Case Study

Case Study :

CenterHung Fan ,spherical Roller bearing,1000 Rpm Speed ,lubrication is oil 

Dominant frequency is 1XTS 

TWF about to be sinsoidal in vertical direction 

PeakVue is concerning and when using correlation Factor it showed a periodic peaks as shown below 

Accelerometer 100mv/g

units mm./sec

Result is :Bearing is rotating on shaft and excessive wear on shaft.

i need your comments.

horizotal (Spectrum mm./sec)img00005

Verticalimg00006Axial img00007

correlation FactorUntitled

Original Post
Hossam 2 posted:

Case Study :

CenterHung Fan ,spherical Roller bearing,1000 Rpm Speed ,lubrication is oil 

Dominant frequency is 1XTS 

TWF about to be sinsoidal in vertical direction 

PeakVue is concerning and when using correlation Factor it showed a periodic peaks as shown below 

Accelerometer 100mv/g

units mm./sec

Result is :Bearing is rotating on shaft and excessive wear on shaft.

i need your comments.Just wondering what kind of comments you need? Root cause analysis?

horizotal (Spectrum mm./sec)

I can not see any images

img00005

Verticalimg00006Axial

 img00007

correlation FactorUntitled

May just be my system for the reason I can not see the images.

Thanks,

Ralph

Ralph Stewart posted:
 

i need your comments.Just wondering what kind of comments you need? Root cause analysis?

actually i have a problem with diagnosing these kind of bearings which lubrication is oil not Grease ,as the G's Level is very low this bearing  was 0.1 G's Rms and the dominant frequency is 1XTS

in technical Associcates of charlote it stated that when bearing turning on shaft it devided them into 4 stages ,as shown in attachment we didnt even reach to stage number 4 but in the stage no.1 at the existance of only 1X we find an excessive wear and feeler gauge 0.2mm could be insterted between bearing inner race and shaft.

Abbreviated Last Measurement Summary
************************************
F1P - Fan Inboard Horz Peakvue .199 G-s
-------------------------------------------------------------------------------
Clarification Of Vibration Units:
Acc --> G-s RMS

Hossam

Attachments

Ralph Stewart posted:

Do you have a Peakvue in the vertical direction, since its amplitude is twice as high as the horizontal?

The vertical velocity data appears to look as though it might have some 1/2x between 6k and 10k cpm frequency.

Actually i never take pkvue at vertical as per emerson recommendation.

I wonder whether the ball and inner&outer race was damage. It look like the problem by rubbing with the shrink fit of inner race is not correct on shaft. I think that 4 stages of bearing damage focus on race, ball, cage of bearing.

I could be total wrong!

Hossam 2 posted:
Ralph Stewart posted:

Do you have a Peakvue in the vertical direction, since its amplitude is twice as high as the horizontal?

The vertical velocity data appears to look as though it might have some 1/2x between 6k and 10k cpm frequency.

Hossman Quote:

"Actually i never take pkvue at vertical as per emerson recommendation."

Hmmmm! "Strange, very strange indeed".

If we look on  page 12(section 2.3.3)  Page 17 Figure 8  Page 18 Figure 9  Page 19 Figure 10   Page 57, Figure 27  Page 58, Figure 28 and29   Page 59, Figure 30                      in the article located at the address shown below.

It seems as though, only in my jokingly manner,   someone forgot to tell Mr. James Robinson and Mr. James Berry this recommendation of no vertical data.

There are several more data sets taken in the vertical direction, further on in the article. Noted only a few in the horizontal direction (maybe 5)

Or maybe there has been an update to the theory in the years since then. ??????? This paper was written and published in 2001.

http://www2.emersonprocess.com...ductionPaper2001.pdf

Personally, I take data where I think it will show me the most information that might help me. Just my opinion and I could be totally wrong.

Thanks and Have a Great Day,

Ralph

....

Ralph Stewart posted:
Hossam 2 posted:
Ralph Stewart posted:

Do you have a Peakvue in the vertical direction, since its amplitude is twice as high as the horizontal?

The vertical velocity data appears to look as though it might have some 1/2x between 6k and 10k cpm frequency.

Hossman Quote:

"Actually i never take pkvue at vertical as per emerson recommendation."

Hmmmm! "Strange, very strange indeed".

If we look on  page 12(section 2.3.3)  Page 17 Figure 8  Page 18 Figure 9  Page 19 Figure 10   Page 57, Figure 27  Page 58, Figure 28 and29   Page 59, Figure 30                      in the article located at the address shown below.

It seems as though, only in my jokingly manner,   someone forgot to tell Mr. James Robinson and Mr. James Berry this recommendation of no vertical data.

There are several more data sets taken in the vertical direction, further on in the article. Noted only a few in the horizontal direction (maybe 5)

Or maybe there has been an update to the theory in the years since then. ??????? This paper was written and published in 2001.

http://www2.emersonprocess.com...ductionPaper2001.pdf

Personally, I take data where I think it will show me the most information that might help me. Just my opinion and I could be totally wrong.

Thanks and Have a Great Day,

Ralph

....

Thanks for your reply

I will attach the paper which tells the horizontal or axial but didnt stated vertical direction..

The point is that the vibration mostly less in vertical direction and too many times i face a less overall values like 3 and 2 and spectrum shows a dominant 1xts but the bearing turning on shaft specially on shafts repaired by belzona 

Regards

Thanks. I guess everyone does what they think is best for them.

But my theory is, sometimes it might be best to think "outside the box" and not stick to the "cut and dried" opinions of others, who may have never encountered something which you or I may be experiencing at any given time.

Thanks and Have a Great Day,

Ralph

 

Ralph Stewart posted:

Thanks. I guess everyone does what they think is best for them.

But my theory is, sometimes it might be best to think "outside the box" and not stick to the "cut and dried" opinions of others, who may have never encountered something which you or I may be experiencing at any given time.

Thanks and Have a Great Day,

Ralph

 

I still need help what is the best way tp monitor the bearings (oil lubricated) as the Gs level is very low as shown and spectrum is tricky 

Thanks in advance

Hossam 2 Quote:    I still need help what is the best way tp monitor the bearings (oil lubricated) as the Gs level is very low as shown and spectrum is tricky 

Thanks in advance

Hosman 2,

I am guessing, when you say, "g's level low", you are still referring to a bearing which is loose on the journal.

IMO, there is really not a set procedure for this analysis of a spherical roller (or ball bearing) looseness on the shaft.

One must look for harmonics of running speed and sub harmonics 1/2x, 1/3x, 1/4x, etc., which are not always very distinct, especially when using an auto scaled amplitude where there are a few amplitudes present which could result in a high vertical scale which "hides" lower amplitudes, which might prove to be important.

See the question in the attached image.

If the 0.05 Gs vertical scale setting is not correct to show the frequency peaks in this zoomed area at about 75% of the full scale plot, adjust it to be about 75%.

Thanks,

Ralph

Spacing between all peaks between 6000 and 11000 cpm

 

Attachments

Photos (1)
Ralph Stewart posted:
Hossam 2 Quote:    I still need help what is the best way tp monitor the bearings (oil lubricated) as the Gs level is very low as shown and spectrum is tricky 

Thanks in advance

Hosman 2,

I am guessing, when you say, "g's level low", you are still referring to a bearing which is loose on the journal.

IMO, there is really not a set procedure for this analysis of a spherical roller (or ball bearing) looseness on the shaft.

One must look for harmonics of running speed and sub harmonics 1/2x, 1/3x, 1/4x, etc., which are not always very distinct, especially when using an auto scaled amplitude where there are a few amplitudes present which could result in a high vertical scale which "hides" lower amplitudes, which might prove to be important.

See the question in the attached image.

If the 0.05 Gs vertical scale setting is not correct to show the frequency peaks in this zoomed area at about 75% of the full scale plot, adjust it to be about 75%.

Thanks,

Ralph

Spacing between all peaks between 6000 and 11000 cpm

 

im so sorry for the late reply but it was my work circumstances.

here is the snapshot please tell me if this is the required snapshot or not 

img00003

Thanks for help

Attachments

Photos (1)

What was the trend of velocity & peakvu

Was it high & later on reduced without any maintenance ?

This must be a taper bore bearing which sits on a taper sleeve with a lock nut & washer

When the bearing is mounted the nut is locked with washer

Did you see the locking when bearing was opened?


im so sorry for the late reply but it was my work circumstances.

here is the snapshot please tell me if this is the required snapshot or not 

 

Thanks for help

 

Thanks Hossam.

See the attached image where I have peaks marked red and green.

I would like for you to measure the RED, using the "set mark" function near the center of area and sideband cursors off the set mark and see if the RED are equally spaced at the true running speed (possibly the true speed is slightly less than the spectrum label of 1000 rpm).

Also do the same to the GREEN and set mark and see if GREEN are also equally spaced at true running speed.

Thanks and Have a Great Day,running speed space

Ralph

Attachments

Photos (1)
Ralph Stewart posted:

im so sorry for the late reply but it was my work circumstances.

here is the snapshot please tell me if this is the required snapshot or not 

 

Thanks for help

 

Thanks Hossam.

See the attached image where I have peaks marked red and green.

I would like for you to measure the RED, using the "set mark" function near the center of area and sideband cursors off the set mark and see if the RED are equally spaced at the true running speed (possibly the true speed is slightly less than the spectrum label of 1000 rpm).

Also do the same to the GREEN and set mark and see if GREEN are also equally spaced at true running speed.

Thanks and Have a Great Day,running speed space

Ralph

this peaks are separated by the running speed of the fan 998 cpm

img00006

Attachments

Photos (1)
Hossam 2 posted:

this peaks are separated by the running speed of the fan 998 cpm

img00006

Thanks Hossam.

Are the peaks I have marked in RED also equally spaced at a frequency equal to running, as are the ones you have marked with the dotted lines labeled "A"?

If the ones marked by the RED "X" are also separated by a frequency equal to running speed, that would make them also one half of running speed, which, IMO, is a good indication there is looseness between the journal and the inner race (or the sleeve and the journal, if the bearing is of the tapered bore type.)

Only my opinion based on experience and I could be totally wrong.

Please see the attach image.

Ralph

Loose on Shaft

 

Attachments

Photos (1)
Ralph Stewart posted:
Hossam 2 posted:

this peaks are separated by the running speed of the fan 998 cpm

img00006

Thanks Hossam.

Are the peaks I have marked in RED also equally spaced at a frequency equal to running, as are the ones you have marked with the dotted lines labeled "A"?

If the ones marked by the RED "X" are also separated by a frequency equal to running speed, that would make them also one half of running speed, which, IMO, is a good indication there is looseness between the journal and the inner race (or the sleeve and the journal, if the bearing is of the tapered bore type.)

Only my opinion based on experience and I could be totally wrong.

Please see the attach image.

Ralph

Loose on Shaft

 

yes the are equally spaced too by the running speed and the difference between the marked A and Red X at is about half the running speed but my question is 

how could this be a early warning but the actual situation is that a worn shaft under bearing about 0.2 mm 

the second question what is the reference that i should call it a looseness as the G level is low and why did you choosed this band from 6000 to 11000 cpm at a previous comment 

thanks 

Hossam 2 posted:

Loose on Shaft

 

yes the are equally spaced too by the running speed and the difference between the marked A and Red X at is about half the running speed but my question is 

how could this be a early warning, By early warning, I simply mean, a signal early enough to possibly schedule a shutdown rather than running until it became obvious that there was a worsening looseness problem, when the 1/2x or even 1/3x energy has moved down  the frequency scale until they (1/2x or 1/3x) are showing between the real 1x and 2x and 3x  harmonics of running speed. Even in this data, the 1/2x was already down to the 6th, 7th, 8th, 9th, etc. harmonic of 1x running speed. Earlier detection might have been possible with Peakvue data taken in the other two directions (Vertical and Axial). Or by simply looking at the velocity data displayed in Gs without the vertical "autoscale" function activated. but the actual situation is that a worn shaft under bearing about 0.2 mm I think I understand what you are saying about "the actual situation", but correct me if I am wrong. I think you mean that at this point in time, when this data was taken, there was already 0.2 mm wear in the shaft. If this is what you are saying (or implying), my guess is that there has been a signal similar to this one which we are discussing and may have possibly been overlooked due to the low amplitudes, which (the amplitudes) may have been even somewhat lower than this data.

the second question what is the reference that i should call it a looseness the "reference" I would use is the fact that there should not be a 1/2x in an "all around" good bearing as the G level is low, IMO, the Gs are possibly going to be "low" since there is a low or small amount of energy being transmitted to the bearing housing, since it (the (energy or amplitude) has to travel from the shaft and inner ring slippage, through the inner ring, through the balls or rollers, through the outer race and through the bearing housing before it gets to the transducer. There is a theory that acceleration does not travel very far from the source. and why did you choosed this band from 6000 to 11000 cpm at a previous comment  I chose this particular band because it appeared to be the area with the most energy within the Fmax data available at the time and even in velocity data there appeared to be something "strange" there. There is nothing magical about choosing this particular area(6K to 11K). There also appeared to be some of the same type data nearly the full length of your spectrum. I guess the "magic" might be understanding that Gs display better in the higher frequency range and not always using the vertical "autoscale" function, and not relying on velocity data too much in the higher frequency area. Sometimes many important things can not be seen and are sometimes never seen until they progress down into the lower frequency range, if using only velocity data. I am not saying you are doing this. Just stating what I do.

This seems to be part of the theory of the Peakvue data, IMO. The wanting to know, " What is happening in the higher frequency range." And this sort of falls back to our short discussion on directional positioning for taking Peakvue data.

Remember that all I have said here is solely my own opinion based on experience and I could be totally wrong.

Thanks and Have a Great Day,

Ralph

thanks 

 

Ralph Stewart posted:
Hossam 2 posted:

Loose on Shaft

 

yes the are equally spaced too by the running speed and the difference between the marked A and Red X at is about half the running speed but my question is 

how could this be a early warning, By early warning, I simply mean, a signal early enough to possibly schedule a shutdown rather than running until it became obvious that there was a worsening looseness problem, when the 1/2x or even 1/3x energy has moved down  the frequency scale until they (1/2x or 1/3x) are showing between the real 1x and 2x and 3x  harmonics of running speed. Even in this data, the 1/2x was already down to the 6th, 7th, 8th, 9th, etc.

ok,,we can tell a question here.. when should i call that a bearing change is a must action.at which point this is what i meant when 1/2x is visible this is what i understood am i right ?

harmonic of 1x running speed. Earlier detection might have been possible with Peakvue data taken in the other two directions (Vertical and Axial). Or by simply looking at the velocity data displayed in Gs without the vertical "autoscale" function activated. but the actual situation is that a worn shaft under bearing about 0.2 mm I think I understand what you are saying about "the actual situation", but correct me if I am wrong. I think you mean that at this point in time, when this data was taken, there was already 0.2 mm wear in the shaft.

yes this is exactly what i meant as the measurement was conducted and wear was already exhibited in shaft

If this is what you are saying (or implying), my guess is that there has been a signal similar to this one which we are discussing and may have possibly been overlooked due to the low amplitudes, which (the amplitudes) may have been even somewhat lower than this data.

the second question what is the reference that i should call it a looseness the "reference" I would use is the fact that there should not be a 1/2x in an "all around" good bearing as the G level is low, IMO, the Gs are possibly going to be "low" since there is a low or small amount of energy being transmitted to the bearing housing, since it (the (energy or amplitude) has to travel from the shaft and inner ring slippage, through the inner ring, through the balls or rollers, through the outer race and through the bearing housing before it gets to the transducer. There is a theory that acceleration does not travel very far from the source. and why did you choosed this band from 6000 to 11000 cpm at a previous comment  I chose this particular band because it appeared to be the area with the most energy within the Fmax data available at the time and even in velocity data there appeared to be something "strange" there. There is nothing magical about choosing this particular area(6K to 11K). There also appeared to be some of the same type data nearly the full length of your spectrum. I guess the "magic" might be understanding that Gs display better in the higher frequency range and not always using the vertical "autoscale" function, and not relying on velocity data too much in the higher frequency area. Sometimes many important things can not be seen and are sometimes never seen until they progress down into the lower frequency range, if using only velocity data. I am not saying you are doing this. Just stating what I do.

i always rely on peakVue data depending on 1Khz and 2Khz filters but what was mentioned that the peakvue is always tricky at the bearings which is submerged in oil or been splashed by oil rather than the bearings running in grease.

This seems to be part of the theory of the Peakvue data, IMO. The wanting to know, " What is happening in the higher frequency range." And this sort of falls back to our short discussion on directional positioning for taking Peakvue data.

i will try the peakvue in axial direction today in bearings i suspect them.

i also use the Autocorrelate function at the timewave form to show the periodic impact this is also very effective i have many cases detected by the auto correlation.

Remember that all I have said here is solely my own opinion based on experience and I could be totally wrong.

Thanks and Have a Great Day,

Ralph

thanks 

 

 

One suggestion for next time, If you create a database and copy the equipment and data of concerned to that copy, you can then zip that database and post it and every one that downloads it can then analyze. See sample attachment.  It shows a crack inner race on a brand new (dropped during installation press roll) on a paper Machine.

Attachments

Add Reply

×
×
×
×