Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
RustyCas

Avatar / Picture

Admin
Registered:
Posts: 1,810
Reply with quote  #1 
I have a customer with a "draw bench" that reduces the diameter of copper wire.  There are 20 spindles, each with a die, turning at increasing speeds from inlet to outlet, at a fixed ratio.  But this is actually a dual-strand machine, so there is an A line and a B line, stacked atop each other. There are also 19 idler shafts, containing a gear that drives the uppers and lowers, and the idlers turn at increasing speeds as well.  So we have 59 shafts (they call them "axles"), with 39 different speeds.  The slowest, the C1 idler turns 92 rpm and the fastest A20/B20 turns 9685 rpm.  These axles are all fitted in a single housing, on about 12" centers. 

The A axles have gears with 29 or 36 teeth.  The B axles (bottom) have 29 and 36 teeth.  The idlers all have 62 teeth; the 'odd' idlers mesh with 4 gears, with the 'even' meshing with 2 gears.

Obviously with speeds ranging from 92 to 9685 rpm you can't use the same setup for each axle.  How would you approach setting up the Fmax for the different axles?  I really don't want a different setup for each axle, and vibration will travel - vibration at any axle may show up at a number of other axles -  so there needs to be some "overlap" in the Fmax's.

Anyone have experience with something similar?

Draw Bench.png 




__________________
"The trend is your friend"
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,398
Reply with quote  #2 
In 1 word no.  But I would tackle it a bit like a paper machine and build all the ratios in the database. You have all the z's so you will know the speeds. And then off course my favourite.  AP set 120x 1600 lines or 150x 3200 lines. Everything will work itself out after that.  All the bearing frequencies will fall into place if present. rgds

Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,398
Reply with quote  #3 
I once worked on a money printing machine.  Had lots of different gears and axes.  Walking through the place was like fort Knox.  Was successful at finding a bearing fault even though I had never taken measurements on the machine before.  It was a weird bearing. Sort of like a needle roller bearing but instead of needles each cage was filled with small balls.  This allowed rotational as well as axial / oscillating motion in the direction of shaft centre line.  Machine speed was measured in notes per minute.  We have some very fancy bank notes here in Australia and make them for New Zealand, Vietnam and quite a few other countries.  They are made from polymer and can go through washing machines.  Last I think it is 10 times the life of paper notes and very difficult to counterfeit.  They wouldn't let me take the machine home. rgds
Walt Strong

Sr. Member
Registered:
Posts: 889
Reply with quote  #4 
"They are made from polymer and can go through washing machines"

Makes money laundering very practical!

Rusty,
You could use airborne ultrasound to scan for suspect bearings and gears without contact. If you don't have an ultrasound meter then use your 1/4" ICP microphone mounted in a sound tube to make noncontact audible sound scans. It might save considerable time and frustration compared to looking for "needle in the haystack" in a lot of vibration spectra.

Walt
Beatnik

Sr. Member / Supporter
Registered:
Posts: 183
Reply with quote  #5 
You'll need a very good definition in your spectrum and twf, including in peakvue. On complex machines that have a low speed output; some gearmeshes, a broken/bad tooths and BPFOs can be very close in frequency. Also it's entirely possible that they have multiple brands of bearings, even on shafts with the same speed and size. I know you always use your speedvue and it will serve you well.

Ask for oil analysis results if they have them and take the temperature of every slow bearing.

Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 599
Reply with quote  #6 
Quote:
Originally Posted by Walt Strong

You could use airborne ultrasound to scan for suspect bearings and gears without contact. If you don't have an ultrasound meter then use your 1/4" ICP microphone mounted in a sound tube to make noncontact audible sound scans. It might save considerable time and frustration compared to looking for "needle in the haystack" in a lot of vibration spectra.

Walt


Walt, what do you mean using a mic mounted in a sound-tube? I'm having trouble visualizing this.
RustyCas

Avatar / Picture

Admin
Registered:
Posts: 1,810
Reply with quote  #7 
Since I didn't have gear info previously, I was using the "forrest fire" approach:  looking for changes in spectra, trending maximum peak acceleration, and using headphones with the 2140 as I collected data.  Trends were increasing, some axles were "noisy", and they were breaking-out (breaking wires) more frequently, so the machine was rebuilt with new bearings.

Most points saw significant improvement, but one of the input axles (drive is in the middle of the unit) jumped way up.  I took two sets of route data, an hour apart, and got almost identical readings, so it's not a an accel mounting error.  But I wonder if the bearing cap may be slightly loose, amplifying the signal.  The only readings I can take are axial, and only on one end of the axle.  Going back today to have another look.  I will also add a thermal scan to my monitoring scheme.

There are 5 of these machines: 2 newer double-strands, a newer single-strand, and 2 ancient singles (one with axles stacked in tall zig-zag pattern).  A while back I told them which axle a noise was coming from on one of the older machines;  didn't have the speeds, didn't know the exact locations of the axles, but I knew it was the axle that was "65% of the input speed" and once they got the gearing info they were able to pinpoint it and correct it.

So now they think I can do that on every machine.  And I probably can, with enough information, and time, and $$$ -- told them it would double their cost and they didn't blink, so...

It's an interesting challenge.  Thanks for your inputs.  Really helps when you work in a vacuum.

__________________
"The trend is your friend"
Walt Strong

Sr. Member
Registered:
Posts: 889
Reply with quote  #8 
Curran,

Microphone is mounted at end of a tube. The concept is to block sound from the side to make Omni-directional microphone more directional. One a recent job, I was measuring vibration surface velocity for sound power radiation and used a soft tube pressing against surface. Photo shows a test setup (sanity check) to compare microphone to accelerometer mounted on a steel flat bar.

Here is an old paper I found after I made mine sound tube:

"USING A "SOUNDTUBE" TO MEASURE NOISE OF
STRUCTURAL SOURCES IN HIGH BACKGROUND NOISE ENVIRONMENTS" 
by
James T. Rainey

Microphone-Sound-Tube and accelerometer on steel plate.png 

Walt
 

OLi

Sr. Member
Registered:
Posts: 1,918
Reply with quote  #9 
I also used similar to hunt ghost's sources for screaming noise in pipe systems. I made mine from PVC tube and found by trial and error that a certain length was a better compromise for directional spread, level etc. I used a noise meter with a signal output but same principle.
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
RustyCas

Avatar / Picture

Admin
Registered:
Posts: 1,810
Reply with quote  #10 
To analyze the old data once I got all the shaft speeds, I just used a spreadsheet.  I tabulated all the major or interesting peaks and then divided each peak frequency by all the shaft speeds. Since most of the normal vibration is gear mesh, the problem shafts were readily identifiable.
__________________
"The trend is your friend"
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,398
Reply with quote  #11 
Way to go Rusty. It's all about getting the correct shaft speeds and everything else falls into place. rgds
Jim Crowe

Member
Registered:
Posts: 53
Reply with quote  #12 
Rusty,

Remember what I always say "The 3 most important things to know when analyzing vibration data is : Know running speed, Know running speed, Know running speed."
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,398
Reply with quote  #13 
Can't agree more, the vast majority of characteristics / faults are simply either integer or non integer multiples of shaft speed.

And speed is usually easy to measure with a CSI Speedvue which can measure speed without having to stop and attach reflective tape. Brilliant tool. rgds
WWST

Sr. Member
Registered:
Posts: 86
Reply with quote  #14 
I have a completely opposing view of this that I'm sure will stimulate a lively discussion. You may want to consider Installing an online system that takes data every 30 minutes. Trend all useful features (acceleration, velocity, displacement, twf pk-pk, CF, autocorrelation, % of synch/non-synch and temp and you're done. We are doing this on 10,000 (soon 30,000) points right now in a somewhat similar (although not as harsh) setup with many rolls close together at different (variable) speeds. Most of the time, we don't know the exact running speed. With all the tends it is fairly easy to identify bad actors that need more analysis. With 24/7 362 monitoring, every week we identify many developing problems in different stages and provide the information directly into their work order system for action and feedback/close out in s cloud-based system accessed by all maintenance personnel.

We have saved this customer in excess of $2 million in downtime already. With 15 more sites coming online this year, we stay busy...

Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.