Heat Transfer · Learn
How a body cools over time — and the single number that tells you whether you can treat it as one uniform lump or must track the gradient inside.
When temperatures change in time, the energy stored in the material matters. An energy balance on a control volume gives the heat equation:
The thermal diffusivity $\alpha$ sets how fast a thermal disturbance spreads: it balances how readily the material conducts ($k$) against how much heat it must store to change temperature ($\rho c$). Metals have large $\alpha$ (respond fast); brick and glass small $\alpha$ (sluggish).
If a body conducts heat internally far faster than it exchanges heat with its surroundings, its temperature stays essentially uniform — just a single number falling in time. An energy balance, with convection to a fluid at $T_\infty$, becomes one ODE:
which solves to a pure exponential decay toward the ambient:
The time constant $\tau$ is how long to fall to $1/e$ ($\approx 37\%$) of the initial gap. Bigger body or lower $h$ → slower cooling.
The lumped assumption holds only when internal conduction resistance is small compared to surface convection resistance. That ratio is the Biot number:
The rule of thumb: if $\mathrm{Bi} < 0.1$, internal gradients are under a few percent and lumped capacitance is accurate. When $\mathrm{Bi} \gtrsim 1$, the surface sheds heat faster than the core can keep up, a real gradient opens, and you must solve the full PDE. (This is exactly the regime of quench hardening: cool the surface of steel fast while the core stays hot.)
Slide $h$ up or switch to glass and watch $\mathrm{Bi}$ cross 0.1 — the lumped (uniform) curve peels away from the true profile, with the error reported live. Open the lab →
Transient problems share a natural clock, the Fourier number:
It's elapsed time measured in units of the diffusion time across the body. When $\mathrm{Fo} \approx 1$, the thermal wave has crossed the object and it's well on its way to equilibrium. Charts and one-term series solutions for slabs, cylinders, and spheres are all written in terms of $\mathrm{Bi}$ and $\mathrm{Fo}$.
For real shapes we step the PDE forward in time on a grid. The simplest scheme is explicit finite differences, but it's only stable if the time step is small enough — for a 1-D surface node with convection,
which is exactly the limit the lab respects so the march never blows up.
EngineeringCandy · Learn · the theory behind the playground