Dear All
Hi!

I have not any noticeable resource about how to set band alarm for spectrum, would you please help me in this subject?

Also What is your opinion about how to set alarm and danger criteria for envelope curve?

Regards,F

The General Motors Vibration Standard was passed around via this board a couple years ago. Evidently GM freely shares this standard. I recall it employs band alarms. I'm not a band alarm guy, however, so I can't speak to its suitability for that purpose.

I believe the posting for the GM Standard was prior to the recent overhaul of this discussion board. I doubt if a search will find it. I will look for it once I return to work next Tuesday. If somebody else has it...cool.

Farahani,

This is a copy of the GM vibration spec.

Ken Culverson

GM_Vibration_Spec_V1.0a-1999.pdf (894 Kb, 125 downloads) GM Vibration Spec V1.0-1999

Dear Dai Wei

Thank you.It is a good source and specially the "aproach" used by it is a clue to extend it for any case.

Let continue this discussion about Envelope Alarm Setting. Some sources suggest a 8dB increase for alarm and 20dB for danger. But I in doubt that if it is applicable and reasonable for all cases or not?

Regards,F

Have you checked out the Technical Associates' Proven Method alarm bands? I find them a good starting point & then use statistical analysis for modification.

Regards,

Aditya

Dear Aditya

Hi!
No, I haven't seen TAC alarm bands. I appreciate you if you explain about it.
Regards,F

This message has been edited. Last edited by: Farahani,

There is a drawback in using "standards" in setting up alarms. Let's say for example one uses API 613 to set up alarms for a gearbox. Working through the calculation the allowable vibration is 1 mil pp so one sets an alarm 50% higher at 1.5 mils. The "example" gearbox runs at 0.4 mil pp but one day suddenly jumps to 0.8 mil pp. If you rely to much on the alarm setting this might not be noticed. You could have a tripling of amplitude that would go unnoticed!

There is a school of thought that recommends setting alarms at say 50% above operating point, in the above example at 0.6 mil pp. This would catch the sudden doubling of amplitude. Yes, nuisance alarms might also be the result and you might have to tweak settings ("seat of the pants" method or as someone suggested statistically). And "yes" again if you have multiple machines of the same type you might have different alarm leves for each machine, but again refinement can often come up with a single setting that is reasonable.

John

quote:

There is a school of thought that recommends setting alarms at say 50% above operating point

Dear John
I agree with you that we should not only rely to standards such ISO 10816, VDI 2056,2059 ,... . I believe that they are useful for example for first setup of a vibration analysis program. But as you say, after some month data collection we have a good data base and we can establish alarms based on statistical methods.
Now, I have some questions:
1- Do you have any justification for using 50% increase as a base? Is it usable for Bearing Condition units?

2- I have some books and references about statistical analysis itself. But would you please introduce for me some resourses about how to use statistical analysis for vibration alarm setting?

Faithfully.
Farahani

This message has been edited. Last edited by: Farahani,

Dear all

Is there anyone to explain the basis of statistical method for alarm setting, or to introduce a good resource for this?

Regards,F

There is an article about alarm settings but for oil analysis if it could help you for vibration analysis:

http://www.noria.com/learning_center/category_article.a...oup=WearDebris[/URL]