Used oil analysis results: keep running, top up or change oil?

This decision rule covers mineral industrial oils in circulating and splash systems: hydraulics, gearboxes, compressors, pumps and turbines. Engine oils, where fuel dilution and base number drive the decision, and synthetic fluids outside ASTM D6224, which the practice tells users to take to the fluid maker, need their own limits.

Which used oil analysis results decide between keep running, top up and change?

Used oil analysis results decide the action through three groups: oil condition (viscosity and acid number) decides whether to change, contamination and wear (water, particles and wear metals) decide whether to find a cause, and oil level alone decides a top-up. NASA describes oil analysis as measuring these same three things: machine wear, lubricant condition and lubricant contamination (NASA, 2008).

The groups matter because each points to a different fix. A change of oil repairs oil condition, since the new charge has its original viscosity and acid number. It does not repair a leaking seal, a failed breather or a bearing that is wearing. ASTM D6224-23, written for auxiliary power plant equipment, says its practice "may be used to perform oil changes based on oil condition and test results rather than on the basis of service time or calendar time" (ASTM D6224-23, 2023).

A top-up belongs to none of the three groups. It answers a low level, and it should follow a result set that is already inside its limits. When a low level comes with water or rising wear metals, the oil level is a symptom of a leak or a consumption problem, and the top-up waits until the cause is known.

The limits have to exist before the sample does. A result read against a limit chosen after seeing it is an opinion, not a decision rule.

How do viscosity, acid number, wear metal and water results in used oil analysis map to an action?

Viscosity, acid number, wear metal and water results in used oil analysis each map to one action: viscosity and acid number outside limits mean change, wear metals above limits mean find the cause, and water above its limit means find and stop the source.

Viscosity is measured against the new-oil baseline. NASA's guide discards or refreshes the oil at a 10 percent change in machines with less than about 19 L (5 US gal) of oil, and notes that viscosity below baseline probably means the wrong makeup oil was added (NASA, 2008). The US DOE guide uses ±15% as the abnormal band (US DOE, 2010).

Acid number tracks oxidation. NASA monitors it relative to the acid number of the new oil (NASA, 2008). The US Bureau of Reclamation says the maximum depends on the oil and its service and should come from the oil maker, and that a sudden increase may mean an operational problem or the end of useful life (USBR, 2024).

Wear metals have no general limit. ASTM D5185-26 states that wear metal concentrations can indicate abnormal wear "if there are baseline concentration data for comparison" (ASTM D5185-26, 2026).

Water has the strictest rule for small systems: NASA says there should be no water present, with 0.1 percent as the minimum the percent solids and water test detects, and that the source must be found and corrected (NASA, 2008).

Used oil analysis action table: keep running, top up, change or find the cause

The used oil analysis action table turns each result into one of four actions, and it works only if the limits in its second column are written down for each machine before the sample is taken. Use the equipment maker's limit where one exists, the laboratory's statistical alarm for wear metals, and the oil maker's maximum for acid number. Where the table quotes a published value, treat it as a default for that equipment class, not a rule for every machine. ASTM D4378-24 calls the values in its own practice "purely indicative" (ASTM D4378-24, 2024).

The table below decides keep running, top up, change or find the cause for each used oil analysis result, with the limit source in the value cell.

The wear metal row matters most: an oil change makes the next report look normal, but the row says to find the cause and resample, because the wear source is still running. Read across each row before acting; a result in limits on its own row can still change the action when another row fails.

When does topping up hide a used oil analysis problem?

Topping up hides a used oil analysis problem when the volume added dilutes the evidence, when the makeup oil differs from the charge, or when the next sample is read without knowing a top-up happened.

The dilution effect is the same one STLE describes for oil changes: fresh oil lowers wear metals and contaminants "simply by dilution" (STLE TLT, 2014). A partial top-up does this in proportion to the volume added, so a reservoir that needs frequent large top-ups will show falling wear metals and water while a seal leaks out oil and lets contaminants in. This is inference from dilution, not a measured result.

Makeup oil also changes the oil itself. NASA reads a viscosity below baseline as probably caused by adding the wrong makeup oil (NASA, 2008). The US Bureau of Reclamation advises that the makeup oil be identical to the oil already in the system, and to check the oil more frequently if a different makeup oil is used (USBR, 2024).

Three habits stop the masking:

• Record each top-up on the sample label. NASA lists the last "topping off" among the information marked on each sample, and says the analyst needs the reservoir volume (NASA, 2008).

• Take a sample after topping off. NASA recommends a periodic analysis following topping off or reconditioning (NASA, 2008).

• Track top-up volume over time. A rising volume is itself a result, and ASTM D4378-24 lists the top-up level among the factors to weigh when interpreting turbine oil results (ASTM D4378-24, 2024).

Why doesn't an oil change fix high wear metals or water in used oil analysis results?

An oil change does not fix high wear metals or water in used oil analysis results because it removes the evidence, not the source. STLE's editor Evan Zabawski wrote that "changing the oil merely makes an oil analysis report look better. It does not fix contamination ingression, or balance, repair or realign parts" (STLE TLT, 2014).

The same editorial explains the arithmetic. Fresh oil lowers wear metals and contaminants by dilution, and over the same interval the system is "practically guaranteed to achieve the similar values again" (STLE TLT, 2014). An oil change therefore buys time and resets the trend. NASA states that a new baseline analysis is necessary following machine repair or oil change out (NASA, 2008), which means the history that showed the rise starts again from zero.

Keeping the oil after fixing the source gives a faster answer. In STLE's breather example, after the repair, one more sample shows whether ingress has stopped, because the values stabilize instead of climbing; after an oil change, it may take three samples to prove the same thing (STLE TLT, 2014). The editorial adds the condition: leave the oil in only if the levels are tolerable, because there is little point keeping it for one more interval if significant damage is occurring.

Oil changes do fix problems of the oil itself. STLE gives the build-up of oxidation by-products such as varnish as the example, while noting a change alone defers the next occurrence (STLE TLT, 2014).

How should a trend change the decision when single used oil analysis results are in limits?

A trend should change the decision when single used oil analysis results are in limits but rising at a rate the machine's own history does not explain: the action moves from keep running to resample sooner and find the cause, even though no limit has been crossed.

Acid number is the clearest case. The US Bureau of Reclamation writes that "Of greatest concern in this test is the rate of increase and not necessarily the neutralization number itself" (USBR, 2024). Two samples at 0.15 and 0.35 mg KOH/g both sit under a maximum of 0.5, but the second rise is the signal (illustrative).

Wear metals work the same way. ASTM D7720-21(2025) guides the statistical evaluation of alarm limits from in-service oil data, with techniques that separate normal variation from "variation that has significant, assignable causes" (ASTM D7720-21R25, 2025). The guide also says its techniques do not supersede limits established and validated by the equipment maker. POLARIS Laboratories notes that laboratories typically set statistical limits for normal wear accumulation from equipment and fluid information, alongside fixed OEM wear limit tables (POLARIS Laboratories, 2023).

A trend is only valid when samples are comparable. Same sample point, same interval, and the same test method: NASA warns, for base number tests, that results can vary between test methods (NASA, 2008). NASA lists two acid number methods, color change (ASTM D974) and electrical conductivity change (ASTM D664), so keep one for the whole trend (NASA, 2008). Changing any of these starts a new baseline.

When should you ask the lab about used oil analysis results before acting?

You should ask the laboratory about used oil analysis results before acting when the result has no baseline to compare with, when two results point to different actions, or when the oil is outside the scope of the limits you hold. In those cases the action table alone cannot rule.

Ask before acting in these situations:

• No baseline. A first sample, a new machine or the first sample after an oil change has no history, and wear metal readings mean little without one (ASTM D5185-26, 2026).

• Conflicting rows. Viscosity in limits but water present, or acid number steady while wear metals rise, needs the lab's reading of which row drives the action.

• Additive depletion. The US Bureau of Reclamation notes that additive content is usually proprietary, so the oil maker may have to be asked how far additives are depleted (USBR, 2024).

• Fluids outside the practice. ASTM D6224-23 does not address synthetic lubricants other than phosphate esters and tells users to consult the manufacturer (ASTM D6224-23, 2023).

• Particle counts against a target code. ISO 4406:2021 specifies only the code for the quantity of solid particles (ISO 4406:2021), so the target code has to come from the equipment maker or the lab.

Lubinpla runs performance evaluation on the same rule, tying each analysis value to a limit, sample and method fixed beforehand, and returns a keep running, top up or change ruling with its evidence for review.