1 Department of Electrical Engineering, Technical University of Denmark2 Center for Electric Power and Energy, Department of Electrical Engineering, Technical University of Denmark
Switching transients in wind turbines, the collection grid, the export system and the external grid in offshore wind farms, during normal or abnormal operation, are the most important phenomena when conducting insulation coordination studies. However, the recommended models and methods from international standards and guidelines are insufficient to take into account considering the special conditions in offshore wind farms. This thesis focuses on the improvement of models and methods used in specialized electromagnetic transient programs. In this thesis, an introduction to offshore wind farm electrical systems and the topic of transients is given in Chapter 1. In Chapter 2 it is described how the component and sub-system models are developed and used from an applied point of view, where common and detail models are mentioned. In Chapter 3 results from frequency domain measurement on transformers and cables are presented. In Chapter 4 results from time domain measurements and simulations of switching operations in offshore wind power grids are described. Specifically, switching operations on a single wind turbine, the collection grid, the export system and the external grid measured in several real offshore wind farms are shown together with simulation results. Switching operations in offshore wind power grids can be simulated with different electromagnetic transient programs. Different programs were used in the project and compared results are included in Chapter 4. Also in Chapter 4 different models were used inthe same program, and in several switching operations sensitivity analysis was made. General models and methods to study switching transient overvoltages in offshore wind farms have been shown in this work. These models and methods are valid for all offshore wind farms, or other systems to be analysed. As long as the required frequency domain measurements are performed in the appropriate components, the information used is correct and the system is created in a suitable simulation tool. The general method used to validate and create detailed models was proven to be very powerful and efficient. This method is also valid for all offshore wind farms, or other systems to be analysed. Once this method has been completed, the predictive simulations can be performed with certainty. The specific method to adapt FRA measurements for transformer wide band modelling proposed in  was used successfully in 100kVA, 4MVA and 120MVA transformers. Based in the positive sequence voltage ratio comparison from different wind farm transformer, it was concluded that the first resonance frequency from low to high voltage side is around 10kHz. This frequency match the low-frequency-end of the frequency range for natural frequencies on large transformer windings according to . On the other hand, after the positive sequence voltage ratio comparison from different wind turbine transformer, it was concluded that the first resonance frequency in some transformers, from low to high voltage side are within range of the frequency range for natural frequencies on medium transformer windings according to . It was found that some of the resonance frequencies from the offline diagnostics of transformers based on SFRA measurements, match the positive sequence voltage ratio resonance frequencies from low to high voltage side. From the frequency variation of equivalent inductance and resistance ratio, it was found that the L/R ratio, R/Rdatasheet ratio and L/Ldatasheet ratio can be calculated from the recommended SFRA measurements and the corrected admittance matrix. In general, the L/R ratio decreases as the frequency increase in all transformers. But the initial magnitude and rate of decrease in frequency from the L/R ratio depends on the transformer rating and technology. In the analysed transformers, the R/Rdatasheet ratio increases as the frequency increase, but the wind farm transformer´s ratio increase more rapidly than the wind turbine transformers. In the analyzed transformers, the L/Ldatasheetratio decrease slightly as the frequency increase, but the error in the wind turbine transformer´s ratio is much higher than in the wind farm transformers. From the time domain validations in wind turbines and collection grid, it was shown that all the measurements could be reproduced with some degree of error using: • standard and detailed transformer models, • PI sections and frequency dependant (phase) models, • simplified and detailed VCB models, • simplified and detailed wind turbine models, • simplified wind farm substation models and • simplified external grid models. From the time domain validations in the export system and external grid, it was shown that all the measurements could be reproduced with dome degree of error using: • standard and detailed transformer models, • PI sections and frequency dependant (phase) models, • simplified circuit breakers models, • simplified wind turbine models, • simplified and detailed wind farm substation models and • simplified and detailed external grid models. In the time domain predictive studies for the collection grid, export system and external grid, standard or simplified models were used except the detailed synchronous machine model and frequency dependant (phase) cable models.
Main Research Area:
Jensen, Kim Høj, Holbøll, Joachim, Sørensen (fratrådt), Troels, Nielsen, Arne Hejde