Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
kawanabadi

Member
Registered:
Posts: 6
Reply with quote  #1 
Dear All,

this is the new topic, as already submitted on thread;

http://www.machineryanalysis.org/post?id=7791266&pid=1293945564#post1293945564



I have recently hydraulic pump damage.  The team already take vibration data a week before the day of accident.

Theses guys are fail to prevent recognize early failure may exist.

By this available data, should it say something or we have to adjust line resolution in order to catch the failure before getting worse.?

Thank you,
Iwan




 
Attached Files
pdf vibration_prior_to_breakdown.pdf (342.00 KB, 30 views)

kawanabadi

Member
Registered:
Posts: 6
Reply with quote  #2 
Dear Ralph, many thanks for your comment,


Ralph Steward, has posting comment and i just replay in blue font.


The data pictures are, at least for me, difficult to read clearly. [smile] >> new file already posted, better clarity.

Your first question "By this available data, should it say something"

I am not sure how the analysis is performed, but it should say something.


Your second question: "we have to adjust line resolution in order to catch the failure before getting worse.?".

It looks like you could use more resolution in the data.


Back to the way the analysis is performed...........

(1) Do they analyze strictly on trend values or do they look at most every waveform and spectrum? >> General speaking; they just relied on trending value only, waveform and spectrum, is performed by my after incident.

(2) In the data your are showing, why or how is the set mark on the 7500 cpm peak and its sidebands chosen? By the software or by the analyst? by the software

(3) Is the 7500 cpm peak really 8.33x or is it 8x? i am not sure, engine has 1800 normal (nominal) rotating speed, where as measured on idle less than 1800 rpm. hydraulic pump have 9 piston, means 9x1800 = 16200 piston phase freq.

(4) Why are the sideband markers measuring such low amplitude peaks rather than the larger ones left and right of the set mark point? this is software, 

(5) Is it possibly that the larger peaks are running speed sidebands off the 7500 cpm peak or actual 1x running speed harmonic peaks, or something else's harmonics? i am not sure, but fill find out

(6) Most of the peaks between 0 and 15000 cpms appear to be modulated by something, in addition to, either being harmonics of possibly running speed, and/or also themselves modulating something. This is where the more resolution would be good, to determine what is riding on these frequencies.>> i just post better clarity. If you provide email, i will send to you higher resolution (original) around 3MB.

Thank you and best regards,
Iwan


Ralph Stewart

Avatar / Picture

Sr. Member
Registered:
Posts: 235
Reply with quote  #3 
Iwan,

My email address is f86sabra at comcast dot net.

" Is the 7500 cpm peak really 8.33x or is it 8x? i am not sure, engine has 1800 normal (nominal) rotating speed, where as measured on idle less than 1800 rpm. hydraulic pump have 9 piston, means 9x1800 = 16200 piston phase freq."

So we are not sure what the 1x is on the spectrum? It shows 900 cpm above the first major frequency. Was this data taken during the "idle" time?

If the 900 cpm is running speed, why does the software say the 9th peak is at 8.33x?

Running speed on this spectrum must have been 833 rpm instead of 900, huh?

833cpm times 9 = 7497 cpm. (8.33x/9.00x = 0.925556 times 900 rpm = 833 rpm), that is if all the peaks up to and including the 9th peak in the spectrum, are at 1x and 1x harmonics.

This seems like the software is assuming running speed is 900 rpm instead of possibly really being 833 rpm. IMO. [confused]

When you say "this is software, does this mean all the analysis is depend on what the software "calculates" rather than "tweaking" to locate the real frequencies positions?

If the analysts are not locating the real running speed before analyzing, that is basically, IMO, a bad thing to do. [smile]

Thanks and Have a Great Day,
Ralph



__________________
Ralph Stewart
AlertAnalytical.com
kawanabadi

Member
Registered:
Posts: 6
Reply with quote  #4 
Dear Ralph,

Yes, the software seems assume, the idle speed of engine as 900 cpm, 
But based on interview with vibration guys, he believed the rpm around 1600 cpm.

I agree with you, this is information is inadequate to perform vibration analysis.


An way, the pump replacement already performed. I just ask them to perform re-measurement as basis data.

Re-measurement along with actual rpm recording.

My point is do we can prevent the same incident and the initial failure could be detect if we perform better vibration measurement and analysis as well.

I will share with you if the result is available.

Thank you and regards,
Iwan

Ralph Stewart

Avatar / Picture

Sr. Member
Registered:
Posts: 235
Reply with quote  #5 
Quote:
Dear Ralph,

Black is your quote. Red is my reply. [smile]

Yes, the software seems assume, the idle speed of engine as 900 cpm, Looks like it is the rpm (833)
But based on interview with vibration guys, he believed the rpm around 1600 cpm. WOW! That frequency does not show in the data as 1x, based on the data you posted, does it?

Unless the rpm was 1666 and the 833 was 1/2 of the running speed. This would make the tall peak that appears to be 9x actually be 4.5x. [confused]


I agree with you, this is information is inadequate to perform vibration analysis.


An way, the pump replacement already performed. I just ask them to perform re-measurement as basis data.

Re-measurement along with actual rpm recording.

My point is do we can prevent the same incident and the initial failure could be detect if we perform better vibration measurement and analysis as well. Quite possibly could have been detected, IMO.

I would never depend on the trend value as the complete and only source of a full analysis on any piece of equipment.

I am not sure what the "normal" good condition data was showing in the past, and this failure could have been a "sudden" unexpected failure. I am not saying it was, but anything is possible.

Are you doing a root cause analysis on the old pump, in an effort to determine what actually happened?


I will share with you if the result is available. Thanks.

Ralph


Thank you and regards,
Iwan

__________________
Ralph Stewart
AlertAnalytical.com
kawanabadi

Member
Registered:
Posts: 6
Reply with quote  #6 
Dear Ralph,
 
Attached is more complete information related with vibration measurement result.
 
in acceleration measurement, it seem there is a trend increase of magnitude and looseness in high frequency.
Does it mean something? 
 
 
Best Regards,
Iwan
 

 
Attached Files
pdf More_Information_vibration_measurement_prior_to_breakdown.pdf (1.62 MB, 21 views)

Edwin

Avatar / Picture

Sr. Member
Registered:
Posts: 131
Reply with quote  #7 
Iwan,
Can I suggest something? If you want to monitor wear and damage with acceleration, make sure you measure the resonance frequency of the moving parts. Normaly that is around 3000 to 3500 Hz. That is why I measure all acceleration up to 4000Hz -240000 cpm (higher gives useless data with magnetic mount). In your measurement up to 2000Hz there is an increase in level and developement in the spectra, but that is only of softer parts, or of the parts getting softer when they are damaged allready. I think if you measured higher frequencies before, you would have discovered problems with lubrication earlier and maybe a simple oil change had avoided this damage. At least the damage would have been predicted by a qualified analyst.

__________________
= Cost Effective Asset Management =
Danny Harvey

Sr. Member / Moderator / Supporter
Registered:
Posts: 1,403
Reply with quote  #8 
"Yes, the software seems assume, the idle speed of engine as 900 cpm"

You should correct that in the database. If it's variable speed, you should verify the operating speed at the time of data collection.

Correct shaft speed is the single most important piece of information in vibration analysis. Without it, it is guess work.

I believe that I would need to spend some time with the pump and engine trying to find settings that might allow me to separate the noises from the diesel engine from those of the hydraulic pump.

More information regarding the drive arrangement, construction and operation of the system would be required.

Do you have a photo of the engine, pump and piping arrangements?

Are you filtering and analyzing the oil and trending that?
RGf

Sr. Member
Registered:
Posts: 78
Reply with quote  #9 
I'm going to assume that this is a 9 piston pump similar to the OilGear machines I have here on my steam turbine hydraulic systems. That being said in conversations with the tech guys at OilGear they really don't have any hard vibration criteria but they agreed that monitoring the 9X running speed peak amplitude vs. the 1X amplitude will give you a good indication of wear either on the cam or the pistons. In my experience this is valid as I removed a pump form service when the 9X area of interest increased to over 50% of the 1X and a tear down revealed a moderate amount of wear on the vitals. After rebuild by OilGear the ratio returned to the "normal" range. As we have installed accelerometers as part of out continuous monitoring system I moved the pump horizontal sensor to a location on the pump body inline with the pump discharge to better detect the 9x peaks and deemphasize the bearing frequencies. I have also found that acoustic monitoring is also a good way to determine machine health. Mine really scream when they get worn.

Attached Images
jpeg Oil_Gear_2.jpg (201.60 KB, 25 views)
jpeg Oil_Gear1.jpg (104.45 KB, 26 views)

Ralph Stewart

Avatar / Picture

Sr. Member
Registered:
Posts: 235
Reply with quote  #10 
kawanabadi,

What did the bearings look like when the root cause analysis was performed?

Thanks,
Ralph

__________________
Ralph Stewart
AlertAnalytical.com
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.