A working oil carries a record of everything happening in the machine around it: fuel leaking past rings, metal wearing off gears, heat oxidising the oil itself, water sneaking in. Oil condition monitoring reads that record to catch a problem before it becomes a failure. What stands out is how much of it you can see from a handful of cheap bulk properties — viscosity and density first among them — without identifying a single molecule.
The baseline is the whole idea
A condition-monitoring number means little on its own. Its power is in the comparison: this sample against a known-good baseline for that oil in that machine, and against the trend of past samples. A viscosity of 95 cSt is neither good nor bad until you know the fresh oil was 100 and last month’s sample was 98. Monitoring is trend-reading, not spec-checking, and the reference is the oil’s own history.1
Viscosity is one of the most telling single numbers. If it has risen, suspect oxidation, soot loading (in a diesel), water, or the wrong top-up oil. If it has fallen, suspect fuel dilution, a sheared VI improver, or again the wrong oil. Density moves with some of the same faults — fuel dilution with a lighter fuel such as diesel lowers both — so density and viscosity together narrow the cause faster than either alone. Both must be read at the same tightly controlled temperature, since a degree or two moves them more than the drift you are trying to detect.
Where the other tests come in
Bulk properties flag that something changed; targeted tests say what. Wear metals — iron from gears, copper from bearings, silicon from dirt or a silicone antifoam additive — are read by ICP-OES, tying oil analysis straight back to the elemental work of the ICP bench.2 Oxidation and nitration are tracked by infrared; acid and base number track the oil’s chemical life; Karl Fischer measures water. Each is a confirmation the bulk trend pointed you toward.3
Using it honestly
The discipline is simple and easy to get wrong. Establish the baseline before you trust a trend. Read the direction of a change as much as its size — up and down mean different faults. And treat a bulk-property alarm as a pointer to a targeted test, not a diagnosis on its own: viscosity told you the oil changed; ICP tells you it was bearing copper. When a cheap, fast measurement can flag an expensive failure weeks early, the cost of the test is small next to the failure it can prevent.
That returns to where the series began. Density started as a way to read a simple solution; viscosity joined it to fingerprint a fluid; here the two, trended against a baseline, become an early-warning system for a machine. One well-watched number rarely tells the whole story — but it reliably tells you when to go looking.