Is their any recommended guidelines regarding Vibration Analysis intervals?


Should motor size or load type such as blower, pump or gear box have any influance over determining if inspections should be performed Monthly, Quarterly or even based on Hour Meters?

Any assistance or links to resources would be appriciated.


Thanks!

As part of our work we often are asked this one and the answer is simply that it depends on how long the machine takes to break!!!

We call the technology of defining the measurement interval POTENTIAL FAILURE ANALYSIS.

In this methodology we identify what failures are likely to occur during the life of the machine. Then we identify how the machine will tell us that a failure mechanism is under way - by that I mean that a dry bearing will exhibit high frequency acceleration vibration or ultrasound in all directions whereas a simple misalignment would show an increase in low frequency synchronous velocity vibration mainly in the axial direction. In other words you have to really understand the failure mechanisms that are likely to occur BEFORE you design your vibration measurements and the intervals.

Once you have decided on the measurement types that you are going to take, you then have to make a value call and make an estimate on how long it will take from the first sign of deterioration that you can measure until the machine fails. I have one customer on an engine cylinder head line who has an internval of one week - based on experience. I have another customer who operates large slow speed boiler induced draught fans who takes readings once every 3 months with good reliability.

I normally recommend starting with an interval that is too tight and then critically examine your measurements that have identified failures. If the machine owners agree then it is a good idea to run one machine to "close to destruction" so that you have a reliable time interval from first indication to failure. If you can't get that (and usually we can't) then we have to look at the trends for as long as we have the machine under our control.

Once you have done this make sure you keep records. It ties in perfectly with ISO9000 as not only do you have an audit trail in case things go wrong (you can go back to the record and see how you can improve the potential failure analysis) but by this process of continual improvement you can verify that you are doing the best that can be done in a regular and repeatable manner.

The whole point of this is that when you take vibration measurements you should not simply be "doing vibration" - you should identify the potential failures and then design the readings around them. It works for all predictive maintenance technologies.

Hi. We are an electrical utility, and we have a regular vibration program. Our vibration tour interval was once every month for the last few years. This was based on the recommendation from the company that originally set up our vibration program. After taking readings this way for the last few years, we determined that the interval on most equipment could be extended. We are now taking readings every 2 months. On some of the critical equipment, we still take readings at greater frequency. My advice would be to start with a shorter interval and based on experience extend it slowly until you have reached an optimum schedule for both predictive maintenance and manpower considerations.

Rhonda Stock
Mechanical Technical Specialist

I have been in vibration monitoring at this mill for over 30 years and started before we had data collectors and computers. This is a question that our condition monitoring team struggles with. Our stated purpose as a condition monitoring team is to find problems or potiential problems early enough to plan, schedule and react to them to minimize any process downtime and maintenance costs. We view our the condition monitoring team as a screen made up of all the technologies (vibration, infrared, etc...), operators, mechanics and management. Each person is a thread in that screen. The more threads you have the tighter the the screen mesh. The tighter the screen mesh and the less likely a potential failure is likely ot pass through it and be missed. Each person that walks by a piece of equipment has uncalibrated instrumentation at their disposal (sight, sound, feel and smell). These persons can help fill the gap between vibration, infrared, etc... inspection intervals. Operators should be physically looking at the machinery they operated once during their shift, the oiler and preventive maintenance mechanic should be walking by that same machine during their routine inspections and report possible problems. The moral of this story is that high tech is great but due to limitations of technologies, time or money to apply them they will never catch all problems before failure occurs unless you have them online with a very high rate of inspection such as every 8 hour shift. There is always a point of deminishing returns.

Mr. Jacques,
RonFrend has given a comprehensive and useful guide.
For emphasis (only), I want a add a comment about equipment criticality, since some others have brought up this subject. Criticality, pre se, has nothing to do with the inspection interval; the only parameter that matters is the rate of degradation from incipiency to functional failure. The rate of degradation doe not depend on criticality, it only depends on the nature of failure and the physicsl(or chemical, biological etc.) process that causes it. So critical items dont need more frequent inspections, but items that degrade more quickly do. However, because critical items tend to have high consequences which we want to avoid, we tend to err on the safe side and take more frequent readings, to avoid situations where the rate of degradation itself is less predictable.

The general principle is, when you dont have any degradation data, start measuring at a relatively high frequency, say, once a week. Plot the readings. Go through one or two complete cycles where you have actually 'caught' each degradation curve, and get an idea of the time it takes to degrade. Divide this time by 3, and you have an initial reading interval for the next cycle. If in the next cycle, you get a different degradation time, stay at this interval. If you get the same timing as before, you can increase the interval by 50% -this is the max. you can go to without the chance that you will fail to catch some failures. On failure with high consequences, stay at the original test intervals even if the degradation time remains as before.

V.Narayan.

I do appriciate your feedback. Thank you