Distribution System Resilience Enhancement Using Movable Energy Resources
Loading...
Authors
Gautam, Mukesh
Issue Date
2022
Type
Dissertation
Language
Keywords
cooperative game theory , distribution systems , graph theory , movable energy resources , reinforcement learning , resilience enhancement
Alternative Title
Abstract
Electric utilities, government agencies, and the public have been concerned about the risks associated with natural disasters and extreme weather events because of their significant impacts on the security, reliability, and resilience of infrastructure systems including power systems. Out of the various components of power systems, the distribution systems have been impacted more because of their characteristics (e.g., presence of scattered and critical loads). Therefore, it is important to develop an approach to enhance the security, reliability, and resilience of distribution systems.Movable energy resources (MERs), as the name suggests, are movable and flexible resources. Movable in the sense that they can be dispatched quickly from staggering locations to power outage locations, and flexible because they can be designed to various sizes and can be quickly integrated into the distribution grid after the occurrence of a natural disaster. These resources can be designed to supply up to a few mega-watts of loads. When there is an island due to an outage and no other means of power distribution service restoration strategies are feasible, MERs can be dispatched to power outage locations to supply at least local and isolated critical loads. However, the effectiveness of the use of MERs is highly dependent on its optimal placement, which needs to be performed very fast after the occurrence of natural disasters. Moreover, the optimal size and number of MERs also play a significant role in post-disaster critical load recovery.
Game theory-based approaches and learning-driven techniques have the potential to assess complex interactions and behaviors of distribution system resources. Cooperative game theoretic approaches based on the Shapley value have the ability to uniquely assign payoff among players of the game taking into account their marginal contributions; and out of various learning driven approaches, reinforcement learning-based approaches have the capabilities to learn from experiences during online operations of power systems. Cooperative game theory- and reinforcement learning- based approaches are, therefore, investigated in this research work to enhance distribution system resilience.
This dissertation is mainly divided into three interdependent tasks for the resilience enhancement of distribution systems through the deployment of MERs. The first task is determination of optimal total size and number of MERs. This task is a long-term planning problem, generally performed annually, and it is a necessary step to ensure the availability of sufficient resources when needed. The graph theory and combinatorial enumeration technique are leveraged in this task to determine optimal total size and number of MERs. The second task is pre-positioning of MERs. This task is a relatively short-term planning problem, generally performed in a few days or a week-ahead manner before the occurrence of extreme events based on weather forecast and monitoring data. The cooperative game theory has been employed for pre-positioning of MERs in this dissertation. The final task is post-disaster routing of MERs. This task is an operational problem, which is performed after the occurrence of extreme events. Post-disaster routing of MERs is implemented if disasters cannot be predicted (i.e., pre-positioning of MERs is not applicable) or if MERs are managed to be deployed after an event not prior to a predicted event. The deep reinforcement learning-based model is leveraged in this dissertation for post-disaster routing of MERs.
In summary, the thesis presented here is that, the combination of planning- and operation-based resilience strategies for the optimal sizing, siting, and routing of movable energy resources will enhance the resilience of electric distribution systems.