Oil & lubrication

Varnish in turbine oil, and why particle counts miss it

Varnish in turbine oil announces itself late: a turbine trips on a sticking servo valve. The oil analysis from three weeks earlier is clean, good particle count, water in range, acid number barely moved. Both are correct. Particle counters cannot see the thing that caused it.

What varnish in turbine oil actually is

Oil oxidises. Oxidation produces polar degradation products of increasing molecular weight, and those products stay dissolved while the oil is hot and has capacity to hold them.

When the oil cools, or when it becomes saturated with them, they come out of solution. They are sticky and strongly polar, so they deposit on metal, preferentially on the coolest surfaces and in the tightest clearances. That is varnish: servo valve spools, bearing surfaces, heat exchanger plates, reservoir walls.

Why the particle count misses it

An automatic particle counter sizes things that block or scatter light. Varnish precursors are dissolved in the oil at operating temperature, there is nothing to block the beam. By the time varnish is a solid particle, it is on a surface, not in the sample.

A clean particle count and a varnish problem are entirely consistent with each other. They measure different things.

This is also why filtration alone does not solve varnish. A filter removes particles from the oil; it does not remove dissolved degradation products, and it does nothing about what has already deposited.

Membrane patch colorimetry

ASTM D7843 gives varnish a number. The sample is aged and cooled under controlled conditions so dissolved products precipitate, filtered through a patch, and the patch colour measured as ΔE, the colour difference from clean.

Commonly used interpretation bands:

ΔEReading
Under 15Normal
15–30Monitor, trend it, sample more often
30–40Abnormal, deposits likely forming
Over 40Critical, expect operational symptoms

Two cautions. The bands are guidance, not a specification, and they behave differently across oil types, group II and group III base oils hold degradation products less readily than group I, so they can show a high MPC while looking otherwise healthy. And the conditioning step matters: results from labs using different ageing protocols are not directly comparable. Trend within one laboratory.

Symptoms that precede the number

  • Servo or proportional valves sticking, especially after a shutdown
  • Longer starts, or trips on cold start after a period at rest
  • Heat exchanger performance falling with no fouling on the water side
  • Filters darkening quickly without the particle count rising
  • Bearing temperatures creeping up a degree or two over months

What actually helps

  • Reduce the rate of oxidation. Heat, water and dissolved metals all accelerate it. A

running-hot cooler is a varnish problem in slow motion.

  • Remove the soluble precursors, not just the particles. That means deposit-control

chemistry or a resin-based medium, chosen against a measured MPC.

  • Clean what has already deposited. New oil in a varnished system re-dissolves deposits

and redistributes them, which is why a flush that skips the deposits often looks worse before it looks better.

  • Confirm with a repeat MPC. If it has not moved, the treatment did not work.

MPC testing is part of Malina's oil and fuel analysis, and deposit-control treatment sits under oil and lubrication. The test method is published as ASTM D7843.


Varnish is one of the few conditions where the machine tells you before the laboratory does. Sticking valves after a shutdown are worth a sample on their own.

Last reviewed 7 August 2026.

Have this checked on your own plant

One reservoir, one sample, one report, and a recommendation you can act on.

Book an oil audit