New tools to reduce the burden of electricity access planning
Around 600 million people in Africa still lack access to electricity. Despite recent progress, electrification efforts face new headwinds since the Covid-19 pandemic, with a growing debt crisis, poor utility financial health, and increased affordability challenges. However, advances in off-grid solutions, in particular solar- and battery-based technologies, with new business models are filling a growing gap in access provision by grid extensions. Based on new IEA data and analysis, in sub-Saharan Africa, off-grid systems accounted for over half of new connections in 2022. Still, closing the access gap requires greater scaling, which today is hindered by traditional planning and customer acquisition approaches, which often relies on workers going village-by-village to assess to the current electrification and energy needs at the community level.
The International Energy Agency, alongside researchers at Massachusetts Institute of Technology, University of Massachusetts Amherst, and Electricity Growth and Use In Developing Economies, developed a model which can address this gap. By mapping which buildings are likely to already have access to electricity today and which do not, the tool also estimates the current or anticipated electricity needs for every building in a country. This open-source model uses satellite images and available footprints of all buildings across Africa, then pairs that with the utility meter data on electricity consumption matched to the geolocation of that building or community. The model then uses Artificial Intelligence (AI) algorithms to learn from high-resolution images of buildings, identifying the patterns that best correlate their image and location to a certain level of energy demand. While impossible to ascertain exactly, the algorithms likely pick up on details that signify whether the building is urban or rural, residential or commercial, and if the community is connected by major roads to markets which could signify relatively high income and ability to pay. When tested, the model was able to identify which buildings have electricity today with over 80% accuracy and provides a 40% error reduction when estimating electricity demand of buildings over the state-of-the-art tools commonly used today.
This tool can, accordingly, be applied to satellite images of entire countries, and produce a significantly improved estimate for planners, utilities, and off-grid solar companies to identify target customers and communities. This significantly reduces the need for extensive on-the-ground surveys, customer acquisition costs and, if trained on a representative sample of communities, can adapt its estimates to the local context, cultures, climate, and other factors on a larger scale. The below schematic gives an overview of the model’s approach and is followed by examples of the new Open Energy Maps tool, released today by the IEA and MIT, which includes maps in Ghana, Senegal, and Uganda, with estimates for electricity demand and electrification stats for all identified buildings in these regions.
Building-level access status and electricity demand estimation model scheme
Identifying unserved buildings and estimating their demand by satellite
Current estimated likelihood of electricity access at building level, selected location
Most regions with an electricity access deficit still lack data from utilities on which buildings and communities are electrified or not. The de facto proxy is utilising nighttime lights from satellite images as an indicator of electrification, which is a common practice for utilities and off-grid developers. While this approach is largely dependent on streetlighting and may miss the increasing number of houses with small off-grid systems, it remains one of the strongest indicators of whether a community is electrified or not. Over the years, the dataset has been enhanced, controlling for other non-electric sources of light like fires and variances over time, such as power outages.
The model builds on the night-time lights methodology, but uses high-resolution images of buildings, their surroundings, and other geolocated datasets such as internet speeds to give a much more granular assessment. Right now, the night-time lights approach gives an assessment of access for each 1 km squared, where the model predicts this at the individual building-level with over 80% accuracy. This approach can help identify settlements where the urban core is electrified, but buildings just a few hundred meters outside of town may remain unelectrified or help identify areas of urban infill and informal settlements which may remain unconnected or underserved. Many of these buildings are among the most interesting customers for utilities and off-grid companies to target, as the cost to reach these customers is low, and their ability to pay may increase faster than other less-connected regions.
Estimated future electricity demand for unserved buildings, selected location
The model also estimates the likely electricity demand of each building, for both buildings with electricity access already and those without. This estimate relies on training the model on high-resolution images of buildings that have been matched with real utility data for that specific building. Perhaps unsurprisingly, many of the buildings that the model identified as higher users of electricity were large industrial or commercial footprints or multi-story buildings. Similarly, buildings in denser, urban areas were correlated with higher-consuming buildings. Interestingly, similar geospatial models which use satellite images to estimate income per capita in different communities had strong overlap with the buildings identified by this model as higher users of electricity.
By combining this model with advanced geographic information systems (GIS)-based least-cost electrification tools, planners can quickly conduct highly precise pre-feasibility studies for grid extension, mini grids, and standalone systems. It also helps companies better target communities and buildings that are likely to have a higher ability to pay, using these communities ideally as anchors for providing additional access to areas that may face a steeper affordability challenge. Additionally, this kind of dataset can help inform other portions of project preparation, such as spatial planning for distribution grids, generators and substations. It can also improve sizing of power systems—especially off-grid solutions— and facilitate more accurate financial forecasts.
An affordability gap revealed
When applied to electrified areas in pilot countries the model revealed the stark disparities between actual electricity consumption and expectations. Electricity consumption in many buildings remained much lower than benchmarks commonly used in the energy space to describe basic levels of household electricity usage, likely due to challenges with affordability of energy and appliances, but also complicated by low reliability of power. Our analysis across Ghana, Senegal, and Uganda revealed that a considerable portion of the population, despite having access to electricity, consumed less than the IEA’s basic and extended bundles of energy consumption. This corresponds to the lower tiers of the Multi-Tier Framework for energy access.
Share of buildings per country per electricity access level, 2022
At the household level, this low consumption suggests that, even where access is available, electricity may not be affordable or reliable enough to support essential activities such as lighting, cooking, and powering devices critical for education and home-based businesses. For off-grid businesses and utilities, this can make households unattractive to connect without greater incentives from governments or other financing mechanisms like results-based finance. However, the disparity though between different regions is also stark, indicating the importance of data catered to reflect the local context, and to update regularly to reflect how changes in tariffs may substantially change the energy demand for each household. The IEA and its partners remain available to engage with other countries and utilities to extend these models to new regions—making a critical public good available to utilities and off-grid developers alike in hopes of lowering the time and costs required to electrify new communities.
Methodological note
The data presented in this commentary has been collected by the IEA from government agencies, existing surveys, and partners. Developed in collaboration with Massachusetts Institute of Technology and University of Massachusetts Amherst and supported by Power Africa, the IEA introduces an open-source building-level electricity access and demand estimation model. This tool aids planners in accurately estimating future energy requirements for sub-Saharan structures. The openly accessible datasets, presently encompassing select countries, comprise polygon vector files.
This initiative, together with the Africa GIS catalogue for Energy Planning is central to the IEA’s partnerships, aiming to support GIS expertise across the continent through dedicated capacity-building and cooperative efforts. It seeks to empower countries with the ability to utilise GIS technology for developing and implementing customised energy access strategies, tailored to meet national challenges. These concerted efforts are crucial in refining energy access initiatives and maximising the use of GIS tools in achieving Sustainable Development Goal 7 (SDG7).
We invite countries interested in this analysis to collaborate towards bridging energy access gaps, aiming for universal coverage. For collaboration, featuring your model or dataset in the Africa GIS Catalogue, or model customisation discussions, contact us at gis@iea.org.