Predictive Maintenance That Works

In the April “Focus on O&M,” we began a series of articles on predictive maintenance (PdM), also known as condition-based maintenance. In that article we introduce PdM as a process where maintenance is performed based on the condition of the equipment rather than on a predetermined interval. We also discussed how a well-oiled PdM program requires an upfront investment in equipment and in the training of technicians in order to reap later benefits. The costs and benefits were also discussed in detail. Given the extremely high cost of an outage in lost energy sales, plus the high cost of replacement power purchases, the cost of a PdM program often is justified solely based on the expected plant reliability improvement.

This installment of the series continues our review of different conditioning-monitoring techniques commonly in use at power plants using any generation technology. In the May issue we began exploring specific PdM techniques with an examination of electrical surge comparison and motor-current signature analysis.

The equipment and techniques discussed are certainly not comprehensive, as new and improved methods and equipment are routinely introduced. We are discussing PdM programs that could be likened to a “basic load” of ammunition, food, and protective equipment for a soldier. You can always carry more, but the basics are sufficient for most missions.

For example, one major element of any robust PdM program is nondestructive testing (NDT) technologies. Within the category of NDT are a number of specific testing approaches used in the plant such a lube oil analysis, thermographic analysis, shock-pulse measurements, ultrasonic analysis, wear-particle analysis, and vibration analysis, to name a few. In this article we begin our discussion of the elements of NDT with a look at the most important aspects of a high-quality oil analysis program.

Routinely Analyze Your Oil

Oil analysis identifies the condition of fluids and lubricants and determines if they are suitable for continued use or should be changed. It also indicates the condition of internal, oil-wetted components, identifying excessive wear. Generally speaking, routine oil analysis will find active machine wear. Oil analysis also can be used to identify the presence of contamination, which can lead to premature failure. Particle contamination is usually credited with 60% to 80% of all machine lubrication-related failures. Oil analysis can be used on machines that have a circulating oil system, including steam or gas turbines, generators, hydraulic systems, diesel or gasoline engines, gearboxes, boiler-feed pumps, and even machine tools.

For the most accurate results, samples should be taken from an active, low-pressure line, ahead of any filtration devices. For consistent results and accurate trending, samples should be taken from the same place in the system each time (using a permanently installed sample valve is highly recommended). Most independent labs supply sample containers, labels, and mailing cartons. If the oil analysis is to be done by a lab, all that is required is to take the sample, fill in information (the machine number, machine type, and sample date), and send it to the lab. Results are normally available within 24 hours of receipt of the sample. If the analysis is to be done onsite, analytical equipment must be purchased, installed, and standardized. Sample containers must be purchased, and a sample information form created and printed.

The most common oil analysis tests are used to determine the condition of the lubricant, excessive wearing of oil-wetted parts, and the presence and type of contamination.

Oil condition is most easily determined by measuring viscosity, acid number, and base number. Additional tests can determine the presence and/or effectiveness of oil additives such as anti-wear additives, antioxidants, corrosion inhibitors, and anti-foam agents. Component wear can be determined by measuring the amount of wear metals such as iron, copper, chromium, aluminum, lead, tin, and nickel. Increases in specific wear metals can mean a particular part is wearing, or wear is taking place in a particular part of the machine. Contamination is determined by measuring water content, specific gravity, and the level of silicon. Changes in specific gravity often mean that the fluid has been contaminated with another type of oil or fuel. The presence of silicon (usually from sand) indicates contamination from dirt.

Spot checking (sampling a system once a year or less) is used primarily to determine whether the fluid or lubricant should be changed, or to confirm if a suspected problem actually exists. For example, if a machine is experiencing noticeable vibration or noise, an oil sample may be taken to confirm if there is bearing damage or excessive wear. Sampling machinery on a periodic basis (once a month or once a quarter) can provide a more subtle indication of lubricant or machine deterioration, or the slow introduction of contamination. Most bearing or gear failures occur after their condition has deteriorated slowly and steadily for a period of months or even years. Contamination may be introduced when oil is added to the system, and periodic monitoring will indicate this. Early warning of contamination allows repairs to be planned during a scheduled shutdown.

Long-term monitoring of oil condition over six or eight sample periods can identify improper maintenance or repair practices. These can include the failure to properly flush out a system after repairs, improper fluid- or lubricant-handling procedures (which introduce water or dirt contamination), or improper filter-handling or -replacement techniques.

Unusually rapid oil degradation can indicate that the oil is not suitable for the equipment or application. For example, a rust- and oxidation-inhibited oil, rather than a straight mineral oil, may be required where there is the possibility of high temperatures or water contamination. Rapid oil degradation may also indicate that the equipment is being operated beyond its original design capacity, creating excessive temperatures or bearing/gear surface loading.

Oil analysis is one of the simplest predictive techniques to use, and certainly one of the least expensive. Independent labs can help select machines and frequencies, suggest which tests to run, supply sample bottle and mailers, interpret the results, and archive data. The maintenance departments of most companies have some experience with oil analysis, if only on a limited basis.

In spite of its low cost and simplicity, oil analysis can be an extremely effective technique, particularly when the data is trended over an extended period of time (12 to 24 months). Trended data can identify poor maintenance and operating practices, which, if corrected, can result in substantial maintenance and operating cost savings.

One company found that contamination levels increased significantly each time oil was added to a gear reducer on a coal-handling system, and the contamination resulted in bearing and gear failures. Upon examination, they found that removing the cover plate to add oil allowed coal dust to fall into the sump. They installed a covered oil reservoir and piped it to all of the gear boxes. Now clean oil can be added by opening a valve, and the incidence of bearing and gear failures has been significantly reduced.

Also, make sure the samples are:

• Taken immediately downstream of the lubricated surfaces.
• Taken during normal operating conditions, including pressures and temperatures.
• Taken at the same location each time.
• Taken after the oil has circulated for a time after an oil change.
• Are representative of the oil in the machine.
• Are placed into a sample collection container that is clean, nonmetallic, and immediately sealed.

Oil analysis can be used only on equipment that contains a circulating oil system. In most cases it can indicate that a problem exists—for example, that there is excessive wear. However, it may not be able to identify the specific cause—what is causing the wear—or which of similar or identical parts are wearing.

Oil analysis is only as good as the timeliness and consistency of the sample. The longer a sample sits before it is shipped and analyzed, the less significant the data; and the value of trended information diminishes quickly if samples are not taken from the same place on the machine each time.

A final thought: Oil analysis is not a standalone program but must be integrated into a comprehensive PdM program. If you only take oil samples when there is a problem, or when there are other system problem indicators (such as high vibration), then you’ll miss a large measure of the program’s benefits. Oil analysis requires constant sampling in order to develop trends that can be used to identify problems before they become chronic. Without knowing what is “normal,” it’s difficult to determine what is “abnormal.” Sample oils frequently: critical equipment at least once a month and noncritical equipment at least once a quarter.

More Coming

In the next segment of “Predictive Maintenance That Works,” we’ll continue our discussion of specific NDT-related condition-monitoring techniques used at power plants and why each should be a part of your PdM program.

— Dr. Robert Peltier, PE is POWER’s editor-in-chief.