Orbit Tutorial

Thanks Steve for some great examples.

Walt

Thanks for sharing Steve. Real world examples are very helpful, especially on unfiltered orbits. TC

Good reference. Im looking forward for sharing on vibration issue at generator.

Thank you Steve.

Thanks, Steve.

I don't know squat about reading orbits and it is nice to get information like this from someone whose abilities I have confidence in.

Steve, thanks for sharing.

I wonder if casing orbit is of any value.
For example, seeing a loop in a journal bearing orbit tells volumes. This is something we can NOT get from another vibration pattern representation. But what about casing orbit?

Can't one do an analysis based just on V, H, and phase readings? Not that I am against orbits in this case, since once you have V, H, and phase, an orbit is just another form, but the question is: 'Does an orbit in this case provide additional info?' Also, most of the time having phase does not add any value.

David

Dave,

I guess we can restart the whole prox probes on fans discussions, but I like orbits. They help me visualize what is going on. If I take bearing cap readings I typically do so with XY accels, and key phasor. Then if I am looking for rolling element bearing problems I look at the spectrum, but if I am look for running speed related stuff like resonance or stiffness issues or misalignment or balance, I look at the orbit.

To me a bearing cap orbit is a 2d ODS.

I am not say that is right or wrong, it is what works for me. I like orbits and so I use them...To each his own...

Steve,

Would please give your reasoning on some Rub section examples.

Was rub in cases ##3,4,6,7 diagnosed solely based on amplitude and available clearance?

Thanks,
Dave

No, in the case of a 1X rub, you really need more info. Startup shutdown data showing residual rotor bow is a great indicator. Sudden change in amplitude and phase that seems to indicate imbalance on a machine that should not go out of balance (like a refrig compressor) is another good indicator.

Here is a paper I did on the subject...

The rub that Steve shows in figure 2 is similar to what Newkirk (GE) investigated (1920's). Today, the Morton effect is popular and has similar characteristics. These rubs can spiral out to very high levels if one is not attentive. These have the effect of locally heating the rotor and causing it to bow, effectively changing the imbalance.

Some excellent stuff, Steve. Thanks for sharing. I think a lot of people miss the usefulness of looking at phase when diagnosing rubs.

Wild Bill,
Good point. One thing that we might need to be clarify is that the Newkirk effect involves local heating of the rotor due to a rub, while the local heating involved with the Morton effect is due to viscous shearing within the bearing oil film.

Just my two cents on a Saturday morning.
Cheers,
Hunter
www.BRGmachinery.com

Steve - thanks for the good stuff. I still your orbit tutorial posted in 2004, but this one has some new stuff added.

Hunter - welcome to the forum. I have read some of your articles on the TAMU Turbo conference website and I enjoyed them.

Thanks for the welcome and kind words, Pete. Those articles are great when you're having trouble going to sleep.

While I've been a board reader for many years, I've decided it's about time that I start trying to contribute something.
All the best,
Hunter
BRG Machinery Consulting

Dear Steve,
Actually I am not familiur with multi keyphesor.
I have seen orbits for one keyphesor. so thanks for ur tutorial. I am experienced with Balancing a 1000 KW Fan with overhung rotor. here i attache some orbits of before and after mass balance. This balancing has been done using Bently Balance and ADRE.

Filtered orbits (MD) like you used for balancing can not have multiple trigger references on the orbit. Much/most of the value of orbits derives from un-filtered orbits.

quote:

Originally posted by Steve Schultheis:
No, in the case of a 1X rub, you really need more info.
Here is a paper I did on the subject...

Steve,

In the referenced paper you mainly showed cases where unfiltered orbit did not carry any useful information (other then large amplitude) of an existing rub (Fig. 6,9,12). Bode plot actually had the clue.

If this is true, how do you explain the fact that a circular orbit has not been practically affected by a radial rub? Is it a reflection of rub severity?

Could small deviations from a circle (flats) in the orbit with no loops always be interpreted as a rub? I do realize that a coastdown test is of much value here but sometimes it is not feasable. Will large amplitude only be sufficient to request a coastdown test?

Dave

Again it is a matter of understanding the machine. If it was a turbine, I would have to suspect a chunk of blade or shroud band had come off. But I also might suspect that a rub was occuring.

The fact is though (as the paper points out) that most rubs happen on startup. It is unusual, although not unheard of, for a rub just to happen after some time on line.

A large 1X vibration and a big elliptical orbit could be due to a dozen or more different things. You can only eliminate possibilities by at least varying speed and load, and it may require a shutdown.

One that I am working right now, the vibration suddenly jumps up on one end, 1X elliptical orbit, hits trip level, the machine shuts down, and then on restart is fine.

That is not unbalance, not instruments, not misalignment. Hmmm, I bet that is a rub...

Dear Steve,

I also saw a 10 MW turbine which rotor was repalced by a new one and after 15 days the turbine was gone for startup and it was tripped due to high vibration of generator shaft.But for 2nd attempt everything is fine.

Steve, could u pls tell me during rub is there always reverse precession?

No, by no means!

In general reverse precession is due to probes being backwards from what you thought.

Reverse precession over a short portion of the rotation (a loop on the orbit) may well be due to rub.

Full reverse precession due to rub is very rare, as it doesnt last long since it wrecks the machine.

Thank you for sharing this information Steve I have read it with interest,
But how strange is this, yesterday I was in a bar and got chatting to a local who asked me what I was doing in his neck of the woods. I replied that I had been carrying out some vibration tests at a local factory.
This fellow said he too carried out vibration testing, opened his laptop and said what do you think of this bode plot?
I looked and said I haven't got a clue He emailed it to me smiled and said I will be here next month try too work it out
Has anybody seen this before or is it a wind up?
Mike

These are not really bode plots since there is no phase information.

Why do you think it's a wind up? There's no reason to believe that the data is not valid, but its difficult to do anything but speculate on what they mean. Whilst the plots can be described, understanding why they are this way will require more data and information.

Both vertical probes show a critical speed at around 2500rpm and the horizontal probes show peaks at around 2000 and 2200rpm. Indicated vibration shows variations of up to 100microns with small changes in speed at the NDE

To get any further needs the following information:

What does the accompanying phase angle data look like?
Do these plots represent startup or shutdown data?
What differences exist between startup and shutdown data?
Are vibration characteristics repeatable from startup to startup and shutdown to shutdown?

What do the equivalent polar plots look like?
What is the real critical speed(s) of this machine?
What do the orbits look like between 3000 and 3600rpm?
What is the rate of change in vibration steps above 3200rpm? (just because they appear to occur very rapidly, each change might have taken place over quite a long period of time if the machine speed was increasing only very slowly.
Are there any other spectral components present?

Any vibration that is not steady can result from a rub or a loose rotating part. The data could also be indicating an instrument fault (loose probe?).

It is difficult to say more. As with all problems, it is virtually impossible to make a sensible diagnosis based on one data format.

Hi MCDM

The plots are labeled "uncompensated fundamental shaft absolute vibration."
I take this to mean no compensation for shaft runout or magnetic variations, 1X, shaft vibration as measured with prox probes. I'm confused at the "absolute" label, since I consider prox probes as measuring shaft vibration a "relative" measurement, that is relative to the housing/prox-mount.

Wouldn't "absolute" require an inertia based reference, like an accelerometer?

Thanks Buzz and Dan for the replies
I am out of my depth with this
I will send an email asking for more details and post

On large turbomachinery, in MANY cases where there is a rub that causes high vibration, the only apparent symptom is a high vibration amplitude. Depending on the relationship to the critical speed, the amplitude can walk straight out at a constant phase or the phase can roll. The orbits at both bearings for the rotor can be almost completely round. This is generally an indication that the shaft is bowing due to a rub that is not immediately adjacent to the bearing. The coast down will typically show a high (sometimes very high) response at the critical speed and possibly a high eccentricity reading on the turbine after shutdown.

Along with the others, I would really like to see the phase info, but I bet the phase is rolling round and round 360 degrees. I have seen bode plots like this. A rub can cause a 1X vector to roll as the bow due to the rub interacts with the residual unbalance.

If you look back in the paper I posted on rubs, you will see a polar plot where the trace spirals around. If you plotted it as a Bode, it would look much like the ones you posted...

Thank you Steve,

for beginner like me, it is very good

Best Regards,

Jono

hi steve..
iam new to vibration field and just entre into industry from collage... i am in oil nd gas field and dont know about orbit,, there is a lot of confudion in orbit, bode and polar plots analysis. pls send me some stuff so i will also be able to understand this orbit analysis, i will be very thankful to you. my email id is akhan.koc@gmail.com and xxman092@gmail.com

quote:

Originally posted by eddy092:
hi steve..
iam new to vibration field and just entre into industry from collage... i am in oil nd gas field and dont know about orbit,, there is a lot of confudion in orbit, bode and polar plots analysis. pls send me some stuff so i will also be able to understand this orbit analysis, i will be very thankful to you. my email id is akhan.koc@gmail.com and xxman092@gmail.com

Not sure if Steve even frequents this board. Last posts that I can find posted by @@ are from 2011.

I suggest you use the "Find" feature. There is enough material on the various plot formats to write a book!