Chapter III · Alcoholic fermentation
How much heat fermentation releases, why a large qvevri or tank heats up more than a small vessel, and how much cooling a fermentation needs to stay within the desired temperature range.
Heat balance: m·cₚ·dT/dt = Q_ferm − U·A·(T − T_env) − Q_cool. Q_ferm = 0.546 kJ per g of sugar fermented; fermentation rate from the previous page's kinetic model, varying with temperature. Heat-transfer coefficients U (W/m²·K): qvevri 1.8, steel 6, barrel 3.5, carboy 5; cₚ = 3.85 kJ/(kg·K).
For each vessel type the peak temperature is computed from 10 to 10 000 litres, using the must and temperatures of the first mode. Surface area grows with volume to the power 2/3, heat release in proportion to volume.
The demand is computed with unlimited cooling holding the temperature at the target. Ice: 334 kJ/kg to melt + warming the water to the target. At home, frozen water bottles are placed in the vessel and swapped 1–3 times a day.
Fermentation is a chemical furnace. Every gram of sugar fermented releases more than half a kilojoule of heat, and if that heat went nowhere, an ordinary must would warm by almost 30 °C during fermentation — to a temperature that kills yeast. In reality the heat passes through the vessel wall to the surroundings, and the temperature of the must is set by the balance of these two flows — generation and loss.
The key player in this balance is the size of the vessel. Heat generation is proportional to volume, heat loss to surface area. A 20-litre carboy has a lot of surface per litre and barely warms up; a 2000-litre qvevri has much less, and its temperature rises by more than 10 °C. On this page we calculate how temperature changes during fermentation, what difference the type and size of vessel makes, and how much cooling is needed.
The warming of must during fermentation has been known since ancient times — the very word "fermentation" (from the Latin fervere, "to boil") describes the phenomenon of bubbles and heat. The first ways of controlling temperature were simple: placing the vessel somewhere cool, in the ground or a cave, using small vessels, pouring cold water over it.
At the end of the 19th century, after the work of Pasteur and Müller-Thurgau, winemakers learned that excess heat stops fermentation and destroys aroma. From the mid-20th century, double-walled (jacketed) steel tanks with a cooling fluid spread, and in the 1960s–70s cool fermentation (12–16 °C) became the basis of the modern white-wine style. In 1980 Roger Boulton linked the kinetic model of fermentation to the heat balance and created a method for calculating the cooling systems of large wineries.
When glucose is converted to ethanol and CO₂, only part of the energy released is stored in ATP; the rest becomes heat. The value accepted in the winemaking literature is ≈ 98 kJ per mole of glucose (23.5 kcal), i.e.:
$$q_{\text{ferm}} \approx 0.546\ \text{kJ/g sugar}$$
The adiabatic limit — the temperature rise if no heat were lost at all:
$$\Delta T_{\text{ad}} = \frac{S \cdot q_{\text{ferm}}}{\rho\,c_p}$$
where $S$ is the sugar (g/L), $\rho \approx 1.08$ kg/L the must density and $c_p \approx 3.85$ kJ/(kg·K) the specific heat capacity. For 216 g/L of sugar, $\Delta T_{\text{ad}} \approx 28$ °C.
The power of heat release is proportional to the fermentation rate: $Q_{\text{ferm}} = r_S \cdot V \cdot q_{\text{ferm}}$, where $r_S$ is the rate of sugar consumption (g/L/s). At 30 g/L/day every 1000 litres releases ≈ 190 W of heat — as much as two old light bulbs, but continuously, day and night.
The change in must temperature is a balance of heat flows:
$$m\,c_p\,\frac{dT}{dt} = Q_{\text{ferm}} - U A\,(T - T_{\text{env}}) - Q_{\text{cool}}$$
The second term is Newton's law of cooling: heat lost through the wall is proportional to the surface area $A$ and the temperature difference. The heat-transfer coefficient $U$ (W/m²·K) depends on the wall material and surroundings: for a steel tank standing in air ≈ 5–8 (natural convection and radiation), for an oak barrel ≈ 3–4 (wood is a poor conductor), for a buried qvevri effectively ≈ 1.5–2.
A vessel's thermal time constant $\tau = m\,c_p / (U A)$ tells us how quickly it "forgets" a temperature change. For a 20-litre carboy it is half a day — the must quickly returns to the surrounding temperature; for a 1000-litre steel tank ≈ 1.4 days; for a 2000-litre qvevri ≈ 6 days — as long as the fermentation itself, so heat accumulates.
For vessels of similar shape, surface area is proportional to volume to the power 2/3: $A \propto V^{2/3}$, and surface per litre $A/V \propto V^{-1/3}$. Make a vessel 100 times bigger and the surface per litre falls 4.6-fold. So:
In a 20-litre carboy the must warms by 1–2 °C during fermentation — the room temperature is in practice the fermentation temperature. In a 225-litre barrel by ≈ 5 °C. In a 1000-litre tank by 5–8 °C, in a 5000-litre tank by 8–12 °C. In a large qvevri, with the pomace, by 10–15 °C.
A buried qvevri is a distinctive thermal system. The clay wall (3–5 cm) and the surrounding earth conduct heat poorly, but the heat capacity of the ground is enormous, and at a depth of 1–2 metres its temperature is almost constant all year — ≈ 12–15 °C in the Georgian lowlands. Steady-state heat loss from a spherical vessel into an infinite medium is $Q = 4\pi k R\,\Delta T$, i.e. an effective $U \approx k/R$; for soil ($k \approx 1–1.5$ W/(m·K)) and a qvevri radius of 0.6–0.8 m this is 1.5–2 W/(m²·K).
As a result a qvevri heats up and cools down slowly during fermentation: at the peak, a 1000–2000-litre qvevri reaches 27–30 °C, and after fermentation it gradually falls to ground temperature and stays stable all winter — ideal conditions for the wine to clear and for malolactic fermentation.
For the aroma of white wine, fermentation is best at 12–18 °C: at low temperature the volatile esters and thiols stay in the wine and less is carried off by CO₂. Red and amber wines need 24–30 °C to extract phenolics (see "Phenolic compounds"). Above 32 °C the danger begins: the toxicity of alcohol rises sharply, fermentation may stick, aroma is lost and the risk from acetic bacteria grows.
Jacketed tanks — a cooling fluid (water, glycol) circulates in the double wall, controlled by a thermostat; the most precise method. Immersion coils and plates ("cooling flags") — movable, usable in any vessel. Pumping the must through a heat exchanger — which also wets the pomace cap. At home: placing the vessel in a cool room or cellar, wrapping it in a wet cloth (evaporative cooling), standing it in a cold-water bath, putting frozen water bottles inside.
Ice equivalent: melting 1 kg of ice absorbs 334 kJ, and warming the meltwater to the target temperature another $4.19\,T_{\text{target}}$ kJ. So 1 kWh of cooling (3600 kJ) is ≈ 8–9 kg of ice. A compressor chiller removes about 3 kWh of heat per 1 kWh of electricity (COP ≈ 3).
| Quantity | Value | Note |
|---|---|---|
| Heat of fermentation | 0.546 kJ/g sugar | ≈ 98 kJ/mol |
| Adiabatic limit (216 g/L) | ≈ 28 °C | with no heat loss |
| Specific heat of must | ≈ 3.85 kJ/(kg·K) | water 4.19 |
| U — steel tank in air | 5 – 8 W/(m²·K) | convection + radiation |
| U — oak barrel | 3 – 4 W/(m²·K) | — |
| U — buried qvevri | 1.5 – 2 W/(m²·K) | effective |
| White wine fermentation | 12 – 18 °C | aroma retention |
| Red / amber | 24 – 30 °C | extraction |
| Danger threshold | ≥ 32 °C | stuck fermentation |
| Ice | 334 kJ/kg | ≈ 8.5 kg per kWh |
The previous page, "Fermentation kinetics", computes the fermentation rate at constant temperature. Here the same model is coupled to the heat balance: temperature rises → fermentation speeds up → heat is released faster. This positive feedback explains why the temperature in a large vessel can rise unexpectedly fast. "Harvest and must yield" already showed the adiabatic limit in the qvevri-filling mode; here it is replaced by realistic dynamics. Chapter V (The qvevri method) covers managing the pomace cap — punching down and stirring — which also evens out the temperature in the qvevri.
Burying the qvevri in the ground is thermal engineering thousands of years old: the earth takes up the heat of fermentation slowly and keeps the wine at a stable, cool temperature through winter. The size of traditional qvevri — usually 500–2000 litres — makes sense from this point of view too: in such a qvevri the fermentation peak is 25–30 °C, which is optimal for making wine on the pomace by the Kakhetian method. In very large qvevri (over 3000 litres) and in a hot autumn the temperature can exceed 32 °C — so such qvevri are sometimes filled less, the pomace is punched down more often, or, in modern winemaking, an immersion cooling coil is used.
At home, in glass carboys and plastic containers (20–60 litres), the must temperature follows room temperature almost exactly — so the main thing is choosing the right room: 15–18 °C for white, 20–25 °C for red.
| Term | Definition |
|---|---|
| Heat of fermentation | Heat released during fermentation per gram of sugar |
| Adiabatic heating | Temperature rise with no heat loss |
| Heat-transfer coefficient (U) | Power lost through the wall per m² per degree |
| Newton's law of cooling | Heat loss is proportional to the temperature difference |
| Thermal time constant | The time for a vessel to approach the surrounding temperature |
| Surface-to-volume ratio | Surface per litre — falls as size grows |
| Jacketed tank | A double-walled tank with a cooling fluid |
| Cooling power | Heat removed by the cooler per second (W, kW) |
| Heat exchanger | A device transferring heat from one fluid to another |
A 2000-litre qvevri buried in the ground (ground temperature 14 °C), must at 22 Brix, initial temperature 18 °C, nitrogen 200 mg N/L. The qvevri surface is ≈ 8.8 m², i.e. 4.4 m² per cubic metre. At the fermentation peak heat release is 0.42 kW, and the temperature reaches 29.3 °C on day 8 — 11 °C above the start, though far below the adiabatic limit (+28 °C). Fermentation finishes in ≈ 11 days — faster than at a constant 20 °C (15 days), precisely because of the warming. With the same must a 1000-litre qvevri warms to 27 °C, a 300-litre one to 23 °C. A 1000-litre steel tank in an 18 °C cellar warms to ≈ 23 °C, a 5000-litre one to 26 °C, a 225-litre barrel to 23 °C, and a 20-litre carboy only to ≈ 20 °C.
If the 2000-litre qvevri must not rise above 22 °C, ≈ 0.28 kW of cooling is needed at the peak and ≈ 22 kWh over the whole fermentation (≈ 7 kWh of electricity with a chiller), the equivalent of ≈ 190 kg of ice: about 135 frozen 1.5-litre bottles, up to 40 a day on the peak days. This shows that ice is impractical for large volumes and mechanical cooling or a lower fill is better; for a 50-litre vessel a few bottles are enough.
1. In the "Temperature curve" mode choose the vessel — qvevri, steel tank, barrel or carboy — and enter its volume. Enter the ground or cellar temperature, the initial must temperature, the sugar and the nitrogen.
2. The chart shows how the must temperature changes during fermentation. If a card warns you in red (≥ 32 °C), there is a danger of a stuck fermentation — cooling or a smaller volume is needed.
3. The "Effect of size" mode shows how the peak temperature rises with vessel size for all four types. It helps you choose the size of vessel.
4. In the "Cooling plan" mode enter the target maximum temperature and the cooling power available. You get how much power and energy the fermentation needs and how many frozen bottles would replace a chiller.
5. For small home vessels (20–60 litres) the main thing is choosing the right room: the must temperature follows the room temperature, only 1–3 °C higher.
1. The heat-transfer coefficients are typical values. For a steel tank U depends strongly on air movement (a fan raises it 2–3-fold) and on whether the wall is wet; for a barrel on wood thickness and moisture. If you measure the temperature in your vessel, adjust U until the model matches.
2. The qvevri model treats the ground as a constant-temperature reservoir. In reality the earth around a qvevri warms gradually during fermentation and the effective U falls over time; neighbouring qvevri also warm one another. So in a cellar with several qvevri the real peak can be 1–3 °C higher.
3. The model treats the must as uniform. In a large vessel, especially near the pomace cap, temperature is uneven: the cap can be 5–8 °C hotter than the bottom layer. Measure under the cap and in the middle; punching down evens out the temperature.
4. Water vapour and ethanol carried off by CO₂ (evaporative cooling) amount to about 3–5% of the fermentation heat and are not in the model — a small, conservative error.
5. When planning a cooling system, size it at 1.3–1.5 times the peak demand: several vessels may ferment at once, and heat-exchanger efficiency falls as the temperature difference shrinks.
1. Why does a 2000-litre qvevri heat up much more during fermentation than a 20-litre carboy?
The surface-to-volume ratio scales as V^(−1/3): make a vessel 100 times bigger and the surface per litre falls 4.6-fold. That is why a large vessel sheds heat poorly.
2. How much would a must with 220 g/L of sugar warm up if no heat were lost at all?
220 g/L × 0.546 kJ/g = 120 kJ per litre; 1.09 kg × 3.85 kJ/(kg·K) ≈ 4.2 kJ/K → ΔT ≈ 28 °C. That would kill the yeast — which is why heat loss or cooling is essential.
3. Why does a buried qvevri stay relatively cool, even though its wall heat-transfer coefficient is low?
The ground around a qvevri has a large heat capacity and an almost constant temperature all year. Fermentation heat passes into it slowly but steadily. In very large qvevri this is not enough, and the temperature exceeds 30 °C.
4. Holding 5000 litres of must at 22 °C needs ≈ 1.3 kW of cooling at the peak. How much ice would replace 1 kWh of cooling?
1 kWh = 3600 kJ; melting ice takes 334 kJ/kg, and warming the meltwater to 22 °C another ≈ 92 kJ/kg. 3600 / 426 ≈ 8.5 kg of ice.