Interpretation of orbital plots

[NovaStark]

Hi all,

I've been through some training with the standard vibrational plots bode, spectrum, and so on. But most courses I have been on never really went in-depth in how interpretation of orbits or even how time waveform analysis is done.

Does anyone know of any books or documents showing examples of how to interpret these or shows examples of what typical plots look like and their causes ? (similar to how there are spectrum charts to assist in diagnosis).

 

[spciesla]

Here are a couple of books that have what you are looking for:

Machinery Malfunction Diagnosis and Correction: Vibration Analysis and Troubleshooting for Process Industries by Bob Eisenmann

https://www.amazon.com/Machinery-Ma...478280608&sr=1-2&keywords=eisenmann+machinery

Fundamentals of Rotating Machinery Diagnostics by Don Bently

https://www.amazon.com/Fundamentals...sr=1-1-fkmr0&keywords=donald+bently+vibration

Additionally, if you are looking for training, you can take the Machinery Diagnostics course from Bently Nevada (GE).

https://ge-energy.turnstilesystems.com/Programs.aspx/MachineryDiagnostics

 

[NovaStark]

Thanks.

I think I did attend that training some years ago. The instructor didn't go much into detail about the different orbitals though.

 

[spciesla]

This forum thread includes an Orbit Primer attachment.

http://www.maintenance.org/topic/orbit-tutorial#8031086423

I am unaware of any Wall Charts for orbits.

 

[NovaStark]

Thanks that helps a bit.

The problem is that most documents that I can find just say something along the lines of this plot is unbalance or shaft bowing however they don't explain how the orbit forms because of it.


And since I don't really understand how certain orbits form because of certain issues, I find the interpretation to be difficult.


Same with time waveform analysis, most instructors don't go through an example and explain anything. They just say that it is a bit more work to interpret but I don't see how you can reach a conclusion based on an orbit or time waveform with no knowledge of the spectrum.

 

[spciesla]

It is hard and somewhat foolhardy to make a conclusion on a machine condition based upon a single plot format. The orbit is a powerful plot format because you are looking at the combination of data from two probes simultaneously. However, this should not take away from reviewing the other plot formats (trend, spectrum, waterfall, shaft centerline, Polar, Bode, Cascade, etc.). Additionally, you need to look at the data from the other bearings and the operational parameters as well (load, flow, temperatures, etc.).

 

[NovaStark]

It is hard and somewhat foolhardy to make a conclusion on a machine condition based upon a single plot format. The orbit is a powerful plot format because you are looking at the combination of data from two probes simultaneously. However, this should not take away from reviewing the other plot formats (trend, spectrum, waterfall, shaft centerline, Polar, Bode, Cascade, etc.). Additionally, you need to look at the data from the other bearings and the operational parameters as well (load, flow, temperatures, etc.).

I 100% agree with that. I just like to understand how certain plots work and how they are formed based on vibration problems. I believe if I am able to visualize what is happening and why an orbital is showing a figure 8 shape for example, I'd be better off in diagnosing problems.

 

[ronfrend]

Orbits are simply a plot of the X displacement sensor output against the Y output in the time domain so effectively you are looking at the movement of the centre line of the shaft. If there is some unbalance but no other loads you would expect the shaft centreline movement to follow a circle. However if there is some misalignment (often called preload) the uneven radial reaction forces cause a reduction of movement in one plane compared to the other so you end up with an oval shape instead of a circle. As the misalignment gets worse the oval orbit gets "pinched" in the middle and you end up with a figure of 8. The worst case I ever had was caused by thermal expansion of an inlet line to a centrifugal compressor and we saw a double figure of 8.
If you have oil whirl giving a whirl frequency at about 0.45X you will see a smaller "loop" hanging off the main unbalance circle and that loop will rotate.
If you have a rub you will see part of the circle is cut off.
If you can visualise the movement of the shaft centreline you are thinking of an orbit. It is an excellent troubleshooting / analysis tool that can help you confirm a spectrum diagnosis.


Ron Frend

http://www.predicon.net

 

[NovaStark]

That's some helpful info ronfrend, but does it still hold for if you filter an orbit by 1X and 2X ? For example, the 1X will typically always be a circle and the 2X an oval/ellipse. However if the 2X shows a figure 8 yet the vibration isn't high should I be concerned (as this means a total shutdown of my plant).

Or like for in your example you had a case of essentially thermal strain due to pipe strain, how would I know if there is excessive loading on the piping which is then being transferred onto a compressor.

 

[ronfrend]

Good question NovaStark. I found the best way to display the most pertinent information was to use a low pass filter to show nothing higher than (about) 10X. Anything more than that and you get a lot of noise that makes it very difficult to see the true orbit. If you filter at 1X or even at 2X you see the effect of unbalance but you miss out on the higher multiples that indicate misalignment.
In the case of my thermally lengthened inlet pipe the compressor ended up tripping. The cause was that the sun roof had been removed (this was in Oman in the summer at 53C) and the heat from the sun caused the expansion. If you have low values of vibration it makes no sense to analyse the heck out of the orbit (or anything else for that matter). As with anything else in vibration analysis use the amplitude to determine what to analyse and only then should you start looking at detail. Personally I have always used two triggers - 1 an absolute trigger so if the vibration amplitude passed a threshold I analyse, or 2 a rate of change trigger so if something started to change amplitude rapidly I analyse. Don't get hung up on minutae - it's far too easy to get bogged down on inconsequential details.

Ron Frend

http://www.predicon.net

 

[NovaStark]

Thanks for the info Ron.