Power density: why land and space matter: Fossil fuels and thermal generation

3.1 Mines and oil fields deliver thousands of watts per square metre

A coal seam several metres thick contains a great deal of energy under each square metre. When a mine removes it at a high rate, the flow of energy per unit of land is enormous. Smil finds that large underground mines working thick seams exceed 10,000 W/m² of the surface land they occupy, because the workings themselves are underground. Most large surface mines operate at about 1,000 to 5,000 W/m², counting the pit, spoil heaps, roads and processing plant. The worst case is mountaintop removal in Appalachia, where thin seams under thick rock mean huge disturbance per tonne, and power density falls to 100 to 200 W/m².

Oil and gas fields are often even more compact. A well pad of a hectare or two can produce oil or gas carrying tens to hundreds of megawatts of energy, and the richest Middle Eastern fields deliver energy at very high rates from relatively few wells. Even including gathering lines, roads and processing plants, oil and gas production commonly runs at thousands of W/m².

India illustrates the point at scale. It mined a record 1,048 million tonnes of coal in the year to March 2025 (provisional figure), much of it from vast open-cast mines in Chhattisgarh, Odisha and Jharkhand. Indian coal is poor, often about 15 MJ per kilogram, yet the mines remain dense energy sources. Imagine a mine producing 50 million tonnes a year from 50 km² of disturbed land. Its output is 750 million GJ a year, a steady 24 GW of heat, or about 475 W/m². That is roughly 50 times the density of a solar park, even before considering that the mine's coal feeds plants turning it into electricity.

The local damage can still be severe. Open-cast mining in India has displaced forest-dwelling and tribal communities, removed forest, and created fires and dust. High power density means a small footprint per unit of energy, not a gentle one.

• Smil, Power Density: A Key to Understanding Energy Sources and Uses (2015) · via Smil Mastery Course, M8 'Power densities of fossil fuels and thermal generation'
• Smil, Oil: A Beginner's Guide (2008) · via Smil Mastery Course, M8 'Power densities of fossil fuels and thermal generation'
• Smil, Natural Gas: Fuel for the 21st Century (2015) · via Smil Mastery Course, M8 'Power densities of fossil fuels and thermal generation'
• Ministry of Coal / PIB, coal production FY 2024-25 (April 2025) · 1,047.57 Mt provisional, https://www.coal.nic.in/sites/default/files/2025-04/PIB2117280.pdf

3.2 Power plants are compact, and nuclear most of all

A large coal-fired power station occupies a site of a few square kilometres, including its boilers, turbine hall, coal yard, cooling towers and switchyard. A 2,000 MW plant running most of the year sends out an average of perhaps 1,200 to 1,500 MW from that site, a power density of hundreds of W/m². Gas-fired combined-cycle plants are smaller still, with no coal yard and no ash, and can reach over 1,000 W/m² on their sites.

The complete chain matters more than the site. Smil adds the land for mining, rail lines carrying coal, ash ponds and the plant itself. Coal-fired electricity then runs at roughly 100 to 1,000 W/m², one to two orders of magnitude above wind and solar. Ash disposal is a significant land claim in India, where coal often contains 35 to 45% ash and plants have accumulated large slurry ponds, some of which have breached and flooded nearby land.

Nuclear power is the densest of all common sources. A reactor site of one or two square kilometres can send out over 1,000 MW day and night. Because uranium's energy density is so high, even the mining, milling, enrichment and fuel fabrication chain occupies little land per unit of electricity. Smil puts nuclear generation including its fuel chain at hundreds to a few thousand W/m². The complications are a matter of accounting: should the exclusion zone around Chernobyl count, and how should a waste repository that must be guarded for millennia be valued? These are real issues, but they are not primarily questions of area.

Transmission adds a land claim to every system. High-voltage lines need cleared corridors several tens of metres wide, and some land use is constrained under them. For fossil and nuclear plants, which can be sited near demand, these corridors are relatively short. For remote solar parks in deserts or offshore wind, they can be hundreds of kilometres long and become a significant part of the system's footprint, and often its slowest part to approve and build.

• Smil, Power Density: A Key to Understanding Energy Sources and Uses (2015) · via Smil Mastery Course, M8 'Power densities of fossil fuels and thermal generation'
• Smil, Energy: A Beginner's Guide (2006) · via Smil Mastery Course, M8 'Power densities of fossil fuels and thermal generation'

3.3 Different accounting rules give very different land figures

In 2022 Jessica Lovering and colleagues published in PLOS ONE the most comprehensive comparison so far of how much land electricity sources use. They gathered about 1,400 observations from real sites in 73 countries and 45 US states, and counted both the land occupied by plants and the land used upstream to produce fuels. The medians differ by a factor of roughly ten thousand: nuclear needs only about 7 hectares for each TWh a year, while biomass grown on dedicated cropland needs some 58,000.

Converted into power density, several of their medians agree with Smil. Ground-mounted solar at about 2,000 ha/TWh/y equals about 5.7 W/m². Wind, counting the full spacing area, at about 12,000 ha/TWh/y equals about 1 W/m². Biomass from dedicated crops comes out at about 0.2 W/m². Nuclear, at 7.1 ha/TWh/y, works out to about 1,600 W/m² of electricity.

Other medians differ sharply. Lovering's coal figure of about 1,000 ha/TWh/y equals only about 11 W/m² of electricity, one to two orders of magnitude below Smil's range and only about twice the density of solar. Natural gas comes out at about 28 W/m² on its direct footprint, but about 6 W/m² if the spacing between wells across a gas field is counted. The main reasons lie in what is counted. Surface coal mines move steadily across the landscape, and a study that counts all land disturbed over a mine's life, including spoil and land awaiting restoration, finds far more area per unit of output than one that counts only the active workings in a given year. Gas field spacing is similar to wind spacing: most land between wells remains in other use.

The lesson is not that one set of figures is right. It is that land comparisons are only meaningful when the rules are stated. Smil's numbers are best read as the density of the core operations; Lovering's as the total area associated with a source, including land affected but not fully used. On either basis nuclear is the most compact source, solar parks are far denser than crop fuels, and wind's footprint depends heavily on whether shared land is counted.

• Lovering, Swain, Blomqvist and Hernandez, Land-use intensity of electricity production and tomorrow's energy landscape, PLOS ONE 17(7): e0270155 (2022) · web
• Smil, Power Density: A Key to Understanding Energy Sources and Uses (2015) · via Smil Mastery Course, M8 'Power densities of fossil fuels and thermal generation'

See it in the table

Check your understanding

A surface mine produces 50 million tonnes a year of coal at 15 MJ/kg from 50 km² of land. What is its approximate thermal power density?

Why does Lovering et al.'s coal figure come out much lower than Smil's?

Which source is the most compact on both Smil's and Lovering's accounting?