Learning Curve Effect on the Global Variable Renewable Energy Deployment

Based on the learning curve effect, renewable energy technologies enter a virtuous cycle

If you improve by 1% every day, you will have improved by 37 times by the end of the year.

The first time I read the quote above, I was awestruck by it. Indeed, learning is a slow but steady process. When someone sets off on a journey of learning a new skill, it is normal that the person experiences relatively slow growth towards the beginning. Once the person gets a good grasp of the basics, the growth is quite steep. And if the person continues to put forth the effort, in the long run, they can gain mastery over the skill. This is the gist of the quote above, which is depicted quite clearly in the plot below.

Illustration of knowledge gain over time through learning. Image by Author.

Learning Curve

In reality, the gain of knowledge or proficiency in a particular domain is not a smooth process. There are several challenges along the way. The curve that resembles the learning journey from a holistic perspective is known as the learning curve. It is a graphical representation of the gain of proficiency or expertise in a domain over the time invested.

The shape of the learning curve could vary depending upon the context. It can be linear, s-curve, or even exponential. The s-curve (sigmoid function) is the generalised form of the learning curve, with slowly accumulating small steps towards the beginning. This is followed by larger steps and then subsequently smaller steps again as learning reaches towards the mature phase.

Illustration of the learning curve. Image by Author.

Experience Curve

The concept of the learning curve is also applicable in economics. Every product in a market goes through four main stages in its lifecycle:

  • Introduction: when the product is launched in the market,
  • Growth: when sales start to increase,
  • Maturity: when the product sales reach their peak,
  • Decline: the product gets old, and a new emerging technology replaces it.

The growth in sales or production volume of a product in its lifecycle is analogous to the knowledge gain in the learning process. When the production volume of a product increases, then the experience related to it also increases leading to efficiency gain over time. Thus, the increase in cumulative production volume leads to a decrease in the cost per unit of the product. This effect is known as the learning curve effect or the experience curve effect.

Oftentimes, the terminologies learning curve and experience curve are used interchangeably depending upon the context.

Illustration of the learning process and the experience curve effect. Image by Author.

Case study of learning curve effect on the variable renewable energy technology deployment

The deployment of renewable energy technologies, particularly Variable Renewable Energy (VRE), grew significantly in the last decade. The energy technologies such as wind and solar power are referred to as VRE because of the properties such as intermittency and uncertainty associated with them. In the following sections, I am going to discuss the trend, drivers, and the effect of the learning curve on VRE deployment from 2010 to 2020.

Global trend in renewable energy

According to International Renewable Energy Agency (IRENA), the total installed capacity of renewable energy technologies more than doubled from approximately 1300 GW in 2010 to nearly 2900 GW in 2020 (IRENA, 2021a). While renewable energy technologies such as hydropower and biomass experienced moderate growth in the last decade, the installation of VRE technologies was exponential over the same period.

Total installed capacity of renewable energy technologies in the world from 2010 to 2020. Image by Author.

Global trend in variable renewable energy

In the last decade, the technology which was deployed the most on the global scale was solar PV. The installed capacity of solar PV grew nearly 18-fold from 40 GW in 2010 to 710 GW in 2020. This was followed by the onshore wind technology of which the installed capacity grew nearly four-fold from 178 GW in 2010 to nearly 700 GW in 2020. While the installed capacity of offshore wind and concentrated solar power (CSP) technologies also grew during this period, the total installed capacity of these technologies remain relatively small. Overall, the installed capacity of VRE increased nearly six and half times from 222 GW in 2010 to 1448 GW in 2020 as shown in the plot below.

Global installed capacity of VRE comparison between 2010 and 2020. Image by Author.

The name of the dataframe used is abc, which looks as in the image below. The dataframe used to get the plot above is available in the gist below.

Dataframe containing total installed capacity of VRE technologies in 2010 and 2020. Image by Author.

Decline in the cost of VRE technologies

The need to decarbonise the energy mix to contribute to climate change mitigation is certainly one of the main drivers of the increased deployment of VRE technologies in the last decade. However, a market is always driven by the cost of the product. Whenever there are multiple options available in the market that meet our objectives, we tend to opt for the most cost-effective one.

The capital cost of VRE technologies fell steadily in the last decade and they continue to drop further today (IRENA, 2021b). Between 2010 and 2020, the capital cost of solar PV decreased by more than five times from 4731 USD/kW to 883 USD/kW. Over the same period, the capital cost of onshore wind declined from 1971 USD/kW to 1349 USD/kW.

Capital cost of installing VRE technologies from 2010 to 2020. Image by Author.

With the plummeting capital costs of these technologies, the cost of generating electricity from these technologies (referred to as Levelised Cost of Electricity (LCOE)) also declined sharply over the years. As a result, VRE technologies have become increasingly cost-competitive with fossil-fuel technologies in recent years. According to IRENA, the global weighted-average LCOE from new capacity addition of solar PV, onshore wind, offshore wind, and CSP technologies declined by 85%, 56%, 48%, and 68% respectively between 2010 and 2020 (IRENA, 2021b).

Global weighted-average utility-scale LCOE of VRE technologies from 2010 to 2020. Image by Author.

The rapid decline in the capital costs and LCOE of renewable energy technologies as compared to non-renewables can be attributed to the fact that renewable energy technologies follow a learning curve, while non-renewables do not. This is because, while the LCOE of fossil fuels and nuclear power depend largely on fuel price and the operating costs, renewable energy technologies do not have fuel costs, have relatively low operating costs, and hence, the LCOE of renewable energy technologies depend largely on the cost of the technology itself.

The principle of the learning curve is that,

“with each doubling of the cumulative installed capacity, the cost declines by a certain average percentage called the learning rate”.

This relation is shown in the mathematical equations below:

Equation showing the relationship between capital costs and cumulative installed capacity at start and end time based on learning parameter. Image by Author.

The learning rate is calculated with the following equation:

Equation showing the relationship between learning rate and progress ratio. Image by Author.

Based on the learning curve effect, renewable energy technologies enter a virtuous cycle. For example, the deployment of solar PV increases to meet the growing electricity demand in a region. With the increase in deployment, the unit cost of installation plummets as a result of mass production i.e., economies of scale, technology improvement, efficiency gain, and competition in the supply chain. Thus, solar PV becomes more cost-competitive in the market, which in turn increases further demand.

Renewable energy technologies become cheaper with increase in cumulative installation and enter a virtuous cycle. Image by Author.

Based on the capital costs and cumulative investment data of VRE technologies, solar PV had a learning rate of 33% while the onshore wind had a learning rate of 18% between 2010 and 2020. This implies that, with each doubling of installed capacity, the unit capital cost of solar PV declined by 33% and that of onshore wind by 18% on average.

Capital costs and cumulative installed capacities of solar PV and onshore wind between 2010 and 2020. Image by Author.

Among all the VRE technologies, solar PV had the highest learning rate (33%) followed by CSP (25%), onshore wind (17%), and offshore wind (10%). This is evident from the steepness of the lines when both the variables are plotted on a logarithmic scale.

If the learning rate of solar PV was only 10% between 2010 and 2020, the capital cost would have declined to only 3055 USD/kW in 2020 from 4731 USD/kW in 2010. If the learning rate was 40%, the capital cost would have declined even further to 567 USD/kW in 2020. This is demonstrated in the plot below.

Capital cost projections for solar PV between 2010 and 2020 with different learning rates. Image by Author.

Conclusion

The traditional electricity market in the world was dominated by fossil fuel technologies. Today renewable energy technologies, particularly VRE, are cheaper than fossil fuels in most countries of the world. The large-scale deployment of solar and wind generation in the past decade has led to a paradigm shift in the power system and electricity markets. How the deployment of VRE and other renewable energy technologies changes the dynamics of merit order and the marginal cost of electricity generation is present in this story.

In this article, I discussed the trend of renewable energy and VRE from a global perspective. I discussed further the concept of the learning curve, why renewable energy technologies follow the learning curve, and how the curve leads the technologies to enter a virtuous cycle of decline in costs and increase in deployment. The notebook and data for analysis in this article are available in this GitHub repository. Thank you for reading!

References

IRENA, 2021a. Trends in renewable energy. Statistics Time Series (irena.org)

IRENA, 2021b. Renewable Power Generation Costs in 2020 (irena.org)

OWID, 2020. Why did renewables become so cheap so fast? — Our World in Data

--

--

--

Your home for data science. A Medium publication sharing concepts, ideas and codes.

Recommended from Medium

Conservancy, FWC team up to rescue hanging pelican on Christmas Eve

Julia Louis-Dreyfus, Zazie Beetz, Ron Howard, Eva Longoria, Winston Duke, Alec Baldwin and more at…

Solar Villages and Farms

Stay Cool All Summer Long

How could rising floodwaters impact your home’s value?

The killing of public joy

We Choose To Reuse & Recycle What We Can For Our Home Garden

We Choose To Reuse & Recycle What We Can For Our Home Garden

Our Planet will Parish if We Don’t Take Action

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Himalaya Bir Shrestha

Himalaya Bir Shrestha

I write about the intersection of data science with energy, climate change, and sustainability in simple words.

More from Medium

Europe’s Most Competitive Football League — A Data Scientist’s Answer

Using Data Science to Develop a Winning Business Strategy, Part 3

How AI can help smart city initiatives

Biased Model Coefficients — (Part 1)