haven't paid any attention to that before... will have to check Monday to see if the sister pumps have a similar sideband or not).

The bottom line for me is that I really can't understand at this point if and how the loose keyway could cause impacting at 1x but that doesn't mean at all that I'm dismissing it... there are plenty of things I can't understand that still happen.

Still interested to hear more comments and experiences and opinions on whether or not the loose keyway could be the culrpit and how that would generate impacting.

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

quote:

Originally posted by electricpete:
...I searched Oli's board but couldn't find it (does anyone have a link?,,,,I'm not eager to recreate the whole thing).

I was questioning my sanity for awhile wondering whether I really posted it. Turns out that I posted it under the title "articles regarding gears vibration analysis?". No wonder I couldn't find it.

Here is that old post including my previous description of vib on this gearbox:

http://www.vtab.se/PHP-NBoard/html/images/materiali/Forum2/HTML/002791.html

The guy names "GSH" had some comments suggesting he had seen torsional oscillation resulting from phase imbalance causing similar symptoms and that was apparently confirmed with a strobe. That doesn't make much sense to me but I'm keeping an open mind. I'm pretty sure we already checked this with a strobe back in 2003 but I'll go take another look.

I assume everyone can access that link without loggin in... let me know if you can't. Thanks yet again Oli for keeping the old reliability-magazine board treasures alive on your board. Maybe one day Terry will take over and give us a little better search feature for that stuff, including search by post originator...one can only hope.

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

The situation of a loose key causing impacting is often further compounded by a gear that is loose on the shaft.

A properly fitted gear will be a shrink fit on the shaft, in which case the purpose of the key is to provide positioning only.

John

Good comment. I think I also need to go back and double-check which keyway the shop was talking about being loose. I heard the news about the loose keyway seond hand and I was assuming it was the coupling hub keyway on the shaft, but I need to double-check that.

quote:

Originally posted by electricpete:

Still interested to hear more comments on whether or not the loose keyway could be the culrpit and how that would generate impacting.

The mechanism of impacting occurring in the case of loose keyway and oscillating torque could be as follows.

The RPM vs. Torque curve in an AC induction motor is sloping down in the active region. At a constant torque the kew is steadily pushing against the keyway wall and no impacting can occur. Now, imagine that load decreased momentarily. The gearbox, due to accumulated kinematic energy and decreased load will speed up for a short moment while motor's speed stays the same at the first moment, thus, releasing force transmitted by the key and dis-attaching from the keyway wall. This will result in moving of the key in the direction opposite to rotation within the loose keyway. Then the motor RPM quickly cathes up causing the key to slam into keyway wall.

This is possibly how every time there is a torque change, the key will execute a "back-and-forth" motion cycle within a loose keyway.

In cases when torque oscillates at 1x the impact will also occur at 1x.

David

Yes, I agree. That's what I meant about the torque transmitted through the key momentarily going through zero due to dynamics of the torsional mass-spring system. Is it likely? Seems like it requires very high oscillating torque or a torsionally resosnant response of the system. If this were the culprit I would expect it to be visible with a strobe, do you guys agree?

The comment which suggested unbalanced phase voltage resulting in 1x oscillating torque I don't believe. I can't think of any physical mechanism for that. I could believe 2*LF torque oscillation (which is known to exist in single-phase motors by the way) or just maybe pole pass oscillation, but not 1x oscillation from this cause.

Looking back at the old thread I see another factor that I had forgotten about. The vibration increases over time after startup (as trended by the plant computer). We did a vibration check using local handheld accelerometer which confirmed what the plant computer was telling us. (I haven't gone back to look to see if the vib on the new gears is changing over time). Does this give you guys any ideas?

I had a case recently resolved including a GBX on in a extruder drive with a DC motor. I got hi bearing values on the mid shaft of the GBX after some 20 years of continous use. It was no real bearing freq.´s to be found but it excited all resonances that could be found in the gbx. So it
was opened and the key holding a gear was very near to fail totally, at that stage the gears looked normal for the time they had been operating. Now the internals is new and it is ok again. Not the same exactly but some impacting was happening due to the failing key. Olov

The coupling hub may have developed some looseness on the shaft if let to move around by the oversize keyway, once its loose through torque misalignment develops and the variation in load created from this condition could be impacting at 1x shaft, was there any evidence of fretting in the hub bore or shaft area?

Captnb

I am glad a motor current spectrum 0f 1/10/2004 was requested. It seems to only show a motor current amplitude ratio of just over 100:1 (~40 dB) between line freq. and the first pole pass SB to the left of line freq. This suggests the possibility of problems with the Rotor (cracked/broken rotor bars, rotor laminations, and/or shorting ring problems). Problems with the rotor can cause torque oscillations that can excite Torsional natural frequencies. Such Torsional nat’l. frequencies can be detected in motor current spectra (they show up as Sidebands around Line Freq.). Looking at your motor current spectrum, please note the freq. at 7159 CPM and compare this to 1 X Line (3598 CPM). This is a difference of 3561 CPM (which is 1 X RPM of your input gear. There is a possible Torsional resonance very near 1 X input gear speed. Normally, to excite Torsional resonance, a speed fluctuation is required. This is indicated in your 11/7/2003 waveform showing the pronounced pulses at once/rev intervals (plus higher frequencies likely pertaining to your gear mesh).

Taking all this into account, it appears slight speed variations may have been exciting Torsional resonance that would pound away on your gears (but the gear component that would take the greatest stress would be your key tying the gear to the shaft. We have often seen waveforms with this pronounced once/rev pulsing that initially indicated cracked/broken gear teeth (however, upon inspection, we have seen problems both with the gear key and with the keyway [as well as the coupling key/keyway]). In any case, waveform g levels exceeding 20 g’s true peak are not healthy and mandate inspection. On occasion, if the keyway itself has been the problem, we have not only seen once/rev pulses in the waveform, we have likewise seen hunting tooth sidebands around 1 X gear RPM as well as hunting tooth sidebands around gear mesh! Thus, the message is that hunting tooth sidebands can suggest problems with either the gear teeth or the keys/keyways. Replacement of the gear set should resolve such problems. We are documenting these types of problems in a “Time Waveform Analysis” paper that we are writing.

Jim Berry

Great comments all the way around. That gives me some things to think about. Jim - I think that my poor quality plot has led you astray - it is actually just over 1000/1 difference (~60dB). That sounds like it will be an interesting paper.