Topology optimization and control of high-renewable power systems

Abstract

Power systems ensure security and economics with sufficient operational flexibility. The ongoing renewable energy transition demands more active and wider leverage of the flexibility of network topology to accommodate high penetration of renewable energy. Meanwhile, how to optimize and control the network to fully exploit this topological flexibility in high-renewable power systems also faces great challenges. Accordingly, in this thesis, we study a series of essential issues of topology optimization and control of high-renewable power systems. These issues cover network connectedness in topology optimization problems, modeling and solving frameworks, topology transition, and novel applications of topology control. Our contributions are as follows:

We resolve the issue of network connectedness in optimal transmission switching problems and their security-constrained versions. We develop an electrical flow based approach to strictly ensure network connectedness in optimal transmission switching problems. Based on this approach, we fully address the issue of network connectedness in security-constrained optimal transmission switching problems, by proposing two network connectedness criteria with their mathematical formulations derived.

We achieve a simultaneous, proper, and efficient treatment of uncertainties of variable renewable energy and N-k security in uncertainty-aware topology optimization. We develop a novel three-stage stochastic and distributionally robust optimal transmission switching model. With a tractable reformulation of the model derived, we develop a solution approach that utilizes strong duality and parallel computing to improve computational efficiency.

We develop a learning-based solution approach to satisfy the strict requirements to solving topology optimization and control problems posed by the renewable energy transition. This approach inherits the capability of the state-of-the-art heuristic approaches to tackle large-scale and highly non-convex cases, but outperforms them in terms of both solution optimality and especially computational efficiency by leveraging a combination of graph neural networks and reinforcement learning.

We set up and solve the topology transition problem in response to increasingly frequent transition from an initial topology to a desired optimal topology. Taking only static factors into consideration, we propose a mixed-integer optimal topology transition model, which can yield satisfactory transition trajectories to overcome the unreliability of ad hoc topology transition. Moreover, we propose a transition-embedded topology optimization model, which always ensures the existence of feasible transition trajectories. Considering dynamic factors further, we develop a methodology to achieve bumpless topology transition by optimally adjusting the auxiliary control resources along with switching transmission lines.

We reveal the potential of topology control for improving two vital aspects of dynamic performance of high-renewable power systems. For low-inertia transmission networks, we derive a closed-form formulation of the H2-norm metric of synchronization performance, which theoretically indicates the potential of transmission switching to improve synchronizability. This revealed potential motivates an H2-norm switching approach for improving synchronization performance. For inverter-based microgrids, we conceptually propose four different topology control schemes for stabilization, with their effectiveness demonstrated numerically.

Overall, this thesis solves a series of essential issues of topology optimization and control in high-renewable power systems and promotes the renewable energy transition from a topological perspective.