Turbine Lube Oil Analysis: What the Lab Report Is Actually Telling You

The oil in a steam turbine does more than lubricate. It carries heat out of the bearings, on many machines it is the working fluid of the governor and trip system, and it passes through every bearing, gear and servo valve on its way round. Almost everything that happens inside the machine leaves a trace in it. A sample taken properly and tested for the right things is the cheapest condition monitoring sensor on the plant.

Most plants already send samples to a laboratory. Fewer read the report as anything more than a traffic light, and fewer still trend it. This article goes through what the common tests measure, what a change in each one means, and what to do when a result moves.

Why the Oil Is a Sensor

Every oil report answers two separate questions, and it helps to keep them apart.

Is the oil still fit for service? Turbine oil degrades by oxidation. Its antioxidant additives are consumed, acids form, and eventually the oxidation products come out of solution as sludge and varnish. The oil also picks up water and air and gradually loses its ability to shed them. Viscosity, acid number, oxidation reserve, varnish potential, water, demulsibility and foaming tests answer this question.

Is the machine healthy? Wear metals, particle counts and contaminants show what the oil has collected on its way round: babbitt from a bearing, iron from a gear or from rust, copper from a cooler, silicon from dust drawn through a breather, sodium from a water leak. Here the oil is simply the messenger.

ASTM D4378 is the standard practice for in-service monitoring of mineral turbine oils in steam, gas and combined cycle turbines used for power generation. It includes sampling and testing schedules through the life of the oil, and it states plainly that the test values it mentions are indicative only: interpretation depends on the type of equipment, the workload, the design of the oil circuit and the top-up rate. That caveat is the philosophy of good oil analysis in one sentence.

Sampling Done Right

A laboratory result is only as good as the sample, and poor sampling produces more false alarms and missed faults than poor testing does.

Sample from the live zone. Draw oil where it is moving and well mixed, ideally from the bearing return line before it reaches the reservoir and upstream of any filter. Oil from the reservoir bottom drain describes the sediment in the reservoir. Oil from downstream of the filter describes the filter.

Same point, same conditions, every time. Trending only works if samples are comparable. Use the same valve, take the sample with the machine at normal load and temperature, and write the load, oil temperature and running hours on the label.

Flush the valve and use clean bottles. Run off enough oil to clear the valve and its dead leg before filling. Use bottles certified clean for particle counting, because a dirty bottle can make a clean system look contaminated.

Get it to the laboratory promptly. Some results, varnish potential among them, depend on how the sample is stored and how long it waits.

Set a frequency and keep to it. The right interval depends on the criticality of the machine and the condition of the oil, and ASTM D4378 gives a schedule to start from. Appearance and water checks are cheap and can be done far more often than the full laboratory suite. When a parameter starts to move, tighten the interval.

A laboratory result is only as good as the sample. Poor sampling produces more false alarms and missed faults than poor testing does. Source: MIMAH engineering practice, with schedules per ASTM D4378.

What Each Test Tells You

Viscosity. Measured at 40 °C and compared with the new oil. Viscosity sets film thickness in every bearing, so a change matters. A rise usually means oxidation or contamination with a heavier oil; a fall usually means contamination with a lighter product or a top-up with the wrong grade. Mixed oils are more common than most plants admit.

Water content. Measured by Karl Fischer titration and reported in parts per million. Water in turbine oil comes from gland steam leakage, condensation and leaking coolers. It promotes rust, speeds up oxidation, weakens the oil film and, in larger amounts, forms emulsions. A sudden rise usually has a mechanical source you can find: a gland seal, a cooler tube, a vapour extractor that has stopped working.

Acid number. The quantity of acidic oxidation products in the oil. It tends to rise slowly and late, so it confirms oxidation rather than warning of it. A step change can also mean contamination.

RPVOT. The rotating pressure vessel oxidation test, ASTM D2272, ages a sample under oxygen pressure at 150 °C in the presence of water and a copper catalyst, and records how long it holds out, in minutes. On in-service oil the standard uses it to assess remaining oxidation life, and the result is read against the value for the same oil when new. A steady decline is normal. A steep one means the oil is being worked hard by heat, entrained air, water or contamination. RPVOT is the best single number for planning an oil change rather than being surprised by one.

MPC varnish potential. Membrane patch colorimetry, ASTM D7843, filters the insoluble material from a sample onto a membrane patch and measures the colour of the patch, reported as a ΔE value. A darker patch means more degradation products ready to form deposits. The method is intended as a trending tool within a wider monitoring programme, and it is not suitable for dyed oils.

Particle count. Particles per millilitre in size bands, reported as an ISO 4406 cleanliness code: three numbers for particles of 4, 6 and 14 microns and larger. Each step in a code number represents roughly a doubling of particle concentration, so moving up one code is a bigger change than it looks. The cleanliness target is usually set by the tightest clearances in the system, which on many turbines are the governor servo valves.

Elemental analysis. Spectrometry measures dissolved and very fine metals and contaminants in parts per million. Tin, and lead on older bearings, points to babbitt; iron to gears, shafts or rust; copper to cooler tubes or bronze parts; silicon to dust; sodium to cooling water or boiler water getting in. Spectrometry sees large particles poorly, so a bearing shedding flakes of babbitt can show in the particle count well before the tin figure moves.

Foaming and air release. Foaming tendency and stability, and air release time, show whether the oil can shed entrained air. Air in the oil degrades pressure control in hydraulic circuits and accelerates oxidation. Contamination and additive depletion both make these results worse.

Demulsibility. How quickly the oil separates from water. Turbine oil must drop water in the reservoir so it can be drained. Oil that has lost demulsibility holds water in suspension and carries it to the bearings.

The common tests on a turbine oil report and what a change in each one means. Read every result as a trend against the new oil, not as a single number. Source: ASTM D4378, D2272 and D7843; ISO 4406; MIMAH engineering practice.

Varnish, and Why It Sticks Valves

Varnish deserves its own section, because it causes failures that look like control faults or bearing faults and are neither.

As turbine oil oxidises it forms degradation products that stay dissolved while the oil is hot. The oil can only hold so much. Where it cools, in the reservoir, the coolers and the low flow corners of the hydraulic system, those products come out of solution and settle as a thin, sticky, amber or brown film on metal. Servo valve spools that sit almost still for weeks at steady load are a favourite place for it.

The consequences are specific. Servo valves and pilot valves stick, because varnish builds in the small clearance between spool and bore. The governor hunts, responds late or does not respond, and a trip valve may close more slowly than it should, which makes varnish a safety problem as much as a reliability one. Bearing temperatures creep up, because a varnish film on the babbitt reduces clearance and insulates the surface from the oil carrying heat away; our article on steam turbine journal bearings covers what that does to the lining. Filters plug early and coolers lose performance, which raises oil temperature and speeds up the next round of oxidation.

Local hot spots accelerate all of this: bearing surfaces, air bubbles compressed in pumps, and static discharge in fine filters degrade the oil where the bulk temperature looks normal. That is why varnish can form in oil whose RPVOT still looks healthy, and why MPC is tested alongside RPVOT rather than instead of it.

Once deposits have formed, an oil change alone rarely clears them. The deposits sit on the metal, and fresh oil takes some back into solution and carries them round again. Treatment means varnish removal equipment on the reservoir, electrostatic or adsorbent resin types, and in bad cases a system flush. The measure of success is an MPC trend that comes down and stays down.

How varnish forms and why it shows up as control and bearing faults that are neither. MPC is tested alongside RPVOT because varnish can form in oil whose RPVOT still looks healthy. Source: MIMAH engineering practice; MPC per ASTM D7843.

Trend, Then Act

The most useful change most plants can make is to plot every parameter against time for each machine, instead of reading each report against a fixed limit.

An absolute limit tells you when something has already gone wrong. A trend tells you when it started. Water content that doubles between two samples, while still under any alarm level, is a gland or cooler problem you can find this month. A slow slide in RPVOT tells you roughly when the oil will need changing, so the change goes into an outage plan instead of forcing one. A particle count that jumps after an overhaul says the flush was not finished.

Build a baseline from the new oil and the first samples after commissioning or an oil change, then set alarm levels relative to it, informed by the ranges in ASTM D4378 and the oil supplier's guidance and adjusted for the machine. When a result moves, work through four steps.

Confirm it. Resample from the same point. A single bad result from a dirty bottle or a stagnant valve is common, and acting on it wastes time.

Find the source. Water points to glands, coolers and vapour extraction. Particles point to breathers, filters and recent maintenance. Wear metals point to specific components. Falling RPVOT and rising MPC point to temperature, air and contamination.

Correct the cause, then the oil. Filtration, dehydration and varnish removal clean the oil. They do not stop the gland leak.

Tighten the interval until the parameter is stable again.

Oil analysis belongs inside a wider condition monitoring programme, alongside vibration and bearing temperature, and it is one of the cheapest steps in the move from calendar based to predictive maintenance. It should feed the outage plan too: the turbine maintenance schedule is easier to hold when the state of the oil is known before the casing comes off. Many of the faults in our list of common steam turbine failures show in the oil first.

MIMAH's rotating equipment team includes oil analysis in the condition monitoring it delivers on turbines and rotating equipment, and reads the results alongside vibration and temperature data. If you have a stack of laboratory reports and no clear view of what they mean, or a governor that has started to stick, get in touch.