Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
Dan Timberlake

Sr. Member
Registered:
Posts: 116
Reply with quote  #1 
Direct coupled, center mounted, 800 HP or so.  Variable speed by VFD.

Some measurements were made using a PC based portable analyzer and a accelerometer magnet mounted on the sleeve bearing  housing. Fairly normal looking spectra resulted. Some 1X, a few other frequencies present, maybe related to decent looking substantial designed isolation mounts, maybe related to running cold with dampers mostly closed. 

A few months later hot commissioning was finally performed. Due to a shipping foul up across the big water the plant's control system was used with permanently mounted accelerometers on the bearing housings.  It is an Emerson controls system, modified to varying degrees by the plant.

The DCS spectra look wacky.  Very low amplitude, with An unending string of harmonics, with a curious superimposed ~15 order modulation. Some measurements are labeled to indicate They use peak-vue on some of their measurements.  But all the measurements "look" similar.

All the time traces look to be a string of 0.12 - 0.17 g impacts with an RMS value about 0.15 , and a crest factor of 7 - 9.

Did someone accidently leave the peak-vue filter on when trying to process basic velocity in/sec Pk and even MILS pk-pk spectra?


Attached Images
png Karl 1 vs Karl II.png (293.48 KB, 40 views)

MarkL

Avatar / Picture

Sr. Member
Registered:
Posts: 1,121
Reply with quote  #2 
If they’re sleeve bearings, isn’t peakvue pretty useless? Well if it’s anything like envelope sleeve bearings don’t really work with it. But that spectra wow its a doozy alright.
John from PA

Sr. Member
Registered:
Posts: 963
Reply with quote  #3 
I truly doubt this is any form of a machine issue.  The spectrum on the left (plant's control system) shows a voltage amplitude of 7 mV and and an overall amplitude of 0.007 in/sec Pk.  The spectrum on the right (traditional analyzer) shows some value of 0.022 in/sec Pk.  It isn't clear if that value is a singular frequency or an overall.  But either of these values are exceptionally low.  Three things could be going on:

1.  The machine isn't running

2.  There is an instrumentation problem (lack of transducer power, incorrect wiring, transducer is laying on the floor, etc.)

3.  It is one of the smoothest machines in my 50 years of experience.

There are likely other things possible but I'd start with a careful check of the instrumentation.
Walt Strong

Sr. Member
Registered:
Posts: 889
Reply with quote  #4 
Looks like junk data. I have not seen a DCS do spectrum data! I would not believe the other data from the InnoAnalyzer without knowing sensor type and locations. Looks like trying to read the Comics when the printing press had a bad day!

Walt
OLi

Sr. Member
Registered:
Posts: 1,915
Reply with quote  #5 
Transducer hanging in the air in the cable is a favorite in a iron ore refinery plant...
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Dan Timberlake

Sr. Member
Registered:
Posts: 116
Reply with quote  #6 
The same Fan company company (but different operator) provided a shop based "bump test" report used an Innomic based PC instrument. The accelerometers listed then were  CTC general purpose AC-102s as I recall.
So I believe the spectrum measurement by the fan company's field service guy used the same gear. I have no real concerns about those measurements.

Accelerometers were provided by the fan co to be permanently installed on the fan bearing housing in machined spot-faces, and used by the plant's DCS. 
I >>think<< they are the same CTC part numbers, but can easily confirm what was actually installed.

I don't know whose decision it was to use peak-vue for any monitoring in the DCS, or anything about how the DCS  is configured normally.

The DCS based spectra were provided under the fan rep's guidance for hot commissioning the fan.
-------------.----------

My dilemma is I have never played with peak-vue, so have no clue what to expect, if a peak vue filter was active, and spectra units were modified to MILS, ips, or even gs, what spectra would likely look like, or even how amplitudes would be affected.

As mentioned by others, I would also expect that conventional measurements on a sleeve bearing  housing would be pretty much devoid of any high frequency activity.

---------.--------

I will check if the  permanent accelerometers are general purpose.
If they are, I now feel justified in asking for more info about the configurations used in the DCS before trying to interpret the hot commissioning spectra provided.

thanks,

Dan T


Walt Strong

Sr. Member
Registered:
Posts: 889
Reply with quote  #7 
DCS Plot review: The PK value in CSI jargon refers to the overall level, typically scaled as pseudo peak or 1.4 x RMS level. There is no indication in the spectrum plot that it is a PeakVue spectrum, so I would forget about that. The spectrum looks like the waveform was a pulse, and therefore some form of electrical noise.

I would expect a more defined/distinctive 1xSS peak in the Innomic spectrum, even if it was low amplitude. What is the frequency resolution? I would not get too carried away trying to interpret either plot without a good back story. Can anyone at the plant at least confirm whether the vibration on the bearing housings "feel" smooth or rough?

Walt
David Eason

Member
Registered:
Posts: 38
Reply with quote  #8 
Dan,
Just a couple general comments about PeakVue. It always uses a filter. Look at the numbers on the right hand side of the spectral data, if it's PeakVue, a filter will be listed in with all of those numbers. There's no filter listed in the data you posted. Next, PeakVue is always in acceleration. The inventor, Dr. Robinson, told me personally to never convert to velocity or displacement. These are just a couple of the standard basics to it.

David Eason
RustyCas

Avatar / Picture

Admin
Registered:
Posts: 1,810
Reply with quote  #9 
I agree with Walt the data on the left is probably junk. They likely have a configuration error. A Peakvue spectrum, regardless of the units used, will indicate as much on the upper right of the plot.

The data on the right looks reasonable, but resolution is awfully coarse from the looks of it. If I am going to display spectral data with such a low bandwidth, I bump the LOR up so that it actually looks like a spectrum rather than an envelope. Peak-peak displacement is 0.53 mils which is indeed smooth for a 1200 rpm machine. I have larger 900 rpm fans that I monitor and balance, and that’s about as low as I’ll see except for when I’ve just balanced one when it can run as low as 0.25 mils.

Referencing the DCS data, you will not see harmonics like that on the housing. On Dodge Sleeve-oils I take a reading on the cooling water nipple and I will see harmonics when the liner is loose in the housing. But it’s just regular 2x, 3x, 4x harmonics typical for most any type of rotating looseness. The side bands (if that’s what I’m seeing) are highly unlikely.

However you can get high frequency vibration in a sleeve bearing. Saw this last week, or rather I heard it using my headphones with a 2140. I don’t use a high Fmax on these bearings, so I didn’t “see” what I was hearing, but a higher Fmax clearly showed a problem. Haven’t analyzed the data yet, but from a visual inspection, I’m fairly certain the housing is “cocked” creating a rub at the ends of the liner. I will be adding a higher Fmax collection for each bearing in the future.

A hard rub in the fan, and maybe a rub in a bearing, can create many, many harmonics though. I doubt you’d see such low vibration levels though, or that a bearing would run long producing that kind of vibration.

__________________
"The trend is your friend"
Dan Timberlake

Sr. Member
Registered:
Posts: 116
Reply with quote  #10 
The suite of  measurements (H, V, A ) on the fan and motor include 1 labeled Peakvue, with an RMS g spectrum and wave from attached .  PV is always on a horizontal position
That is why I wondered about it's (PV's) effect on conventional measurements if accidentally left on.

thanks,

Dan T

Attached Images
png curious hot report.png (613.59 KB, 25 views)

RustyCas

Avatar / Picture

Admin
Registered:
Posts: 1,810
Reply with quote  #11 
The horizontal & "Peakvue" spectra and waveforms (see above) are nearly identical, so were taken with the same setup, which I don't think is PeakVue in either instance.  I think they just have the setup wrong. I don't believe any of the data is valid.

Perhaps the newer version of CSI's software is "new and improved" and no longer indicates when a plot is PeakVue or not, but below is what I get when I plot PeakVue data.

PeakVue Example.png


__________________
"The trend is your friend"
David Eason

Member
Registered:
Posts: 38
Reply with quote  #12 
Dan,
Rusty is right, they don't have PeakVue set up correctly. Notice in Rusty's screenshot that the third line on the right hand side says, "(PkVue-HP 1000 Hz)". That's the filter setting. Also, a PeakVue waveform will be rectified.

David Eason
Dan Timberlake

Sr. Member
Registered:
Posts: 116
Reply with quote  #13 
thanks all,

I will decline commenting on the report, other than to request information about the set-ups behind the measurements.

Dan T
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.