So this is our new home. I wonder what this button does?

Since Terry created this category, I thought I'd post this again over here to get things going.

There's a cooling tower gearbox that's recently started showing input speed harmonics on the output shaft. It's a double reduction, first reduction is 3.75:1, second 4:1, 15:1 overall. I don't see any gearmesh or anything else--just all output and input. There's also an audible growl now that just doesn't sound right. I'm thinking the VFD could be causing some chatter. Tomorrow I'd like to try changing the carrier frequency since I've seen this work twice before on smaller gearboxes.

Anything else I could try? They're talking about pulling the gearbox and I'd hate for them to jump the gun.

Can't get the photo album thing to work so I'm back to attaching one file.

VFD's can often make them sound funny. Coupling problems can make a strange sound too.
Have you been inside the tower yet? When I have one that's questionable, I shut em down, and go in to check for seal leaks, backlash, couplings, etc.
Can you sample the oil for wear metals?

Hi Patrick,

Are we able to make the images smaller so they will fit on the screen rather than having to scroll up or down and left or right to see all? The old board had this trouble sometimes but the image was just too large when uploaded.

Ralph,

I didn't do anything with the image file. And I agree, it's a pain to move the picture around. I just attached it, thinking it would just be a link to download--at least that's what it did the first time. I'm not sure why it loaded the image into the text. I'm still trying to figure out the photo album thing. It looks like it posts a thumbnail.

Stan,

Haven't been in the tower yet--maybe today. We've requested a sample to be drawn and sent out.

Thanks for the help.

Neither this board or the previous boards will resize the images. A good shareware tool to adjust image sizes is Irfanview at:

http://irfanview.com/

Images should in general probably be limited in size to 800x600 pixel or less. It also will convert image formats from a long list of formats to formats that the board accepts and you can als use it to crop and do other edits before posting. It's also useful for improving the image quality of not-so-good images to enhance viewing by resampling and sharpening after copy the board images to the clipboard.

As far as the data in the graph, it looks more like an imbalance or coupling wobble problem of the input shaft rather than impacting.

quote:

The old board had this trouble sometimes but the image was just too large when uploaded.

quote:

Neither this board or the previous boards will resize the images.

I agree the board doesn't resize, as I stated in my previous post. The size has to be set before uploading.
Just trying to get everyone's attention before too many are uploaded to large.

I agree with Duncan on the gearbox spectrum.

Looks like the input is just shaking the whole gearbox from some 1X input problem (balance, misalignment, etc.)rather than an internal input shaft problem being transfered to the output shaft/bearings/gears.

I guess that's what still confuses me a bit about peakvue. I thought it was only supposed to show the resulting impacts, friction, etc. I guess what you're saying is the multiples in the FFT could be stemming from a balance or misalignment problem.

Didn't get inside the tower today--maybe tomorrow.

If the load zone revolves around the outer race because of unbalance, wobbling shaft/coupling, etc, that produces a 1x signal in the envelope spectrum from friction in the moving load zone. If the envelope detector see a 1x harmonic series in it's input band that significant compared to the friction noise, peakvue might see the difference in those signals as a 1x envelope component because it and other simple demodulation schemes can't distinguish between the broadband signals that bearings produce and ordinary vibration signals. You could take a spectrum that includes at theast the low end of the peakvue passband and see if you see significant 1x harmonic series components. If so, you could move the low frequency corner of the input filter higher, perhaps to something like 5 kHz and see if you see the same looking envelope spectrum. Of course, all the levels should be much lower but if you compared the ratio of the 1x component to the surrounding noise floor in each, this would help make a judgement. This really shouldn't be a problem on a cooling tower.

Lifting up and down on a blade tip should not have a click or clunk, except on a gearbox with mighty loose bearings.

Patrick, what does your non-Peakvue spectrum show? I can't see the value of using Peakvue if your normal spectrum shows a distinct problem. I think the 1x output is likely just fan imbalance which will show harmonics if the 1x amplitude is high enough. The 1x input is probably a driveshaft/coupling problem. Coupling wear/looseness will give imbalance of the driveshaft.

I definitely would not pull the gearbox based on just the peakvue spectrum. What set them on that course of action in the first place?

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

Here's the regular velocity spectrum. It shows the same info as the peakvue but not as crisp. The output is 116 rpm--with speeds that low I tend to look at the peakvues more than the regular.

The growling sound lit the fire under them. All my trends are basically flat except the HFD and peakvue overall--they took off. I told them I wouldn't pull the box without a little more investigation inside it first.

So they pulled the gearbox--three missing teeth on the input shaft pinion. I wasn't there when they pulled it so I don't have any pictures or anything.

Will you post the PeakVue waveform? It may be clearer. Try the circular plot after you locate turning speed, it works well for cracked gear teeth sometimes.

What is your maximum Fmax on the input point?

How many teeth on the input gear?

You say "I don't see any gearmesh or anything else", was your non Peakvue data Fmax long enough to capture 3 or 4 harmonics of the first gearmesh frequency?

Did you take a Peakvue with a HP filter at least passed the 1st gearmesh frequency and an Fmax long enough for 2 or 3 gearmesh frequencies in the Peakvue?

Bill,

The attached pdf has the peakvue TWFs from the input gear in circular and nor-circular formats.

Ralph,

The input pinion has 16 teeth and mates with a 60 tooth gear. Second shaft pinion has 23 teeth and mates with 92 teeth on the bull gear. My Fmax on the input shaft is 80X for the regular point. For the PV point, 2000 Hz HP (maybe too high?), 70X Fmax and 1600 LOR.

I know not everything is textbook, but I really expected something in the data to go with what we heard (the growl). And like I said above, the only trends that really moved were on the output shaft.

Thanks for the help.

Binder1.pdf (499 Kb, 49 downloads) Input Gear

Hi Patrick,

Looks like your Fmax was plenty long. 80X should have shown nearly 5 times gearmesh.

Strange there wasn't any gearmesh showing with even a slight input speed sideband increase, especially around fundamental gearmesh and the 1st harmonic of gearmesh.

quote:

I know not everything is textbook,------

When we reach a point that we think we know everything about this field, we are only fooling ourselves. This field is a never ending learning process, no matter how long we have been doing it. Always something coming up that doesn't act like the textbook says it should.

Thanks, Ralph.

It's during these situations I get introspective and ask 'could I have done anything different?'

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

Were the three missing missing teeth in a row - next to each other - or were they scattered around the pinion?