Convection

Look around a heated room and you are looking at deliberate convection engineering. Radiators are mounted low and air conditioners high, for the same reason in mirror image: a heater must sit at the bottom so its warm, buoyant air can rise and set a current turning through the whole volume, while a cooler must sit at the top so its cold, dense air can fall. Reverse them and you get stable stratification — a hot ceiling, cold feet and a bill for nothing. The chimney draught, the fins on a motorcycle engine, the fan over a laptop's heat sink and the pump driving coolant through a car radiator are the same physics with the volume turned up: raise $h$, thin the film, carry the heat away. And where convection is unwanted it is engineered out — the vacuum in a thermos flask exists precisely because a vacuum has no fluid to circulate, while double glazing, a duvet and a polar bear's fur all trap air in spaces too narrow to turn over. On the planetary scale, convection is the weather: the trade winds, the monsoon and the Hadley circulation are convection cells the size of continents, powered by the plain fact that the equator is hotter than the poles. The oceans run their own version, warm surface water flowing polewards, cooling, sinking and returning at depth on a thousand-year circuit that sets the climate of northern Europe. Deeper still, the mantle creeps in convection cells that drive the plates and raise mountains, while the churning liquid iron core generates the magnetic field that shields us — and the granules mottling the Sun's surface are the tops of convection cells, each the size of a country.

Put a saucepan of water on a hob and ask how the heat reaches the surface. Conduction is the obvious guess, and it is hopeless: water conducts about as badly as brick ($k \approx 0.6\ \mathrm{W\,m^{-1}K^{-1}}$), so a column of water heated only by conduction from below would take an age to warm through. Yet the pan boils in minutes. Something is cheating, and what it cheats with is motion. The mechanism runs in four beats. One: the fluid at the bottom is heated, its molecules speed up, push their neighbours apart, and the fluid expands. Two: expansion lowers the density — the same mass now fills more volume, and $\rho = m/V$ has no say in the matter. Three: the cooler fluid around it is denser, so gravity pulls it down harder, and as it sinks it squeezes the light warm fluid upwards — Archimedes' upthrust, nothing more exotic: the warm parcel is now lighter than the fluid it sits in, so it floats, exactly as a cork floats in water. Four: the risen fluid gives up its heat at the top, cools, grows dense again and sinks elsewhere, closing the loop. That loop is the convection current, and it keeps turning for as long as the bottom is hot and the top is cool. Notice what the mechanism demands. It demands a fluid, whose parts can slide past one another: heat a steel bar and its atoms vibrate harder, but they are locked into a lattice and go nowhere, so a solid can only ever conduct. It demands matter: across the vacuum gap of a thermos flask, or the void between the Sun and your skin, there is nothing to carry, so convection is switched off and only radiation survives. And it demands gravity — on the Space Station a candle flame makes no plume at all, but sits as a dim spherical ball, starved of the oxygen a convection current would have brought it. Convection needs a fluid that can move; conduction needs matter that can stay still; radiation needs nothing at all.

The purple ribbon is not a metaphor; it is the standard school demonstration, and it is the whole topic in one image. The dye marks the water. The water is what moves. The heat is merely a passenger. Nearly every wrong answer in this chapter comes from forgetting that line and imagining that "heat" is a substance which rises of its own accord.

Convection is driven by density, and density is governed by temperature. For a gas the link is exact. A parcel of air in a room is not sealed in a box: it is surrounded by the rest of the room's air, so any excess pressure it develops is relieved instantly by expanding, and it therefore warms at constant pressure. The ideal-gas law $PV = nRT$, with $P$ and the molecule count $n$ fixed, says the volume grows in proportion to the absolute temperature; and since the parcel's mass cannot change, its density $\rho = m/V$ must fall in exactly the same proportion:

where: $\rho_1, \rho_2$ — densities of the gas before and after heating, in $\mathrm{kg\,m^{-3}}$; $T_1, T_2$ — the corresponding absolute temperatures, in kelvin (not optional — the relation is nonsense in Celsius); pressure and amount of gas held fixed. The physical reading: hot gas is thin gas. Warm the air over a radiator from $20\ \mathrm{^\circ C}$ ($293\ \mathrm{K}$) to $40\ \mathrm{^\circ C}$ ($313\ \mathrm{K}$) and its density falls to $293/313 = 0.936$ of its old value — a drop of $6.4\%$, from about $1.21$ to $1.13\ \mathrm{kg\,m^{-3}}$. That $6.4\%$ is the entire engine of a heated room. It does not sound like much. It is more than enough. Liquids expand too, but grudgingly: their molecules already touch. The school relation uses the coefficient of volume expansion $\gamma$:

where: $\rho_0$ — the density before heating; $\gamma$ — the coefficient of volume (cubical) expansion, in $\mathrm{K^{-1}}$ ($\approx 2.1\times10^{-4}\ \mathrm{K^{-1}}$ for water near room temperature); $\Delta T$ — the temperature rise in kelvin (identical to the rise in $\mathrm{^\circ C}$, being a difference). The physical reading is a lesson in orders of magnitude: heat water by the same $20\ \mathrm{K}$ and $\rho = 1000/(1 + 2.1\times10^{-4} \times 20) = 995.8\ \mathrm{kg\,m^{-3}}$, a fall of just $0.42\%$ — some $15$ times feebler than the gas. Yet water is some $830$ times denser than air, so every cubic metre that does rise carries far more energy with it: hence the ferocity of a rolling boil, and hence water-cooled engines. Buoyancy now finishes the job. A warm parcel of volume $V$ displaces cool fluid of density $\rho_{\text{cool}}$, so Archimedes gives it an upthrust $\rho_{\text{cool}} V g$ while its weight is only $\rho_{\text{warm}} V g$. The upthrust wins, and Newton's second law on the parcel's own mass gives it an initial upward acceleration $a = (\rho_{\text{cool}}/\rho_{\text{warm}} - 1)\,g = (T_{\text{warm}}/T_{\text{cool}} - 1)\,g$ — derived in full in the card below. The temperature ratio is the buoyancy. Air at $40\ \mathrm{^\circ C}$ in a $20\ \mathrm{^\circ C}$ room gets $a = (313/293 - 1) \times 9.8 = 0.67\ \mathrm{m\,s^{-2}}$, about $7\%$ of $g$ (taking $g = 9.8\ \mathrm{m\,s^{-2}}$, the CBSE and ICSE value; JEE and NEET often allow $10$) — a gentle push, but a relentless one applied to every cubic centimetre of heated air for as long as the heater is on. Read it backwards and it says when convection dies: set $T_{\text{warm}} = T_{\text{cool}}$ and $a$ vanishes. No temperature difference, no buoyancy, no current.

If buoyancy alone turns the loop, the convection is natural (or free): a radiator warming a room, a mug steaming, the sea breeze, the Earth's mantle. Drive the fluid past the surface with a fan, a pump or the wind and it is forced. The physics is identical — heat still rides on moving fluid — but forced convection is far faster, because you no longer wait for a $0.67\ \mathrm{m\,s^{-2}}$ nudge to shift the fluid; you hurl cold fluid at the surface. Both obey the one equation every heating engineer uses:

where: $H$ — the rate of convective heat loss, in watts (never joules); $h$ — the convective heat-transfer coefficient, in $\mathrm{W\,m^{-2}K^{-1}}$; $A$ — the surface area washed by the fluid, in $\mathrm{m^2}$; $\Delta T = T_{\text{surface}} - T_{\text{fluid}}$ — the excess of the surface over the bulk fluid, in kelvin (equal to the number of degrees Celsius, being a difference). The physical reading: loss is proportional to how hot, how big, and how vigorously stirred. A radiator of $A = 1.5\ \mathrm{m^2}$ whose surface sits at $60\ \mathrm{^\circ C}$ in a $20\ \mathrm{^\circ C}$ room, losing heat naturally with $h = 10\ \mathrm{W\,m^{-2}K^{-1}}$, delivers $H = 10 \times 1.5 \times 40 = 600\ \mathrm{W}$. Note that $H \propto \Delta T$ to the first power: that linearity is the engine behind Newton's law of cooling, and it is what separates convection from radiation, whose power climbs as $T^4$. Carry the magnitudes of $h$ in your head, because examiners test the comparison and never the digits: still air, natural convection, $h \approx 5$–$25$; air blown by a fan, tens to a couple of hundred; water, hundreds; boiling or condensing water, thousands. A fan that lifts $h$ from $10$ to $40\ \mathrm{W\,m^{-2}K^{-1}}$ multiplies the loss by four, with no change whatever to the temperature of the surface. Why should $h$ care about the wind at all? Because a fluid does not slide freely over a solid — it sticks to it. The air actually touching a radiator is not moving, and just outside it lies a thin sluggish film the current never properly stirs, across which the heat must first conduct before the moving fluid beyond can carry it away. That film is the bottleneck. It is Ludwig Prandtl's boundary layer of 1904, and treating it as a still slab of thickness $\delta$ and applying Fourier's law to it gives $h \approx k_{\text{fluid}}/\delta$ — so the punchline of the whole topic is that a fan does not improve the air's conductivity; it scours the film thinner. Natural convection with $h = 10$ implies $\delta = 0.024/10 = 2.4\ \mathrm{mm}$ of still air; the fan reaches $h = 40$ and shaves it to $0.024/40 = 0.6\ \mathrm{mm}$. Wind chill, blowing on hot soup, engine fins and a laptop's fan are one idea — strip away the still air insulating the surface — and a woollen jumper is the same idea in reverse, dividing air into pockets too small to circulate. An insulator is a machine for making $\delta$ enormous.

The largest convection cells you will ever meet are made of air, and the board favourite is the coastal breeze. It turns on one fact: water is far harder to warm than land. Water's specific heat capacity is $4180\ \mathrm{J\,kg^{-1}K^{-1}}$ against roughly $800$ for dry sand and rock, so the same sunshine raises the land's temperature more than five times as fast. By mid-afternoon the land is the hotter surface: air above it warms, thins and rises, cooler denser air over the sea slides in beneath to replace it, and you feel a sea breeze blowing off the water onto the beach — the low branch of a loop whose return flow runs back out to sea high overhead. After sunset it reverses. The land, holding little heat, cools quickly; the sea is now the warmer surface; air rises over the water, and a gentle land breeze blows offshore.

The rule that survives every version of this question is written along the bottom of the figure: at ground level the wind blows from the cooler surface towards the hotter one, because it is over the hotter surface that air is rising and leaving a gap. Hold that and you can reconstruct the sea breeze, the land breeze, the monsoon (a sea breeze the size of a subcontinent) and the draught up a chimney without memorising any of them.

The third row is the one students never expect and examiners adore. Heat a fluid from the top and you get no convection current — you can keep ice at the bottom of a test tube whose top is boiling. Convection needs the heavy fluid to sit above the light one; that is the unstable arrangement gravity is itching to overturn.

Board level — the radiator in a cold room (CBSE/ICSE/IGCSE pattern)

A radiator has surface area $A = 1.5\ \mathrm{m^2}$. Its surface sits at $60\ \mathrm{^\circ C}$ while the room air is at $20\ \mathrm{^\circ C}$. For natural convection in air take $h = 10\ \mathrm{W\,m^{-2}K^{-1}}$. Find (a) the rate at which it delivers heat to the room by convection, (b) the energy delivered in one hour, and (c) the new rate once the room has warmed to $30\ \mathrm{^\circ C}$.

Step 1. Get $\Delta T$ right: it is the excess of the surface over the air, $\Delta T = 60 - 20 = 40\ \mathrm{K}$. Do not add $273$ — this is a difference, and a difference of $40\ \mathrm{^\circ C}$ is exactly $40\ \mathrm{K}$.

Step 2. (a) Equation (3): $H = hA\Delta T = 10 \times 1.5 \times 40 = \boxed{600\ \mathrm{W}}$. Note the unit — a rate, in watts, not an amount of energy.

Step 3. (b) Energy is rate times time, with the time in seconds: $t = 1\ \mathrm{hour} = 3600\ \mathrm{s}$, so $Q = Ht = 600 \times 3600 = 2.16\times10^{6}\ \mathrm{J} = \boxed{2.16\ \mathrm{MJ}}$, which is $0.6\ \mathrm{kWh}$ on the electricity bill.

Step 4. (c) As the room warms, the excess shrinks: $\Delta T = 60 - 30 = 30\ \mathrm{K}$, so $H = 10 \times 1.5 \times 30 = \boxed{450\ \mathrm{W}}$ — a fall of $25\%$. A radiator grows less effective the better it has done its job, which is why a heated room settles at a steady temperature instead of warming for ever.

Competitive level — the buoyancy engine, and switching on the fan (JEE-Main/NEET pattern)

Air touching the same radiator is warmed from the room's $20\ \mathrm{^\circ C}$ to $40\ \mathrm{^\circ C}$ at constant atmospheric pressure. Take $g = 9.8\ \mathrm{m\,s^{-2}}$ and $k_{\text{air}} = 0.024\ \mathrm{W\,m^{-1}K^{-1}}$. (a) By what percentage does its density fall? (b) What initial upward acceleration does buoyancy give it? (c) A $5\ \mathrm{cm}$ layer of the same air, held perfectly still, would conduct how much heat from that $1.5\ \mathrm{m^2}$ radiator — and how does that compare with the $600\ \mathrm{W}$ it loses by convection? (d) A fan raises $h$ from $10$ to $40\ \mathrm{W\,m^{-2}K^{-1}}$: what is the new loss?

Step 1. (a) Kelvin, at once — that is the question. $T_1 = 293\ \mathrm{K}$, $T_2 = 313\ \mathrm{K}$, and at constant pressure equation (1) gives $\rho_2/\rho_1 = T_1/T_2 = 293/313 = 0.936$. The density falls by $1 - 0.936 = 0.064 = \boxed{6.4\%}$. (A candidate who writes $20/40$ gets $50\%$ and has destroyed the question.)

Step 2. (b) The buoyancy relation: $a = (T_2/T_1 - 1)g = (313/293 - 1) \times 9.8 = 0.0683 \times 9.8 = \boxed{0.67\ \mathrm{m\,s^{-2}}}$ upwards — roughly $7\%$ of $g$. Small, but applied continuously to every parcel of heated air.

Step 3. (c) Still air can only conduct, so this part needs Fourier's law, not equation (3): $H = kA\Delta T/L = 0.024 \times 1.5 \times 40/0.05 = 1.44/0.05 = \boxed{28.8\ \mathrm{W}}$. Against the $600\ \mathrm{W}$ that the moving air carries, the ratio is $600/28.8 = \boxed{20.8}$ — circulating the air transports about twenty-one times more heat, with the same air, the same area and the same $\Delta T$; the only change is that the fluid was allowed to travel. That single ratio is why convection is a topic in its own right, and why a vacuum flask, with no fluid at all, beats every insulator made of matter.

Step 4. (d) Equation (3) with the new $h$: $H = 40 \times 1.5 \times 40 = \boxed{2400\ \mathrm{W}}$, four times the natural value, because $H \propto h$ at fixed $A$ and $\Delta T$. Physically, using $h \approx k/\delta$, the fan has thinned the stagnant film from $\delta = 0.024/10 = 2.4\ \mathrm{mm}$ to $0.024/40 = 0.6\ \mathrm{mm}$. Nothing about the air changed; only the film did. Timing note: under two minutes, and nothing here needs a calculator beyond the division $293/313$. Part (a) is a one-line ratio and the marks live in remembering kelvin; part (d) is pure proportionality — $h$ quadruples, so $H$ quadruples, and you should never re-substitute. If an MCQ offers you $50\%$ in part (a) or $600\ \mathrm{W}$ in part (d), those are the Celsius trap and the "forgot $H \propto h$" trap sitting in plain sight.

"Metals convect." They do not — metals are solids and they conduct. Convection needs a fluid free to move. This is the commonest wrong option in CBSE, ICSE and NEET items on the three modes.

"Heat rises." Heat is not a substance and has no buoyancy. Warm fluid rises, and it rises because it is less dense. Boards mark the reasoning, not the conclusion, and the word the examiner hunts for is density.

Kelvin in the density law, Celsius in the loss law. $\rho \propto 1/T$ needs absolute temperature ($293\ \mathrm{K}$, not $20$). But $H = hA\Delta T$ needs a difference, where $40\ \mathrm{^\circ C} = 40\ \mathrm{K}$ and adding $273$ is simply wrong. Two temperatures, two rules, one question.

Treating $h$ like $k$. $k$ is a property of the material; $h$ is a property of the situation — flow, geometry, orientation, fluid. Change the wind or turn the plate on its side and $h$ changes. There is no such thing as "the $h$ of copper".

Forgetting that $H$ is a rate. Watts, not joules; for an energy, multiply by the time in seconds ($1$ hour $= 3600\ \mathrm{s}$). And heat entering at the top gives stable stratification and no current at all — always ask where the heat goes in.

Board scope. CBSE, ICSE, NEET and IGCSE examine convection qualitatively — mechanism, diagrams, applications, comparisons. JEE-Main lists it in Unit 7 alongside conduction and radiation but rarely asks for a number; when it does, it is $H = hA\Delta T$ or a density-ratio argument exactly like the one above. Nusselt and Grashof numbers are on no school syllabus.

Texts. NCERT Physics Class 11, Ch. 11 (Thermal Properties of Matter) — convection, natural and forced · Cambridge IGCSE Physics 0625, Thermal Physics (convection currents, sea breezes) · Halliday, Resnick & Walker, "Conduction, Convection and Radiation".

Provenance. Standard results. A verifier should recompute: the radiator $H = hA\Delta T = 10 \times 1.5 \times 40 = 600\ \mathrm{W}$, its $Q = 600 \times 3600 = 2.16\ \mathrm{MJ} = 0.6\ \mathrm{kWh}$ in one hour, and $450\ \mathrm{W}$ once the room reaches $30\ \mathrm{^\circ C}$ (a $25\%$ fall); the constant-pressure density drop $1 - 293/313 = 6.4\%$ ($\rho$ falling from $1.21$ to $1.13\ \mathrm{kg\,m^{-3}}$) with the matching volume expansion $313/293 = 1.068$; the buoyant acceleration $(313/293 - 1) \times 9.8 = 0.67\ \mathrm{m\,s^{-2}}$ and the $1.2\ \mathrm{s}$ climb through half a metre; the water comparison $1000/(1 + 2.1\times10^{-4} \times 20) = 995.8\ \mathrm{kg\,m^{-3}}$, a fall of $0.42\%$; the still-air conduction $0.024 \times 1.5 \times 40/0.05 = 28.8\ \mathrm{W}$ and the ratio $600/28.8 = 20.8$; the fanned loss $40 \times 1.5 \times 40 = 2400\ \mathrm{W}$; the film thicknesses $0.024/10 = 2.4\ \mathrm{mm}$ and $0.024/40 = 0.6\ \mathrm{mm}$; the heat-capacity ratio $4180/800 = 5.2$; and the Prandtl numbers $\approx 0.7$ (air) and $\approx 7$ (water).

Prerequisites. Conduction & Thermal Conductivity — the mode convection is always contrasted with, and the source of the $k$ that sets $h$. Leads to. Newton's Law of Cooling — the linear $H \propto \Delta T$ of equation (3) is exactly what makes that law work. Go deeper (undergraduate). The full theory of natural convection, the Rayleigh criterion for when a current starts at all, and the Nusselt and Grashof numbers belong to fluid mechanics — none of it needed at school.