Global Electricity Review 2024

Downloads

Report - Global Electricity Review 2024 - PDF (43 MB) Annex - Global Electricity Review 2024 - PDF (3 MB) GER2024 Data (Charts) - XLSX (4 MB) GER2024 Data (Monthly) - CSV (26 MB) GER2024 Data (Yearly) - CSV (20MB) Methodology - Ember Electricity Data - PDF (695 KB)

Methodology

Generation, imports and demand

Annual data from 2000 to 2022 is gross generation, taken primarily from the Energy Institute Statistical Review of World Energy, the Energy Information Administration (EIA), Eurostat and IRENA. 2023 data is an estimate of gross generation, based on generation gathered from monthly data. This estimate is calculated by applying absolute changes in monthly generation to the most recent annual baseline.

Net imports from 1990 to 2022 are taken from the EIA and Eurostat, with recent data estimated in the same manner as generation. Demand is calculated as the sum of generation and net imports, and where possible validated against published direct demand figures. Because it uses gross generation and does not include transmission and distribution losses, it will tend to be higher than end-user demand.

Monthly data is gathered for 80 countries from 70 sources, including national transmission system operators and statistical agencies, as well as data aggregators such as ENTSO-E. In some cases data is published on a monthly lag; here we have estimated recent months based on our generation model. These cases are flagged in the dataset.

Monthly published data is often reported provisionally, and is far from perfect. Every effort has been made to ensure accuracy, and where possible we compare multiple sources to confirm their agreement. A full detailed methodology can be found here.

Emissions

Ember’s calculations for emissions are continually improving, but may be conservative or otherwise uncertain in ways we describe below. These figures aim to include full lifecycle emissions including upstream methane, supply chain and manufacturing emissions, and include all gases, converted into CO2 equivalent over a 100-year timescale.

Emissions can vary over time as power station efficiency changes, and as different fuel qualities are used. Therefore, we report emissions values by fuel type, and emissions intensity by country. These values are calculated by multiplying our generation numbers by emissions factors taken from a number of sources, detailed below. We aim where possible to capture variance between geographies and over time in emissions intensity from different fuels. We have recently updated this approach and are actively working to improve it. If you have any comments or suggestions for improvement, please email [email protected].

Our sources and methodology for different fuels are described below. All factors we use are for net generation. Where we report gross generation, we adjust our factors by 6% for thermal fuel sources and 1% for others, following a standard conversion approach.

Coal

Data is taken from Gibon et al. 2022 (UNECE) and the Global Energy Monitor Coal Plant Tracker (GEM). UNECE provides lifecycle emissions factors for different fuel types for the year 2020 for each REMIND region. UNECE reports values for different technologies using bituminous coal. We derive factors for different coal grades based on IPCC 2005 direct combustion emissions factors. Using country-level annual technology and coal grade mixes from GEM capacity data, we estimate blended emissions factors per country per year for coal.

Gas

Country-level factors are taken from Jordaan et al. 2022, and are for generation for the year 2017. Two sets of factors are provided; we use the ones that attempt to account for combined heat and power. For smaller countries where no data is available, a world average number is used.

Nuclear and wind

We use region-level data from UNECE.

Bioenergy, hydro, solar, other renewables and other fossil fuels

We use data from the IPCC AR5 WG3 Annex III (2014). These are global estimates for the year 2020; we use midpoint lifecycle factors. These are:

Bioenergy: 230 g/kWh
Hydro: 24 g/kWh
Solar: 48 g/kWh
Other renewables: 38g/kWh
Other fossil: 700/kWh

Caveats

This approach attempts to account for some geographical and temporal variance in emissions factors. It is a work in progress, and figures may differ from reality for a number of reasons. Some of these are listed below:

Coal: UNECE base factors are for coal plants in the year 2020. They do not capture operational efficiency losses associated with older plants or intra-technology efficiency differences. Finally, we make assumptions to derive factors for coal grades other than lignite, including identical combustion efficiencies and upstream emissions per MWh generated.

Gas: Our gas factors are specific to the year 2017, so do not account for temporal variations in plant efficiencies or methane leakage rates. The methodology in Jordaan et al. 2022 also prefers to underestimate methane emissions where there is doubt. In general there is very significant uncertainty around methane emissions rates, even in countries that prioritise collecting this data. Some authors believe that emissions rates are significantly higher than assumed in our factors.

Time horizon: Upstream methane emissions for gas and coal generation are calculated on a long-term basis assuming methane is 21 times as potent as CO2. However, the short-term impact of methane is actually four times higher, at 86 times the potency of CO2. See this page for more information.

Solar and wind: Recent efficiency improvements have seen wind and solar emissions intensity drop, as energy output has increased relative to emissions from manufacturing. Our numbers may therefore be higher than reality. We also do not currently capture geographical variation in emissions intensity within REMIND regions; this can be significant, as countries with lower annual solar capacity factors will have proportionately higher lifecycle emissions.

Bioenergy: Our value is very likely to be a significant underestimate of the actual emissions caused by bioenergy generation. The emissions intensity of bioenergy is highly dependent on the feedstock, how it was sourced, and what would have happened had the feedstock not been burnt for energy. The IPCC figure we use is for dedicated energy crops and crop residues, rather than the much more commonly used woody or forest biomass, which has been shown to carry a greater risk of high-carbon outcomes. In certain cases, bioenergy can have a carbon intensity significantly greater than coal. Bioenergy is also frequently cofired with fossil fuels. We have disaggregated these wherever possible, but in certain cases recorded bioenergy generation may include some co-firing. In these circumstances, actual emissions will be higher than we estimate.

Hydro and other renewables: Hydropower emissions are generally very low, but can vary based on emissions during construction and biogenic emissions, and so in a small number of cases can be much higher than our value. Similarly, other renewable sources such as geothermal can in rare outlier cases have high emissions.

Gross and net generation: In the EU, we report net generation for monthly data and gross generation for yearly data. For gross generation, we perform the conversion described above, which may introduce some error.

Combined heat and power (CHP): In many cases, thermal power plants produce both heat and electricity. Our coal factors are based on only the electricity produced by such plants, ignoring heat. It may not therefore be fair for our dataset to include all emissions attributed to co-firing plants, which actually have greater efficiency than reported when considering total useful energy output. Our gas factors account for CHP.

2024 generation and demand forecast

The 2024 demand and generation forecast is based on the following assumptions:

Demand: 3.3% increase from 2023, based on IEA projections.
Solar: BNEF regional 2024 capacity addition projections (mid-case scenario). Regional average load factors are used based on 5-year average of weather and timing of additions. Prediction range is based on Monte Carlo analysis varying capacity deployment and panel performance.
Wind: Global capacity additions of 115 GW, from GWEC. Assumes 27% average global capacity factor.
Hydropower: Assumes capacity factor recovers halfway towards pre-2021 global average.
Nuclear: 2024 estimates calculated for the US based on EIA data, and for China based on capacity additions. Rest of world based on expected capacity additions.
Other renewables: Follows historic trend.

Missing solar disaggregation

Missing solar is disaggregated as follows:

National insolation and timing effects (i.e. the changes from a country having a less sunny year or unusually late installations) are estimated in individual countries where data is available, covering 75% of global capacity, and scaled up to a global figure. National insolation effects are recorded as “less sunny weather”.
Global insolation and timing effects (i.e. the changes caused by a global capacity shift towards countries which are less sunny or generally install later in the year) are calculated. Global insolation effects are recorded as “installations in less sunny countries”.
Global and national timing effects are combined as “late year additions”.
“Underreported solar generation” refers only to marginal underreporting – that is, it is an explanation of why 2023 departed from the historic trend rather than a comprehensive estimate of missing reporting. It is estimated from:
- Solar generation in countries where we know a portion is not reported (e.g. in Spain)
- Capacity installed in countries which do not report up-to-date generation (primarily in the Middle East and Asia)

Demand disaggregation

Electricity demand from EVs is estimated based on the number of vehicles sold, multiplied by the annual average electricity consumption for key vehicle types: passenger cars, light commercial vehicles, trucks, buses, and two-wheelers, including battery-electric and plug-in hybrid vehicles. Key sources on EV sales: China (passenger cars, light commercial vehicles, trucks, buses, two-wheelers), India, EU, US, Japan, rest of the world.
Electricity demand from heat pumps estimated based on the number of units sold, multiplied by annual average consumption. Key sources on heat pump sales: EHPA, AHRI, Rosenow et al, JRAIA, IEA.

Other notes

Solar insolation data is from the ERA5 reanalysis dataset, extracted from Teal.
Solar capacity installations are from BNEF. In all cases solar capacity is measured in DC rather than AC.

Acknowledgements

Authors and other contributors

Małgorzata Wiatros-Motyka, Nicolas Fulghum, Dave Jones, Katye Altieri, Richard Black, Hannah Broadbent, Chelsea Bruce-Lockhart, Matt Ewen, Phil MacDonald, Kostansa Rangelova, Sarah Brown, Libby Copsey, Reynaldo Dizon, Sam Hawkins, Leo Heberer, Sanghyun Hong, Rosamond Hutt, Uni Lee, Aditya Lolla, Josie Murdoch, James Robinson, Neshwin Rodrigues, Chris Rosslowe, Oya Zaimoglu.

Peer reviewers

Harry Benham (Ember), Kingsmill Bond (RMI), Nathaniel Bullard (Business Climate Pte. Ltd), Duncan Burt (Reactive Technologies Limited), Kanika Chawla (SEforALL), Toby Lockwood (Clean Air Task Force), Hannah Ritchie (Our World in Data), Oliver Then (vgbe energy e.V.), Bryony Worthington (Ember).