Nine Reasons Why Oil Analysis Programs Fail

Ashley Mayer

Many oil analysis programs are failures. Maybe not total failures, but they are almost certainly not delivering the results they could be and should be. Here are some of the most common reasons why oil analysis programs do not live up to their expectations.

1) Poor Analysis Strategies
A commonly beholden oil analysis strategy is to take samples only when the machine is suspected of having a problem, or when some other condition monitoring (CM) technique, such as vibration, triggers an exception. Oil analysis, like most other CM tools, is a trending game. If the only samples that are ever analyzed are problems, then without a background of the known-normal situation, the diagnosis is likely to be unreliable. So, don’t only take samples when problems are suspected; make sure to sample regularly and frequently.

Comparison of vibration and oil analysis
This pie chart provides a comparison of vibration and
oil analysis and initial problem detection.

2) Analysis Intervals are too Infrequent
Another common pitfall is performing analysis too infrequently. This is often for cost reasons, but this is usually a short-sighted justification. Just look at the cost of an oil analysis program using correct analysis frequencies and compare it to the potential savings. It will quickly become obvious that taking shortcuts here is not justified.

If cost is still an issue, then consider either reducing the test scope on alternate samples or perhaps consider some on-site oil testing. There are basic, cheap tests which are easily and inexpensively performed, yet are considerably powerful.

Tools exist to determine required analysis frequencies, and a good rule of thumb is: critical equipment on a monthly basis and less critical equipment on a three-monthly basis.

3) Poor Sampling Techniques
Oil analysis is like a sausage machine: What you get out is what you put in. An oil analysis program is only as good as the oil sampling program. Extracting the sample is probably the weakest link in the oil analysis chain, so it’s worth taking some time and effort to make sure that representative samples with a minimum of data disturbances (i.e. sample contamination) are being extracted.

There are basically three ways to take a sample: from the drain port; using a drop-tube and vacuum pump; or from a dedicated sample valve, possibly with a vacuum pump. The first two methods are the most likely to produce a contaminated sample, and it is fair to say that between them they account for more than 95 percent of samples extracted.

As a matter of priority, try to install dedicated sample valves in the correct locations. Ensure that sampling procedures are documented for consistency reasons, and make certain that the staff entrusted with this vital task is properly trained on how to accomplish it.

4) Delay in Getting the Samples to the Laboratory
Basically, the information contained in your sample becomes obsolete almost as soon as the sample has been taken. So waiting for days, or even weeks or months, to dispatch the samples to the laboratory is not going to help in limiting the degree of obsolescence. Don’t wait until you have a full box of samples before you send them off to the lab – and don’t try to skimp by using ground, rather than overnight, delivery service.

5) Delays in Getting Results Back
Delays in getting the results back can be caused by the laboratory or by results being received but not disseminated to the correct party internally. You should expect to have a report in your hand within 24 to 48 hours of the sample reaching the laboratory, assuming no specialized testing is necessary. If this is not the case, then check to see where the bottleneck is located. You might need to consider changing laboratories, or you might need to improve the speed of internal communication within your organization.

6) Poor Information Submitted to the Lab
As mentioned previously, oil analysis is a trend game. And if you cannot maintain the trend, then the usefulness of the technique is compromised. There should be a consistent means in place to describe machines and components during sample submission. If this is not done, then what occurs is known in the trade as a “split history”: two or more instances of the component exist in the database with samples split between them. It might be a good idea to ask your laboratory to supply you with a list of all machines and components that they have on record for you. Study it and ask them to merge any split histories you find.

The analogy of an oil sample being to a machine what a blood sample is to a human being is often used. Oil analysis is mechanical pathology. Extending the analogy a bit further, when you give a blood sample to your doctor, you also impart to him or her other useful information by talking to them. The same goes for machines. You should let your laboratory know about any unusual operating characteristics or recent maintenance carried out. It will help with the interpretation.

7) Lack of Correct Tests
When a sample is analyzed by the laboratory, typically a test slate appropriate to the type of component is applied. Some of these test slates are more comprehensive than others; but in almost all cases, there are many more tests available that could usefully be employed, particularly when abnormal operation is observed or suspected.

Unfortunately, correct test selection is something that you will likely need to drive yourself rather than relying on it just getting done automatically. Build up a good contact at your laboratory, someone with whom you can call and discuss results and, possibly, alternate testing. The other option is to make yourself familiar with all of the tests available so you know what to ask for and when to ask for it.

8) Poor Interpretation of the Tests
There are many reasons for poor interpretation of the results. They include: unfamiliarity with the equipment in question by the diagnostician; decisions made on the basis of poor or inaccurate supplied information; incorrect test slate selection; and conservative or vague interpretations, possibly fueled by litigation concerns. Like test slate selection, result interpretation is something for which you will have to be the final arbitrator.

Having someone in the organization who can pick up a report and interpret it in the context of the environment is absolutely essential. This is a skill which can easily be developed with a minimal investment in training and certification.

9) Failure to Integrate with Other CM Technologies
Oil analysis does overlap with other condition monitoring technologies such as vibration analysis and infrared thermography, but the overlap is small. A study at the Palo Verde Nuclear Generating Station some years ago showed that oil analysis and vibration analysis only agreed with each other 27 percent of the time when monitoring bearing failures. It’s not surprising that there is such divergence between the tests given that they are monitoring very different phenomena.

Oil analysis will most successfully be used when it is integrated with other CM technologies. To do so requires someone with a good knowledge of the subject and who knows its strengths and weaknesses.

Conclusion
Maybe it’s time to look at your oil analysis program and check to see if it’s giving you the value you expect. In the current economic climate, there is little tolerance for waste, and turning an asset with current mediocre performance into a star with just a little effort may be just what the mechanical pathologist ordered.

Subscribe to Machinery Lubrication

About the Author