Modeling and Hierarchical Coordinated Control Approach for Parallel Operated DC DC Converters in a DC Microgrid

Abstract

DC microgrids are gaining significant attention over AC microgrids due to their ability to reduce conversion steps and enhance overall system efficiency. Contemporary loads, such as electric vehicles (EVs), battery energy storage (BES), and data centers, are predominantly DC in nature. Similarly, distributed energy resources (DERs), including photovoltaic (PV) systems, ultracapacitors, and fuel cells, can seamlessly integrate with suitable power electronic interfaces. These advancements necessitate proper control of power electronic interfaces operating in parallel within microgrids (MGs). A critical challenge in DC MGs is achieving optimal load sharing and voltage regulation, especially given the unequal ratings of DERs, sensor inaccuracies, and cable resistances. In the literature, several types of control techniques are provided, such as centralized, decentralized, and distributed control, for the parallel operation of the DC-DC converter. Out of these, decentralized control has become very popular because it does not rely on the communication network. While decentralized linear droop control is an effective technique for managing DC MGs, conventional approaches suffer from current-sharing errors and inadequate voltage regulation under heavy loading conditions. To address these limitations, this thesis proposes a coordinated droop control method incorporating three droop curves, two nonlinear (NL) and one linear, designed for the parallel operation of DC-DC boost converters sourced with DC source under varying loading conditions. The proposed controller dynamically transitions between the droop curves based on load variations, ensuring enhanced performance. A detailed analysis is conducted to derive droop coefficients for the different curves, achieving superior current-sharing accuracy and improved voltage regulation, even with line resistances. The effectiveness of the proposed method is demonstrated through MATLAB-based simulations using a 10 kW DC MG setup as the test platform.