Optimal Control of Commercial Office Battery Systems, and Grid Integrated Energy Resources on Distribution Networks
The proliferation of new sensing and actuation technologies presents new opportunities for enhanced supervision, optimization, and control of energy systems. In the commercial office environment, the use of smart power strips, uninterruptible power supplies, and advanced building energy management systems is growing as a means to implement control of energy efficiency measures and demand response. In electricity distribution networks, phasor measurement units and inverters are enabling utility operators to monitor and manage distribution systems more effectively. The physics or operational constraints of systems within the commercial office environment or distribution networks are often nonlinear or nonconvex, and thus difficult to incorporate into optimization programs. This dissertation presents research into modeling and optimal control of two nonlinear energy systems.
At the commercial office scale, we discuss the development and implementation of optimal control of plug loads and office scale battery storage. Building upon successes in optimal control of plug loads, we propose a model predictive controller (MPC) for the incorporation of battery storage. We derive a model of an off the shelf battery storage system through experimental data, and discuss extensions to allow controllable charging. We investigate two methods to solve the nonlinear and binary MPC, and simulations show the promise dynamic programming method.
At the distribution network scale, we discuss power flow models for optimization of grid distributed energy resources (DER), and techniques for solving nonlinear optimal power flow (OPF) problems. First, we study semidefinite programming as a method for solving nonlinear OPFs for control of voltage phasors. Simulation results motivate the development of novel models. We then derive a linearized unbalanced power flow model (LUPFM) for use in convex optimal power flow (OPF) formulations. The LUPFM builds upon previous work by adding a relationship between voltage phasor and complex power flow. A study into the LUPFM accuracy shows its fidelity for benchmark networks. We discuss two applications of the LUPFM. The first is balancing of voltage magnitude on a distribution network, and simulations are successful for both radial and mesh networks. The second application is minimization of voltage magnitude and angle difference for switching operations. Simulations show the success and potential of the LUPFM for OPF control of voltage phasors.