Why do lube oil samples taken before and after the separator look the same?
Last verified: 2026-08-06
Two oil samples, drawn on the same day from the same medium-speed engine — one upstream of the centrifugal separator, one downstream — come back from the laboratory with results that barely differ. The reasonable conclusion is that the separator has stopped earning its keep, and attention turns to the machine. CIMAC's own recommendation on system-oil treatment names that conclusion, and names it as the wrong one. Separator efficiency is generally not reflected in routine lube oil analysis at all. The samples match because almost every number on a standard report tracks something a centrifuge was never going to remove.
What does a matching pair of samples actually prove?
Very little, and CIMAC says so directly. Its section on the effect of separation on lube oil analysis states that comparing results from samples drawn before and after the centrifugal separator may reveal no differences, and that this "can lead to the wrong interpretation and conclusion that the centrifugal separator is not working correctly."
That sentence is doing more work than it first appears. It is not saying the test is imprecise. It is saying the test is aimed elsewhere.
A centrifugal separator sorts by density difference and particle size — it throws denser, larger material outward and lets the oil pass. Routine analysis, meanwhile, usually reports viscosity, BN, water, insolubles, oxidation and inorganic constituents. Those two lists overlap far less than the layout of a report suggests.
Why does base number sit still across the separator?
Because separation does not act on it. CIMAC states that BN is normally not influenced by separation.
Base number measures the oil's reserve of alkalinity, held against acid attack. In a medium-speed engine, sulphuric acid formed from fuel sulphur reaches the oil in the cylinder or via blow-by into the crankcase, and is neutralised by the over-based additives, forming calcium sulphate. On high-sulphur heavy fuel oil, that calcium sulphate is often one of the major contaminants in the charge. BN falls because acid consumed the reserve — a chemical event in the engine, not a mechanical one in the bowl.
Oxidation is blunter still. CIMAC states that oxidation products are dissolved or dispersed in the oil and therefore cannot be removed by the separation process. A number that cannot move across a separator cannot report on one.
Viscosity behaves the same way, and for a stated reason: it is mainly influenced by contaminants that separation cannot fully remove.
Why does the insolubles number move so little?
This is the one that misleads hardest, because insolubles sound exactly like the thing a centrifuge exists to take out.
The test creates much of what it measures. An insolubles method adds a solvent — pentane or heptane — and weighs what precipitates. CIMAC's point is that only a small portion of that material was insoluble in the oil before the solvent was added, and only that small portion was ever available to be separated. The main fraction is not affected by the separation process at all.
The gap between the two is roughly an order of magnitude. In CIMAC's figures, pentane insolubles are usually in the order of 0.05–2.0%, while the separable sludge is in the order of 0.005–0.05% by weight.
So a separator can remove essentially all of the sludge available to it and still barely disturb the number the report prints.
Water is subject to a similar floor. CIMAC states that a reduction in water can only be observed once the water content in the oil is above approximately 0.2%. Below that, removal is happening where the routine figure cannot display it.
| What the routine report shows | What separation actually does to it |
|---|---|
| BN (base number) | Normally not influenced by separation |
| Oxidation products | Dissolved or dispersed — cannot be removed by separation |
| Viscosity | Driven mainly by contaminants separation cannot fully remove |
| Insolubles (pentane) | Only the small pre-existing fraction is separable — 0.005–0.05% against a reported 0.05–2.0% |
| Water | Reduction observable only above approximately 0.2% content |
| Inorganic constituents | Standard method reads particles below 5–10 µm; separation works best above 5 µm |
Figures and scope statements as given in CIMAC Recommendation No. 24 (2005), §5.5.6.
Which number actually demonstrates the separator is working?
CIMAC names one method: measuring the number of particles in relation to particle size.
The result explains the whole problem. Large particles — above roughly 5 micrometres — are usually removed very effectively, while removal efficiency for smaller particles is quite low. And CIMAC adds the line that turns this from a curiosity into a reason to care: bigger particles matter more, because bigger particles are more harmful.
Set that against how inorganic constituents are normally measured. CIMAC names ASTM D5185 for that purpose — citing its 2002 edition, this being a 2005 document — and notes that the method can only measure particles smaller than about 5–10 µm. Larger particles can be measured if the sample is ashed, but that is not done in routine analysis.
The separator is most effective on exactly the particles the routine method is least equipped to see. The instrument that would prove the machine is working is blind to its main achievement.
CIMAC is candid about the catch: particle counting by size is available in only a limited number of laboratories.
So what does genuinely govern how well the separator works?
Separation efficiency is real and variable — it simply is not what the routine report is showing. Three things move it.
Temperature moves it most. The recommended processing temperature for a lube oil cleaning system is 95 ± 2 °C, with a maximum not exceeding 100 °C so that any water present does not boil. The mechanism is Stokes' Law: raising temperature lowers both the density and the viscosity of the oil, which raises the settling velocity of suspended material.
The sensitivity is severe, and CIMAC prints it with two exclamation marks. Lowering separation temperature from 95 °C to 90 °C means throughput has to be reduced by 22% to maintain the same separation efficiency. Five degrees costs close to a quarter of the effective capacity.
Which machine is fitted also matters. A purifier runs a continuous water outlet held by a water seal that depends on a gravity disc — sized to oil density, oil viscosity, separation temperature and flow rate. A clarifier has neither water seal nor gravity disc, so no continuous water outlet; it accumulates water and solids at the bowl periphery and discharges intermittently under a timer or water-sensing device.
And some contaminants resist regardless. Fuel components are hard to remove because fuel and lube oil share compositional characteristics. Heavy components — mainly asphaltenes — can be removed to a degree, but CIMAC reports that investigations found separator efficiency at removing asphaltenes is very low, influenced in part by how dispersant the lubricant is. In that same context the recommendation notes that dispersancy improving compatibility with heavy fuel oil could carry a side effect: some loss in centrifugal separation efficiency.
Is there a setting the documentation can tell you to use?
No, and the absence is deliberate rather than an omission.
Auxiliary-engine cleaning systems are often arranged so one separator serves a single engine at a time, and discharge time may need shortening during the first hours of a cleaning phase to cope with high initial contaminant levels. On the interval itself CIMAC declines to generalise: "The appropriate discharge interval has to be found by trial and error."
The variables that decide it are the installation, the fuel in use, the water ingress rate actually present, and the separator's own manufacturer instructions — the recommendation is explicit that maintenance and operation should follow the separator manufacturer's recommendations.
Water ingress is worth isolating for its own reasons. CIMAC lists fresh-water routes into system oil as leakage from coolers, leakage from cylinder cooling jackets, condensation of blow-by vapour after the engine stops — and, notably, "poorly set or optimised centrifugal separators." Where water content is high, the recommendation warns that the overbased additive can precipitate as insoluble calcium carbonate, and that alkaline detergents and other additives are sensitive to depletion by water. Where water and oil form a stable emulsion that will not settle or centrifuge out, load-carrying capacity falls and the oil has to be renewed.
What can be settled in writing before the oil is ordered?
The following is our own recommendation to specifiers, not CIMAC guidance — the recommendation addresses treatment systems, not procurement documents.
A lubricant data sheet can be asked to state water separability, or demulsification, behaviour as a declared property with its test method named, rather than left to be discovered in service. Where a lubricant is offered on the strength of its heavy-fuel-oil compatibility, the same documentation can be asked what that dispersancy implies for separation efficiency, since the recommendation identifies the two as capable of pulling against each other. And because a base number means little without the method that produced it, the BN test method named on the certificate of analysis is worth matching to the method the laboratory will use for trending.
Those are questions a supplier can answer in writing before a charge is committed. Suppliers listing trunk piston and marine engine oils on Altonex Global can be asked for exactly that documentation through an RFQ, alongside the specification itself.