—
The bar is logarithmic. It has to be: at −20 °C a gear oil is five orders of magnitude thicker than petrol, and on a linear scale every other fluid would be an invisible sliver.
—
Pressure drop, roughness and fittings live in the pipe flow calculator — this panel is only about which regime the fluid is in, because that is what viscosity decides.
Click a column to sort. Greyed rows are outside that fluid’s valid temperature range — frozen, boiling, or below its pour point — and the tool will not quote a number it does not believe.
| Fluid | ν cSt | μ cP | ρ kg/m³ | SG | ν40/ν100 | ISO VG |
|---|
Every oil datasheet quotes kinematic viscosity at 40 °C and 100 °C. Type yours in and it joins the chart and the table.
—
Pump curves, pressure drop and Reynolds number want kinematic (ν, cSt). Force, torque and shear calculations want dynamic (μ, cP). They differ only by density: μ = ν·ρ. Both are shown here at whatever temperature you pick, because quoting either without a temperature is meaningless.
The number is the oil’s nominal kinematic viscosity in centistokes at 40 °C, ±10%. So an ISO VG 46 sits between about 41 and 51 cSt at 40 °C and says nothing directly about its viscosity anywhere else. SAE grades are a different scheme measured at different temperatures.
Because doing it properly needs data this tool does not have. ASTM D2270 requires tabulated L and H values for ν₁₀₀ below 70 cSt; the published polynomial form is only valid above 70 and returns a negative L for an ISO VG 46, which is nonsense. Rather than print a number dressed up as a standard, the tool shows the raw ν₄₀/ν₁₀₀ ratio — the same information, without the false authority.
Each fluid uses the relation that fits it. Oils use ASTM D341, the industry’s own, fitted through the two points every datasheet quotes. Water and seawater use a Vogel correlation, within 2% of published values from 0 to 100 °C. Thin liquids use Andrade, within 1% for ethanol and 5% for mercury. They are exact at 40 and 100 °C by construction and best near them.
Because the fluid does. Each one carries its own valid range — freezing and boiling points, pour points, spec limits — and outside it no correlation means anything. Water stops at 0 and 100 °C; heavy fuel oil does not start until 30 °C, because below that it is a solid rather than a slow liquid.
No. Use the actual product datasheet. These are representative values for common fluid types, and real products vary with additive package, batch and age — honey, olive oil and heavy fuel oil especially. This is for understanding and sizing intuition, not for signing anything off.
Mercury is thirteen times denser than water and flows just as freely. Honey is barely denser than water and will not pour. Density is mass per volume; viscosity is internal friction. Nothing links them, which is why the tool shows both and why μ = ν · ρ is a conversion rather than a relationship.
Every oil is roughly the same when hot — that is what the grade is chosen to guarantee. SAE 30 and 5W-30 are identical at 100 °C. At −20 °C the monograde is about nine times thicker. Everything a multigrade oil is for happens at the cold end of this chart, not the hot one.
Push a viscous fluid through a pipe in laminar flow and the rate goes as the fourth power of the bore. Halve the diameter and you get a sixteenth of the flow for the same pressure. It is why a slightly undersized hydraulic line is not a slightly slower one, and why warming the oil is usually cheaper than a bigger pump.
“ISO VG 46” means 46 cSt at 40 °C, ±10%. Quoting a viscosity without its temperature says almost nothing — over the range on this page the same oil varies by a factor of a thousand. If a spec gives one number and no temperature, it is not a spec.
Viscosity against temperature is not one curve family. A lubricating oil and a glass of water disagree so completely about the shape that any correlation good for one is visibly wrong for the other. This tool picks the relation per fluid, by a mechanical rule rather than by taste, and tells you at the top of the chart which one you are looking at.
The petroleum industry's own viscosity–temperature relation, and the one every oil datasheet is built on:
log₁₀ log₁₀ (ν + 0.7) = A − B · log₁₀ T
Two points give A and B, which is exactly what a datasheet provides. The double logarithm is not decoration. A single exponential term cannot hold its shape across 170 °C, and the failure is not subtle: fitted through 40 and 100 °C and extrapolated to −20, a single-term fit reports a hydraulic ISO VG 46 as 770 cSt where D341 gives 4,518. That is not a rounding difference. It is the difference between an oil that will pump on a cold morning and one that will not.
That error is what this page used to ship. The sharpest way to see it: at −20 °C D341 makes SAE 30 about nine times thicker than 5W-30, and that gap is the entire reason multigrade oil exists. The old fit showed 3.4×, quietly understating the point of the product category.
Water is the most anomalous common liquid there is, and no petroleum correlation fits it. Fitted through the same two anchors, both D341 and a single-term fit come out 22–23% low at 0 °C. So water and seawater use the Vogel-type correlation instead, which holds to within 2% from 0 to 100 °C against published values.
Its density is not a constant either. Water uses Kell's equation, which is a rational function rather than a polynomial — a plain fifth-order polynomial tried here was 11 kg/m³ out at 100 °C, and since μ = ν·ρ a wrong density silently corrupts every dynamic viscosity on the page. Kell also reproduces the thing that makes water strange: the density maximum at 3.98 °C, not at freezing. No expansion coefficient can do that, which is why every other fluid here gets one and water does not.
Petrol, kerosene, ethanol, mercury and glycol coolant are simple low-viscosity liquids, and simple liquids genuinely do follow Arrhenius behaviour. Checked against published values, this is within 1% for ethanol and 5% for mercury. D341 is also mathematically undefined below 0.3 cSt, since log₁₀(ν + 0.7) goes to zero there and the outer logarithm diverges — so the thin fluids could not use it even if it fitted.
The old version of this page let one slider run −20 to 150 °C for everything, so it would cheerfully quote you a viscosity for water at 150 °C and mercury at −20. Each fluid now carries its own range — freezing and boiling points, pour points, spec limits — and outside it the tool greys the row and refuses the number. Heavy fuel oil starts at 30 °C because that is roughly its pour point: below that it is not a slow liquid, it is a solid, and no correlation describes it.
Two-point fits, and a fit is only as good as its two points. No additives and no shear thinning — a multigrade's VI improver behaves differently in a loaded bearing than in a viscometer, and that difference is the entire specialism of lubricant engineering. Everything is at 1 atm; viscosity rises steeply with pressure and none of that is here. Honey, olive oil and heavy fuel oil vary enormously batch to batch and should be read as representative rather than specific.
Viscosity Index is deliberately absent. ASTM D2270 needs tabulated L and H values for ν₁₀₀ below 70 cSt, and the published polynomial form is only valid above 70 — it returns a negative L for an ISO VG 46, which is nonsense. Rather than print a number dressed up as a standard it does not implement, the tool shows the raw ν₄₀/ν₁₀₀ ratio, which is the same information without the false authority.
This page has changed in 11 archived releases, but each of those was a site-wide pass, so none is attributable to this tool on its own and none is listed here. That is not a claim that the tool never changed — a release that reworked many pages at once may well have altered this one too. The changelog has them.