Stan said "I do not believe condition monitoring with portable data collectors has been "one upped" by permanent systems, I believe we've:

- been "one upped" by NOT demonstrating our ongoing value,

- been complacent in continuing to purchase and use instruments and software from companies who have lost sight of who made their companies successful,

- worked too hard finding minute details, and in retrospect, have lost sight of the big picture,

- developed condition monitoring into it's own little niche market, instead of making it a direct part of maintenance."

All I can say is AMEN!

To be sure, automation and computerization will continue their creep into all aspects of our lives.

My crystal ball looks more like Ali's. Gradual evolution but no dramatic changes in 10 years+. Just my gut feel.

In addition to the upfront costs of on-line monitoring, let's not forget that in addition to the analysts who still reveiew that data (albeit much more efficiently), there are people who have to troubleshoot and maintain the computers. And surprise!... every few years your hardware or software is going to obsolete... more $. At least that's the way it seems with our Bently stuff. (my impression based on my coworker's comments, although I don't work with that system). Perhaps if it's just a few analog overalls per machine on the plant computer, then its much more manageable, but also less useful.

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

Something that gets a little lost in this discussion, is there are many, many small companies and plants that understand the need for predictive maintenance (and related technologies) but who either can't afford to do it in-house, or don't want to. These companies/plants don't go through the typical bust/boom cycles that large corporations do. When things get a little slow, they don't throw the baby out with the bathwater. If you do a good job for them, prove your worth, and don't get greedy, you are there until you decide to quit. I've been at one such plant, a closely held corporation, monthly for 11 years now. Another customer has been bi-monthly for 10 years. There are many, many such facilities that are underserved by the "legacy companies", or any PdM company/consultant. I am proof positive that you can make a good living serving such clients, just doing basic vibration work (PdM, trouble-shooting, alignment, balancing). That situation is not going to change.

That's perhaps a little different perspective on "the market", but it's out there if you know where to look.

BTW, I've never had a customer ask me for information to allow them to do a "cost benefit analysis"... if any of them have ever done it, they've not told me anything about it. Believe it or not, there are a lot of corporations that don't measure every aspect of their business. Sometimes they just "know" there is value in an activity.

Who here has ever done a formal cost benefit analysis for periodically changing the oil in their car? Anyone? Then why do you change the oil every 3000 miles? Because you 'know' it's a good idea. But how do you know?

Rusty,
Yes there is a market out there, but from my travels around the world, I would guesstimate that for every datacollector in the hands of a competent such as yourself, there are at least 20 being used by people who really don't have a clue about either the machines they are monitoring or machine vibration, and that's where the problem lies - these setups bring no value to industry.

I would suggest a similar (or higher) competent/inept owner ratio for online permanently installed systems.

Before the inevitable protests begin, just remember that I am talking about worldwide usage (Africa Asia and many parts of Europe) not just the US

Is it true that not many understand the subject of vibration because it's still subjective and not well defined yet? The diagnostics have many possibilities or causes of why the vibration occurs that need years of learning and experience.

Josh, it is not a matter of vibration analysis being subjective, and it is not a matter of vibration analysis being insufficiently defined. We have been studying it for 100 years.

The problem is that any real vibrating mechanical system is complex, and so problems don't always fit into simple neat formulas (or wall chart analysis).

Rusty, I think the cost benefit analysis on a oil change is done subconsiously. Everyone has a pretty good idea of the cost of a new engine, and compared with $100 per year in oil changes, plus the intrinsic cost of being carless it is well worth it. That doesn't mean the cost benefit analysis was not done.

When the vibe tech is finding crucial stuff every month, there is not much need to do a cost analysis either, it is pretty easy to calculate it in your head. Just beware when the plant gets pretty reliable or if the only calls you are making are early stage bearing failures, that will carry on for months.

quote:

Then why do you change the oil every 3000 miles? Because you 'know' it's a good idea. But how do you know?

So, you don't go the full 15,000 miles on the BMW

I go 10,000 on my VW diesel (synthetic oil)...

quote:

Originally posted by Steve Schultheis:
I go 10,000 on my VW diesel (synthetic oil)...

Me too, but my oil changes end up over $100 a year.

Steve,

Your comment sounds like many complex mechanical systems are not fully studied and defined even after 100 years.

I think it's very interesting what a cardiologist charges to do and then interpret an EKG... and an EKG is soooooo simple compared to most all machines that we analyze. And of course unlike machinery, nearly everyone has the same EKG.

Rusty. could you email me. Vibeguy2004@yahoo.com

Sorry Josh,

If that is what you got from my comment, that is not what I meant. We understand how things work very well. The problem is that we make measurements on very complex systems, but we never gather enough data to fully characterize those systems. We try to infer the health of a machine based on a few measurements, and for the most part that works fine, but when it is something complex, that approach breaks down. We need more data but we don't have it, and it is hard to get it.

quote:

Originally posted by Josh:
Is it true that not many understand the subject of vibration because it's still subjective and not well defined yet? The diagnostics have many possibilities or causes of why the vibration occurs that need years of learning and experience.

I am pretty much in agreement with that, from the standpoint that a canned expert program that would attempt to diagnose machines using cookie-cutter rules would be very ineffective. The analyst has numerous other things that add to the diagnosis. The history of the equipment. A comment from the maintenance guy or the operator. A smell or observation noted during walk-around. Some but not all of these can be retained when someone is in an office monitoring machines.

I also agree heartily with Rusty that decisions are often made primarily on gut feel without a lot of economic analysis, simply because there often is not a tremendous amount of hard data to quantity the relationship between maintenance and monitoring strategies and future equipment performance/costs.

quote:

I also agree heartily with Rusty that decisions are often made primarily on gut feel without a lot of economic analysis, simply because there often is not a tremendous amount of hard data to quantity the relationship between maintenance and monitoring strategies and future equipment performance/costs.

Pete, what I really meant to imply is that decisions are often made without hard data because you don't actually need hard data, just as Steve confirms in the case of changing the oil in your car. Sometimes the right thing to do is so obvious, you don't need to analyze it to death. The best companies, and the best managers, don't waste time analyzing the obvious. They decide quickly and move on, and if their decision proves to be wrong, they change course just as quickly, and move on once more.

As an example, when we encounter a machine with a high 1x vibration (perhaps confused by other metrics) we can analyze it to death, or we can simply align it and then attempt to balance it (assuming this can be done quickly and conveniently). If aligning and balancing it doesn't work, we've at least eliminated 2 of the most probable causes. Often we can actually correct the problem in less time than it would take to do a rigorous analyses. You wouldn't approach every situation like this, but frequently it's the best approach.

Problem solving and decision making are 2 main soft skills being quite rigorously taught in career development.

Oil change for personal vehicles is easy to decide because it's cheap and nobody will question any consequences after that, not even the wife I guess.

However, oil change for many vehicles eg for military, transportation companies, etc, maybe the maintenance manager will need to think further about lubrication strategy in general such as to do some simple oil tests.

So what will we say if someboby says oil analysis is an overkill exercise because of the simple thought that oil analysis is about seeing the colour change to recommend oil change? For personal vehicles, maybe yes but not for industrial plants I guess.

For the 1x vibration case, cause elimination is quite common in vibration analysis from my experience. It's the old police detective investigation technique! For small equipment, perhaps we want to plunge into the problem straight away but maybe not for big turbomachineries for which we would want to pinpoint the cause rather accurately.

Anyway, there are many styles of problem solving and decision making. Check this out if you wish:
http://en.wikipedia.org/wiki/Myers-Briggs_Type_Indicator

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

Steve,

You said we don't have enough data and measurement points for complex systems. Why is this? I thought the first thing to do when preparing the vibration monitoring strategy is to come up with appropriate measurement points. I am thinking about complex systems in normal industrial plants, which I think have sufficient measurement points. Pls give examples of complex systems without enough measurement points that you mentioned above.

Ok, I am working with a compressor right now that has high 1X vibration (fairly nice ellipse on the orbit) on the coupling end of the machine. Levels on the other end are not bad. The vibration runs along at normal levels, then suddenly jumps to danger level and remains at danger level until the machine is tripped. Upon restart the level is back to normal. After a few weeks running, it suddenly jumps back up to danger level. The sensors available are Xy probes at both ends, plus thrust.

So what is the issue, there are many things that can cause high 1X vibration:

Unbalance,
misalignment
rub
runout
rotor bow
Coupling problem
bent shaft
resonance
Loose probes
Probe resonant to 1X
Impeller problem
Bearing clearance or load issues

I can rule out some of these, but with out more data, and perhaps without overhauling this machine, I will not know for sure. I can make some intelligent guesses based on this limited data but I will not know.

There are lots of problems where a couple points on the bearings is insufficient. For instance a support resonance problem might need a ODS analysis to resolve it. Erik Concha has a post elsewhere on this board where there is a casing vibration that seems to be from the attached piping. That is going to require a lot of data.

For simple problems, like unbalance, or rolling bearing failures, a couple measurement points per bearing is fine. But that will not always be enough for a more complex problem.

So Josh, why don't you make your position clear to me. You seem to be arguing both sides.

Is vibration analysis an art? or a hard science?

I would say the really good analyst uses experience and intuition to take limited data and make a solid conclusion as to the nature of the malfunction. It is not pure calculation.

quote:

I would say the really good analyst uses experience and intuition to take limited data and make a solid conclusion as to the nature of the malfunction. It is not pure calculation.

I concur with Steve's assessment. Further, the reason we often don't have "complete data" is economics. Unless you are a corporate employee with lots of time to study a problem, or -- if a consultant -- have a client with very deep pockets, you are limited as to how much "study" can be done of a problem. If a customer calls me in and says "this fan has high vibration", he's thinking 8 hours (max) + travel to "fix" the problem. He's not going to give me 2 or 3 days + expenses. So based on experience and intuition, you "attack" the problem in the way you think will fix it most quickly. Usually (80%) you are right. Sometimes (20%) it doesn't go smoothly and it takes a little longer and they are OK with that because you busted your butt (you did, didn't you?) to get it done quickly. But if you stood around, or just took your time collecting and analyzing data, they may not be so understanding.

Of course I'm not talking about complex problems with multi-million dollar machines... that's a whole different scenario, but less common to most of us.