A Tool for Probabilistic Evaluation of Microgrid Operating Strategies with Demand Side Management

2019-01-18T21:45:39Z (GMT) by Jesse Thornburg
Smart meters in microgrids enable fine-grained monitoring and control of individual building<br>demands. Possibilities are being explored globally to leverage these smart meter capabilities for<br>demand side management (DSM), particularly to advance rural electrification in emerging contexts.<br>Simulating different microgrid configurations and operating strategies before implementing<br>them is valuable for providing reliable service to customers and keeping expenses low. This<br>dissertation focuses on the design, development, and implementation of a simulation tool that<br>quantitatively compares microgrid operating strategies and sizing options. To this end, the tool’s<br>pre-processing engine accepts arbitrary parameters for probability distributions to characterize<br>loads and nondispatchable supplies (e.g., wind and PV). This dissertation presents methods to<br>compute probability distributions for the aggregate system demand from individual load distributions<br>that characterize each consumer. Further computational methods are given to derive<br>probabilistic estimates of aggregate loads reduced by DSM. These probability underpinnings<br>create effective system-level models for simulation studies of energy management schemes. The<br>models of aggregate load behavior and probabilistic supply are used in a simulation model that<br>includes dispatchable generation and storage components to perform Monte Carlo simulation<br>studies.<br>This dissertation describes the rationale and modeling parameters for different components<br>in the simulation tool. The tool allows different rule-based energy management strategies to<br>be implemented and compared, with certain supplies and storage options being controllable<br>while others are driven by external factors. To account for the wide range of possible outcomes<br>that occur in a real-world system, the tool is fundamentally probabilistic and runs its<br>MATLAB/Simulink-based microgrid model with Monte Carlo methods. The loads can be designated<br>Markovian or independent-in-time. The Energy Manager makes dispatch decisions with<br>limited knowledge of the rest of the system, similar to controllers in real-world microgrids. The<br>Energy Manager also limits certain loads with DSM to meet system goals, e.g., to reduce the<br>incidence of power cuts or limit fuel-burning generation. The Energy Manager can prioritize<br>renewable generation, energy storage, etc. as desired by the microgrid operators. To demonstrate<br>the simulation tool, this dissertation concludes with case studies based on a microgrid in Rwanda. The case studies provide examples of how smart meters, which are able to control<br>residential demand, can benefit microgrid operations. The deployment of DSM strategies using<br>smart meters is shown to reduce the occurrence and duration of power cuts when system<br>demand exceeds the total available supply. <br>