Could you boil chai with body heat? The thermodynamics

One resting adult gives off about 80 W, enough energy to boil a cup of chai in 9 minutes. The second law of thermodynamics says no number of hugging friends will ever do it, unless you add a machine.

A glass tumbler of frothy Kerala-style milk tea
Photo: Navaneethpp, CC BY 4.0, via Wikimedia Commons

You cannot boil chai with body heat by contact, however many friends hug the kettle: heat only flows from hotter to colder, so the chai can never get warmer than the people warming it, about 37°C at the core. In energy terms a resting adult gives off roughly 80 W, enough to boil a 150 mL cup in about 9 minutes, so the energy is there. Turning it into boiling heat needs a heat engine and a heat pump, and the laws of thermodynamics then set the bill at about 10 people with perfect machines, or around 27,000 people with today's body-heat generators.

Calculated in this post from Byrne et al. (2005), where 1 MET is 1 kcal per kg per hour, Proto et al. (2018), who measured about 5 µW per module at rest, and the Carnot limits, for this what-if's premises of a 150 mL cup, a 25°C room, a 70 kg adult and a 5-minute wait.

Key takeaways

This is a lovely question because the naive maths says yes and the physics says no. The difference between the two is the second law of thermodynamics, the rule that makes coffee go cold and never the other way round.

Here is the plan:

  1. do the naive energy sum;
  2. show exactly where it breaks;
  3. design the contraption that would actually work.
45 kJto bring 150 mL of chai from 25°C to boiling in Bengaluru
~80 Wresting heat output of a 70 kg adult (1 MET)
96.9°Cboiling point of water at Bengaluru's 920 m altitude (calculated)
3.9%Carnot limit for turning body heat into work in a 25°C room

How much heat does a cup of chai need?

A 150 mL cup of chai needs about 45 kJ to go from room temperature to a boil in Bengaluru, about the energy a 1,500 W electric kettle delivers in 30 seconds.

Calculated in this post from water's specific heat, about 4,184 J/(kg·K) (Wikipedia), the 96.9°C boiling point worked out below from Bengaluru's 920 m altitude (Wikipedia), and this what-if's premises of a 150 mL cup starting at 25°C and an illustrative 1,500 W kettle.

Chai is mostly water and milk, so I use water's specific heat, 4.18 kJ/kg·K (milk is slightly lower). Water boils lower in the hills, so the target temperature depends on where you stand. Two equations get it:

  1. Air pressure at altitude. The barometric formula with a scale height of about 8,400 m (an approximation for the lower atmosphere):
    \[ P = P_0 e^{-z/H} = 101.3\ \text{kPa} \times e^{-920/8400} \approx 90.8\ \text{kPa} \]

    Calculated in this post from the barometric formula, with a scale height of about 8.4 km and sea-level pressure of 101,325 Pa (Wikipedia) and Bengaluru's 920 m altitude (Wikipedia).

  2. Boiling point from the Clausius–Clapeyron relation, which links vapour pressure to temperature through the latent heat of vaporisation, \(L\) = 40.66 kJ/mol:
    \[ \frac{1}{T_b} = \frac{1}{373.15\ \text{K}} - \frac{R}{L}\ln\frac{P}{P_0} \quad\Rightarrow\quad T_b \approx 370.1\ \text{K} = 96.9^\circ\text{C} \]
  3. Heat for the cup.
    \[ Q = m c \Delta T = 0.15\ \text{kg} \times 4180\ \tfrac{\text{J}}{\text{kg·K}} \times (96.9 - 25)\ \text{K} \approx 45{,}000\ \text{J} \]

    Calculated in this post from water's specific heat, about 4,184 J/(kg·K) (Wikipedia), the 96.9°C boiling point worked out in the step above (with its latent heat of 40.66 kJ/mol (Wikipedia)), and this what-if's premises of a 0.15 kg cup starting at 25°C.

At Leh, about 3,500 m up, the same formulas give a boiling point near 88.5°C, which is why tea brewed there tastes thinner and why pressure cookers matter in the hills.

Several large tea kettles boiling on a stove in New Delhi
A tea stall's kettles run on gas burners delivering kilowatts. A human body offers about 80 watts.Photo: Shubham Sharma on Unsplash, CC0, via Wikimedia Commons

How much heat does a human body give off?

A resting adult releases roughly 70 to 80 W of heat, because almost all the energy from food ends up as heat. By the classic definition, one metabolic equivalent (1 MET) is 1 kcal per kg per hour, which for a 70 kg person is

\[ P = 70\ \text{kg} \times 1\ \tfrac{\text{kcal}}{\text{kg·h}} \times \frac{4184\ \text{J}}{3600\ \text{s}} \approx 81\ \text{W} \]

Byrne and colleagues measured 769 adults and found real resting energy use averaged 0.84 kcal/kg/h, so the textbook value overestimates it by about 20% (the paper's wording), which gives about 68 W for the same 70 kg person. I use 81 W as the generous figure. So the naive answer is

\[ t = \frac{Q}{P} = \frac{45{,}000\ \text{J}}{81\ \text{W}} \approx 555\ \text{s} \approx 9\ \text{minutes} \]

One person, nine minutes, one cup. If only it were that easy.

The story

Someone has boiled water with muscle power before, and the muscles belonged to horses. In 1798 Count Rumford, Benjamin Thompson, was overseeing cannon boring at the Munich arsenal. He sank a brass gun barrel in a box holding 18.77 pounds of water at 60°F, fitted a deliberately blunt steel borer, and had two horses turn the barrel at about 32 revolutions a minute.

After an hour the water stood at 107°F; after two hours, 178°F. At 2 hours 30 minutes it boiled. Rumford wrote of the surprise on the bystanders' faces at seeing cold water boil without any fire, and estimated the heat at more than nine wax candles burning together. The key was the direction: horses supplied work, and work can be turned into heat at any temperature. Body heat, as we will see, is a different currency.

Rumford: An Inquiry concerning the Source of the Heat which is excited by Friction, Philosophical Transactions (1798), transcribed by J. Denker
Metabolic equivalent: one size does not fit allByrne NM, Hills AP, Hunter GR, Weinsier RL, Schutz Y · Journal of Applied Physiology · 2005

The 1-MET convention (3.5 mL O₂ per kg per minute, or 1 kcal per kg per hour) traces back to one 70 kg, 40-year-old man. The authors measured resting metabolic rate by indirect calorimetry in 642 women and 127 men aged 18 to 74. Average resting energy cost was 0.84 ± 0.16 kcal/kg/h and oxygen use 2.6 ± 0.4 mL/kg/min, so the convention overstates resting oxygen use by 35% and energy by 20%. The limit: a mostly American sample that included many people with obesity, so body composition drove much of the variation.

Read the paper

Why can't hugging the kettle ever boil it?

Because the second law of thermodynamics forbids heat from flowing on its own from a cooler body into a hotter one, and a human core sits at about 37°C, 60 degrees short of boiling.

Rudolf Clausius stated it in the 1850s: heat cannot pass by itself from a colder to a warmer body. Put a kettle against a crowd and heat flows into it only while the kettle is cooler than their skin.

As the kettle warms, the flow slows and stops at the point where the kettle matches the skin. With \(N\) people, the steady temperature is

\[ T_{kettle} = \frac{N\,G_{skin} T_{skin} + G_{air} T_{room}}{N\,G_{skin} + G_{air}} \;\xrightarrow{N\to\infty}\; T_{skin} \]

where \(G\) is the thermal conductance of each contact. Adding friends only makes the kettle approach skin temperature faster. It never passes it.

Core temperature in healthy people is held near 37 ± 0.5°C at rest, and skin is cooler than the core, so a crowd-warmed chai tops out as lukewarm milk. Even a feverish crowd only gets you a few degrees more.

Heat only flows downhill on its own 96.9°C 37°C 25°C Chai at the boil Body (core) Room air free, by itself ✕ forbidden without work (Clausius) A hugged kettle stops at skin temperature, 60 degrees short
The second law as a ladder. Body heat falls freely to the cooler room but cannot climb the 60 degrees to the kettle by itself. Temperatures from Hymczak et al. (core) and the boiling-point calculation above.

Energy is not the problem

Notice what the second law does not say. It does not say the 80 W are lost. It says they arrive as low-grade heat, at a temperature too close to the room's to push anything hotter.

Energy is plentiful; usable temperature is scarce.
Core Temperature Measurement: Principles of Correct Measurement, Problems, and ComplicationsHymczak H, Gołąb A, Mendrala K, Plicner D, Darocha T, Podsiadło P, Hudziak D, Gocoł R, Kosiński S · International Journal of Environmental Research and Public Health 18(20):10606 · 2021

This review of clinical thermometry sets out how core temperature is regulated and measured. Core temperature is held close to 37 ± 0.5°C at rest with a daily swing of 0.25 to 0.5°C. At rest the body sheds heat mainly by radiation (about 65%), then evaporation (about 20%), convection (10 to 15%) and conduction (about 2%), which is the channel a hugged kettle would rely on. The limit: a clinical review focused on measurement sites and devices, not on heat output in watts.

Read the paper (free)
A human hand shown side by side in visible light and in thermal infrared
A thermal camera shows the hand as the warmest thing in the room, yet still far cooler than boiling. Radiation is the body's main way of shedding heat at rest.Photo: Amfeli, CC BY 4.0, via Wikimedia Commons

What is the least number of friends it could take?

With perfect machines, about 10 people. The trick is to use body heat to run a heat engine, then use its work to run a heat pump that pushes heat uphill into the kettle.

  1. Heat engine, body to room. The Carnot efficiency, the best any engine can do between two temperatures, is
    \[ \eta_C = 1 - \frac{T_{room}}{T_{body}} = 1 - \frac{298\ \text{K}}{310\ \text{K}} \approx 3.9\% \]
    So at most 3.1 W of work per 81 W person.
  2. Heat pump, room to kettle. A perfect heat pump delivering heat at 96.9°C from 25°C air has a coefficient of performance
    \[ \text{COP}_C = \frac{T_{hot}}{T_{hot} - T_{room}} = \frac{370.1}{370.1 - 298.2} \approx 5.1 \]

    Calculated in this post from the Carnot heating COP, Thot/(Thot − Tcold) (Wikipedia), the 96.9°C boiling point worked out above and this what-if's 25°C room.

  3. Useful heat per person. \( 81 \times 0.039 \times 5.1 \approx 16 \) W, delivered at boiling temperature.
  4. Power to boil in 5 minutes. \( 45{,}000\ \text{J}/300\ \text{s} = 150 \) W, so \( N = 150/16 \approx 9.3 \), call it 10 friends.

That ten is a hard floor, set by Carnot. No inventor can beat it, because a machine that did could be run backwards to break the second law.

The best possible machine, per person Body: 81 W of heat at 37°C Carnot engine η = 3.9% 78 W to room 3.1 W of work Heat pump COP = 5.1 13 W from room 16 W into the kettle at 96.9°C 150 W for a 5-minute cup ÷ 16 W ≈ 10 people
The Carnot floor, drawn as a chain. Each perfect engine skims 3.9% of body heat as work; each perfect heat pump multiplies that work 5.1 times as heat at boiling temperature. Numbers from the steps above; flows rounded.

How many friends would it take with real technology?

Around 27,000 friends with today's wearable thermoelectric generators and a plain heater, or about 10,700 with a good real heat pump, because real devices capture a tiny fraction of the Carnot limit.

Calculated in this post from Proto et al. (2018), who measured about 5 µW per module at rest, the module size and body area assumed in the steps below, and the 150 W needed to boil the cup in 5 minutes (45 kJ over 300 s, worked out above).

A thermoelectric generator (TEG) is a Peltier module run in reverse: a temperature difference across it produces a voltage. Proto and colleagues strapped one to arms and legs and measured about 5 µW at rest and up to about 50 µW while cycling, with only about 2°C across the device.

  1. Per module. 5 µW at rest. I assume a standard 40 mm × 40 mm module, 16 cm².
  2. Whole-body suit. A body surface of about 1.8 m² (my assumption for an adult) fits about 1,125 modules: \( 1125 \times 5\ \mu\text{W} \approx 5.6\ \text{mW} \) per person.
  3. Plain resistive heater. \( N = 150\ \text{W} / 0.0056\ \text{W} \approx 26{,}800 \) people.

    Calculated in this post from Proto et al. (2018), who measured about 5 µW per module at rest, the module count and body area assumed in the two steps above, and the 150 W worked out earlier.

  4. Add a real heat pump with COP 2.5 (my assumption, half of Carnot): \( N \approx 10{,}700 \).
Thermal Energy Harvesting on the Bodily Surfaces of Arms and Legs through a Wearable Thermo-Electric GeneratorProto A, Bibbo D, Cerny M, Vala D, Kasik V, Peter L, Conforto S, Schmid M, Penhaker M · Sensors 18(6):1927 · 2018

The team strapped a commercial Peltier module (TES1-12704) to the biceps, forearm, thigh and calf of male volunteers at about 23 to 25°C room temperature. At rest it produced about 2.6 to 5 µW depending on the site; walking raised that to about 18 to 25 µW, jogging about 30 µW, and cycling a median of 37 µW with peaks near 50 µW. The temperature difference across the module never exceeded 2.5°C. The limit: very few volunteers, one module type, and absolute power rather than power per square centimetre.

Read the paper (free)
People needed to boil a 150 mL cup in 5 minutes
See the data as a table
People (people)
Perfect engine + perfect heat pump10
Today's TEG suit + real heat pump (COP 2.5)10700
Today's TEG suit + plain heater26800

Source: Calculated from Byrne et al. 2005 and Proto et al. 2018 · Hugging the kettle directly is not shown: by the second law it never boils, at any number of people.

Why real devices fall so far short

Two things make the real number so much worse than the floor:

  • Tiny temperature difference. A TEG on skin sees only about 2°C across it, not the 12°C between core and room, because skin, the module and the air all add thermal resistance.
  • Weak materials. Today's best materials convert a small slice of even that.

New materials are closing the gap. A 2025 stretchable thermoelectric rubber from Peking University reached a figure of merit of 0.49 at room temperature, and a 2025 Korean ionic film lit an LED from a 1.5°C difference.

How many friends to boil your chai?

What changes if the assumptions move?

The verdict does not move, but the head count swings by orders of magnitude with the harvesting technology and the waiting time.

ChangeEffect on friends neededWhy
Wait 60 minutes instead of 512× fewer (about 2,200 with TEG + heater)Power needed falls from 150 W to 12.5 W, though heat loss from the kettle now dominates
Everyone cycles instead of restingAbout 7× fewerProto measured about 37 µW cycling vs 5 µW resting
Cold room, 15°C instead of 25°CCarnot floor drops to about 7A bigger body-to-room gap makes the engine better, though the heat pump works harder
Sea level, 100°C boilAbout 4% more heat\(\Delta T\) rises from 72 to 75 K
Real resting output, 68 W instead of 81 WFloor rises to about 11Byrne's measured average
Kettle heat loss of about 36 W at boiling (estimate)About 10 to 25% more for a 5-minute boilNewton's law of cooling, \(hA\Delta T\) with \(h\) ≈ 10 W/m²·K and \(A\) ≈ 0.05 m²
Verdict: possible with changes

By touch alone it is impossible: the second law of thermodynamics caps the chai at the temperature of the people warming it, near 37°C, at any crowd size. With a heat engine and a heat pump in between it becomes possible, and the Carnot limits (3.9% engine efficiency, COP of 5.1) set the floor at about 10 people for a 5-minute cup. With today's wearable thermoelectric modules, which harvest about 5.6 mW per resting person, it takes roughly 27,000 people, or about 10,700 with a good heat pump. At an assumed ₹150 per module, kitting out that crowd would cost several hundred crore rupees for one cup.

Calculated in this post from the Carnot heating COP, Thot/(Thot − Tcold) (Wikipedia) and Proto et al. (2018), who measured about 5 µW per module at rest, for this what-if's premises of a 150 mL cup, a 25°C room and a 5-minute wait. The ₹150 module price is an assumption, not a quoted price.

A cup of masala chai with milk on a saucer
The prize: 150 mL of chai, 45 kJ of heat, and a reminder that a gas flame at around 2,000°C is what makes boiling easy.Photo: Miansari66, Public domain, via Wikimedia Commons

Once you do get it boiling, what happens next is chemistry: why milk sometimes splits when ginger goes in is explained in why milk curdles in ginger chai, and the brewing itself in how to brew masala chai. The leaf's own story is in CTC versus orthodox tea. More from the Taste Lab is on the Rasa world page.

What if next?

Questions people ask

How much heat does the human body produce?

At rest, roughly 70 to 80 W for a 70 kg adult. The textbook 1 MET is 1 kcal/kg/h (about 81 W), but measurements in 769 adults averaged 0.84 kcal/kg/h, so the textbook value overestimates by about 20%.

Can body heat boil water?

Not by direct contact. Heat flows only from hot to cold, so water warmed by bodies can never exceed body temperature. It needs an engine and a heat pump in between.

How much electricity can body heat generate?

Very little with current wearables. A Peltier module on the arm produced about 5 µW at rest and up to about 50 µW while cycling. The theoretical Carnot limit between a 37°C body and a 25°C room is 3.9%.

At what temperature does water boil in Bengaluru?

About 96.9°C at the city's roughly 920 m altitude, by the barometric formula and the Clausius–Clapeyron relation. At Leh it is near 88.5°C.

Sources

  1. Byrne NM et al.: Metabolic equivalent: one size does not fit all, Journal of Applied Physiology (2005) (1 MET definition; measured 0.84 kcal/kg/h in 769 adults)
  2. Proto A et al.: Thermal energy harvesting on the bodily surfaces of arms and legs through a wearable thermo-electric generator, Sensors (2018) (5 µW at rest, up to 50 µW cycling, under 2.5°C across the module)
  3. Hymczak H et al.: Core temperature measurement, principles of correct measurement, problems, and complications, IJERPH (2021) (core 37 ± 0.5°C; heat-loss pathway shares)
  4. Peking University News: Thermoelectric elastomer offers new energy solution for wearables (2025) (ZT 0.49)
  5. Electronics For You: Body heat film powers wearables (2025) (UNIST ionic film, 1.5°C difference)
  6. Rumford (Benjamin Thompson): An Inquiry concerning the Source of the Heat which is excited by Friction, Philosophical Transactions of the Royal Society (1798), transcription (18.77 lb of water, two horses, boiled at 2 h 30 min; the story)
  7. Wikipedia: Benjamin Thompson (Munich arsenal, blunt borer, caloric theory)
second law of thermodynamicsCarnot efficiencyheat pumpthermoelectric generatormetabolic rateboiling pointClausius-Clapeyronchai
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Foodblog.

Keep reading in Foodblog

From the other blogs