It is a truism that a badly run oil analysis program does more harm than good. On more than one occasion, I've recommended that a client terminate their current oil analysis activities until they are in a better position to make the technology actually work for them, rather than impede them. Trying to force a technology into a situation which it's not suited for is like trying to force a square peg into a round hole; it's a waste of time, money and resources. If you are receiving 80 percent of your samples back from the lab marked as critical, you probably need to take a long, hard look at what you are doing.
So what are the elements of an effective oil analysis program?
The first step is to ensure that the sample extraction is performed correctly. I've discussed this in a previous column, so I won't say any more on this except to reiterate that an oil analysis program is only as good as the oil sampling program.
Sampling at the correct frequency is vitally important, and from my experience, oil analyses are usually carried out far less frequently than is ideal. Oil analysis is like an insurance policy - if you don't pay the premiums you won't get the payouts. There are a couple of different models for determining sampling frequency - these will be the subject of a future column. A rough rule of thumb is that critical equipment should be sampled at least once a month. Noncritical equipment, for which oil analysis is warranted, should be sampled no less frequently than once every three months. Unless you have specific goals in mind, sampling only once every six months or only once a year is probably a waste of resources - maybe a different condition-monitoring technology would be more effective in these instances.
Good Interpretation Skills
This is also a subject I've written about in my previous columns. Interpretation is a team effort among the laboratory staff, possibly external consultants, and the people dealing intimately with the machine on a day-to-day basis. A good interpretation can be made only if information is freely available and shared. Plant personnel have a more intimate knowledge of the assets than any diagnostician off-site, so it is important to encourage their interpretation skills and seek their input.
Once an interpretation (or diagnosis) is made, it's time to develop a solution. The diagnosis is the problem statement, and the prognosis (or solution) needs to be created. This is the time to gather data that may or may not have been available at the time of the analysis, but can be used to help identify the solution. The ability to think laterally can be of great benefit here. Use the experiences and knowledge of the plant personnel, particularly of those who have been around for some time. A similar situation might have happened 30 years ago, and unless you ask the right people the right questions, you could find yourself reinventing the wheel. Include other condition-monitoring technologies in determining your solution.
Once a diagnosis has been made and a maintenance intervention completed, keep records of what was done. If the next analysis shows that the problem persists, you'll know what not to do the next time. Share this information with the laboratory - it will help guide them on their next prognosis.
If you've detected a problem and performed some maintenance work to correct it, it's important to determine whether your work was successful. The only way you will know this is with a rapid follow-up sample, or a series of rapid follow-up samples. Don't stick to the normal sample frequency here; probably a good place to start is by halving it. Also, treat every set of results as critical until you are absolutely sure the problem has been taken care of.
Where to Start
Assuming that you do not have an oil analysis program, or the existing oil analysis program is in serious need of an overhaul, where would you start in trying to improve it? There are two options to consider, and the one chosen will likely be a function of the way responsibilities are divided within the plant. The first option is to choose the most critical machines, and implement the oil analysis program fully and correctly on these machines. How many machines you choose will obviously be a function of resources, but I suggest choosing five or ten machines and getting their program as near to perfect as possible. Once this has been done, choose another five or ten, and carry on in this manner. The second option is to choose an entire section of the plant, maybe a section plagued with frequent problems, or a newly commissioned area which you want to keep on the straight and narrow right from the start. Whichever approach you choose, make sure your goals are clearly stated and that you have a realistic chance of succeeding with the resources available.