TL;DR:
- MIT is using AI to optimize campus energy use and reduce emissions.
- Smart thermostats, once used in homes, are now applied to an entire campus.
- The project aligns with MIT’s Climate Action Plan and started in 2019.
- Research focuses on predicting optimal temperature set points.
- Initial pilot in Building 66 expanded from classrooms to entire buildings.
- Physics-based models and external data inform AI algorithms.
- AI balances thermal comfort with energy efficiency.
- Scalability is a key goal for broader campus impact.
- Collaboration with Schneider Electric expedites implementation.
- Successful programs could turn MIT’s campus into a virtual energy network.
Main AI News:
Smart thermostats have revolutionized how we manage energy in our homes, adapting to occupancy patterns and preferences with machine learning. Now, MIT is pushing the boundaries of this technology, applying it to the vast and complex ecosystem of a campus. In a cross-departmental endeavor, MIT is pioneering AI building controls that respond in real-time to internal and external factors, aiming to slash energy consumption and emissions across its campus.
The Energy Challenge
MIT’s sprawling campus posed a unique energy challenge. Existing building management systems struggled to adapt quickly to fluctuations in occupancy or external factors such as weather forecasts or grid carbon intensity. This inefficiency led to excessive energy use, often resulting in suboptimal room temperatures. MIT researchers have now embarked on a mission to employ AI to predict and establish optimal temperature set points at the individual room level, considering various factors. This approach aims to enhance energy efficiency without the need for manual intervention.
Les Norford, a professor of architecture at MIT, explains, “It’s not that different from what folks are doing in houses, except we have to think about things like how long a classroom may be used in a day, weather predictions, the time needed to heat and cool a room, the effect of the sun, and neighboring classrooms.” These considerations form the core of the research and pilot programs conducted by Norford and his team.
Exploring AI Possibilities
The initiative aligns with MIT’s Climate Action Plan, which calls for the exploration of AI’s potential to reduce on-campus energy consumption. Vice President for Campus Services and Stewardship Joe Higgins initially proposed the idea in 2019, sparking a series of events that led to this groundbreaking project. Undergraduate and graduate student researchers began running differential equations and managing pilots to test the feasibility of the concept. Jeremy Gregory, executive director of the MIT Climate and Sustainability Consortium, joined the project, leveraging MIT’s research community to tackle real-world challenges.
Expanding the Scope
Initial pilots focused on thermostat set points in one building, but MIT soon realized that classrooms offered a more complex testing ground. Building 66, a mixed-use facility housing classrooms, offices, and labs, became the expanded testing site. The sheer number of classrooms on campus provided abundant opportunities to gather data and fine-tune the parameters of the project.
Developing Smarter Building Controls
The development of smarter building controls begins with a physics-based model using differential equations to understand how objects heat up or cool down, store heat, and distribute it across a building. External data, including weather, grid carbon intensity, and classroom schedules, are integrated into the AI algorithms. The AI then calculates optimal thermostat set points, balancing the thermal comfort of occupants with energy use efficiency. Real-life testing includes feedback from building occupants to ensure comfort. AI algorithms translate this learning into energy and carbon emission savings.
Scaling Up for Impact
While current pilots focus on six classrooms within Building 66, the long-term goal is to implement AI controls across entire buildings. The potential energy savings could be substantial, contributing significantly to MIT’s decarbonization goals. The scalability of these algorithms is crucial to achieving this impact, making them applicable to numerous rooms and buildings on campus.
Operational staff play a pivotal role in connecting research with real-world operations. The Building Management Systems (BMS) team quickly integrated AI systems with existing controls, enabling rapid pilot deployment. As the pilots near completion, the BMS team has identified 50 additional buildings where the technology can be installed to further energy savings. Collaboration with Schneider Electric, a building automation company, has expedited the deployment of new control algorithms in Building 66 classrooms.
Expanding Possibilities
Successful completion of these programs could transform MIT’s campus into a virtual energy network. Thousands of thermostats could coordinate as a unified virtual entity, reducing the need for carbon-intensive power plants during peak times and optimizing power grid energy use. This not only promises substantial energy savings but also brings MIT closer to its decarbonization goals.
As MIT’s AI-driven campus management pilots continue, they exemplify the university’s commitment to being a “test bed for change.” This project not only showcases the potential of cutting-edge research but also serves as a model for harnessing campus resources as a living laboratory. MIT is at the forefront of sustainability and innovation, leading the way toward a greener future.
Conclusion:
MIT’s pioneering use of AI in campus energy management not only aligns with sustainability goals but also showcases the potential for broader market applications. As businesses and institutions seek to reduce energy consumption and emissions, AI-driven solutions, similar to MIT’s approach, could revolutionize energy management, leading to increased efficiency and reduced environmental impact. This initiative serves as a compelling example of the transformative power of AI in addressing complex challenges.
