For critical assets, we all know that condition-based maintenance (CBM) is the most effective reliability strategy. But for CBM to be truly effective, we must have both predictive and proactive data. But not just any old data; the data must provide sufficient information so we can make a judgment call on whether a machine is running normally, is starting to show signs of deterioration, or is close to failure. As a former colleague of mine Drew Troyer likes to say, "Data is the difference between deciding and guessing."
So what data do we need to effectively conduct CBM on gearboxes? When it comes to predictive maintenance (PdM) data, there are, of course, a number of choices, from vibration analysis to thermography to ultrasonics to oil analysis. With such a wide array of options, the most obvious question is, which one's better? The answer, of course, is it depends. It depends on the machine, the failure mode and the degree to which the failure has progressed. Take, for example, the issue of soft foot on a gearbox caused by a poor base mount. A vibration analyst skilled in doing phase analysis can easily pick up on this problem using vibration analysis long before the problem shows up in oil analysis. But turn that around; incipient gear wear caused by lack of oil film or incorrect oil viscosity will always show up first in oil analysis. For gearboxes, like every other asset, a variety of PdM tools are required to effectively address all possible and likely failure modes.
Having said that - and here's where I admit there might be some bias - in my opinion, pound-for-pound oil analysis is a far more effective predictive tool for gearboxes than any other PdM technology. Typical for most common gearbox failures, oil analysis will identify a problem well in advance of vibration analysis. To illustrate this, I always think back to an example from early on in my career from a large Renk gearbox at a cement plant. From oil analysis, we found a slow increase in iron wear month-by-month with no apparent change in vibration signature. But fully four months to the date from when the first abnormal oil sample appeared, the vibration analyst started to pick up evidence of a cracked inner race on the input shaft bearing. Once diagnosed, the problem was corrected during the plant's annual maintenance shutdown with no unscheduled downtime.
In this example, oil analysis was what I like to call the "sharp stick." It prodded the maintenance team into action to pay more attention to this critical gearbox, while vibration analysis provided a level of diagnostic specificity - cracked inner race on the input shaft bearing - that oil analysis likely would never provide. But for oil analysis to be effective, there are certain rules we have to apply. First and foremost, we have to have a good sample. The old adage, garbage in, garbage out, certainly holds true for oil analysis. Beyond sample quality, performing the right series of test is also critical. Let's look at the critical success factors for oil analysis on gearboxes.
Sampling Industrial Gearboxes
Gearboxes are the most common industrial component type that requires oil sampling. Unfortunately, they are also the most frequently incorrectly sampled component. The most common method by which gearboxes are sampled is to use a vacuum sampling gun and flexible plastic tube inserted through the fill port or breather port, a method often referred to as drop-tube sampling. Not only does this allow for the introduction of contaminants through the open port, the use of flexible tubing can also result in tremendous variation within the oil level (top, middle, bottom) and within the gearbox (bottom, gear case wall, etc.) itself.
A far better option for sampling is to install a dedicated sample port in the correct location. For wet sump circulating systems, the best place for a sample port is after the oil pump, but before any full flow filters (Figure 1). Taken from this location, the maximum amount of information about the degree of wear debris, contamination and fluid condition will be obtained.
Where no oil circulation system exists and oil is distributed by internal splash, sample ports should be installed on a suitable location using an existing port. Oftentimes, the dipstick port, level port, or drain port all can be modified with suitable sampling hardware to effectively extract a representative sample (Figure 2). Ideally, the sample tube should be installed such that the sample is being taken approximately halfway up the oil level, at least 2 inches or 5 cm away from the walls and any rotating elements within the gearbox. Once installed, the sample can be extracted using an appropriate threaded adapter coupled to a standard vacuum oil sampling gun. Facilities that have switched from drop tube sampling to the use of a dedicated sample valve often see an overall drop in baseline values for wear debris and contamination, as well as a significant reduction in sample-to-sample baseline variability (noise), significantly increasing the sensitivity of oil analysis as a diagnostic tool.
Compared to other predictive maintenance tools, such as vibration analysis and thermography, oil analysis is typically a leading indicator of a wear problem, often showing up weeks or months in advance of any significant change in vibration signature. This is especially true for low speed gears and bearings in a multi-reduction gear drive where vibration analysis is typically less sensitive due to difficulties with slow speed accelerometers. While oil analysis will typically show the problem first, vibration analysis is far better at localizing the exact component and failure mode. But both are required for an effective predictive maintenance program.
Test for Gearboxes
Aside from sample integrity, key to oil analysis for gear drives is selecting the correct series of tests. While standard tests, such as elemental analysis and viscosity, are important, it's equally important to look at other test methods to optimize the detection capability of oil analysis. This is particularly true for wear debris detection. When gearboxes start to wear, most wear debris is in the excess of five to 10 microns in size. At this particle size, the instruments used to provide elemental wear debris analysis (e.g., ppm of iron, copper, etc.) are blind, meaning a problem may go undiagnosed. Under these circumstances, large particle tests, including ISO particle counting or ferrous density analysis, is strongly recommended. To further diagnose active wear problems, access to wear particle morphology (size, shape, color, etc.) using either analytical ferrography or automated particle shape classification (e.g., LaserNet Fines) is also highly recommended.
While an annual or semiannual oil change is fine for smaller gearboxes, it makes sense for larger gearboxes to use oil analysis for condition-based oil changes, changing the oil based on oil analysis rather than time. But for condition-based oil changes to be effective, we need to look beyond viscosity. Typically viscosity will only change when the oil is 90 percent to 95 percent degraded, which is too late to prevent deposit and other long-term problems from occurring. Instead, we should use a combination of acid number and Fourier transform infrared spectroscopy (FTIR) oxidation as an indicator of lubricant health. Since many gear oils start with high acid numbers in the 0.5 to 4.0 mg/KOH range due to the extreme pressure (EP) or anti-wear (AW) additives used in these oils, FTIR is a good backup test, provided a good new oil baseline has been taken.
For gearboxes where water is a problem, we also need to be careful about the test method used. While FTIR or a simple crackle test can be good indicators of gross water ingression, the Karl Fischer moisture method (ASTM D6304) is preferred where accurate, trendable data is required. An ideal test slate for industrial gearing should include the tests listed in Table 1.
Sampling frequencies for gearboxes also should be considered carefully. While quarterly sampling may be adequate for slower turning gearboxes (< 50 RPM on the lowest speed shaft), for higher speed gear drives or very critical applications, monthly or even biweekly sampling is recommended. This is particularly true for newer gear drives where power densities are much higher and main gears are far more sensitive to particles, moisture and other poor lubricant conditions.
Gearboxes are commonplace across many different industries. Because of their size and apparent robust nature, gearboxes are often overlooked when it comes to precision lubrication. But with a little care and attention, focusing on getting the right lubricant in the right place at the right time while maintaining the oil in a clean, dry, cool condition, gear drives can provide reliable operations for many years.
“R.A.I.” the Reliability.aiTMChatbot
You can ask "R.A.I." anything about maintenance, reliability, and asset management.