MULTI-DOMAIN MODELING, SIMULATION AND OPTIMIZATION OF INTEGRATED ENERGY SYSTEMS
Open Access
- Author:
- Anbarasu, Saranya
- Graduate Program:
- Architectural Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- March 24, 2025
- Committee Members:
- Satadru Dey, Outside Unit & Field Member
Wangda Zuo, Chair & Dissertation Advisor
Greg Pavlak, Major Field Member
James Freihaut, Major Field Member
Jiazhen Ling, Special Member
Sen Huang, Special Member
James Freihaut, Program Head/Chair - Keywords:
- Decarbonization
Decentralization
Smart Communities
Integrated Energy Systems
Modelica
Reduced order modelingOptimization
Reinforcement Learning
Sustainability.
Decarbonization
Decentralization
Integrated Energy Systems
Modelica
Reduced-order models
Optimization
Reinforcement learning
Sustainability
Real-world case study
Smart communities
Neural networks
Interdependencies - Abstract:
- The global energy sector is undergoing a profound transformation driven by the need for decarbonization, decentralization, and digitalization. By 2050, electricity consumption in buildings is set to rise by 50%, with transportation electrification contributing to a 27% increase in global electricity demand over the next decade. Coupled with this, the growing digitalization, particularly the energy consumption of data centers, is expected to exacerbate pressures on grid infrastructure. This shifting landscape demands a reevaluation of traditional energy systems that are increasingly incompatible with these new challenges. In response, the focus is shifting toward integrated, multi-energy networks that seamlessly combine renewable generation, energy storage, and demand-side resources. This dissertation responds to these evolving demands by developing innovative methodologies aimed at enhancing the design, operation, and control of Integrated Energy Systems (IES). These systems, which optimize the flow of multiple energy sources, carriers, and technologies, promise more efficient, flexible, and sustainable energy production and consumption. By creating dynamic, high-fidelity Modelica models for district and community-scale energy systems, the research makes significant strides in simulating complex, multi-domain energy processes, including thermal, electrical, and hydraulic systems. To ensure practical applicability, the research proposes model reduction strategies, such as spatial aggregation and the use of artificial neural networks (ANNs), to accelerate simulations by up to 2000 times. Moreover, the research also addresses the optimization of existing district energy systems (DES), demonstrating how model-based optimization can significantly reduce fuel consumption, emissions, and operational costs. By introducing Reinforcement Learning (RL)-based energy management strategies, the study offers a path to further reduce operational costs, peak demand, and emissions while improving resource allocation. Finally, by exploring the interconnections between energy and other infrastructure systems in future smart communities, this dissertation offers a multi-domain framework that provides new insights into the cascading impacts of disruptions. This work represents a crucial step toward building adaptive, resilient, and sustainable energy systems for the future.