Abstracts Category : Other

Process Systems Engineering of Microgrid Energy Networks: Design, Scheduling, and Supervisory Control

by Michael Zachar

Proliferation of small, distributed power generation has the potential to reduce losses from electricity transportation, alleviate congestion, enable high efficiency cogeneration systems, and serve as a way to harvest inherently dispersed renewable feedstocks. Unfortunately, relative to traditional power plants, distributed generation tends to have high capital costs, and the power output of distributed renewable generation (i.e. based on wind and solar) is inherently stochastic and intermittent. Multiple distributed generation technologies can be combined into a single system (i.e. a microgrid) to take advantage of synergies and improve the overall performance. However, this introduces a challenging design problem with a wide variety of generation and storage technology alternatives to choose from. In addition, distributed generation systems must be designed and operated so there is no disruptive impact on the existing infrastructure. Nonetheless, distributed generation can be an important part of an overall strategy to improve the sustainability and efficiency of power supply. This thesis addresses important practical problems related to the integration of distributed generation in the form of microgrid power systems using techniques from the Process Systems Engineering field. The problem of optimal microgrid design is investigated to (i) determine how public policy can drive microgrid adoption, (ii) quantify how geographic location and customer type impact microgrid efficacy and technology selection, and (iii) identify recurring motifs/trends in the technology selection and unit sizing. Then, the problem of optimal scheduling and supervisory control of a microgrid is considered to develop a framework for non-disruptive interaction with the existing electrical infrastructure. In particular, (i) a novel market structure for microgrids is formulated, (ii) a hierarchical supervisory control system is developed which utilizes stochastic optimization, and (iii) this control system is tested using a detailed, virtual microgrid simulation. Optimal microgrid system design was addressed with the application of mathematical optimization, specifically mixed integer linear programming. The problem considered was designing a system which provides both power and heat. Technologies considered include renewable generation, fuel-based cogeneration, and storage. In addition, the microgrid was assumed to be connected to the existing electrical infrastructure (i.e. power can be imported). This design problem was solved for a variety of policy scenarios, in different geographic locations, and for different types of customers (i.e. different load profiles). Important design parameters that were studied include the cost of energy supply, the integration of renewable power, the emissions associated with energy supply, and the optimal level of self-generation. The design results for different geographic locations and load profiles were then used to develop and train a heuristic procedure that can serve as a surrogate for detailed