India uses about 124 million tonnes of milled rice a year by USDA's 2026 estimate, or roughly 3.4 lakh tonnes a day. Poured into one heap, that day's rice would form a cone about 54 m tall, as high as an 18-storey building, and 173 m across at the base. Cooking it would take about 70 GWh of heat and 6.8 lakh tonnes of water, and cooking it as a single pot would take about a century, because heat creeps through packed cooked rice at the pace of conduction.
Calculated in this post from USDA's 124 million tonnes of milled-rice consumption for 2025/26, a bulk density of 800 kg/m³ and an angle of repose of 32° (both inside the ranges measured by Bhat and Riar), and with these example cooking inputs: 2 kg of water per kg of rice, heated from 25°C to 100°C.
- 3.4 lakh tonnes of milled rice leave India's stock every day (USDA, 2025/26).
- 54 m tall, 173 m wide: the one-day cone at rice's 32° angle of repose.
- About 70 GWh of heat and 6.8 lakh tonnes of water would cook it.
Calculated in this post from USDA's 124 million tonnes a year and the example cooking inputs in the energy section below.
- About a century to cook it as one pot, because heat only crawls through packed rice by conduction.
- Cube-root scaling: a whole year's rice in one cone is only about 7 times taller.
Rice is the one food most Indian kitchens cook every single day, and big numbers like "124 million tonnes" mean nothing until you can see them. So I'll do what an engineer would do with a heap of grain:
- use its bulk density to get a volume;
- use its angle of repose to get a shape;
- use a few heat-transfer numbers to see what cooking it would involve.
How much rice does India eat in a day?
About 3.4 lakh tonnes. USDA's New Delhi office estimates India's milled-rice consumption and residual use at 124 million tonnes for the 2025/26 marketing year, and forecasts 128 million tonnes for 2026/27.
"Consumption and residual" includes some feed, seed and losses, not only plates of rice, so this is the rice that leaves the national stock each day.
Spread over roughly 1.45 billion people (an assumption for 2026), it is about 234 g each, a little over one cup of raw rice per person.
In August 2010 the trouble was a pile too big to store. Court-appointed food commissioners told the Supreme Court that about 50,000 tonnes of government grain had already gone bad, and that about 1.35 lakh tonnes had been lying in the open in Punjab alone since 2008–09. News reports showed wheat and rice rotting in the open in Punjab, Uttar Pradesh, Haryana and Rajasthan; in Jaipur, an FCI godown held liquor bottles while wheat bags outside rotted in the rain.
On 13 August a bench of Justices Dalveer Bhandari and Deepak Verma told the government to give the excess grain to the hungry free rather than let it rot. Food minister Sharad Pawar called that a "suggestion" he could not implement, and called the reports exaggerated. On 31 August the bench answered: "Tell your minister that free food distribution is our order, not a suggestion by us."
Governance Now: SC slaps Pawar; says free food distribution is our order (31 August 2010)See the data as a table
| Consumption and residual (million tonnes) | Milled production (million tonnes) | |
|---|---|---|
| MY 2024/25 | 121.4 | 150.2 |
| MY 2025/26 | 124 | 152 |
| MY 2026/27 (forecast) | 128 | 150 |
Source: USDA FAS New Delhi, Grain and Feed Annual IN2026-0019 (Post estimates) · Marketing year runs October to September. Exports take most of the gap; stocks are near record highs.
The report sets out USDA's production-supply-distribution tables for Indian rice. For MY 2025/26 the Post estimates milled production at 152 MMT from 52 million hectares, consumption and residual at 124 MMT, exports at 24 MMT and ending stocks at 52 MMT. For MY 2026/27 it forecasts production of 150 MMT, consumption of 128 MMT and record exports of 25 MMT, as the government releases more grain at subsidised prices to run down large stocks. The limit: these are attaché estimates, not official Government of India figures, and "residual" hides losses and feed use.
Read the report (free, via APEDA)How tall would the pile be?
About 54 m, with a base 173 m wide, because loose rice settles into a cone whose sides slope at its angle of repose, close to 32 degrees.
The angle of repose is the steepest slope a heap of grains can hold before the surface slides. It comes from friction between grains, the same Mohr–Coulomb friction that governs sand dunes and soil banks.
Pour rice from a point and it builds a cone at that angle, whatever the quantity. That gives the geometry:
- Volume from bulk density. Bhat and Riar measured bulk densities of 750 to 872 kg/m³ for Indian brown rice cultivars; I use 800 kg/m³:
\[ V = \frac{\dot m}{\rho_b} = \frac{3.40\times10^{8}}{800} \approx 4.25\times10^{5}\ \text{m}^3 \]
- Cone shape. A cone of height \(h\) and slope angle \(\theta\) has radius \(r = h/\tan\theta\), so
\[ V = \frac{\pi r^2 h}{3} = \frac{\pi h^3}{3\tan^2\theta} \quad\Rightarrow\quad h = \left(\frac{3V\tan^2\theta}{\pi}\right)^{1/3} \]
- Plug in \(\theta\) = 32° (\(\tan\theta\) = 0.625):
\[ h = \left(\frac{3 \times 4.25\times10^5 \times 0.390}{\pi}\right)^{1/3} \approx 54\ \text{m}, \qquad r = \frac{54}{0.625} \approx 87\ \text{m} \]
- Footprint. \( \pi r^2 \approx 23{,}600\ \text{m}^2 \), about 2.4 hectares, roughly three football pitches.
The authors measured engineering properties of 12 traditional paddy cultivars and their brown rice from Kashmir, in triplicate on 1 kg lots. Brown rice bulk density ranged from 750 to 872 kg/m³, true density from 1,263 to 1,465 kg/m³, porosity from 32% to 48%, and angle of repose from 28.5° to 36.2°; paddy was lighter (478 to 621 kg/m³) and steeper (33.7° to 46.1°). These are the numbers that size silos, hoppers and conveyors. The limit: traditional temperate cultivars, not the high-yield varieties that fill most of India's supply.
Read the paper (free)How much would the pile weigh down on itself?
At most about 4.2 bar under the peak, the weight of a 54 m column of rice, a load grain kernels carry routinely in stores and silos.
Calculated in this post as ρgh, from a bulk density of 800 kg/m³ (inside the 750 to 872 kg/m³ range measured by Bhat and Riar) and the 54 m cone height worked out above.
Grains are not a liquid
That is an upper bound. Granular materials do not behave like liquids: grains wedge against each other and form arches that carry part of the load sideways, which is why a tall silo's floor feels far less than \(\rho g h\). Engineers call this the Janssen effect.
A free cone has no walls to lean on, so the arches redirect load outward instead, and the pressure under the very top is lower than the simple column suggests. Either way, rice kernels survive it: grain is routinely stored in heaps and silos of this scale.
A pile this size would breathe
Stored grain respires slowly, and a heap 54 m deep traps heat and moisture, so warehouse rice is kept in bags with aisles and fumigated stacks rather than one mountain. India's real stock, roughly 52 million tonnes at the end of 2025/26 by USDA's estimate, is about 150 of these daily cones.
How much energy would it take to cook?
About 70 GWh of heat, the output of a 3 GW power station running for a full day, plus 6.8 lakh tonnes of water.
Calculated in this post from the daily rice (USDA's 124 million tonnes a year divided by 365) and the example inputs below; the power-station comparison is just 3 GW × 24 hours = 72 GWh.
- Recipe. Assume a common absorption method, 2 kg of water per kg of rice, heated from 25°C to 100°C. I take raw rice's specific heat as about 1.6 kJ/kg·K (an assumption for low-moisture grain) and water's as 4.18 kJ/kg·K.
- Heat per kg of rice.
\[ q = (1 \times 1.6 + 2 \times 4.18)\ \tfrac{\text{kJ}}{\text{K}} \times 75\ \text{K} \approx 747\ \text{kJ per kg rice} \]
- Whole pile. \( 3.40\times10^{8} \times 747\times10^{3} \approx 2.5\times10^{14}\ \text{J} \approx 70\ \text{GWh} \).
- As LPG. At 46 MJ/kg and an assumed 60% stove efficiency, about 9,200 tonnes of LPG, before any heat lost while simmering.
Calculated in this post from the 70 GWh above, the net calorific value of LPG (propane 46.3 and butane 45.8 MJ/kg in the Flogas data sheet) and an assumed 60% stove efficiency (an example input).
- Water. \( 2 \times 3.40 \) lakh t ≈ 6.8 lakh tonnes, enough to fill about 270 Olympic swimming pools at 2,500 m³ each.
Calculated in this post as 2 kg of water per kg of the 3.40 lakh tonnes of rice from USDA's 124 million tonnes (1 tonne of water is about 1 m³), divided by the pool volume.
Real cooking uses more energy than this, because pots lose heat for 15 to 25 minutes while the rice simmers, and pressure cookers recover some of it. The pressure-cooker side of that story is in how many whistles dal and rice need.
Pile and cook India's rice
How long would it take to cook as one pot?
Decades, by conduction, because once rice swells into a packed mass, heat can only seep through it at about 1.5 × 10⁻⁷ m²/s, and the centre of a pot this size is 50 m from the nearest hot wall.
Imagine a cylindrical pot holding the rice and its water: about 11 lakh m³, roughly 100 m wide and 140 m tall. The cooking goes in two phases:
Calculated in this post as the rice volume (about 4.25 lakh m³, from the 800 kg/m³ bulk density in Bhat and Riar) plus 6.8 lakh m³ of water; the 100 m by 140 m shape is an imagined example, not a real pot.
- Convection. At first the water circulates freely and heats far faster than conduction would allow.
- Lock-up. Rice starch gelatinises between roughly 70 and 80°C, and the Wariyah study found that above that point water uptake switches from a first-order process to a constant-rate one controlled by the swelling grain. The grains swell, soak up the free water and lock into a dense, sticky mass.
From then on convection stops and the Fourier clock takes over:
I borrow \(\alpha\) from potato, which has a similar water and starch content; treat the answer as an order of magnitude. The edges would be burnt while the middle had not yet reached cooking temperature.
Each grain itself needs only 15 to 23 minutes in boiling water to lose its white core.
The problem is entirely getting heat to the grain, which is why India cooks its rice in hundreds of millions of small pots every day rather than one big one.
Three Ethiopian rice varieties were parboiled with soaking at 40 to 80°C and steaming for 10 to 50 minutes, then tested. Minimum cooking time, measured by pressing 2 g samples cooked in boiling water between glass plates until no white core remained, ranged from 16 to 23 minutes for Gumara and Edget and 15 to 20 minutes for Narica-4, rising with harsher parboiling. Mean water uptake ratios ran from 3.25 to 4.20. The limit: parboiled African varieties, so treat the times as typical, not specific to Sona Masoori or basmati.
Read the paper (free)What changes if the assumptions move?
The height is remarkably robust, because it scales with the cube root of the volume.
| Change | Height | Base diameter |
|---|---|---|
| Baseline: 800 kg/m³, 32° | 54 m | 173 m |
| Shallow slope, 28.5° | 49 m | 181 m |
| Steep slope, 36.2° | 60 m | 164 m |
| Light rice, 750 kg/m³ | 55 m | 177 m |
| Dense rice, 872 kg/m³ | 53 m | 168 m |
| 2026/27 forecast, 128 Mt/yr | 55 m | 175 m |
| A full year in one cone | about 390 m | about 1.2 km |
Doubling the rice raises the cone only by \(2^{1/3}\), about 26%. A year's rice, 365 times the volume, is only about 7 times taller.
Calculated in this post from the cone formula above, in which height grows with the cube root of the volume at a fixed angle of repose (32°, inside the range measured by Bhat and Riar): the cube root of 365 days of rice is about 7.1.
Piling it is entirely possible: 3.4 lakh tonnes of milled rice settles into a cone about 54 m tall and 173 m wide, its slope fixed at about 32° by grain friction, with no more than about 4 bar under the peak. Cooking it as one pot is the impossible part: after the starch gelatinises, conduction through 50 m of packed rice would take on the order of a century. Split into ordinary pots, it needs about 70 GWh of heat and 6.8 lakh tonnes of water, which is roughly what Indian kitchens already do every day.
Calculated in this post from USDA's 124 million tonnes a year, a bulk density and angle of repose inside the ranges measured by Bhat and Riar, and the example cooking inputs above.

Why some rice cooks fluffy and some sticky is down to amylose, explained in the science of rice from basmati to Sona Masoori, and how quickly that rice reaches your blood is in the glycaemic index of Indian staples. One of the grandest things India does with rice is told in the history of biryani in India. More from the Taste Lab is on the Rasa world page.
What's new
- USDA forecasts record Indian rice exports of 25 MMT in 2026/27 as stocks pile up USDA FAS via APEDA, Apr 2026
- Government to sell FCI rice to ethanol distilleries at ₹2,320 per quintal under OMSS 2026-27 Rural Voice, Jul 2026
What if next?
- What if idli batter fermented for a year? Two-thirds of idli batter is rice. Leave it alone long enough and the microbes take over.
- What if you cooked biryani with a lightning bolt? A billion joules arrives in microseconds. Rice needs minutes.
- What if you ate only one taste for a week? After 234 g of rice a day, what happens to a tongue on a single note.
Questions people ask
How much rice does India consume per year?
About 124 million tonnes of milled rice in the 2025/26 marketing year by USDA's estimate, including some feed and losses, with 128 million tonnes forecast for 2026/27. India produced about 152 million tonnes.
What is the angle of repose of rice?
Measured values run from about 28° to 36° for brown rice and 34° to 46° for paddy, depending on variety and moisture, according to a study of 12 Indian cultivars.
What is the bulk density of rice?
About 750 to 870 kg/m³ for brown rice and 480 to 620 kg/m³ for paddy in the same study, so a cubic metre of milled rice weighs roughly 800 kg.
How long does rice take to cook?
Minimum cooking times of 15 to 23 minutes in boiling water have been measured for several varieties, judged by pressing grains until no white core remains.
Sources
- USDA FAS New Delhi: India Grain and Feed Annual, IN2026-0019 (2026) (124 MMT consumption 2025/26, 128 MMT forecast, 152 MMT production, 52 MMT stocks)
- Bhat FM, Riar CS: Cultivars effect on the physical characteristics of rice of temperate Kashmir, Journal of Food Science and Technology (2016) (bulk density 750–872 kg/m³, angle of repose 28.5–36.2°)
- Meresa A et al.: Effect of parboiling conditions on physical and cooking quality of selected rice varieties, International Journal of Food Science (2020) (minimum cooking time 15–23 min)
- Wariyah C, Anwar C, Astuti M, Supriyadi: Kinetics of water absorption in rice, Agritech 27(3) (2007) (first-order uptake below gelatinisation, zero-order above)
- Moro Pisco JF, Salas-Valerio WF: Thermal diffusivity of whole potato, Revista Colombiana de Investigaciones Agroindustriales (2020) (α ≈ 1.5 × 10⁻⁷ m²/s, used as a proxy)
- Rural Voice: Government to sell FCI rice for ethanol at ₹2,320/qtl, OMSS policy 2026-27 (2026)
- Governance Now: SC slaps Pawar; says free food distribution is our order (31 Aug 2010) (50,000 tonnes spoiled, 1.35 lakh tonnes in the open in Punjab, the 13 and 31 August orders; the story)
