Research
Glacian is grounded in over a decade of research spanning thermodynamics, high-fidelity physics-based modeling, and AI/ML optimization for mission-critical cooling. These foundations are reinforced by federally funded research from the U.S. Department of Energy, the National Science Foundation, the Department of Defense, JPMorgan Chase, ASHRAE, and other government and private agencies. Since 2013, we have received over $8 million in non-dilutive research grants. Our research outcomes have been recognized and featured by the U.S. Department of Energy for advancing next-generation data center cooling efficiency.
2025
Penn State: Development of Digital Twin for Liquid Cooled Data Centers
This project develops a next-generation digital twin for liquid-cooled data centers, integrating detailed thermal-fluid modeling with electrical systems to enable holistic optimization. The model leverages high-fidelity, equation-based simulation (Modelica) to accurately represent liquid cooling behavior under real operating conditions, supporting advanced design and intelligent control.
2025
NSF: Translation Potential of an Innovative Cooling Software for Efficient Data Centers
This project is to refine the commercial viability of the AI-powered cooling optimization software. This effort shifted the platform's focus from demonstrating energy cost savings to emphasizing revenue maximization for data center owners by freeing up power capacity for increased revenue-generating IT load.
2025
NSF: Novel Fault Detection and Diagnostics System for Data Center Resilience
This project is to enhancing data center resilience against cascading mechanical failures and external threats, such as malicious cyberattacks.3 The novel FDD system utilizes a Digital Modeling Tool to simulate the entire facility and proactively predict rare or unanticipated events that conventional AI systems struggle to forecast, ensuring cooling systems remain online and functional.
2020 - 2025
DOE: Optimal Co-Design of Integrated Thermal-Electrical Networks and Control Systems
This project is to create an open-source modeling and optimization platform for the optimal design and retrofit of Grid-interactive Efficient District (GED) energy systems, integrating renewable energy sources. The research includes developing integrated thermal-electrical network models—such as the CEAIM model—to simulate how data centers can reliably function and interact with the electric grid through services like load shifting.
2025
Penn State: Energy-Efficient Software for Data Center Cooling Optimization
The project is to accelerate the development of the technology to a stage sufficient for attracting commercial licensees or building a professionally funded startup, focusing on the dynamic adjustment of cooling to maximize profit.
2017 - 2022
ASHRAE: Development of Near-Optimal Control Sequence for Chiller Plants with Water Side Economizer Using Dynamic Models
This project develops near-optimal advanced control sequences for chiller plants with water-side economizers, enabling holistic optimization of cooling system power and energy in 504 scenarios across different plant configurations, load profiles, and U.S. climate zones. The results demonstrate up to 15% energy reductions and promote practical implementation of high-performance control logic in mission-critical cooling systems.
2016 - 2020
DOE: Physics-based Models for Data Center Cooling
This project is to develop the first practical, holistic software tool that couples the modeling of internal airflow management with external cooling systems to enable global optimization. The tool uses a self-learning regression model (ISAT-FFD) for high-speed airflow prediction, demonstrating significant energy saving potential (e.g., 53% to 74% in case studies) and allowing for autonomous optimal operation.
​
2013 - 2016
JP Morgan Chase and DOD: Model Predictive Control for Chilled Water Plants
This project develops a dynamic-optimization platform based on Python and Modelica for chilled-water plants that enables continuous tuning of control settings and system configurations. The tool supports flexible modeling of multiple chillers, pumps, and loops, and facilitates real-time model predictive control (MPC) for improved operational efficiency.
