Why Has Gas Turbine Lube Oil Turned Black With Clean Codes?

Both readings are true. The colour and the number sheet are measuring different populations of material. Work through the checks below in order and each one either closes a branch or hands you the next test.

Why does the oil look black when the particle count is clean?

An automatic particle counter reports the ≥4 µm, ≥6 µm and ≥14 µmIt is blind to everything smaller. The chromophores that turn a bright red oil brown and then black are oxidation intermediates, additive degradation products and soft carbonaceous soot in the 0.05-1 µm range, plus species that are fully dissolved. They pass a 3 µm element without touching it and they never reach a counter channel.

Aminic antioxidant degradation products are the worst offenders because they are intensely coloured. A few parts per million will take a water-white base oil to dark brown with no measurable move in viscosity or acid number. So the clean count tells you there is no hard particulate loading and no filter breakthrough. It says nothing about how hard the oil has worked.

Check before moving on: pull a live sample from the running header upstream of the filter - not a sump drain, not a dead leg - after flushing five to ten line volumes. Stand it beside a retain of the new oil in identical clear bottles. Then pull the same sample through a 0.45 µm membrane patch. A stained patch with clear filtrate means the colour is suspended and filterable. A clean patch with black filtrate means the colour is dissolved, and no mechanical filter on site will touch it.

If the bulk oil temperature is normal, where is the heat?

Header and sump RTDs read a mixed average of several thousand litres. Oil degrades at surfaces and inside bubbles, at temperatures the bulk instrument will never see. Three localised mechanisms produce black oil while the trend line stays flat:

• Micro-dieseling. Entrained air is carried into a high-pressure zone - pump inlet, gear mesh, orifice - and compressed adiabatically. The bubble interior reaches ignition temperature and burns a shell of oil around it. Output is black carbonaceous soot, sub-micron, distributed evenly. Signature: dark oil, black specks embedded in filter media, degraded air release and foam on the reservoir.
• Hot-spot degradation. A bearing drain line running against an uninsulated hot surface, a failed heat shield, a scavenge line that stays full and cooks between starts. Signature: varnish laid down on the cold side of the system - servo spools, control valve bores, cooler tubes, bearing shells - and a burnt smell in the sample.
• Electrostatic discharge. High flow velocity through tight non-conductive filter media with a low-conductivity Group II or III base stock generates charge faster than it can bleed off. The spark carbonises oil inside the element. Signature: short element life, pinholes in the media, black flecks, audible crackling at the filter housing.

Check before moving on: trend bearing metal temperatures and header-to-return ΔT against the commissioning baseline; IR scan drain lines, cooler shells and filter housings at load; cut open a used element and inspect the pleats under light; note filter ΔP history and element service life. Carbon specks in the pleats or a shortened element life points at heat or discharge, not at contamination.

Did something get into the oil?

On a pipeline compressor station the sump is not a sealed system. Bearing housings breathe, seals leak both directions, coolers develop tube leaks, and overhaul crews carry lubricants that were never on the approved list. Each ingress path has a fingerprint.

Check before moving on:run one panel on both stations covering ICP, FTIR, Karl Fischer and copper strip, and run the identical panel on the retained new oil.

Why did two stations turn at the same time?

Two separate sumps on two separate sites changing colour together is a strong signal, and it points away from a machine fault unless the units are genuinely identical in duty, ambient and oil age. Look first at what the two sites share off the machine: the oil batch and delivery lot, the top-up drum or tote, the transfer pump and hoses, the filter element part number if a media change was made, the flushing procedure, and the contractor and window of the last overhaul. A single contaminated tote or a filter media substitution will present on every unit it touched, at the same time, with identical symptoms.

Check before moving on: pull batch and lot numbers for the last three deliveries into each station and locate the supplier retain samples. Run MPC (ASTM D7843), antioxidant voltammetry (ASTM D6971, RULER) and ICP on the retain. If the new oil already reads high on MPC or low on antioxidant, the investigation is finished on the supply side.

Does black oil warrant a change-out?

Colour is not a condemning limit and never has been. ASTM D1500 colour is recorded for trend, not for decisions. These are the properties that condemn a charge:

Use ASTM D4378, the practice for in-service monitoring of mineral turbine oils, as the framework, and take the actual numeric trigger from the turbine OEM's condemning limits table and the oil supplier's report. Then follow the branch:

1. All properties hold, MPC low. The colour is cosmetic - trace degradation products or a shifted dye. Keep the charge, shorten the sampling interval, close the file.
2. MPC high, everything else in spec. Varnish potential, not end of life. Fit a side-stream kidney loop sized to turn the reservoir over several times per day, using ion-exchange or electrostatic/balanced-charge media. Pleated filters and centrifuges pass soluble varnish untouched.
3. RPVOT and antioxidant down, TAN climbing. The charge is finished and the colour is the symptom. Drain, flush, refill. Topping up a charge with no antioxidant reserve buys nothing.
4. Fingerprint says foreign lubricant, water or gas ingress. Repair the ingress path first. Incompatible additive packages require a full drain and flush; filling clean oil into a leaking system just re-blackens it.

Check before moving on: put the OEM condemning limits table and the laboratory report side by side and mark each property pass or fail. If nothing is exceeded, the oil stays in, and that decision is now documented.

How do you prove the decision held?

One pitfall to set expectations for first: ion-exchange media strips the red dye along with the varnish precursors. Oil that comes back from a clean-up loop lighter than new is normal and is not a second problem.

1. Retain one litre of the in-service charge and one litre of the new oil in sealed amber glass, labelled with unit, sump, date and running hours.
2. Fix the sampling method: same live port upstream of the filter, same flush volume, same 30-day interval, both stations.
3. Trend the panel each interval - MPC, antioxidant, TAN, viscosity, water, ICP, copper strip. Record ASTM D1500 colour, act on none of it alone.
4. Verify the machine side: filter ΔP and element life back at baseline, no carbon specks in a cut element, bearing metal temperatures unchanged, zero servo or control-oil sticking events.
5. Close the ingress paths that the fingerprint identified - pressure-test the cooler water side and watch sump water by Karl Fischer, confirm seal buffer gas differential and vent routing, and reissue the station's approved lubricant list to the overhaul contractor.
6. Take the kidney loop offline and pull the next 30-day sample. If MPC and antioxidant hold at the post-clean values with viscosity and TAN unchanged against the retain, the colour bodies were carried in and have been removed. A rebound within two to four weeks means the machine is still making them, and the search goes back to heat and ingress.

FAQ