Solar photovoltaic (PV) is increasingly competitive with other forms of electricity generation, and is the low-cost option in many applications ( high confidence). Costs have declined by 62% since 2015 ( high confidence) and are anticipated to decline by an additional 16% by 2030 if current trends continue (low confidence, medium evidence). Key areas for continued improvement are grid integration and non-module costs for rooftop systems ( high confidence). Most deployment is now utility-scale ( high confidence). Global future potential is not limited by solar irradiation, but by grid integration needed to address its variability, as well as access to finance, particularly in developing countries ( high confidence).

The global technical potential of direct solar energy far exceeds that of any other renewable energy resource and is well beyond the total amount of energy needed to support ambitious mitigation over the current century ( high confidence). Estimates of global solar resources have not changed since the IPCC’s Fifth Assessment Report (AR5) (Lewis 2007; Besharat et al. 2013) even as precision and near-term forecasting have improved (Diagne et al. 2013; Abreu et al. 2018). Approximately 120,000TW of sunlight reaches the Earth’s surface continuously, almost 10,000 times average world energy consumption; factoring in competition for land use leaves a technical potential of about 300 PWh yr –1 (1080 EJ yr –1) for solar PV, roughly double current consumption (Dupont et al. 2020). The technical potential for concentrating solar power (CSP) is estimated to be 45–82 PWh yr –1 (162–295 EJ yr –1) (Dupont et al. 2020). Areas with the highest solar irradiation are: western South America; northern, eastern and southwestern Africa; and the Middle East and Australia (Figure 6.7) (Prăvălie et al. 2019).

In many parts of the world, the cost of electricity from PV is below the cost of electricity generated from fossil fuels; in some, it is below the operating costs of electricity generated from fossil fuels ( high confidence). The weighted average cost of PV in 2019 was USD68 MWh –1, near the bottom of the range of fossil fuel prices (IRENA 2019b). The cost of electricity from PV has fallen by 89% since 2000 and 69% since AR5, at a rate of –16% per year. The 5:95 percentile range for PV in 2019 was USD52–190 MWh –1 (IRENA 2021b). Differences in solar insolation, financing costs, equipment acquisition, installation labour, and other sources of price dispersion explain this range (Nemet et al. 2016; Vartiainen et al. 2020) and scale. For example, in India, rooftop installations cost 41% more than utility-scale installations, and commercial-scale costs are 39% higher than utility-scale. Significant differences in regional cost persist (Kazhamiaka et al. 2017; Vartiainen et al. 2020), with particularly low prices in China, India, and parts of Europe. Globally, the range of global PV costs is quite similar to the range of coal and natural gas prices.

PV costs (Figure 6.8) have fallen for various reasons: lower silicon costs, automation, lower margins, automation, higher efficiency, and a variety of incremental improvements (Fu et al. 2018; Green 2019) (Chapter 16). Increasingly, the costs of PV electricity are concentrated in the installation and related ‘soft costs’ (marketing, permitting) associated with the technology rather than in the modules themselves, which now account for only 30% of installed costs of rooftop systems (O’Shaughnessy et al. 2019; IRENA 2021b). Financing costs are a significant barrier in developing countries (Ondraczek et al. 2015) and growth there depends on access to low-cost finance (Creutzig et al. 2017).

CSP costs have also fallen, albeit at about half the rate of PV: –9% yr –1 since AR5. The lowest prices for CSP are now competitive with more expensive fossil fuels, although the average CSP cost is above the range for fossil-based power generation. Other data sources put recent CSP costs at USD120 MWh –1, in the middle of the fossil range (Lilliestam et al. 2020). Continuing the pace of change since AR5 will make CSP competitive with fossil fuels in sunny locations, although it will be difficult for CSP to compete with PV and even hybrid PV-battery systems. CSP electricity can be more valuable, however, because CSP systems can store heat longer than PV battery systems.

The share of total costs of PV-intensive electricity systems attributed to integration costs has been increasing but can be reduced by enhancing grid flexibility ( high confidence) (Sections 6.4.3 and 6.6, and Box 6.8). The total costs of PV include grid integration, which varies tremendously depending on PV’s share of electricity, other supply sources like wind, availability of storage, transmission capacity, and demand flexibility (Heptonstall and Gross 2020). Transmission costs can add USD1–10 MWh –1 or 3–33% to the cost of utility-scale PV (Gorman et al. 2019). Distributed (rooftop) PV involves a broader set of grid integration costs – including grid reinforcement, voltage balancing and control, and impacts on other generations – and has a larger range of integration costs from USD2–25 MWh –1, which is –3% to +37% (Hirth et al. 2015; Wu et al. 2015; Gorman et al. 2019). Other meta-analyses put the range at USD1–7 MWh –1 in the USA (Luckow et al. 2015.; Wiser et al. 2017), while a comprehensive study put the range at USD12–18 MWh –1 for up to 35% renewables and USD25–46 MWh –1 above 35% renewables (Heptonstall and Gross 2020). Increased system flexibility can reduce integration costs of solar energy (Wu et al. 2015) including storage, demand response, sector-coupling (Brown et al. 2018; Bogdanov et al. 2019), and increase complementarity between wind and solar (Heide et al. 2010) (Sections 6.4.3 and 6.4.4).

Since solar PV panels have very low operating costs, they can, at high penetrations and in the absence of adequate incentives to shift demand, depress prices in wholesale electricity markets, making it difficult to recoup investment, and potentially reducing incentives for new installations (Hirth 2013; Millstein et al. 2021). Continued cost reductions help address this issue of value deflation, but only partially. Comprehensive solutions depend on adding transmission and storage (Das et al. 2020) and, more fundamentally, adjustments to electricity market design (Roques and Finon 2017; Bistline and Young 2019).

The most important ways to minimise PV’s impact on the environment lie in recycling materials at end of life and making smart land-use decisions (medium confidence). A comprehensive assessment of PV’s environmental impacts requires lifecycle analysis (LCA) of resource depletion, land-use, ecotoxicity, eutrophication, acidification, ozone, and particulates, among other things (Mahmud et al. 2018). LCA studies show that solar PVs produce far less CO2 per unit of electricity than fossil generation, but PV CO2 emissions vary due to the carbon intensity of manufacturing energy and offset electricity (Grant and Hicks 2020). Concerns about systemic impacts, such as reducing the Earth’s albedo by covering surfaces with dark panels, have shown to be trivial compared to the mitigation benefits (Nemet 2009) (Box 6.7). Even though GHG LCA estimates span a considerable range of 9–250 gCO2 kWh –1 (de Wild-Scholten 2013; Kommalapati et al. 2017), recent studies that reflect higher efficiencies and manufacturing improvements find lower lifecycle emissions, including a range of 18–60 gCO2 kWh –1 (Wetzel and Borchers 2015) and central estimates of 80 gCO2 kWh –1 (Hou et al. 2016), 50 gCO2 kWh –1 (Nugent and Sovacool 2014), and 20 gCO2 kWh –1 (Louwen et al. 2016). These recent values are an order of magnitude lower than coal, and natural gas and further decarbonisation of the energy system will make them lower still. Thin films and organics produce half the lifecycle emissions of silicon wafer PV, mainly because they use less material (Lizin et al. 2013; Hou et al. 2016). Novel materials promise even lower environmental impacts, especially with improvements to their performance ratios and reliability (Gong et al. 2015; Muteri et al. 2020). Higher efficiencies, longer lifetimes, sunny locations, less carbon-intensive manufacturing inputs, and shifting to thin films could reduce future lifecycle impacts.

Another environmental concern with large PV power plants is the conversion of land to collect solar energy (Hernandez et al. 2015). Approximately 2 hectares of land are needed for 1 MW of solar electricity capacity (Perpiña Castillo et al. 2016; Kabir et al. 2018); at 20% efficiency, a square of PV panels of 550 km by 550 km, comprising 0.2% of Earth’s land area, could meet global energy demand. Land conversion can have local impacts, especially near cities and where land used for solar competes with alternative uses, such as agriculture. Large installations can also adversely impact biodiversity (Hernandez et al. 2014), especially where the above-ground vegetation is cleared and soils are typically graded. Landscape fragmentation creates barriers to the movement of species. However, a variety of means have emerged to mitigate land use issues. Substitution among renewables can reduce land conversion (Tröndle 2020). Solar can be integrated with other uses through ‘agrivoltaics’ (the use of land for both agriculture and solar production) (Dupraz et al. 2011) by, for example, using shade-tolerant crops (Dinesh and Pearce 2016). Combining solar and agriculture can also create income diversification, reduced drought stress, higher solar output due to radiative cooling, and other benefits (Elamri et al. 2018; Hassanpour Adeh et al. 2018; Barron-Gafford et al. 2019). PV installations floating on water also avoid land-use conflicts (Sahu et al. 2016; Lee et al. 2020), as does dual-use infrastructure, such as landfills (Jäger-Waldau 2020) and reservoirs where evaporation can also be reduced (Farfan and Breyer 2018).

Material demand for PV will likely increase substantially to limit warming to well below 2°C, but PV materials are widely available, have possible substitutes, and can be recycled (medium confidence) (Box 6.4). The primary materials for PV are silicon, copper, glass, aluminium, and silver – the costliest being silicon, and glass being the most essential by mass, at 70%. None of these materials is considered to be either critical or potentially scarce (IEA 2020e). Thin-film cells, such as amorphous silicon, cadmium telluride and copper indium gallium diselenide (CIGS), use far less material (though they use more glass), but account for less than 10% of the global solar market. Other thin-films, such as those based on perovskites, organic solar cells, or earth-abundant, non-toxic materials such as kesterites, either on their own, or layered on silicon, could further reduce material use per energy produced (Box 6.4).

After a typical lifetime of 30 years of use, PV modules can be recycled to prevent environmental contamination from the toxic materials within them, reusing valuable materials and avoiding waste accumulation. Recycling allows the reuse of nearly all – 83% in one study – of the components of PV modules, other than plastics (Ardente et al. 2019) and would add less than 1% to lifecycle GHG emissions (Latunussa et al. 2016). Glass accounts for 70% of the mass of a solar cell and is relatively easy to recycle. Recycling technology is advancing, but the scale and share of recycling is still small (Li et al. 2020d). By 2050, however, end-of-life PV could total 80 MT and comprise 10% of global electronic waste (Stolz and Frischknecht 2017), although most of it is glass. IEA runs a programme to enable PV recycling by sharing best practices to minimise recycling lifecycle impacts. Ensuring that a substantial amount of panels are recycled at end of life will likely require policy incentives, as the market value of the recovered materials, aside from aluminium and copper, is likely to be too low to justify recycling on its own (Deng et al. 2019). A near-term priority is maximising the recovery of silver, silicon, and aluminium, the most valuable PV material components (Heath et al. 2020).

Many alternative PV materials are improving in efficiency and stability, providing longer-term pathways for continued PV costs reductions and better performance ( high confidence). While solar PV based on semi-conductors constructed from wafers of silicon still captures 90% of the market, new designs and materials have the potential to reduce costs further, increase efficiency, reduce resource use, and open new applications. The most significant technological advance within silicon PV in the past 10 years has been the widespread adoption of the passivated emitter and rear cell (PERC) design (Green 2015), which now accounts for the majority of production. This advance boosts efficiency over traditional aluminium backing by increasing reflectivity within the cell and reducing electron-hole recombination (Blakers 2019). Bifacial modules increase efficiency by using reflected light from the ground or roof on the backside of modules (Guerrero-Lemus et al. 2016). Integrating PV into buildings can reduce overall costs and improve building energy performance (Shukla et al. 2016). Concentrating PV uses lenses or mirrors that collect and concentrate light onto high efficiency PV cells (Li et al. 2020a). Beyond crystalline silicon, thin films of amorphous silicon, cadmium telluride, and copper indium gallium selenide (among others) have the potential for much lower costs while their efficiencies have increased (Green et al. 2019). Perovskites, inexpensive and easy to produce crystalline structures, have increased in efficiency by a factor of six in the past decade; the biggest challenge is light-induced degradation as well as finding lead-free efficient compounds, or establishing lead recycling at the end of the lifecycle of the device (Petrus et al. 2017; Chang et al. 2018; Wang et al. 2019b; Zhu et al. 2020). Organic solar cells are made of carbon-based semiconductors like the ones found in the displays made from organic light emitting diodes (OLEDs) and can be processed in thin films on large areas with scalable and fast coating processes on plastic substrates. The main challenges are raising the efficiency and improving their lifetime (Ma et al. 2020; Riede et al. 2021). Quantum dots, spherical semi-conductor nanocrystals, can be tuned to absorb specific wavelengths of sunlight, giving them the potential for high efficiency with very little material use (Kramer et al. 2015). A common challenge for all emerging solar cell technologies is developing the corresponding production equipment. Hybrids of silicon with layers of quantum dots and perovskites have the potential to take advantage of the benefits of all three, although those designs require that these new technologies have stability and scale that match those of silicon (Chang et al. 2017; Palmstrom et al. 2019). This broad array of alternatives to making PV from crystalline silicon offer realistic potential for lower costs, reduced material use, and higher efficiencies in future years (Victoria et al. 2021).

Besides PV, alternative solar technologies exist, including CSP, which can provide special services in high-temperature heat and diurnal storage, even if it is more costly than PV and its potential for deployment is limited. CSP uses reflective surfaces, such as parabolic mirrors, to focus sunlight on a receiver to heat a working fluid, which is subsequently transformed into electricity (Islam et al. 2018). Solar heating and cooling are also well established technologies, and solar energy can be utilised directly for domestic or commercial applications such as drying, heating, cooling, and cooking (Ge et al. 2018). Solar chimneys, (still purely conceptual), heat air using large transparent greenhouse-like structures and channel the warm air to turbines in tall chimneys (Kasaeian et al. 2017). Solar energy can also be used to produce solar fuels, for example, hydrogen or synthetic gas (syngas) (Montoya et al. 2016; Nocera 2017; Detz et al. 2018). In addition, research proceeds on space-based solar PV, which takes advantage of high insolation and a continuous solar resource (Kelzenberg et al. 2018), but faces the formidable obstacle of developing safe, efficient, and inexpensive microwave or laser transmission to the Earth’s surface (Yang et al. 2016). CSP is the most widely adopted of these alternative solar technologies.

Like PV, CSP facilities can deliver large amounts of power (up to 200 MW per unit) and maintain substantial thermal storage, which is valuable for load balancing over the diurnal cycle (McPherson et al. 2020). However, unlike PV, CSP can only use direct sunlight, constraining its cost-effectiveness to North Africa, the Middle East, Southern Africa, Australia, the Western USA, parts of South America (Peru, Chile), and the Western part of China (Deng et al. 2015; Dupont et al. 2020). Parabolic troughs, central towers and parabolic dishes are the three leading solar thermal technologies (Wang et al. 2017d). Parabolic troughs represented approximately 70% of new capacity in 2018 with the balance made up by central tower plants (Islam et al. 2018). Especially promising research directions are on tower-based designs that can achieve high temperatures, useful for industrial heat and energy storage (Mehos et al. 2017), and direct steam generation designs (Islam et al. 2018). Costs of CSP have fallen by nearly half since AR5 (Figure 6.8) albeit at a slower rate than PV. Since AR5, almost all new CSP plants have storage (Figure 6.9) (Thonig 2020).

Solar energy elicits favourable public responses in most countries ( high confidence) (Mcgowan and Sauter 2005; Ma et al. 2015; Hanger et al. 2016; Bessette and Arvai 2018; Jobin and Siegrist 2018; Roddis et al. 2019; Hazboun and Boudet 2020). Solar energy is perceived as clean and environmentally friendly with few downsides (Faiers and Neame 2006; Whitmarsh et al. 2011b). Key motivations for homeowners to adopt PV systems are expected financial gains, environmental benefits, the desire to become more self-sufficient, and peer expectations (Korcaj et al. 2015; Vasseur and Kemp 2015; Palm 2017). Hence, the observability of PV systems can facilitate adoption (Boudet 2019). The main barriers to the adoption of solar PV by households are its high upfront costs, aesthetics, landlord-tenant incentives, and concerns about performance and reliability (Faiers and Neame 2006; Whitmarsh et al. 2011b; Vasseur and Kemp 2015).

Wind power is increasingly competitive with other forms of electricity generation and is the low-cost option in many applications ( high confidence). Costs have declined by 18% and 40% on land and offshore since 2015 ( high confidence), and further reductions can be expected by 2030 (medium confidence). Critical areas for continued improvement are technology advancements and economies of scale ( high confidence). Global future potential is primarily limited by onshore land availability in wind power-rich areas, lack of supporting infrastructure, grid integration, and access to finance (especially in developing countries) ( high confidence).

Energy from wind is abundant, and the estimated technical potentials surpass the total amount of energy needed to limit warming to well below 2°C ( high confidence). Recent global estimates of potentially exploitable wind energy resource are in the range of 557–717 PWh yr –1 (2005–2580 EJ yr –1) (Eurek et al. 2017; Bosch et al. 2017, 2018; McKenna et al. 2022), or 20–30 times the 2017 global electricity demand. Studies have suggested that ‘bottom-up’ approaches may overestimate technical potentials (Miller et al. 2015; Kleidon and Miller 2020). But even in the most conservative ‘top-down’ approaches, the technical wind potential surpasses the amount needed to limit warming to well below 2°C (Bosch et al. 2017; Eurek et al. 2017; Volker et al. 2017). The projected climate change mitigation from wind energy by 2100 ranges from 0.3°C–0.8°C depending on the precise socio-economic pathway and wind energy expansion scenario followed (Barthelmie and Pryor 2021). Wind resources are unevenly distributed over the globe and by time of the year (Petersen and Troen 2012), but potential hotspots exist on every continent (Figure 6.10) as expressed by the wind power density (a quantitative measure of wind energy available at any location). Technical potentials for onshore wind power vary considerably, often because of inconsistent assessments of suitability factors (McKenna et al. 2020). The potential for offshore wind power is larger than for onshore because offshore wind is stronger and less variable (Bosch et al. 2018). Offshore wind is more expensive, however, because of higher costs for construction, maintenance, and transmission. Wind power varies at a range of time scales, from annual to sub-seconds; the effects of local short-term variability can be offset by power plant control, flexible grid integration, and storage (Barra et al. 2021) (Section 6.4.3). In some regions, interannual variations in wind energy resources could be important for optimal power system design (Wohland et al. 2019a; Coker et al. 2020).

Wind power cost reductions (Figure 6.11) are driven mainly by larger capacity turbines, larger rotor diameters and taller hub heights – larger swept areas increase the energy captured and the capacity factors for a given wind speed; taller towers provide access to higher wind speeds (Beiter et al. 2021). All major onshore wind markets have experienced rapid growth in both rotor diameter (from 81.2 m in 2010 to 120 m in 2020) (IRENA 2021b), and average power ratings (from 1.9 MW in 2010 to 3 MW in 2020). The generation capacity of offshore wind turbines grew by a factor of 3.7 in less than two decades, from 1.6 MW in 2000 to 6 MW in 2020 (Wiser et al. 2021). Floating foundations could revolutionise offshore wind power by tapping into the abundant wind potential in deeper waters. This technology is particularly important for regions where coastal waters are too deep for fixed-bottom wind turbines. Floating wind farms potentially offer economic and environmental benefits compared with fixed-bottom designs due to less-invasive activity on the seabed during installation, but the long-term ecological effects are unknown and meteorological conditions further offshore and in deeper waters are harsher on wind turbine components (IRENA 2019c). A radical new class of wind energy converters has also been conceived under the name of airborne wind energy systems that can harvest strong, high-altitude winds (typically between 200–800m), which are inaccessible by traditional wind turbines (Cherubini et al. 2015). This technology has seen development and testing of small devices (Watson et al. 2019).

Wind capacity factors have increased over the last decade (Figure 6.11). The capacity factor for onshore wind farms increased from 27% in 2010 to 36% in 2020 (IRENA 2021a). The global average offshore capacity factor has decreased from a peak of 45% in 2017. This has been driven by the increased share of offshore development in China, where projects are often near-shore and use smaller wind turbines than in Europe (IRENA 2021b). Improvements in capacity factors also come from increased functionality of wind turbines and wind farms. Manufactures can adapt the wind turbine generator to the wind conditions. Turbines for windy sites have smaller generators and smaller specific capacity per rotor area, and therefore operate more efficiently and reach full capacity for a longer time period (Rohrig et al. 2019).

Electricity from onshore wind is less expensive than electricity generated from fossil fuels in a growing number of markets ( high confidence). The global average LCOE onshore declined by 38% from 2010 to 2020 (Figure 6.11), reaching USD0.039 kWh –1. However, the decrease in cost varies substantially by region. Since 2014, wind costs have declined more rapidly than the majority of experts predicted (Wiser et al. 2021). New modelling projects onshore wind LCOE of USD.037 kWh –1 by 2030 (Junginger et al. 2020a), and additional reductions of 37–39% have been predicted by 2050 (Wiser et al. 2021). The future cost of offshore wind is more uncertain because other aspects besides increases in capacity factors influence the cost (Junginger et al. 2020b).

The cost of the turbine (including the towers) makes up the largest component of wind LCOE. Total installed costs for both onshore and offshore wind farms have decreased since 2015 (Figure 6.11), but the total installed costs for onshore wind projects are very site- and market-specific, as reflected in the range of LCOEs. China, India, and the USA have experienced the largest declines in total installed costs. In 2020, typical country-average total installed costs were around USD1150 kW–1 in China and India, and between USD1403–2472 kW–1 elsewhere (IRENA 2021b). Total installed costs of offshore wind farms declined by 12% between 2010 and 2020. But, because some of the new offshore wind projects have moved to deeper waters and further offshore, there are considerable year-to-year variations in their price (IRENA 2021b). Projects outside China in recent years have typically been built in deeper waters (10–55 m) and up to 120 km offshore, compared to around 10 m in 2001–2006, when distances rarely exceeded 20 km. With the shift to deeper waters and sites further from ports, the total installed costs of offshore wind farms rose, from an average of around USD2500 kW–1 in 2000 to around USD5127 kW–1 by 2011–2014, before falling to around USD3185 kW–1 in 2020 (IRENA 2020a). The full cost of wind power includes the transmission and system integration costs (Sections 6.4.3 and 6.4.6). A new technology in development is the co-location of wind and solar PV power farms, also known as hybrid power plants. Co-locating wind, solar PV, and batteries can lead to synergies in electricity generation, infrastructure, and land usage, which may lower the overall plant cost compared to single technology systems (Lindberg et al. 2021).

Wind power plants pose relatively low environmental impact, but sometimes locally significant ecological effects ( high confidence). The environmental impact of wind technologies, including CO2 emissions, is concentrated in the manufacturing, transport, and building stage and in disposal as the end-of-life of wind turbines is reached (Liu and Barlow 2017; Mishnaevsky 2021). The operation of wind turbines produces no waste or pollutants. The LCA for wind turbines is strongly influenced by the operating lifetime, quality of wind resources, conversion efficiency, and size of the wind turbines (Kaldellis and Apostolou 2017; Laurent et al. 2018). All wind power technologies repay their carbon footprint in less than a year (Bonou et al. 2016).

Wind farms can cause local ecological impacts, including on animal habitat and movements, biological concerns, bird and bat fatalities from collisions with rotating blades, and health concerns (Morrison and Sinclair 2004). The impacts on animal habitats and collisions can be resolved or reduced by selectively stopping some wind turbines in high-risk locations, often without affecting the productivity of the wind farm (de Lucas et al. 2012). Many countries now require environmental studies of impacts of wind turbines on wildlife prior to project development, and, in some regions, shutdowns are required during active bird migration (de Lucas et al. 2012). Offshore wind farms can also impact migratory birds and other sea species (Hooper et al. 2017). Floating foundations pose lower environmental impacts at build stage (IRENA 2019c), but their cumulative long-term impacts are unclear (Goodale and Milman 2016). Recent studies find weak associations between wind farm noise and measures of long-term human health (Poulsen et al. 2018a, b, 2019a, b).

Public support for onshore and particularly offshore wind energy is generally high, although people may oppose specific wind farm projects ( high confidence) (e.g., Bell et al. 2005; Batel and Devine-Wright 2015; Rand and Hoen 2017; Steg 2018). People generally believe that wind energy is associated with environmental benefits and that it is relatively cheap. Yet, some people believe wind turbines can cause noise and visual aesthetic pollution, threaten places of symbolic value (Devine-Wright and Wiersma 2020; Russell et al. 2020), and have adverse effects on wildlife (Bates and Firestone 2015), which challenges public acceptability (Rand and Hoen 2017). Support for local wind projects is higher when people believe fair decision-making procedures have been implemented (Dietz and Stern 2008; Aitken 2010a). Evidence is mixed whether distance from wind turbines or financial compensation increases public acceptability of wind turbines (Cass et al. 2010; Rand and Hoen 2017; Rudolph et al. 2018; Hoen et al. 2019). Offshore wind farms projects have higher public support, but can also face resistance (Bidwell 2017; Rudolph et al. 2018).

Common economic barriers to wind development are high initial cost of capital, long payback periods, and inadequate access to capital. Optimal wind energy expansion is most likely to occur in the presence of a political commitment to establish, maintain, and improve financial support instruments, technological efforts to support a local supply chains, and grid investments integrate VRE electricity (Diógenes et al. 2020).

Integrated whole-system approaches can reduce the costs of low-carbon energy system transitions ( high confidence). A lack of flexibility in the electricity system may limit the cost-effective integration of technologies as part of broader net-zero energy systems. At the same time, the enormous latent flexibility hidden in heating and cooling, hydrogen, transport, gas systems, and other energy systems provides opportunities to take advantage of synergies and to coordinate operations across systems (Martin et al. 2017; Zhang et al. 2018; Martinez Cesena and Mancarella 2019; Pavičević et al. 2020; Bogdanov et al. 2021) (Figure 6.16).

Sector coupling can significantly increase system flexibility, driven by the application of advanced technologies (Clegg and Mancarella 2016; Heinen et al. 2016; Bogdanov et al. 2019; Solomon et al. 2019; Zhang et al. 2019b; Zhang and Fujimori 2020; Zhao et al. 2021). For example, district heating infrastructure can generate both heat and power. Cooling systems and electrified heating systems in buildings can provide flexibility through preheating and precooling via thermal energy storage (Z. Li et al. 2016; G. Li et al. 2017). System balancing services can be provided by electric vehicles (EVs) based on vehicle-to-grid concepts and deferred charging through smart control of EV batteries without compromising customers’ requirements for transport (Aunedi and Strbac 2020).

Hydrogen production processes (power-to-gas and vice versa) and hydrogen storage can support short-term and long-term balancing in the energy systems and enhance resilience (Stephen and Pierluigi 2016; Strbac et al. 2020). However, the economic benefits of flexible power-to-gas plants, energy storage, and other flexibility technological and options will depend on the locations of VRE sources, storage sites, gas, hydrogen, and electricity networks (Jentsch et al. 2014; Heymann and Bessa 2015; Ameli et al. 2020). Coordinated operation of gas and electricity systems can bring significant benefits in supplying heat demands. For example, hybrid heating can eliminate investment in electricity infrastructure reinforcement by switching to heat pumps in off-peak hours and gas boilers in peak hours (Fischer et al. 2017; Dengiz et al. 2019; Bistline et al. 2021). The heat required by direct air carbon capture and storage (DACCS) could be effectively supplied by inherent heat energy in nuclear plants, enhancing overall system efficiency (Realmonte et al. 2019).

Rather than incremental planning, strategic energy system planning can help minimise long-term mitigation costs ( high confidence). With a whole-system perspective, integrated planning can consider both short-term operation and long-term investment decisions, covering infrastructure from local to national and international, while meeting security of supply requirements and incorporating the flexibility provided by different technologies and advanced control strategies (Zhang et al. 2018; O’Malley et al. 2020; Strbac et al. 2020). Management of conflicts and synergies between local district and national level energy system objectives, including strategic investment in local hydrogen and heat infrastructure, can drive significant whole-system cost savings (Zhang et al. 2019b; Fu et al. 2020). For example, long-term planning of the offshore grid infrastructure to support offshore wind development, including interconnection between different countries and regions, can provide significant savings compared to a short-term incremental approach in which every offshore wind farm is individually connected to the onshore grid (E3G 2021).

Flexibility technologies – including energy storage, demand-side response, flexible/dispatchable generation, grid-forming converters, and transmission interconnection – as well as advanced control systems – can facilitate cost-effective and secure low-carbon energy systems ( high confidence). Flexibility technologies have already been implemented, but they can be enhanced and deployed more widely. Due to their interdependencies and similarities, there can be both synergies and conflicts for utilising these flexibility options (Bistline et al. 2021). It will therefore be important to coordinate the deployment of the potential flexibility technologies and smart control strategies. Important electricity system flexibility options include the following:

•Flexible/dispatchable generation. Advances in generation technologies, for example, gas/hydrogen plants and nuclear plants, can enable them to provide flexibility services. These technologies would start more quickly, operate at lower power output, and make faster output changes, enabling more secure and cost-effective integration of VRE generation and end-use electrification. There are already important developments in increasing nuclear plants flexibility (e.g., in France (Office of Nuclear Energy 2021)) and the development of small modular reactors, which could support system balancing (FTI Consulting 2018).

•Grid-forming converters (inverters). The transition from conventional electricity generation, applying mainly synchronous machines to inverter-dominated renewable generation, creates significant operating challenges. These challenges are mainly associated with reduced synchronous inertia, system stability, and ‘black start’ capability. Grid-forming converters will be a cornerstone for the control of future electricity systems dominated by VRE generation. These converters will address critical stability challenges, including the lack of system inertia, frequency and voltage regulation, and black start services while reducing or eliminating the need to operate conventional generation (Tayyebi et al. 2019).

•Interconnection. Electricity interconnections between different regions can facilitate more cost-effective renewable electricity deployment. Interconnection can enable large-scale sharing of energy and provide balancing services. Backup energy carriers beyond electricity, such as ammonia, can be shared through gas/ammonia/hydrogen-based interconnections, strengthening temporal coupling of multiple sectors in different regions (Bhagwat et al. 2017; Brown et al. 2018) (Section 6.4.5).

•Demand-side response. Demand-side schemes – including, for example, smart appliances, EVs, and building-based thermal energy storage (Heleno et al. 2014) – can provide flexibility services across multiple time frames and systems. Through differentiation between essential and non-essential needs during emergency conditions, smart control of demands can significantly enhance system resilience (Chaffey 2016).

•Energy storage. Energy storage technologies (Section 6.4.4) can act as both demand and generation sources. They can provide services such as system balancing, various ancillary services, and network management. Long-duration energy storage can significantly enhance the utilisation of renewable energy sources and reduce the need for firm low-carbon generation (Sepulveda et al. 2021).

A digitalised energy system can significantly reduce energy infrastructure investments while enhancing supply security and resilience ( high confidence) (Andoni et al. 2019; Strbac et al. 2020). Significant progress has been made in the development of technologies essential for the transition to a digitalised energy control paradigm, although the full implementation is still under development. Electrification and the increased integration of the electricity system with other systems will fundamentally transform the operational and planning paradigm of future energy infrastructure. A fully intelligent and sophisticated coordination of the multiple systems through smart control will support this paradigm shift. This shift will provide significant savings through better utilisation of existing infrastructure locally, regionally, nationally, and internationally. Supply system reliability will be enhanced through advanced control of local infrastructure (Strbac et al. 2015a). Furthermore, this paradigm shift offers the potential to increase energy efficiency through a combination of technologies that gather and analyse data and consequently optimise energy use in real-time.

The transition to advanced data-driven control of energy system operations (Cremer et al. 2019; Sun et al. 2019a) will require advanced information and communication technologies and infrastructure, including the internet, wireless networks, computers, software, middleware, smart sensors, internet of things components, and dedicated technological developments (Hossein Motlagh et al. 2020). The transition will raise standardisation and cyber-security issues, given that digitalisation can become a single point of failure for the complete system (Ustun and Hussain 2019; Unsal et al. 2021). Implementing peer-to-peer energy trading based on blockchain is expected to be one of the key elements of next-generation electricity systems (Qiu et al. 2021). This trading will enable consumers to drive system operation and future design, increasing overall system efficiency and security of supply while reducing emissions without sacrificing users’ privacy (Andoni et al. 2019; Ahl et al. 2020). When deployed with smart contracts, this concept will be suitable for energy systems involving many participants, where a prerequisite is digitalisation (e.g., smart meters, end-use demand control systems) (Juhar and Khaled 2018; Teufel et al. 2019).

New sources of flexibility and advanced control systems provide a significant opportunity to reduce low-carbon energy system costs by enhancing operating efficiency and reducing energy infrastructure and low-carbon generation investments, while continuing to meet security requirements ( high confidence). In the USA, for example, one study found that flexibility in buildings alone could reduce US CO2 emissions by 80 Mt yr –1 and save USD18 billion yr –1 in electricity system costs by 2030 (Satchwell et al. 2021). Key means for creating savings are associated with the following:

•Efficient energy system operation. Flexibility technologies such as storage, demand-side response, interconnection, and cross-system control will enable more efficient, real-time demand and supply balancing. This balancing has historically been provided by conventional fossil-fuel generation (Nuytten et al. 2013).

•Savings in investment in low-carbon/renewable generation capacity. System flexibility sources can absorb or export surplus electricity, thus reducing or avoiding energy curtailment and reducing the need for firm low-carbon capacity such as nuclear and fossil-fuel plants with CCS (Newbery et al. 2013; Solomon et al. 2019). For example, one study found that flexibility technologies and advanced control systems could reduce the need for nuclear power by 14 GW and offshore wind by 20 GW in the UK’s low-carbon transition (Strbac et al. 2015b).

•Reduced need for backup capacity. System flexibility can reduce energy demand peaks, reducing the required generation capacity to maintain the security of supply, producing significant savings in generation investments (Strbac et al. 2020).

•Deferral or avoidance of electricity network reinforcement/addition. Flexibility technologies supported by advanced control systems can provide significant savings in investment in electricity network reinforcement that might emerge from increased demand, for example, driven by electrification of transport and heat sectors. Historical network planning and operation standards are being revised considering alternative flexibility technologies, which would further support cost-effective integration of decarbonised transport and heat sectors (Strbac et al. 2020).

Pumped hydroelectric storage (PHS). PHS makes use of gravitational potential energy, using water as the medium. Water is pumped into an elevated reservoir using off-peak electricity and stored for later release when electricity is needed. These closed-loop hydropower plants have been in use for decades and account for 97% of worldwide electricity storage capacity (IRENA 2017b; IEA 2018b). PHS is best suited to balancing daily energy needs at a large scale, and advances in the technology now allow rapid response and power regulation in both generating and pumping mode (Valavi and Nysveen 2018; Dong et al. 2019; Kougias et al. 2019). The construction itself can cause disruption to the local community and environment (Hayes et al. 2019), the initial investment is costly, and extended construction periods delay return on investment (Section 6.4.2.3). In addition, locations for large-scale PHS plants are limited.

Advanced pump-turbines are being developed, allowing both reversible and variable-speed operation, supporting frequency control and grid stability with improved round-trip efficiencies (Ardizzon et al. 2014). New possibilities are being explored for small-scale PHS installations and expanding the potential for siting (Kougias et al. 2019). For example, in underwater PHS, the upper reservoir is the sea, and the lower is a hollow deposit at the seabed. Seawater is pumped out of the deposit to store off-peak energy and re-enters through turbines to recharge it (Kougias et al. 2019). Using a similar concept, underground siting in abandoned mines and caverns could be developed reasonably quickly (IEA 2020h). Storage of energy as gravitational potential can also be implemented using materials other than water, such as rocks and sand. Pumped technology is a mature technology (Rehman et al. 2015; Barbour et al. 2016) and can be important in supporting the transition to future low-carbon electricity grids (IHA 2021).

Batteries. There are many types of batteries, all having unique features and suitability, but their key feature is their rapid response time. A rechargeable battery cell is charged by using electricity to drive ions from one electrode to another, with the reverse occurring on discharge, producing a usable electric current (Crabtree et al. 2015). While lead-acid batteries (LABs) have been widely used for automotive and grid applications for decades (May et al. 2018), LIBs are increasingly being used in grid-scale projects (Crabtree et al. 2015), displacing LABs. The rapid response time of batteries makes them suitable for enhanced frequency regulation and voltage support, enabling the integration of variable renewables into electricity grids (Strbac and Aunedi 2016). Batteries can provide almost all electricity services, except for seasonal storage. LIBs, in particular, can store energy and power in small volumes and with low weight, making them the default choice for EVs (Placke et al. 2017). EV batteries are expected to form a distributed storage resource as this market grows, both impacting and supporting the grid (Staffell and Rustomji 2016).

Drawbacks of batteries include relatively short lifespans and the use of hazardous or costly materials in some variants. While LIB costs are decreasing (Schmidt et al. 2017; Vartiainen et al. 2020), the risk of thermal runaway, which could ignite a fire (Gur 2018; Wang et al. 2019a), concerns about long-term resource availability (Olivetti et al. 2017; Sun et al. 2017), and concerns about global cradle-to-grave impacts (Peters et al. 2017; Kallitsis et al. 2020) need to be addressed.

The superior characteristics of LIBs will keep them the dominant choice for EV and grid applications in the medium term ( high confidence). There are, however, several next-generation battery chemistries (Placke et al. 2017), which show promise ( high confidence). Cost reductions through economies of scale are a key area for development. Extending the life of the battery can bring down overall costs and mitigate the environmental impacts (Peters et al. 2017). Understanding and controlling battery degradation is therefore important. The liquid, air-reactive electrolytes of conventional LIBs are the main source of their safety issues (Janek and Zeier 2016; Gur 2018), so all-solid-state batteries, in which the electrolyte is a solid, stable material, are being developed. They are expected to be safe, be durable, and have higher energy densities (Janek and Zeier 2016). New chemistries and concepts are being explored, such as lithium-sulphur batteries to achieve even higher energy densities (Van Noorden 2014; Blomgren 2017) and sodium chemistries because sodium is more abundant than lithium (Hwang et al. 2017). Cost-effective recycling of batteries will address many sustainability issues and prevent hazardous and wasteful disposal of used batteries (Harper et al. 2019). Post-LIB chemistries include metal sulphur, metal-air, metal ion (besides lithium) and all-solid-state batteries.

Compressed air energy storage (CAES). With CAES, off-peak electricity is used to compress air in a reservoir – either in salt caverns for large-scale or in high-pressure tanks for smaller-scale installations. The air is later released to generate electricity. While conventional CAES has used natural gas to power compression, new low-carbon CAES technologies, such as isothermal or adiabatic CAES, control thermal losses during compression and expansion (Wang et al. 2017c). Fast responses and higher efficiencies occur in small-scale CAES installations, scalable to suit the application as a distributed energy store, offering a flexible, low-maintenance alternative (Luo et al. 2014; Venkataramani et al. 2016).

CAES is a mature technology in use since the 1970s. Although CAES technologies have been developed, there are not many installations at present (Wang et al. 2017b; Blanc et al. 2020). While the opportunities for CAES are significant, with a global geological storage potential of about 6.5 PW (Aghahosseini and Breyer 2018), a significant amount of initial investment is required. Higher efficiencies and energy densities can be achieved by exploiting the hydrostatic pressure of deep water to compress air within submersible reservoirs (Pimm et al. 2014). CAES is best suited to bulk diurnal electricity storage for buffering VRE sources and services, which do not need a very rapid response. In contrast to PHS, CAES has far more siting options and poses few environmental impacts.

Liquid air energy storage (LAES). LAES uses electricity to liquefy air by cooling it to ­­–196°C and storing it in this condensed form (largely liquid nitrogen) in large, insulated tanks. To release electricity, the ‘liquid air’ is evaporated through heating, expanding to drive gas turbines. Low-grade waste heat can be utilised, providing opportunities for integrating with industrial processes to increase system efficiency. There are clear, exploitable synergies with the existing liquid gas infrastructure (Peters and Sievert 2016).

LAES provides bulk daily storage of electricity, with the additional advantage of being able to capture waste heat from industrial processes. This technology is in the early commercial stage (Brandon et al. 2015; Regen 2017). Advances in whole systems integration can be developed to integrate LAES with industrial processes, making use of their waste heat streams. LAES uniquely removes contaminants in the air and could potentially incorporate CO2 capture (Taylor et al. 2012).

Thermal energy storage (TES). TES refers to a range of technologies exploiting the ability of materials to absorb and store heat or cold, either within the same phase (sensible TES), through phase changes (latent TES), or through reversible chemical reactions (thermochemical TES). Pumped Thermal Energy Storage (PTES), a hybrid form of TES, is an air-driven electricity storage technology storing both heat and cold in gravel beds, using a reversible heat-pump system to maintain the temperature difference between the two beds and gas compression to generate and transfer heat (Regen 2017). TES technologies can store both heat and cold energy for long periods, for example, in underground water reservoirs for balancing between seasons (Dahash et al. 2019; Tian et al. 2019), storing heat and cold to balance daily and seasonal temperatures in buildings and reducing heat build-up in applications generating excessive waste heat, such as data centres and underground operations.

TES can be much cheaper than batteries and has the unique ability to capture and reuse waste heat and cold, enabling the efficiency of many industrial, buildings, and domestic processes to be greatly improved ( high confidence). Integration of TES into energy systems is particularly important, as the global demand for cooling is expected to grow (Elzinga et al. 2014; Peters and Sievert 2016). Sensible TES is well developed and widely used; latent TES is less developed with few applications. Thermochemical TES is the least developed, with no application yet (Prieto et al. 2016; Clark et al. 2020). The potential for high-density storage of industrial heat for long periods in thermochemical TES (Brandon et al. 2015) is high, with energy densities comparable to that of batteries (Taylor et al. 2012), but material costs are currently prohibitive, ranging from hundreds to thousands of dollars per tonne.

Flywheel energy storage (FES). Flywheels are charged by accelerating a rotor/flywheel. Energy is stored in the spinning rotor’s inertia which is only decelerated by friction (minimised by magnetic bearings in a vacuum), or by contact with a mechanical, electric motor. They can reach full charge very rapidly, their state of charge can be easily determined (Amiryar and Pullen 2017), and they operate over a wide range of temperatures. While they are more expensive to install than batteries and supercapacitors, they last a long time and are best suited to stationary grid storage, providing high power for short periods (minutes). Flywheels can be used in vehicles, but not as the primary energy source.

Flywheels are a relatively mature storage technology but not widely used, despite their many advantages over electrochemical storage (Dragoni 2017). Conventional flywheels require costly, high tensile strength materials, but high-energy flywheels, using lightweight rotor materials, are being developed (Hedlund et al. 2015; Amiryar and Pullen 2017).

Supercapacitors – also known as ultracapacitors or double layer capacitors (Scap). Supercapacitors consist of a porous separator sandwiched between two electrodes, immersed in a liquid electrolyte (Gur 2018). When a voltage is applied across the electrodes, ions in the electrolyte form electric double layers at the electrode surfaces, held by electrostatic forces. This structure forms a capacitor, storing electrical charge (Brandon et al. 2015; Lin et al. 2017) and can operate from –40°C to 65°C.

Supercapacitors can supply high peaks of power very rapidly for short periods (seconds up to minutes) and are able to fulfil the grid requirements for frequency regulation, but they would need to be hybridised with batteries for automotive applications. Their commercial status is limited by costly materials and additional power electronics required to stabilise their output (Brandon et al. 2015). Progress in this area includes the development of high-energy supercapacitors, LIB-supercapacitor devices (Gonzalez et al. 2016), and cheaper materials (Wang et al. 2017a), all providing the potential to improve the economic case for supercapacitors, either by reducing manufacturing costs or extending their service portfolio.

Redox flow batteries (RFB). Redox flow batteries use two separate electrolyte solutions, usually liquids, but solid or gaseous forms may also be involved, stored in separate tanks, and pumped over or through electrode stacks during charge and discharge, with an ion-conducting membrane separating the liquids. The larger the tank, the greater the energy storage capacity, whereas more and larger cells in the stack increase the power of the flow battery. This decoupling of energy from power enables RFB installations to be uniquely tailored to suit the requirements of any given application. There are two commercially available types today: vanadium and zinc bromide, and both operate at near ambient temperatures, incurring minimal operational costs.

RFBs respond rapidly and can perform all the same services as LIBs, except for onboard electricity for EVs. Lower cost chemistries are emerging, to enable cost-effective bulk energy storage (Brandon et al. 2015). A new membrane-free design eliminates the need for a separator and also halves the system requirements, as the chemical reactions can coexist in a single electrolyte solution (Navalpotro et al. 2017; Arenas et al. 2018).

Power to fuels (PtX) (see also Section 6.4.3.1). The process of using electricity to generate a gaseous fuel, such as hydrogen or ammonia, is termed power-to-gas (PtG/P2G) (IEA 2020h). When injected into the existing gas infrastructure (Section 6.4.5), it has the added benefit of decarbonising gas (Brandon et al. 2015). Electricity can be used to generate hydrogen, which is then converted back into electricity using combined-cycle gas turbines that have been converted to run on hydrogen. For greater compatibility with existing gas systems and appliances, the hydrogen can be combined with captured carbon dioxide to form methane and other synthetic fuels (Thema et al. 2019), however, methane has high global warming potential and its supply chain emissions have been found to be significant (Balcombe et al. 2013).

PtX can provide all required grid services, depending on how it is integrated. However, a significant amount of PtX is required for storage to produce electricity again (Bogdanov et al. 2019) due to the low roundtrip efficiency of converting electricity to fuel and back again. However, portable fuels (hydrogen, methane, ammonia, synthetic hydrocarbons) are useful in certain applications, for example, in energy systems lacking the potential for renewables. The high energy density of chemical storage is essential for more demanding applications, such as transporting heavy goods and heating or cooling buildings (IEA 2020h). Research is needed into more efficient and flexible electrolysers which last longer and cost less (Brandon et al. 2015).

Hydrogen and reversible hydrogen fuel cells (H/RHFC). Hydrogen is a flexible fuel with diverse uses, capable of providing electricity, heat, and long-term energy storage for grids, industry, and transport, and has been widely used industrially for decades (Section 6.4.5.1). Hydrogen can be produced in various ways and stored in significant quantities in geological formations at moderate pressures, often for long periods, providing seasonal storage (Gabrielli et al. 2020). A core and emerging implementation of PtX is hydrogen production through electrolysers. Hydrogen is a carbon-free fuel holding three times the energy of an equivalent mass of gasoline but occupying a larger volume. An electrolyser uses excess electricity to split water into hydrogen and oxygen through the process of electrolysis. A fuel cell performs the reverse process of recombining hydrogen and oxygen back into water, converting chemical energy into electricity (Elzinga et al. 2014). Reversible hydrogen fuel cells (RHFCs) can perform both functions in a single device, however, they are still in the pre-commercial stage, due to prohibitive production costs.

Hydrogen can play an important role in reducing emissions and has been shown to be the most cost-effective option in some cases, as it builds on existing systems (Staffell et al. 2018). Fuel cell costs need to be reduced and the harmonies between hydrogen and complementary technologies, such as batteries, for specific applications need to be explored further. Hydrogen can provide long-duration storage to deal with prolonged extreme events, such as very low output of wind generation, to support resilience of future low-carbon energy systems. Research in this technology focuses on improving roundtrip efficiencies, which can be as high as 80% with recycled waste heat and in high-pressure electrolysers, incorporating more efficient compression (Matos et al. 2019). Photo-electrolysis uses solar energy to directly generate hydrogen from water (Amirante et al. 2017).

Table 6.6 | Technical characteristics of a selected range of battery chemistries, categorised as those which precede LIBs (white background), LIBs (yellow background) and post LIBs (blue background).

Note: With the exception of the All-solid-state batteries, all use liquid electrolytes. Source: aMahmoudzadeh et al. 2017; bManzetti and Mariasiu 2015; cPlacke et al. 2017; dNykvist and Nilsson 2015; eCano et al. 2018; f Bloomberg Energy Finance, 2019; gYou and Manthiram 2017; hFotouhi et al. 2017; iIRENA 2017b; jYang et al. 2020.

Public awareness and knowledge about electricity storage technologies, their current state, and their potential role in future energy systems is limited (Jones et al. 2018). For instance, people do not perceive energy system flexibility and storage as a significant issue, or assume storage is already taking place. Public perceptions differ across storage technologies. Hydrogen is considered a modern and clean technology, but people also have safety concerns. Moreover, the public is uncertain about hydrogen storage size and the possibility of storing hydrogen in or near residential areas (Eitan and Fischhendler 2021). Battery storage both on the household and community level was perceived as slightly positive in one study in the UK (Ambrosio-Albala et al. 2020). However, financial costs are seen as a main barrier. The potential of EV batteries to function as flexible storage is limited by the current numbers of EV owners and concerns that one’s car battery might not be fully loaded when needed.

Hydrogen is a promising energy carrier for a decarbonised world (Box 6.9). It can be utilised for electricity, heat, transport, industrial demand, and energy storage (Abdin et al. 2020). In low-carbon energy systems, hydrogen is expected to be utilised in applications that are not as amenable to electrification, such as a fuel for heavy-duty road transport and shipping, or as a chemical feedstock (Schemme et al. 2017; Griffiths et al. 2021). Hydrogen could also provide low-carbon heat for industrial processes or be utilised for direct reduction of iron ore (Vogl et al. 2018). Hydrogen could replace natural gas-based electricity generation (do Sacramento et al. 2013) in certain regions and support the integration of variable renewables into electricity systems by providing a means of long-term electricity storage. Hydrogen-based carriers, such as ammonia and synthetic hydrocarbons, can likewise be used in energy-intensive industries and the transport sector (Schemme et al. 2017; IRENA 2019b) (e.g., synthetic fuels for aviation). These hydrogen-based energy carriers are easier to store than hydrogen. At present hydrogen has limited applications – mainly being produced onsite for the creation of methanol and ammonia (IEA 2019c), as well as in refineries.

Low- or zero-carbon produced hydrogen is not currently competitive for large-scale applications, but it is likely to have a significant role in future energy systems, due to its wide-range of applications ( high confidence). Key challenges for hydrogen are: (i) cost-effective low/zero carbon production; (ii) delivery infrastructure cost; (iii) land area (i.e., ‘footprint’) requirements of hydrogen pipelines, compressor stations, and other infrastructure; (iv) challenges in using existing pipeline infrastructure; (v) maintaining hydrogen purity; (vi) minimising hydrogen leakage; and (vii) the cost and performance of end uses. Furthermore, it is necessary to consider the public perception and social acceptance of hydrogen technologies and their related infrastructure requirements (Iribarren et al. 2016; Scott and Powells 2020).

Hydrogen production. Low- or zero-carbon hydrogen can be produced from multiple sources. While there is no consensus on the hydrogen production spectrum, ‘blue’ hydrogen (Goldmann and Dinkelacker 2018) generally refers to hydrogen produced from natural gas combined with CCS through processes such as steam methane reforming (SMR) (Sanusi and Mokheimer 2019) and advanced gas reforming (Zhou et al. 2020). Low-carbon hydrogen could also be produced from coal coupled with CCS (Hu et al. 2020) (Table 6.7). Current estimates are that adding CCS to produce hydrogen from SMR will add on average 50% on the capital cost, 10% to fuel, and 100% to operating costs. For coal gasification, CCS will add 5% to the capital and fuel costs and 130% to operating costs (Staffell et al. 2018; IEA 2019d). Further, biomass gasification could produce renewable hydrogen, and when joined with CCS could provide negative carbon emissions. ‘Green’ hydrogen (Jaszczur et al. 2016) is most often referred to as hydrogen produced from zero-carbon electricity sources such as solar power and wind power (Schmidt et al. 2017) (Table 6.8). Nuclear power could also provide clean hydrogen, via electrolysis or thermochemical water splitting (EERE 2020). Hydrogen can even be produced by pyrolysis of methane (Sánchez-Bastardo et al. 2020) – sometimes called ‘turquoise’ hydrogen, solar thermochemical water splitting, biological hydrogen production (cyanobacteria) (Velazquez Abad and Dodds 2017) – and microbes that use light to make hydrogen (under research) (EIA 2020).

Table 6.7 | Key performance and cost characteristics of different non-electric hydrogen production technologies, including carbon capture and storage (CCS).

Source: aCSIRO 2021; bIEA 2020; cIRENA 2019; dHydrogen Council 2020; eCCC 2018; f BEIS 2021; gIshaq et al. 2021; hAl-Mahtani et al. 2021; iIEA 2019.

* USD per GBP exchange rate: 0.72 (August 2021); LHV: Lower Heating Values; Long-term refers to 2040 and 2050 according to different references.

Table 6.8 | Efficiency and cost characteristics of electrolysis technologies for hydrogen production.

Source: aCSIRO 2021; bIEA 2020; cIRENA 2019; dHydrogen Council 2020; eCCC 2018; f BEIS 2021; gIEA 2019; hChristensen 2020.

* USD per GBP exchange rate: 0.72 (August 2021); †The cost of hydrogen production from electrolysers is highly dependent on the technology, source of electricity, and operating hours, and some values provided are based on the assumptions made in the references.

Hydrogen energy carriers. Hydrogen can be both an energy carrier itself, be converted further into other energy carriers (such as synthetic fuels) and be a means of transporting other sources of energy. For example, hydrogen could be transported in its native gaseous form or liquefied. Hydrogen can also be combined with carbon and transported as a synthetic hydrocarbons (Gumber and Gurumoorthy 2018) (IRENA 2019d) as well as be transported via liquid organic hydrogen carriers (LOHCs) or ammonia (IRENA 2019d). For synthetic hydrocarbons such as methane or methanol to be considered zero carbon, the CO2 used to produce them would need to come from the atmosphere either directly through DACCS or indirectly through BECCS (IRENA 2019b). LOHCs are organic substances in liquid or semi-solid states, which store hydrogen based on reversible catalytic hydrogenation and de-hydrogenation of carbon double bounds (Niermann et al. 2019; Rao and Yoon 2020). Hydrogen produced from electrolysis could also be seen as an electricity energy carrier. This is an example of the PtX processes (Section 6.4.4), entailing the conversion of electricity to other energy carriers for subsequent use.

Ammonia is a promising cost-effective hydrogen carrier (Creutzig et al. 2019). Onsite generation of hydrogen for the production of ammonia already occurs today, and the ammonia (NH3) could be subsequently ‘cracked’ (with a 15–25% energy loss) to reproduce hydrogen (Hansgen et al. 2010; Montoya et al. 2015; Bell and Torrente-Murciano 2016). Because the energy density of ammonia is 38% higher than liquid hydrogen (Osman and Sgouridis 2018), it is potentially a suitable energy carrier for long-distance transport and storage (Salmon et al. 2021). Moreover, ammonia is more easily condensable (liquefied at 0.8 MPa, 20°C), which provides economically viable hydrogen storage and supply systems. Ammonia production and transport are also established industrial processes (about 180 MMT yr –1) (Valera-Medina et al. 2017), and hence ammonia is considered to be a scalable and cost-effective hydrogen-based energy carrier. At present, most ammonia is used in fertilisers (about 80%), followed by many industrial processes, such as the manufacturing of mining explosives and petrochemicals (Jiao and Xu 2018). In contrast to hydrogen, ammonia can be used directly as a fuel without any phase change for internal combustion engines, gas turbines, and industrial furnaces (Kobayashi et al. 2019). Ammonia can also be used in low- and high-temperature fuel cells (Lan and Tao 2014), whereby both electricity and hydrogen can be produced without any nitrogen oxide (NOx) emissions. Furthermore, ammonia provides the flexibility to be dehydrogenated for hydrogen-use purposes. Ammonia is considered a carbon-free sustainable fuel for electricity generation, since in a complete combustion, only water and nitrogen are produced (Valera-Medina et al. 2017). Like hydrogen, ammonia could facilitate management of VRE, due to its cost-effective grid-scale energy storage capabilities. In this regard, production of ammonia via hydrogen from low- or zero-carbon generation technologies along with ammonia energy recovery technologies (Afif et al. 2016) could play a major role in forming a hydrogen and/or ammonia economy to support decarbonisation. However, there are serious concerns regarding the ability to safely use ammonia for all these purposes, given its toxicity – whereas hydrogen is not considered toxic.

In general, challenges around hydrogen-based energy carriers – including safety issues around flammability, toxicity, storage, and consumption – require new devices and techniques to facilitate their large-scale use. Relatively high capital costs and large electricity requirements are also challenges for technologies that produce hydrogen energy carriers. Yet, these energy carriers could become economically viable through the availability of low-cost electricity generation and excess of renewable energy production (Daiyan et al. 2020). A key challenge in use of ammonia is related to the significant amount of NOx emissions, which is released from nitrogen and oxygen combustion, and unburned ammonia. Both have substantial air pollution risks, which can result in lung and other injuries, and can reduce visibility (EPA 2001). Due to the low flammability of hydrogen energy carriers such as liquefied hydrogen (Nilsson et al. 2016) and ammonia (Li et al. 2018), a stable combustion (Lamas and Rodriguez 2019; Zengel et al. 2020) in the existing gas turbines is not currently feasible. In recent developments, however, the proportion of hydrogen in gas turbines has been successfully increased, and further development of gas turbines may enable them to operate on 100% hydrogen by 2030 (Pflug et al. 2019).

Long-distance hydrogen transport. Hydrogen can allow regional integration and better utilisation of low- or zero-carbon energy sources (Boxes 6.9 and 6.10). Hydrogen produced from renewables or other low-carbon sources in one location could be transported for use elsewhere (Philibert 2017; Ameli et al. 2020). Depending on the distance to the user and specific energy carrier utilised (e.g., gaseous hydrogen or LOHC), various hydrogen transport infrastructures, distribution systems, and storage facilities would be required (Hansen 2020; Schönauer and Glanz 2021) (Figure 6.17).

Hydrogen can be liquefied and transported at volume over the ocean without pressurisation. This requires a temperature of –253°C and is therefore energy-intensive and costly (Niermann et al. 2021). Once it reaches its destination, the hydrogen needs to be re-gasified, adding further cost. A demonstration project is under development exporting liquid hydrogen from Australia to Japan (Yamashita et al. 2019). Hydrogen could also be transported as ammonia by ocean in liquid form. Ammonia is advantageous because it is easier to store than hydrogen (Zamfirescu and Dincer 2008; Soloveichik 2016; Nam et al. 2018). Liquid ammonia requires temperatures below –33°C and is therefore more straightforward and less costly to transport than liquefied hydrogen and even liquefied natural gas (Singh and Sahu 2018). A project exporting ammonia from Saudi Arabia to Japan is under consideration (Nagashima 2018). LOHCs could also be used to transport hydrogen at ambient temperature and pressure. This advantageous property of LOHCs makes them similar to oil products, meaning they can be transported in existing oil infrastructure including oil tankers and tanks (IEA 2019; Niermann et al. 2019). A project is under development to export hydrogen from Brunei to Japan using LOHCs (Kurosaki 2018).

Intra-regional hydrogen transportation. Within a country or region, hydrogen would likely be pressurised and delivered as compressed gas. About three times as much compressed hydrogen by volume is required to supply the same amount of energy as natural gas. Security of supply is therefore more challenging in hydrogen networks than in natural gas networks. Storing hydrogen in pipelines (linepack) would be important to maintaining security of supply (Ameli et al. 2017, 2019). Due to the physics of hydrogen, in most cases exiting gas infrastructure would need to be upgraded to transport hydrogen. Transporting hydrogen in medium- or high-pressure networks most often would require reinforcements in compressor stations and pipeline construction routes (Dohi et al. 2016). There are several recent examples of efforts to transport hydrogen by pipeline. For example, in the Iron Mains Replacement Programme in the UK, the existing low-pressure gas distribution pipes are being converted from iron to plastic (Committee on Climate Change 2018). In the Netherlands, an existing low-pressure 12 km natural gas pipeline has been used for transporting hydrogen (Dohi et al. 2016).

To bypass gas infrastructure in transporting hydrogen, methane can be transported using the existing gas infrastructure, while hydrogen can be produced close to the demand centres. This approach will only make sense if the methane is produced in a manner that captures carbon from the atmosphere and/or if CCS is used when the methane is used to produce hydrogen.

Bulk hydrogen storage. Currently, hydrogen is stored in bulk in chemical processes such as metal and chemical hydrides as well as in geologic caverns (Andersson and Grönkvist 2019; Caglayan et al. 2019) (e.g., salt caverns operate in Sweden) (Elberry et al. 2021). There are still many challenges, however, due to salt or hard rock geologies, large size, and minimum pressure requirements of the sites (IEA 2019c). Consequently, alternative carbon-free energy carriers, which store hydrogen, may become more attractive (Lan et al. 2012; Kobayashi et al. 2019).

Given the significant geographical variations in the efficiency of renewable resources across different regions and continents, electricity transmission could facilitate cost-effective deployment of renewable generation, enhance resilience and security of supply, and increase operational efficiency ( high confidence). The diurnal and seasonal characteristics of different renewable energy sources such as wind, solar, and hydropower can vary significantly by location. Through enhanced electricity transmission infrastructure, more wind turbines can be deployed in areas with high wind potential and more solar panels in areas with larger solar irradiation. Increases in electricity transmission and trade can also enhance operational efficiency and reduce or defer the need for investment in peaking plants, storage, or other load management techniques needed to meet security of supply requirements associated with localised use of VRE sources. Increased interconnectivity of large-scale grids also allows the aggregation of ‘smart grid’ solutions such as flexible heating and cooling devices for flexible demand in industrial, commercial, and domestic sectors (Hakimi et al. 2020) and EVs (Muratori and Mai 2020; Li et al. 2021). In general, interconnection is more cost-optimal for countries that are geographically close to each other and can benefit from the diversity of their energy mixes and usage (Schlachtberger et al. 2017). Such developments are not without price, however, and among other concerns, raise issues surrounding land use, public acceptance, and resource acquisition for materials necessary for renewable developments (Capellán-Pérez et al. 2017; Vakulchuk et al. 2020).

A number of studies have demonstrated the cost benefits of interconnected grids in a range of geographical settings, including across the USA (Bloom et al. 2020), Europe (Newbery et al. 2013; Cluet et al. 2020), between Australia and parts of Asia (Halawa et al. 2018), and broader global regions, for example between the Middle East and Europe or North Africa and Europe (Tsoutsos et al. 2015). While there is growing interest in interconnection among different regions or continents, a broad range of geopolitical and socio-techno-economic challenges would need to be overcome to support this level of international cooperation and large-scale network expansion (Bertsch et al. 2017; Palle 2021).

Status of electricity transmission technology. Long-distance electricity transmission technologies are already available. High voltage alternating current (HVAC), high-voltage direct current (HVDC), and ultra HVDC (UHVDC) technologies are well-established and widely used for bulk electricity transmission (Alassi et al. 2019). HVDC is used with underground cables or long-distance overhead lines (typically voltages between 100–800 kV) (Alassi et al. 2019) where HVAC is infeasible or not economic. A project development agreement, worth approximately USD17 billion, was signed in January 2021 that would connect 10 GW of PVs in the north of Australia via a 4500 km 3 GW HVDC cable to Singapore, suggesting that this would be cost effective (Sun Cable 2021). In September 2019, the Changji-Guquan ±1,100 kV UHVDC transmission project built by State Grid Corporation of China was officially completed and put into operation. The transmission line is able to transmit up to 12 GW over 3341 km (Pei et al. 2020). This is the UHVDC transmission project with the highest voltage level, the largest transmission capacity, and the longest transmission distance in the world (Liu 2015).

Other technologies that could expand the size of transmission corridors and/or improve the operational characteristics include low-frequency AC transmission (LFAC) (Y. Tang et al. 2021 ; Xiang et al. 2021) and half-wave AC transmission (HWACT) (Song et al. 2018; Xu et al. 2019). LFAC is technically feasible, but the circumstances in which it is the best economic choice compared to HVDC or HVAC still needs to be established (Xiang et al. 2016). HWACT is restricted to very long distances, and it has not been demonstrated in practice, so its feasibility is unproven. There are still a number of technological challenges for long-distance transmission networks such as protection systems for DC or hybrid AC-DC networks (Chaffey 2016; Franck C. et al. 2017), improvement in cabling technology, and including the use of superconductors and nanocomposites (Ballarino et al. 2016; Doukas 2019), which require advanced solutions.

Challenges, barriers, and recommendations. The main challenge to inter-regional transmission is the absence of appropriate market designs and regulatory and policy frameworks. In addition, there are commercial barriers for further enhancement of cross-border transmission. The differing impacts of cross-border interconnections on costs and revenues for generation companies in different regions could delay the development of these interconnectors. It is not yet clear how the investment cost of interconnections should be allocated and recovered, although there is growing support for allocating costs in accordance with the benefits delivered to the market participants. Increased cross-border interconnection may also require new business models which provide incentives for investment and efficient operation, manage risks and uncertainties, and facilitate coordinated planning and governance (Poudineh and Rubino 2017).

Optimising the design and operation of the interconnected transmission system, both onshore and offshore grids, also requires more integrated economic and reliability approaches (Moreno et al. 2012) to ensure the optimal balance between the economics and the provision of system security while maximising the benefits of smart network technologies.

A wide range of factors, including generation profiles, demand profiles circuit losses, reliability characteristics, and maintenance, as well as the uncertainties around them will need to be considered in designing and operating long-distance transmission systems if they are to be widely deployed (Djapic et al. 2008; Du 2009; De Sa and Al Zubaidy 2011; E3G 2021). Public support for extending transmission systems will also be crucial, and studies indicate that such support is frequently low (Vince 2010; Perlaviciute et al. 2018).

The impacts of climate change on hydropower will vary by region. High latitudes in the northern hemisphere are anticipated to experience increased runoff and hydropower potential. For other regions, studies find both increasing and decreasing runoff and hydropower potential. Areas with decreased runoff are anticipated to experience reduced hydropower production and increased water conflict among different economic activities ( high confidence).

Hydropower production is directly related to the availability of water. Changes in runoff and its seasonality and changes in temperature and precipitation intensity will influence hydroelectricity production (IHA 2019). In general, increased precipitation will increase water availability and hydropower production. Increased precipitation intensity, however, may impact on the integrity of dam structures and affect power production by increasing debris accumulation and vegetation growth. Additionally, increased precipitation intensity results in the silting of the reservoirs or increases the amount of water spilt, resulting in erosion (Schaeffer et al. 2012; IHA 2019). Climate change will likely lead to higher air temperatures, resulting in more surface evaporation, less water storage, and loss of equipment efficiency (Ebinger and Vergara 2011; Mukheibir 2013; Fluixá-Sanmartín et al. 2018; Hock et al. 2019). Climate change may alter the demands for water use by other sectors that often rely on stored water in multi-purpose reservoirs, and may therefore generate conflicts over water use. The increased need for water for irrigation and/or industry can affect the availability of water for hydropower generation (Spalding-Fecher et al. 2016; Solaun and Cerdá 2017). Higher temperatures increase glacier melt, increasing water availability for hydropower while the glaciers exist. Changes in the timing of snow and ice melt may require upgrading in storage capacity and adaptation of the hydropower plant management for fully exploiting the increase in water availability.

The conclusions regarding climate change impacts on hydropower vary due to differences in modelling assumptions and methodology, such as choice of the climate and hydrological models, choice of metrics (e.g., projected production vs hydropower potential), level of modelling details between local and global studies, reservoir operation assumptions. Also important is how hydropower production matches up with other reservoir purposes, accounting for other water and energy users, and how the competing uses are impacted by climate change (van Vliet et al. 2016b; Turner et al. 2017). Nonetheless, analyses consistently demonstrate that the global impact of climate change on hydropower will be small, but the regional impacts will be larger, and will be both positive and negative (Figure 6.20). Gross global hydropower potential in the 2050s has been estimated to slightlydecrease (Hamududu and Killingtveit 2012) between 0.4% (for the low-emission scenario) and 6.1% (for the highest-emission scenario) for the 2080s compared to 1971–2000 (van Vliet et al. 2016a).

Regional changes in hydropower are estimated from 5–20% increases for most areas in high latitudes (van Vliet et al. 2016b; Turner et al. 2017) to decreases of 5–20% in areas with increased drought conditions (Cronin et al. 2018). Models show a consistent increase in streamflow and hydropower production by 2080 in high latitudes of the northern hemisphere and parts of the tropics (Figure 6.20) (e.g., central Africa and southern Asia) while decreasing in the USA, southern and central Europe, Southeast Asia and southern South America, Africa and Australia (van Vliet et al. 2016c,a). Decreases in hydropower production are indicated for parts of North America, central and southern Europe, the Middle East, central Asia and Southern South America. Studies disagree on the changes in hydropower production in China, central South America, and partially in southern Africa (Hamududu and Killingtveit 2012; van Vliet et al. 2016b; Solaun and Cerdá 2019; Fan et al. 2020).

Climate change will not substantially impact future wind resources and will not compromise the ability of wind energy to support low-carbon transitions ( high confidence). Changing wind variability may have a small-to-modest impact on backup energy and storage needs (low confidence); however, current evidence is largely from studies focused on Europe.

Long-term global wind energy resources are not expected to substantially change in future climate scenarios (Karnauskas et al. 2018; Pryor et al. 2020; Yalew et al. 2020). However, recent research has indicated consistent shifts in the geographic position of atmospheric jets in the high-emission scenarios (Harvey et al. 2014), which would decrease wind power potentials across the Northern Hemisphere mid-latitudes and increase wind potentials across the tropics and the Southern Hemisphere. However, the climate models used to make these assessments differ in how well they can reproduce the historical wind resources and wind extremes, which raises questions about the robustness of their predictions of future wind resources (Pryor et al. 2020).

There are many regional studies on changes in wind resources from climate change. For Europe, there is medium evidence and moderate agreement that wind resources are already increasing and will continue to increase in Northern Europe and decrease in Southern Europe (Carvalho et al. 2017; Devis et al. 2018; Moemken et al. 2018). For North America, the various studies have low agreement for the changes in future wind resources in part because the year-to-year variations in wind resources are often larger than the future change due to climate change (Johnson and Erhardt 2016; Chen 2020; Costoya et al. 2020; Wang et al. 2020b). Studies show increases in future wind resources in windy areas in South America (Ruffato-Ferreira et al. 2017; de Jong et al. 2019). No robust future changes in wind resources have been identified in China (Xiong et al. 2019). However, none of the global or regional studies of the effects of climate change on wind resources considers the fine-scale dependence of wind resources on the topography and wind direction (Sanz Rodrigo et al. 2016; Dörenkämper et al. 2020) or the effect of expanding wind energy exploitation (Volker et al. 2017; Lundquist et al. 2019). There is limited evidence that extreme wind speeds, which can damage wind turbines, will increase due to climate change (Pes et al. 2017; Pryor et al. 2020). Nevertheless, projected changes in Europe and North America – regions where the most extensive analysis has been undertaken – are expected to be within the estimates embedded in the design standards of wind turbines (Pryor and Barthelmie 2013).

Future wind generation in Europe could decrease in summer and autumn, increasing in winter in northern-central Europe but decreasing in southernmost Europe (Carvalho et al. 2017). Towards 2100, intra-annual variations increase in most of Europe, except around the Mediterranean area (Reyers et al. 2016), but this may reflect natural multi-decadal variability (Wohland et al. 2019b). Wind speeds may become more homogeneous over large geographical regions in Europe due to climate change, increasing the likelihood of large areas experiencing high or low wind speeds simultaneously (Wohland et al. 2017). These changes could result in fewer benefits in the transmission of wind generation between countries and increased system integration costs. Europe could require a modest increase (up to 7%) in backup energy towards the end of the 21st century due to more homogeneous wind conditions over Europe (Wohland et al. 2017; Weber et al. 2018). However, other studies report that the impact of climate change is substantially smaller than interannual variability, with no significant impact on the occurrence of extreme low wind production events in Europe (Van Der Wiel et al. 2019). If European electricity systems are designed to manage the effects of existing weather variability on wind power, they can likely also cope with climate change impacts on wind power (Ravestein et al. 2018). Changes in wind-generation variability caused by climate change are also reported for North America (Haupt et al. 2016; Losada Carreño et al. 2018), with modest impacts on electricity system operation (Craig et al. 2019).

Climate change is not expected to substantially impact global solar insolation and will not compromise the ability of solar energy to support low-carbon transitions ( high confidence). Models show dimming and brightening in certain regions, driven by cloud, aerosol and water vapour trends (Chapter 12 of IPCC AR6 WGI). The increase in surface temperature, which affects all regions, decreases solar power output by reducing the PV panel efficiency. In some models and climate scenarios, the increases in solar insolation are counterbalanced by reducing efficiency due to rising surface air temperatures, which increase significantly in all models and scenarios (Jerez et al. 2015; Bartók et al. 2017; Emodi et al. 2019). Increases in aerosols would reduce the solar resource available and add to maintenance costs (Chapter 12 of IPCC AR6 WGI).

In many emission scenarios, the effect on solar PV from temperature-induced efficiency losses is smaller than the effect expected from changes on solar insolation due to variations in water vapour and clouds in most regions. Also, future PV technologies will likely have higher efficiency, which would offset temperature-related declines (Müller et al. 2019). Cloud cover is projected to decrease in the subtropics (around –0.05% per year), including parts of North America, vast parts of Europe and China, South America, South Africa and Australia (medium agreement , medium evidence). Thus, models project modest (<3%) increases in solar PV by the end of the century for southern Europe, northern and southern Africa, Central America, and the Caribbean (Emodi et al. 2019). There are several studies projecting decreasing solar production, but these are generally influenced by other factors, for example, increasing air pollution (Ruosteenoja et al. 2019). The multi-model means for solar insolation in regional models decrease 0.60 W m –2 per decade from 2006 to 2100 over most of Europe (Bartók et al. 2017), with the most significant decreases in the Northern countries (Jerez et al. 2015).

Climate change can affect biomass resource potential directly, via changes in the suitable range (i.e., the area where bioenergy crops can grow) and/or changes in yield, and indirectly, through changes in land availability. Increases in CO2 concentration increase biomass yield; climate changes (e.g., temperature, precipitation, and so on) can either increase or decrease the yield and suitable range.

Climate change will shift the suitable range for bioenergy towards higher latitudes, but the net change in the total suitable area is uncertain ( high confidence). Several studies show northward shifts in the suitable range for bioenergy in the northern hemisphere (Tuck et al. 2006; Barney and DiTomaso 2010; Bellarby et al. 2010; Hager et al. 2014; Wang et al. 2014a; Preston et al. 2016; Conant et al. 2018; Cronin et al. 2018), but the net effect of climate change on total suitable area varies by region, species, and climate model (Barney and DiTomaso 2010; Hager et al. 2014; Wang et al. 2014a).

The effect of climate change on bioenergy crop yields will vary across region and feedstock ( high confidence); however, in general, yields will decline in low latitudes (medium confidence) and increase in high latitudes (low confidence) (Haberl et al. 2010; Cosentino et al. 2012; Preston et al. 2016; Cronin et al. 2018; Mbow et al. 2019). However, the average change in yield varies significantly across studies, depending on the feedstock, region, and other factors (Beringer et al. 2011; Kyle et al. 2014; Mbow et al. 2019; Dolan et al. 2020). Only a few studies extend the modelling of climate change impacts on bioenergy to quantify the effect on bioenergy deployment or its implications on the energy system (Calvin et al. 2013, 2019; Kyle et al. 2014; Thornton et al. 2017). These studies find that changes in deployment are of the same sign as changes in yield; that is, if yields increase, then deployment increases.

Some of the uncertainty in the sign and magnitude of the impacts of climate change on bioenergy potential is due to uncertainties in CO2 fertilisation (the increase in photosynthesis due to increases in atmospheric CO2 concentration) (Haberl et al. 2011; Bonjean Stanton et al. 2016; Cronin et al. 2018; Solaun and Cerdá 2019; Yalew et al. 2020). For example, earlier studies found that, without CO2 fertilisation, climate change will reduce global bioenergy potential by about 16%; with CO2 fertilisation, however, climate change increases this potential by 45% (Haberl et al. 2011). However, newer studies in the USA find little effect of CO2 fertilisation on switchgrass yield (Dolan et al. 2020). There is also a considerable uncertainty across climate and crop models in estimating bioenergy potential (Hager et al. 2014).

The operation of thermal power plants will be affected by climate change, deriving from changes in the ambient conditions like temperature, humidity and water availability (Schaeffer et al. 2012) ( high confidence). Changes in ambient temperature have relatively small impacts on coal-fired and nuclear power plants (Rankine cycle); however, gas-fired power plants (Brayton or combined-cycle) may have their thermal efficiency and power output significantly decreased (De Sa and Al Zubaidy 2011; Schaeffer et al. 2012). Droughts decrease potential cooling water for thermal power plants and increase the probability of water outlet temperatures exceeding regulatory limits, leading to lower production or shutdowns. Thermal power utilisation has been reported to be, on average, 3.8% lower during drought years globally (van Vliet et al. 2016c), and further significant decreases in available thermal power plant capacity due to climate change are projected (Koch et al. 2014; van Vliet et al. 2016b; Yalew et al. 2020). An increase in climate-related nuclear power disruptions has been reported in the past decades globally (Ahmad 2021).

Carbon capture may increase cooling water usage significantly, especially in retrofits, with up to 50% increase in water usage for coal-fired power plants globally, depending on the CCS technology (Rosa et al. 2020) (Section 6.4). In Asia, planned coal capacity is expected to be vulnerable to droughts, sea level rise, and rising air temperatures, and this may be exacerbated by incorporating carbon capture (Wang et al. 2019c). Recently, however, studies have proposed designs of CCS with a minimal increase in water requirements (Magneschi et al. 2017; Mikunda et al. 2021).

Older thermal power plants can be retrofitted to mitigate climate impacts by altering and redesigning the cooling systems (Westlén 2018), although the costs for these solutions may be high. For example, dry cooling may be used instead of once-through cooling; however, it lowers thermal efficiency and would leave plants vulnerable to ambient temperature increase (Ahmad 2021). Closed-circuit cooling is much less sensitive to water temperature than once-through cooling (Bonjean Stanton et al. 2016). Modifying policies and regulation of water and heat emissions from power plants may also be used to mitigate plant reliability problems induced by climate change (Eisenack 2016; Mu et al. 2020), albeit with potential impacts for other water users and ecology. Improvements in water use and thermal efficiencies and the use of transmission capabilities over large geographical regions to mitigate risks on individual plants are also possible mitigation options (Miara et al. 2017).

Net-zero energy systems will use far less fossil fuel than today ( high confidence). The precise quantity of fossil fuels will largely depend on the relative costs of such fuels, electrification, alternative fuels, and CDR (Section 6.6.2.4) in the energy system ( high confidence). All of these are affected by regional differences in resources (McGlade and Ekins 2015), existing energy infrastructure (Tong et al. 2019), demand for energy services, and climate and energy policies. Fossil fuel use may persist, for example, if and where the costs of such fuels and the compensating carbon management (e.g., CDR, CCS) are less than non-fossil energy. For most applications, however, it is likely that electrification (McCollum et al. 2014; Madeddu et al. 2020; Zhang and Fujimori 2020) or use of non-fossil alternative fuels (Zeman and Keith 2008; Graves et al. 2011; Hänggi et al. 2019; Ueckerdt et al. 2021) will prove to be the cheapest options. Most residual demand for fossil fuels is likely to predominantly be petroleum and natural gas given their high energy density (Davis et al. 2018), while demand for coal in net-zero energy systems is likely to be very low (Luderer et al. 2018; Jakob et al. 2020, Section 6.7.4) ( high confidence).

There is considerable flexibility regarding the overall quantity of liquid and gaseous fuels that will be required in net-zero energy systems ( high confidence) (Figure 6.22 and Section 6.7.4). This will be determined by the relative value of such fuels as compared to systems which rely more or less heavily on zero-emissions electricity. In turn, the share of any fuels that are fossil or fossil-derived is uncertain and will depend on the feasibility of CCS and CDR technologies and long-term sequestration as compared to alternative, carbon-neutral fuels. Moreover, to the extent that physical, biological, and/or socio-political factors limit the availability of CDR (Smith et al. 2015; Field and Mach 2017), carbon management efforts may prioritise residual emissions related to land use and other non-energy sources.

Net-zero energy systems will rely on decarbonised or net-negative CO2 emissions electricity systems, due to the many lower-cost options for producing zero-carbon electricity and the important role of end-use electrification in decarbonising other sectors ( high confidence).

There are many possible configurations and technologies for zero- or net-negative-emissions electricity systems ( high confidence). These systems could entail a mix of variable renewables, dispatchable renewables (e.g., biomass, hydropower), other firm, dispatchable (‘on-demand’) low-carbon generation (e.g., nuclear, CCS-equipped capacity), energy storage, transmission, carbon removal options (e.g., BECCS, DACCS), and demand management (Luderer et al. 2017; Bistline et al. 2018; Jenkins et al. 2018b; Bistline and Blanford 2021b). The marginal cost of deploying electricity sector mitigation options increases as electricity emissions approach zero; in addition, the most cost-effective mix of system resources changes as emissions approach zero and, therefore, so do the implications of electricity sector mitigation for sustainability and other societal goals (Mileva et al. 2016; Bistline et al. 2018; Sepulveda et al. 2018; Jayadev et al. 2020; Cole et al. 2021). Key factors influencing the electricity mix include relative costs and system benefits, local resource bases, infrastructure availability, regional integration and trade, co-benefits, societal preferences and other policy priorities, all of which vary by country and region (Section 6.6.4). Many of these factors depend on when the net-zero point is reached (Figure 6.22).

Based on their increasing economic competitiveness, VRE technologies, especially wind and solar power, will likely comprise large shares of many regional generation mixes ( high confidence) (Figure 6.22). While wind and solar will likely be prominent electricity resources, this does not imply that 100% renewable energy systems will be pursued under all circumstances, since economic and operational challenges increase nonlinearly as shares approach 100% (Box 6.8) (Frew et al. 2016; Imelda et al. 2018 b; Shaner et al. 2018; Bistline and Blanford 2021a; Cole et al. 2021). Real-world experience planning and operating regional electricity systems with high instantaneous and annual shares of renewable generation is accumulating, but debates continue about how much wind and solar should be included in different systems, and the cost-effectiveness of mechanisms for managing variability (Box 6.8). Either firm, dispatchable generation (including nuclear, CCS-equipped capacity, dispatchable renewables such as geothermal, and fossil units run with low capacity factors and CDR to balance emissions) or seasonal energy storage (alongside other balancing resources discussed in Box 6.8) will be needed to ensure reliability and resource adequacy with high percentages of wind and solar (Jenkins et al. 2018b; Dowling et al. 2020; Denholm et al. 2021) though each option involves uncertainty about costs, timing, and public acceptance (Albertus et al. 2020).

Electricity systems require a range of different functional roles – for example, providing energy, capacity, or ancillary services. As a result, a range of different types of generation, energy storage, and transmission resources may be deployed in these systems (Baik et al. 2021). There are many options for each of these roles, each with their strengths and weaknesses (Sections 6.4.3 and 6.4.4), and deployment of these options will be influenced by the evolution of technological costs, system benefits, and local resources (Fell and Linn 2013; Hirth 2015; Bistline et al. 2018; Mai et al. 2018; Veers et al. 2019).

System management is critical for zero- or negative-emissions electricity systems. Maintaining reliability will increasingly entail system planning and operations that account for characteristics of supply- and demand-side resources (Hu et al. 2018). Coordinated planning and operations will likely become more prevalent across portions of the electricity system (e.g., integrated generation, transmission, and distribution planning), across sectors, and across geographies (EPRI 2017; Konstantelos et al. 2017; Chan et al. 2018; Bistline and Young 2019) (Section 6.4.3).

Energy storage will be increasingly important in net-zero energy systems, especially in systems with shares of VRE ( high confidence). Deployment of energy storage will vary based on the system benefits and values of different options (Arbabzadeh et al. 2019; Denholm and Mai 2019). Diurnal storage options like lithium-ion batteries have different value than storing and discharging electricity over longer periods through long-duration energy storage with less frequent cycling, which require different technologies, supporting policies, and business models (Gallo et al. 2016; Blanco and Faaij 2017; Albertus et al. 2020; Dowling et al. 2020; Sepulveda et al. 2021) (Section 6.4.4). The value of energy storage varies with the level of deployment and on the competitiveness of economic complements such as VRE options (Mileva et al. 2016; Bistline and Young 2020) and substitutes such as flexible demand (Brown et al. 2018; Merrick et al. 2018), transmission (Schlachtberger et al. 2017; Brown et al. 2018; Merrick et al. 2018; Bistline and Young 2019), trade (Bistline et al. 2020b), dispatchable generators (Hittinger and Lueken 2015; Gils et al. 2017; Arbabzadeh et al. 2019), direct air capture (DAC) (Daggash et al. 2019), and efficiencies in system operations (Tuohy et al. 2015).

The approach to other sectors could impact on electricity sector planning, and the role of some technologies (e.g., hydrogen, batteries, CCS) could depend on deployment in other sectors. CCS offers opportunities for CO2 removal when fuelled with syngas or biomass containing carbon captured from the atmosphere (Hepburn et al. 2019); however, concerns about lifecycle environmental impacts, uncertain costs, and public acceptance are potential barriers to widespread deployment (Section 6.4.2). It is unclear whether CDR options like BECCS will be included in the electricity mix to offset continued emissions in other parts of the energy system or beyond (MacDowell et al. 2017; Bauer et al. 2018 a; Luderer et al. 2018). Some applications may also rely on power to fuels (PtX) electricity conversion to create low-emissions synthetic fuels (Sections 6.6.2.6, 6.4.4, and 6.4.5), which could impact on electricity system planning and operations. Additionally, if DAC technologies are used, electricity and heat requirements to operate DAC could impact electricity system investments and operations (Realmonte et al. 2019; Bistline and Blanford 2021a).

Without additional efforts to reduce emissions, it is very unlikely that energy system CO2 emissions will decrease sufficiently to limit warming to well below 2°C ( high confidence). Scenarios assuming improvements in technology but no additional climate policies beyond those in place today provide a benchmark for comparison against energy-related CO2 emissions in mitigation scenarios (Figure 6.26). Emissions in these reference scenarios increase through 2050 but span a broad range (Riahi et al. 2017; Wei et al. 2018) (Chapter 3, Figure 3.16). The highest emission levels are about four times current emissions; the lowest are modestly below today’s emissions. Emissions in these scenarios increase in most regions, but they diverge significantly across regions (Bauer et al. 2017). Asia and the Middle East and Africa account for the majority of increased emissions across these scenarios (Figure 6.27). While it is unlikely that there will be no new climate policies in the future, these scenarios nonetheless support the conclusion that the energy sector will not be decarbonised without explicit policy actions to reduce emissions.

Warming cannot be limited to well below 2°C without rapid and deep reductions in energy system GHG emissions ( high confidence). Energy sector CO2 emissions fall by 87–97% (interquartile range) by 2050 in scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot and 60–79% in scenarios limiting warming to 2°C (>67%) with action starting in 2020 (Figure 6.26). Energy sector GHG emissions fall by 85–95% (interquartile range) in scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot, and 62–78% in scenarios limiting warming to 2°C (>67%) with action starting in 2020 (Figure 6.26). In 2030, in scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot, net CO2 and GHG emissions fall by 35–51% and 38–52% respectively. ­Key characteristics of emissions pathways – the year of peak emissions, the year when net emissions reach zero, and the pace of emissions reductions – vary widely across countries and regions. These differences arise from differences in economic development, demographics, resource endowments, land use, and potential carbon sinks (Schaeffer, et al. 2020; Schreyer, et al. 2020; van Soest, Heleen et al. 2021) (Figure 6.27, Figure 6.28, Box 6.11). If countries do not move quickly to reduce emissions – if reductions are delayed – a more rapid energy transition will subsequently be required to limit warming to 2°C or lower (Rogelj et al. 2015a, 2018a; IPCC 2018).

The timing of net-zero energy system emissions varies substantially across scenarios. In scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot (2°C (>67%)), the energy system reaches net-zero CO2 emissions (interquartile range) from 2060 onwards (2080–). (Figure 6.28). However, net emissions reach near-zero more quickly. For example, in scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot (2°C (>67%)) net energy system CO2 emissions drop by 95% between 2056 and 2075 (2073 and 2093). Net full economy GHG emissions reach zero more slowly than net CO2 emissions. In some scenarios, net energy system CO2 and total GHG emissions do not reach zero this century, offset by CDR in other sectors.

The timing of emissions reductions will vary across the different parts of the energy sector (Figure 6.28). To decarbonise most cost-effectively, global net CO2 emissions from electricity generation will likely reach zero before the rest of the energy sector (medium confidence). In scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot (2°C (>67%)), net electricity sector CO2 emissions (interquartile range) reach zero globally between 2044 and 2055 (2052 and 2078) (Figure 6.28). It is likely to be less costly to reduce net CO2 emissions close to or below zero in the electricity sector than in other sectors, because there are relatively more low-emissions options in electricity. Sectors such as long-distance transport, air transport, and process heat are anticipated to face greater challenges to decarbonisation than the electricity sector (Clark and Herzog 2014; Rogelj et al. 2015b, 2018b; IPCC 2018; Luderer et al. 2018).

In addition, there are potential options to remove CO2 from the atmosphere in the electricity sector, notably BECCS, which would allow electricity sector emissions to drop below zero. Without CDR options, electricity sector emissions may not fall all the way to zero. If CDR is accomplished in other sectors and not in electricity, some fossil fuel plants may still lead to positive net electricity sector CO2 emissions, even in net-zero economies (Bistline and Blanford 2021b; Williams et al. 2021a).

We lack sufficient understanding to pin down precise dates at which energy system CO2 emissions in individual countries, regions, or sectors will reach net zero. Net-zero timing is based on many factors that are not known today or are bound up in development of key technologies, such as energy storage, bioenergy, or hydrogen. Some countries have low-carbon resource bases that could support deep emissions reductions, while others do not. Timing is also affected by the availability of CDR options, whether these options are in the energy sector or elsewhere, and the discount rate used to assess strategies (Bednar et al. 2019; Emmerling et al. 2019). Moreover, while many scenarios are designed to minimise global mitigation costs, many other frameworks exist for allocating mitigation effort across countries (van den Berg et al. 2019) (Chapter 4).

There are multiple technological routes to reduce energy system emissions (Section 6.6). Here we discuss three of these: (i) decarbonising primary energy and electricity generation; (ii) switching to electricity, bioenergy, hydrogen, and other fuels produced from low-carbon sources; and (iii) limiting energy use through improvement of efficiency and conservation. CDR is discussed in Section 6.7.1.3 Fossil fuel transitions are discussed in Section 6.7.4.

Decarbonising primary energy and electricity generation. Limiting warming to well below 2°C requires a rapid and dramatic increase in energy produced from low- or zero-carbon sources ( high confidence). Low- and zero-carbon technologies produce 74–82% (interquartile range) of primary energy in 2050 in scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot and 55–68% in scenarios limiting warming to 2°C (>67%) (Figure 6.29). The share of low-carbon technologies in global primary energy supply today is below 20% (Chapter 3, Section 6.3, and Figure 6.29). The percentage of low- and zero-carbon energy will depend in part on the evolution of energy demand – the more that energy demand grows, the more energy from low- and zero-carbon sources will be needed, and the higher the percentage of total primary energy these sources will represent.

Low- and zero-carbon sources produce 97–99% of global electricity in 2050 in scenarios limiting warming to 1.5°C (>50%) with no or limited overshoot and 93–97% in scenarios limiting warming to 2°C (>67%) (Figure 6.29) (medium confidence). Decarbonising electricity generation, in tandem with increasing use of electricity (see below), is an essential near-term strategy for limiting warming. The increase in low- and zero-carbon electricity will occur while electricity demand grows substantially. Studies have projected that global electricity demand will roughly double by 2050 and quadruple to quintuple by 2100 irrespective of efforts to reduce emissions (Bauer et al. 2017; Luderer et al. 2017; IEA 2019a).

Renewable energy, especially generation from solar and wind, is likely to have an important role in many low-carbon electricity systems. The contributions of wind and solar electricity will depend on their levelised costs relative to other options, integration costs, system value, and the ability to integrate variable resources into the grid (Section 6.6). Electric sector technology mixes will vary by region but will typically include additional resources such as hydropower, nuclear power, fossil generation with CCS, energy storage resources, and geothermal energy, among others. Contributions of different options vary widely across scenarios based on different assumptions about these factors (Figure 6.30).

Nonetheless, it is likely that wind and solar will dominate low-carbon generation and capacity growth over the next couple of decades due to supporting policies in many countries, and due to their significant roles in early electric sector decarbonisation, alongside reductions in coal generation (Bistline and Blanford 2021b; Pan et al. 2021). Clean firm technologies play important roles in providing flexibility and on-demand generation for longer durations, though deployment of these technologies is typically associated with deeper decarbonisation levels (e.g., beyond 70–80% reductions), which are likely to be more important after 2030 in many regions, and with more limited CDR deployment (Baik et al. 2021; Bistline and Blanford 2021a; Williams et al. 2021a).

A combination of factors – user, business, cultural, regulatory, and transnational – will hinder low-carbon energy transitions. Strong path dependencies, even in early formative stages, can have lasting impacts on energy systems, producing inertia that cuts across technological, economic, institutional and political dimensions ( high confidence) (Rickards et al. 2014; Vadén et al. 2019) (Chapter 5).

Energy systems exemplify the ways in which massive volumes of labour, capital, and effort become sunk into particular institutional configurations (Bridge et al. 2013, 2018). Several embedded factors affect large-scale transformation of these systems and make technological diffusion a complex process:

•User environments affect purchase activities and can involve the integration of new technologies into user practices and the development of new preferences, routines, habits and evenvalues (Kanger et al. 2019).

•Business environments can shape the development of industries, business models, supply and distribution chains, instrument constituencies and repair facilities (Béland and Howlett 2016).

•Culture can encompass the articulation of positive discourses, narratives, and visions that enhance cultural legitimacy and societal acceptance of new technologies. Regulatory embedding can capture the variety of policies that shape production, markets and use of new technologies.

•Transnational community can reflect a shared understanding in a community of global experts related to new technologies that transcends the borders of a single place, often a country.

While low-carbon innovation involves systemic change (Geels et al. 2018), these are typically less popular than energy supply innovations among policymakers and the wider public. Managing low-carbon transitions is therefore not only a techno-managerial challenge (based on targets, policies, and expert knowledge), but also a broader political project that involves the building of support coalitions that include businesses and civil society (moderate evidence, high agreement ).

Low-carbon transitions involve cultural changes extending beyond purely technical developments to include changes in consumer practices, business models, and organisational arrangements. The development and adoption of low-carbon innovations will therefore require sustained and effective policies to create appropriate incentives and support. The implementation of such policies entails political struggles because actors have different understandings and interests, giving rise to disagreements and conflicts.

Such innovation also involves pervasive uncertainty around technical potential, cost, consumer demand, and social acceptance. Such uncertainty carries governance challenges. Policy approaches facing deep uncertainty must protect against and/or prepare for unforeseeable developments, whether it is through resistance (planning for the worst possible case or future situation), resilience (making sure you can recover quickly), or adaptation (changes to policy under changing conditions). Such uncertainty can be hedged in part by learning by firms, consumers, and policymakers. Social interactions and network building (e.g., supply and distribution chains, intermediary actors) and the articulation of positive visions, such as in long-term, low-emission development strategies, all play a crucial role. This uncertainty extends to the impacts of low-carbon innovations on energy demand and other variables, where unanticipated and unintended outcomes are the norm. For instance, rapid investments in public transport networks could restrict car ownership from becoming common in developing countries (Du and Lin 2017).

Current investments in fossil infrastructure have committed 500–700 GtCO2 of emissions, creating significant risks for limiting warming to 1.5°C (Callaghan 2020) ( high confidence). These current investments combined with emissions from proposed fossil infrastructure exceed the emissions required to limit warming to 1.5°C (medium confidence). Existing coal- and gas-fired electricity generation accounts for 200–300 GtCO2 of committed emissions. Emissions from coal generation are larger than for gas plants (Smith et al. 2019; Tong et al. 2019). The lifetime of coal-fired power plants is 25–50 years, creating long-lasting risks to climate goals (Erickson and Tempest 2015). Gas-fired power plants are younger on average than coal-fired power plants. Industry sector lock-in amounts for more than 100 GtCO2, while buildings and transport sector together contribute another 50–100 GtCO2 (Erickson and Tempest 2015).

Lock-in is also relevant to fossil resources. Both coal and gas exploration continue, and new permits are being issued, which may cause economic (Erickson et al. 2018) as well as non-economic issues (Boettcher et al. 2019).

The nature of lock-in varies across the energy system. For example, lock-in in urban and transport sectors is different from the electricity sector. Broadly, urban environments involve infrastructural, institutional, and behavioural lock-in (Ürge-Vorsatz et al. 2018). Addressing lock-in in these sectors requires action by multiple stakeholders and is unlikely with just technological evolution (Table 6.11).

Committed carbon emissions are unevenly distributed. The disproportionate high share of committed emissions in emerging economies is the result of rapid growth in recent years, which has led to a comparably young fossil infrastructure with substantial remaining life (Shearer et al. 2017). Mature industrialised countries tend to have older infrastructures, part of which will be up for retirement in the near future (Tong et al. 2019). Coal-fired power plants currently planned or under construction are associated with 150–300 GtCO2, of which about 75% and about 10% are located in Asia and the OECD respectively (Edenhofer et al. 2018; Pfeiffer et al. 2018). If implemented, these new fleets will further shorten all coal plants’ lifetimes by another 10 years for meeting climate goals (Cui et al. 2019).

Despite the imperative to reduce use of fossil fuels and the multiple health and other benefits from closing coal-based infrastructure (Portugal-Pereira et al. 2018; Liu et al. 2019a; Karlsson et al. 2020; Rauner et al. 2020; Cui et al. 2021), coal power plants have continued to be commissioned globally (Jewell et al. 2019; Jakob et al. 2020), most notably in Asian countries. Gas power plants also continue to be built. In many regions, new fossil electricity generation exceeds needed capacity (Shearer et al. 2017).

Existing policies and the NDCs are insufficient to prevent an increase in fossil infrastructure and associated carbon lock-in ( high confidence) (Bertram et al. 2015; Johnson et al. 2015). Current investment decisions are critical because there is limited room within the carbon budget required to limit warming to well below 2°C (Kalkuhl et al. 2019; Rosenbloom 2019). Delays in mitigation will increase carbon lock-in and could result in large-scale stranded assets if stringency is subsequently increased to limit warming (Box 6.11). Near-term implementation of stringent GHG mitigation policies are likely to be most effective in reducing carbon lock-in (Haelg et al. 2018). Near-term mitigation policies will also need to consider different energy transition strategies as a result of different resources and carbon budgets between countries (Lucas 2016; Bos and Gupta 2018).

Near-term policy choices are particularly consequential for fast-growing economies. For example, Malik et al. (2020) found that 133 to 227 GW of coal capacity would be stranded after 2030 if India were to delay ambitious mitigation through 2030 and then pursue an ambitious, post-2030 climate strategy. Cui et al. (2021) identified 18% of old, small, inefficient coal plants for rapid near-term retirement in China to help achieve air quality, health, water, and other societal goals and a feasible coal phase-out under climate goals. Comparable magnitudes of stranded assets may also be created in Latin America when adding all announced, authorised, and procured power plants up to 2060 (González-Mahecha et al. 2019). Options to reduce carbon lock-in include reducing fossil fuels subsidies (Box 6.3), building CCS-ready facilities, or ensuring that facilities are appropriately designed for fuel switching (Budinis et al. 2018). Substantial lock-in may necessitate considerable deployment of CDR to compensate for high cumulative emissions.

Past and present energy sector investments have created technological, institutional, and behavioural path dependencies aligned towards coal, oil, and natural gas ( high confidence). In several emerging economies, large projects are planned that address poverty reduction and economic development. Coal infrastructure may be the default choice for these investments without policies to invest in low-carbon infrastructure instead (Joshua and Alola 2020; Steckel et al. 2020). Path dependencies frequently have sustainability implications beyond carbon emissions. (Box 6.2 and Section 6.7.7). There are several SDG co-benefits associated with decarbonisation of energy systems (Section 6.7.7) (Sörgel et al. 2021). For example, coal mining communities frequently experience significant health and economic burdens from resource extraction.

Climate policy will be particularly effective if it targets key factors inhibiting, enabling, and motivating mitigation behaviours. As barriers differ across mitigation options, regions, and groups, tailored approaches are more effective (Grubb et al. 2017). When people face important barriers to change (e.g., high costs, legal barriers), policy would be needed make low-carbon actions more attractive, or to make high-carbon actions less attractive. As people generally face multiple barriers for change, combinations of policies would be more effective (Rosenow et al. 2017).

Financial incentives can motivate mitigation actions (Santos 2008; Thøgersen 2009; Bolderdijk et al. 2011; Eliasson 2014; Maki et al. 2016), particularly when actions are costly (Mundaca 2007). In many countries, more residential solar PV were installed after the introduction of favourable financial schemes such as feed-in-tariffs, federal income tax credits, and net metering (Wolske and Stern 2018). Similarly, many programs have promoted the installation of lower-carbon household options such as heat pumps, district heating, or solar water heaters across Europe, the Asia-Pacific and Africa (Hu et al. 2012; Sovacool and Martiskainen 2020; Ahmed et al. 2021). Yet, financial incentives may underperform expectations when other factors are overlooked. For example, people may not respond to financial incentives when they do not trust the organisation sponsoring the programme, or when it takes too much effort to receive the incentive (Mundaca 2007; Stern et al. 2016a). Financial incentives are more effective if combined with strategies addressing non-financial barriers.

Communicating financial consequences of behaviour seems less effective than emphasising social rewards (Handgraaf et al. 2013) or benefits of actions for people (e.g., public health, comfort) and the environment (Bolderdijk et al. 2013; Asensio and Delmas 2015, 2016; Schwartz et al. 2015; Ossokina 2020). Financial appeals may have limited effects because they reduce people’s focus on environmental consequences, weaken intrinsic motivation to engage in mitigation actions, provide a licence to pollute (Agrawal et al. 2015; Bolderdijk and Steg 2015; Schwartz et al. 2015), and because pursuing small financial gains is perceived not worth the effort (Bolderdijk et al. 2013; Dogan et al. 2014).

Providing information on the causes and consequences of climate change or on effective mitigation actions increases people’s knowledge and awareness, but generally does not promote mitigation actions by individuals (Abrahamse et al. 2005) or organisations (Anderson and Newell 2004). Fear-inducing representations of climate change may inhibit action when they make people feel helpless (O’Neill and Nicholson-Cole 2009). Energy-related advice and feedback can promote energy savings, load shifting in electricity use and sustainable travel, particularly when framed in terms of losses rather than gains (Gonzales et al. 1988; Wolak 2011; Bradley et al. 2016; Bager and Mundaca 2017 ). Also, credible and targeted information at the point of decision can promote action (Stern et al. 2016a). Information is more effective when delivered by a trusted source, such as peers (Palm 2017), advocacy groups (Schelly 2014), and community organisations (Noll et al. 2014), and when tailored to actors’ personal situations and core values (Daamen et al. 2001; Abrahamse et al. 2007; Bolderdijk et al. 2013; Boomsma and Steg 2014; Wolsko et al. 2016; van den Broek et al. 2017). This explains why home energy audits promoted energy savings (Delmas et al. 2013; Alberini and Towe 2015), and investments in resource efficiency and renewable energy generation (Kastner and Stern 2015).

Energy use feedback can promote energy saving behaviour within households (Fischer 2008; Grønhøj and Thøgersen 2011; Delmas et al. 2013; Karlin et al. 2015; Zangheri et al. 2019) and at work (Young et al. 2015), particularly when provided in real time or immediately after the action so that people learn the impact of different actions (Abrahamse et al. 2005; Faruqui et al. 2009; Delmas et al. 2013; Yu et al. 2015; Stern et al. 2016a; Tiefenbeck et al. 2016). Energy labels (Banerjee and Solomon 2003; Stadelmann 2017), visualisation techniques (Pahl et al. 2016), and ambient persuasive technology (Midden and Ham 2012) can encourage energy savings as they immediately make sense and hardly require users’ conscious attention. Feedback can make people aware of their previous mitigation behaviours, which can strengthen their environmental self-identity, and motivate them to engage in other mitigation actions, to act in line with their self-image (Van der Werff et al. 2014).

Social influence approaches that communicate what other people do or think can encourage mitigation actions (Clayton et al. 2015), as can social models of desired actions (Osbaldiston and Schott 2012; Abrahamse and Steg 2013; Sussman and Gifford 2013; Wolske et al. 2020). Feedback on one’s own energy use relative to others can be effective (Nolan et al. 2008; Allcott 2011; Schultz et al. 2015), although not always, and effect sizes are small (Abrahamse and Steg 2013) compared to other types of feedback (Karlin et al. 2015).

Interventions that capitalise on people’s motivation to be consistent can promote mitigation actions (Steg 2016). Examples are commitment strategies where people pledge to act (Abrahamse and Steg 2013; Lokhorst et al. 2013), implementation intentions where they additionally explicate how and when they will perform the relevant action and how they would cope with possible barriers (Bamberg 2000, 2002; Rees et al. 2018), and hypocrisy-related strategies that make people aware of inconsistencies between their attitudes and behaviour (Osbaldiston and Schott 2012).

Bottom-up approaches can promote mitigation action (Abrahamse and Steg 2013). Indeed, community energy initiatives can encourage members’ low-carbon behaviour (Middlemiss 2011; Seyfang and Haxeltine 2012; Abrahamse and Steg 2013; Sloot et al. 2018). Organisations can promote mitigation behaviour among their employees and customers by communicating their mission and strategies to mitigate climate change (Ruepert et al. 2017; van der Werff et al. 2021).

Default options, where a preset choice is implemented if users do not select another option, can promote mitigation actions such as energy savings, green electricity uptake, and meat-free options (Pichert and Katsikopoulos 2008; Bessette et al. 2014; Campbell-Arvai et al. 2014; Kunreuther and Weber 2014; Ölander and Thøgersen 2014; Ebeling and Lotz 2015; Liebe et al. 2018; Liebe et al. 2021).

Public acceptability reflects the extent to which the public evaluates climate policy, mitigation options, and system changes (un)favourably, which can shape, enable, or prevent low-carbon energy system transitions. Public acceptability of policy and mitigation options is higher when people expect these have more positive and less negative consequences for self, others, and the environment (Perlaviciute and Steg 2014; Demski et al. 2015; Drews and Van den Bergh 2016). Public opposition may result when a culturally valued landscape is affected by renewable energy development (Warren et al. 2005; Devine-Wright and Howes 2010), particularly when place-based identities are threatened (Devine-Wright 2009, 2013; Boudet 2019). Acceptability can increase after a policy or change has been implemented and the consequences appear to be more positive than expected (Schuitema et al. 2010; Eliasson 2014; Weber 2015; Carattini et al. 2018); effective policy trials can thus build public support.

Next, climate policy and low-carbon options are evaluated as more fair and acceptable when costs and benefits are distributed equally, and when nature, the environment and future generations are protected (Schuitema et al. 2011; Drews and Van den Bergh 2016). Compensating affected groups for losses due to policy or systems changes enhanced public acceptability in some cases (Perlaviciute and Steg 2014), but people may disagree on which compensation would be worthwhile (Aitken 2010b; Cass et al. 2010), on the distribution of compensation (Devine-Wright and Sherry-Brennan 2019; Leer Jørgensen et al. 2020), or feel they are being bribed (Cass et al. 2010; Perlaviciute and Steg 2014). Pricing policies are more acceptable when revenues are earmarked for environmental purposes (Steg et al. 2006; Sælen and Kallbekken 2011) or redistributed towards those affected (Schuitema and Steg 2008).

Climate policy and mitigation options, such as renewable energy projects, are also perceived as more fair and acceptable when the public (Dietz 2013; Bidwell 2014; Bernauer et al. 2016b) or public society organisations (Terwel et al. 2010; Bernauer et al. 2016b) could participate in the decision-making (Arvai 2003; Devine-Wright 2005; Terwel et al. 2012; Walker and Baxter 2017; Perlaviciute and Squintani 2020). People are more motivated to participate in decision-making on local projects than on national or general policy goals (Perlaviciute and Squintani 2020). Public acceptability is also higher when people can influence major rather than only minor decisions, particularly when trust in responsible parties is low (Liu et al. 2019a). Public participation can enhance the quality and legitimacy of decisions by including local knowledge and views that may otherwise be missed (Dietz 2013; Bidwell 2016).

Public support is higher when people trust responsible parties (Perlaviciute and Steg 2014; Drews and Van den Bergh 2016; Michaels and Parag 2016; Jiang et al. 2018; Liu et al. 2019a). Public support for unilateral climate policy is rather strong and robust (Bernauer et al. 2016a), even in the absence of reciprocal commitments by other states (Bernauer and Gampfer 2015).

Public acceptability of climate policy and low-carbon options differs across individuals. Climate policy and low-carbon options are more acceptable when people strongly value protecting other people and the environment, and support egalitarian worldviews, left-wing or green political ideologies, while acceptability is lower when people strongly endorse self-centred values, and support individualistic worldviews (Dietz et al. 2007; Perlaviciute and Steg 2014; Drews and Van den Bergh 2016). Similarly, public decision-makers support climate policy more when they endorse environmental values (Nilsson et al. 2016). Climate and energy policy is more acceptable when people are more concerned about climate change (Hornsey et al. 2016), when they believe their actions would help mitigate climate change, and feel responsible to mitigate climate change (Steg 2005; Eriksson et al. 2006; Jakovcevic and Steg 2013; Drews and Van den Bergh 2016; Kim and Shin 2017; Ünal et al. 2019).