I spend most of my time on the vibration board. Please excuse me if my question has been adquately answered in recent postings.

An electrician friend recently returned from some training in which the topic of diagnosing motor rotor bar faults was discussed. One of the presenters explained that it is possible to determine the number of rotor bars that are faulted. I believe he was looking at a current signature to do so.

My understanding is that rotor bar faults show up as pole pass sidebands around 60 Hz in the current or torque signatures. The severity of the fault is evidenced by the relative size of the 60 Hz peak vs the sidebands. Is there some evidence that can quantify the number of bars faulted? It seems like a rhetorical question in that it probably doesn't matter how many are faulted...simply knowing the severity is sufficient to make a decision. As a discussion point, however, is it possible to determine the number of bars faulted.

George,

Check out this post:

http://maintenanceforums.com/eve/forums/a/tpc/f/7161085912/m/7801080543

Especially Epete's post:

quote:

Originally posted by electricpete:
The attached article was at one time posted here for free download:
http://turbolab.tamu.edu/pubs/papers.html
Since they only keep the last three years, it's no longer available there.

Equation 2 on page 3 of 8 gives the following formula to estimate the number of broken bars:
n_broken=2*R/[10^(db/20)+pp]
where
n_broken is the estimated number of broken bars
R is the total number of rotor bars
pp is number of pole pairs (i.e. for 4-pole motor pp=2)
dB = dB down in curent at full load

There are a number of other similar formula's around.

Hello , everyone !! With great pleasure I 'm reading your post and I learned too much about motor testing.I have been worked with Microlog CMVA 60 and I found more than 10 motor with broken bars in my company but I didn't read anything from SKF about prediction number of broken bars. Did it this formula is relyable ? Sorry for mu bad English . .

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

The concept of all of the formulae is to 'estimate' the number of broken rotor bars based upon the amplitude (in dB) of the Pole Pass Frequency sidebands in relation to the current. Because of the variety of rotor bar to stator winding ratio's depending upon the motor manufacturer, the methods are used more to show the severity of the condition of the rotor.

Work performed by Howard Haynes at Oak Ridge National Laboratories in the mid-1980's in the development of Electrical Signature Analysis (of which Current Signature Analysis is a portion) for the detection of broken rotor bars for MOV's (Motor Operated Valves) in the nuclear power industry came up with a 'pass/fail' value of -35dB down from the peak current.

However, it is important to note that when you take current data from the Current Transformer of a motor (ie: a medium voltage motor) that the PPF and other peaks will be lower than if you were to take current directly. The difference will vary by the type and quality of the CT.

Howard

OK...I learned a bunch of stuff:

1. Thanks, Steve, for citing Pete's equation for estimating the number of damaged rotor bars from the current signature. It seems there is no way to specifically identify the number of broken rotor bars...instead just estimate the number of damaged bars by applying the dB down number to the equation Pete cited. I was fishing for the possibility that several broken bars would show up as 2nd or 3rd order PPF sidebands, but that's just my guess...not empirically proven.

2. If Arie is out there...I compliment your technique to use vibe signal enveloping to view the modulation frequency of RBPF. I think that is pretty damn clever. I had not considered that RBPF would modulate at PPF in the presence of damaged rotor bars. Do you suggest a way to determine severity of a potential rotor bar fault by looking at the PPF peak in the enveloped spectra?

3. One of Pete's postings cited a technical artical by Msr Thomson and Orpin. They apply a dB down criteria to determine acceptable airgap eccentricity. In this case they measure the amplitude difference between RBFF and 2xLF sidebands. Is there a thumbrule for acceptability here?

Thanks to all,
George

PS...Steve, I heard you talked to my boss, Mr Arnold, last week. It seems you may come up to see us. I look forward to it.

George,
Yes, I am still there behind the dikes and feet dry.
Thanks for your compliment. But thanks to envellopping technology too!
From field experience I found out that the RBPF modulation index (= amplitude of PP modulation frequency / amplitude of RBPF) expressed in [%] is often well in line with stator current AM index. So I stick to the rule: 1 % or -40 dB is a borderline condition.
The 2*Fline AM index IMO has nothing to do with rotor bar condition. It is an indication of airgap non-uniformity or rotor excentricity. Believe me, not many motors are concentric! The 2*Fline AM index is always there and normally does not harm induction motor performance and reliability.
Regards,
Arie Mol, NL

George - We are supposed to work out the details of our visit when Howard returns. He indicated that you may be coming down here for the Vibe Institute meeting at the end of November. I hope to see you there.

Arie...glad that you noticed our discussion. Thanks for responding.

OK, Arie...I want to try your enveloping idea on some of my route motors. If you have the patience, I hope to explain my understanding of your technique...then ask a couple questions to make sure I evaluate the data as you recommend.

I will collect a data point on the center of the stator housing...hopefully where it communicates best with the stator core. I will use acceleration enveloping (I am an SKF user) with a frequency band that includes RBPF. The display will be 0-200 HZ, max number of lines, 4 averages. I will look for PPF near the lower end of the enveloped spectra...indicating its modulation of RBPF.

Assuming my description above is correct...I have a couple questions about evaluating the data. You advised that 40dB delta between RBPF and PP modulation is a good threshold. Is it reasonable to measure the PP modulation amplitude from the enveloped spectra? Would it be reasonable to compare it to the RBPF amplitude measured from a separate survey of the stator housing...this time setting up the data collection as a simple acceleration spectra, likely 0-4000Hz, 800 lines? Any other hints you have to evaluate this vibration data to evaluate for rotor faults is very interesting to me. Thanks, in advance, for your attention.

George

George,
Let's see how this develops:
http://maintenanceforums.com/eve/forums/a/tpc/f/7161085912/m/2981009643
Arie

Thanks, Arie...I will follow it with interest.
George

George - How can I get in touch with you (email)? I know the company extension (@---.com) from Howard's email, but I need the first half.

Steve

I am "gxd1". Howard is "hra1". Look forward to hearing from you.
George

George,

I know this is an old subject, but do you have the capability of using PDMA gear? Anything I suspect with a motor I get our guys involved since they get a better look than I can with the data collecter.

Thanks for the hint, Mick. Our electric shop uses something that I believe is equivalent, or at least similiar to PDMA stuff. We use test equipment provided by Baker. The online tester is called Explorer. It uses both current and voltage inputs to evaluate percent full load, current signatures, torque spectra, et al. Our problem is that we don't put the necessary time into collecting enough data to adequately trend it all.