Rotating machinery imbalance

[JMarkWolf]

Forgive my multiple posts but I have been directed to a couple different forums to better answer my question.

I have an embedded DFT that identifies the velocity magnitude and phase of the heavy spot on rotors, from time-series accelerometer and key-phasor data.

I wish to validate and eventually certify these results.

Can someone on the forum refer me to an individual or lab that offers such services?

 

[rob768]

Being Dutch-based, I guess we are located to far away, but we could do that. (

http://www.technofysica.nl)

 

[Tmoose]

I'm probably misinterpreting something, but the subject says unbalance, and the sensor is described as an accelerometer, suggesting bearing casing vibration measurement.

Are you looking for certification of the sensors and instrumentation? Or for the amount of residual unbalance of the rotor?

Vibration levels can be, and often are related to unbalance. But are not automatically equivalent to unbalance. One way to Establish the relationship involves a survey made by placing trial masses on the rotor at multiple locations and recording the resulting 1X vibration and phase

 

[JMarkWolf]

Rotating machinery, as in airplane propellers and helicopter rotors, determining magnitude of imbalance (velocity) and location of heavy spot (phase).

I have in-flight time-series accelerometer and key-phasor data collected from aircraft.

My embedded DFT is giving me a report as to above parameters. My immediate need is to validate the report provided by my embedded DFT algorithm (determine degree of accuracy).

 

[electricpete]

determining magnitude of imbalance (velocity)

I'm with Tmoose - your terms are not clear. Velocity is not a unit of imbalance. Inch-ounce might be a unit of inch-ounce or gram-cm.

If you are balancing the rotor by itself, you can draw a correlation using influence coefficient. You can also move a known unbalance (inch-ounce) around the rotor and plot the magnitude response... it will have a steady component and a component which varies sinusoidally with position. The sinusoidally varying position is due to your known added unbalance and can be used as conversion to convert the dc component of velocity (unknown unbalance) into a known unbalance.

As far as validating FFT... still don't know what you're after.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[GregLocock]

Your algorithm should be self certifying, since if you use it to balance a machine then the balancing process will either work or it won't.

Or you could add a known imbalance to a rotor for which you have already have established the influence coefficients.

Sorry this doesn't actually answer your question

Incidentally in one of your previous posts you mention that you want to find a lab near Detroit to check this for you. I'd look in the yellow pages for an instrument calibration service.





Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Strong]

Mark,

Here is a good calibration lab that may help:

http://wccl.com/contacts.htm

I assume you want to verify that instrument is workly correctly first and then get a certified electronic calibration (see lab link). One practical way to test end-to-end functions is to compare to another known instrument with measurments on a rotor kit or low cost balance simulator. You could then verify amplitude, speed/frequency, and phase by testing at different speeds and with different weight amounts and at different weight locations. I use a Sears variable speed grinder with a custom 6" diameter aluminum disk with several tapped holes for set-screw weights. You could beg, borrow, rent, or steal a balance analyzer or pay someone like me to do the functional tests and then go for lab calibration and certificate. You cannot get an unbalance weight display without first "calibrating" the machine with a known unbalance weight.

Walt

 

[JMarkWolf]

Hi Walter

We have talked in the past.

Yes this is correct. I have a commercial balancer (which may or not need calibration) with which to compare the results from my embedded algorithm, and the reports are very close, within 30 degrees and 0.02ips, repeatable with real-time acquisition from my desktop rotor simulator, and 10 degrees and similar phase discrepancy from previously acquired flight data. But I'd like to determine which is closer in order to characterize my algorithm.

Initially, I'd like to find someone whom I can email my acquisition data to, who is set up to analyze it.

 

[Strong]

Mark,

What is the format of the raw data? 30-degrees phase is not very close,unless it is a consistant shift that can be compensated by calculation. Does your vibration sensor or speed sensor have a phase lag that is different from the "commercial balancer"?

Walt

 

[GregLocock]

The guts of the maths, if you have phase information, is a two liner from memory. If you post the inputs and outputs many people here could have a look. Or if you are feeling shy just post the inputs (unbalanced vibration, and the trial mass experiments)

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[JMarkWolf]

The data format is an Excel file, one column for the key phasor strobe and one column for the raw accel data.

The 30 degree discrepancy appears to be quite repeatable. I suspect it has to do with the integrated Bessel filter in the MEMS accelerometer used to capture one data set, vs. the ICP accelerometer with hardware filtering used in the commercial balancer to capture the other data set.

A representative data set is attached.

The data parameters are:

sample rate = 1mS
accelerometer sensitivity = 1200mV/g
magnitude of the raw data is in volts.


Tell me the velocity in inches per second, and the phase in degrees.

 

[electricpete]

A representative data set is attached.

Nope, no data attached.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[JMarkWolf]

After posting last evening, data was attached, but is absent this morning.

Attempting again.

 

[JMarkWolf]

After checking repost, data is indeed attached with "little paper clip" icon.

We'll see if it remains this time.

 

[JMarkWolf]

Greg Locock said "The guts of the maths, if you have phase information, is a two liner from memory."

Hi Greg

I'd like to hear anything you can recall about this.

 

[Strong]

Mark,

The file cannot be downloaded because the path is on your computer and not accessible.

Walt

 

[JMarkWolf]

Oops. Misunderstood how the upload works. My mistake.

Should be there now.

 

[GregLocock]

All I got was two long streams of numbers enigmatically labelled ch13.





Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[GregLocock]

OK, for some reason your tacho pulse channel is a different length to the data channel.

All we could derive from this would be the phase between the pulse and the same frequency of the data channel, it tells you nothing about the balance.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Strong]

Mark,

Both channels have to be sampled simultaneously, therefore the time record should have the same number of samples for both channels (tacho and vibration). The tacho and accelerometer angular positions need to be known relative to the known heavy spot to verify that the vibration vector is correct.

Walt