Efficiency, waste heat and recovery: Efficiency gains in history and energy intensity
Understanding Energy / Module 12: Efficiency, waste heat and recovery Part 2 of 5Efficiency gains in history and energy intensity
Better converters built the modern world as much as more fuel did. Steam engines, lamps, furnaces and ammonia plants all became many times more efficient. At the scale of whole economies, these gains show up as falling energy intensity, the energy used per unit of GDP. This part traces that history and today's pace.
2.1 History's converters improved tenfold or more, again and again
Smil argues that the long rise of industrial civilisation owes as much to better machines as to more coal and oil. The steam engine is the classic case. Thomas Newcomen's atmospheric engines of the early eighteenth century turned only about 0.5 to 1% of their coal into work. James Watt's engines of the 1780s, with a separate condenser, reached about 2 to 5%. By 1900 the best triple-expansion steam engines reached about 15 to 20%, and modern supercritical steam plants reach about 45%. Each step cut the coal needed per unit of work several times over.
Light followed the same path, only faster. Candles and oil lamps turned a tiny fraction of their fuel into light. Gas mantles, carbon-filament bulbs, tungsten bulbs, fluorescent tubes and now LEDs each multiplied the light per unit of energy. An LED today gives dozens of times the light per watt of an early electric bulb.
Industry shows the same pattern. Making a tonne of pig iron took about 200 GJ around 1800, mostly as charcoal or coke. A good integrated mill today needs under 20 GJ per tonne of finished steel, a tenfold fall. After 1950, natural gas feedstock and large single-train plants more than halved the energy needed per tonne of ammonia.
These gains share a feature. They came from many small improvements over decades, not from one breakthrough, and they were driven as much by cost and reliability as by a wish to save energy. They also came with huge growth in total use. Each improvement made the service cheaper, which opened new markets. Efficiency and expansion have usually travelled together, a point that returns in the discussion of rebound.
| Service | Early converter | Modern converter | Gain |
|---|---|---|---|
| Work from coal | Newcomen engine, about 0.5 to 1% | supercritical steam plant, about 45% | about 45 to 90 times (calculated) |
| Electricity from gas | early gas turbine, under 20% | combined cycle, about 60% | about 3 times |
| Light from electricity | carbon-filament bulb | LED | dozens of times |
| Iron and steel | about 200 GJ per t of pig iron, 1800 | under 20 GJ per t of steel | about 10 times |
| Ammonia | early coal-based plants | modern gas plants, about 28 GJ per t | more than 2 times since 1950 |
Rounded values; the gains compound with growth in scale and falling costs.
Source: Smil, Energy and Civilization: A History (2017), via Smil Mastery Course, M1 'Conservation and efficiency'- Smil, Energy and Civilization: A History (2017) · via Smil Mastery Course, M1 'Conservation and efficiency'
- Smil, Energy: A Beginner's Guide (2006) · via Smil Mastery Course, M1 'Conservation and efficiency'
- Smil, Still the Iron Age: Iron and Steel in the Modern World (2016) · via Smil Mastery Course, M11 'Steel today: energy, stocks and the future'
2.2 Energy intensity measures an economy's efficiency, but it hides a lot
Energy intensity is total primary energy divided by GDP, often in megajoules per dollar. A lower number means an economy produces more output for each unit of energy. Smil shows that intensity usually follows a hump. It rises in early industrialisation, as countries build steel mills, cement works, railways and power grids. Then it falls as economies mature, shift towards services, and adopt more efficient machines.
China is a striking case. In China's Past, China's Future, Smil shows that its energy intensity began falling even before the 1979 reforms and then dropped steeply through the 1980s and 1990s, unusually fast for a country still industrialising. India's intensity has also been falling for decades.
Smil warns that the number must be taken apart before comparing countries. It depends on whether GDP is converted at market exchange rates or at purchasing-power parity, which can change a poor country's apparent intensity by a factor of two or three. It depends on whether traditional biomass, such as firewood and dung, is counted. It depends on climate: a cold country needs more heating. It depends on structure: an economy with aluminium smelters looks more intensive than one with software firms. And it depends on trade.
That last point matters most for rich countries. When Britain or the United States imports steel, cement, chemicals and manufactured goods instead of making them, the energy used to make them is counted in China, India or Vietnam. Domestic intensity falls, but the world's energy use does not. Part of the decline in rich countries' intensity is efficiency; part is offshoring.
The measure is still useful. Tracked over time within one country, with structure and trade in mind, it shows whether an economy is getting more out of each unit of energy.
- Change in energy intensity (energy per unit of GDP)
- True efficiency
- Better engines, motors, boilers and lamps
- Less waste in buildings and industry
- Structure
- Shift from heavy industry to services
- Growth or decline of energy-intensive sectors
- Trade
- Importing energy-intensive goods lowers domestic intensity
- Exporting them raises it
- Measurement
- Exchange rates versus purchasing-power parity
- Whether traditional biomass is counted
- Climate and weather
- True efficiency
- Smil, Energy at the Crossroads: Global Perspectives and Uncertainties (2003) · via Smil Mastery Course, M13 'Energy intensity'
- Smil, China's Past, China's Future: Energy, Food, Environment (2004) · via Smil Mastery Course, M13 'Energy intensity'
2.3 Global efficiency is improving at 1 to 2% a year, half the rate governments promised
The IEA tracks global progress on efficiency through the rate at which primary energy intensity falls. Its Energy Efficiency 2025 report found that progress was about 1% in 2024 and was set to reach about 1.8% in 2025, the best year since the late 2010s. That sounds positive, but it is well below the target. At the COP28 climate conference in 2023, governments agreed to double the global rate of improvement by 2030, to about 4% a year. Since 2019 the average has been about 1.3% a year, down from about 2% in 2010 to 2019.
Why the slowdown? Industry is one reason. Around two-thirds of the growth in global final energy demand since 2019 came from industry, and industrial energy intensity has improved by under 0.5% a year recently, against almost 2% in the previous decade. Electricity is another. Demand for it has grown two to three times faster than total energy demand since 2019, and in some regions the extra power has come from less efficient fossil plants.
There are bright spots. The IEA estimates that India's energy intensity improved by over 4% in 2025, above its average since 2019, and China's by over 3%. Efficiency-related investment was set to reach almost 800 billion US dollars in 2025, over 70% above 2015, and about 18 million people worked in efficiency jobs in 2024. Two-thirds of end-use investment, however, is still in China, the United States and the European Union.
For investors the gap between 1.8% and 4% is the market. Closing it means replacing motors, boilers, air conditioners, vehicles and industrial furnaces faster than they would wear out, and capturing waste heat that is now thrown away.
| Period or year | Average improvement per year |
|---|---|
| 2010 to 2019 | about 2% |
| 2019 to 2024 | about 1.3% |
| 2024 | about 1% |
| 2025 (estimate) | about 1.8% |
| COP28 goal for 2030 | about 4% |
| India, 2025 | over 4% |
| China, 2025 | over 3% |
See it in the table
Gas Natural gasAl AluminiumSt SteelW TungstenCem CementNH3 AmmoniaCheck your understanding
Why does a rich country's falling energy intensity overstate its efficiency gains?
Offshoring steel, cement and manufacturing lowers domestic energy use per dollar without lowering global use. Smil stresses that such comparisons must be taken apart.What rate of global energy intensity improvement did governments agree at COP28 to reach by 2030?
The goal is to double the historical rate to about 4% a year. Progress was about 1.8% in 2025 and averaged 1.3% since 2019.Which statement best describes how efficiency gains have happened historically?
Steam engines, lamps, steel and ammonia all improved step by step over decades, and each gain made the service cheaper, which expanded use.