Committee:

Dr. Sakis Meliopoulos, ECE, Chair , Advisor

Dr. Daniel Molzahn, ECE

Dr. Andy Sun, ISyE

Dr. Santiago Grijalva, ECE

Dr. Lukas Graber, ECE

Abstract: Climate change has increased the frequency and intensity of severe weather conditions leading to catastrophic power interruptions.  To mitigate these interruptions, the adoption of microgrids (µGs) emerged. A µG is a cluster of interconnected loads and distributed energy resources (DERs) that may be managed collectively to achieve given operational objectives. Concurrently to the adoption of µGs, large amounts of renewable resources have been integrated into the grid. Renewable resources are characterized by variable and uncertain power output which creates operational challenges to grid operators. Grid operators are faced with an increasing need for flexible resources that are able to absorb the variability and uncertainty in operation. Part of the need can be met by µGs; a µG may be optimized to provide different ancillary services to the grid. We propose a microgrid energy management system (µGEMS) that optimally plans the operations, and control the DERs while committing, holding, dispatching, and maintaining different ancillary services for the grid in a reliable and economical manners. Reserve, regulation, and voltage support services can be supplied simultaneously via the µGEMS. The proposed µGEMS may be used to commit the services a Day-Ahead (DA) in advance to dispatch, or in Real-Time (RT) (i.e., DA and RT Commitments). Commitment rules that the µGEMS can consider include minimum acceptable capacities, required time to respond, and required time to maintain. We model the µG as an AC network using the current formulation to obtain a model that is mostly linear. Bus voltage, circuit loading, and point of common coupling (PCC) power factor limits are enforced during the commitment, the holding, the dispatching, and the maintaining stages of services. The proposed µGEMS consists of a collection of interacting optimization problems each with a certain task, planning horizon, and frequency of solve. The optimization problems are generally mixed-integer quadratically constrained programming (MIQCP) problems. A solution methodology for the optimization problems is proposed based on successive linear programming (SLP) which promises efficient handling of discrete variables.