| Summary: | For years, large power systems have been predominantly managed using the very well known synchronous machine on the generation side. With the increasing penetration of load and generation interfaced by converter-based systems, the conventional syn-chronous machine is being gradually replaced by these new devices. However, this slow and steady change of the generation technology has led to side-effects which can affect to the local stability and, if no remedial action is taken, system-wide stability. The main objective of this thesis is to understand, expose and overcome the weak-nesses of converter-dominated grids within a laboratory environment. To do so, a converter of 10kVA has been built implementing the standard D-Q axis Current Injection (DQCI) control and, for first time to date, the Virtual Synchronous Machine with Zero Inertia (VSM0H). Further developments have been made to this control strategy to implement this theoretical algorithm into a real system. The solutions proposed in this thesis for converter-dominated grids are based on the Grid Forming Nodes (GFN) solution. This composition suggests the usage of a combination of DQCI and Virtual Synchronous Machine (VSM) converters to achieve stability within converter-dominated grids. However, there is limited understanding about VSM functioning. Initially, the application of VSM-style converters was proposed as a natural replacement of the real synchronous machine present in the system without understanding which are the specific weaknesses that affect to this type of grid. Due to the aforementioned reasons, this thesis aims to provide a more scientific and rigorous explanation of why the GFN solution can provide stability for converter-dominated power systems. Using the converter built, it is possible to recreate scenarios where the technical challenges involved in power systems can be understood. Going from well-established solutions for relatively low Levels of Penetration (LoP), such as grid-friendly or smart converters, to an insightful explanation of why these solutions are not valid for converter-dominated grids with high values of LoP. This thesis puts more emphasis on the latter case, assessing experimentally the GFN solution for representative scenarios and control strategies used in the converter built explicitly for this project. Initially, three experiments are conducted to identify the weaknesses of converter-dominated grids. During these initial scenarios, it has been found that DQCI converters behave as current sources from the grid point of view, following an already imposed voltage signal. As current sources, DQCI converters do not provide effective voltage regulation. On the contrary, GFN/VSM converters behave as voltage sources behind an inductance from the grid point of view. Thus, they contribute to the entire grid stability by assuring the local voltage stiffness. However, further experiments with unbalanced and non-linear loads have found that this capability is finite and directly dependent on the equivalent converter filter impedance. During the execution of the experiments, there have been additional contributions explained throughout this thesis: The virtual resistor technique. Incorporating an additional feed-forward term into the inner current loop, the stabilizing effect of the parasitic inductor resistance is simulated. As result, the inner control loop operates in a more stable manner. The explanation, implementation and validation of a novel control strategy which makes the converter operate with variable speed response depending on the grid events. This adaptive functioning leads to a more robust operation for both the converter and the grid. A new control strategy based on the VSM0H (so called in this thesis as VSM0H+). Measuring the negative sequence voltage present in the grid, it is possible to add a compensation term which can provide an almost perfect voltage regulation under the presence of heavily unbalanced loads.
|
|---|
