A machine-learning approach to predicting Africa’s electricity mix based on planned power plants and their chances of success

Energy scenarios, relying on wide-ranging assumptions about the future, do not always adequately reflect the lock-in risks caused by planned power-generation projects and the uncertainty around their chances of realization. In this study we built a machine-learning model that demonstrates high accuracy in predicting power-generation project failure and success using the largest dataset on historic and planned power plants available for Africa, combined with country-level characteristics. We found that the most relevant factors for successful commissioning of past projects are at plant level: capacity, fuel, ownership and connection type. We applied the trained model to predict the realization of the current project pipeline. Contrary to rapid transition scenarios, our results show that the share of non-hydro renewables in electricity generation is likely to remain below 10% in 2030, despite total generation more than doubling. These findings point to high carbon lock-in risks for Africa, unless a rapid decarbonization shock occurs leading to large-scale cancellation of the fossil fuel plants currently in the pipeline. Energy scenarios can sometimes miss lock-in caused by planned power plant projects and uncertainty around their realization. Here, Alova et al. build a machine-learning model that predicts Africa’s electricity generation mix in 2030 based on the commissioning chances of planned projects.