An innovative approach for the optimal sizing of mini-grids in rural areas integrating the demand, the supply, and the grid

On the path to universal electricity access, the most significant challenge lies in the electrification of remote rural villages where connection to the national grid is a durable but prohibitively expensive solution. With decreasing costs of renewable technologies, autonomous mini-grids combined with solar home systems constitute today an economically affordable and robust electrification option. In this paper, we elaborate a novel methodology that automatically designs and estimates the cost of the optimal mini-grid to install in a village, that requires only some geographical information with a delimitation of its roofs. In a first step, we use machine-learning algorithms to predict the demand of each building. In a second step, we develop a mathematical optimization approach to best design the mini-grid where generation and storage assets as well as the reticulation of the grid are jointly optimized. Our methodology has many advantages: first, by automating the full process, the calculation time of the electrification cost is drastically reduced by many orders of magnitude and the methodology can easily be deployed to any village/region. Second, our approach can reduce investment costs by more than 20% when compared to existing benchmarks. Finally, it can help agencies to efficiently assess the electrification costs of many regions and support them in the energy access planning.