Hydro-Québec, one of the largest producers of hydroelectricity globally, is pioneering the integration of artificial intelligence (AI) into its grid management systems. This important initiative is intended to support grid reliability and avoid blackouts, particularly with increasing demand for energy. Hydro-Québec is adopting more sophisticated AI models for load forecasting. The strategy is intended to improve operational efficiency as they plan and adjust to an ever-changing energy landscape.
We realized the need for this innovation on May 22, 2024 during an extreme heat wave. That’s when the utility’s oldest legacy forecasting model couldn’t account for an atypical surge in demand. This lack of oversight required big finger-pointing course corrections from the operators, further proving the inadequacies of outdated technology. Hydro-Québec’s AI model was able to avoid this pitfall, showing the potential for more advanced predictive capabilities.
Hydro-Québec is embarking on an exciting technological path. They’re quick to point out that AI is only one tool in a much larger toolkit meant to develop flexible and resilient grid operations.
The Need for Better Load Forecasting
Load forecasting has long been a crucial function for utilities such as Hydro-Québec. Traditional forecasting methods have increasingly become ill-equipped to meet the complexities of modern energy demands. This last concern is particularly acute with the energy demands spurred by AI tech.
“Load forecasting has always been an important function. AI is both a challenge and an opportunity on both sides of that equation. It is complicating load forecasting because of the demand it generates; however, we now have this opportunity to use this new tool to do better load forecasting.” – Vijaykar
As the potential use of AI ramps up, so does the pressure on the grid. It equally is an opportunity to realize incredible efficiency savings by being much more accurate at forecasting load need. Hydro-Québec’s deep commitment to integrating AI into its operations demonstrates a visionary acknowledgment of these challenges.
During peak periods, as often witnessed during heatwaves, demand can spike above normal seasonal or weekly patterns. The legacy models Hydro-Québec had used in the past didn’t adapt well, resulting in tremendous operational chaos. One particularly ambitious model crashed and burned on May 22. This unfortunate incident emphasizes a critical need to shift from outdated solutions to smarter alternatives.
“At a minimum, AI can help to run systems more efficiently. At a maximum, it will avoid catastrophic blackouts.” – Vijaykar
Hydro-Québec’s strategy includes a bottom-up plan to have regionally relevant AI applications implemented over more than 350 substations by 2028. This strategy will help make sure that demand is forecasted and managed correctly at a very granular level.
Implementing AI Across the Grid
Hydro-Québec’s journey into AI innovation truly began in 2018. To illustrate their approach, they released a proof of concept using a very lightweight neural network model. What began as a simple load forecasting pilot for a single substation has evolved into a much more complex and far-reaching initiative.
Current AI models do not work in a linear model. They take advantage of constraint models based on the Easy Nonlinear LeastSquares Inequality Programme (ENLSIP) algorithm. These models can hold a complex mixture of functions and variables that are constantly updated in order to align with real-time data streams.
“We look at patterns and try to find a model that gives a curve to match, then we adjust parameters depending on the day, and fit them to the mathematical model until it looks right.” – Clermont
Hydro-Québec applies sophisticated modeling to produce short-term forecasts of as little as 36 hours. They can additionally forecast hourly emissions variations up to 12 days ahead by applying machine learning techniques to meteorological data. Further out forecasts though are based largely on past weather trends.
The new AI integration allows Hydro-Québec to make real-time, dynamic decisions based on future demand signals coming from the electrical grid. It rapidly recalibrates load forecasts, often with minimal advanced notice from utility operators. This maintains control of the grid, even if the world around them changes.
“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.” – Clermont
That flexibility will be absolutely essential as the energy sector undergoes unprecedented transformation driven by technology and climate change.
Challenges Ahead for Hydro-Québec
Though the AI’s capabilities seem promising, much work still lies ahead before these systems can be fully integrated into the day-to-day operations of Hydro-Québec. One significant hurdle is managing the vast amounts of data generated by over four million smart meters deployed across its service area.
“Having to deal with data from more than four million smart meters is another ball game.” – Hydro-Québec’s spokesperson
The complexity of this data necessitates continuous improvement in AI systems and collaboration with meteorologists to refine forecasting models further. Hydro-Québec will subsequently keep improving its services until 2026 and 2027. Specific attention will be paid to using smart meter data and building prototypes to support renewable energy forecasting.
“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. They need to really kick the tyres on new technological solutions.” – Vijaykar
Hydro-Québec is making significant leaps with their AI projects. The firm further emphasizes that artificial intelligence and machine learning are complementary to existing grid planning and forecasting techniques—not replacements.
“AI complements but does not replace grid planning and forecasting. AI-driven orchestration works best alongside grid forecasting, market signals and human operational oversight.” – Adkins
This multi-faceted approach goes a long way in building grid management resilience. In doing so, it addresses the growing uncertainty of energy needs fueled by renewables.