Historical transitions in national power generation
Power generation is the primary use of coal today, accounting for over 60% of total global consumption[1]. 1.5°C pathways published by the IPCC[2] see very rapid reductions in global coal use during the 2020s, especially in power generation, compared to less severe reductions in oil and in fossil methane gas (hereafter referred to as gas) (Fig. 1). However, with around half of the world’s coal use in China, and a quarter in other non-OECD countries[3], this rate of global reduction will be felt especially in those countries, where coal often provides a majority of power generation. To consider the feasibility of this rate of change, we examine reductions at the country level in TIAM-UCL and compare with past transitions across 144 countries, using historical generation data from the IEA[4].
In this study we focus on two important general dimensions of system inertia that make transition more difficult: the size of the system and the degree of the incumbent technology’s dominance within it, though feasibility is also affected by more specific factors including degree of stranded assets[5],[6], age of generation fleet[7], growth rate of generation[8] flexibility of the electricity system[9], and wider institutional capacity[10].
The size of the generation system is important because larger systems carry more inertia[11],[12], as in the cliché of turning a supertanker. A larger system has more power plants, more employees and more or larger companies, each slowing the process of transition.
One lesson of the energy transitions literature[13],[14] is that market share rather than absolute amount of generation shapes the pace of transition. This reflects the simple intuition that a system can adapt faster if it is less dependent on a technology. We focus on the percentage-point decline in a technology’s or fuel’s share of generation, to partly capture this dependence: a percentage-point decline can only be high where a technology’s share is initially high. (Using a negative compound growth rate would have omitted this dimension.)
Fig. 2 plots countries’ fastest 10-year reduction in a fuel or technology’s share of generation, against size of the generation system. The dashed line shows the least-squares best fit, and may be interpreted as an “average” pace of all countries’ fastest transitions, relative to the size of their generation systems. The solid line is the closest line to the x-axis that contains all the points, and represents the “world record” of the fastest transitions by the fastest countries.
Several observations may be made. First, the graph confirms the prior finding[15] that faster transitions are possible in smaller power systems. To see this intuitively, the “transition” can be almost instantaneous in small countries, such as Malta’s 2012-17 conversion of Delimara, its sole power station, from heavy oil to gas. As systems get larger, less rapid transitions have been achieved.
Second, there is a large spread. Some countries’ power systems have been highly stable over the period, especially where they are dominated by a single source. For example, in Norway, where hydro has consistently provided above 94% of power, the fastest transition was a mere 5% decline in 2000-2010. Other countries have gone through two or more major transitions, such as Togo, Malaysia, Denmark and Japan.
Third, the countries closest to the “world record” are all relatively wealthy, reinforcing the causal link between socioeconomic capacity and phaseout feasibility[16],[17]. Among poorer countries, the fastest transitions were generally driven primarily by external events, and often to the countries’ cost. For example, modestly-growing gas and renewables gained a 22% share of Jamaica’s generation only after oil generation collapsed due to oil price rises[18].
Modelled coal decline pathways under Paris-aligned scenarios
As noted above, 1.5°C pathways in IAMs see rapid reductions in global coal use during the 2020s. We use the global energy system model TIAM-UCL – which delivers similar coal decline rates to other IAMs in its default 1.5C scenarios (Fig. 1) – to examine the implied pace of decline in more detail at national levels. TIAM-UCL divides the world into eight single countries (Canada, China, India, Japan, Mexico, South Korea, UK and USA) and eight multi-country regions.
Fig. 3a superimposes the modelled 2020-30 change in coal power on the framework of historical transitions, for the ten largest coal-power-consuming countries. For the countries that are not represented as a separate region - South Africa, Australia, Indonesia, Germany and Russia - we assume their shares of their regions’ coal generation and total generation remain constant.
For China, India and South Africa, coal declines around twice as fast as any decline seen for any country, relative to its size. Comparable pace has occurred only in much smaller countries: for example, China would need to reduce its coal power in the 2020s as quickly as Belgium in the 1960s (Fig. 2) – a system just 0.4% of the size of China’s. In addition, China, India and South Africa must not only reduce coal’s share of generation by 20%, 25% and 30% (respectively) in a decade, but must do so for three decades in a row.
Fig. 3b plots these same 10 countries, but with coal power phased out by the 2030/2050 deadlines adopted by the PPCA, assuming a linear decline in coal’s share of generation to zero on those dates. This is instructive, because it reflects a phaseout pace at the high end of political ambition in the real world. In this case, most of the top 10 countries are just inside or just outside the “world record”; only Indonesia and Russia decline more slowly than the average line. This suggests that for most of the largest coal consumers, the PPCA timelines are close to the limits of feasibility based on historical precedent, such that it is hard to imagine a faster phaseout. However, Indonesia, Russia and some smaller or less coal-dependent power systems may be able to move faster than this timeline. If so, this has implications for where, how and how much mitigation efforts must be made, which we turn to in the next section.
Consequences of slower coal phaseout for 1.5°C pathways
We now assess the implications for the wider energy system transition if rates of coal decline in key coal consuming regions are constrained. We constrain the model to phase out coal power no faster than the PPCA timeline in each country/region, and compare 1.5°C scenarios with and without this constraint (see Methods).
The slower decline in coal generation leads to a change in fossil fuel primary energy, as shown in Fig. 4a. Under the PPCA scenario, slower coal decline is compensated by more rapid reduction in gas and oil, with gas peaking earlier in 2025, compared to 2030 in the unconstrained case. As for where this is produced, the acceleration of decline in oil and gas is unevenly distributed between regions. The United States, Australia and Mexico see the largest proportional reductions in cumulative gas production over the period 2020-2050 with respectively 20%, 12% and 11% (Fig. 4c). Western Europe (Norway) and the United States the largest proportional reductions in cumulative oil production with 19% and 12% respectively (Fig. 4d).
A key impact of the constraint is a shift in mitigation effort at the regional level. A slower decline in coal-dominated non-OECD countries such as China and India (with a 2050 PPCA coal phaseout deadline) means that other regions will need to mitigate faster. This is shown in Fig. 5a, where the annual rate of change of CO2 emissions between 2020 and 2045 is barely changed in China, India and Other Global South, whereas it accelerates from -10.4% to -16.6% in the United States, from -8.1% to -12.5% in Europe, and from -7.7% to -10.7% in Other Global North.
From a sectoral perspective, with the increase in power sector emissions, it is the transport sector that provides the majority of additional mitigation required, with a contribution also from the industry sector (Fig. 5b, 5c). Overall, the effect is that cumulative emissions are higher to 2045 under the PPCA case but are then balanced by additional mitigation in 2050 and beyond, as shown by the ‘net change’ trend line.
More specifically, the transport reductions come primarily from faster reduction in oil use in cars (mostly before 2035), road freight (2035-2045) and shipping (2040 to 2050) (Fig. 5d).