Everything above waits for heat to leave through the wall. Ice in the drink does not wait: melting it absorbs about 80 times more heat than warming the same mass of water by one degree, and it happens in the minutes the cubes take to disappear. The cost is that you end up drinking the meltwater. This uses the same drink and the same container you chose above.
Not Newton’s law on one average temperature: a 2D axisymmetric ring-and-layer grid. Each cell exchanges heat with its neighbours and with the environment.
An empirical Rayleigh-number correlation inflates the effective conductivity (clamped 1–7×) to stand in for buoyant internal mixing; a layer check lets warmer fluid rise.
Natural (Churchill–Chu) in a still fridge; forced (Churchill–Bernstein-style, over a cylinder) with a fan, a breeze or an ice-water bath: dramatically stronger.
Stefan–Boltzmann with representative emissivities. Usually small, but rarely negligible for a still fridge.
Bi = hL/k, internal over external resistance. A small Bi means the surface is the bottleneck and the whole drink cools together; a large Bi means the surface sheds heat faster than the interior can supply it, so the core lags and the lumped approximation stops being trustworthy. A stirred, salted bath reaches Bi of about 6, which is why the core and surface lines split so far apart there. The dashed “lumped” line is Newton’s law; the solid “mean” line is this simulation’s spatially-resolved average: the gap is the detail a simple formula misses.
Melting a gram of ice absorbs about 334 J; warming a gram of water a degree absorbs 4.2. Ice in the drink delivers that in the minutes it takes to melt rather than the hour a fridge needs, but the meltwater stays: taking a 355 mL can from 22°C to 8°C needs roughly 51 g of ice and leaves the drink about 12% water. Two cubes only get you to 13°C, and you cannot go below freezing with ice alone: that is what salt is for.
A drink does not pass through its freezing point, it stops there: each gram of ice gives up 334 J, so the curve flattens near the bottom. Alcohol and sugar depress the point and spread it over a range (a lager goes at about −2°C, cola nearer −3, wine around −5.5, neat whisky not until about −27), because the first ice is pure water and leaves a more concentrated, colder-freezing remainder: slush long before a block. Water expands about 9% on freezing and a sealed container has only about 4% headspace: a can domes then splits a seam, glass goes from intact to shrapnel with no warning, PET usually stretches. An open glass cannot burst.
A sealed can only exchanges heat. An open glass also exchanges water with the room: a hot drink evaporates and cools quickly, but a cold one sits below the dew point, so moisture condenses onto it and releases latent heat back into the drink: the puddle under your pint. A pint takes measurably longer to chill on a humid day, which is why the humidity slider appears for open vessels.
It depends on how dry the air is. Evaporation drives the surface toward the wet-bulb, not the air temperature: on a 30°C day at 15% humidity the wet-bulb is about 14°C and the wrap genuinely chills; at 50% it is 22°C and the wrap merely stops the drink warming; in saturated air it does nothing.
A still fridge relies on natural convection, cold air sinking past the container under gravity; a fan or an ice-water bath forces fluid past the surface, far more effective. A bottle’s neck has less surface area relative to volume than a can’s body, so the model uses the real tapered geometry. Right after a warm drink goes in, the surface responds at once while the core lags: a can “cold to the touch” may not be cold all the way through.
Water's convective heat-transfer coefficient is roughly two orders of magnitude higher than air's, even still air against forced air. That's why the classic party trick — ice water plus salt, or just ice water alone — beats any fridge for speed, even a fan-forced one. Both baths are in the Environment list, and the salt slider is real physics rather than a gesture: brine follows the measured freezing curve down to the eutectic at −21.1°C, and a 20% bath roughly halves the time to target again.
Usually, yes — and the section above works out what it costs you. Melting ice absorbs about eighty times more heat per gram than warming the same water by a degree, so it acts almost instantly. But the meltwater stays in the glass: taking a 355 mL can from 22°C to 8°C needs roughly 51 g of ice and leaves the drink about 12% water. Two cubes only get you to 13°C. And you cannot go below freezing with ice alone, however much you add — that is what the salt is for.
It depends entirely on how dry the air is, which is why the answer people give is so inconsistent. Evaporation drives the surface toward the wet-bulb temperature, not the air temperature. On a 30°C day at 15% humidity the wet-bulb is about 14°C and the wrap genuinely chills. At 50% it is 22°C and the wrap merely stops the drink warming up. In saturated air it does nothing whatsoever. Pick Wet wrap + breeze and move the humidity slider to see it collapse.
The simulation now tells you, because it models the phase change rather than sailing through it. Your drink does not freeze at 0 °C — a lager goes at about −2 °C, cola nearer −3, wine around −5.5, and neat whisky not until about −27, which is why the freezer never touches it. Alcohol and sugar push the point down and smear it over a range, so you get slush long before you get a solid block.
Set the environment to a freezer and the page will tell you when ice starts forming, when it turns slushy, and when it freezes through — and the timer will tell you in real time, even with the tab in the background.
Because water is the awkward substance that expands when it freezes, by about 9%. A sealed container has only its headspace to absorb that — typically 4% or so — and once the ice needs more room than there is, something has to give.
What gives depends on what it is made of, which is why the warning differs by container. Aluminium is ductile: a can domes its base, bulges, and only then splits a seam. Glass does not yield at all, so a bottle goes from intact to shrapnel with no warning stage, and it is the one the page flags earliest. A PET bottle usually survives by stretching. An open glass cannot burst at all — the ice simply rises.
Because the top of it is not a wall. A sealed can exchanges only heat, through metal. An open glass also exchanges water with the room, and on a cold drink that runs backwards: the surface is below the dew point, so moisture condenses onto it and releases latent heat into the drink. That is the same physics as the ring your pint leaves on the table, and it is a real warming term — a pint takes several minutes longer to chill in humid air than in dry. The humidity slider appears whenever you pick an open vessel, for exactly that reason.
Heat has to conduct (and buoyantly mix) its way from the centre out to the wall before it can leave. Right after you put a warm drink in the cold, the surface responds almost immediately while the core lags well behind — which is exactly why a "cold to the touch" can isn't always cold all the way through yet.
Conduction moves heat between adjacent fluid cells inside the drink. Convection is heat leaving through the container wall into moving air or water outside. Radiation is the (usually small) contribution from the container's surface glowing in the infrared toward its cooler surroundings — included here because it's rarely negligible for a still fridge.
A bottle's neck has far less surface area relative to its volume than a can's cylindrical body, so heat has fewer places to escape from near the top. The simulation models the actual tapered geometry — body, shoulder, and neck — rather than treating every container as a plain cylinder.
Releases in which this page changed, newest last. Derived from the archived copy of every release, not from notes written afterwards — so it reflects what actually shipped. Site-wide passes are left out; they are in the full changelog.