Thermal convection is heat carried by the bulk movement of a fluid, and the rate at which a surface hands heat to the fluid washing it is Q = h · A · (Ts − Tinf) — Newton's law of cooling. The quantity that does the work is h, the convective heat transfer coefficient in W/(m2·K): the identity of the fluid and the briskness of its flow are both buried inside that single number, and it is the one figure a reader usually has to estimate. Note the slug: this page is convection at a surface, while /calculators/thermal-conduction is conduction through a slab of known thickness — one letter apart, two different formulas. This free calculator solves the relation for the heat rate, the coefficient, the area or either temperature, names the band h belongs in, and shows every step.
Each button writes the coefficient, the area and the two temperatures into the four boxes, so the answer changes the moment you press it. The line underneath quotes back whatever the widget then works out, so nothing in it is stored text. All five sets of figures are stated givens chosen to make a point, not measurements of any named object.
Pick a case above, or type your own figures.

The thermal convection calculator is a free online tool for the rate at which a surface hands heat to the fluid moving past it: Q = h · A · (Ts − Tinf), in watts. Give it the convective heat transfer coefficient h, the wetted area A and the two temperatures, and it returns the heat rate; point the Solve for menu at any one of those four instead and it runs the same relation in that direction.
Read the slug before you read the formula. This page is convection — Newton's law of cooling, at a surface washed by a fluid. /calculators/thermal-conduction is conduction, Fourier's law through a solid slab of known thickness, and the two addresses differ by a single letter. If your problem gives you a thickness and a conductivity rather than a coefficient, you are on the wrong one of the two.
The quantity that does all the work here is h, and it is not a constant. Its size is set by how fast the fluid is moving and by which fluid it is — neither of which this formula carries — so this page gives it as indicative, order-of-magnitude bands rather than as a looked-up number: still air about 2 to 25, moving air about 10 to 200, water about 50 to 10000 W/(m²·K). A chip beside the answer names which of those bands your figure falls in, and because the bands overlap, two names at once is a real answer rather than a fudge.
In natural convection h grows with the temperature difference, which makes the formula only approximately linear. Take the temperature difference on a 0.30 m isothermal plate in air up eightfold, from 10 K to 80 K, and the heat rate at 1 m² climbs not eight times but 13.356 times, from 33.2 W to 443 W — because h drifts from 3.32 to 5.54 W/(m²·K), a factor of 1.670, while you do it. Anyone who treats the coefficient as a looked-up constant is wrong by that gap, and the five-quarters power rule that nearly describes it is an asymptote, not an identity.
Both temperature boxes carry one scale on purpose, and nothing on this page adds 273.15. The formula uses only the difference between the two temperatures, so a surface at 60 °C into a fluid at 20 °C gives exactly the same heat rate as 333.15 K into 293.15 K — both pairs differ by 40. A kelvin figure may therefore be typed straight in, provided both boxes use it; a half-converted pair would be wrong by 273.15 and the page would have no way to tell.
A negative heat rate is kept and labelled rather than hidden. It means the fluid is the warmer of the two and the surface is gaining heat: at h = 25 W/(m²·K) over 1.5 m², a surface at 20 °C in a fluid at 60 °C reads −1500 W. The chip labelled direction of heat flow turns that sign into words, so it cannot be read as a slip.
Some states have no answer and the page refuses them rather than inventing one. With the two temperatures equal the heat rate is an exact 0 W, which is a result — but asking the same state for h or for the area puts that zero difference in a denominator, and the panel says so instead of printing an infinity. An area of zero and a result below absolute zero are refused the same way. A coefficient of zero is not: no convection at all is a real state rather than a missing one, so that case returns an honest 0 W and the band chip says exactly that.
What the page does not do is worth naming as plainly as what it does. It has no time input, so it says nothing about how long an object takes to cool or about the shape of that approach; it carries no U-value and no multi-layer wall; and it does not work h out from a geometry or a flow speed, because that is the hardest part of the problem and hiding it behind three boxes would be worse than asking for it. It also assumes the fluid is moving at all: a wide layer of still air heated from below does not convect until the Rayleigh number passes about 1708, and until then heat crosses it by conduction only.
| Symbol | Quantity | Default unit | Also accepts | Example value |
|---|---|---|---|---|
| Q | Convective heat rate | W | kW | 1500 |
| h | Convective coefficient | W/m²·K | kW/m²·K | 25 |
| A | Area wetted by the fluid | m² | cm² | 1.5 |
| Ts | Surface temperature | °C | — | 60 |
| Tinf | Bulk fluid temperature | °C | — | 20 |
Q = hA(Ts − Tinf). If your problem hands you a thickness and a conductivity instead of a coefficient, you want the thermal conduction calculator — one letter away in the address bar, and a different formula.h, the area, the surface temperature and the fluid temperature; whichever you choose becomes the headline and the other four become your input boxes. Switching the menu refills every box with this page's defaults, which are a consistent set.25 W/(m2·K), and the box menu also offers the kilowatt option kW/m2·K. This is the input most often guessed, so a chip names the indicative band it falls in instead of implying there is one correct value.1.5 m2, with cm2 on the menu. Use the area actually touched by the moving fluid, not the whole outside of the object, and keep it strictly positive — zero area is no surface at all.60 for the surface and 20 for the fluid. Both boxes carry one scale deliberately: the formula uses only the difference, so kelvin figures work provided both boxes use them, and nothing here adds 273.15.1500 W. A minus sign is kept rather than hidden: it means the fluid is the warmer of the two and the surface is gaining heat.0 W in the heat-rate direction, which is an answer. Ask the same state for h or the area and the difference is now a denominator: the panel refuses rather than printing an infinity.Two neighbouring quantities are worth separating from this one before you start. The chip labelled heat flux is this page's heat rate divided by its area, and if that is the quantity you actually want on its own, the heat flux calculator does it directly. And below the onset of convection, a still fluid layer passes heat by conduction only, which is Fourier's law rather than anything on this page.
If you want the mechanism argued out in prose, with the threshold and the worked problems, read the thermal convection guide, which this tool is the arithmetic half of. For how convection sits beside the other two ways heat moves, the guide to conduction, convection and radiation is the comparison page, and it is where the three-way contrast belongs rather than here.
h = 25 W/(m2·K), A = 1.5 m2, Ts = 60 °C and Tinf = 20 °C. These were chosen because the arithmetic is exact: a 40 K difference, 37.5 W/K of conductance and a round 1500 W.The table holds the coefficient at 25 W/(m2·K) and the area at 1.5 m2, keeps the fluid at 20 °C, and moves the surface temperature alone. Every headline and chip cell was read out of the running widget rather than worked out by hand. Where a cell and the tool ever part company, believe the tool.
| Surface temperature | Difference chip | Headline heat rate | Flux chip | Direction chip |
|---|---|---|---|---|
| 20 | 0.00 K | 0 | 0 W/m² | no net transfer |
| 25 | 5.00 K | 187.5 | 125 W/m² | surface to fluid |
| 30 | 10.00 K | 375 | 250 W/m² | surface to fluid |
| 40 | 20.00 K | 750 | 500 W/m² | surface to fluid |
| 50 | 30.00 K | 1125 | 750 W/m² | surface to fluid |
| 60 | 40.00 K | 1500 | 1000 W/m² | surface to fluid |
| 0 | -20.00 K | -750 | -500 W/m² | fluid to surface |
Three things are worth reading off that table. The heat rate is exactly proportional to the difference while h is held fixed, so doubling 20 K to 40 K doubles 750 W to 1500 W. The first row is the only one where the answer is an exact zero, and the last row is the only one where the direction chip changes its mind.
That strict proportionality is also the table's limitation, and it is worth saying plainly. It holds because h was typed in once and never updated, which is what every substitution into this formula quietly assumes. In real natural convection the coefficient grows with the difference, so the right-hand end of a table like this understates the heat rate.
Switching the Solve for menu runs the same row backwards. From 1500 W with A = 1.5 m2, 60 °C and 20 °C the coefficient mode returns 25 W/(m2·K); from the same heat rate with h = 25 the area mode returns 1.5 m2; and with both given, the two temperature modes return 60 °C and 20 °C.
| Symbol | What it is | Unit | On this page |
|---|---|---|---|
| Q | The convective heat rate: watts, not joules. Negative means the surface is gaining heat | W | Box menu reads W or kW. Headline 1500 at the defaults, -1500 with the two temperatures swapped, an exact 0 when they are equal. Default 1500 when it is an input. |
| h | The convective heat transfer coefficient: how many watts cross each square metre for each kelvin of difference | W/(m2·K) | Box menu reads W/m2·K or kW/m2·K. 25 is the default. Indicative bands only: still air about 2 to 25, moving air about 10 to 200, water about 50 to 10000. |
| A | The area of the surface actually in contact with the moving fluid. Strictly positive | m2 | Box menu reads m2 or cm2. 1.5 m2 is the default, which is 15000 cm2. |
| Ts | The temperature of the surface itself, not of the body behind it | °C | One scale only, on purpose. 60 °C is the default; a kelvin figure may be typed straight in provided the fluid box matches it. |
| Tinf | The temperature of the bulk fluid, far enough away to be unaffected by the surface | °C | 20 °C is the default. The subscript is short for infinity — the conventional mark for far from the surface, which is where this temperature must be measured. |
| ΔT = Ts − Tinf | The temperature difference: the only way the two temperatures enter the formula | K | A chip beside the answer, never an input: 40.00 K at the defaults. A kelvin and a degree Celsius are the same size, so a difference needs no conversion. |
| q = Q / A | The heat flux: the same heat rate expressed per square metre of surface | W/m2 | A chip beside the answer: 1000 W/m2 at the defaults, 35000 W/m2 in the water case. |
Written out, Q = hAΔT looks like a law. It is closer to a definition: everything difficult about the problem — how fast the fluid moves, how thick the sluggish layer against the surface is, whether the flow is smooth or churning — has been collected into the single coefficient h. The formula is exact once you know h, and knowing h is the whole task.
That matters most in natural convection, where the fluid moves only because the heat has made it buoyant. Raise the temperature difference and you do not merely multiply the driving term: you also speed the fluid up, which thins the layer against the surface and raises h itself. So the heat rate grows faster than the difference does.
The size of that effect is worth a number. For an isothermal vertical plate 0.30 m tall in air, raising the difference from 10 K to 80 K raises h from 3.32 to 5.54 W/(m2·K) — a factor of 1.670 — and so raises the heat rate at 1 m2 from 33.2 W to 443 W. That is 13.356 times, where a fixed h would have predicted 8. The heat rate grows roughly as the five-quarters power of the difference rather than in proportion to it, and roughly is the honest word: the correlation behind those figures is only asymptotically a power law.
Force the fluid along with a pump or a fan and the picture changes, because the speed no longer comes from the heat. That is the useful half of the natural-versus-forced distinction: a pumped flow is much closer to the constant h the formula appears to promise. The forced-flow side of the story, where the transition to churning flow is set by the speed rather than the temperature, is the Reynolds number simulator's ground rather than this page's.
water because 500 is above the moving-air band entirely.h as a given, and h is the hard parth understates the answer.h is what gets inferred, which is what the Solve for menu is really for. Measure 150 W off a 0.25 m2 plate sitting 60 K above its air and the coefficient mode returns 10 W/(m2·K), which the band chip places in still or barely moving air. That is the honest direction of the inference: the rate is the observation and the coefficient is the conclusion.h = 25 W/(m2·K) and a 40 K difference, 1500 W needs 1.5 m2. Halve the coefficient, as a still room rather than a draughty one would, and the same rate needs twice the area — which is why fins exist.
-1500 W, and the direction chip reads fluid to surface: the surface is gaining heat. The sign is the answer here, which is why it is not hidden behind a magnitude.For the definition, the threshold argued out in prose and a full set of worked problems, read the thermal convection guide, which this tool is the arithmetic half of. The guide to conduction, convection and radiation sets the three mechanisms side by side, the Fourier's law guide covers the conduction case a still layer falls back on, and the heat flux guide covers the per-square-metre reading this page prints as a chip.
Five tools sit beside this one. The thermal conduction calculator does Fourier's law through a slab; the heat flux calculator does watts per square metre on its own; the thermal diffusivity calculator does how fast a temperature change spreads; the heat transfer simulator puts all three mechanisms on one screen; and the Fourier's law simulator runs the conduction half interactively. The full physics lab library and the calculator index are open too.
It works out the convective heat rate between a surface and the fluid washing it, from Q = hA(Ts - Tinf), where h is the convective heat transfer coefficient in W/(m²·K), A is the wetted area in square metres and the two temperatures are in degrees Celsius. The same relation is rearranged four more ways, so you can enter any four of the five quantities and read the fifth. The answer is a rate in watts, not an amount of energy in joules.
It is Q = hA(Ts - Tinf), usually called Newton's law of cooling. Q is the heat rate in watts, h the convective heat transfer coefficient in W/(m²·K), A the area in contact with the fluid and Ts - Tinf the temperature difference between the surface and the bulk fluid. It is a definition of h as much as a prediction of Q, which is why h is where all the awkward physics of the flow ends up.
Watts per square metre per kelvin, written W/(m²·K). Read the unit as the answer to a question: how many watts cross each square metre for each kelvin of temperature difference. Because a kelvin and a degree Celsius are the same size, the unit is unchanged whether the difference is quoted in one or the other.
Either, as long as both boxes use the same scale, because the formula only ever uses the difference between them. A surface at 60 and a fluid at 20 degrees Celsius gives exactly the same heat rate as 333.15 and 293.15 kelvin, since both pairs differ by 40. The boxes are labelled in degrees Celsius and nothing on this page adds 273.15, so a mixed pair would be wrong by that amount and the page would have no way to tell.
In natural convection the fluid moves because the heat itself makes it buoyant; in forced convection a pump or a fan moves it regardless. The formula on this page is the same for both, but the value of h is not: forcing the fluid along raises it, often by an order of magnitude. The deeper consequence is that natural-convection h depends on the temperature difference, while a pumped flow is much closer to the constant the formula appears to assume.
Because in natural convection the fluid speed is produced by the temperature difference, so raising the difference drives a brisker flow, and that brisker flow leaves a thinner sluggish layer for the heat to cross. On a 0.30 m vertical plate in air, raising the difference from 10 K to 80 K raises h by a factor of 1.670 and the heat rate by a factor of 13.356 rather than the factor of 8 a constant h would predict. That is why this page gives h as indicative bands and says what it is doing.
Yes, and it is a result rather than an error. A negative Q means the fluid is warmer than the surface, so the heat is flowing from the fluid into the surface: at h = 25, A = 1.5 m², a surface at 20 and a fluid at 60 degrees Celsius the answer is -1500 W. The chip beside the answer turns that sign into words so it cannot be misread.
For a fluid layer heated from below there is a genuine threshold, and below it nothing moves: the Rayleigh number has to exceed about 1708, and heat crosses by conduction alone until it does. With 15 K across a still-air gap that means a gap narrower than about 10.4 mm cannot convect at all. The same 15 K in water convects down to about 2.01 mm, which is why an air gap insulates and a water gap does not.
Because working h out needs the geometry, the fluid properties and the flow regime, and a page that invented one from three boxes would be hiding the hardest part of the problem. What this page does instead is take h as a stated input, name the band it belongs to, and solve the relation in any of its five directions. The simulator linked from this page is where the coefficient itself is built up from the fluid and the temperature difference.