Artificial intelligence’s (AI’s) potential to be a game-changer in reducing load-shedding is being undermined by a lack of investment and the right skills, as South Africans are experiencing a harsh cold snap, and there is no guarantee that Eskom can keep the lights on.
While AI offers promising tools to improve grid efficiency and reliability, these benefits are negated by systemic challenges that have long plagued the country’s electricity infrastructure.
The adverse effects of load-shedding are severe. A South African Reserve Bank bulletin has estimated that load-shedding reduces GDP growth by between 0.7 and 3.2 percentage points.
Independent energy expert Chris Yelland told ITWeb that AI can improve efficiency and response times in the electricity ecosystem, helping to manage outages more effectively.
For example, in Johannesburg, outages are currently logged via an app with City Power instead of digital tools being used to help with faster incident management. “AI has the potential to be a game-changer,” Yelland says.
Renewable energy company GoSolr supports this view, highlighting that “massive strides in technology and innovation are driving us forward,” especially in solar energy.
A GoSolr “Light Paper” points out how AI can enhance the accuracy of electricity devices and boost power source efficiency. “We see shining beacons of light where some countries are really getting it right, so we know it can be done,” the report states.
AI can help predict faults and manage spare parts inventory more effectively, Yelland adds. “It can help you with keeping sufficient spares and making sure that you know clearly what you have to order for spares.”
Vladimir Milovanovic, VP of power systems for Anglophone Africa at Schneider Electric, explains that AI, combined with machine learning, enables a predictive, proactive approach. This optimises limited field resources, reduces outages, improves operational efficiency and boosts customer satisfaction.
Academic research backs this up. A 2022 thesis by Nomea Shakoane at the University of the Witwatersrand identifies AI solutions, such as predictive maintenance, fault detection and power demand monitoring, as crucial priorities for Eskom.
However, AI’s promise is hollow without addressing deeper issues. Yelland says: “If you don’t have the money to buy a spare [part], AI doesn’t solve underlying problems that have been festering for years.”
Chronic underinvestment, misallocated budgets and a shortage of skilled personnel blunt the impact of technological advances, he notes.
GoSolr’s report echoes this, warning that, while technology and innovation exist, South Africa is “not utilising it or moving as fast as we can”.
CEO Andrew Middleton warns that “our energy crisis is still apparent. We can’t rely on Eskom to solve it – the utility is slow-moving and doesn’t have the resources.”
Yelland says Eskom doesn’t have enough reserves should there be a generating unit or grid failure, like thieves stealing pylons. The most recent Statistics South Africa data on power output shows that Eskom’s electricity generation remained flat year-on-year in April, when Johannesburg weather lows were four degrees warmer than now.
The recent resurgence of load-shedding, including stage six outages in March and unexpected returns last month, also shows that outages can happen without warning.
Yelland cautions against overconfidence in predicting breakdowns. “These are unplanned breakdowns, and they are kind of random; they can happen at any time.”
More worrying is the state of municipal infrastructure. “Right now, municipalities are a greater source of outages than Eskom generation capacity shortages.”
Share