Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment   Page 1 of 2      1   2   Next
arie mol

Avatar / Picture

Member
Registered:
Posts: 27
Reply with quote  #1 
Does a 'haystack' at the root of a frequency peak in vibration frequency spectrum points towards a structural resonant condition close to this frequency?
Any document available that supports this diagnosis?
John from PA

Sr. Member
Registered:
Posts: 810
Reply with quote  #2 
Best to add a representative plot and the basic machine details.  Depending on where the frequency is, relative to the anticipated machine frequencies, it might mean a few different things.
electricpete

Sr. Member
Registered:
Posts: 517
Reply with quote  #3 
I agree, specific spectrum would help.

I also agree something resembling a haystack may be associated with resonance. I had talked about how resonance may show up as an increase in the noise floor in vicinity of a peak here - discussion starts at bottom of the page and continues to the next page where there is an example rolling element defect spectrum where resonance amplification shows up well in the noise floor on log scale. 

https://www.machineryanalysis.org/post/balancing-a-cooling-tower-fan-10253775?highlight=noise+floor&trail=15
arie mol

Avatar / Picture

Member
Registered:
Posts: 27
Reply with quote  #4 
ePete, thanks for the link suggestion.
Shurafa2

Sr. Member
Registered:
Posts: 78
Reply with quote  #5 
Quote:
Originally Posted by arie mol
Does a 'haystack' at the root of a frequency peak in vibration frequency spectrum points towards a structural resonant condition close to this frequency?


I would say, this question implies some oversimplification because the answer can be yes and no. I'll explain what I mean.

Based on my limited experience, there are resonance cases that manifest this symptom but most of the structural resoance cases I saw did not. Also, there are other fault reasons that generate a similar symptom like vane pass frequency with turbulence.

If the spectrum you analyze contains a high 1X vibration amplitude with a significant haystack around the peak, there might be additional reasons with the resonance that contribute to the resultant vibration. For example, are you sure the speed is constant? Does the spectrum have enough resolution? Did you use reasonable averaging? Be careful when using log scales if you are only familiar with linear scales.

Quote:
Originally Posted by arie mol
Any document available that supports this diagnosis?


I saw recently Mobuis Institute materials that explain the structural vibration with haystack around the resonance/forcing frequency. I do not know if the document has a reference tag or if it is in the public domain.

Regards- Ali M. Al-Shurafa

JasonTranter

Member
Registered:
Posts: 26
Reply with quote  #6 
What we try to explain is that if you happen to see that the region around the peak is elevated, or the peak appears a little broad, it may indicate the presence of a natural frequency. We then recommend toggling to log amplitude scale and you may see a bigger "hump".

We also explain that "haystacks" can represent that noise is being generated (e.g. from a lubrication issue or cavitation) which excites a natural frequency and thus you get resonance.

Sorry for the poor graphs attached...

Attached Images
png Haystack.png (63.87 KB, 15 views)
png Hump near peak.png (30.28 KB, 14 views)

Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,245
Reply with quote  #7 
Hi All,

Attached is an example of a small haystack which is a press nip resonance.  30 Aug & 19 Sep are low machine speeds, the others which show the BPFO are at higher speeds.

1. In orders the BPFO harmonics nicely line up, in Hertz they don't
2. The hump around 165 Hz is the resonance.
3. harmonic falling in the resonance range is amplified.
4. Fault severity is easily over estimated due to the harmonic falling in resonance is amplified.
5. @Jason - you have a better example of this in your archives.  same set of rolls, same bearing - different paper machine.  That one shows how a different harmonic falls in the resonant range at different machine speed. You even have a photo of damage found and how that call was not overestimated.  The resonance really pushed levels through the roof.
6.See how much lower levels are when machine speed is considerably lower.
7. A plot in log vertical axis. 
8. Fun and games these paper machines.
rgds
ps. Another reading to come 30 Sep. Replacement on 3 Oct.

 
Attached Files
pdf Press Roll History of data.pdf (194.97 KB, 16 views)

fburgos

Sr. Member
Registered:
Posts: 443
Reply with quote  #8 
nice example thanks vibe-rate, can yo post the acceleration witout log scale, plot one is velocity you have the label acceleration units.

I like both plots, in orders bpfo line and plot in herz the resonance line up.
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,245
Reply with quote  #9 
You are right fburgos, I'll fix when I have a bit more time. The resonance is also very clear in the non Hurtz plots, pun intended.
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,245
Reply with quote  #10 
All fixed. rgds
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,245
Reply with quote  #11 
Bit of background.  Site was planning to change major roll next to this one and didn't want to do this one because an extra crew was required, cranage problems and safety / complications / compromising of 2 crews working along side. So I was asked will it last.

Standard response was how long is a piece of string.  But more seriously, let's see what happens over time. Gut feel is that it will last until following shut first few weeks of November.

So monitoring every few days, and taking into account machine speed differences - it is looking better every day that no significant change is observed.  So confidence increases all the time.

Then maintenance plan changed completely (think 10 engineers involved - plenty of nappies being handed out (diapers for the US people) and now roll will be changed early October.  We'll see.  I'll try to source photo of the damage there but I suspect the inboard raceway will have shed a relatively small area of surface.  Think an area 75mm x 50mm. Or 3 x 4" or thereabouts.  Will aim to get data in hours in the lead up of replacement.  See if I can get a photo I'll share. rgds
OLi

Sr. Member
Registered:
Posts: 1,494
Reply with quote  #12 
I have a customer that will pull the next yankee bearing a bit earlier next time, raceway looked like a gravel path... but it lasted as long as they wanted, like a year more than you really wanted but if you have luck... I have another where the data skyrocket and he pulled it early and was just as happy for that it was like not a minute to early.. So trend is everything IMHO and fault propagation.
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Vibe-Rater

Avatar / Picture

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,245
Reply with quote  #13 
I agree OLi,

I find quite often when an outer race fault develops (not so much inner race), then it stays the same for a long time, sometimes it deteriorates straight away. Experience is the answer for keeping your cool and seeing the deterioration rate.  No one teaches that.  And you can tell young players by that very fact.  They jump at things particularly Peakvue and the like.  I help out at quite a few internal departments and find myself saying ... hang on hang on ... lets see..  a lot.  Guess it keeps me in a job.
Danny Harvey

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,222
Reply with quote  #14 
"I find quite often when an outer race fault develops (not so much inner race), then it stays the same for a long time, sometimes it deteriorates straight away. Experience is the answer for keeping your cool and seeing the deterioration rate.  No one teaches that.  And you can tell young players by that very fact.  They jump at things particularly Peakvue and the like.  I help out at quite a few internal departments and find myself saying ... hang on hang on ... lets see..  a lot.  Guess it keeps me in a job."

Sometimes experienced people wait too long, too.

I was called to do analysis on a huge Yankee bearing a few year back. I didn't know that these folks had their own guy who had years of history on the bearing.  I checked it out and sure enough, bpfo harmonics in both the PV and Velocity spectra and in the twf's.  

Then I went back to the guys office and he showed me the data history which showed bpfo harmonics in velocity spectra going back 18 months and he said it went back further. He was using the same equipment and software I was using but using SST and a 500 mv/g accel instead of PeakVue. This seems to be typical of smaller in-house programs. They are frozen in time to where they were first trained. His was with Mastertrend and prior to PeakVue.

But he had found and tracked this defect for over 2 years and couldn't make the decision on when to make the change.  They called me on Christmas Eve during the scheduled outage and wanted me to tell them that it would make it until the July outage.

It looked to me to be starting to lose it's geometry and had been for about 4-6 months so I told them I doubted that it would and they had better at least get the bearing on site and start trying to find some potential down time.

It made it another 8 days before it turned a roller.

So here was an experienced guy who waited too long. I don't recall what his twf's looked like from the 500 mv/g accels but what started out a clean harmonics of bpfo in his spectra had gotten steadily more complex with multiple sets of sidebands and random noise.  (That's what these particular haystacks were.)  I don't make a call based solely on spectra but I don't think this fellow ever looked at twf's. He hadn't been trained that way. 
RustyCas

Avatar / Picture

Admin
Registered:
Posts: 1,566
Reply with quote  #15 
One of the first jobs I did when I went out on my own was a bearing on a sheet rubber calendar stack. They “suspected” they had a bad bearing, but wanted confirmation. This was pre-PeakVue and I too used a 500 my sensor and SST to determine there was a severe outer race defect on this bearing running 11 rpm. 1x BPFO was dominant, with multiple harmonics. Though I’d never analyzed a bearing this big or this slow, it was an easy call.

A month later I went to take a look as they were replacing it. Before seeing the now removed bearing, I noticed piles of something in the bearing housing....”What the heck?” I scooped out a handful of oily metal filings. I asked a nearby mechanic if they’d seen any of this before. He said, “Oh, we’ve been pulling that off the screens in the oil return for a year or so.”

__________________
"The trend is your friend"
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.