Oil analysis is a laboratory test of a lubricant sample that reads the health of both the oil and the machine it came from. A standard report checks three things: fluid condition (viscosity and acidity), contamination (water and dirt particles), and wear debris shed by internal parts. Because the oil touches every internal surface, its chemistry is a running diary of what is happening inside equipment you cannot open while it runs.
That makes oil analysis one of the workhorse condition-based maintenance techniques, alongside vibration and thermography. The report can look intimidating, columns of elements, ratios, and codes, but it is organized around those three simple questions. This guide walks through how to read the parameters that matter most: viscosity, TAN, water, and the ISO 4406 particle count, and how to turn them into action instead of filing them.
What are the three things an oil report checks?
Every standard report, whatever the lab's layout, is answering three questions. Read it in this order and it stops being a wall of numbers:
- Is the oil still good? Fluid-health tests, viscosity, total acid number (TAN), oxidation, and additive levels, tell you whether the lubricant can still do its job. This is the oil's own condition.
- What has gotten into it? Contamination tests, water content, particle count (ISO 4406), plus dirt, coolant, and fuel, tell you what is fouling the oil. Contamination usually kills machines faster than the oil simply wearing out.
- Is the machine wearing abnormally? Wear-metal analysis measures iron, copper, chromium, aluminum, and other elements the machine sheds. A spike in a particular metal points to a particular part. (Reading wear metals in depth is its own subject; here we focus on fluid health and contamination.)
How do you read viscosity and TAN?
Viscosity is the single most important number on the report, because viscosity is the property the oil was chosen for, its resistance to flow is what keeps a load-bearing film between moving parts. The lab compares the sample's viscosity to the new-oil baseline. A change within about 5% is normal, up to roughly 10% is often still acceptable, and a change beyond 10% above or below baseline is a common alarm point. Direction tells the story: viscosity rising usually means oxidation, contamination, or the wrong (heavier) oil was topped up; viscosity falling usually means fuel dilution or a lighter oil got in. Wrong viscosity means the film can fail, so this number gets read first.
TAN, the total acid number, measures acidity, reported as the milligrams of potassium hydroxide needed to neutralize one gram of oil. As oil oxidizes with heat, time, and contamination, it forms acids, and TAN climbs. Those acids corrode metal and mark the oil as nearing the end of its useful life. As with viscosity, TAN is judged against the new-oil baseline and, more importantly, against its own trend: a steadily rising TAN is a countdown to an oil change, and a sudden jump signals something changed in the machine or the operating conditions.
How do you read water and the ISO 4406 particle count?
Water is more destructive than most people expect, and it does damage well before the oil looks milky. Even dissolved water, invisible in the sample, accelerates additive depletion, promotes oxidation, and attacks bearing surfaces. That is why typical alarm limits sit low, often between 50 and 300 ppm for most systems, with gear oils frequently changed once water passes roughly 500 ppm (0.05%). Any free or emulsified water is an unambiguous problem to chase to its source, a failed seal, a breather pulling humid air, a leaking cooler.
Solid particle contamination is graded with the ISO 4406 cleanliness code a three-number code such as 18/16/13. The three numbers are range codes for the count of particles 4 microns and larger, 6 microns and larger, and 14 microns and larger, per millilitre. The scale is logarithmic, so each one-number step roughly doubles the particle count in that size band, which is why dropping even two code numbers is a large real cleanliness gain.
How do you actually read an oil analysis report?
- Check the header first. Confirm the sample is from the right machine, the right compartment, and a sensible interval, with the correct oil identified. A mislabeled sample produces confident nonsense.
- Read viscosity against baseline. Is it within about 10% of new-oil? If not, the oil's core function is at risk, find out whether it is oxidation, contamination, or the wrong top-up.
- Scan the contamination block. Water in ppm and the ISO 4406 code. High water or a climbing particle code often explains a wear trend elsewhere on the report.
- Read TAN and oxidation for oil life. A rising acid number says the lubricant is aging out, independent of contamination.
- Glance at wear metals for a story. A spike in one metal points at one part; broad low-level wear is usually normal. Match the metal to the components it comes from.
- Compare to the trend, then act. One report is a snapshot; the value is in the slope across samples. Decide: keep running, resample sooner, filter or dry the oil, change it, or open a maintenance work order.
| Parameter | What it reveals | Typical alarm signal |
|---|---|---|
| Viscosity | Whether the oil can still form a film | Beyond about ±10% of new-oil baseline |
| TAN (acid number) | Oxidation and acid formation; oil aging | Rising trend above baseline |
| Water content | Ingress from seals, breathers, coolers | Often 50–300 ppm; any free water |
| ISO 4406 code | Solid particle cleanliness | Climbing code vs. target for the system |
Why does sampling technique decide whether the report is worth reading?
A perfect lab cannot fix a bad sample, and most disappointing oil programs fail here rather than in the analysis. Three habits separate reports you can trust from noise. First, sample from the same point every time, ideally a dedicated sampling valve in a live, turbulent zone of the system, not the bottom of a tank where debris settles or a drain plug that grabs sludge. Consistency is what makes two reports comparable. Second, sample while the machine is warm and running, or shortly after shutdown, so the debris and water are still suspended in the oil rather than lying at the bottom.
Third, keep the sample clean. A particle count is only meaningful if the dirt in the bottle came from the machine and not from a grubby container or a wiped-off valve. Flush the sampling point, use clean bottles, and cap them immediately. Finally, sample on a schedule set by the asset's criticality, critical or hard-to-replace equipment on tighter intervals, minor assets less often, because the entire value of oil analysis is the trend, and a trend needs regular, comparable points. A single heroic sample after a machine already sounds sick tells you far less than routine sampling that caught the slope early.
What the numbers say
- Contamination control is where oil analysis pays back, and it is well documented that condition monitoring converts failures into planned work. The U.S. Department of Energy's FEMP O&M guidance, maintained by PNNL, reports that condition-based maintenance, the family oil analysis belongs to, saves 8–12% over preventive-only programs and offers an opportunity versus reactive operation that can exceed 30–40% (PNNL, O&M Best Practices: Maintenance Approaches).
- The particle-count grading itself is not a lab convention but an international standard: ISO 4406, the cleanliness-code method for hydraulic and lubrication fluids which is what to cite when someone asks where an 18/16/13 comes from or why each step doubles the count.
Where does oil analysis fit in a reliability program?
Oil analysis is one instrument in a condition-monitoring program and it is strongest where lubrication and contamination drive failures, bearings, gearboxes, hydraulics, large compressors. It pairs naturally with vibration analysis, which reads mechanical faults the oil cannot, and it is only as good as the sampling discipline behind it: clean, consistent, scheduled samples from the same point, trended over time. Sloppy sampling produces reports nobody can trust. The broader case for catching degradation before failure is in our predictive maintenance and equipment reliability guides, and the day-to-day practices that keep oil clean and correct are covered in lubrication management. Contamination and lubrication issues are also among the leading entries in bearing failure modes.
As with every condition technique, the payoff lives in the trend, and the trend only exists if results are captured against the right asset and kept over time instead of stapled into a binder. Plants that fold oil-analysis results into one operational layer, the way Harmony pulls floor data together, can see a water-ingress or oxidation trend building weeks before it costs a gearbox, and turn it into planned work. For how one plant built that foundation of trustworthy floor data, see the CLS case study.