Quebec's winters are notoriously long, cold and energy intensive. And even though Hydro-Québec provides abundant, relatively inexpensive electricity, waste and efficiency remain serious concerns.
A new Concordia study clarifies energy consumption patterns by looking at neighbourhood-level usage in Quebec's major urban centres and socio-demographic data derived from Canada's 2021 census, such as employment and income levels, car ownership, average age and household size. The researchers say this method could help utility managers, policy planners and government authorities better understand which demographics consume the most household energy and where usage is at its peak. The data could also be used in devising more effective and equitable energy strategies.
Hydro-Québec provided the researchers with hour-by-hour usage data from residential smart meters for the full four years between 2019 and 2023. The team then broke down that data for Montreal, Trois-Rivières and Quebec City according to the Forward Sortation Areas — Canada Post's geographical subdivisions used to sort and deliver mail as efficiently as possible.
Using sophisticated models and advanced machine learning tools, the researchers determined which variables drove energy consumption the most across both long-term heating patterns and short-term daily use.
"Energy demand is not just about the building, but about the people who reside in it," says lead author Masood Shamsaiee , a PhD student at the Next-Generation Cities Institute . "You can have two neighbourhoods with similar building profiles, but if two different types of people live in them, they are going to have two completely different consumption patterns."
The study was published in the journal Energy and Buildings . It was co-authored by Ursula Eicker , a professor in the Department of Building, Civil and Environmental Engineering .
Age, income, employment as key drivers
To observe long-term patterns, the researchers used a method called change-point analysis to determine when heating systems were activated as outdoor temperatures changed. They then applied a machine-learning model to measure how different socio-demographic factors influenced heating behaviours.
For short-term patterns, the team grouped daily electricity-use profiles into clusters and used another machine-learning model to determine which social characteristics best predicted each pattern. AI analyses were used so the models could reveal not just predictions, but also which factors mattered most.
The analysis revealed strong links between energy use and social characteristics. Higher-income neighbourhoods and areas with larger households tended to have higher baseline electricity use and stronger increases in consumption as heating demand rose. Lower-income neighbourhoods, however, often activated heating earlier in the season, which may reflect less efficient buildings due to poorer insulation and older windows.
Neighbourhoods with older populations tended to have higher electricity use per person, likely because residents spend more time at home and prioritize indoor comfort. Areas with a higher population of non-Canadians, newer homes, high-rise apartments, younger residents and crowded living conditions averaged lower use.
Daily routines also played a role in consumption demands. Areas with high employment levels and car-dependent lifestyles showed strong peaks in the morning and evening, when residents left for and returned from work. Neighbourhoods with higher unemployment or more walkable environments tended to have flatter electricity-use patterns throughout the day.
"I hope this study adds a human element to the different models informing policy makers," says Shamsaiee. "This can be used as a tool to help utility companies like Hydro-Québec develop better, more detailed efficiency programs while at the same time delivering a fairer and more equitable distribution system."
Read the cited paper: " Socio-Demographic insights on urban building energy consumption "
