MIT-Comillas’ Reference Electrification Model (REM) determines the most cost-effective strategy to provide energy access in areas that lack it. We are collaborating with the MIT-Comillas Universal Energy Access Group [1] to build modern mapping platforms that put this data in the hands of policymakers, energy planners, and entrepreneurs to better connect rural communities.

The basic input data for REM are fuel costs, solar irradiance, grid extent and reliability, consumer demand and infrastructure costs. It then uses a deep learning-based computer vision system that the MIT-Comillas team developed to locate buildings in satellite imagery. With this infrastructure information and geodata, REM generates detailed engineering designs, providing recommendations on the best ways to connect rural communities through a combination of grid extensions, microgrids, and stand-alone systems.

We are working with the MIT-Comillas team to make these unique and valuable insights more accessible. Lightweight web maps can easily get these localized plans to regional power authorities and development organizations in the field. Further opening the data can show where off-grid approaches are viable intermediary solutions for electrification, motivate essential policy decisions, and encourage off-grid energy providers to work in the places they can have the greatest impact.

Example REM outputs for the Vaishali district in Bihar, India.

REM enables users to perform sensitivity analyses to external factors like demand level, grid reliability, fuel and technology cost, and cost of non-served-energy. In this example, you can see how the model projections change with the price of diesel. As diesel prices rise, it becomes more and more cost efficient to connect people to the existing grid (this is because many of the solar-and battery-powered microgrids also use diesel for backup power). But even with high diesel prices, there are some areas that are better served through off-grid power models. We built this map using Mapbox GL. Mapbox GL uses your computer or phone’s powerful graphics card to quickly process huge amounts of data. This enables us to take complex model results and render them on the fly into a highly detailed, interactive web map based on input from the user. The extremely efficient vector tiles that Mapbox GL uses for geodata makes the tool usable even on slower Internet connections.

1.1 billion people still lack access to electricity and frequently use dirty kerosene lamps for lighting. Nearly 3 billion rely on harmful biomass fuels for cooking, and these activities contribute to air pollution that kills millions of people per year. Expedited electrification will decrease risks of respiratory disease, improve income generation prospects, and enable children to study longer.

Immediately extending the existing grid isn’t always the right answer. Depending on complex geographic and economic factors, it may be more cost-effective and sustainable to build microgrids and stand-alone systems instead. If the grid’s reach eventually extends to these areas, connecting grid-compatible off-grid solutions may encourage infrastructure reuse and promote a cleaner generation mix as well. Satellite imagery, machine learning, powerful modeling tools, and modern mapping technology can empower infrastructure planners to design off-grid electricity solutions that are viable, cost-effective, and can better serve communities.

Acknowledgements

Data preparation resources for the REM runs shown were supported by the MIT Tata Center for Technology and Design.

1. The MIT-Comillas Universal Energy Access Research group currently works with policymakers, NGOs, and private companies in India, Rwanda, and Uganda to explore paths towards universal electricity access. Given the multivariate objectives inherent in sustainable development, the group is investigating how regulation and advanced planning tools have the potential to enable game-changing technologies and business models for electrification.