I’m able to continue serving my customers till as late as 10pm. This extension has doubled my daily income!
Nancy Amito, a small restaurant owner in Palabekogilli Trading Centre, a largely refugee-populated village in Lamwo district in Northern Uganda
A lot of people take something as simple as flipping on a switch when dusk arrives for granted but for communities such as Nancy’s this is but a luxury. 27% of Uganda’s population has access to any kind of electricity, the number drops to 10% in rural areas. One promising remedy lies in making better use of Uganda’s existing but often untapped renewable energy sources such as solar, wind and biomass.
To provide power to unelectrified areas, the government needs to plan how best to use the available resources. In some cases, the electricity grid is to be extended, and in others there can be alternative solutions such as solar mini-grids or solar home systems. This planning can be difficult as there is limited data on exactly how large settlements are, how much commercial activity there is in each, or whether there are schools, hospitals, or administrative buildings with specific requirements. This is where data-driven technology, for example processing satellite images, can help.
To pave the way for a just energy transition, GIZ’s FAIR Forward – AI for All project partnered up with Sunbird AI and the Ugandan Ministry of Energy and Mineral Development to support the implementation of the National Electrification Strategy (NES) that aims at connecting 10 million households to the most optimal energy source by 2030. The non-profit AI firm Sunbird AI used satellite imagery and various other data sources to develop a site identification tool capable of giving out recommendations for renewable energy sites in Lamwo district. This helps public officials to better plan and implement the optimal energy supply in rural areas so that people like Nancy receive the reliable electricity access they need for lighting their homes and running their businesses.
The Lamwo Electrification App
The tool draws from Google’s Open Buildings dataset and helps to visualize building outlines. Machine learning methods then detect and sort the types of buildings, for example permanent or non-permanent buildings. Information on the layout of settlements as well as shapes and sizes of buildings provides rich insights on electricity demand. The site identification tool uses all these data sources to give out concrete recommendations for suitable types of electrification options for each village, as well as where these could be best placed.
According to Denis Ongola, an on-site electrician for Winch Energy, the solar-powered mini-grids have so far been installed in 25 villages across Lamwo. These mini-grids produce between 20 to 80 kwh worth of energy, serving between 50 to 200 households according to a village’s needs. “Unlike before, community members are now running their businesses 24/7. Some of these businesses include millet and maize grinding mills, salons, welding and retail shops. Members even commune at newly-opened sports-centers to enjoy football matches.”
Lamwo’s Chairperson Mr. Oyet Sisto stresses that citizens prefer mini-grid connections because they are more reliable, and technicians are present in the villages in case any problems arise. After a field visit to Lamwo, Sunbird’s Executive Director, Dr. Ernest Mwebaze, noted how increased electrification through mini-grids had several positive effects. More lighting provided a stronger sense of security because people no longer moved in complete darkness. Furthermore, a higher availability of mobile banking services such as Airtel and EquiDuuka eased the burden on people to travel to bank branches just to make a simple money transfer.
As we continue making concerted efforts towards a just energy transition powered by renewables, one thing is clear: There’s a good chance that we can use AI to help us find clean and sustainable energy sources and create significant economic benefit for rural communities.
Find out more about the work of FAIR Forward!