Energy statistics measure the same flow at different points, and mixing them up causes endless confusion. A chart can show coal supplying a quarter of world energy while another says electricity is only a fifth of what consumers use, and both can be right. The difference lies in where you count: at the mine and the wellhead, at the meter, or at the point where the energy finally does something useful. This guide walks through the stages, where the losses go, and the accounting choices that make wind, solar and nuclear look bigger or smaller depending on who publishes the numbers.
Four stages of energy
Energy passes through up to four stages on its way from nature to you:
| Stage | What it means | Examples |
|---|---|---|
| Primary | Energy as extracted or captured, before conversion | Crude oil, raw coal, natural gas, uranium, wind, sunlight, river flow, firewood |
| Secondary | Energy carriers produced from primary sources | Petrol, diesel, electricity, refined LPG, hydrogen, district heat |
| Final | Energy delivered to end users, measured at the meter or pump | Electricity into a factory, petrol into a car tank, gas into a boiler |
| Useful | The service delivered after the end-use device | Heat in a room, motion of wheels, light, cold air from an air conditioner |
Official statistics usually stop at final energy (sometimes called total final consumption). Useful energy is rarely measured at national level because it depends on millions of devices, each with its own efficiency. See the glossary entries for primary energy and final energy for short definitions.
How big the losses are
The biggest gap sits between primary and final energy in electricity generation. A coal plant converts roughly 33–45% of the fuel’s heat into electricity; the rest leaves through cooling towers and the stack. A combined-cycle gas plant reaches 55–63%. Refineries are more efficient, typically losing around 5–10% of crude’s energy in processing, and gas pipelines and electricity grids lose a few percent more in transport.
The second big gap sits between final and useful energy, inside the devices we use:
| End-use device | Typical efficiency (final to useful) |
|---|---|
| Petrol car engine, tank to wheels | 20–30% |
| Diesel truck engine | 35–45% |
| Electric car, battery to wheels | 85–90% |
| Condensing gas boiler | 85–95% |
| Heat pump | 250–450% (COP 2.5–4.5, since it moves heat from outside) |
| Incandescent bulb (light output) | about 2–5% |
| LED lamp | about 30–40% |
Put the two gaps together and most primary energy never becomes a useful service. Lawrence Livermore National Laboratory’s annual US energy flow chart typically shows roughly two-thirds of US primary energy ending up as “rejected energy”, mostly waste heat from power plants and vehicles.
Final energy by sector and carrier
According to IEA energy balances, world total final consumption runs at somewhat more than two-thirds of total energy supply. The rough split by sector looks like this:
| Sector | Approximate share of world final energy |
|---|---|
| Industry (energy use) | about 30% |
| Transport | about 25–30% |
| Residential buildings | about 20% |
| Commercial and public buildings | about 8% |
| Non-energy use (petrochemical feedstocks, bitumen, lubricants) | about 9–10% |
| Agriculture, fishing, other | about 2–3% |
Shares are indicative ranges from recent IEA balances; individual years move by a point or two.
By carrier, oil products remain the largest slice of final energy, at around 40%, because transport runs almost entirely on them. Electricity supplies about 20% of final consumption worldwide, natural gas around 15–16%, coal (mostly in industry) around 10% and biomass, district heat and other sources the rest. That 20% for electricity is the number to watch: every scenario that reaches net zero pushes it to 50% or more by mid-century.
Counting non-fossil electricity: three methods
Fossil fuels are easy to count as primary energy because they have a heat content. Wind, solar and hydro produce electricity directly, with no fuel, so statisticians must decide what to record. Three conventions exist:
- Physical energy content method (IEA, Eurostat). Wind, solar and hydro electricity count one-to-one: 1 kWh produced is 1 kWh primary. Nuclear and geothermal are counted by the heat produced in the reactor or reservoir, so nuclear electricity is multiplied by roughly 3 (assuming 33% efficiency).
- Direct equivalent method (UN statistics, IPCC reports). All non-combustible electricity counts one-to-one, nuclear included.
- Substitution method (used for years by the Energy Institute Statistical Review and by Our World in Data’s main charts). Non-fossil electricity is counted as the fossil fuel a thermal plant would have needed to produce it. The Energy Institute assumed a thermal efficiency close to 40% in recent years, so 1 kWh of wind becomes about 2.5 kWh of primary energy.
None of these is wrong. They answer different questions. The physical method follows the energy flow. The substitution method answers “how much fossil fuel does this source displace?”
Worked example: 1 TWh of electricity, three ways
Solar farm, 1 TWh: physical content / direct equivalent: 1 TWh = 3.6 PJ substitution at ~40%: 1 ÷ 0.40 = 2.5 TWh = 9.0 PJ
Nuclear plant, 1 TWh: physical content (33%): 1 ÷ 0.33 ≈ 3.0 TWh ≈ 10.9 PJ direct equivalent: 1 TWh = 3.6 PJ substitution at ~40%: 2.5 TWh = 9.0 PJ
The same terawatt-hour of solar power can look 2.5 times larger depending on the method. A country moving from coal to solar can see its primary energy fall sharply under the physical method even if electricity use stays flat. Most of that “drop” is coal plant waste heat that no longer exists.
Why the method matters in practice
Primary energy targets are written into law in some places. The EU’s energy efficiency directive sets caps on both primary and final energy consumption, and building regulations in many European countries use primary energy factors to rate homes. A factor of 2.5 for grid electricity, set when grids were coal-heavy, penalises electric heating compared with gas. Several countries have cut their factors (Germany uses 1.8, for example) as grids have decarbonised. If you compare a heat pump with a gas boiler, the heat pump vs gas boiler calculator works on final energy and cost, which is what you actually pay.
Headline shares change too. Under the substitution method, low-carbon sources supplied roughly 18% of world primary energy in 2023 according to the Energy Institute, and fossil fuels about 81.5%. Under the physical method, the low-carbon share looks a few points smaller because wind, solar and hydro shrink to their electricity output. Neither number tells you the share of electricity, which Ember puts at about 41% low-carbon in 2024.
Units you will meet
Primary energy appears in exajoules (EJ), million tonnes of oil equivalent (Mtoe), quadrillion BTU (quads) or terawatt-hours. One EJ is about 23.9 Mtoe or 278 TWh. The energy units cheat sheet lists the full set of factors, and the energy unit converter does the arithmetic. For a broader picture of how these totals have grown, see the history of energy.
At world scale, the Energy Institute put 2023 primary energy consumption at about 620 EJ, a record, and reported growth of around 2% in 2024. IEA total energy supply comes out somewhat higher partly because it includes traditional biomass, such as firewood and charcoal used for cooking, which the Energy Institute leaves out. Always check the method and coverage before comparing two charts.
Electrification and the shrinking primary total
Because fossil systems waste so much heat, electrifying with clean power cuts primary energy faster than it cuts services. Replace a petrol car (25% efficient) with an EV charged from wind (no combustion loss, ~85% efficient from grid to wheels after charging losses), and the primary energy for each kilometre drops by roughly 70–80% under the physical method. Swap a gas boiler for a heat pump with a COP of 3 and the final energy for heating falls by about two-thirds.
This is why serious scenarios show global energy demand falling even as the economy grows. The IEA’s net zero pathway, for instance, has total energy supply declining through the 2030s while services keep rising. Compare the drivetrain numbers yourself in the EV vs petrol car calculator, or check a country’s mix on the world energy map.
Quick reference
- Primary is energy as found in nature, final is what reaches the meter or pump, useful is the service delivered.
- Power plants and engines throw away most of the primary energy as heat; roughly two-thirds is lost in a fossil-heavy system like the US.
- Electricity is about 20% of world final energy but a much larger share of primary energy because of conversion losses.
- Wind and solar look up to 2.5 times bigger under the substitution method than under the physical content method. Check which one a chart uses.
- Electrification lowers primary energy demand because it removes combustion losses, so falling primary energy does not mean fewer services.
Calculators for this topic
Frequently asked questions
What is the difference between primary and final energy?
How much primary energy does the world use?
Why do wind and solar look small in primary energy charts?
What is useful energy?
Is final energy the same as energy consumption on my bill?
Why does electrification reduce total energy use?
Sources & references
- Energy Institute — Statistical Review of World Energy 2024 and 2025
- IEA — World Energy Balances: overview and methodology
- Our World in Data — Primary energy: the substitution method explained
- Lawrence Livermore National Laboratory — Energy flow charts
- Eurostat — Energy balance guide and primary energy factors
Figures in this guide are typical ranges drawn from the sources above; your own numbers depend on location, prices and equipment. Use the linked calculators for site-specific estimates.