Event Date
Bridging the Implementation Gap: Scalable Control and Reliability in Thermal Energy Systems
Matt Ellis, Associate Professor, UC Davis Chemical Engineering
Abstract
Achieving optimal and reliable operation in thermal energy systems requires low-cost, easily deployable, and scalable modeling and control strategies. While economic model predictive control (MPC) is uniquely suited to energy systems because it can dynamically handle time-varying pricing and demand, its implementation complexity remains a major barrier to widespread adoption. This talk presents recent advances that address these implementation bottlenecks and reliability modeling challenges across two key domains. First, to overcome the primary bottleneck preventing large-scale MPC deployment in building HVAC systems—control-oriented modeling—we introduce two novel techniques: a hybrid framework integrating physics-based and data-driven methods, and a fully data-driven approach. To address model structure selection, we also present a method for simultaneous topology identification and parameter estimation in thermal resistance-capacitance (RC) networks. Second, to address mission-critical performance in high-density liquid cooling systems used in data centers, we present a two-level directed graph technique for availability and reliability modeling in single-loop liquid-cooled data centers.
Bio
Matthew J. Ellis received a B.S. in Chemical and Biological Engineering from the University of Wisconsin—Madison in 2010 and a Ph.D. in Chemical and Biomolecular Engineering from the University of California, Los Angeles in 2015. After graduating, Dr. Ellis joined the Advanced Development Team at Johnson Controls (JCI), where he co-designed and developed the core patented economic model predictive control (EMPC) algorithm used within JCI’s Central Plant Optimization product. He served as the technical lead responsible for developing and applying an EMPC system to building-side HVAC systems. In 2019, he joined the Department of Chemical Engineering at UC Davis. He currently is an Associate Professor at UC Davis. Dr. Ellis’s work has resulted in numerous publications, over 50 granted non-provisional U.S. patents, and a co-authored monograph entitled Economic Model Predictive Control: Theory, Algorithms and Chemical Process Applications. His current research focuses on both fundamental and applied research in the broad area of control and optimization of process and energy systems, specifically developing rapidly deployable methods, control-oriented modeling approaches, and operational technology-based technologies that enable enhanced cybersecurity of control systems. His work addresses a wide range of applications, including chemical and other industrial processes, thermal systems (buildings, district heating and cooling systems, and data center cooling systems), and other energy storage system applications.