quote:

Why not measure the bearing centerline position to determine this?

I suspect that most equipment running with plain bearings is not equipped with prox probes, nor will it ever be because of cost. There are many, many facilities that simply don't have the money for "proper" monitoring of plain bearings, nor the personnel to do so. So many of us have to make do with what we have.

Duncan points out that slot-pass vibration can be highly variable, and indeed this is true, especially on motors with rolling-element bearings. That's why, when you DO see a motor with plain bearings where the slot pass vibration trends steadily upward, it can be significant. As pointed out, "bad" babbit bearings can be very difficult to detect, so I look for anything that is trendable.

If you are taking bearing cap or motor housing readings on such a motor (typical of walk-around programs), it only takes a few seconds to take an extra reading, or even just to set up a parameter band that looks at slot pass frequency. Anything is better than nothing.

CMT, A lot can be learned from the failed bearings, especially if they weren't completely destroyed. Perhaps you can answer the question posed to you about how the failure occurred if the damage is examined. If you have the bearings, post some pictures. Thanks.

If the maintenance decision has been made that it is not worth instrumenting a machine properly to monitor and to understand and to prevent failures, then that is the way it is. Failures should be accepted as part of doing business, and this should have been part of the decision not to instrument.

The question as to cost value involves not just the instrumentation vs. maintenance costs/savings. The impact of downtime for un-spared or partially spared equipment has to be assessed.

If you can't catch a problem because of this, this is not a failure of the PM program. Rather it results from an economic decision. If the decision not to instrument did not come from a proper analysis, then there is another job for a consultant to sell, assessing the equipment instrumentation/maintenance requirements.

Anything can be difficult to detect without the proper measurements. Fluid film bearing issues are easy to detect given the proper measurements.

I used to do some measurements at a chemical plant (long ago) where I had to power up the proximity probes and take the measurements locally. It is much better when they are wired to a monitor in an air conditioned enviornment. Even if the machine is not continuously monitored, having the correct instrumentation is essential.

That's a good comparision of alternative approaches that I hadn't thought of (but makes sense)

1 – no window, pretrigger, allow to die within window

or

2 – random impacting ... expect rindows to go beyond FFT window and handle that with a window function.

If we use a window, wouldn't we rather select a window optimized for frequency accuracy (like Kaiser Bessel on the Entek Datapak). The Hanning as I understand it is a compromise between magnitude accuracy and frequency accuracy, but magnitude seems not terribly important on a bump test.

I realize this probably is a very very very small error compared to other larger sources of error, but just wanted to ask if any reason not to select the Kaiser-Bessel isntead?

Bill - I think with what I know of your background, you are assuming that the only way to handle this problem is with proximity probes which is the answer that most practitioners would give. However, there is a substantial body of work that says this can be handled with accelerometer transducers and current common data collection hardware that doesn't require expensive equipment modification to install proximity probes.

This message has been edited. Last edited by: Duncan Carter,

Pete – You may have posted in the wrong place.

For frequency accuracy the uniform (rectangular) window is good, but the signal doesn't die at the ends of the samples. The Kaiser Bessel window depends upon parameters; I don't know if the analyzer you mention allows you to set those parameters. You don't (always) kill off entirely the ends of the sample. The Kaiser window at one extreme would be a rectangular window and the other a Gaussian window – the Gauss window might work well, too.

Suitably picked a Kaiser Bessel window should work fine as would a number of window functions, but the analyzer must have that window. The Hann window is available most of the time. With a tool like Matlab (trademarked) or Scilab (trademarked) one can try a variety of windows, but restricted to a particular analyzer one has limitations.


If by my background, you mean that it includes extensive work with fluid film bearings you would be correct, Duncan.

Bill - yes, that is what i meant.

Thanks William. You're right I posted in the wrong place.

I was aware Kaiser Bessel is a very versatile filter which is why I mentioned the Entek version that is tuned for frequency accuracy.

Thinking some more, I think Hanning window in conjunction with a small bin width would be fine for this purpose, given all the other errors and the fact that the damped peak may be somewhat wide to begin with. It's not worth spending any more energy on that aspect.

Usually, when talking about accuracy and windows one means for a periodic force/response. Impacts are different.

And don't laugh! But just to track shaft position and installing eddy current probes --- externally mount 1 probe in the vertical and monitor the DC .... but a reference has to be known from reliable info or lift check.

So if the decision is not to monitor rotor dynamics, a cheaper alternative to monitor shaft position can be; mount a probe externally and periodically plug-in and check position - one probe per brg and one driver - cost is not too high for most.

quote:

Failures should be accepted as part of doing business

So, Bill, does this mean that if I am asked to monitor plain-bearing machines that I'm just supposed to tell my customer that since they didn't buy it with the proper instrumentation that "failures" are inevitable, and sorry I can't help you... ??

If the customer has not considered the options, then it is about time to perform an assessment. Presumably, evaluating the instrumentation/value is part of what an outside consultant could/should bring to the table.

Fortunately with fluid film bearings, they should not fail if designed and operated properly, and this may have been evaluated during an assessment. However, the bearings don't know this – They really don't have much more inherent intelligence (without instrumentation) than rolling element bearings, and issues like alignment and lubrication can interfere with their infinite life design. On bigger machines with turning gear, this is a wear mechanism if there is no lift oil.

More cognitive activities like defining the maintenance strategy should have more value both to the customer and the vendor than activities such as data collection. At least, that has been my experience. Depending upon the maintenance strategy, a properly instrumented machine may have no instrumentation, but don't expect miracles from the walk around measurement system if such a program is part of the maintenance strategy.

Someone at sometime made a decision to use a walk around program or this wouldn't be a topic of discussion. How did they make that decision? What value did they expect, and should they expect? What value do they need? These are areas to add value to the customer. I am sure most of us have seen monitoring programs with areas of little or no value and questionable need. At the same time these programs may provide value in certain areas, but the condition monitoring program needs greater accuracy. One needs to assess the CM program as to how it is meeting the requirements and possibly reevaluate the needs periodically.

If the risks and cost of unpredicted failures are unacceptable to the customer, a case can be made to instrument the machine. RCFA's are likely to be more accurate and easier with instrumented equipment.

I've seen a number of outside contractors (some internal) get caught up in the data collection and analysis as being the job. Don't get me wrong, they do a good job collection data, but the goal is to enhance profitability, i.e. add value. I got to do a number of reliability assessments for rotating equipment last year, and the value provided by the outside CM programs was often questionable; although some produced good looking reports.

One question I asked was if they trusted and followed the recommendations of the provider. This produced some interesting answers. Not following recommendations was not that uncommon, and often resulted from following recommendations without positive results. I saw reports that indicated problems had been corrected when there was no action; it may be possible to have self healing machinery, but we don't buy it that way.

If in the judgment of CM provider that they do not have the instrumentation to walk on water for the customer, then the customer should understand this or made to understand this (It may take some of the tools used for a bump test – instrumented hammer not required.)