This thesis focuses on the modelling and analyses the responses of a grid-following converter, which is a key component in modern power systems and in the integration of renewable energy sources. Grid-following converters play a crucial role in integrating renewable energy sources, such as solar and wind, into the grid. The primary goal of the research was to develop and investigate a simulation model of the converter in Matlab/Simulink and to compare the responses using two different simulation approaches: EMT (Electromagnetic Transients) and RMS (Root Mean Square). These two approaches allow different level of accuracy and computational time, which is crucial for the analysis of dynamic phenomena and long-term trends in the grid.
The initial phase of the research, the key assumptions and parameters for the model were determined. The converter was assumed to operate in a stiff grid, meaning voltage and frequency remain constant regardless of load changes. In addition, we assumed the converter operates in grid-following mode, tracking the grid’s voltage and frequency rather than actively regulating them, as grid-forming converters do. Thus, it can be said that a grid-following converter determines the injection or absorption of power into the grid or from the grid. The basis for developing the model was a mathematical description of a converter’s operation in the two axis (dq) frame, where elements such as the synchronization mechanism (PLL) and control loops were modelled using differential equations. The model was built in Matlab/Simulink environment, with most of the parts built from scratch, without relying on pre-built libraries. The EMT model employed detailed time-domain simulations of transient phenomena, using microsecond-level time steps, while the RMS model was simplified, capturing RMS values of voltages, currents, and power while neglecting fast transients. This simplification enabled significantly faster simulations, albeit at the cost of precision.
As part of the research, the key elements of the inverter were implemented, one of which is Phase-Locked Loop (PLL), which ensures synchronization with grid voltage and frequency. The PLL can be modelled using a PI controller and mathematical transformations (Park and Clarke transformations) to switch from a three-phase system to the dq reference frame. Of course, it was not necessary to model it in the EMT system as a pre-prepared model already exists in Simulink. Current control loops were implemented using PI controllers to regulate current in the d and q axes, ensuring the converter injects or absorbs active and reactive power as required by the consumers and the availability of a renewable energy source. The power stage of the converter included an LCL filter model to reduce harmonics and prevents distortion, which improves the regulation quality. The EMT model demonstrated a high degree of accuracy during sudden changes in power demand. The RMS model was successfully developed, to the point where its accuracy is comparable to that of the EMT model. Moreover, the simulation duration in the case of RMS approach is reduced by approximately ten times, while the accuracy does not degrade significantly. A comparison of the results from both approaches has shown that each excels in its own domain. EMT simulations are more detailed but require longer simulation duration, whereas RMS simulations are faster to execute but less accurate in analysis of transient phenomena.
This research enables the application of the obtained models in the design and optimization of the operation of a grid-following converter in various scenarios occurring in the power grid. Also, it provides an opportunity for further analysis of more complex networks, where dynamic phenomena and stability studies are included, or even interactions between multiple converters.
|
