I have attached some info on a three stage Centac at one of my customers sites. The problem I have is a trend of overall on the second stage axial bearing. I have 3 Wilcoxen 10mv/g high freq transducers permanetly mounted on the bearing retainers, and they were mounted in the same location we used to stick this same transducer on when we magnet mounted it. As all of you know, you can't see much with that little thing on that little magnet. We tried beeswax, but what a hassle. Finally just spot faced and tapped.
The problem I have is I can't relate the increasing trend to a specific component.
Have a look and tell me what I am missing.
I have suspicions of air flow, but am not sure enough to have it torn apart for turning vanes.
The client had IR come in and take some readings. They provided a report which I have attached in my document for comparison. I have a problem with the look of their data (appears to be magnet mount with a lot of noise).
What is going on here?
D

Centac_Comparison.pdf (1,343 KB, 90 downloads)

I believe the last 4 colums are pressure drop and possibly operating hours.
The trend is increasing, which frequency/s are increasing. This style of compressor typically has sleeve bearing for the stages.
Can you post any hisotical waterfall plots?

Here is a waterfall plot of the 2nd stage axial.
I did a frequency trend of all three predominate frequency and no pattern (increasing trend) was seen. Mostly up and down.

Waterfall_from_Feb.pdf (26 KB, 40 downloads)

Looking just at the waterfall plot, I'm not really seeing any significant changes, right? What is the Fmax of the time waveform from which the trend value (overall g's ?) is being calculated? Is it the same Fmax that the spectra are calculated from? Keep in mind that at high frequencies it only takes tiny amounts of energy to generate high acceleration levels, which can skew overall g levels especially if there is any kind of broadband activity. I usually trend Maximum Peak Acceleration instead.

What are the radial values doing? What is the trend of the prox probes?

Rusty,
F-max of spectra is 3.6 mil with 12,800 lines. The TWF itself has 1.8 mil with 800 lines.

I am trending peak accel. I changed it (double clicked on peak to change to RMS) to match what the IR folks had done for comparison.

I know it only takes a little bit to change the peak level, that is why I am wondering if degradation of the turning vanes would mess up air going into 2nd stage and make a higher level (broadband type noise).
Quite frankly I'm at a loss as to why it has been climbing so consistently.

D

The listed gear teeth all have a common factor of 3. I would have thought that should not be standard, i.e. having a common factor.

Did the modifications include changing the gear teeth combinations, possibly to change the speed of a component? Having a common multiple means that the teeth on one gear only mesh with 1/3 of those on the other. This can give a wear pattern dependent upon how one assembles the gears to each other. Without match marking, the gears can be mated in a different manner after a maintenance.

I agree Bill. That is why I was not happy with the numbers given to me, but I can't dispute it. I never saw the gears.
The original numbers I had were 380, 39 and 27. But after the repair, they (ones doing the repair) said they were 381, 39, and 27.
I asked them about the 1/3 wear pattern, and was met with silence
I'm not sure now exactly what they changed to get the increased capacity. Putting 1 more tooth on the bull gear certainly wouldn't do it.

D

Interesting Dave! Can you email me the spec lib and template so I can have a look? Thanks

Greg

Dave,

Every Centac that I cover, (only about a dozen and only 2 different sizes) has a 380 tooth bull gear. Further, I don't see how they could add one tooth to the bull gear and make no changes to the pinions. You are adding diameter to the bull gear but not taking it away from the pinions so the center distance HAS to change if you want to maintain proper pitchline, right?.

And unless they changed impellers, you're right, 1 tooth would give you 1/380th more air with the same impeller.

Greg,

I'll see what I can do about e-mailing you the library. It's about 10 meg zipped up. Hope mother Rockwell will allow it. I couldn't get on the FTP incoming site last week. Will try again and let you get them from there if e-mail doesn't work. But since you are a high ranking official of the mother ROckwell software group, perhaps you can help me get back on?

Danny,

I too am familiar with the 380 tooth bull gear, but took what I was told and plugged it in.
I have changed my version of the template back to 380. I believe there was some mis quoting going on. I can't believe someone with the resources of IR would put out a set of gears with a 1/3 combination, especially turning the speed these puppies turn.

Have you ever seen a trend go up like this one without spectra to tell the tale?

By the way, Rusty asked about the prox probe readings. All three were read by IR (I don't have a plug in for the box, yet )and they were of very low amplitude. Spectra did not look bad (big 1X on first stage, harmonics at very low amplitude on 2nd and third stages).
I will attach a pdf of those readings and a regular f-max with 50 mv transducer radial readings.

D

Here are the displacement readings taken from the prox probes, and a horizontal reading taken from the flange of the bearing retainer.

2nd_Stage_Horizontal.pdf (773 KB, 26 downloads)

While looking at the individual peaks for Greg, I happened to think of the band cursor, and did a frequency trend on a band that contained the 1st stage vane pass peak, a mystery peak, and the gear mesh peak. Those three peaks together as a band have the correct slope onthe trend as shown in the attachment. I still need to get enough info I suppose to figure out what the mystery peak is. I guess I'm going to have to get on the phone.

D

The_trend_of_the_frequencies_which_contain_Gear_Mesh.pdf (35 KB, 17 downloads)

We monitor more of these gadgets - IR Centac,Cooper Cameron, Atlas Copco, etc., than just about anybody in the world. Case readings using seismic accels beat nothing....but not by much. The casings are very stiff, the frequencies are high, and trying to determine rotor behavior from casing data is largely science fiction......although it induces a entirely different set of problems, prox probe data is the only indicator that really means anything.......

Just to elaborate......on Joy/Cameron/Cooper (or whoevcer they are this week), recall that in many cases (model and aerodynamics dependent), the fundamental of bullgear mesh in many cases exceeds 30 KHz......seismic coupling, etc., is essentially beyond the range of current accelerometer technology.......

As usual, machine designers are not particularly concerned with our problems.....

quote:

recall that in many cases (model and aerodynamics dependent), the fundamental of bullgear mesh in many cases exceeds 30 KHz

So, if your bull gear runs at 60 Hz this means that it has more than 500 teeth???

quote:

Originally posted by Don Rainey:
Just to elaborate......on Joy/Cameron/Cooper (or whoevcer they are this week), recall that in many cases (model and aerodynamics dependent), the fundamental of bullgear mesh in many cases exceeds 30 KHz......seismic coupling, etc., is essentially beyond the range of current accelerometer technology.......

Don,

I'm not real sure where your at on this. This centac has 380 teeth, with 22.8 kHz gear mesh, which I can see just fine with a mounted transducer.
Now if I can see the peaks, and trend over time, why could I not spot a potential problem.
From other Centacs I have monitored, we have picked out worn gears (bull gear, which I am now beginning to wonder about this one) which had sidebands of running speed, which became multiple sidebands. Again, we had no problem seeing either the gear mesh or sidebands, especially when it unloaded.
So why is this method science fiction?

If the sleeve bearings begin to wear, I can certainly see it in the prox probes, but I doubt very seriously I can see much of anything to do with the compressor wheels or bull gear with them.
Enlighten me

Dave