Year: 2019
Type: dissertation
Abstract: Electric power systems are shifting away from conventional fuel-burning generation and moving towards renewable energy generation. Flexibility from energy storage and demand response is needed to accommodate the variability of renewable power sources. A key challenge to demand response is the uncertainty of loads. In this thesis, we design online optimization-based algorithms for demand response. Using the performance-guaranteed frameworks we develop, we can utilize flexible loads for demand response in real-time while accounting for uncertainty. We first design online convex optimization algorithms for power setpoint tracking with flexible loads. We deal with different types of feedback from the loads: full and limited feedback and two intermediary feedback levels. This is done to account for different plausible communication scenarios. We apply our approaches to thermostatically controlled loads and charging electric vehicles. We then formulate a two-level method for the energy management of multi-energy buildings. A scheduling level and a tracking level are utilized to leverage the building's several sources of flexibility for providing ancillary services while satisfying its different energy requirements. Next, we establish a predictive online convex optimization framework in which estimates about future rounds are used to improve the performance of online convex optimization algorithms. We apply this framework for frequency regulation and curtailment, and observe performance improvement over standard algorithms. Thereafter, we extend the multi-armed bandit framework to an instance where the number of arms to play evolves according to a wide-sense stochastic process. We use this extension to curtail flexible loads' power consumption in response to random power imbalance due to renewables. Our approaches rely only on algebraic operations and projections onto a compact and convex sets. They thus are computationally efficient and can be scaled up to very large numbers of loads for real-time decision making.
Source: TSpace (University of Toronto)
Author: Antoine Lesage‐Landry
Language: English
Topic: Smart Grid Energy Management
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3003483093