This article first appeared in February’s print edition of Business Monthly.
In 2023, power outages for two hours were, for the first time, a fact of life in Egypt beyond the summer months when air conditioner use drops. In December, Asharq Business, Bloomberg’s Arabic news portal, reported that load shedding would continue until March due to a decrease in fuel for generators.
Integrating artificial intelligence (AI) into the national power grid could significantly reduce those blackouts. “AI holds promise … for optimizing energy generation, distribution and consumption,” FDM Group, a UK-based consultancy, said in September,
AI is also increasingly essential as “power systems become vastly more complex as demand for electricity grows and decarbonization efforts ramp up,” said Vida Rozite, a policy analyst with the International Energy Agency (IEA), in November. “The result is a vastly greater need for information exchange — and more powerful tools to plan and operate power systems as they keep evolving.”
Such integration will be challenging, as governments need to ensure digital, legislative and human capabilities can cope with the significant influx of sensitive data generated from integrating AI into the power grid.
Getting it right means “increasingly sophisticated AI-driven solutions [will] improve the efficiency of … energy sources, enhance grid stability and reduce greenhouse emissions,” said FDM Group.
AI grid
AI technologies in smart grids “predict consumption patterns using historical and real-time data, which can help allocate resources more efficiently,” FDM Group noted. That is particularly useful during “sudden periods of high demand. AI can improve the distribution of electricity, ensuring that power is directed where it’s needed most” and lessen the chance of blackouts.
Such AI systems also could also predict equipment failures. “Machine learning can analyze large amounts of data … such as usage stats, weather and historical maintenance records to predict breakdowns,” said FDM Group, “This approach minimizes downtime, reduces repair costs and improves the overall reliability of [the] energy infrastructure.”
Using AI in power grids means “automating manual inspections of millions of power lines, poles and mounted devices,” Marc Spieler, senior managing director at NVIDIA, a producer of computer hardware, said in a July blog.
Implementation should be straightforward. Yehu Ofer, CEO of Odysight, a developer of AI-powered tools, told Reuters in December that utility companies need only fit optical sensors connected to a computer with the AI algorithm at critical points of failure to “see issues before they become problems. I can change the model of maintenance by not replacing things before they need to be replaced.”
Demand prediction
Using AI-powered systems to predict peak demand patterns and adjust supply significantly benefits governments. “Many traditional approaches to forecasting [electricity demand] only can look at a small number of parameters with small datasheets,” Karen Panetta, a fellow at the Institute of Electrical and Electronics Engineers, told Reuters in December. “AI can allow us to explore relationships and … redistribute energy most efficiently.”
That would ultimately lower operational costs and, more importantly, “help make the shift toward renewable energy sources,” said FDM Group.
Another reason for making AI forecasts necessary is that power grids now use fossil fuels alongside renewable sources. The Egyptian government plans to generate 42% of the country’s electricity from renewables by 2035, up from 22% in 2022. “With renewables, we have two variables,” Sherif El-Mashad, digital lead at ABB Electrification, told Reuters in December. “We will need to have precise predictions, but at the same time, we need to have orchestration between all the different elements at play.”
Sustainable supply
Ensuring a stable energy supply from renewable sources while minimizing dependency on non-renewables is another perk of using AI in power grids. “This can be complicated … since the sun doesn’t always shine, and wind doesn’t always blow,” noted Rozite of the IEA. “That’s where machine learning can play a role, [matching] variable supply with rising and falling demand … allowing it to be integrated more easily into the grid.”
Kristjan Jansons, CEO of MindTitan, an AI startup, told Reuters in December, “AI can help select sites where wind or solar facilities could be installed.” The result would be more scalability of clean energy output that is cost-effective.
FDM Group said AI could determine the best times to store renewable energy, when to release it, and how much to distribute. It would be particularly crucial for hospitals, data centers, and emergency services.
AI also is “profoundly” transforming the oil and gas exploration sector, said FDM Group. “AI can identify potential oil and gas reserves that may have gone unnoticed using traditional methods [and] guiding exploration toward the most promising prospects.” That is vital for Egypt, as nearly 75% of its electricity supply still relies on fossil fuels.
Egypt also could benefit from AI at its Dabaa nuclear power plant. AI’s “role in nuclear power plants is indispensable, as it … helps prevent accidents while maintaining the reliable generation of clean energy,” said FDM Group.
AI headwinds
FDM Group said the biggest challenge facing governments is the cost of implementing AI into existing infrastructure.
The second is that AI systems used in utilities would have access to “vast amounts of sensitive data, including grid information, customer data, and operational details,” FDM Group said. “Ensuring the security of this data is paramount … Compliance with data privacy regulations … adds an extra layer of complexity.”
The third issue facing the government is a shortage of AI professionals who also understand the energy sector, said FDM Group. “This scarcity … can slow down the adoption and development of AI solutions in the industry.”
Another problem is that AI-powered electricity grids require fewer workers, a concern for emerging economies seeking to create jobs. Manoj Sinha, CEO of Husk Power Systems, an Indian mini-grid operator, told Reuters in December that since adopting AI, his company has hired 1,000 workers instead of about 3,500.
Jansons of MindTitan added that AI isn’t suitable for every situation a power grid faces. “It needs to be used on [parts] maintained often enough so that small tweaks … enable big gains. Wind turbines are suited to predictive maintenance because they are standardized.”
Another problem is integrating AI into existing power grids requires significant lead time. “The energy sector is very stable,” El-Mashad of ABB Electrification told Reuters. “Companies need to be fully confident that AI-based technologies are 100% reliable and robust before they deploy them … The worst-case scenario, where lights go off as the result of a mutant [AI] algorithm, is certainly to be avoided at all costs.”
Rozite also noted that AI uses “more energy than other forms of computing — a crucial consideration as the world seeks to build a more efficient [decarbonized] energy system.” He estimated that “training a single [AI] model uses more electricity than 100 US homes consume in an entire year.”
To overcome those headwinds, governments need to agree on developing “mechanisms for data sharing and governance,” Rozite said. “A coordinated global approach can enable internationally applicable and replicable solutions, transfer learning globally, and expedite the energy transition while reducing costs.”
