Integration of Renewable Energy Sources into Power Grids Applying Distributed Virtual Inertia and Model Predictive Controls

Abstract

The current energy arena is changing, from fossil fuel-based generation to power electronic converter-interfaced renewable generation. Hence, the power system inertia and short-circuit current gradually reduce, making low-inertia grids more sensitive to frequency disturbances (i.e., power mismatch between generation and demand) and jeopardizes system stability. This thesis develops control methods for grid-following and grid-forming converters employed toward more power electronic-based generators. The thesis contributions are divided into two main approaches. First, the distributed virtual inertia method, a grid-following converter solution aimed at synthetic inertia provision, is studied in detail. It is depicted that this method has two drawbacks: (1) small-signal stability analyses affirm that a local mode associated with the controller is prone to become unstable when the converter operates in weak grids, and (2) the DC-link voltage is not reverted to its reference value after the power mismatch between generation and demand occurred in the host grid. Herein, the aforesaid problems are addressed properly; efficient compensators are proposed which eliminate the adverse impact of distributed virtual inertia gain on the converter stability in weak grid connections. Moreover, the distributed virtual inertia controller is modified so as not to affect the outer-loop voltage regulator after transients. Then, the DC voltage restoration is possible. Second, the conventional primary control, i.e., inner-loop cascaded linear controller and outer-loop droop, used in islanded AC microgrids is discussed. In sum, this approach has inferior dynamic response and rapid rate of change of frequency following perturbations. Accordingly, the thesis addresses these issues by introducing a modified virtual synchronous generator control. A Laguerre functions-based discrete-time model predictive controller with a multiobjective cost function is incorporated as the heart of the control system which supersedes the inner loop for hierarchical linear controllers of grid-forming converters. This yields realizing large prediction horizons, improved dynamic performance (very short rise time and slight overshoot), and inherent overcurrent protection in the case of fault or overloading without sacrificing the controller robustness.

Finally, the merits of proposed techniques are verified by comparisons with corresponding primary methods. And, detailed model simulations are conducted in MATLAB/Simulink to show the efficacy of the proposed controllers.