Pulkit Nahata obtains his PhD: congratulations!

Abstract

Thrust by the ever growing need to leverage the benefits of Renewable Energy Sources (RESs), to rein in climate change and electricity costs, and to guarantee uninterrupted power supply to areas lacking electric infrastructure, power generation is becoming increasingly distributed. Central to this shift in operational exemplar are microGrids (mGs), commonly recognized as small-scale electric networks integrating Distributed Generation Units (DGUs) and loads. Compatible with both AC and DC paradigms, mGs can operate either connected to, or detached from, the main grid. While operating as stand-alone, islanded entities, next-generation mGs should not only mirror an autonomous power system complete with advanced controls, but also allow DGUs and loads to enter/leave over time with minimal supervision efforts. The primary objective of this doctoral thesis is therefore the design and development of hierarchical control schemes for the overall operation and control of IslandedmGs (ImGs) with flexible structures—or, with no reference topologies. The control methodologies devised as part of this work attend to some of the key challenges in the control of AC and DC ImGs, while doing away the limitations of several existing schemes.
This thesis comprises three parts. The first part delineates a passivity-based approach to thedesign of scalable primary controllers ensuring voltage stability in DC ImGs, and voltage and frequency stability in AC ImGs. Unlike most primary controllers in the literature, our passivating regulators guarantee offset-free tracking of reference signals, while factoring in power line dynamics and nonlinear loads. Furthermore, the proposed primary local controllers—completely decentralized from the standpoint of both design and structure—can always be synthesized. This brings about a true Plug-and-Play (PnP) functionality, that is, DGUs can plug-in to, as well as plug-out of, the ImG network without having any bearing on its stability. The second part focuses on DC ImGs and, with a view to their safe, reliable, and efficient operation, details higher-level supervisory control structures capable of seamlessly interfacing with the primary controllers developed before. Our supervisory controllers take two forms: (i) consensus-based distributed controllers; and (ii) flexible EnergyManagement System (EMS), along with an intermediary layer drawing on power-flow equations. In the former case, we consider the objective of voltage balancing and proportional current sharing and, without assuming any timescale separation, prove that the desired coordinated behaviors are achieved in a stable fashion. Compatible with that of the primary regulators, the design of the secondary regulators is fully decentralized, facilitating PnP operations. The supervisory control architecture in the latter case aims at properly defining mG internal voltages, efficiently coordinating DGU operations, and minimizing mG operation costs, while taking into consideration the non-deterministic absorption/production of loads and renewables. Any meaningful planning and optimization task pertaining to temporally varying mGs and distribution networks alike hinges on network identification—the knowledge of the admittance matrix capturing topological information and line parameters of an electric network.
Keeping this is view, in the third part we set out a data-driven, online network identification procedure using phasor measurements of voltages and currents. We also take advantage of tools from theory of Optimal Experiment Design (OED) to accelerate the convergence of the
proposed identification algorithm. All theoretical results and design approaches presented in the thesis are corroborated by means of extensive numerical simulations capturing realistic mG benchmarks.