I have several bearings that have harmonics of turning speed. In the peak-vue data taken at 500 hz. 1600 lines with a 500 high pass filter.
The data will have multiple harmonics of turning speed of the shaft.
This is on the felt rolls and dryer cans of a paper machine. What experiences do you have using peak-vue?

The speed of your machine would be a factor in choosing the filters. If you are running 1000 feet per minute or less, 500/500 may be good. I am more aquainted with 2K to 4K machines, and that requires 1000/1000. Harmonics of that nature would not concern me (maybe some imbalance?). The peakvue's main purpose for me is inner race defect. Outer race will show up in the regular readings just as readily. Peakvue is a valuable tool, but perfecting it for your equipment will take some trial and error.

Peakvue is seeing something equal to running speed in the higher frequency area, probably sidebands around gearmesh or inner race defect or some other carrier.

If running speed was extending that far out in the higher frequency on its own as harmonics, you probably could see it in normal data as a 1X. You probably could see where the running speed is with an Fmax of about 1K 3200 lines.

Being an old dryer can man I would guess from the information that you have gearmesh sidebanded by 1X, even on the felt rolls. If not, you may have bearing trouble but probably not if showing on more than one point scattered out on the machine.

Ray,

Your setup looks OK to me for this application. In general, 1x in PeakVue indicates either friction modulated at 1x or impacting at 1x. I suggest looking into normal vibration data taken with Fmax greater then 500 HZ in order to identify the carrier frequency and possibly adjust HP filter.

Also look at the TWF in PeakVue. It will give you some insight. I you can, please post your data.

David

quote:

The peakvue's main purpose for me is inner race defect. Outer race will show up in the regular readings just as readily. Peakvue is a valuable tool, but perfecting it for your equipment will take some trial and error.

RNL,
We take Peakvue on our paper machines and pick up Outer Race readings before they show up on regular readings. Most of the readings are taking at 1000hz for peakvue and regular. We have started backing some of them off to 500hz on both. We have had 2 bearings fail in the last 2 years on felt rolls that showed OR harmonics on Peakvue, but nothing on the regular reading. We were having greasing problems(wash up water getting into the grease) and I think they failed faster than a normal progressive failure. We take readings once a week on the paper machines.

Hi JimB -
I find that very interesting. Do you have any data or pics from the failure you could post? I will check my stuff tommorrow, but I vaguely remember one felt roll I pulled for outer race in just my peakvue readings, and all I could find was dried grease. I am leary to pull on pk vu alone unless it is bpfi. I have not noticed alot of difference between pk vu 1000 or 500 hz on felt rolls. It does seem to help some on goulds pumps (1200 to 1800 rpm), but both work. Have you had any experience dealing with 3600 speed pumps? I think my next step is to try a 2000 hz reading, but no time to get involved yet.

Ray, maybe this does not matter, but since Ralph mentioned gearmesh, is this TS or DS of the PM?

In case anyone is interested. Note that the reading before replacement showed nothing. ???

peak_vue.xls (1,471 Kb, 54 downloads) peakvue

RNL,

Thanks for sharing the data history.

Peakvue is a good tool in its place but sometimes its place is not always where we want to place it as is, not to say yours is the wrong place.

I still have a big gripe about the HP and Fmax ruling.

Ray,

If you can email me an extracted database with some of your data (say, 2 or 3 months data on several points in question), I would be more than glad to look and see what I think I see.

rnl,
I use Peakvue also on a papaermachine. If you don't use the waveform as your severity calculator, you'll probably call em too early. The spectrum will show you the frequency of interest, but not necessarily how bad it is. Look at the Pk amplitude in the waveform. Also, autocorrelate your waveform just for frequency verification. Your 1000 hz hp is fine for the roll that you have shown in your example. One thing to watch is, if you notice the defect "diminishing", then you may want to lower your filter at that time. On some slow turning rolls, I will even acquire a 100-600 hz band pass, but usually for these, you have to rely almost solely on the waveform, because you may only have the first harmoninc of the bearing frequency show up due to only a 100Hz fmax on the spectrum.

Scott

Thank you Ralph. But I don't think the IS dept is going to let me do any extracting. If there is some filters or fmax that you would like me to experiment with, email me direct : rnl1@newpagecorp.com

Thank you Viberscott. I have not used the waveform in any way. As soon as I get the opportunity, I will go back and see if that would have helped. I do have some slow speed stuff that I would like to start getting peakvue on.

quote:

Thank you Ralph. But I don't think the IS dept is going to let me do any extracting.

Now that is wierd. Do they not want any "free" help on analyzing or setup? Maybe they would feel more comfortable hiring me to do some help, huh?

You must be on a network version, huh? I have a client on network and I have had to get the IT, or whatever they call themselves, guy to enter his password so many times on extracting and merging databases during database reconstruction at their plant that he finally gave us the password. Well maybe the your company doesn't want to "send out" any setups or vib data in fear someone will "steal" some ideas, huh?

Oh Well, to each their own.

As far as Fmaxs, etc., I don't have any suggestions at this time, being totally in the dark on where the IS department is on their control of the database functions. The dept might not want any outside suggestions, huh?

Sorry.

Thanks for the input about peak-vue. So, the harmonics seen in pk-vue are not that important?
Inner race defects with high impacting in the TWF ( 3-5 G's ) is the real value of using pk-vue? Filters used -- depends on the speed?
Am I understanding the rules of using pk-vue?

Fractional gearmesh is a problem in the cans and felt rolls.

Thanks for the responses, please send more,
Ray

quote:

So, the harmonics seen in pk-vue are not that important?

In basic terms, the presence of harmonics indicates the signal was "strong". The PV methodology holds the highest amplitude in each time block defined by 1/Fmax. This will stretch the short duration, periodic pulse and make it look like a square wave. An FFT of a square wave will have the fundamental frequency and many harmonics.

Here is a question I have for you all using PeakVue on the slow movers. I recently met someone who used a low Fmax like 20ish and a 2KHP filter. The Data that I saw was very impressive. I know that here we are talking a little faster speed here but was just wondering if you all have ran across that. Most of the time that I used the 500HP and the 1kHP I never really saw a difference. Are you all seeing a difference when using that. I have also used the 100-600BP and 500-1000BP and seen 1x harmonics and found a bearing that had 10Mil of excessive looseness in it. Most of the time when I have used PeakVue and seen 1x Harmonics is was due to looseness- housing to bearing or shaft to bearing. Is this something any of the rest of you have experienced. Thanks!

quote:

Here is a question I have for you all using PeakVue on the slow movers. I recently met someone who used a low Fmax like 20ish and a 2KHP filter.

OJU - A little experiment. Maybe it helps, maybe not. It may not be applicable to what someone told you about using peakvue on low speed stuff. I did not do a 50/2000, sorry. I assume they were talking about using peak vue to determine greasing schedule?. Band pass is over my head. I did see where the different BP's were an option. But is that really peakvue? I thought peakvue was a snapshot of impacting at a high frequency. For this experiment I used one of my slower pumps that I know has an inner race defect currently. CSI AP sets would only allow me to go as low as 500 filter. The 50 f-max/500 hp filter only gave the 1st fundamental of bpfi, mixed with a bunch of other harmonics that I do not recognize. The 100/500 was a little better, but not the classic bpfi peakvue that I am used to. The 500/500 is what I would call the best. The 500/1000 is ok, & you could get by with it. The 500/2000 loses the bpfi to the point that you would be taking a chance of not catching the defect.

50-500.xls (273 Kb, 11 downloads)