Layering AI into HVAC Systems Shows Reduction in Carbon Emissions

Heating and cooling systems are just one of the many new ways that AI can be integrated into schools. According to a new study from Schneider Electric's Sustainability Research Institute, AI-powered HVAC systems in schools can lead to significant carbon emissions savings. The research found that Swedish schools using AI-powered HVAC could reduce their emissions by about 60 times the amount of carbon produced during the AI system's deployment process. A comparative analysis between Stockholm, Sweden, and Boston, Mass., showed that implementing the AI-powered HVAC solution in Boston could yield seven times the carbon emission savings as Stockholm.

Below is a Q&A with Rémi Paccou, Director of Sustainability Research at Schneider's Research Institute and the study's author. He participated in an email Q&A with Spaces4Learning to discuss the study's background and some of its major findings.

At the broadest level, what are some of the benefits of integrating AI into HVAC systems?

AI-powered HVAC systems offer significant advantages through dynamic optimization beyond conventional fixed schedules. These systems demonstrate remarkably accurate indoor climate forecasting capabilities with reduced computational demands while achieving substantial energy reductions through techniques like Deep Reinforcement Learning. They provide dual benefits of enhanced indoor climate control and energy savings by continuously analyzing multiple variables affecting comfort with unprecedented precision. AI-driven systems adapt autonomously through experiential data, optimizing performance across diverse building types while reducing reliance on skilled engineers. This approach enables buildings to navigate increasing system complexity while providing flexible control methods that respond to changing conditions in real time.

What are some of the benefits specific to schools?

Educational facilities present unique opportunities for AI-powered HVAC optimization due to their predictable yet variable occupancy patterns. The Stockholm study demonstrated significant reductions in district heating (3.12%) and electricity consumption (8.93%) across educational properties. These systems accommodate school schedules, with the capacity to hard-code holidays and vacations while learning occupant behavioral patterns during these periods. Beyond energy savings, schools benefit from improved indoor comfort, which research suggests positively impacts learning outcomes. The system's ability to precisely control temperature and CO2 levels is particularly valuable in educational settings where concentration and cognitive performance are essential for academic achievement.

Could you tell us about the recent case study in Sweden? How many schools participated? What was the methodology? What were the effects, and what were the some of the indirect effects?

The four-year study in Stockholm examined 87 of 120 educational properties managed by SISAB. The research compared energy consumption data from 2019, before the deployment of AI, with data from 2023, following AI implementation. To ensure accurate results, the study deliberately excluded intermediate years affected by the pandemic. The methodology employed was based on ITU-T L.1480, a recognized international standard for ICT effects assessment (see page 17). The AI solution utilized 9,901 sensors, including 5,496 for temperature and 4,405 for CO2, to optimize building operations. Key findings revealed a reduction in district heating by 2,388 MWh (3.12%) and a decrease in electricity consumption by 3,527 MWh (8.93%). These reductions resulted in a total carbon emission decrease of 259.17 tCO2e, comprising 109.87 tCO2e from heating and 149.30 tCO2e from electricity. Indirect effects observed included accelerated obsolescence of controllers due to increased write cycles, a minimal increase in server usage, and potential, though unquantified, reductions in maintenance-related travel.

How could moving this solution to schools in Boston lead to significant carbon emissions savings?

The comparative analysis between Stockholm and Boston revealed remarkably different potential impacts based on regional energy profiles. Implementing identical AI-powered HVAC solutions in Boston schools could yield carbon emission savings of 1,765.88 tCO2e annually—approximately 7 times higher (604% greater) than Stockholm's 250.6 tCO2e. This striking disparity stems primarily from differences in energy generation profiles. While Sweden's electricity mix is predominantly low-carbon (95.88% from hydropower, nuclear, wind, and solar), Boston's more carbon-intensive energy sources multiply the impact of each unit of energy saved. This comparison demonstrates how contextual factors significantly influence environmental returns on technological investments, making AI-powered HVAC particularly valuable in regions with higher-carbon energy profiles.

What are some of the system-wide challenges in integrating AI systems into legacy HVAC infrastructure?

The Stockholm study identified several critical integration challenges associated with AI deployment. Firstly, memory limitations in legacy controllers proved to be a significant obstacle. Specifically, with its 100,000 write cycle limit, the EPROM memory faced premature obsolescence due to the AI system's 15-minute setpoint adjustments compared to the previous monthly adjustments. This resulted in an accelerated replacement cycle, increasing annual greenhouse gas emissions by 7.3 tCO2e. Secondly, the AI solution's computational demands increased CPU and memory usage on existing automation servers by 5-10%, adding complexity to integration. Thirdly, network infrastructure requirements for data transmission, involving daily transfers of 333.8 MB for training and 3.62 MB for inference, presented additional integration challenges. Finally, the study emphasized the need for skill evolution. AI augments, rather than replaces, on-site technicians, requiring specific training programs and tailored work adaptation plans.

Would certain types of schools (universities over K-12, or large facilities over small facilities) benefit from using AI-powered HVAC systems more than others?

While the Stockholm study included a heterogeneous portfolio of educational facilities (from 100 to 48,000 square meters), it doesn't explicitly compare facility types. However, the findings suggest institutions with more significant energy expenditures would realize greater absolute savings. The carbon cost-benefit ratio exceeded 1:60, indicating substantial returns regardless of facility size. The research indicates buildings with more complex HVAC systems and higher operational variability may benefit most from AI optimization. Additionally, facilities in regions with carbon-intensive energy sources would see significantly magnified environmental benefits, as the Stockholm-Boston comparison demonstrates. Ultimately, the AI solution's effectiveness appears more contingent on the existing energy profile and HVAC complexity than the specific educational facility type.

Featured

  • Fort Collins to Convert 1980s Office Park into Junior High School

    The Liberty Common School, a charter-public school in Fort Collins, Colo., recently broke ground on an adaptive reuse project that will convert an 1980s-era office park into a 45,000-square-foot junior high school for seventh- and eighth-grade students, according to a news release.

  • Image courtesy of MiEN Company

    6 Ways to Pull Off a Major District Construction Project

    Designing and building a large-scale project on a K–12 campus is a monumental undertaking that requires the right blend of ideas, funding, design and execution to get it right. The process also relies on multiple partners, each of which has to handle its respective aspect of the project while also keeping the district’s broader mission and goals in mind.

  • Exhale Fans Launches New Generation of Bladeless Ceiling Fan

    Exhale Fans recently announced the launch of Gen. 5 of its flagship HVAC delivery product, the world’s only bladeless ceiling fan, according to a news release. The fan provides users with a 10% savings on HVAC energy costs and requires no renovations to current systems to install.

  • Abstract colorful arrows in front of a contemporary university building

    Spaces4Learning 2025 Trends in Higher Education

    With 2025 well underway, it’s time to take a look at some broader trends submitted by you, our Spaces4Learning readership. We asked for your thoughts on topics like classroom design, health & safety, materials & construction, and technology in both K–12 and higher-education environments. Below is a roundup of 2025 trends in higher education from the experts in the trenches.