A unified probabilistic assessment of wind reserves for islanded microgrids

Abstract

This thesis presents an analytical and numerical framework for the unified probabilistic assessment of wind reserves with a focus on the applications of wind generation in islanded microgrids. A multivariate nonparametric kernel density estimation algorithm is proposed to generate probabilistic models of a site’s wind resource, electrical demand and the performance of installed wind generation. These models are numerically combined to evaluate the capability of wind generation to act as a dynamic reserve by predicting its performance when used for demand response, secondary generation and frequency regulation in an islanded microgrid. The proposed modeling framework captures multivariate cross-correlation, nonstationary environmental and load behavior, as well as multimodality in their underlying probability distributions. A case study is conducted using field data from Cartwright in order to validate the proposed algorithms. The case study results include probabilistic predictions of wind generation effectiveness for varying load profiles and generation capacity. PLEXIM simulation software is used to implement a model microgrid to demonstrate the integration of wind generation and its regulatory capabilities. The proposed algorithm has applications in power system planning and operation, and it provides probabilistic data for use in energy management and optimization of microgrids.