After a recent rebuild by my client, one compressor is showing a large increase in amplitude at 2 x primary stage running speed. No sidebands are apparent and there is no data available that would go out far enough for gearmesh or vane passing frequencies. An axial spectrum on the primary stage is attached.

Actual motor speed is 3570 rpm and the bull gear is 508 teeth with 33 tooth on the primary pinion and 24 tooth on the secondary. The bull gear is mounted on rolling element bearings that the manufacturer will not identify. The other shafts are sleeve bearings. They would not supply the vane counts for the impellers either.

Any ideas?

cooperturbo.doc (33 Kb, 76 downloads)

Danny,

We have 12 units. 2 High Press and 10 LP compressors. The config you have unfortunately is unlike any of ours. If that peak out at 75kcpm is 1 x rpm of the 2nd stage, the only thing you can do is have them verify that it is within spec for balance quality. If not they should come out and field balance it.

Mick,

The amplitude at 2nd stage running speed is about .03 ips and that has held steady through the rebuild.

The one I am more concerned about is the .075 ips at about 110000 cpm which is 2 x running speed for the primary stage. That is brand new and has never appeared before.

Thanks,

Danny

Danny,

It would be really advisable to get a good high freq accel and look at GMF and a couple of it's harmonics. Looks like GMF is over 30 kHz. PCB makes an ICP accel with a resonance freq over 60 kHz. If I remember correctly you use CSI equipment and I cant remember what it's max freq is. I bet you have some pretty high G levels on that compressor. Ten G's (peak) and above are probably a concern.

Skip

I connected to the proximity probes through the test jack on the BN vibration transmitters and captured some data. I think it has a 4-20 ma output, but couldn't confirm it, so I can't really trust the amplitudes of displacement that i measured. I don't have the BN part number because my wife has my car.

Anybody got experience with these?

Dan, we used to build some of the gearing for Cooper when I was with Philadelphia Gear. It was not uncommon to get a 2X of shaft speed from a loose wheel. I would carefully review you mounting procedures, verify if necessary. This is a critical step as these wheels have quite a bit of bore growth at speed due to their shaft speed. Some manufacturers had us (again PGC) actually go to a "polygon" shaped shaft extension to insure an improved fit at speed.

John

John,

What are you referring to as a wheel?

Thanks,

Danny

Danny,

You should get good displacement data off the BNC connector from the Bently probes. I can't remember which sensitivity was used on the Coopers, though.

Skip

I got what looks like the genuine spectral shape the you would expect, by the amplitudes shown are miniscule which makes me wonder about the sensitivity also.

I'll talk to someone at Cameron (Cooper) and maybe they will know or be able to find out.

Thanks for the help.

Bill,

I share your curiousity. I'm sure John will elaborate next opportunity.

Dan, sorry about the terminology. By wheel I meant the compressor impellor.

Bill, the technology of the polygon fit still exists, either as a 3-lobe or 4-lobe fit. The company has changhed hands but still exists in Millville NJ under the new name General Polygon Systems. Website is http://www.generalpolygon.com/history.htm

Dan, by mentioning the "polygon" fit I didn't want to infer that Cooper used that method. My Philadelphia Gear days were 20+ years ago and I don't really remember. But PGC made gearing for Joy, Cooper, Air Products, etc. and on at least some of them the polygon fit method of mounting the compressor impellors was used.



John

Danny,

Most of the prox probes use a 200mv/mil sensitivity. If you connected to the BNC jack (buffered output) you were probably reading the direct displacement value, but that depends on what type transmitter they had in place.
Bentley has different systems of course. What kind was it, an 1800, 3300, or ?

Dave

Before I get too far into the data I collected from the Vibration Transmitters, let me ask a general question.

I collected more data with a different analyser and probes and got about the same results as far as the 2 x primary stage so I'm pretty sure that it isn't sensor resonance. There is a large difference in mass between the two collection devices, so they probably have very different natural frequencies.

This vibration is very predominant (about 8:1 vs h and v) in the axial direction. The proximity probes are mounted radially. Since they measure realtive shaft position, are they going to measure axial vibration at all?

I found out that these are eddy-current probes and I think it is 4-20 mA output. They give the following equation:

Vibration=((mA-4)/16)x 5

so I guess for example a 16 mA output it would be:

16-4=12/16=.75 x 5 = 3.75 mils?

I set up my analyser for displacement probes with 100mV/mil output.

Any suggestions?

Thanks,

Danny

Danny,

Do you know the model of monitor these probes are mounted to?
Do you happen to know the model # of probes (stupid question I guess, but thought I would ask).
Most all of the prox probes are eddy current (unless it's a shaft rider of some sort). Generic sensitivity is 200 mv/mil.
No, a radial mounted prox probe does not measure in the axial direction. They have other style probes that do that on a sloped surface or a shoulder (think thermal growth or relative movement measurement of a turbine). You can also mount one axially to look at end of shaft for thrust.

Dave

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

Dave,

That's what common sense told me, but I have learned that common sense doesn't always reflect the real story.

These prox probes pass through a BN Vibratin Transmitter and it says that it has a 4-20 mA output. It doesn't say where that output will be found though. There is a test jack (where I connected), which is a bnc connector and I now think 4-20 mA. There is also a terminal block where I think you can get 4-20mA or maybe 0-5 vdc. I couldn't wire into those or check them, but I'll probably get some more info pretty soon.

Anyway, It doesn't really matter since what I'm looking for is not going to be seen by these probes anyway.

Maybe they have a bolt-on mounting option of some sorts.

Thanks,

Danny

Bill,

The formula Danny is refering to I believe is just finding the linear relation of some value (his ex. 16) to an out put of 4-20 which 4 would signify 0 Mils Pk-Pk and 20 ma would signify 5 mils Pk-Pk.
That is assuming his example transducer is 0 - 5 mils. Pk-Pk.

Danny,

Some of the transmitters have the capability of outputing both an AC and DC voltage, as well as a 4-20 ma. The 4-20 is most commanly used for driving a chart recorder, vib meter on a panel, or input into DCS, PLC, etc.
It is possible the signal conditioner you're hooked to provides all three. Most generally (again I'm talking generic type conditioners here), the BNC output is a buffered output of the 0-5V picked up across the load resistor (24V supply generally 450 ohm resistor) that the 4-20 is running through. The 4-20 is generally picked up from screw terminals. However, again this is very generic. WIthout knowing the specifics of the model transmiter/conditioner you're hooked to, hard to tell.
And anyhow, as you said, those prox probes are probably not going to show you much at the frequencies you're talking about.

Merry Christmas to you and your family. I appreciated your recent hospitality.
Happy New Year also
Dave

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

Bill,

Thanks for your response.

Dave's explanation of the formula agrees with my understanding and like we have both concluded, the point is moot due to the fact that the probes are not oriented to measure axial movement and that is what I'm concerned about. I would like to get a complete explanation from the compressor manufacturer of BN for future reference and expect I will after the Holidays.

Dave,

I am probably working with nothing out of the ordinary, just don't know for sure. The only thing I'm really sure of is that I didn't use the right settings.

When you said that the prox probes aren't goning to show much at the frequencies I'm talking about did you mean in the direction I'm talking about?

Merry Christmas and Happy New Year right back at you. You are certainly welcome to enjoy our hospitality anytime.

Danny

Bill,

That explains your point better. Thanks.

Danny

Are you working today?

Everyone seems to be missing the point on the transmitters. The 4-20 Ma reading is connected to the DCS for a direct vibration reading. Use 200 mv/mil when taking vibration readings using the adaptor. The adaptor is mostly, I believe, to prevent ground loops. I hate those GD#$%^& transmitters.

Mark,

Hear Hear

Dave

I have gotten nowhere with this with my client, the rebuilders, or the manufacturer other than to ask me what the output of the vibration tramsmitters is. The fact that the proximity probes are not set up to measure axial vibration and this is only axial vibration seems lost on everyone.

I have made a third contact with the supply side folks and asked him to find someone who is familiar with analysis of a vibration spectrum. I would hope that they have at least one person who can do more than ask what the vibration tramsmitters say.

I'm used to varying degrees of cooperation with manufacturers and understand their positions. Big companies sometimes move slowly and I am patient.

What gets me is a client who pays $40,000 for a compressor rebuild and doesn't seem to care when there is something out of the ordinary. The 1 year warranty is an easy thing to beat in industry and after that it will be too bad if something happens.

Any new ideas as to what might be the cause of the axial at 2 x primary?

Thanks again,

Danny