Photo: © Shutterstock – FuzzBones

Analyzing global energy scenarios

Coal has peaked worldwide and it won’t come back, and that is as official as it gets.

It is confirmed by the International Energy Agency (IEA) in its December 2018 Coal report. The IEA has been a staunch coal champion since its inception in 1974, as part of its drive for security of energy supply.

“Global coal demand will be stable though 2023,” says the International Energy Agency in its annual coal report: Coal 2018 – Analysis and forecasts to 2023.

This admission stands in stark contrast to the IEA’s previous predictions. In 2014 it forecast a global coal demand of 6462 million tonnes for 2019, an increase of almost 16 per cent from 2013.

Less than five years ago, the conventional wisdom was that coal would keep on increasing for a long time to come.

In the real world, it peaked in 2013 at 5588 Mtce1, fell between 2014 and 2016 and even after an uptick in 2017 it was 4 per cent down on 2013.

Coal is the worst fuel in almost every respect regarding the climate, environment and health.

The difference between +16 per cent and -4 per cent is big: 3 billion tonnes of CO2 per year, assuming that coal remains flat from 2017 to 2019. Because many of the dirtiest coal power plant have been retired and the newer plants are somewhat cleaner, the health benefits of declining coal are even greater.

The political lesson is just as striking. King Coal is not invincible. Even in China, which uses half of the coal in the world, coal has declined since 2013. It will continue to do so, slowly, according to the IEA.

The agency also predicts declines in Europe, North America and Japan, but expects coal to increase substantially in India and Southeast Asia.

But every new prediction is lower than the previous one. In December 2016, three years after the peak, the IEA reckoned that global demand would be 5469 Mtce in 2019. Two years later that was adjusted to 5389 Mtce, and so on. It has consistently overestimated coal and underestimated renewables.

The economic and political forces against coal are growing stronger, as can be followed on CoalWire at Every issue has more bad news for coal: big financial institutions are no longer financing coal, coal power projects are being abandoned, power plants, mines and harbours are being closed. In the background, the costs for wind, solar and batteries are falling below the remaining coal power plants.

In just the two months since early December, 3 GW of coal power projects have been axed in Japan, 1 GW in Turkey, 3 GW in Thailand, and 12.5 GW are to be phased out in Germany by 2022.

Much of this was not known when the IEA made its forecasts.

So it may be overestimating once again.

The IAE was formed in 1974 as a response from the OECD countries after the 1973 the oil price shock, as a kind of intergovernmenal think tank for what were then known as the rich countries. Ever since then it has promoted coal and nuclear, and to some extent energy efficiency, in order to decrease dependence on oil. It produces a large volume of data and reports, the best known of which is the

World Energy Outlook every November. The most recent report appointed itself “The gold standard of energy analysis” on its title page.

If you overlay the curves of the IEA forecasts for coal, the earlier curves point almost straight up, while the later ones become flatter and flatter, like the quills of a porcupine (figure).

The curves for solar and wind form a similar pattern, but the actual values are always higher than IEA’s forecasts.

The flaw in the IEA forecasting method is also reflected in oil price forecasts. In 2004, it was forecast that the price would stay at $22 in 2006–2010 and then rise to $29 by 2030. In reality, it rose to $110, and then fell to $45 by 2016 The track record is not great.

Every year the IEA scenarios have overestimated the amount of coal and nuclear power and underestimated wind and solar. In other words they have been a conservative force. This may be because they have tuned the models to get these results, but could also be because this type of model always shows more resistance to change than we see in real life. They assume that the present energy mix represents an equilibrium that is expensive to deviate from.

The IEA is not alone in this. The forecasts of the US Energy Information Agency and  BP show exactly the same flaw, and many national scenarios and even NGO scenarios have persistently underestimated the force for change.

When the IEA was first set up, and for many years afterwards, oil was the yardstick for measuring all energy. Everything was measured in tonnes of oil equivalents. The phrase “security of supply” reflected the unquestionable demand of the US, Europe and Japan for access to oil at a price that they felt to be right.

Oil is much less important today. Europe uses far less oil2 than it did in 1973, despite enormous growth and a massive increase in transport.

An underlying assumption since the models were first developed is that growth in GDP controls the “demand” for energy sources. This assumption worked fairly well during the development phase of poorer countries, but shows a poor match for developed countries. The geography and history of countries plays a much smaller role now than in the early days of these scenarios.

If you want to cut carbon dioxide emissions, roughly the same formula applies everywhere. We know today that wind and solar power work all over the world. This means we can get as much electricity as we need without fossil fuels or nuclear power3, and that this electricity will be relatively cheap4. Electricity from wind and sun may not always be available when and where we need it, but there is plenty of space for solar and wind within the existing energy system, especially in countries that have hydropower or can use hydropower from other countries. The share of solar and wind power could be still greater with demand management and moderate grid expansion. In dry, sunny countries, concentrated solar power (CSP) can provide energy balancing for several hours.

Hydrogen from electrolysis5 can enable further balancing, along with electric cars and storage batteries.

Batteries are much too expensive for main grid balancing, but can be economically viable for overstretched local grids or as an alternative to otherwise unavoidable grid expansion. By using methods such as demand management and battery storage, local grids can solve the problems of the main grid faster and cheaper than new 400 kilovolt lines. That is what “smart grid”, microgrids and “grid edge” is all about.

Lighting was previously a major consumer of electricity, but with the adoption of technology such as LEDs, timers, presence-sensing and daylight monitoring lighting accounts for no more than 6 per cent of US electricity.6

Heating and cooling used to be major consumers of fuel or electricity, depending on the local climate and technical solutions. This is no longer the case. Heat pumps can be used to provide heating and cooling everywhere. There is no strong reason to expand district heating and cooling. Everything can be done locally, within months rather than years. The need for heating and cooling also depends on the performance of building climate shells and ventilation systems. Whatever the climate, it makes sense to have draught-proof, well-insulated buildings with good windows, which either keep the cold out and the heat in, or vice versa. Such buildings must also have effective ventilation systems and efficient refrigerators, freezers and stoves.

The entire transport sector can in principle be powered by electricity, or hydrogen. The technology is there.

Industries that need heat can get it from electricity instead of fossil fuels. Steel can be produced from ore using electricity or hydrogen. In the near future, primary aluminium could be produced using electricity7, instead of electricity and carbon electrodes, as now.

Biofuels will be used, if nothing else in the form of waste (banana peels, straw, dung, wastewater, residue from the pulp industry), but there is no reason to increase their use significantly, as this leads to conflicts with biodiversity.

Some modellers see “optimisation” as the key, but this is unnecessarily sophisticated. None of the positive developments we have seen – the growth of wind power, solar power and efficiency improvements such as Energy Star – have been optimised. They were an expression of political will. The use of solar panels in Germany was far from the optimum solution: it was an absurdly expensive technology in a country where the sun hardly shines and where peak consumption occurs in January, when solar panels provide almost nothing. Nevertheless, political will has transformed the market. Solar panels, wind power and efficient office machinery are taking market share everywhere. In 2017, worldwide8, solar power grew 35,2%, wind power 17,3 %, much more than gas power (+1,4%) and coal power (+3,2 % an uptick in a declining trend). Oil power decreased 7,6 %.

The vision of the environmental movement could now be close to that of the electricity industry in the 1970s: total electrification. They wanted all this electricity to come from nuclear power, for example, whereas NGOs want renewable electricity. The plan of action could be to pile on more sun and wind, and eliminate fossil fuels everywhere.

The usefulness of huge complex computer models is questionable, if we already know what we have to do.

 As we can see in many parts of the world (China, Germany, Denmark and California) rapid growth in renewables does present some problems, but it also delivers solutions.9

In addition to more electricity we also need R&D in certain areas, for example in demand management, greehouse gas reduction in agriculture and some parts of the process industry. But these must be prioritised on the basis of our existing knowledge, not a computer model.

They base their results on macro-economic data, such as GDP growth, population and the oil price.

This is doomed to fail, because there is not much connection between GDP growth and energy or electricity growth, other than in low-income countries

Every year the IEA World Energy Outlook scenarios have overestimated the use of coal and nuclear power and underestimated wind and solar power. They have thus been a conservative force. This may be because they have tuned the models to get these results, but could also be because this type of model always shows more resistance to change than we see in real life.

The models are often based on a business-as-usual scenario in which you change the requirements and get new results. But there is no such thing as “business as usual”. Business is always changing. A plausible future model made 10 March 2011 might have forecast that by 2025 Japan would have 60 nuclear reactors and Germany 20. A few weeks later it was clear that Germany would not have any nuclear power and Japan would have 0–20 reactors by this time.

This also illustrates the danger of trying to incorporate political circumstances into national forecasts. It is clearly a political challenge to phase out nuclear power in France, Sweden or Finland because the nuclear power industry is a big political issue, in the same way as coal power in Poland. But if we make these concerns part of “national circumstances”, coal and nuclear will never be phased out.
A few years ago the German lignite industry was planning its operations beyond 2050. But now the phase-out has begun and its final year is set as 2038. The environmental movement is fighting for earlier closure.

National scenarios do not need to be overly refined. Geographical differences can be ironed out through trade. Denmark does not have any hydropower, but could still get 100 per cent (66 per cent wind and solar projected for 2021) of its electricity from renewables over a 12-month period, thanks to hydropower from Norway and Sweden. If there is already some excess capacity, as in Denmark, then the problem already has a solution. Many new distribution lines are already being built, for example between Norway and the UK, and in Germany.

Bottom-up analysis is scientifically sound, but complex and difficult to communicate to politicians and the public. The question is whether there is any great need.The entire analytical apparatus is based on the idea of a “primary energy demand” that can be divided between energy sources. But why? Electricity is electricity and fuel is fuel! No one can picture a mix of electricity and fuel.

If we are going to do modelling it should be energy backcasting. Start with the 1.5°C target and work out what it will take to get there. The most promising path does not have to be linear, and is unlikely to be the EU’s “walk first, run later”. The better approach would be “Run first, walk later”, as most of the technology is already familiar, cheap and can be applied quickly. We can do the unfamiliar, difficult and expensive bits later, as they may require research, development and scaling up of technologies, as well as testing of new incentives before they can be applied in full scale.

Fredrik Lundberg

1 Million tonnes of coal equivalent, the energy contents of as many million tonnes of standard coal.
2 See BP Statistical Review of World Energy June 2018
3 See for example  Mark Z. Jacobson et al, also
4 As for actual current cost of electricity, see
5 For hydrogen in industry, see Hydrogen cars exist, such as Toyota Mirai, but are not quite commercial.
8 BP Stats op cit

Figure: Share of electricity generation from renewables , except hydro power in different years. Dotted lines indicate projections made by IEA in the years labelled, solid line gives the actual shares according to their definitions.


In this issue