Future of Buildings: Graduate Insights and Innovations

“Learning to Store: Physics-Aware Neural Control Activates Phase Change Materials in Buildings”
Presenter: Yongyun Jin, Ph.D. Candidate, Mechanical and Aerospace Engineering

As cities face rising temperatures and growing energy demands, buildings must become more adaptive, efficient, and resilient. In this seminar, we explore how advanced control systems and smart materials can work together to meet that challenge. We introduce a cutting-edge framework that combines bio-based phase change materials (PCMs) with neural networks to create a new kind of building intelligence—one that doesn’t just react to heat but learns how to store and manage it. This approach treats PCMs not as passive thermal buffers, but as programmable thermal assets. At the core of our method is a differentiable predictive control (DPC) strategy driven by a physics-consistent neural network (PCNN). This system learns building thermodynamics in a way that respects physical laws, enabling it to optimize energy costs, reduce peak loads, and maintain comfort, even during extreme heat events. Through simulations using EnergyPlus, we demonstrate that this method can cut energy costs by up to 44%, reduce peak loads by 35%, and lower indoor peak temperatures by nearly 4 °C. These results show how materials and machine learning can co-evolve to support the next generation of climate-resilient buildings.

About the speaker: Yongyun Jin is a second-year Ph.D. student at Syracuse University’s Built Environment Science and Technology (BEST) Lab, directed by Professor Bing Dong. His research focuses on advanced building materials and intelligent control systems.

Hosted by the ASHRAE Student Branch at Syracuse University, “Future of Buildings: Graduate Insights and Innovations” presentations are held every other Thursday at 727 E. Washington St. Lunch is provided and free parking is available onsite. Please RSVP to attend.

Learn more