Hydro-Québec is advancing AI technologies in load forecasting and load management to incorporate AI into its future operations. The utility AI-powered tech for short-term forecasting, a 36-hour horizon, helps inform storm response. It produces hourly forecasts out to 10-12 days into the future based on meteorologists’ predictions. The company is leveraging historical weather data to generate longer-term outlooks, projecting load forecasts for up to 42 days. This cutting-edge strategy directly connects forecasted energy demand with measured, real-world energy demand. It trains on high stakes, inflection points when differences can lead to very expensive mistakes.
The utility’s deployment of AI is really in response to the challenges that come during unexpected and extreme weather. When an unexpected heatwave hit on May 22, 2024, Hydro-Québéco’s old forecasting models were unable to adapt. They completely underestimated a drop in load on the grid, underscoring how important advanced solutions are needed right now. Hydro-Québec in particular has been at the cutting edge of improving forecasting. This new feature will allow operators to react more quickly, avoid unnecessary disruptions, and keep the grid running smoothly.
Advancements in AI Forecasting Capabilities
Hydro-Québec’s initial foray into AI started with a proof of concept that employed a neural network at one substation. The company spent years on research before rolling out deep neural networks throughout the company’s operations. The AI models approach the old art of forecasting with an air of brashness. They are extraordinary in atypical circumstances that deviate from historical patterns.
By 2028, Hydro-Québec will implement a regional approach to load forecasting. On the initiative side, this plan will address 350+ substations. This new initiative is designed to help them harness AI more deeply across their operations, including strengthening their ability to predict energy demands days in advance. Aroon Vijaykar, Hydro-Québec’s commercial business lead, highlighted the heavy-footed approach that utilities need to be careful to avoid when they start pursuing new technologies.
“Utilities are understandably conservative about the system. Unlike the tech industry, which can afford to move fast and break things, for utilities one mistake can be catastrophic.” – Aroon Vijaykar
With every step that the organization takes towards deeper AI integration, the key tenet continues to be about improvement and learning. Hydro-Québec’s lead author of the UNECE Task Force on Digitalisation in Energy, Sylvain Clermont, noted the importance of training AI systems effectively.
“When they get confident, Hydro-Québec will stop using the old model. It is a training question; the AI needs to be trained.” – Sylvain Clermont
Addressing Challenges and Opportunities
Hydro-Québécois understanding of complexities AI creates, especially around data hygiene and complexity are evident. More importantly, it views cannabis associated fantastic opportunities to boost cost and accuracy of load forecasting. With over four million smart meters contributing data, the utility faces unique hurdles in processing and analyzing this information effectively.
Vijaykar explained the dual nature of AI’s impact on load forecasting:
“Load forecasting has always been an important function. AI is both a challenge and an opportunity on both sides of that equation.” – Aroon Vijaykar
By utilizing AI-driven orchestration alongside traditional grid planning and human operational oversight, Hydro-Québec aims to create a flexible and resilient energy management system. This rigorous approach guards against one-sided AI solutions, so even as AI improves forecasting effectiveness, it does not supplant critical human intuition and institutional knowledge.
“AI complements but does not replace grid planning and forecasting,” – Adkins
As Hydro-Québec evolves alongside an increasingly decentralized and digital energy landscape, AI integration will be key to its success. This landscape is dominated by a rapid expansion of renewable energy. This monumental shift brings new layers of variability and uncertainty that today’s forecasting models have a hard time overcoming.
“The increasing penetration of renewables introduces variability and uncertainty into grid operations, making traditional forecasting and control methods less effective.” – Adkins
Future Directions for Smart Grid Management
As for what’s next, Hydro-Québec has ambitious plans for its AI projects. In addition, the organization is committed to growing its operational AI capacity. Between 2026 and 2027, it will focus on increasing the use of smart meter data and creating a prototype for renewable energy forecasting. All of these initiatives are meant to improve overall grid efficiency and reliability—notably, still a broad goal within FERC’s Regional Transmission Organizations.
Clermont highlighted the importance of leveraging historical data to inform mathematical models used for forecasting:
“With experience, you have a lot of historic curves, so our mathematical models are quite good for regular patterns.” – Sylvain Clermont
He noted that unexpected events can challenge these models’ accuracy:
“Then comes something totally out of the box – whether extreme weather or something else that you have never experienced in the past – and your model will be off.” – Sylvain Clermont
Hydro-Québec is honing its AI implementation to improve operational efficiency. Besides improving overall reliability, this forward-looking approach avoids the potential for disastrous failures resulting from wrong load forecasts.
“At a minimum, AI can help to run systems more efficiently. At a maximum, it will avoid catastrophic blackouts.” – Aroon Vijaykar