Stand in front of an open refrigerator and you feel cold air spill out, but hold your hand a few centimetres from the shelf and the chill is gentler. Switch on a fan blowing the same cold air and your hand feels icy almost at once. The air temperature has not changed. What changed is how vigorously the air carries heat away from your skin — and that vigour is exactly what the Nusselt number measures.
Convective heat transfer is hard to predict from first principles because it depends on a moving fluid, a boundary layer, and a surface all at once. Engineers handle it with dimensionless correlations, and the Nusselt number sits at the centre of that approach. This article explains what the Nusselt number is, how a correlation produces it, and how to turn it into the heat-transfer coefficient you actually need.
Why this calculation matters
Almost every thermal design includes a surface losing or gaining heat to a moving fluid: a CPU heat sink, a car radiator, a boiler tube, a building wall in the wind. The rate of that exchange is set by the convective heat-transfer coefficient h, and h is notoriously difficult to compute directly. It depends on velocity, fluid properties, surface geometry, and whether the flow is laminar or turbulent.
The Nusselt number is the device that makes the problem tractable. Decades of experiments have been distilled into correlations that give Nu as a function of the Reynolds and Prandtl numbers. Once you have Nu, the heat-transfer coefficient follows from a single algebraic step. Get the Nusselt number right and the rest of the thermal calculation — the heat duty, the surface temperature, the required fin area — falls out cleanly. Get it wrong and a heat exchanger ends up undersized or a hot component runs past its temperature limit.
The core formula
The Nusselt number is the dimensionless ratio of convective heat transfer to conductive heat transfer at a surface:
Nu = h * L / k
Here h is the convective heat-transfer coefficient in W/m^2K, L is a characteristic length in metres, and k is the thermal conductivity of the fluid in W/mK. Because the numerator and denominator carry the same units, Nu is a pure number.
The physical reading is direct. Imagine a stagnant layer of the same fluid conducting heat across the length L. The Nusselt number tells you how much more effectively the actual moving fluid transfers heat than that hypothetical motionless layer would. A Nusselt number near 1 means the fluid is barely moving and conduction dominates. A large Nusselt number means convection is carrying heat far more effectively than conduction alone.
In practice you rarely measure Nu — you predict it from a correlation. For turbulent flow inside a pipe, the widely used Dittus-Boelter correlation gives:
Nu = 0.023 * Re^0.8 * Pr^n
where Re is the Reynolds number, Pr is the Prandtl number, and n is 0.4 for heating the fluid or 0.3 for cooling it. Other geometries — flat plates, cylinders in cross-flow, banks of tubes — have their own correlations, but the workflow is identical.
Once a correlation has delivered Nu, the definition is rearranged to extract the coefficient you came for:
h = Nu * k / L
That single rearrangement is the whole point of the Nusselt number. It converts a dimensionless experimental result into the dimensional quantity that heat-transfer calculations actually use.
A worked example
Consider forced air flow over a flat plate. Suppose a correlation has already delivered a Nusselt number of Nu = 100. The air has a thermal conductivity of k = 0.026 W/mK, and the characteristic length of the plate is L = 0.5 m. Find the convective heat-transfer coefficient.
Step 1 — recall the definition and rearrange it. The Nusselt number is Nu = h*L/k, so the coefficient is:
h = Nu * k / L
Step 2 — substitute the values.
h = 100 * 0.026 / 0.5
Step 3 — evaluate.
h = 2.6 / 0.5 = 5.2 W/m^2K
So this forced-air flow over the plate produces a convective coefficient of 5.2 W/m^2K. That number can now drive the rest of the analysis: multiply by the surface area and the temperature difference between the plate and the air, and you have the convective heat rate.
It is also worth reading the Nusselt number itself. A value of 100 means the moving air transfers heat about a hundred times more effectively than a still air layer of the same thickness would by conduction alone. A Nusselt number near 1 would mean conduction dominates and the fluid is doing almost nothing; a value of 100 signals vigorous convection.
Common mistakes
Using the fluid's conductivity, not the solid's. The k in Nu = h*L/k is the thermal conductivity of the fluid in the boundary layer — the air, water, or oil doing the convecting — not the conductivity of the wall material. Reaching for the metal's conductivity is a common and large error.
Choosing the wrong characteristic length. L must match the length used to define the correlation. For pipe flow it is usually the inside diameter; for a flat plate it is the distance along the surface; for a cylinder in cross-flow it is the outside diameter. Substituting an inconsistent L corrupts both Nu and the resulting h.
Applying a correlation outside its valid range. The Dittus-Boelter correlation is for fully developed turbulent flow within stated Reynolds and Prandtl bounds. Feed it a laminar Reynolds number and the Nusselt number it returns is meaningless. Always check the range a correlation was fitted for.
Confusing the Nusselt number with the Biot number. The two look almost identical — both are h*L/k. The crucial difference is which conductivity appears: the Nusselt number uses the fluid conductivity, while the Biot number uses the solid conductivity. They answer different questions and are not interchangeable.
Forgetting the heating-versus-cooling exponent. In correlations like Dittus-Boelter, the Prandtl exponent n changes depending on whether the fluid is being heated or cooled. Using the wrong value shifts the predicted Nusselt number, and therefore h, in the wrong direction.
Try the interactive NovaSolver calculator
Working a single Nusselt number by hand is straightforward, but seeing how it responds across a range of flow conditions is where intuition is built. The Nusselt Correlation Simulator on NovaSolver evaluates the Dittus-Boelter correlation for turbulent pipe flow: set the Reynolds number, the Prandtl number, the fluid thermal conductivity, the pipe diameter, and whether the fluid is being heated or cooled, and it returns the Nusselt number, the heat-transfer coefficient h, the wall heat flux, and the thermal boundary-layer thickness, with a live Re-Nu log-log chart.
Related calculators
- Forced Convection calculator — apply Nusselt correlations to externally driven flows over plates and tubes.
- Natural Convection calculator — for buoyancy-driven flows, where the Nusselt number depends on the Rayleigh number instead.
- 1D Heat Conduction calculator — the conduction side of the picture, useful for comparing the two transport modes the Nusselt number balances.
You can browse the rest in the thermal calculators hub.
Closing note
The Nusselt number is the hinge that connects experimental heat-transfer data to working design numbers. It is just a ratio — convection over conduction — but that ratio lets a correlation deliver a dimensionless result that a single algebraic step turns into the coefficient h. Predict Nu from the right correlation, keep the fluid conductivity and the characteristic length consistent, and the convective part of any thermal design becomes a calculation you can defend rather than a guess.
Top comments (0)