Oil analysis types

Lubricating oil analysis comes in three families, and each one asks a different question about the machine: is the oil still good (physicochemical), what got into it (contamination), and is the machine wearing out (wear). It works like a blood test: a small sample of the oil circulating in the machine shows the health of the oil and of the parts it has passed through. This article explains what each analysis measures, why a single one decides little, and which one to order for each machine.

What is lubricating oil analysis for?

Lubricating oil analysis is for deciding, with measured data, two things: whether the oil can stay in the machine and whether the machine is wearing out. It is like a routine blood test, which finds the problem before the person feels any symptom.

It all starts with the sample: a small bottle of oil, in the amount the laboratory asks for the chosen tests, taken while the machine is running or right after it stops. The sample is always taken at the same point, in a place where the oil is circulating, and never from the bottom of the reservoir, where dirt settles and distorts the reading.

The laboratory compares this used-oil sample, the oil that has already worked in the machine, with new oil, the oil that came out of the drum. New oil is the reference: each result is worth how far it has moved away from it.

The same set of tests has two uses. On new oil, it confirms that the product delivered is the one ordered, as shown in the article on the lubricant laboratory and its tests. On used oil, it measures how much the oil has changed inside the machine. This second use prevents both losses: changing good oil too early and breaking the machine with spent oil.

To read what changed, you first need to know what happens to the oil in there.

How does oil degrade inside the machine?

Oil degrades in five ways at once, and each one leaves a mark the laboratory can read. You already know the first one from the kitchen: butter left out of the fridge turns rancid, because the fat reacts with oxygen in the air, and heat speeds that reaction up. Oil suffers the same, plus four other things.

Before the five ways, one word: additive. It is the chemical added to oil in a small dose to give it a quality or strengthen one, such as protection against wear, rust or foam.

1. Heat and oxygen attack the oil molecules, a reaction called oxidation, which leads to acids and larger molecules. The leftovers become sludge, a pasty deposit, and varnish, a thin amber-to-dark-brown film stuck to the parts, which leads to stuck valves and clogged filters.

2. Outside fluids get into the oil. Water arrives through the heat exchanger (the equipment that cools the oil with water), through condensation in the reservoir or through a seal. Another oil gets in by mistake during top-up, and solvent gets in during cleaning. In an engine, fuel and radiator coolant also get in. This leads to rust, spoiled additive and an oil whose recipe differs from the original.

3. Hard particles from outside, such as dust, sand and metal chips left over from assembly or maintenance, enter the clearances (the minimal gap that lets one part move against another) and scratch the metal. Each scratch releases more particles, in wear that feeds itself.

4. The machine's own parts wear and leave metal in the oil.

5. Additives are used up while they protect, and some, made of long molecules, break under the mechanical stress between the parts, which leads to an oil that protects less.

The five ways fit into three questions, and each question is a family of analysis:

• what happened to the oil itself (ways 1 and 5): physicochemical analysis;

• what got in from outside (ways 2 and 3): contamination analysis;

• what came out of the machine (way 4): wear analysis.

Figure 1. The five ways oil changes inside the machine and the family of analysis that reads each one.

How this shows up in your plant: dark oil with a burnt smell, cloudy or milky oil in the sight glass (the little window on the reservoir), a filter that clogs before its time, operating temperature creeping up, foam in the reservoir. By the time the symptom reaches the sight glass, the analysis had been showing the change for a while.

What does physicochemical oil analysis show?

Physicochemical analysis shows whether the oil itself has changed, and two readings almost always decide: viscosity and acidity.

Viscosity is how much the oil resists flowing: honey is thick, water is thin. It is what keeps the oil film, the thin layer that separates one part from another. The laboratory measures the sample and sees how far it has moved from new oil, up or down.

Viscosity that goes up points to oxidation, mixing with a thicker oil, soot (in an engine) or water trapped in the oil as tiny droplets. Viscosity that goes down points to mixing with a thinner oil, solvent or fuel getting in, or a long-molecule additive broken by the stress between the parts. These are the ways you have just seen, read in a single number.

Figure 2. The viscosity of used oil compared with new oil: each side has its own causes.

The viscosity of new oil is in the product name, through the ISO VG grade; the article on what ISO VG means explains how to read that number.

The second reading is acidity. Oxidation produces acid, so acidity rises as the oil ages. The technical name is acid number, or TAN (total acid number). What counts is the difference from new oil, because some additives start with an acidity of their own.

In engine oil, the reading that weighs most is the alkaline reserve, or TBN (total base number): the part of the oil that neutralizes the acid formed when fuel burns. When it runs out, the acid starts attacking the metal.

For those who need the number: viscosity is measured in centistokes (cSt) at 40 °C, and also at 100 °C for engine oil. ISO 3448 allows ±10% around the nominal value in new oil: an ISO VG 68 falls between 61.2 and 74.8 cSt. That range applies to oil delivered in drums.

How far used oil may move from new oil is agreed per machine and per oil type, using ASTM D6224 (a guide for in-service gear, hydraulic, diesel engine, pump, compressor and electrohydraulic control oil) and ASTM D7720 (how to set those values statistically). TAN and TBN come out in mg of KOH per gram: TAN by NBR 14448 (the Brazilian standard equivalent to ASTM D664), TBN by NBR 5798 (equivalent to ASTM D2896).

What do infrared and spectrometry show in oil?

Infrared shows what changed in the oil molecules, and spectrometry shows which chemical elements are in the oil and how much of each. Both read the chemistry from the inside.

Infrared works like a fingerprint. A light the eye cannot see, infrared, passes through the oil, and each type of bond between the atoms of the molecules absorbs a band of that light. The pattern of what was absorbed, called a spectrum, is the oil's fingerprint. Compared with the spectrum of new oil, it shows oxidation, a lot of water, soot and coolant (in an engine), antiwear additive being used up, and mixing with another oil.

Spectrometry is like fireworks, whose color comes from the metal salt mixed into the charge: strontium salt gives red, copper salt gives blue or green.

In the laboratory, the sample is heated in a plasma, a gas at thousands of degrees. Each chemical element (iron, copper, zinc, each type of atom) emits light in its own colors, most of them invisible to the eye, and the instrument measures how much there is of each color. From that comes the amount of each element.

That amount is given in ppm, parts per million: 1 ppm is 1 milligram in 1 kilogram of oil.

Additives have their own elements: antiwear additive carries phosphorus and zinc, and other additives carry calcium, barium and boron. If used oil shows additive elements different from those of new oil, the most common cause is mixing or topping up with another oil.

One caution when reading: spent additive usually stays in the oil, and spectrometry counts the element whether it is working or not. So stable phosphorus and zinc do not prove that the antiwear additive is active; infrared completes the reading. The same instrument also measures the metals that come off the parts, and they return further on, under wear.

For those who need the number: infrared of used oil is read by trend, against new oil, by ASTM E2412; oxidation has its own method, ASTM D7414.

Varnish risk is measured by membrane patch colorimetry (ASTM D7843), and the reserve of antioxidant is followed by voltammetry (ASTM D6971); both standards were written for turbine oil and are also used, by trend, on other circulating oils. Plasma spectrometry (ICP) of used oil follows ASTM D5185.

How does the analysis measure water in oil?

Water is measured by a chemical test called Karl Fischer, a reaction made only for water, which gives the content in ppm and reveals water long before the oil turns cloudy.

Water gets in by the ways you have already seen: heat exchanger, seal and condensation. Condensation is deceptive: a glass of cold water gets wet on the outside, and a reservoir that heats and cools every day collects water on the inside in the same way.

Why not trust the eye? Oil only turns cloudy when water passes what it can dissolve, and that limit changes with temperature and rises in aged oil: old oil holds more water without clouding, and that water is already attacking the machine. So a clear oil when hot may turn cloudy as it cools, and a clear oil may already have too much water.

In circulating oil (the oil that passes through the machine in a closed loop, as in the bearings of a paper machine), it also matters whether it still releases water in the reservoir instead of forming a milky mixture with it, the emulsion. This ability to separate water is called demulsibility.

For those who need the number: 1% water is 10,000 ppm. Coulometric Karl Fischer follows ASTM D6304, with the sample injected directly or through an oven. Some sulfur compounds interfere; in heavily additized oil, the laboratory uses the oven, which evaporates the water so that only water reaches the measuring cell.

Demulsibility follows NBR 14172 (equivalent to ASTM D1401); detergent hydraulic oil is the exception, made on purpose to hold the little water present in fine droplets.

How does the analysis count particles in oil?

An automatic counter passes the oil in front of a light and counts each particle that casts a shadow, sorting the particles by size. The ones that matter are like a grain of sand in a shoe: tiny and able to hurt.

They are measured in micrometers (µm). A micrometer is one thousandth of a millimeter, and a human hair is about 70 micrometers thick. Between the parts of a piston pump or a servo valve (a precision hydraulic valve), the clearance is a few micrometers; a particle the size of the clearance scratches the part and can jam the valve.

The result is written as a code, the ISO 4406 code: three numbers, one for particles from 4 micrometers up, another from 6 up and another from 14 up, in each milliliter of oil, such as 18/16/13.

One rule is enough to read the code: each number indicates the range in which the count falls, and each extra number means about double the particles. The lower the number, the cleaner the oil. The 18/16/13 in the example means more than a thousand particles of 4 µm and up in each milliliter; the acceptable limit depends on the machine and is agreed with the Technical Team.

Figure 3. The 18/16/13 code: one number for particles from 4, 6 and 14 micrometers up, and each extra number means about double the particles.

Cleanliness matters most in the hydraulic circuit, and the article on types of hydraulic oils shows that circuit and its clearances.

For those who need the number: automatic light-blockage counting follows ISO 11500, with the counter calibrated by ISO 11171 (that is what the "(c)" in "µm(c)" indicates). By table 1 of ISO 4406:2021, 18/16/13 means, in each milliliter, more than 1,300 up to 2,500 particles from 4 µm(c), more than 320 up to 640 from 6 µm(c) and more than 40 up to 80 from 14 µm(c).

Free water droplets and air bubbles also cast shadows and count as particles. ISO 4406 is the way to write the result; the counting itself follows other standards. For oil with water or very dark oil, ASTM D7647 dilutes the sample before counting, and ASTM D7596 counts by imaging.

A particle that comes from outside is dirt; one that comes out of the machine itself tells another story.

Which analysis shows machine wear?

Wear shows up in two analyses: spectrometry, which measures how much metal is in the oil, and ferrography, which shows the shape of the metal particles. Think of an eraser that leaves crumbs when it rubs out: a part that wears leaves metal in the oil, and the type of metal points to the part.

Spectrometry uses the same plasma instrument seen above, now reading the wear metals:

• iron (Fe): gear, shaft, bearing, engine cylinder liner (the cylinder inside which the piston runs);

• aluminum (Al): piston, housing (the case that encloses the machine);

• copper (Cu): bronze bushing (the ring that supports the shaft), which also releases tin (Sn), brass cage of the bearing (the part that keeps the balls apart), heat exchanger tube;

• lead (Pb) and tin (Sn): the soft alloy that lines the plain bearing (the part in which the shaft turns supported on a film of oil, without balls);

• silicon (Si): dust that got in through the breather (the opening of the reservoir through which air enters and leaves) or through the seal, after subtracting the silicon of the antifoam (the additive against foam), which is already in new oil.

Figure 4. The metal points to the part: iron, aluminum, copper, lead and tin, and silicon with the aluminum of dust that comes in through the breather.

A reading rule helps: silicon and aluminum rising together point to dust, because mineral dust is rich in silicon and aluminum.

Spectrometry has a blind spot: particles larger than a few micrometers partly escape the measurement, and they are precisely the ones from severe wear. For those there is ferrography.

In ferrography, a magnet pulls the iron particles onto a glass slide and lines them up by size; those of other metals, such as bronze and aluminum, stay beside them, scattered. The analyst examines everything under the microscope, and the shape tells the cause:

• flat flake: fatigue, a part that peels from being loaded over and over, in a bearing or a gear tooth;

• spiral, like the metal ribbon that comes off a lathe: cutting by a hard particle;

• large particle, with parallel scratches on the surface and straight edges: severe sliding, metal against metal, with the oil film broken.

Ferrography comes in when iron rises and you need to know why. In a gearbox, where the iron comes from the tooth, the choice of oil that holds back this wear is in the article on which oil to use in an industrial gearbox.

For those who need the number: spectrometry of used oil follows ASTM D5185, which calibrates with metal dissolved in oil and warns that it does not quantify insoluble particles; the limit of a few micrometers comes from the predictive maintenance literature. Analytical ferrography follows ASTM D7690, which identifies the particles under the microscope and does not measure concentration.

So is high iron in a sample already a reason to stop the machine?

Why does a single analysis decide so little?

Because one analysis is a photograph, and the decision needs the sequence. A temperature of 37.5 °C, measured once, says little; a person with 36.8, 37.2 and 37.8 °C on three days in a row shows that something is getting worse. Samples from the same point, at regular intervals, make up the trend.

An example, only to illustrate: iron at 40 ppm says little on its own. The same iron rising from 10 to 20 and then to 40 ppm in three samples at equal intervals shows wear accelerating: at each interval, the iron rises more than in the previous one.

Figure 5. Example: the same iron at 40 ppm, alone and as the third of three samples at equal intervals, in a trend that rises and crosses the alert limit agreed for the machine.

The trend is read against the alert limit: the value agreed for that machine, above which you act. With both, the decision falls between keeping the oil, fixing the cause or changing the oil. This is condition-based oil change: changing when the analysis shows it is time, not by the calendar. It matters most where stopping the machine is costly, as the article on lubricants for the tissue industry shows.

The exception is a result very far from new oil, such as visible water, which calls for action on the first sample. And the trend only holds with comparable samples: same point, same machine condition, clean and labeled bottle.

For those who need the number: ASTM D7720 shows how to set alert limits statistically, and ASTM D6224 is the guide for in-service gear, hydraulic, diesel engine, pump, compressor and electrohydraulic control oil.

Which oil analysis to order for each machine?

The right analysis is the one that reads what most threatens that machine: particles in the hydraulic system, tooth wear in the gearbox, oxidation in the compressor, water in circulation, acid and soot in the engine.

For the compressor, the article on air compressor oil shows why heat weighs so much. If your machine is not in the table, the Technical Team builds the plan with you.

How does Kelpen Oil do oil analysis?

Kelpen Oil analyzes oil in its own laboratory in Arujá (SP), Brazil, which does quality control of new and used lubricants, with 25 analyses by ASTM, NBR and ISO standards. The same laboratory analyzes the lubricant in use in your operation, as a paid service, one-off or under contract.

It is worth sending a sample of the new oil along, or at least the product name and the batch, because each result is read against new oil. The table shows what the laboratory measures in used oil.

Frequently asked questions

How do you do a lubricating oil analysis? Take the sample with the machine running or right after it stops, always at the same point, in a clean bottle labeled with the machine, the date and the hours of oil use. Send it to the laboratory with the oil name, so that the result can be compared with new oil.

Where can you get a lubricating oil analysis done? At a laboratory that works by standards (ASTM, NBR or ISO) and states the method of each test in the report. The Kelpen Oil laboratory, in Arujá (SP), Brazil, analyzes the lubricant in use in your operation, as a paid service, one-off or under contract.

Are the types of transformer oil analysis the same? No. Insulating oil has its own tests and standards: dielectric strength (NBR IEC 60156, the Brazilian version of IEC 60156), which shows whether the oil still insulates against electricity, and dissolved gas analysis (NBR 7070), which points to electrical discharge or abnormal heating inside the transformer. Even water, measured by Karl Fischer, has its own standard for insulating oil (NBR 10710).

Does the color of the oil show whether it is good? Not on its own. Oil darkens with use, and some darken early without losing function; color helps to notice change, and what decides the oil change is viscosity, acidity, water, particles and metals, compared with new oil. ASTM color is measured by NBR 14483 (equivalent to ASTM D1500).

Glossary

• Alkaline reserve (TBN): the part of engine oil that neutralizes acid; when it runs out, the acid attacks the metal.

• Demulsibility: the ability of the oil to release water instead of forming a milky emulsion with it.

• TAN: total acid number, the amount of acid in the oil, which rises as it oxidizes.

• ISO 4406 code: three numbers that summarize how many particles of 4, 6 and 14 micrometers and up there are in each milliliter of oil.

• Spectrometry: a test that measures the elements in the oil, additive and wear, by the light each one emits in a plasma.

• Ferrography: microscope examination of wear particles separated by a magnet, which shows the cause of wear by their shape.

Talk to the Technical Team

The Kelpen Oil Technical Team evaluates the machine, the oil in use, the hours since the last change and what earlier analyses showed, and builds the analysis plan with you. The Arujá (SP) laboratory does the analysis of used oil, as a paid service. Before ordering the first analysis, choose the sampling point for each machine and keep a sample of the new oil: they are the reference for every reading that follows. Send this article to whoever handles maintenance at your company.