Hi Fellow,
I have a tough question for the wise old men of the vib. Community: Is it possible to use the TK17 Torsional Vibration Signal Conditioner with an optical device at lower speed than 400 rpm even if there is sufficient even? Why? I suspect torsional resonance on our dry Ball mill and I am trying to figure out a way to prove the management that finite element analysis is necessary to understand the phenomenon. Maybe one of you may help me find an alternative solution.

Thanks, Marcel

Marcel,

I don't have this instrument, but I checked the specifications at http://www.bently.com/prod/products/datasheets/141581.pdf

I am not sure why there is a 400 rpm limit. The lower limit of 8000 counts/min may be more important. If your encoder had 199 teeth (max allowed) then you could possibly go down to 40 rpm shaft speed. Why not put torsional sensor on motor shaft that is at higher speed? The motor shaft should "see" a system torsional (motor-gearbox-bowl-rollers). An optical encoder should work instead of Bently prox probes, sine the output is typically TTL compatible (0-5 V).

I have measured torsional vibrations on a coal mill with an optical encoder on motor shaft and connected to a Copp-Tek F-V converter. I also have a strain gage-telemetry system. Either one of these methods is probably easier to mount than the Bently system. I have also seen torsional vibrations by motor current spectrum analysis.

What are the symptoms of your mill vibrations? Do they occur at low load?

Good luck trying to model torsional response of a mill, because of the unusual roll-bowl boundry conditions. I would make torsional measurements first, and if resonance is present, then try making a model.

Walt
w_f_strong {at} msn dot com

The TK-17 has three limitations on measurement range, ALL of which must be met.

Shaft speed must be between 400-20,000 RPM
Number of teeth must be between 20-199
Tooth passage frequency (teeth/sec) must be between 240-25,000

The manual reads "These three ranges are all dependent on each other, so be careful that your setup meets all three requirements."

HI Walt,
The Ball Mill was first a wet design that was converted to dry milling about 35 years ago. We experience repeated tooth breakage and an open condition of the split joint of the girth gear for as long as I have been here (8 years). Two years ago, the management "invest" a lot of money to refurbish the mill. We machined both head and shell of the mill, poured new concrete on the base, changed pinion and girth gear with new superior delta design (Falk). Within 6 month the girth joints opened, pinion tooth broke with overall of 1.18 ips in axial pinion bearing. The girth gear was aligned and pinion was changed. It runs well for a while than a month ago a tooth broke in the joint of the girth without any sign. The girth gear was changed under warranty and it now generates a very weird spectrum. (See the attached files) Ghost frequency?

There was report in the old days of harmonics in the current, uneven wear of the pinion, joints bolts breakage, tooth breakage, girth and pinion and cracked motor's shaft. Last year I looked for resonance on all Ball mill's moving parts without success but a learning experienced. (1.18 overall was resonant axial condition of the outboard pinion's bearing housing) There is thermo expansion issue of the shell and the head due to dry milling but it cannot explain every things. The only thing left is torsional vibration either resonant or induce by the tumbling effect of the load. Last year I try to take some torsional reading by connecting, a contact tachometer to my data collector that gave some result but not strong enough to convince anybody. The interpretation of the square waveform was too shaky though there was sideband of various frequencies in the FFT. How can I transform a frequency modulated square waveform into a frequency-modulated sine wave proportional to angular speed of the shaft? Current FFT is an option but I am convinced that it will be challenged. Now there are so many people's egos involve that it is difficult to have any serious conversation about the subject. I really need to strong in my recommendation!

This message has been edited. Last edited by: Marcel B,

ball_mill__8.doc (2,900 Kb, 37 downloads)

Marcel,
I could not download your file, possibly blocked by firewall, etc..

If your contact tachometer output is only 1-pulse per revolution, then it is not adequate as a torsional sensor. You cannot do an FFT of the tachometer pulse train, because it needs to be converted to an analog voltage by frequency to voltage converter or with special signal processing software. Copp-Tech makes a good F-V converter.

In addition to measurement issues, it appears you may have a credibility problem with management. This may be solved by greater study and practice on your part or by hiring an expert for assistance. It looks like the modifications/repairs to date have not reslved the root cause of the failures.

Walt
w_f_strong {at} msn [dot] com

Hi gentlemen,
I am truly sorry; the document posted was too big. I will try a smaller file. I hope that that one will work. I am currently working with the electrician to set up the CT so I can acquire a spectrum of the current. Do you thing it is possible to take the current phase angle? Maybe by connecting the PT (with a transfo) into the tach input? I am still thinking of a way to convince my boss to buy a Copp-Tek converter that would be a very easy way to take torsional vib.

Regards, Marcel

Ball_mill__8.doc (1,428 Kb, 25 downloads)

Marcel,

The modulation sidebands around the tooth mesh frequency are spaced at 6x girth gear speed. Is the girth gear made in 6-segments or other structural members for support? Dimetral variations or stiffness variations spaced at 6 per girth gear circumference could cause amplitude modulation of tooth mesh.

Walt

Walt,
I send a new spectrum and you can clearly see that the sidebands are spaced at 2x girth gear, which correspond to the split joints of the gear. It is more pronounced at 3x split joint frequency.

Hi fellows,
Got some more questions and information for you. I took current and tension signature yesterday on the synch motor. The amplitude values are off because I had to use a home made set up (Tension and current on CT and PT with Differential probe SI9000 at 1/200 from Tie Pie engineering and current with AC clamp Hioki 9010-10) I got some very interesting spectrum but I˜m not sure if it is high enough to continue investigating. I joint a report of with associated spectrum and two files, one with the literature about torsion angle and means to compute the angle using Bessel function and another with the calculation of my result. Since it is not possible to joint more than one file at the time, I will post it on the next two replies. I assume that the twist angle of the rotor is proportional with the motor's current. I am not sure how the magnetic flux changes with the stator impedance, proportional or exponential with regard of the stator's current.. The sidebands at 558cpm come from the lifter liners inside the mill. There are 36 lifter liners within the circumference of the mill that turns at 15.5 rpm. 15.5rpm x 36 liners = 558cpm I suspect the frequency at 818.5 cpm to be from resonant condition. It is well visible in my vibration spectrum and is not associated with any mechanical phenomenon yet.

I count on you, the old and wisdom to get some advises.

Best regard, Marcel

ball_mill__8.PDF (425 Kb, 20 downloads)

Hi, from marcel

torsiona.pdf (715 Kb, 23 downloads)

Last one!

Beta.xls (20 Kb, 21 downloads)

I'm having a hard time following the discussion. That's a lot of info and a lot of data. Can you explain in simple terms the purpose of your Bessel function calculation?

Just thinking out loud. I wonder if strobe check would provide any clue? Freeze the shaft and look for movement at your suspect frequency. Absence of visible evidence probably doesn't prove anything, but if you see something, that's a start.

Also Engineering dynamics has some papers on torsional vibration.
http://www.engdyn.com/papers/papers_torsional.htm
From the table of contents, click on a link, then look for pdf icon on upper right side to download.

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

quote:

I joint a report of with associated spectrum and two files, one with the literature about torsion angle and means to compute the angle using Bessel function and another with the calculation of my result

I tryed help Martin Moore using Bessel function. View next link example on July 24 , 2005 10:37 A.M. Read aspects 1 and 2


http://maintenanceforums.com/eve/forums/a/tpc/f/375...641067351#5641067351

Regards and have a nice day

This message has been edited. Last edited by: Miguel Kovac,

An off-topic question

Miguel - how did you create that link that takes us directly to your post within the thread?

Electricpete, I put mouse arrow on my name and click, after select view post by Miguel Kovac then I search article wich I am interested, copy on a paper each one http://---letters and numbers----. After in the post I write http://---each one--- handly

HI Pete,
I am no teacher so I will do my best to be clear and if any member of the board notices any flaw on my explanation, please correct me for the benefit of everybody including me.
From my understanding, sidebands are a modulation (variation) of a carrier frequency. It has two components: frequency and amplitude. Assuming that the frequency or amplitude variation of the carrier frequency is function of a variation of the momentary angular speed of the shaft than Bessel function may be use to compute the angular amplitude or torsion angle of a specific sidebands. Angular amplitude of torsion is what ips is to vibration. No analysis I have performed so far has explained the failure of the girth gear so I am exploring torsion vibration to pursuit the root cause analysis.

Clear as mud,
Best regard, Marcel

Marcel,

I recommend a structural vibration test on Girth Gear using impact-response method. I have worked on a few gear wear/failure projects where resonant amplification was a significant factor. An unusual tooth contact pattern can sometimes be traced to gear/tooth natural frequency.

Also, remember that sidebands around a carrier frequency can be from amplitude modulation or frequency modulation. I would consider amplitude modulation more common for gear tooth forces.

Walt

HI Walt,
I tested the natural frequency of many components of the ball mill and did not get any correlation with the spectrum's ghost frequency. I performed that test with the help of Proaxion tech. in Montreal; we were able to detect many deflection mode and shape of the girth gear. Unfortunately they were present in the spectrum but at low amplitude and didn't match any ghost frequency. Maybe there is some things I didn't see? I will send you the result; maybe you will be able to dig something out of it. Too bad for quantity of data but it has been going on for the past 5 years.

Best regard, Marcel

wabush.xls (54 Kb, 11 downloads)