An s-curve model for deployment of new energy technologies based on analysis of past rates of technology deployment was undertaken nearly a decade ago [i], Its assessments can now be compared with what has happened since given an extra decade or so of historical data, and the implications for policy reconsidered in this light.
The analysis showed that new technologies typically start with a period of exponential growth, increasing by about an order of magnitude per decade. When technologies reach around 1% of world energy supply their growth becomes more slower in percentage terms (though not necessarily linear). The post-exponential rate of growth and the saturation point for each technology on its s-curve are less clearly defined.
A wide variety of energy technologies conform well to this model (see chart), because it takes a few decades to build the scale of industry necessary to provide 1% of the world’s energy. After this long replacement cycles in the energy sector (typically 20-40 years) and competition with incumbent infrastructure limit the rate of further growth.
The triangles on the chart show the actual energy consumption for 2017. The projections have held up well for solar and wind, which have both shown exponential growth, with solar growing faster than an order of magnitude per decade, as projected.
Other low carbon technologies have been much slower to develop. Nuclear has stayed roughly constant as projected. A slight decline (not visible on this scale) is more than accounted for by a fall in output in Japan following the Fukushima disaster.
Use of biofuels has increased substantially, roughly doubling over the last decade, but much less rapidly than the near order of magnitude that the projections implied.
CCS has made very little progress, and is now about an order of magnitude less than projected. It is likely to fall much further behind projections for 2025 given the current rate of project development. The extremely rapid scale-up shown in the projections now seems wildly optimistic.
Those technologies closest to conventional energy technologies (CCS, biofuels) have thus grown relatively slowly, while newer technologies (solar and wind) have followed a much faster track.
Rates of deployment of energy technologies projected from 2006. Triangles indicate 2017 actuals.
Source: Kramer,GJ and Haigh,M. No quick switch to low-carbon energy, Nature Vol462, 2009 (For some reason CCS seems to have been plotted as energy in rather than electricity out, so if the line for CCS reached the same annual energy as solar or wind it would still be generating only about a third as much electricity. Teh triangle for actual CCS is on the basis of energy out. It would be approcimately 0.3 higher on the vertical axis on an energy in basis.)
The model implies that groups of technologies still in the early phases of deployment, including CCS, concentrated solar thermal power, and geothermal will take several decades to reach very large scale. They will thus probably only be in a position to make a really large contribution to emissions abatement around the middle of this century (if at all).
This has important implications for emissions tracks consistent with limiting temperature rises to 1.5 degrees. Many of these show very large amounts of bioenergy with CCS providing net removal of carbon dioxide (“negative emissions”) after 2050. This would require renewed emphasis on early scale-up to remain achievable, though even then substantial barriers would remain.
This emphasises the importance of deployment of those technologies that are already at scale (wind and solar in particular). Continuing improvements in energy efficiency (despite rebound effects) and the use of natural gas in power generation also have an important role to play in emissions reductions pathways. And of course the sooner the scale-up of early-stage technologies such CCS begins the earlier they will be able to make a more material contribution, so starting now remains valuable.
There are couple of important caveats to this analysis. While the authors refer to the patterns as “laws” they are observed regularities rather than absolute constraints. Some technologies have particular factors associated with their deployment not captured by the model. For example, the reduced rate of growth of nuclear from the mid-1980s was driven by a particular confluence of political and economic factors, and its future growth is similarly subject to political and economic constraints in many places, although it is favoured in others. The analysis also does not go back far enough to show all of the very rapid increase in oil use in the two decades after the end of the second world war.
Solar PV has quite different supply side characteristics to other energy technologies, being much more scalable. Energy efficiency technologies also have different characteristics, as the authors of the modelling work acknowledge. Other demand side technologies such as electric vehicles seem also seem likely to be able to scale up somewhat more rapidly than these projections suggest, with major implications for the energy mix. And, while some storage technologies might take time to reach scale, lithium ion batteries seem likely to be able to continue to grow very rapidly as there production is also scalable, although there may be some supply chain constraints that may partially limit this. These imply different prospects for deployment in these cases.
In view of the time required to build scale in new technologies, few energy policies seem more important than those that encourage continuing reductions in costs and increases in the rate of deployment of technologies already at scale, including wind and solar PV, along with continuing improvements in energy efficiency.
Updated 25th October 2018
[i] Kramer,GJ and Haigh,M. No quick switch to low-carbon energy, Nature Vol462, December 2009