MODELING AND CONTROL OF ISLANDED MICROGRID WITH INVERTER-INTERFACED PV-BATTERY SOURCES

Abstract

In recent times, with the increased demand for using Renewable Energy Resources (RESs), Microgrids (MGs) have emerged to be an integrated framework for supply and Demand Side Management (DSM) at the low voltage distribution network. These MGs constitute of several components, viz., Microsources (MSs), storage devices, and different types of loads. Also, the inherent generation intermittency of RESs along with no or low inertia of interfaced Power Electronic (PE) converters, makes the MG a highly oscillating system that is easily vulnerable to instability. Hence, for smooth and coordinated operation, dedicated control strategies and robust stability analysis tools are inevitably used. For performing these tasks under different operating scenarios, accurate mathematical models of MGs are need to be developed. Unfortunately, due to its complex SoSs architecture, modeling of MGs leads to the formation of a higher-order system having a large number of states. Moreover, as these MGs are Multi-Input Multi-Output (MIMO) systems, its handling becomes a tedious task and simultaneously leads to increased controller design complexity. Furthermore, the order of these controllers is kept around the systems’ order, while a higher-order controller by default has stability issues. In the present work, three important aspects of MGs, i.e., modeling, order reduction, and controller design have been studied. In classical control theory, Model Order Reduction (MOR) is used to obtain an equivalent lower-order model (by preserving the dominant states), which are used for the controller design process, time and frequency domain studies, stability analysis, etc. Now, there are several MOR methods in the control literature, with different levels of computations, and selecting a suitable one for obtaining a Reduced-Order Model (ROM) for a specific application is quite ambiguous. Also, as the islanded MGs have very low or negligible inertia and highly oscillating systems where it is difficult to distinguish between the slow and fast state dynamics as the gap might not be substantial. Therefore, a comparative insight into the qualitative usage of various MOR techniques concerning its applications on an islanded MG system is presented. Standard performance evaluation and analytical tools from the control theory are considered for investigating the efficacy, applicability, and inferences of these MOR methods as per their required application. Moreover, issues of using ROMs for online control purposes are identified, and a simplified full-scale model is developed and used for the same. i Another important issue of inverter-interfaced MG systems, i.e., the assumption of inverters’ input terminal voltage to be constant is addressed here. The small-signal modeling of Photovoltaic (PV) and battery type islanded MG system is developed where the dc-bus voltage at the input side of the inverter is to be variable. This includes the generation uncertainties of the PV based RESs whose output varies according to the solar irradiance. Also, the developed simplified MG model is a lower order system that avoids any extra computations and resource utilization. This also fills a gap regarding the inclusion of the storage dynamics by using the State-Space Model (SSM) of a battery system. The developed SSM of the PV-Battery type islanded MG system is verified by applying the Hardware-in-loop (HIL) simulations over the Matlab/Simulink and the Real-Time Digital Simulator (RTDS) platform. Later on, a robust decentralized voltage and frequency (V-f) control for the developed simplified SSM of the islanded MG system is proposed. The Linear Quadratic Regulator (LQR) principle is used for the real-time V-f control considering both generation intermittency and load disturbances. Here, the time-varying generation and load arbitrages are accounted for without considering the forecast information, therefore, concerns of huge data handling and computations are reduced. Also, the decentralized scheme includes the effects of parameter uncertainties due to the interconnected dynamics of neighboring buses. The LQR controller is realized based on the output-feedback mechanism, hence, it does not need any state estimators, which immensely increases the computations. As the MGs are mainly established on the LV distribution networks, a more efficient resistive or reverse droop scheme is realized instead of the conventional/regular droop. This assists in better power-sharing among the MSs. The same islanded MG model is used to develop a Power Management System (PMS) based on the distributed Model Predictive Control (MPC) strategy. Here, the generation and load disturbances are taken into account as per the forecasted generation and load profiles. The loads are divided into critical and noncritical/curtailable types. The PMS extracts maximum power from renewables and minimum wear and tear of storage units, i.e., the longevity of batteries with the least number of charging/discharging cycles and ramping constraints. Also, from the loads perspective, the pre-emptive shedding of noncritical loads are applied to achieve reliable and longer duration of power supply to the critical ones. This work uses the simplified lower-order MG model that eases the computational burden. The adaptive MPC algorithm implements the changing operating scenarios in real-time by updating all the involved state variables. The test ii case scenarios for 24 and 48 hrs duration for both summer and winter season are realized to verify the performance of the proposed MPC scheme. Hence, in this thesis important aspects concerning the MG modeling, order reduction, and controller design features are studied. Its outcomes will assist the power and energy researchers in studying various MG operations from modeling and control perspective.