A simple s-curve model of solar deployment shows continued strong growth.
In my previous post I looked at the IEA’s projections for solar PV. These always project no growth (or even a reduction) in the rate of installation, whereas in practice the rate of installation keeps growing rapidly. I commented in the post that the growth of solar deployment did not seem likely to stop any time soon. So how fast might solar PV continue to grow?
To estimate how fast solar PV deployment will grow, I’ve adopted a simple logistic function (s-curve) model for the deployment of solar. This type of function is widely used to model the growth of new technologies[i]. The results for two scenarios are shown in the chart below together with actual annual deployment to date. Both scenarios fit the historical data well, and are similar for the next few years, but then diverge significantly.
Scenarios for deployment of solar PV
Source for historical data: BP statistical review of world Energy to 2017, estimate for 2018 based on data in previous post.
The low case is based on an electricity system continuing to grow at current rates, with solar taking an increasing share, and deployment eventually reaching 300GW p.a. (see notes below for more on this). The base case assumes a larger role for the power sector in the energy mix, as decarbonisation drives the electrification of end use, and solar deployment eventually reaches 50% more than in the low case, at 450 GW p.a..
These projections show deployment in another 4 to 6 years reaching more than double its 2018 rate of just over 100GW. This compares with the 3 years it took to double from 50GW to its present size. By 2030 solar is generating 3600 to 4500TWh p.a., around 12-15% of electricity consumption[ii].
Of course this highly stylised analysis only gives an indication of scale, and even greater growth is possible. However I have not included a higher scenario, as these scenarios already represent continued very rapid growth. This will require continuing attention to how solar can best be integrated into wider energy systems, including through the greater use of battery storage.
Adam Whitmore – 6th February 2019
Notes: Developing indicative markers for eventual industry size
A low case is estimated by looking at the size of the power sector. This requires (in very approximate numbers) about 1000TWh of new and replacement generation each year over the next couple of decades. If a third of this were to be solar it would eventually grow to about 330TWh p.a. of this, or about 300GW p.a.. It seems unlikely that solar’s share of new capacity would in the long run be less than this given its cost competitiveness and scalability.
This scenario appears roughly in line with Shell’s Sky scenario. Both suggest that by 2035 Solar PV generation will be a factor of a little over 20 higher than in 2015.
However, this eventual rate of deployment may be an underestimate. Decarbonising the energy system will require widespread electrification of end use, and so much more of the world’s energy will come from low carbon electricity. For this analysis I’ve chosen a figure for eventual installation rate 50% greater than in the low case, reaching of 450GW p.a.. This represents one possibility within the range of scenarios for more ambitious decarbonisation, and higher estimates are possible.
[i] A logistic function is often used to model deployment of new technologies based on a range of examples, and I’ve previously used this type of model to look at electric vehicle growth – see here including examples of previous technology transitions. The analysis presented here updates my previous analysis of solar in both data and approach, given the additional data available since that was completed.
[ii] World electricity consumption was 21,000TWh in 2015, https://www.statista.com/statistics/280704/world-power-consumption/ growing at 2.6% p.a. over 2010 to 2015. Assuming this growth rate is maintained electricity consumption will reach around 31,000TWh by 2030. BP’s review of energy suggests a lower growth rate, with around 2000TWh less demand in 2030 than in the case used here, presumably reflecting greater efficiency.