What this calculator estimates
An ice cube left out on a worktop melts in something between twenty minutes and a couple of hours, and the difference comes down to far more than the temperature of the room. What the ice is sitting on matters enormously. So does whether the air is moving, how cold the ice was when it came out of the freezer, and — above all — how big the cube is.
This calculator estimates how long a piece of ice takes to melt completely. Enter the size across, say where it is and how warm the surroundings are, and it returns a best estimate with an honest range either side of it. The range is not padding: heat transfer into a melting object is genuinely hard to pin down, and a single confident number would be misleading.
It also produces a result most people find surprising. A cube, a sphere and a short cylinder of the same width all take the same time to melt, even though they contain very different amounts of ice. That is not a quirk of the arithmetic — it falls straight out of the physics, and it explains why the ice spheres sold for whisky work, and by how much.
The ice and its surroundings
How the calculation works
Melting ice does something helpful: it holds still at 0 °C. However warm the room is, the surface of the ice stays at freezing point until the last of it has gone, because all the incoming energy goes into breaking the bonds of the crystal rather than into raising the temperature. That makes the temperature difference driving the melt constant, and the sums much simpler than they look.
Step one — how much energy is needed. Turning ice at 0 °C into water at 0 °C takes about 334,000 joules per kilogram. This is the latent heat of fusion, and it is a large number: melting a kilogram of ice takes roughly the same energy as heating a kilogram of water from 0 °C to 80 °C. Ice that comes out of the freezer below 0 °C needs warming first, at about 2,100 joules per kilogram for each degree, which this calculator adds on.
Energy needed per kilogram = 334,000 + 2,100 × (0 − starting temperature in °C)Step two — how fast energy arrives. Heat crosses into the ice through its surface, at a rate set by the temperature difference and by how well the surroundings deliver heat. That second quantity is the heat transfer coefficient, written h, in watts per square metre per degree. Still air is poor at it. Water is far better. Metal in direct contact is better still.
Energy arriving each second = h × surface area × (surrounding temperature − 0 °C)Step three — putting them together. Set the energy arriving equal to the energy needed, and something tidy happens: the surface area cancels out. What is left says that the surface of the ice retreats inwards at a steady rate, no matter how big the piece is or what shape it is.
Surface retreat rate = h × temperature difference ÷ (density of ice × energy needed per kilogram)
Melting time = characteristic half-thickness ÷ surface retreat rateThe characteristic half-thickness is three times the volume divided by the surface area. For a cube it works out at half the side length; for a sphere, half the diameter; for a cylinder as tall as it is wide, half the diameter again. All three are the same number, which is why all three melt in the same time.
The formula is standard physics and the constants — the density of ice, the latent heat of fusion, the specific heat capacity of ice — are measured quantities that are not in dispute. The uncertainty is concentrated almost entirely in h. Published values for natural convection in air span roughly 5 to 25 W/m²K, but a real ice cube is rarely floating in still air: it sits in a puddle of its own meltwater on a surface that conducts heat into it. The figures used here are effective values that bundle all of that together, chosen so that the answers match what people actually observe. They are the honest weak point of the calculation, which is why the result is given as a range.
Why shape matters less than you would think
The surface of a melting piece of ice retreats at a steady rate, so the melting time depends on how far the surface has to travel to reach the middle — and nothing else. A cube 30 mm across, a sphere 30 mm across and a cylinder 30 mm across all have 15 mm of ice between the surface and the centre, so all three melt in the same time.
They do not contain the same amount of ice, though.
| Shape, 30 mm across | Volume | Mass | Melting time |
|---|---|---|---|
| Cube | 27.0 cm³ | 24.8 g | Same for all three |
| Cylinder, as tall as it is wide | 21.2 cm³ | 19.4 g | Same for all three |
| Sphere | 14.1 cm³ | 13.0 g | Same for all three |
The sphere holds barely half the ice of the cube and lasts exactly as long. Per gram it is nearly twice as durable.
That is the wrong comparison for most practical purposes, though. If you are choosing between shapes for a drink, the sensible question is which lasts longer for the same amount of ice — the same amount of eventual dilution. Compared on equal volume, a sphere lasts about 24% longer than a cube, because a sphere has the smallest surface area of any shape enclosing a given volume.
Yes, and by a real but modest margin. Roughly a quarter longer than the same volume of ice as a cube, in the same glass. The much larger effect is simply using a bigger piece: one large cube beats several small ones by far more than any shape ever will, because doubling the width doubles the melting time while giving you eight times as much ice. Marketing that implies a dramatic advantage for the shape alone is overstating a genuine but small effect.
Worked example: an ice cube on a plate
A cube from a standard domestic ice tray, 25 mm along each side, taken straight from a freezer at −18 °C and left on a plate in a kitchen at 20 °C with the windows shut.
Likely range: 47 minutes to 1 hour 41 minutes
Cube, 25 mm across
Volume: 15.6 cm³ | Mass: 14.3 g
Energy needed to melt it: 5.3 kJ
Warming from −18 °C to 0 °C accounts for 10% of that.
Half the mass will have gone after about 13 minutes.
By 32 minutes only about an eighth of it is left.
The surface retreats at a steady 11.6 mm per hour throughout. The melting only looks as though it slows down because so little remains.
The same volume of ice formed into a sphere would take about 80 minutes — roughly 24% longer. A sphere 25 mm across, however, would melt in the same time as this one, because it has the same distance from surface to centre. Why shape matters less than you would think.
A wide estimate, based on an effective heat transfer coefficient of 35–75 W/m²K for those surroundings. In a drink the liquid cools as the ice melts, so real times run longer than shown. Not for food safety, medication or anything that matters — see when not to use this calculator.
Working through it by hand: the cube holds 15.6 cm³ of ice weighing 14.3 g. Melting it needs 334,000 joules per kilogram, plus another 37,800 to bring it up from −18 °C to freezing point, giving 371,800 joules per kilogram or about 5.3 kJ in total — roughly what a small torch bulb uses in half an hour.
The distance from surface to centre is 12.5 mm. With an effective heat transfer coefficient in the region of 55 W/m²K and a 20 degree temperature difference, the surface retreats at about 11 mm per hour, so 12.5 mm takes a little over an hour.
The instructive part is what happens if you change one thing. Move the same cube onto a metal baking tray and the estimate roughly halves, because the tray conducts heat into the base far better than the plate does. Put it in a glass of water instead and it drops to a quarter of an hour. Nothing about the ice has changed at all — only what is delivering heat to it.
Understanding the result
Two things are worth reading carefully.
The range matters more than the midpoint. The spread reflects genuine uncertainty in how well heat reaches the ice, not rounding. A result of "about 50 minutes, likely 35 minutes to 1 hour 15" means the physics is settled and the surroundings are not. If you need a figure to plan around, use the fast end — ice melting sooner than expected causes more problems than ice lasting longer.
Most of the mass goes early. Because the width shrinks steadily while the volume depends on width cubed, the melting is heavily front-loaded:
| Time elapsed | Ice remaining |
|---|---|
| 21% of the total | Half the mass gone |
| Half the total | Only 12.5% of the mass left |
| 75% of the total | Under 2% left |
This is why ice seems to vanish quickly and then leave a stubborn little sliver behind for ages. The sliver is not melting more slowly — the surface is retreating at exactly the same rate it always was. There is simply very little left to notice.
Treat the answer as an order-of-magnitude guide. It is reliable enough to tell you whether something takes minutes or hours, and to compare one situation against another. It is not accurate enough to plan anything that matters on, and it must not be used for food safety, medication storage or anything where being wrong has consequences. See the section below.
When not to use this calculator
This is a curiosity and a teaching tool. There are several situations where it looks relevant and is not, and one of them genuinely matters.
Not for deciding whether food is safe. This is the important one. The calculator tells you when ice has finished melting; it tells you nothing useful about whether frozen food has reached an unsafe temperature. Food becomes hazardous well before the last ice crystal disappears, because the outside warms into the range where bacteria multiply long before the centre has thawed. The Food Standards Agency advises defrosting food in a fridge rather than at room temperature, and once defrosted, cooking it within a day. Do not use a melting-time estimate as a substitute for that guidance, and do not use it to judge whether a freezer's contents have survived a power cut.
Not for medication or medical supplies. Insulin, some vaccines and various other medicines have specific storage requirements, and whether an ice pack has melted is not the same question as whether the contents stayed within range. Follow the patient information leaflet, or ask a pharmacist.
Not for cold chain, laboratory or engineering work. Anything that has to be documented, certified or relied upon needs measurement, not estimation. The uncertainty in the heat transfer coefficient alone is a factor of two or more.
Not for large blocks of ice. The model assumes heat arrives at the surface faster than it travels inwards, which holds for cubes and spheres up to a few centimetres. For a block the size of a bag of party ice or larger, conduction through the ice itself becomes the limiting factor and this calculation will underestimate the time, in some cases substantially.
Not for ice in the open in freezing weather. If the surroundings are at or below 0 °C the ice will not melt at all, and the calculator will say so rather than returning a number.
Things that affect the result
- What the ice is resting on. The single largest factor after size. A metal tray conducts heat into the contact face far faster than air delivers it to the other five, and can halve the melting time compared with a wooden board.
- The puddle. Meltwater that stays around the base keeps it in good thermal contact with the surface and speeds things up. Ice on a wire rack, where the water drains away, lasts noticeably longer than ice in a saucer.
- Air movement. A fan or an open window makes a substantial difference. Still air is a poor conductor, and the thin layer of cold air that forms around the ice is doing real insulating work until something disturbs it.
- The drink cooling down. In a glass, the liquid does not stay at room temperature — it drops quickly as the ice absorbs heat from it. The temperature difference collapses and the last of the ice can take several times longer than a constant-temperature calculation suggests. For drinks, treat the answer as a lower bound.
- How cold the ice started. Ice at −18 °C needs about 11% more energy than ice already at 0 °C. Real but modest, and much smaller than most people assume.
- Air bubbles and clarity. Cloudy ice from a domestic freezer is full of trapped air, which makes it slightly less dense and a little quicker to melt than the clear ice made by directional freezing.
- Humidity. In humid air, water vapour condenses on the cold surface and releases its own latent heat as it does so, speeding up the melt a little.
- Salt or alcohol. Both lower the freezing point, so ice in a spirit or in salty water melts faster than the same ice in plain water — the ice can no longer sit comfortably at 0 °C.
Common mistakes
- Assuming the time scales with volume. It scales with width. A cube twice as wide holds eight times as much ice but takes only twice as long to melt.
- Expecting the last piece to hold on. The final sliver disappears at exactly the same surface retreat rate as the first millimetre. It just looks slower because so little is left.
- Reading the midpoint as a prediction. It is the middle of a wide range, not an answer to the nearest minute.
- Using it on a drink and expecting accuracy. The drink cools as the ice melts, which the model does not track. Real times in a glass run longer.
- Confusing melted with thawed, or thawed with safe. These are three different things, and only the first is what this calculator estimates.
- Measuring the wrong dimension. For an oblong piece, the number that matters is roughly the smallest width, because that is the shortest route to the middle.
Frequently asked questions
How long does a standard ice cube take to melt at room temperature?
A typical domestic ice cube is around 25 mm across, and on an ordinary plate in a still room at 20 °C it takes roughly an hour, with anything from about 45 minutes to nearly two hours being plausible depending on the surface and the air movement. On a metal tray it can be half that. In a drink it is more like ten to twenty minutes, and in a drink being stirred, under five.
Why does a bigger ice cube last so much longer?
Because melting works inwards from the surface at a steady rate, so the time depends on how far the surface has to travel to reach the centre — which is half the width. Double the width and you double the time, but you get eight times as much ice. This is the main reason bars use large cubes and spheres for spirits: a single big piece cools the drink for far longer while adding less water in the process.
Do ice spheres really last longer than cubes?
Compared with the same volume of ice in cube form, yes, by around 24%. A sphere has the smallest surface area of any shape enclosing a given volume, so heat gets in more slowly. The effect is real but modest, and it is much smaller than the effect of simply using a larger piece. A sphere and a cube of the same width take the same time to melt, but the sphere contains only about half as much ice.
Does hot water freeze faster than cold water?
That is a different question — the Mpemba effect — and it remains genuinely contested. Some experiments have reported hot water freezing first under particular conditions, and various explanations have been proposed involving evaporation, dissolved gases, convection and supercooling. Careful attempts to reproduce it have often failed. It is not settled science, and it does not affect melting in any case.
Why does ice melt faster on metal than on wood?
Metal conducts heat hundreds of times better than wood, so it delivers heat from the room into the ice through the contact face very efficiently, while the wood essentially insulates. This is also why a metal worktop feels colder than a wooden one at the same temperature — it is pulling heat out of your hand faster. The demonstration works in reverse with ice.
Does salt make ice melt faster or slower?
Faster, in the sense that it disappears sooner, though the mechanism is not extra heat. Salt lowers the freezing point of water, so ice in salty water can no longer sit stably at 0 °C and melts even though the surroundings are cold. This is why salt is spread on icy roads. It is also why an ice bath with salt in it gets colder than one without — the melting draws energy out of the water around it.
How accurate is this estimate?
Good enough to distinguish minutes from hours and to compare situations against one another, and not much better than that. The physics and the constants are solid; the weak point is the heat transfer coefficient, which depends on details the calculator cannot know — how much meltwater is pooling, exactly what the surface is made of, whether there is a draught. The range shown reflects that honestly, and the true answer can still fall outside it.
Can I use this to work out whether my freezer food is still safe after a power cut?
No, and please do not. Whether food is safe depends on the temperature it reached and for how long, not on whether ice crystals remain, and the outside of a package warms into the hazardous range long before the centre thaws. The Food Standards Agency publishes guidance on power cuts and frozen food; follow that instead. If you are in any doubt about a particular item, the safe course is to throw it away.
Is what I enter stored?
No. Everything is calculated in your browser and nothing is transmitted or retained.
Related tools
References
These organisations publish the guidance behind this page.
- National Physical Laboratory — the UK's national measurement institute, on thermal properties and measurement uncertainty
- Institute of Physics — teaching resources explaining latent heat, specific heat capacity and changes of state
- Royal Society of Chemistry — the physical properties of water and ice
- Food Standards Agency — guidance on defrosting food safely, chilled storage and frozen food after a power cut
- Met Office — background on ice, freezing conditions and why salt is used on roads
External guidance changes. Check the current position at the source before relying on it for a decision that matters.