Hold your hand near a hot mug of coffee and you feel warmth. Blow gently across the same mug and it cools noticeably faster, even though the air temperature has not changed. The physics behind that everyday observation is convection: heat carried away from a surface by a moving fluid. The faster the fluid moves and the better it carries heat, the quicker the surface loses energy.
This article explains how to quantify that process. We will work through Newton's law of cooling, see why the convective heat transfer coefficient is the term that does all the heavy lifting, and run a clean numerical example you can reuse as a sanity check.
Why this calculation matters
Convection sets the pace for a huge fraction of thermal engineering. A CPU heat sink, an engine radiator, a building wall, a heat exchanger, a power transistor on a circuit board — each survives or overheats depending on how effectively a surrounding fluid strips heat away. If the calculation is off, parts run hot, efficiency drops, and service life shortens.
The difficulty is that convection is not a fixed material property the way thermal conductivity is. It depends on the fluid, on how fast that fluid moves, on the geometry of the surface, and on whether the flow is laminar or turbulent. A vertical plate in still air and the same plate in a 5 m/s wind can differ in cooling rate by an order of magnitude. Getting a defensible number means understanding what governs the convective heat transfer coefficient, not just plugging into a formula.
The core formula
The starting point is Newton's law of cooling, which gives the rate of convective heat transfer from a surface to a surrounding fluid:
Q = h * A * (T_s - T_inf)
Here Q is the heat transfer rate in watts, A is the surface area in contact with the fluid, T_s is the surface temperature, and T_inf is the temperature of the fluid far from the surface. The term h is the convective heat transfer coefficient, in W/m^2K. It is the proportionality constant that captures everything about how well the fluid carries heat away.
The equation itself is simple. The hard part is h. Unlike conductivity, it is not tabulated for a material — it has to be estimated for each situation. Typical ranges give a feel for the scale:
Free convection, gases: 2 - 25 W/m^2K
Free convection, liquids: 50 - 1000 W/m^2K
Forced convection, gases: 25 - 250 W/m^2K
Forced convection, liquids: 100 - 20000 W/m^2K
Boiling and condensation: 2500 - 100000 W/m^2K
In practice, h is found through dimensionless analysis. The Nusselt number Nu relates convective transfer to pure conduction across the same fluid layer:
Nu = h * L / k so h = Nu * k / L
where k is the fluid's thermal conductivity and L a characteristic length. The Nusselt number itself comes from empirical correlations built on the Reynolds number Re (flow inertia versus viscosity) and the Prandtl number Pr (momentum versus thermal diffusivity). For forced convection in a pipe, the Dittus-Boelter correlation is a common choice; for a flat plate or a cylinder in cross-flow, other correlations apply. Each is a fitted curve from experiment, valid only over the range it was measured.
A worked example
Consider a heated surface losing heat to surrounding air. The surface sits at T_s = 80 degC, the air is at T_inf = 20 degC, the convective heat transfer coefficient is h = 25 W/m^2K, and the exposed area is A = 0.5 m^2. Find the rate of convective heat loss.
Step 1 — find the temperature difference.
T_s - T_inf = 80 - 20 = 60 K
A difference in degrees Celsius equals the same difference in kelvin, so no conversion is needed for the gap itself.
Step 2 — apply Newton's law of cooling.
Q = h * A * (T_s - T_inf)
Q = 25 * 0.5 * 60
Q = 750 W
So the surface sheds 750 watts to the surrounding air. That single number tells you a great deal: it is the heat that some source — an electrical element, a fluid stream, a chemical reaction — must continuously supply just to hold the surface at 80 degC. Cut the air movement and h might fall to 5 W/m^2K, dropping Q to 150 W and letting the surface temperature climb. Add a fan and h could rise past 100 W/m^2K, and the same surface would shed several kilowatts. The geometry and temperatures stayed fixed; only h moved, and it moved the answer with it.
Common mistakes
Treating h as a material constant. It is the most frequent error. Thermal conductivity belongs to the material; the convective coefficient belongs to the whole situation — fluid, speed, geometry, and flow regime together. Copying an h value from an unrelated problem rarely survives scrutiny.
Mixing up the temperatures. T_inf is the bulk fluid temperature well away from the surface, not the temperature in the thin boundary layer touching it. Using a near-wall value collapses the driving temperature difference and underpredicts the heat transfer.
Ignoring radiation. Newton's law of cooling covers convection only. For a hot surface in air, radiative loss can be comparable, especially at higher temperatures and lower air speeds. If you need the total, add the radiative term separately.
Using a correlation outside its validity range. Dittus-Boelter, flat-plate, and cylinder correlations each carry limits on Reynolds number, Prandtl number, and geometry. Apply one outside its tested band and the Nusselt number — and therefore h — can be off by a wide margin.
Forgetting the flow regime. Laminar and turbulent flow give very different convective coefficients. A small change in velocity that pushes the flow across the transition can change h far more than the velocity change alone would suggest.
Try the interactive NovaSolver calculator
Estimating h by hand means chasing fluid properties and choosing the right correlation, which is exactly where errors creep in. The Forced Convection Heat Transfer Calculator on NovaSolver handles that for you: pick a fluid (water, air, engine oil, or liquid sodium) and a geometry (internal pipe, flat plate, or cylinder in cross-flow), set the velocity, characteristic length, and surface and fluid temperatures, and it returns the Reynolds number, Prandtl number, Nusselt number, the convective coefficient h, the heat flux, and the flow regime — with log-scale charts comparing all four fluids at once.
Related calculators
- Natural Convection Calculator — for the no-fan case, where buoyancy alone drives the flow and h is governed by the Rayleigh number.
- 1D Heat Conduction Calculator — for the conduction half of the problem, getting heat from inside a solid to the surface where convection takes over.
- Fin Efficiency Calculator — for extended surfaces that boost area A when h alone cannot move enough heat.
You can explore the full set in the thermal engineering tools hub.
Closing note
Newton's law of cooling looks almost too simple — heat rate equals h times area times temperature difference. The honesty of the formula is that it pushes all the complexity into one term. The convective heat transfer coefficient is where the fluid, the flow, and the geometry all meet, and it is where careful engineering pays off. Treat h as a quantity to be reasoned about rather than copied, match your correlation to the geometry and flow regime, and remember that radiation may be waiting alongside it. Get h right and the rest of the thermal analysis follows cleanly.
Top comments (0)