Cooldown Time Calculation: Formulas, Worked Example, Pitfalls
Cooldown time is how long a hot object takes to fall from a starting temperature to a target temperature in a given environment. For a small solid with a uniform temperature, the answer is a single equation: t = τ · ln[(T₀ − T∞) / (T − T∞)], where τ = ρ·V·c / (h·A). For anything bigger, slower to conduct, or full of air, you need to check whether that shortcut holds. This guide gives you the equation, a worked example with real arithmetic, the checks that decide whether to trust it, and the cases where a different method is the better choice.
TL;DR
For a solid with negligible internal temperature gradients, cooldown time is the thermal time constant τ = ρVc/(hA) multiplied by ln[(T₀ − T∞)/(T − T∞)].
Check the Biot number first. If Bi = hLc/k is below about 0.1, the lumped method is reasonable. If not, use the Fourier number and a transient conduction solution.
Convection coefficient and radiation dominate the error. Doubling h roughly halves the cooldown time, and radiation adds a linearized coefficient you should not ignore at high temperature.
Enclosures and equipment rooms need a thermal mass term, or your estimate will be too optimistic.
Our toolkit puts conduction, convection, and radiation in one Excel-based workflow with a built-in property database, so you can skip rebuilding the spreadsheet each time.
Table of Contents
What cooldown time actually means
“Cooldown” gets used for several different problems, and the right formula depends on which one you have:
A solid part in air or fluid: a machined block, a pipe section, a heat sink, a forged billet. This is transient conduction plus convection and radiation at the surface.
A molded plastic part: cooling time inside a mold is governed by wall thickness and thermal diffusivity. MISUMI’s tutorial works through an ABS part with a 1.5 mm wall, and Toray’s molding guide shows how cooling time fits into total cycle time.
A room or enclosure after cooling fails: this is really a heat-up calculation, with air volume, equipment load, and thermal mass.
A system that must be pulled down to a setpoint: cryogenic or process equipment where you cool the vessel as well as its contents.
This article concentrates on the first case because it is the foundation for the others. It then covers enclosures and points you to the right method for the rest.
The lumped capacitance method
The lumped capacitance method assumes the whole body sits at one temperature at any instant. University course notes describe it as valid when there are no significant thermal gradients, so heat stored can be tracked with a single temperature. A University of Washington lab report states the same assumption: temperature is spatially uniform during the entire heating or cooling process.
With that assumption, an energy balance on the body gives an exponential decay:
(T − T∞) / (T₀ − T∞) = exp(−t / τ)
Solving for time:
t = τ · ln[(T₀ − T∞) / (T − T∞)]
with the time constant
τ = ρ · V · c / (h · A)
where:
Symbol | Meaning | SI unit |
ρ | Density | kg/m³ |
V | Volume | m³ |
c | Specific heat | J/(kg·K) |
h | Overall surface heat transfer coefficient | W/(m²·K) |
A | Exposed surface area | m² |
T₀ | Starting temperature | °C or K |
T∞ | Surroundings or fluid temperature | °C or K |
T | Target temperature | °C or K |
The ratio V/A is the characteristic length Lc. Write τ = ρ·c·Lc/h and you can see what drives cooldown: thick, dense, high-heat-capacity parts cool slowly, and a higher coefficient cools them faster.
Use temperature differences, so Celsius and kelvin give the same answer. Keep every other quantity in SI before you start, and convert at the end.
Check the Biot number before you trust it
The shortcut only holds when heat moves through the solid much faster than it leaves the surface. The test is the Biot number:
Bi = h · Lc / k
where k is the thermal conductivity of the solid. A common rule of thumb treats Bi below about 0.1 as acceptable for the lumped approach. The ratio compares internal conduction resistance to surface resistance. Small Bi means the surface is the bottleneck, so the interior stays close to uniform.
When Bi is larger, the surface cools faster than the core, and one temperature no longer describes the body. Then you need the Fourier number, Fo = αt/L², which is dimensionless time based on thermal diffusivity α. Portland State’s transient conduction animation defines it for a sphere and shows how the dimensionless temperature evolves, including the point where the lumped analysis applies.
Quick intuition: metals in still air almost always pass. Ceramics, polymers, concrete, and anything cooled by water or boiling often fail.
Worked example: an aluminum sphere in still air
The property values below are typical textbook-style assumptions chosen for illustration. Replace them with your own material data and a measured or correlated h.
Problem: A solid aluminum sphere, 25 mm radius, starts at 150 °C. The room is at 25 °C. How long until it reaches 40 °C?
Assumed inputs
ρ = 2,700 kg/m³
c = 900 J/(kg·K)
k = 237 W/(m·K)
h = 15 W/(m²·K), a combined natural convection and radiation value for still air
Step 1: characteristic length. For a sphere, V/A = R/3 = 0.025 / 3 = 0.00833 m.
Step 2: Biot number. Bi = 15 × 0.00833 / 237 = 0.00053. That is far below 0.1, so the lumped method is fine.
Step 3: time constant. τ = 2,700 × 900 × 0.00833 / 15 = about 1,350 s, or 22.5 minutes.
Step 4: temperature ratio. (40 − 25) / (150 − 25) = 15 / 125 = 0.12.
Step 5: time. t = 1,350 × ln(1 / 0.12) = 1,350 × 2.12 = about 2,860 s, or roughly 48 minutes.
Step 6: test sensitivity. Add a fan and assume h rises to 50 W/(m²·K). τ falls to about 405 s and the cooldown to about 860 s, or 14 minutes. Bi becomes 0.0018, still fine.
Three things stand out. First, you never reach T∞ exactly, so the time to get within a degree or two of ambient grows quickly. Second, going from 40 °C to 30 °C would take another 1,350 × ln(15/5) = about 1,480 s, longer than a third of the whole cooldown to 40 °C. Third, the answer scales almost linearly with 1/h, which is why the choice of h matters more than the third digit of density.
Getting h right: convection and radiation together
Most cooldown estimates go wrong at the coefficient, not the algebra.
Convection. Natural convection in air gives low h, in the single digits to low tens of W/(m²·K). Forced air, liquids, and phase change push it far higher. Lecture notes on heat transfer background, such as this Michigan Tech handout, define h through Newton’s law of cooling and show how it is measured, which is a useful reminder that h is an empirical quantity tied to geometry and flow.
Radiation. Radiation to the surroundings is easy to linearize. Radiative loss follows the Stefan-Boltzmann law with a fourth-power temperature dependence, but near a given temperature you can define h_rad ≈ 4εσT³ with T in kelvin. For ε = 0.8 and a surface near 350 K, that gives roughly 7.8 W/(m²·K), which is comparable to natural convection. Add h_conv and h_rad for the combined h in the example above.
At several hundred degrees Celsius, radiation can dominate, and the coefficient changes as the part cools. In that case, step through time in small intervals, recalculating h each step, instead of using a single exponential.
A practical note on surface finish: polished metals have low emissivity, so they cool more slowly by radiation than the same part painted or oxidized. If you assumed ε = 0.8 for bare aluminum, you probably overestimated the cooling.
When the lumped method fails
Recognize these cases and switch methods:
Bi above about 0.1. Use a transient conduction solution with the Fourier number, or a numerical model.
Variable properties. Specific heat and conductivity change with temperature, especially for cryogenic work and for polymers near glass transition. A single c value can mislead over a 300 K span.
Contact with other bodies. If the part sits on a plate or fixture, conduction through the interface matters. Our guide to thermal contact resistance explains how to estimate the joint’s effect.
Complex flow. If you do not know h because the airflow is recirculating or confined, CFD is the right tool. See how enclosure heat dissipation tips from simple estimates to simulation.
Dimensional changes during cooling. Parts shrink as they cool. If tolerances matter, see how to calculate thermal expansion. NIST’s metrology lecture gives the length correction as ΔL = α(T − 20)L, and notes CTE varies by alloy and process.
Cooling can also create stress when one region contracts faster than another. Our article on thermal stress calculation shows how a 60 °C change turns into 144 MPa.
Cooldown after cooling fails: enclosures and equipment rooms
The opposite question comes up in data rooms, electronics cabinets, and labs: how long before a space gets too hot if cooling stops? It is the same energy balance run in the heating direction, but air volume and thermal mass drive the result.
TechTarget’s 2024 guide to UPS backup time gives a rule of thumb for small spaces: Tmin = (Volume × 0.41 × TD / Watts) + τMass. Volume is in cubic feet, TD is the temperature differential between the maximum rated temperature and the air delivery temperature, Watts is the IT load, and τMass is a thermal mass constant between 0.5 and 1.5 minutes. The metric version uses a factor of 25.5 with cubic meters.
An illustrative run, with inputs I chose for the example: a 2,000 ft³ room, a 20-degree allowable rise, a 10,000 W load, and τMass of 1 minute. That gives 2,000 × 0.41 × 20 / 10,000 = 1.64 minutes, plus 1 minute, so about 2.6 minutes. The point is how short it is, and how much of it the mass term represents.
The same source warns that skipping the mass term produces overly optimistic and potentially dangerous predictions. Treat the formula as a screening estimate. For a real facility, validate with a simulation. Our comparison of data center CFD software alternatives covers the options by type.
Heat load comes first
Any of these estimates is only as good as the watts you feed it. For electronics, start from the dissipated power of each component. A walkthrough of power dissipation formulas for engineers is a useful way to build that load figure before you estimate how fast an enclosure climbs or recovers.
Ambient temperature and batteries
If the cooling failure coincides with a power failure, batteries become part of the problem. C&D Technologies notes that, as a general rule for VRLA batteries, about 10 °C (18 °F) above the reference temperature can cut service life nearly in half. Their paper also cautions against relying on generic correction factors when manufacturer curves exist. A hot room therefore shortens both the time you have and the battery capacity you assumed.
UPS planning guides also give design targets. One UPS planning and installation guide mentions designing equipment to operate at 50 °C (122 °F) for periods when normal cooling or ventilation has failed. Check your equipment’s own rating, since that is the number that sets your TD.
What not to use: cooling degree days
Searches for cooldown calculations sometimes land on cooling degree days. They are a different tool. The U.S. Department of Energy’s Future of Air Conditioning report is a useful reference on cooling capacity and units, including that one refrigeration ton equals 12,000 Btu/hr or 3.52 kW thermal. Cooling degree days measure demand for energy over a season, using the gap between mean outdoor temperature and a base, typically 65 °F. They say nothing about how fast a specific object or room cools, and they ignore humidity and building physics. Keep them for annual energy estimates.
A practical checklist
Define the system: solid part, enclosure, or process vessel.
Get material properties at the average temperature of the cooldown, not only at the start.
Compute Lc = V/A and estimate h from convection and radiation.
Calculate Bi. If it is above 0.1, switch to a transient conduction method.
Compute τ and then t for your target temperature.
Run a sensitivity check: h ±50%, emissivity high and low, ambient ±5 °C.
Compare with one measurement, even a rough thermocouple trace. The slope of ln(θ) versus time gives you τ directly.
If cooling matters for safety or tolerance, add margin or run a simulation.
Step 7 is the one most people skip. A single measured cooldown curve calibrates h better than any correlation, and you can reuse that h for nearby conditions.
Our perspective
Most of the cooldown questions we see are not hard in principle. They go wrong because the spreadsheet silently assumes constant properties, a single h, or a uniform temperature, and nobody checks. We built our toolkit around the idea that conduction, convection, and radiation belong in one place, with variable material properties and a built-in property database, so the checks are part of the workflow instead of an afterthought.
Our advice: do the lumped hand calculation first, because it tells you what dominates. Then, if the Biot number is marginal or the airflow is unclear, move up to a more detailed model. Do not pay for a heavy tool before you know which assumption is carrying your answer. And do not skip the measurement. Students and small teams especially should treat any cooldown prediction as a hypothesis until one thermocouple trace agrees with it.
— Joel
Run the numbers in the Jewlz toolkit
If you would rather not rebuild the iteration each time, the Jewlz Engineering Toolkit is an Excel-based tool that combines conduction, convection, and radiation analysis with a built-in property database and a wide temperature range. For heat transfer problems like the one above, start on the thermal analysis page. Basic tools are free, so you can try your own part before committing to anything.
FAQ
What is the formula for cooldown time?
For a body at uniform temperature, t = (ρVc / hA) · ln[(T₀ − T∞) / (T − T∞)]. The first term is the thermal time constant. It only applies when the Biot number, hLc/k, is small, typically below about 0.1.
How long does it take for an object to cool to room temperature?
It never fully reaches it in theory, because the decay is exponential. In practice, after about three time constants the temperature difference is down to around 5% of the start, and after four to five it is nearly gone. The example above has a 22.5-minute time constant, so “close to room temperature” takes over an hour.
Why is my calculated cooldown faster than the measured one?
The usual causes are an overestimated h, an emissivity that is too high for a polished surface, or ignored contact with a warm fixture. Another is a Biot number above 0.1, where the core lags the surface. Measure one cooldown curve and back out h from the slope.
Can I use cooling degree days to estimate cooldown time?
No. Cooling degree days describe seasonal energy demand relative to a base temperature, not the thermal response of a specific object or space. Use the energy balance and time constant approach instead.
How do I estimate how long a server room lasts after cooling fails?
A screening formula uses room volume, allowable temperature rise, IT load, and a thermal mass term of roughly 0.5 to 1.5 minutes. For a critical facility, confirm the result with CFD or a measured test, and check battery and equipment temperature ratings.
Sources
Recommended
Comments