The drum of ISO VG 46 ran out on a Friday. The maintenance team topped the reservoir up with ISO VG 150 because that is what the store had, and the arithmetic looked harmless: 80 litres of 46 plus 20 litres of 150 is 67 cSt on paper and 57 cSt by the blending number, either of which is close enough to 68, which is close enough to 46 for a hydraulic system that is not doing anything clever.
Two things were wrong with that calculation. The viscosity blending number says the mixture is 60 cSt, not 66. And viscosity was never the risk. The risk was the 200 millilitres of silicone antifoam in the VG 150, at a concentration that would not register on any specification sheet and would quietly take the reservoir's air release time from four minutes to fourteen.
Three kinds of incompatibility, three different tools
Buyers and maintenance engineers use one word for three unrelated problems, which is why arguments about "compatible" oils never converge. Sort them first, and each one gets its own answer.
| Type | What actually happens | Can it be calculated? | How it is settled |
|---|---|---|---|
| Physical | Viscosity, density and flash point of the blend differ from either component | Yes, from the blend rules | Arithmetic, then confirmation by ASTM D445 |
| Chemical | Additives from the two oils react, flocculate or drop out as sediment | No | ASTM D7155 mix test at realistic ratios, plus filtration and ICP |
| Material | Seals, hoses, paints and coatings swell, shrink, harden or soften | No | ASTM D471 immersion of the actual elastomer |
Most mixing disasters in the field are type two or three. Almost all of the discussion beforehand is about type one, because it is the only one that fits on a calculator.
The viscosity blend number, and why the arithmetic average lies
Viscosity does not blend linearly. The industry standard workaround is the Refutas viscosity blending number, VBN:
VBN = 14.534 x ln(ln(viscosity + 0.8)) + 10.975
Convert each component to VBN, average the VBNs by mass or volume fraction, then convert back. Doing that on the two oils in the opening example:
| Mixture | Arithmetic average (cSt at 40 C) | VBN blend (cSt at 40 C) | ISO VG band |
|---|---|---|---|
| 100% VG 46 | 46 | 46 | VG 46 (41.4 to 50.6) |
| 95% VG 46 + 5% VG 150 | 51.2 | 48.4 | Still inside VG 46 |
| 50% VG 46 + 50% VG 150 | 98 | 80 | Between VG 68 and VG 100 |
| 100% VG 150 | 150 | 150 | VG 150 (135 to 165) |
Two lessons fall out of the table. The first is that a routine 5% top-up does not move a fluid out of its grade even when the top-up oil is three times as viscous, so the panic about a small cross-contamination is usually misplaced. The second is that losses grow with the viscosity ratio: at 50:50 the arithmetic average overstates the blend by 18 cSt, which is the difference between ordering a VG 100 and receiving something closer to a VG 68. The bigger the ratio, the worse the linear approximation, and the more it matters, because a hydraulic system specified for VG 46 and running at VG 80 will lose volumetric efficiency, run hotter and leak at every worn clearance.
Sediment: the part that has to be measured
ASTM D7155 exists because the chemical half of the problem has no formula. Two oils can be individually excellent and mutually destructive: calcium sulfonate, calcium salicylate and calcium phenate detergents are not interchangeable in a mixture, and combinations of detergent and dispersant chemistry that are stable separately can form a fine suspension or a hard deposit when blended. Additive packages can also be diluted below the level at which they function, which is the quiet version of the same failure - no sediment, no visible change, and an oil that oxidises twice as fast as either parent.
The test is deliberately simple, which is why there is no excuse for skipping it. Blend the two oils at the ratios that reflect what will actually happen, hold them at an elevated temperature, and look. The ratios in common use are 50:50, 95:5 and 5:95; the 5:95 case is the one that corresponds to a top-up, and it is the one most often omitted. Observe appearance, haze, separation and sediment after settling, and if the application justifies it, filter the blend and weigh the residue. A blend that turns hazy in 24 hours at 70 C is not going to behave better in a 2,000-litre reservoir over a year.
Residual old oil matters here because the drain is never complete. A hydraulic reservoir, a gearbox sump and a long pipe run retain something in the region of 5 to 10% of the charge, so the real ratio in the machine is rarely the ratio in the test bottle. When a change of product is planned, the test should include the old oil at its realistic residual percentage, not at zero.
The antifoam problem, and other ppm-level contaminants
Silicone antifoams are effective at tens of parts per million, which is precisely why they are dangerous in a shared system. A drum that contains 100 ppm of polydimethylsiloxane is, to the next oil, a 100 ppm contaminant. Silicone is also surface active in the wrong way: it can suppress foam in the donor oil while degrading air release in the recipient, because the mechanisms are different. Foam is measured by ASTM D892; air release by ASTM D3427; and a fluid that passes one and fails the other is a real and common outcome after a top-up from the wrong drum.
The same logic applies to other trace additives. A tackifier or friction modifier carried over at a few hundred ppm changes the friction characteristics of a fluid whose specification says nothing about friction. A dye carried over is cosmetic. A rust inhibitor is not.
Seals follow polarity, not viscosity
Elastomer swell is driven by how much the oil and the rubber like each other, which is a polarity and aromaticity question rather than a viscosity question. Moving a system from a mineral oil with a moderate aromatic content to a highly refined white mineral oil or a synthetic base stock reduces the aromatic fraction, and the seal that was comfortably swollen at, say, plus 8% can shrink toward plus 2%, at which point a worn shaft seal starts to weep. Move in the other direction, into an ester, and the same seal can swell far more than intended, soften and extrude.
This is measurable, and it is measured with ASTM D471: immerse a specimen of the actual seal compound - NBR, FKM, EPDM, PTFE - in the candidate fluid at the temperature and duration that represent service, typically on the order of 70 hours at 100 C, and record volume swell, hardness change and tensile change. Generic compatibility charts are a starting point and nothing more, because compound formulation within one elastomer family moves the answer by several percent. An EPDM seal and a mineral oil are a bad pair regardless of viscosity; an ester and an EPDM seal are worse.
What to re-measure after a blend
| Property | Method | Why it is on the list |
|---|---|---|
| Kinematic viscosity at 40 C | ASTM D445 | Confirms the blend number, and catches additive-induced thickening |
| Flash point | ASTM D92 | A low-flash component in a high-flash oil is a fire risk carried silently |
| Pour point | ASTM D97 | Wax interaction can raise it well above either parent |
| Demulsibility | ASTM D1401 | Mixed emulsifier systems are the classic cause of stable emulsions |
| Foam and air release | ASTM D892, D3427 | The ppm-level contamination described above shows up here first |
| Filterability | ISO 13357 | Catches gels and floc that visual inspection misses |
| Cleanliness | ISO 4406 | Sediment that passes a 40 micron eye still blocks a 3 micron servo valve |
| Additive elements | ICP, with FTIR for chemistry | Shows whether zinc, phosphorus and calcium survived at the expected ratio |
The procedure that costs one afternoon
- Blend 500 mL at the realistic ratio, including residual old oil at 5 to 10%.
- Hold at 70 C for 24 hours, then inspect for haze, separation and sediment; filter and weigh if the application is critical.
- Measure viscosity, flash point, demulsibility and air release on the blend, not on the components.
- Immerse the actual seal compound for 70 hours at service temperature and compare swell against the incumbent oil.
- If the blend passes, keep 500 mL sealed as a reference and record batch numbers of both oils. If a problem appears in six months, that bottle is the only way to find out which drum was in the machine.
The rule that comes out of all of this is unglamorous: mix by test, not by grade. Two oils of the same ISO grade and the same API base oil category can still be chemically incompatible, and two oils of different grades can be perfectly miscible. The grade predicts the viscosity of the blend reasonably well once you use the blending number. It predicts nothing else.
Supply
Our base stocks and additives cover the three families that most often meet in a mixed system. Mineral: white mineral oil 5#, 10#, 22# and 32#, all CAS 8042-47-5. Finished hydraulic fluids: HD HM T10, HD HM T22 and HD HM T100. Esters and polyol intermediates: butyl oleate (142-77-8), butyl stearate (123-95-5), pentaerythrityl tetraisononanoate (93803-89-5), pentaerythritol (115-77-5) and dipentaerythritol (126-58-9). Slip and lubricity additives: oleamide (301-02-0) and stearamide (124-26-5). Waxes: paraffin wax 60 (8002-74-2).
Top-ups and changeovers sit next to two other problems worth reading together: refractometer factors and dilution and tramp oil, fines and filter ratings. This article is about two oils meeting; those two are about one oil degrading inside a system. Technical data sheets, typical COAs and blending guidance are available through the enquiry form on any product page. If you are planning a product changeover, tell us the incumbent oil and the seal compound and we will tell you which of the three compatibility questions actually applies to your machine.