Search

Synchronous-synchronous rotary frequency converters are in the Norwegian traction power system applied for converting three-phase AC voltage at 50 Hz to single-phase AC voltage at 16 2/3 Hz. The converters consist of a three-phase synchronous motor and a single-phase synchronous generator combined on a common shaft. The three-phase motor is fed from the public grid and drives the single-phase generator. The motor-to-generator number of pole ratio is three-to-one, enabling the single-phase generator to generate the voltage at a frequency one-third of that applied to the three-phase motor. This Master's Thesis describes three approaches for modeling single-phase synchronous generators in rotary frequency converters. Model 1 is developed by applying one armature winding combined with rotor windings identical to the three-phase machine's rotor configuration. The equations are used directly, and not transformed to a common reference frame. Test results obtained from the implemented model present a rotary converter model experiencing initial conditions that are destabilizing the converter. Model 2 views the behavior of the armature single-phase winding's pulsating MMF distribution as the result of two fictitious three-phase machines. Each machine induces a rotating MMF distribution. Equations are developed for each machine individually. They are decoupled from each other but interacts with their common rotor circuits. Successful model implementation has not been obtained, due to the simulation not converging to final solutions. Model 3 applies a three-phase synchronous machine with one open-circuited phase and necessary parameter adjustments for obtaining the behavior of the single-phase machine. Test results present the converter model behaving as expected during the loaded steady-state performance. The rate of decay of symmetrical fault current is faster than for sets of parameter adjustments carried out in the literature.

Modern wind farms are increasing in size and complexity. As a result, increased power injected from large offshore wind farms is having a significant impact on the stability and power quality of the grid. Harmonics are a fundamental aspect to be evaluated during power quality assessment. Therefore, a thorough understanding of their behavior is needed to successfully plan an offshore wind farm. In this master thesis, the production and propagation of harmonic components in offshore wind farms are investigated. All relevant theory required to fully understand harmonics and how they can be filtered is included. A guide for simulating the electrical components of an offshore wind farm, such as voltage source converters, cables, and transformers, is also provided. A Simulink model of an offshore wind farm is built using data from Anholt Offshore Wind Farm in Denmark. The wind turbine in this model is simulated as a voltage source converter with an LCL-filter. A d − q reference frame is used in the control system of the converter, with a pulse width modulator to control the switches and a phase locked loop to synchronize the converter with the grid. The turbine itself is connected to a simplified offshore grid with trans- formers and subsea cables. Nine different scenarios (running the turbines at half power, grid harmonics and adding a switch-on delay among others) are simulated on this model, each de- signed to mimic factors present in real offshore wind farms. In addition, active filtering is im- plemented to limit specific harmonic frequencies. Results from all nine scenarios show that harmonics vary greatly in the presence of each simulated factor. There are two kinds of current harmonics produced by the wind turbine; high- order switching harmonics and low-order harmonics produced by the phase locked loop. While most of the scenarios had a constant harmonic output from the switches, the harmonics created in the phase locked loop varied as much as 120%, depending on the conditions under which the turbine is running. While it is documented that the phase locked loop can cause harmonics and instability in the system, this thesis investigates specifically what amplifies harmonics by investigating several practical scenarios in offshore wind farms.

In this thesis work, the modular multilevel converter was investigated. More specifically, an analytical study of circulating current suppression methods for low-bandwidth systems was undertaken. A comparative study of two frequency adaptive digital controllers proposed in earlier research papers has been evaluated to check which yields the most efficient controller. The modular multilevel converter (MMC) is a converter aimed for high voltage direct current applications. Compared to other voltage source converters, the MMC has several distinct advantages; high efficiency, scalability, modularity, and superior harmonic performance. One of the key advantages is its modular design. The redundancy is proportional to the number of submodules, and having multiple submodules in series gives high redundancy. If one submodule were to fail, the converter could continue to operate with high efficiency by bypassing the failed sub-module. However, its modular design does also have some drawbacks. First, the more submodules, the higher switching losses. However, the switching losses can be reduced by lowering the switching frequency. By reducing the switching frequency, the system s bandwidth reduces as well. Consequently, the controller s ability to suppress disturbances reduces. Second, a circulating current is introduced to the system. The circulating current is responsible for transferring power from the dc-link to the ac-side, and also balance the energy of the capacitors. If not controlled, the circulating current will be predominant of a second harmonic, which results in higher resistive losses. The additional resistive losses affect the lifetime of semiconductor devices and passive components due to the temperature rise. To suppress the second harmonic during grid frequency fluctuations, studies of a digital frequency adaptive controller was undertaken. A proportional resonant controller was digitally realised in an infinite impulse response (IIR) filter. By only changing one variable, the filter adapts its resonant peak according to the measured grid frequency. A phase-locked loop was used in combination with a look-up table to measure the grid frequency and to give a new parameter for the IIR filter. A comparative study of two frequency adaptive digital PR controllers has been undertaken. Both methods are used to change the resonant peak of the IIR filter. The first method proposed in [1] focuses on changing ω0, while the second method proposed in [2] focuses on changing the sampling time. Both controllers were simulated in a system where the grid frequency changed from 50 Hz to 52 Hz, resulting in a 4 Hz jump in the second harmonic. A fast Fourier transform (FFT) analysis reveals that both methods work with high efficiency. However, in medium- and low-bandwidth systems the method proposed in [1] was able to suppress the second harmonic more effectively.

More than half of all hydro generator failures are caused by insulation breakdown. Insulation failures inflict serious damage on the generator and represent significant operating costs for utility companies. One of the main reasons for insulation breakdowns is the irreversible degradation caused by internal partial discharges (PD) resulting from voids in the insulation system. Therefore, it is crucial that methods for condition assessment of generator insulation can identify and analyze the partial discharge activity in these voids. The main purpose of this thesis is to facilitate condition assessment of generator bar insulation by measuring partial discharges resulting from voids in insulating materials. The aim is to clarify how important PD parameters relate to the void size and the discharge mechanisms in the voids. A part of the work also involves comparing the experimental results with the theoretical model of internal partial discharges: the Abc Model. In addition, possibilities and limitations of partial discharge measurements as a diagnostic tool are examined. The test samples consist of both real generator bar insulation and specially designed laboratory samples containing disk-shaped voids of different diameters. A thin layer of aluminum was applied by vacuum evaporation to the cavity surfaces of one sample group to investigate the effects of increased surface conductivity. Electrical detection of partial discharges was performed using a conventional measuring circuit, and the PD activity was analyzed in the form of inception voltage, as well as the discharge magnitude and discharge frequency at a voltage frequency of 50 Hz. The inception voltage is found to decrease with increasing void diameter due to lower field enhancement in larger voids. In the case of the laboratory samples, the discharge magnitude is generally increasing with increasing diameter. The discharge magnitude is larger when the void surfaces are conducting since the discharge area then is equal to the void surface area. However, for samples made of aged generator insulation, the discharge magnitude is constant regardless of void diameter. More importantly, it was not possible to conclusively distinguish the PD activity in the voids from the PD activity inherent in the insulation. The discharge frequency tends to increase with increasing void size. This can be explained by the differences in the electric field strength in the different cavities at the specific voltage level, and the occurrence of parallel discharges. Higher void surface conductivity generally leads to lower discharge frequency. In conclusion, the PD activity is seen to depend on the void size. The theoretical model is successful in describing the PD activity in voids with conducting surfaces, but fails to describe the PD activity in aged generator insulation. PD measurements performed on the laboratory samples can detect the voids and assess the relative void size. However, the voids in aged generator insulation cannot conclusively be detected using the chosen PD approach. This represents an important limitation of the PD method.

The growing need of battery energy storage to support the increase in renewable production in the electricity mix has brought the attention to the modular multilevel converter (MMC) for medium voltage applications. Modular multilevel converters have been verified the most promising battery interface as it has very high efficiency, excellent AC waveforms, scalable and modular structure and it allows for the use of low rating semiconductors. In addition it provides redundancy due to the ability of distributing the batteries into each submodule. A drawback with the MMC, is that when half-bridge submodules are used, the converter can not handle DC faults, hence it is necessary to understand the dynamics of the converter during various faults. In this thesis a MMC interfacing a 1 MW battery to the grid is built in \textit{Matwork's Matlab/Simulink}. Both the converter and control system is evaluated and designed in order to evaluate post-fault currents and voltages. The converter was designed with a DC-voltage of 15 kV and five submodules in each arm. The degree of complexity chosen for this model is the submodule level switched model type, including IGBT, reverse diodes and SM capacitors. A high carrier frequency and phase-shifted pulse-width modulation eliminates the need of submodule energy balancing. Five short circuit fault locations have been investigated, two inside the submodule, over one arm, on the DC-link and between two phases on the AC side of the converter. No fault handling strategies, AC- or DC breakers are implemented in the model. The simulation of short circuit faults inside the submodules reviled that five submodules is not enough to ensure redundancy. The importance of developing fast responding DC breakers was demonstrated as the upper converter arm on phase \textit{a} was short-circuited. As this fault only includes one phase, the first post fault zero-crossing appears to late to prevent damage on the converter. DC breakers are also necessary to prevent the extremely high discharge current from the battery to the fault during the short circuit fault on the DC-link. During the short circuit fault between two phases, all semiconductors will be damaged by the large current if no fault handling is applied, but on the other hand the zero-crossing appears fast enough to isolate the converter from the grid before the damage is done. The most critical faults are the short circuit faults across one arm and across the DC-link because they both requires DC circuit breakers, which is not yet commercialized.

Variable flux permanent magnet synchronous machines (VF-PMSMs) use a novel operation concept based on changing the magnetization level of the permanent magnets in accordance with variations in speed and torque. This may improve the efficiency of wide-speed operation, reducing or removing the need for flux weakening current to counter-act excessive back-emf at high speeds. Magnetization change in these machines is achieved through short d-axis current pulses into the stator windings. The technology has gain increased attention the last few years, having numerous publications attempting to solve the various challenges related to design and operation. This paper will first give a conceptual introduction to VF-PMSMs and briefly summarize some of the literature published on various designs. Then, partly based on the findings of others, two simple VF-PMSM rotors are designed and built to fit a pre-existing stator. A rotor built with only strong samarium-cobalt magnets had its viability as a VF-PMSM experimentally tested. Results show that the magnet material used is stronger than ideal and that laminating the rotor may be crucial for functional VF-PMSM designs.

In this text three different solid state circuit breakers (SSCB) for medium voltage direct current (MVDC) systems have been investigated, namely the interrupting, the current limiting and the resistive topologies. In particular, design strategies for each breaker topology has been proposed emphasizing the optimal design of passive elements with regard to different design goals. These design goals include the minimization of total breaking time, maximum fault current, ratings and required number of power semiconductor devices, size of passive components, on-state losses and transient response during no-fault conditions while at the same time maximizing the reliability and controllability of the system. The breaker topologies designed using the proposed strategies have been simulated for a variety of cases to investigate different design trade-offs. The trade-offs found in the results shows how important it will be to design the breakers and optimize them according to the specific application to which it is to be used. Furthermore the improvement of important parameters with regard to applied power electronic devices have been investigated. How these improvements could potentially affect the performance of the breakers is an important result of the performed analysis' and simulations. The effects of external parameters such as the sensing and coordination equipment and strategies are also discussed and showed to be of great importance. The findings of the literature study performed in context to this thesis will be shortly summarized in the theory part. These findings include topics on general MVDC systems, its applications and fault mechanisms as well as solid state circuit breakers found in literature. Relevant theory of power semiconductor devices and the modeling of RLC circuitry will also be included. Analyzing and designing the different breaker topologies will be the main part of this text. The topologies are analyzed using state-space representation and design strategies are proposed for each topology. These design strategies also sheds light on different limitations regarding available technology. To verify the proposed design strategies, a simulation of a simplified MVDC transmission system has been designed in Simulink. Using this system, all the topologies have been simulated using comparable parameters derived from commercially available technology. The results of the simulations are compared and any peculiarities discussed. In particular the feasibility of the different topologies for different applications are discussed. Specific suitability between topologies and applications are discussed as well as possible novel improvements of the topologies. A possible novel topology integrating the current limiting circuit breaker with a buck converter is proposed as a result of the the performed analysis. This however without further in-depth investigation of proposed topology.

Gas insulated components are important in the modern transmission system. The most common gas used in high pressure insulation is Sulphur Hexafluoride (SF6). SF6 has beenclassified as a greenhouse gas and a potential replacement is CO2, either by itself or in mixtures with other gases. The insulation properties of gas insulation depends on several parameters, e.g. surface roughness. An electrode with a built-in precision motor was used in a quasi-uniform gap. This thesis aims to investigate the impact of needle protrusions with lengths in the range 0.05 – 2 mm in SF6 at 4 bar absolute pressure and CO2 at 6 bar absolute pressure. Another aim is to investigate the applicability of the enlargement law. The results with a single needle were used to create predictions, according to the enlargement law, for 20 and 100 identical needle protrusions. The electrode used allowed needle arrays of 20 and 100 needles to be tested with needle lengths in the range 0.05 – 2 mm. The enlargement law was also investigated for two sandblasted surface areas. The results in SF6 showed a decrease in the 50 % average background breakdown field, E50, with increasing needle lengths for both polarities. With 20 needles E50 decreased from the single needle results and agreed well with the predictions for both polarities. With positive polarity and 100 needles, there was no significant change in the results between the 20 and 100 needles and there was a discrepancy with the predictions. With 100 negative needles there was a good agreement between predictions and results, showing a decrease from the results with 20 needles. The sandblasted surfaces showed a decrease from the small area (2.383 cm2) to the large area (64.465 cm2). The reduction in E50 showed good agreement with predictions with negative and disagreed significantly with positive polarity. The obtained results in CO2 showed an insensitivity to single needle protrusion with needle lengths shorter than 1 mm with positive and 0.75 mm with negative polarity. With 20 needles, there was an increased sensitivity and reduction for both polarities. E50 with positive polarity decreased with increasing needle lengths. Negative polarity had a significant decrease of E50 from 0.1 to 0.5 mm where it seemed to saturate between 0.5 – 2 mm needle lengths. For 20 needles both polarities agreed well with predictions. With 100 needles, there was no longer an agreement between predictions and results for either polarities. The results with positive polarity decreased further. With negative polarity similar behavior as with 20 needles was seen, where the saturation occurred at 0.2 mm needle lengths and ended at the same field strengths as with 20 needles. The sandblasted surfaces showed much more of a decrease than what was predicted based on the small area. The experimental results were analyzed with Turnbull’s algorithm to determine the empirical cumulative distribution function, which was fitted with the three parameter Weibull distribution. From the distribution functions the 2 % average background breakdown field was found. Possible physical explanations and breakdown mechanisms of the results were discussed. The results showed that SF6 was more sensitive to the needle protrusion than CO2. The main conclusion is that the enlargement law must be used with care. It is possible that the mechanisms involved in breakdowns scale differently than just with increased number of needles or area and require further investigation to understand.

Microgrids are low voltage distribution networks comprising various distributed generators, storage devices and controllable loads that can operate as a controlled entity, either interconnected or isolated from the main distribution grid. Successful operation of microgrids can increase the reliability and efficiency of the energy system, lower operational costs and facilitate the implementation of renewable energy sources. However, long-established regulation strategies used in conventional power systems might not be feasible in microgrids, due to the new system configuration, the variety of resources utilized, and the considerable presence of power electronics. A control system for a microgrid model has been developed. The model consists of a generating unit (PV-array), a storage unit (battery), a distribution system and loads, as well as the power electronic interfaces between the AC- and the DC-components. The control is performed through inverter control of the microgrid units in a master-slave structure. The control functions implemented are current control, power control, voltage control, and maximum power point tracing of the PV-array. The control system was developed using the rotating dq-frame coordinate system, and implemented in the MATLAB/Simulink simulation tool. Most simulations concerned events that spanned less than 3 seconds, and it took approximately 5-10 seconds to perform the simulations. The control system was shown to enable optimized efficiency of the PV-array, while at the same time regulate the amplitude and frequency of the load voltage in the presence of a variable load condition and a variable supply. The control system was feasible during both grid-connected and islanded operation, and was able to adapt to islanding events with the microgrid stability preserved. An advantage of the master-slave structure utilized is that it enabled reference tracking with zero steady-state error. A disadvantage with the strategy is the need for two distinct controllers on the master unit (the battery), with following dependence on proper coordination of battery controller modes. The application potential of the developed simulation model was demonstrated. Simulations concerned transient dynamics during important events, such as the islanding of the microgrid during power exchange with the utility, and islanding that did not coordinate with the microgrid master controller. It was also demonstrated how the current controller time constant affected the battery's voltage reference tracking ability. As the time constant increased, the reference tracking ability became more sensitive to the load conditions.

With the beginning implementation of a smart grid in industrialized countries increasing the load flexibility of customers, as well as communication with the grid operators, a potential market for bidirectional on-board chargers (OBC) is emerging. The OBC is the electric vehicle charger located within the vehicle, operating as a link between the DC-voltage battery and the AC grid voltage in electricity sockets anywhere, thus working at a universal voltage range (85Vrms-265Vrms) with rated power in the range of 1.9kW to 3.5kW. To maximize the real power drawn by these OBCs, they are designed with power factor correction (PFC) converters at the AC grid interface, shaping the drawn AC current sinusoidally and in phase with the AC voltage, to maximize the power factor. Thus, a bidirectional OBC would necessarily require a bidirectional PFC. In this master s thesis, a simulation model is developed for a bidirectional totem-pole PFC, attempting to verify the use of this topology as the mentioned bidirectional OBC PFC. The work is focused on describing the basic bidirectional operation of the bidirectional totem-pole PFC, deriving expressions for currents, voltages and losses, as well as for the gate logic and control system. Average current mode control is used, with cascaded current and voltage control including PI controllers and Symmetrical optimum tuning. To verify the simulation model for the power levels and demands of an OBC, the main components are dimensioned to achieve 98% efficiency at 230Vrms, 3.5kW and 90kHz switching frequency, implementing GaN eHEMTs in the fast-switching leg of the totem-pole. At last, simulations are performed, showing that the PFC is able to achieve a power factor of 0.998 and a THD of 5% in both G2V and V2G, without a DM filter on the AC side.

As a result of the continuously increasing carbon emissions, the global demand for cleaner energy is rising. Grid-connected photovoltaic (PV) systems is an attractive solution to provide green energy, and PV technology has therefore gained increased attention which has resulted in higher efficiency PV panels. PV plants have benefited with increased annual energy yield, and the never before has solar energy contributed with more new power capacity to the power grid. However, variable power production due to unpredictable weather conditions pose a challenge for optimal planning, grid operation, and grid stability. A replacement of the classic inverter with Modular Multilevel Converter (MMC) got introduced as an attractive topology for PV plants due to a high scalability that facilitates a direct connection to medium voltage grid without the use of a step-up transformer. The required voltage level is achieved by designing the MMC with the correct number of submodules. It also provides high MPPT tracking, which gives the MMC several advantages over conventional PV inverters. The isolation in such a configuration can be provided between the PV strings and the submodules of the MMC by utilizing isolated DC-DC converters. Different isolated full-bridge DC-DC converter configurations with high efficiency are investigated in this study. The single-unit full-bridge configuration provides a simple structure, but requires high voltage and current ratings due to the high power transfer. By utilizing a parallel- or series-connection of converters, the bulk transmission of power to the submodules is distributed, which increases the overall reliability of the system. If the PV string operating voltage is not equal to the submodule capacitor voltage, then a step-up in voltage is necessary. However, the PV strings are in this study designed such that the HF transformer turns ratio is 1:1 in all configurations. The parallel-arrangement avoids the parallel-connection of switching devices in the DC-DC converter in order to handle the high currents. SiC MOSFETs have been utilized in the full-bridge due to the high voltage rating. Simulations have been performed at different DC-link voltages, at many different switching frequencies and for the entire operating range of 0.1-1 per unit power in order to find the efficiency. A 50\% reduction in the current through each full-bridge results in a 50\% reduction in the total copper losses. The copper losses are proportional to the square current, and by reducing the current by 50\%, the total copper losses is reduced by 50\%. However, the voltage rating is the same when assuming an unity voltage transformer ratio. In case of the series-arrangement, the voltage stress on the switching device is reduced considerably. Due to the reduced voltage, an utilization of GaN devices is possible. GaN switching devices have a lower voltage rating than SiC. However, they have shown a superior performance, especially at high-frequency operation. High-frequency operation is beneficial due to reduced size of the HF transformer. The efficiency of the GaN-based series-connected full-bridge DC-DC converter was obtained by calculations of the switching and conduction losses by assuming a worst-case duty ratio of 0.5. The results gave a high efficiency of 98\% for a operating range up to 200 kHz switching. The obtained efficiency curves were utilized in a simulation model in order to find the overall energy efficiency and the levelized cost of energy (LCOE) of the MMC PV plant configurations. These configurations were also compared to a central inverter (CI) PV configuration and a multi-string central inverter (MSCI) PV configuration in order to state the benefits of utilizing the MMC as inverter in medium voltage PV plants. The MMC topologies, as well as the MSCI topology, provides lower MPPT losses due to the distributed MPPT in comparison with the CI configuration. The MMC configurations had a better energy efficiency for all the different DC-DC converter configurations compared to both the CI and the MSCI configuration. For the MMC utilizing the single-unit DC-DC configuration, the energy efficiency was obtained to be 95.28\% for the best-case switching frequency at 1 kHz, and 94.40\% for the worst-case switching frequency at 20 kHz. The CI and MSCI efficiency was in comparison 89.63\% and 90.11\% respectively. However, the MMC configurations did not account for losses in the HF transformer in the DC-DC configurations. It was assumed that these losses do not account for a great part of the total losses in the PV plant, and the results would have been almost the same when accounting for them. For the MMC utilizing a parallel-connection of DC-DC converters, the energy efficiency was obtained to be 95.25\% at 10 kHz switching. In the MMC utilizing the GaN-based series-connection of DC-DC converters, the energy efficiency was calculated to be 95.01\% at 20 kHz switching, and 94.39\% at 200 kHz switching. The levelized cost of energy (LCOE) was also found for all the different PV configurations. The LCOE was shown to be 6.25\% lower for the MMC utilizing the single-unit DC-DC converter compared to the CI configuration, and 6.21\% and 5.76\% lower for the MMC utilizing the parallel- and series-connection of DC-DC converters respectively. This shows the great benefits of utilizing the MMC instead of the conventional PV inverters. Another important aspect is that partial shading is not considered in these simulations. The MMC configurations will increase the MPPT substantially, resulting in a even higher extracted energy compared to the CI configuration during partially shaded and non-equalized irradiance conditions in the PV plant. The parallel- and series-connection of DC-DC converters increases the MPPT more than the single-unit DC-DC converter configuration. This will result in a substantially decrease in the LCOE of the MMC PV configurations compared to the CI configuration. A higher number of series- or parallel-connected DC-DC converter may result in higher efficiency and lower losses, but at the cost of a more complex system and a higher price due to more components. The number of series- and parallel-connected devices must therefore be selected with care. The simulations on the proposed MMC PV plant configurations verified the improved energy efficiency and the lowered LCOE when utilizing MMC over the CI and the MSCI configurations in medium voltage PV plant applications.

As the share of variable renewable energy in the power system rises, the demand for balancing technologies to compensate for fluctuation in production increases. Pumped-storage hydropower has significant potential globally and is seen as the most mature storage technology available. Modern pump-storage plants utilize a power electronic interface between the machine and power system to control power consumption in pump mode. Introducing power electronic converters between the machine and grid causes the rotating mass of the machine to be independent of the frequency in the power grid. Therefore, the machine loses its ability to contribute to the inertia in the power system. To solve this, the converters can be controlled such that the effect of inertia is synthesized, often termed synthetic inertia. A number of methods to emulate inertia exist, with a broad span in complexity and design. The most straightforward solutions release stored kinetic energy based on the frequency derivative or deviation. Other more complex solutions mimic the behavior of a synchronous machine, often referred to as virtual synchronous machine (VSM). The VSM method enables the converter to operate as a grid-connected synchronous machine with a speed governor and excitation system. In this thesis two methods are developed; classical torque control with supplementary inertia controllers and virtual synchronous machine control. The methods are tested in a simulation model consisting of a converter-fed hydro-electric power plant connected to a simplified grid. It is shown that both methods performed as intended, emulating the effect of inertia during a power imbalance event. The classical torque control and VSM control improve the frequency nadir by 36.4% and 54.5% respectively based on the steady-state frequency for a given test case. The VSM method is found superior in the context of inertia support due to fast power control. Also, the VSM inherits a number of other advantages such as the possibility for dynamic inertia support. Dynamic inertia support allows the VSM to change the amount of synthetic inertia in the grid during operation, which can improve small-signal stability.

The generator is a critical component in the hydropower power plant, and detection of its fault conditions is important for safe and cost-effective operation. A new proposed method for detection of rotor interturn short circuits (ITSC), based on detecting harmonic sidebands produced around the main harmonic (the 50[Hz]) and its third harmonic (the 150[Hz]) when there is a fault on the generator, might be used for detection of eccentricity fault conditions on hydro generators [29]. This master thesis investigated the ability of the rotor interturn short circuit detection method to detect eccentricity and differentiate between eccentricity and rotor interturn short circuit faults. The thesis also investigated the method s capability to detect rotor ITSC with static eccentricity. The method was investigated through simulations using the finite element simulation software Ansys Maxwell. The simulations has been made on a model based on technical drawings of a real generator, and three models with slight modifications in generator topology. The results of the simulation showed that the method is highly dependent on generator topology, and is unsuited for some generator topologies. For the cases it is suitable it can detect rotor ITSC with static eccentricity even at low levels of fault severity. Dynamic eccentricity can be detected as it entered critical levels. Mixed eccentricity can be detected, but some critical levels of the fault can go unnoticed by the method. The method is unsuited for detection of static eccentricity. There has also been found a way to differentiate between eccentricity and rotor interturn faults based on the amplitude of two of the harmonic sidebands.

This thesis delves into the two fields of signal analysis and small-signal stability of power systems. Due to the extensive deployment of PMUs in today s Smart Grid, new possibilities arise to assess the small-signal stability by measurement based techniques. The benefits of this approach are many, the most compelling being that the power system has become too complex to be accurately modelled with a component-based approach. A whole range of methods are available for the purpose of signal analysis in power systems, all with various assumptions, strengths and weaknesses. Insight into the analysis methods and the measurement data in itself is important for choosing the appropriate techniques for a given signal. The methods described in this thesis are: Prony s method, Robust Recursive Least Squares (RRLS) and the FFT based Welch s method. Prony s method is known as a ringdown (post-disturbance transient) analyzer, while the two latter are known as ambient analyzers. To improve estimation robustness and accuracy, pre-filtering is implemented with the Empirical Mode Decomposition (EMD). This technique works as a non-linear and non-stationary band-pass filter, effectively extracting the desired frequency spectrum in the electro-mechanical range. This is validated using Welch s method, which reveals negligible change in the power spectral density of the investigated frequency range. Both Prony s method and RRLS assume an underlying parametric model, and require specifying a model order. This issue is transformed into a benefit, extracting the consistent information from multiple analyses with varying model order. Clustering is used as post-processing intelligence to identify the dominant modes of the estimation. The third technique, Welch s method, is used for validation purposes. The methods and data are contextualized; subsequently, the techniques are described with relevant theory, and thoroughly tested for both simulated and real-world data. The estimation from ringdown data is compared to the estimation from ambient data, which in real-time scenario often is the only option for evaluation. The three analyzers contribute to mutual validation of the modal content, improving the credibility of the estimates. For simulated data, the measurement based estimation is also evaluated against eigenvalue inspection of the linearized power system model. This thesis is based in part on a conference paper to be published at SPEEDAM 2018. This article describes theory of Prony s method, as well as the EMD and Clustering technique. Since then, the thesis has evolved in scope and depth. The results from detailed testing show that the investigated combination of methods performs well on real-world PMU-measurements. Prony s method identifies the modes of ringdown data and the RRLS method identifies the modal content in the ambient data. The damping ratio is in general slightly underestimated in the ambient data compared to ringdown data. However, both give a good indication of the modal content. A comprehensive toolbox of the methods mentioned above has been indigenously created (in Python) as a part of thesis work; the toolbox utilizes weaknesses of some methods, e.g. the model order selection problem in Prony s method and RRLS, as input to other methods, such as the clustering technique. The result is an autonomous algorithm, taking care of each submethod s deficiency.

A framework for loss analysis using COMSOL and LiveLink for Matlab is presented in this thesis. The loss analysis is based on Steinmetz' theory on iron core losses. IEC standard 60404-8-4:2013 gives a method of presenting loss data for laminated cores in W/kg, when a sinusoidal flux density is assumed. The variables in the standard is the frequency and the peak of the flux density. These data form the basis for the loss analysis when the iron core is used in an application, such as an electrical machine. The data are used to create functions with the same variables as the standard contains, such as the Bertotti equation. The method for loss analysis is based on code written by Havez et al. The method uses a simulated model in COMSOL and fetches the data from the iron core of the model. The simulation needs to include a sweep through the time domain to get the sufficient field distributions. These data are post processed in Matlab. Using the FEM elements, the frequency and the peak of the flux density is calculated for each individual element, so the loss distribution can be found together with the total losses. It is recommended to use the average of the flux density in every corner of the element instead of the mphinterp function to find the typical flux density of the element. The mphinterp function is proven to be slow and having issues with rotating machinery. The peak of the flux density for an element is the maximum of the typical value over a time sweep, when the DC offset is removed. The frequency is calculated by using the Fast Fourier Transform. It is presented a method that omits the problem with the FFT function being dependent on how much of a signal is included in the function. The method is to carry out multiple FFT analyses on different lengths of the flux density signal and to use the median of the last part of the frequencies calculated. The method is consistent if at least four periods of the length of the flux density vector signal is included in the analysis. If the time used on the loss analysis should be limited, the effort should be put on the COMSOL simulation, rather than reducing the post processing time. This is because the post processing time is significantly shorter than the simulation duration in COMSOL. The loss analysis must be carried out on at least 6/14th of a rotation to avoid large deviations above +/- 11\% from the loss result from a full rotation. A method for extraction of the relative permeability from a BH curve is presented. The method is based on linear regression and use the slope of the BH curve to calculate the relative permeability. A loss analysis based on a model where relative permeability is used is carried out. This analysis shows that the entirety of the BH curve must be included in the linear regression, if the results should be close to the results obtained using the BH curve.

The world is critically dependent on a transition from fossil fuels to renewable energy. Microgrids represent a promising systematic approach to increasing the reliability of the power system and solving the challenge of integrating intermittent renewable energy sources into the existing grid. Several renewable energy sources, such as wind and photovoltaic, utilize power electronic converters, representing distinctly different characteristics compared to the technology used in the conventional power system of today. Among several traits, the converter's inherent ability to compensate harmonics is a cost-effective solution to meet the persistent challenge of harmonic distortion, which is a typical power quality issue in microgrids. However, uneven line impedances between the converters and the PCC lead to imbalance between the harmonic currents supplied by the converters, potentially harming the converter supplying most of the harmonic load. This Master's thesis investigates a harmonic sharing scheme to meet the challenge of harmonic current sharing imbalance. A simple AC microgrid microgrid in island mode is developed, consisting of two converters, their individual LCL filters, uneven line impedances and one non-linear load. In addition, the control system of the microgrid's converters is developed and elaborated. It consists of primary control, including current, voltage and droop control, as well as control measures to achieve harmonic compensation of the capacitor voltage for both converters. This way, the issue of uneven harmonic current sharing is displayed, as the converter facing the lowest line impedance supplies the majority of the harmonic load. A virtual impedance-based harmonic sharing scheme is then implemented in the control system of the converter facing the lowest impedance, to achieve better sharing of the harmonic load. Specifically, the sharing of the 1st, 5th and 7th harmonic current component is studied. The implemented harmonic sharing scheme is displayed with simulation results in a case study, proving the feasibility of the elaborated control system. The 5th and 7th harmonic currents are observed to be shared almost perfectly, while the fundamental component sharing is improved substantially, but still has a value of 0.033 p.u. in constant offset. Additional key performance indicators (KPIs), namely apparent power, THD and load parameters, are simultaneously monitored to give a broad assessment of the harmonic sharing scheme. In this case, all KPIs are deemed adequate, resulting in assessing the harmonic sharing scheme as successful, but with room for improvement in terms of fundamental current sharing. Three additional cases are displayed to evaluate the control system's resilience towards certain system alterations. The first of the three additional cases investigates harmonic compensation of the PCC voltage. Isolated, the scheme is successful, as the THD of the PCC voltage is lowered from 16.2\% to 8.20\%. However, the system as a whole is not compatible with this configuration, ending in unsatisfactory performance of all other KPIs than the THD. The second of the additional cases studies the effect of implementing opposite droop compared to a conventional droop scheme. The difference in system responses between the two implementations is observed to be marginally different, with the opposite droop delivering an improvement of fundamental harmonic current sharing. Specifically, opposite droop achieves a mismatch in fundamental current sharing of 0.013 p.u., compared to the corresponding number of 0.033 p.u. for the conventional droop. The last of the additional cases simulates the connection of a linear load to the PCC. The harmonic sharing remains efficient, showing only a decrease in performance for the fundamental current component. However, the load experiences large ripples. E.g., the voltage has a ripple of approximately 131 V, constituting a suboptimal loading scenario. Hence, as a whole, the control system is not compatible with connecting an additional linear load between two phases. Summarized, the harmonic sharing scheme proposed and developed demonstrates efficient harmonic sharing. However, it has been found that the proposed scheme may reduce its effectiveness when system conditions change substantially. Hence, recommendations for further work are provided which include enhancing robustness of the harmonic sharing approach.

This thesis focuses on power sharing in low-voltage, islanded ac microgrids. It is motivated by the challenges related to climate change. To achieve a successful integration of renewable energy resources in the power grid, it is likely with an increased use of small distributed generation units. A promising way of organizing these units is by the use of microgrids. With microgrids disconnected from the main grid, several challenges related to current, voltage and power flow control arise. These challenges are studied in this thesis, where a microgrid with two distributed generation units and one load is investigated while operating in island mode. The two distributed generation units in the system are implemented with significantly different output impedances, in order to examine the unequal power distribution. Distributed control systems are developed for this system, especially focusing on the use of virtual impedances for improved power sharing. Proposed virtual impedance methods are presented. These methods reduce the effect the virtual impedances have on the voltage level. Three different cases are considered, each of which divided into two subcases. In case 1, the microgrid system is assumed to have predominantly resistive output impedances. The subcases 1 a) and 1 b) consider the use of opposite- and conventional droop control algorithms for controlling the voltage and frequency levels. The microgrid considered in case 2 is similar to the one in case 1, but with predominantly inductive output impedances. As in case 1, the conventional and opposite droop control algorithms are utilized in case 2 a) and 2 b), respectively. Case 3 is a reproduction of case 2, but in the selection of virtual impedances, the physical output impedances are estimated to be 25 % lower than the actual values. Simulation models are carried out in all of the cases by utilizing the software MATLAB/Simulink. Each simulation case considers the use of physical output impedance only, an existing virtual impedance method and a proposed virtual impedance method. The proposed virtual impedance methods show promising results in all cases, where the power sharing performances are the same as for the existing methods. The voltage drops due to the use of virtual impedances are however significantly reduced. All of the simulation cases show a reduction of at least 90 %. This master's thesis is a part of the research project FME CINELDI, and the results of this thesis are planned to be verified through laboratory experiments.

For some high power applications, as for instance in adjustable speed hydro, ASH, synchronous machines are used for power generation. In this master`s thesis, flux models and sensorless control methods for a synchronous generator with damper windings, developed in the specialisation project autumn 2017, has been carefully analysed. A drive is classified as sensorless when no sensors are used for measuring the position or speed, but the position and speed are rather estimated based on the measured voltage and current supplied to the electric machine. Higher reliability, higher robustness, and a reduced size and cost are motivating factors for running electric machines sensorless. The main focus has been on sensorless control at low speed operation, and on parameter sensitivity of the flux models. Steady state analysis of the current model and the voltage model, which are used to estimate the stator flux linkage, were performed by analytic expressions in Matlab, while the dynamic analysis and the analysis of the sensorless operation were performed by simulations in Simulink. Estimated stator flux linkage and estimated rotor position was used as input to the control system. It was illustrated that the voltage model is sensitive to the estimated stator resistance during low speed operation, and that the sensitivity increases with the generated torque. During steady state operation, only the magnitude of the estimated stator flux linkage was affected by the erroneously estimated resistance, while the angle of the estimated stator flux linkage was unaffected due to the unity power factor control. During dynamics, an erroneous estimation of the stator resistance made the flux linkage estimate drift, resulting in oscillating errors in both the estimated stator flux linkage amplitude and in the estimated stator flux linkage angle. The concept of stator flux linkage drifting, the consequences of it and methods to compensate for it, have been thoroughly explained and illustrated. It has been illustrated how the voltage model is less sensitive to the stator resistance at higher frequencies, and that the speed at the instant when the drifting occurs determines how much the flux linkage will drift. Once the flux linkage has drifted, the offset established is not reduced by increasing the speed, and hence there is a need for drift compensating techniques. Sensorless control without drift compensation did not yield satisfying control, and instability occurred frequently. Changes in torque reference, changes in stator resistance and driving through zero speed are typical scenarios that trigger drifting. It was found that the current model is more sensitive to the q axis inductance while generating a high torque, while the sensitivity to the d axis inductance was highest at lower torques. Steady state analysis revealed that the current model is not completely independent of the speed, as field weakening affects the sensitivity of the model. When using drift compensation, the machine was able to operate sensorless at low speeds, and to drive through zero speed. The generated torque followed its reference less accurately at low speeds, especially when driving slowly through zero speed while generating a high torque. Providing a torque at standstill for a longer time was not successfully achieved. Finally, attempts were made to improve the sensorless control by combining the flux models in different ways, to obtain a less parameter sensitive closed loop observer. It was concluded that the closed loop observer eliminates the need for an external drift correction method. By further modifications, better flux estimates were obtained for low speeds, both during the first step in torque reference and through zero speed, assuming that the inductances are known with high accuracy.

This thesis is centered around the possibilities of utilizing the active filter capability of a Voltage Source Inverter (VSI) to mitigate harmonic distortion in a marine power system. Hotel loads in marine vessels constitute an increasing part of the installed power onboard, as ships are increasingly electrified. Hotel loads represent the loads that do not contribute to the propulsion of the vessel, e.g. lighting or cranes. The system studied in this thesis consists of a power supply supplying hotel loads in the MW range. The power supply, which is connected to the internal DC-grids of the ship, consists of a VSI and a LCL-filter. The problem is related to the highly distorting hotel loads propagating harmonic distortion into the system. The harmonic distortion is mainly produced from the diode rectifier stages in Variable Frequency Drives (VFDs). The VFDs are used to control AC-motor torque and speed by varying frequency and voltage at the motor terminals. The main objective of this thesis is to ensure a voltage waveform at the terminals of the power supply within the requirements on Total Harmonic Distortion (THD) set by Det Norske Veritas (DNV). The worst-case THD obtained in the system exceeds the requirements. Measures must be taken in order to lower the harmonic content at point B and fulfill the requirements. An active harmonic filter integrated to the control of the VSI was developed in order to mitigate the harmonic distortion. Three cases have been simulated, with different degrees of harmonic filtering. The applied active filter features great performance and obtains a low THD well within the demands. However, it comes with a price in terms of increased power losses and the current rating of the components. Efforts were therefore made to decreasing the power losses and the system currents. Case 2 was devoted to finding the minimum power losses when still operating within the THD requirements. In the last case, a proposed degree of harmonic mitigation was simulated. This case yielded a worst-case THD well within the requirements, while still keeping both power losses and system currents within acceptable limits. It was concluded that the active filter capability of the Voltage Source Inverter is sufficient to mitigate the harmonic distortion in the system. However, the results are strongly dependent on the parameter data; further research should therefore put more efforts into gaining more precise estimations of these data.

The recent commercialization SiC technology, has made it desirable to use SiC in electrical motor drives for marine applications, due to the superior material qualities of SiC compared to conventional Si based technology. This Master thesis will present an analysis of the switching performance of a SiC MOSFET module from ROHM Semiconductor, and an accompanying investigation of how a high performance SiC based motor drive can be realized. As part of the thesis work, the many critical components of an electrical motor drive system was investigated, and this will be described in this thesis. Two printed circuit boards were produced to enable the SiC modules to be operated using a motor controller. In addition, a low inductive three phase DC-bus was designed, produced and tested. The three phase DC-bus design will be described through theoretical considerations, before an experimental analysis of the stray inductance of the DC-bus is presented. It will be shown that a low inductive design can be realized through the use of planar copper conductors, and the DC-bus design has good inductive capabilities compared with a single half-bridge DC-bus which was used in the experimental SiC characterization. The SiC module performance characterization was conducted by performing double-pulse testing, both in simulations and in laboratory experiment. It was shown that the presence of stray inductive elements in the circuit causes voltage overshoots up to 110 V over the nominal voltage of 600 V and long lasting ringing. The switching characteristics calculated from these tests will be presented, and the circuit dynamics giving rise to these effects discussed. A comparison between the simulation results and experimental results was made, and it will be shown that while the simulation model captures some switching characteristics well, it fails to model the large voltage overshoots observed in laboratory experiments. Further more, it was investigated if switching performance could be improved through the introduction of snubber circuits, in particular RC turn-off snubbers and DC-snubbers. Both the simulation results and the experimental results showed that the RC turn-off snubbers failed to remove the overvoltages completely, but was able to reduce it to 35 V with a snubber damping constant of 0.4. This comes at the cost of increased switching losses however, and it is clear that a trade off must be made.

Deep sea mining (DSM) is an emerging technology. The ever-increasing world-wide need for minerals in growing markets and industries of developing countries has made DSM financially reasonable. As there are few known system concepts for DSM it is important to investigate the different possibilities of system design and operation of the future DSM production systems. This thesis has, by investigating R&D and system concepts that are under development, presented different solutions for the electrical distribution system of a DSM production system capable of operating in the Norwegian Sea. In this thesis is it assumed that the DSM production system will consist of three subsea mining machines (referred to as seafloor production tools (SPTs)) designed to excavate, process and collect ore at the seafloor. The electrical power system is estimated to operate at depths up to 3,500m, where the production system will mine seafloor massive sulfide deposits formed through hydrothermal activity. A positive displacement pump, powered by pressurized seawater from the topside installation, is used for all the proposed system topologies. Firstly, simplified models where made in order to investigate three different system topologies of powering the SPTs. In the first model the different components of the SPTs where to be powered individually, thus reducing the number of subsea components by placing the frequency converter of the components at the TS. The simulation proved that powering the SPT components individually would result in excessive power losses, thus ruling out this system topology. In addition the machines of the SPTs perform better when the converters are installed on board the SPTs. Two other system topologies where proposed. The first is powering each of the three SPTs individually from the topside installation. The second is having a combined power distribution to a subsea distribution station which distributes the power to the SPTs. Both these system topologies where deemed suitable for powering a DSM system. Therefore, four simulation models where made in order to investigate the possibilities of power distribution to a DSM production system. The simulation software PowerFactory is used. Two models utilizes AC distribution to the subsea components of the production system, and the two other models use DC distribution. The AC models have a distribution voltage of 6.6kV operating at 60Hz and the DC models operate at +/- 3kV. The voltage levels are chosen in order to obtain a realistic comparison between the two. The simulations performed in this thesis proves that the distribution losses will be lower when powering the subsea mining machines through a subsea distribution station compared to the individual power distribution. In addition will the losses be even lower if DC distribution is utilized. By reducing the power losses the production system will be less expensive to operate due to lower fuel consumption. Harmonic content will also be an issue for the proposed AC system topologies, and measures should be done in order to reduce the influence of such power quality pollution. Although the combined distribution to a subsea station will have the lowest power loss, this system will have a large investment cost as the number of subsea components are increased. Advanced analysis will therefor be necessary to evaluate which system topology should be used for powering a DSM power system. However, due to the large power requirement of possible DSM in the Norwegian Sea, a combined distribution to a subsea distribution station will most likely be recommended for DSM operations in the Norwegian Sea.

Denne masteroppgaven har analysert funksjonaliteten til utviklede use cases ved å utføre forsøk i laboratoriet. Use casene har blitt utviklet for en batteripakke plassert i en husstand med PV produksjon tilknyttet strømnettet gjennom en plusskundeavtale. Use casene som har blitt testet og analysert er "Håndtering av plusskundens PV produksjon" og "Regulering av effekt trukket fra nettet". Use case "Håndtering av plusskundens PV produksjon" har som mål å øke eget forbruk av sluttkundens PV produksjon ved bruk av en batteripakke. Use case "Regulering av effekt trukket fra nettet" har som mål å bruke batteripakken til å holde effekten trukket fra nettet under en bestemt effektgrense, slik at lastprofilen til husstanden er mer forutsigbar fra nettets perspektiv, uten de store svingningene. Det har blitt utviklet to modeller i NI LabVIEW, en for hvert use case. Laboratorieoppsettet hvor kodene har blitt testet i, består av et batterisystem levert av Eltek, kalt Smart Storage Enabler, og en PV emulator. PV emulatoren er brukt til å simulere et solcelleanlegg med PV produksjon. Smart Storage Enabler består av tre toveis-invertere, kalt Ipack, en DC-DC inverter som PV emulatoren er koblet til og åtte blybatterier. Hvert batteri har en nominell spenning på 12 V og nominell kapasitet på 190 Ah. Datagrunnlaget som har blitt brukt i forsøkene er timesverdier for last, solinnstråling og temperatur. Ved bruk av innebygde, matematiske modeller har PV emulatoren generert I-V og P-V kurver basert på solinnstråling og temperatur. I et videre studie, er det anbefalt å utføre forsøkene for et tidsintervall med høyere oppløsning, som 15-minuttsverdier eller 1-minuttsverdier, for å avdekke hvordan kodene og batteriet takler større fluktuasjoner i produksjon og forbruk. Denne oppgaven har bidratt til at utstyret i laboratoriet er operativt, at utviklede use cases er implementert i NI LabVIEW til fungerende modeller og at inverterne i systemet kan svare på kommandoer og utføre spesifikke funksjoner fra NI LabVIEW via en CAN protokoll. Resultatene viser potensialet for forbrukerfleksibilitet hos en plusskunde med PV produksjon og energilagring. Gjennom utviklede use cases og forsøk i laboratoriet, har det også blitt demonstrert ulike måter fleksibiliteten til en batteripakke i samsvar med forbruk og produksjon kan utnyttes på. For use case "Håndtering av plusskundens PV produksjon" kom det frem at det økonomiske potensialet basert på egenforbruket for testperioden, april til og med september 2015, er 2205.53 NOK uten bruk av batteri og 2291.97 NOK med bruk av batteri. Dette tilsvarer en økning på 4\%. Med andre ord, er installasjon av en batteripakke i systemet ikke lønnsomt med tanke på investeringskostnadene til batteriet. Imidlertid er det viktig å påpeke at batteriet kan gi andre nytteverdier, som å dekke effekttopper på vinteren og begrense innmatet effekt. Videre, ble det konkludert med at den totale energien matet inn i nettet synker med 33\% med bruk av batteri, fra 597.11 kWh til 397.49 kWh. For use case "Regulering av effekt trukket fra nettet" kom det frem at det er hensiktsmessig å variere effektgrensene avhengig av hvilken årstid det er. Det ble også konkludert med at batteriet ikke klarer å kompensere for de største svingningene over effektgrensen. Ideelt sett skulle batteriet ha kompensert for all svingning over effektgrensen, slik at netto effekt trukket fra nettet ville vært lik effektgrensen og null over grensen. Ved effektgrensen på 1.2 kW, ble det trukket totalt 426.54 kWh fra nettet etter at effekten opp til effektgrensen, PV produksjonen og utlading av batteriet var blitt trukket ifra, mens for effektgrensen på 0.75 kW ble det trukket totalt 537.78 kWh fra nettet. I tillegg til å ha denne rapporten som et resultat, har masteroppgaven resultert i to modeller i NI LabVIEW, med hver sin funksjonalitet som utforsker fleksibiliteten til et PV/batteri-system hos en plusskunde. Modellene gir derfor muligheter for videre forskning på forbrukerfleksibilitet hos sluttkunder. Det er også mange muligheter for videre arbeid med modellene, som å optimalisere kodene, utvikle kodene til å måle batteriets ladetilstand og dermed også bevare batteriets livssyklus. Det er også anbefalt, i et videre arbeid, å kombinere kodene som er utviklet, samt inkludere en parameter som teller måned og har tilgang til historisk forbruk og produksjon, slik at hvilken kode som skal kjøres bestemmes ut ifra dette.

When Europe is transitioning from fossil fuels to renewables, hydropower plants will have an important role in the integration of large quantities of fluctuating sources in the power system, such as wind and solar. With the development of variable speed of hydrogenerators, the objective is to further improve the flexibility and efficiency of the plants. If applying this technology in hydropower plants, Norway can be a green battery in the European power system. This thesis focuses on the development and preparation of a lab set-up with a converter fed synchronous machine, for research on variable speed of hydrogenerators. This includes testing and documentation of the synchronous machine, modelling and simulation of the lab set-up, development and simulations of possible operating scenarios which later can be tested in lab. A large part of the work covers determination of parameter values of the 8kVA synchronous machine. Several tests have been conducted in lab, to determine its most important parameter values, while other parameters were estimated based on typical values. Those were needed for modelling of the machine, and tuning of the control system. A simulation model has been developed to simulate variable speed control of the lab machine, where the machine is fed by a two-level voltage source converter. The objective has been to make the simulation model representative to the lab set-up so that it can be relevant for further lab research. The simulations show that the converter set-up is suitable for adjustable speed control of the synchronous machine. Synthetic inertia contribution by a converter fed synchronous machine has been simulated, by implementing a control loop that deaccelerates the machine in case of a frequency drop in the grid. This was done by letting the frequency signal pass through a derivative controller, which either affects the torque or the speed reference in case of a frequency drop. Highest impact was achieved by letting the derivative controller affect the speed reference, where the power boost lasted 0.7s and peaked at 18\% increase in the presented simulation. A significant impact was also achieved by affecting the torque reference for the inner control loop. However, this impact only lasted 0.1s, since the speed controller quickly compensated for the injection from the derivative controller. For further work, these operating scenarios can be tested in the lab set-up.

Battery energy storage systems (BESS) employing secondary battery technology may offer the greatest potential for large-scale integration of renewable energy sources (RES), and is considered a key element for enabling smart grids for future power systems. The two main components of a BESS are the power conversion system (PCS) and the battery management system (BMS). When selecting a PCS topology, the most important characteristic are the reliability and efficiency of the converter. The converter technologies commonly used for PCS have been the conventional two- and three-level voltage source converters (VSCs), however the PCS topology chosen for the BESS in this thesis is the modular multilevel converter (MMC). The MMC offers several advantages over the conventional VSCs, such as modularity and scalability, low total harmonic distortion (THD), excellent efficiency and reliability. The battery technology chosen for the BESS is based on the lithium-ion battery technology. Lithium-ion offers desirable characteristics such as high efficiency, good response time, reasonably high charge and discharge rates, small dimensions and/or low weight and little maintenance requirements. All of the abovementioned converter and battery characteristics imply that the full MMC-BESS scheme will become highly efficient and reliable with a high round trip efficiency, well-suited for large-scale BESS applications. The system in which the MMC-BESS is investigated, comprises of an offshore wind farm, a two-level VSC-based HVDC transmission system and the AC grid on land. Based on the investigated cases, this thesis can be divided into two main parts: design of the MMC-BESS, and modeling and simulation of the MMC-BESS. The design-part is mainly focusing on a more realistic approach of the MMC-BESS configuration by performing various calculations regarding both battery and converter requirements. The initial MMC-BESS configuration was the conventional, half-bridge MMC with centralized batteries connected to the common DC-link. This approach may offer the simplest solution to implement, however it negatively affects the reliability and efficiency of the full scheme. Furthermore, to mitigate the drawbacks of the centralized battery-approach, an MMCBESS configuration based on distributed batteries was proposed. This approach utilizes the MMC modular structure, as well as improving both the reliability and efficiency. Moreover,by placing bidirectional DC-DC converters, base on the dual active bridge (DAB) topology, in the interface between the converter submodules and the batteries, a galvanic isolation of the batteries could further improve both the safety and reliability of the scheme. Finally, a proposed approach with the main purpose of reducing space, components and cost, was to replace the two-level VSC stations of the HVDC transmission system with the MMC-BESS based on distributed batteries employing DAB converters. This solution offers the possibility to also improve the performance of the HVDC system. However, more effort should be made in verifying these assumptions. The modeling and simulation-part analyzed the initially proposed MMC-BESS configuration in the design-part. The MMC was modeled as a simplified MMC, with an AC-side and DC-side representation and a power balance equation coupling the two sides together. The modeling approach was based on the scientific paper draft presented in [70], as well as the paper presented in [25]. The centralized battery was modeled according to the linear battery model with a small equivalent series resistance (ESR). A state-of-charge (SoC) measurement system was modeled as a part of the BMS, to monitor the charging and discharging processes of the centralized battery. Furthermore, the HVDC transmission system was provided by Prof. Tedeschi and Post.doc. Sanchez Acevedo at the Department of Electric Power Engineering, NTNU. Moreover, the offshore wind farm and the AC grid on land where implemented as ideal AC voltage sources. The simulation verified the MMC s ability to provide a controlled bidirectional power exchange with the centralized battery. The results showed the MMC operating satisfactorily with an active power flow in the range of 1.0 and -1.0 pu, and well-tuned current and energy controllers. The SoC measurements further verified the charging and discharging processes of the battery, where the battery SoC varied between 100% and 73% SoC. The MMC-BESS ability to provide grid support for variable wind power delivered to the AC grid was also investigated. The results showed the MMC-BESS was able to inject active power to the AC grid, as well as absorb active power from the wind farm. The SoC varied between 100% and 85% throughout the simulation. The last case to be investigated, was a comparison between the MMC-BESS presented in this thesis, where the scheme was connected to the AC-grid, and a BESS solution connected to the HVDC-link of the transmission system. The latter solution is based on the master s thesis of Marta N. Hellesnes,in [38]. Both solutions where able to support the AC grid by smoothing the varying wind power from the offshore wind farm.

Low-voltage direct current (LVDC) microgrid has recently been recognized as a feasible solution for next generation electrical power distribution network, due to several benefits that it offers an increasing amount of facilities. However, LVDC microgrids suffers from crucial concerns, which needs to be properly addressed before being conceived as a commercially and economically available solution for future power distribution. LVDC protection technology lacks experience, standards and guidelines, resulting in typical expensive and overdimensioned electrical components comprising LVDC systems. It is necessary to develop an economical DC protection solution, which is capable of conducting with low losses and rapid short-circuit currents interruption, without being destroyed in the process. A promising circuit breaker (CB) based purely on power electronics, comprising 2 antiparallel connected commercially unavailable reverse blocking integrated gate commutated thyristors (RB-IGCTs) has been developed by ABB. Replacing the unavailable RB-IGCT devices in the original CB design with insulated gate bipolar transistors (IGBTs) or asymmetric integrated gate commutated thyristors (A-IGCTs), is evaluated. Based on proper parameterization, two CB configurations are presented, comprising either 12 parallel connected 3.3 kV rated IGBTs from Infineon, or 8 parallel connected 4.5 kV rated A-IGCTs from ABB. These configurations are compared against each other, using simulation models assembled in Matlab® Simulink. The higher voltage ratings of the A-IGCT allowed the CB to utilize larger metal oxide varistors (MOVs), which reduces the energy dissipation period with approximately 1 ms. The A-IGCT based CB is indicated to be more expensive, due the CB configuration being 60 % heavier than that of the IGBT based CB. On the other hand, the A-IGCT based CB is able of conduct nominal current with 23 % of the losses to that of the IGBT based CB. Since excessive conduction losses are one of the major disadvantages of solid-state CB, the 4.5 kV A-IGCT is concluded to be the most promising semiconductor device for the CB design. The preferred CB is later utilized to protect three simplified isolated LVDC systems modelled in Simulink against short-circuit faults. Its breaking performance indicated that the A-IGCT based CB might be overdimensioned for the considered LVDC system, but is however conceived as a feasible solution. The promising CB configuration comprising 4.5 kV rated A-IGCTs was able of interrupting faults of 17.5 kA at 3000 V within approximately 2 ms of fault injection.

Revolve NTNU har de siste årene bygget en bil som både går fortere og er lettere. For at dette skal skje blir alle komponenter i racerbilen presset til sitt ytterste. Dette har medført at batteripakken, som er den tyngste og en av de mest sentrale delene i bilen, utvikler mye varme på grunn av mengden battericeller men og strømstyrke. Det er denne utfordringen som har vært inspirasjon til masteroppgaven. Den omhandler testing av battericeller og moduler med fokus på varmeutvikling. Oppgaven er skrevet i samarbeid med NTNU Revolve. Det er batterier som er brukt i racerbilen Revolve- team 2017 som er benyttet i undersøkelsene. Enkeltceller, celler i parallell og en hel modul har blitt testet. I masteroppgaven er det blitt sett på nye type studier av battericeller ved bruk av elektrokjemisk impedans spektroskopi for å kunne få en enda mer detaljert forståelse av hvordan battericellene oppfører seg ved forskjellige temperaturer. En har også sett på nye metoder for kjøling av battericeller ved bruk av Phase-change material som er ikke-elektrisk ledene. Dette gjør at du kan ha de i direkte kontakt med battericellene. Eskalering av testingen for å se på en hel modul har og blitt gjort, for å få et mer realistisk blide av hvordan forholdene er i batteripakken i en racerbil og for å få en bedre forståelse av hvordan cellen blir påvirket av hverandre ved stor påkjenning av strøm og høye omgivelsestemperaturer. Det er og blitt sett på forskjellige kjølemetoder, i situasjoner som simulerer forholdene i racerbilen. Resultatene av modultestingen har vist at differansen på temperaturen på utsiden av modulen sammenlignet med innsiden er på hele 13.7C. Resultatene har og vist at det er rundt senter cellene er varmest mens de blir kaldere jo nærmer ytterkant av modulen de kommer. En ser og at temperaturen er høyere i midten av cellene sammenlignet med lengre nede når strøm er påsatt. Når modultestene har kjørt rene opp og utladningsstrømmer har det vist seg at strømretningen påvirker og varmeutviklingen i modulen. Ved kjøling ved hjelp av vind har det vist seg at parallell kjøling er det som gir best resultater for varmeutvikling. Det har og blitt sett på entropieffekten i en modul. Det har vist seg at ved 2C ladningmed 36C omgivelsestemperatur har den kaldeste temperaturen i modulen vært på 34C. Testing av PCM har vist at nok volum av denne væsken vil gi en bra kjøleeffekt. Ved elektromagnetisk

Due to an increased focus on the development of high-power marine microgrids, new vessels are being installed with large power systems including highly fluctuating loads operating at high efficiency and low emissions. The introduction of a battery energy storage system in a shipboard DC grid allows for operating the diesel gensets at variable speeds, not being locked to a fixed frequency or speed. The transient performance of the system can also be improved and an optimal load leveling of the different prime movers can be achieved by the proper implementation of a battery. Thus, the system can operate efficiently with highly fluctuating loads while keeping the fuel consumption low. In order to integrate the batteries into the shipboard DC grid, high-power bidirectional DC/DC converters are needed. Many converter configurations are available, yielding different advantages and disadvantages. Thus, each different system should be evaluated in order to see which converter topology is best suited for that specific system. The stress levels on the switching devices are also higher than before, which can be seen as a result of the utilization of higher voltages and currents in the shipboard grids. Isolated bidirectional DC/DC converters for high power applications can provide the proper connection, but along with the increased loads comes higher requirements to component ratings as well as a more complex control system. A high-power medium-voltage DC/DC converter suitable for implementing such a battery is analyzed in this thesis after explaining the shipboard electrical system. The converter has a rated power of 4MW with an equal input and output voltage of 1100V and with a high-frequency transformer providing galvanic isolation between the shipboard power grid and the battery. Its performance is evaluated by means of modeling it in Matlab/Simulink where the focus has been on the switch resonance components in order to achieve a high-performance, soft-switched converter. A PI controller is also implemented in order regulate the output voltage of the converter for different load levels. As some of the converter requirements are a robust design, high operating efficiency and a fast response, the PI controller is tuned in order achieve this and it is seen that the converter responds fast with small deviations in parameter values for load variations of several megawatts. A gate signal generator is also made, where the output of the PI controller is used as input, delaying the gate signals to switches in either the primary or secondary side, depending on the power flow direction. The model is used to analyze the current waveforms through the switches and the transformer, and by doing measurements and calculations based on the simulation results, the efficiency of the converter is calculated at different load steps ranging from 0 to 4MW. It is seen that the converter operates at an efficiency of 97.2% at an output power of 4MW with increasing efficiency for lower load levels due to the lower conduction losses. However, the high-frequency transformer is assumed ideal in this thesis and thus the efficiency will be somewhat lower with this included.

This thesis aims to address the challenge of suppressing the circulating current. Principle of operation and mathematical derivations, common control methods for inner dynamics and software/hardware algorithm implementation is presented, with focus on the control algorithm repetitive control for suppression of the circulating current. This algorithm utilizes the internal model principle for suppression of the periodic circulating current. Algorithm derivation and stability is analyzed in the discrete time domain and performance confirmed with MATLAB Simulink®. Algorithms are investigated and discussed both for sequential software implementation on embedded processor and as hardware logic implementation on FPGA. A model based design approach is used in MATLAB Simulink® to define system requirements, create the analyzed algorithms, simulate and verify the design, and automatically generate code for the system. MATLAB Embedded Coder® is used to generate C/C++ code for embedded processor implementation, while MATLAB HDL Coder® is used to generate synthesizable HDL code for IP core generation. The third party synthesis tool Xilinx Vivado Design Suite is used to interface the designed IP core into a complete processing system and implement the IP core onto a FPGA. Processor in the Loop (PIL) simulations is used to look at algorithm execution time in processor, while MATLAB and Vavado generate reports on hardware logic resource utilization and critical path estimations.

The growing need and importance of Battery Energy Storage Systems (BESS) in grid support applications as well as in the transportation sector, has brought attention to the adaption of the Modular Multilevel Converter (MMC) to be used in BESS. The MMC is already known for its high efficiency, high voltage capability and good harmonic performance, but the modular arrangement of the converter also allows the battery to be implemented in a distributed manner. Known as a split battery topology or split storage. Connecting individual battery units directly into the submodules of the MMC. Opening the possibility of high AC-side voltage without a transformer even with low battery voltage. However, the current flowing through the submodules inherently consists of oscillating components on top of the DC-component. This thesis show that these are at fundamental and 2nd harmonic frequencies with amplitude of two and one times the DC-component respectively. It is assumed that this is unacceptable due its negative impact on battery performance and lifetime. In most literature this is solved by interfacing the batteries with the submodules using a DC-DC converter. But this thesis explores the possibility of mitigating these components using a control technique and an interface consisting of a filter based on passive components. Therefore, the proposed technique is two-legged. 1. injecting a circulating current that cancels the 2nd harmonic power fluctuation in the arms and hence submodules. Removing the need for 2nd harmonic filtration. 2. Employing a resonant filter to mitigate fundamental component together with a lowpass filter. The performance and feasibility of this technique is analyzed theoretically and through simulations. Resonant branch of the filter is analyzed in lab experiment. And results indicate that this could be an attractive solution, especially in applications where high reliability is of concern.

This thesis investigates the use of low frequency alternating current (LFAC) for the longdistance transmission of the power extracted from offshore wind farms. It is based on the measurement of power flow in different nodes in the system simulating an offshore wind farm being built outside the coast of Denmark, Horns Rev 3 (HR3). The results for scenarios simulating different lengths of the offshore cable and the operating frequency of the system are compared. Reactive power compensation is used to improve the power flow of the system and to counteract the reactive current from the capacitance in the long offshore cable. Power flow, voltage- and current limitations of the offshore cable, and varying the reactive power compensation in the different nodes are used to assess if the different scenarios described are feasible or not. Considerations related to size and operation at reduced frequency on the different power system components are mentioned and discussed briefly. The thesis does not include any detailed analysis with regards to a possible increase in cost for the different components operating at reduced frequency. The thesis concludes that the use of LFAC gives reductions regarding power losses, improving the power system efficiency. For the real case of HR3, using LFAC excludes the need for reactive power compensation. The thesis also proves that it is possible to transmit substantial amounts of power over longer distances than any other AC-cable in operation today. It is proved that using LFAC and reactive power compensation from shore, it is possible to transmit over 370 MW to the grid with an overall efficiency of more than 93 %, at a distance of 200 km using a 220 kV offshore cable. Including reactive power compensation from the offshore end of the cable, the distance can be increased to 300 km, delivering over 370 MW of power to the grid with an overall efficiency of 91 %. However, the weight and size of LFAC transformers are considerably higher than the equivalent 50-Hz components, with up to 2.7 times the weight, this would be a challenge for the construction of the offshore substation transformers. Beyond the scope of this thesis further investigations are needed regarding costs for the different components of the system, as this is not included.

The idea to couple a gear magnetically has been around for some time, and has gained increased interest in recent years. By manipulating air gap field harmonics, magnetic gearing has been shown to have comparative, or even better performance than its mechanical counterpart, while circumventing typical mechanical disadvantages. Recently, magnetic gearing has been integrated into Permanent Magnet motors to obtain "pseudo" direct-drives that result in efficient high-torque density at low speed electrical machines. This thesis investigates such a promising machine topology, using the Finite Element Analysis tool COMSOL Multiphysics. The general magnetically geared Partitioned Stator Switched Flux Permanent Magnet machine was found to be able to meet the requested specifications of 200kNm and 4.2MW, with challenges being low power factor and high cogging torque. This machine is a very interesting concept that may be used for low-speed high-torque application when lower power factor can be managed(i.e. power electronics).

As the performance of silicon power semiconductors is close to the theoretical limit, other semiconductor materials are sought to improve power electronics system efficiency. Devices made with the wide bandgap material silicon carbide (SiC) are promising due to the possibility of significantly improved system efficiency and reduced system volume and weight. However, reliability is still a key issue to be dealt with before wide-spread use of the devices may take place. In this thesis two of the largest reliability concerns of the SiC MOSFET are evaluated by investigating discrete devices. The issues are the threshold voltage instability and the reliability of packaging of the devices. Dedicated test benches were established to study the issues. The tested devices indicate that the drift of the threshold voltage is still an issue for SiC MOSFETs. The methodology of performing power cycling tests was investigated. This accelerated life time stress test can be utilized to evaluate packaging reliability. The tests were performed by using the Vsd(T)-method to evaluate the junction temperature. Several devices were tested, all of which indicated failure by bond-wire lift off. The results furthermore indicate that power cycling of SiC MOSFETs is affected by the threshold voltage instability. In particular, a reduction of the on-state voltage was observed at the beginning of almost all the tests, which was attributed to the instability of the threshold voltage. The results call for a discussion as to how power cycling of SiC MOSFETs ought to be performed, in order to reduce the influence of the threshold voltage drift. A suggestion for reducing the influence of the latter is also given.

Distribusjonsnettet i Norge er delt inn i 139 konsesjonærområder. Hver av disse er pålagt av NVE å rapportere data om feil og avbrudd. På grunn av rivende utvikling innenfor datateknologi knyttet til "det digitale skiftet" åpner det seg nye muligheter for anvendelse av kartfestede data. Formålet med denne oppgaven har vært å undersøke mulighetene til å benytte geografiske informasjonssystemer (GIS) for å presentere feilstatistikk på nye måter. Måten distribusjonsnettet er strukturert på gjør det mulig å sammenligne feilstatistikk mellom nettselskap. I denne oppgaven er det undersøkt om geografiske data kan brukes som basis for en slik sammenligning. Det ble derfor utviklet et dashbord i ArcGIS. En bakgrunnsstudie av feilanalyse og en praktisk utforming av verktøyet "dashbord" med påfølgende vurdering utgjør kjernen i oppgaven. Bakgrunnsstudiet gir en oversikt over dagens rapporteringssystem for feil og avbrudd i kraftsystemet (FASIT), som lagrer store mengder nyttig informasjon om nettselskapene. TrønderEnergi ble kontaktet og brukt som modell i arbeidet. Oppgaven fokuserer på kart som medium og bruk av geografiske data NVE har om nettselskapene i dag. Den praktiske utformingen bygger på erfaringer fra bakgrunnsstudiet til å utvikle en dashbordvisning for sammenligning av feilstatistikk mellom selskaper. I motsetning til tradisjonell feilstatistikk som tar for seg alle nettselskaper samlet sett, vil et dashbord kunne gi rom for å analysere individuelle forskjeller mellom nettselskap. Den praktiske utformingen av dashbordet ble begrenset av mangelfull programmeringserfaring med GIS som er tidkrevende og komplisert, samt mangel på kartfestede data. Erfaringen med dashbordet viser at man kan sammenligne feilstatistikk fra ulike nettselskap. Løsningen ble presentert for TrønderEnergi som var positive til sammenligning ved bruk av geografiske rammevilkår. For feilstatistikkens del ble det slått fast at koordinatfesting av feil er en forutsetning for å kunne gjøre inngående sammenligninger. Vurderingen avdekker også at flere forhold i feilrapporteringen og kilden til anleggsdata (eRapp) bør forbedres for å effektivt kunne produsere god feilstatistikk for nettselskapene. Oppgaven belyser også muligheter for videreutvikling og gir eksempler på andre kartkilder knyttet til geografiske, topografiske eller klimatiske faktorer som kan brukes i feilanalyse. Hovedkonklusjonen for oppgaven er at det er et behov i nettselskapene til å kunne nyttiggjøre seg av egen feilstatistikk og sammenligne seg med andre selskap. Økt bruk av GIS innenfor kraftbransjen vil på sikt kunne bidra til å styrke arbeidet i mange av de utfordringene bransjen står ovenfor.

The purpose of this thesis is to study the power quality in a stand-alone wind- powered water injection system for enhanced oil recovery. Water injection is crucial for boosting the recovery rate in mature oil reservoirs. The concept of a stand-alone wind-powered water injection system is to utilise a single wind turbine to solely power the water injection pump in an installation which is physically disconnected from the mother platform. The concept is possible due to recent technological developments within offshore floating wind power and advances within power electronic converters. The main objectives have been to develop a model in MATLAB/Simulink, implement coordinated control systems and study the power quality during common, and important scenarios. The system is powered by a 6 MW permanent magnet synchronous generator which is directly driven by the wind turbine. The generator is interfaced with the microgrid main bus through a 2-level back-to-back voltage source converter, in order to obtain variable speed operation of the turbine. The main load is the water injection system which is driven by a 3 MW induction machine with pump loading. The induction machine is also interfaced with the microgrid through a 2-level back-to-back voltage source converter for speed control. The battery storage is responsible for supplying the fixed critical loads during low wind conditions, and to control the main bus voltage magnitude and frequency. The wind turbine is tested through several common wind scenarios. The performance is shown to be a trade-off between the ability to follow the optimal speed and fluctuations in produced power. The turbine is further tested for fault ride-through capability and harmonic distortion. Due to the slow pressurisation in hydrocarbon reservoirs, water is not required to be injected with a fixed rate. Thus, the water injection is performed only when there is wind power available. However, the wind is volatile, which makes it critical for the induction machine to be able to follow a rapidly fluctuating speed reference. This has been analysed in-depth. Another critical event of a microgrid is the capability of performing a black start. The battery has been tested and shown capable of establishing and maintaining the rated voltage and frequency of the microgrid. The results obtained are thoroughly discussed throughout the thesis, simplifications and problems encountered have been highlighted.

This paper will compare the surface permanent magnet (SPM) machine to the interior permanent magnet (IPM) machine with fractional slot concentrated winding (FSCW). Key features is to achieve high field weakening with low losses, while keeping a high torque at rated speed. A walk through of essential theory concerning field weakening will be done. The application builder in COMSOL 5.2 was used to quickly do changes in the model. The Livelink Matlab extension was also used for post-processing the results and estimating the core losses.The results show that the I-shaped IPM machine has the lowest losses and both double layer and single layer winding can achieve high field weakening. The I-shaped rotor also has low total harmonic distortion and torque ripple. The V-shaped IPM rotor shows it difficult to optimize because of its many geometric variations and non linear behavior. The SPM machine has a higher power factor then the IPMs, due to the higher inductance in the IPM machines.

The Large Hadron Collider (LHC) utilizes superconducting Main Dipole Magnets that bend the trajectory of the beam. The LHC ring is electrically divided into eight octants, each allocating a 7 km chain of 154 Main Dipole Magnets. Dedicated detection and protection systems prevent irreversible magnet damage caused by quenches. Quench is a local transition from the superconducting to the normal conducting state. Triggering of such systems, along with other failure scenarios, result in fast transient phenomena. In order to analyze the consequence of such electrical transients and failures in the dipole chain, one needs a circuit model that is validated against measurements. Currently, there exists an equivalent circuit of the Main Dipole Magnet resolved at an aperture level. Each aperture model takes into account the dynamic effects occurring in the magnets, trough a lossy-inductance model and parasitic capacitances to ground. At low frequencies the Main Dipole Magnet behaves as a linear inductor. Cable eddy current losses are demonstrated by a flattening of the transfer function impedance in the 30 − 50 Hz range. The time constant of such losses is dictated by the parallel resistance, and the relative size of the loss is given by a scaling parameter. Capacitive effects become dominant around 10 kHz. Across the dipole magnet there is a resistor connected in parallel to dampen voltage waves. Simulations of an Main Dipole Magnet in OrCAD Cadence PSpice, using the present parameters, and measurements from the LHC give a clear discrepancy. This necessitated an updated fit and three methods tailored to obtain each parameter were developed. Firstly, the inductance value was obtained estimated from the initial slope of the impedance plot. Secondly, the numerical method chosen for the parameter fit is Particle Swarm Optimization. The algorithm iteratively minimizes the error between measurements and the analytical impedance transfer function, making it possible to estimate the value of the parallel resistor and the scaling factor. Finally, parasitic capacitance to ground was determined with Finite Element Method in COMSOL, as it is challenging to extract parameters from high frequency measurements. Values from measurements verify this method of estimating capacitance. The measurements of the Main Dipole Magnets were performed while connected to the rest of the dipole magnet chain, which influenced frequency response measurements. Hence a proposal on how to reduce the sensitivity to this influence is outlined. Moreover, the method of fitting was found to be modular, meaning each Main Dipole Magnet can be fitted individually in the dipole magnet chain. This is significant as it is not necessarily possible to perform measurements on a stand-alone magnet. To crosscheck the validity of the method, Particle Swarm Optimization fits from stand-alone measurements and measurements from the dipole magnet chain were compared. Both values of the parallel aperture resistance and scaling factor were different for the two cases. Compared to the operating point of the Main Dipole Magnet, measurements were performed at 1 A, resulting in 20 % lower inductance than nominal value. Through COMSOL simulations persistent magnetization was found to be the dominating cause. Furthermore, at 1 A the magnet is in the so-called Meissner phase, creating nonlinearites in the persistent magnetization. Simulations indicate that this distorts the mid-range frequency AC characteristics represented by the parallel aperture resistance and scaling factor. However, measurements outside the Meissner phase are expected to provide similar parameters values to that of the working point of the LHC. Overall, the approach presented has shown promising results and can be translated to a general Section method for fitting electrical parameters for accelerator magnets.

In hydro power applications, the synchronous generator is widely used. As the speed of the synchronous generator is fixed to the frequency of the grid, the hydro turbine must operate at fixed speed independent of load and water flow. Earlier studies of the topic have concluded that the efficiency of the turbine can be increased by introducing a variable speed generator. The doubly-fed induction generator (DFIG) is widely used in wind power applications and is capable of operating at speeds +/-30% around synchronous speed. However, this introduces a power electronic converter for controlling the rotor currents and slip rings are used for transferring the power into the rotor circuit. This adds complexity to the system which may cause needs for more maintenance and more expensive equipment. This work continues a work started in a Master thesis in 2015, where the control strategies for a DFIG used in hydro power applications were developed. Simulations were used to study the behavior of the configuration, and a lab setup where a asynchronous motor was used as turbine was made. The lab setup is improved and the complete control system is rebuilt. A Speedgoat real-time controller is chosen as the controller for the lab setup. The programming of the controller is done by the use of Simulink, which is a well known tool for simulations. The signal conditioning system for speed, voltage and current measurements and solutions for interfacing the converters are made from scratch. The motor drive controlling the asynchronous motor which is used for simulating the turbine, is controlled by the Speedgoat controller. Through lab tests, the functionality of the lab setup is verified by changing speed and torque. The result of the work, is a lab setup which can be used for testing the DFIG in a hydro power application. Since the hardware is made and found to be working well, it is possible to develop the setup further in the future mainly by doing improvements in software.

Compact high-efficiency power converters is a hot topic in electrical power engineering. So is also the case for electric and plug-in hybrid electric vehicle battery chargers. Increasing the switching frequency in the power converters will reduce the size of passive components, but this also results in lower efficiency. The progress being made in wide bandgap semiconductors introduce possibilities for lower loss, and thus improved efficiency at higher switching frequencies. Silicon carbide is such a semiconductor material, whose electrical and thermal capabilities trumps those of a classical silicon semiconductor. To investigate whether utilizing a silicon carbide semiconductor in an on-board charger for electric and plug-in hybrid electric vehicles can provide lower switching loss, a silicon carbide MOSFET is tested against a silicon MOSFET. The focus is on the hard switched application of the power supply s power factor corrector. A continuous-conduction mode DCDC boost converter is designed and used to perform the test. The converter is designed for an input voltage of 230V, output voltage of 350V and rated for 3500W. Emphasis has been made on parasitic elements and their effect on switching behavior. Three different circuits was developed through the course of the thesis, with improvements in layout implemented for each step. Significant improvements was seen in gate drive circuitry and the overall converter from the first design to the second, and improvements in the gate circuitry was seen from the second design to the third. From this, it is concluded that for best performance, the gate driver circuit should be as compact as possible, with short conduction paths and close to the MOSFET gate pin. The return form the MOSFET source to the gate driver ground should be as large as possible, and directly underneath the gate signal path. SMD components with low parasitic inductance and capacitance should be used in the gate driver circuit. The Si and SiC MOSFET show different switching behavior. The main difference is at turn-off, regarding delay and voltage rise waveform. At the same gate resistance and gate voltage, the turn-off delay time for the Si MOSFET is almost three times longer than the SiC MOSFET delay time. While the SiC MOSFET turn-off voltage showed a linear increase the entire rising period, the Si MOSFET voltage rise was more exponential. Accurately determining losses in a MOSFET, and how well it is capable of operating, is a challenging exercise. Distortion caused by the probes while measuring, bandwidth limitations in the probes and oscilloscope, time delays in the probes and improper components and design can all contribute to erroneous results. Nevertheless, it was concluded that in this circuit, with these MOSFETs, the SiC MOSFET had a switching loss between 106 µJ and 138 µJ, which was between 0,7 and 0,75 times lower than the Si MOSFET switching loss. With the conduction loss for the MOSFETs taken into account, which was found to be 2,2 W for the SiC MOSFET and 7,5 W for the Si MOSFET, at nominal operation and a selected operating temperature of 125 ºC, the optimal switching frequency was found to be between 165 kHz and 215 kHz. This is based on a power loss cap of 25 W.

Denne besvarelsen har kommet med forslag til metoder for å optimalisere driften av distribusjonstransformatorer ved hjelp av trådløse målere og kontinuerlig overvåkning. Forarbeidet i prosjektoppgaven undersøkte nytten av temperaturmåling og levetidsberegning på papirisolasjon. Det viste seg å gi urealistiske og unøyaktige resultater, noe denne hovedoppgaven forsøkte å rette opp i. Det ble undersøkt om man kunne gjøre mer realistiske levetidsberegninger fra tilgjengelige måleverdier av temperatur og belastning på transformatoren. Med utgangspunkt i dette ble det laget en helseindeks for å representere hvordan tilstanden til transformatoren var. Den rangerer hver parameter som måles og samler det i en sluttvurdering. Det er ikke så mye data som blir målt på dette tidspunktet, så det ble lagt til muligheter som spenningsmåling, befaringer, spenning og fuktighet for fremtidige målinger. Ved å overvåke helseindeksen og indikatorer som maks og gjennomsnittlig last/temperatur kan det tas bedre avgjørelser om transformatoren trygt kan stå flere år uten å byttes ut. Siden en del transformatorer byttes ut før teknisk endt levetid, enten på grunn av havari som lyn eller kun på grunn av høy alder, ble det utviklet en kalkulator for å blant annet finne optimal merkeytelse på en transformator. Antagelsen var at det kunne være mer lønnsomt å kjøre en billig transformator hardere enn å ha en dyr, lavt lastet, transformator havarere med mye levetid igjen. Ved å redusere merkeytelsen på en transformator med så lite at man antar endringen i svikt var tilnærmet lik null ble det demonstrert at man kunne redusere kostnader og samtidig ha liten innvirkning på levetiden. Helseindeksen ble inkorporert i kalkulatoren som sammen med en pålitelig lastprofil kan gi en løpende tilstandsvurdering på kalkulatoren man regner på. På den måten kan man stille helsen opp mot den økonomiske analysen. Kalkulatoren åpner også for å eksperimentere med forskjellige scenarioer som uforventet lastøkning, toppeffekts-situasjoner, økt aldringsrate på grunn av fuktighet og lignende for å se hvordan det forandrer levetiden og indikatorer som maks last og temperatur. Oppgaven har prøvd å finne en pålitelig modell for transformator-temperatur basert på utetemperatur og last for å kunne brukes i kalkulatoren og beregne levetiden til papirisolasjonen selv uten direkte målinger. På den måten kan kalkulatoren, med gode antakelser, brukes på en større del av transformatorparken som ikke har målere installert. Det viser seg å være vanskelig å få til en objektiv og generell helseindeks som gjelder for alle typer transformatorer på grunn av alle faktorene som spiller inn på levetiden. Det blir derfor presentert alternative målere som kan gjøre tilstandsvurderingen mer nøyaktig. Det vanskelige spørsmålet er om investeringen i målere gjør opp for de reduserte kostnadene i å la transformatorene stå lengre ute eller bruke optimaliserte merkeytelser.

A parametrized 2D model of a permanent magnet synchronous machine is developed in the finite elements method software COMSOL Multiphysics, along with a graphical user interface to control it. The GUI provides automatic construction of geometry based on specified input parameters, application of boundary conditions and equations, and post-processing of results. Algorithms are implemented to apply automatic periodicity settings and winding layouts for single layer and double layer integer and fractional slot windings. The output of the model is validated by comparing to analytical formulas and to other studies with known input parameters and results. The purpose of developing the the application is educational, allowing simulations to be performed without requiring in-depth knowledge of FEM. A range of sample studies are presented in this respect, to indicate possible areas of application within the educational setting. It is found that simulations can be performed very efficiently, indicating that the application could provide a valuable tool for studying electrical machines. Depending on further development and implementation of additional functionality, the application could probably also be used in the process of designing machines.

Støtte fra Enova til utbygging av private solcelleanlegg har stimulert til en økning i antall solcelleanlegg i Norge [19]. Med en økning av distribuert generering oppstår det utfordringer forbundet med spenningskvalitet og sikkerhet hos nettselskapene, og det er det som diskuteres i denne oppgaven. Sikkerheten til instanser som jobber på linjer med høy og lav spenning er viktig. Det er derfor med et kapittel som omhandler personsikkerhet i forbindelse med solcelleanlegg. Det eksisterer løsninger for å sikre avstenging av solceller ved brann og øydrift. Dette er med på å ivareta sikkerheten til brannmenn og energimontører. Det er likevel ikke påbudt med slike løsninger på alle anlegg, og derfor må brannmannskap behandle branner på solcelleanlegg som branner i elektriske anlegg for øvrig. På samme vis har energimontører sine rutiner for å hindre elektriske støt. Denne rapporten tar for seg to anlegg. Det første anlegget er lokalisert på Myrholt, Steinkjer, og er et anlegg som er typisk for privatboliger (3-5 kWp). Det andre anlegget er bygget på Overhalla Barne- og Ungdomsskole og har en effekt på over 45 kWp. Det som skiller disse anleggene, foruten om størrelsen, er styrken på nettet de er tilknyttet. Begge anleggene leverer til nettet når produksjonen er større enn forbruket, og begge er derfor definert som plusskunder [19]. Nettstyrken på Myrholt er såpass svak at det oppstår problemer i flere av parameterne i FoL. Hovedsakelig oppstår de fleste bruddene på grunn av laster som slås ut og inn, og ikke på grunn av produksjon i anlegget. Likevel har det oppstått et resonansfenomen som medvirker til å bryte kravet til flimmer. Dette resonansfenomenet er et resultat av svakt nett og inverterens kraftelektronikk. Et lignende fenomen har ikke blitt observert i nettet på Overhalla Barne- og Ungdomsskole. På Overhalla Barne- og Ungdomsskole er det ikke blitt observert noen spesiell påvirkning i spenningskvalitet på grunn av solcelleanlegget. Variasjoner i spenningskvalitet ser ut til å være et resultat av lastendringer i nettet for øvrig. Raske simuleringer i NETBAS har vist at anlegget må mangedobles i størrelse for at stasjonære spenningsnivå skal bli et problem. Dersom solcelleanlegg settes opp i husstander med omkringliggende sterke nett, vil ikke solceller være betydelig merkbare, så lenge anleggene forholder seg i størrelser som passer til forbruket, typisk 3-5 kWp for en vanlig bolig [19]. Konklusjonen blir derfor at solcelleanlegg ikke byr på større problemer enn en tilsvarende stor last vil gjøre. Før analysen av anleggene kommer et kapittel som omhandler simulerte situasjoner hvor nettet er svakt, og flere i et nabolag ønsker å bygge ut solcellesystemer. Her oppstår det høye spenninger i lavlastperioden, og noen av nodene i simuleringen bryter grenser satt av FoL. Simuleringen viser at for å unngå for høye stasjonære spenningsnivå når det produseres solkraft fra mange plusskunder samtidig, kan det bli nødvendig med struping av produksjon. Annet mulig tiltak er automatisk trinning av transformator i tilknytning til det utsatte nettet. Simuleringen viser også at nodenes spenning ligger på nedre spenningsgrense i tunglast. Dersom en hadde hatt måledata fra flere plusskunder med solcellesystem i samme nabolag, hadde det vært mulig å si noe mer konkret om effektene som oppstår i nettet på grunn av produksjonstopp fra flere anlegg samtidig.

In this thesis, the performance of the full Silicon Carbide (SiC) half-bridge power module BSM120D12P2C005 from Rohm Semiconductor is investigated. A laboratory circuit enabling double-pulse tests of the half-bridge module is designed and built. Building such a laboratory circuit demands careful analysis of the challenges and dangerous aspects that might arise. Consequently, the first part of the thesis discusses relevant background theory on SiC material properties, SiC MOSFETs and SiC SBDs. This is followed by general theory on different power converters, as well as analysis of switching transients and power losses in MOSFETs and SBDs. A theoretical efficiency comparison of three-phase inverters consisting of eight different state-of-the-art SiC half-bridge modules and one state-of-the-art Si IGBT half-bridge module is provided. It is shown in this comparison that all the SiC modules achieve an efficiency of 98 %, while the Si IGBT module gives an efficiency of approximately 93 % at 50 kHz. Subsequently, general theory on important considerations when designing a converter circuit for hard-switching SiC modules is presented. This is followed by a detailed description of the laboratory setup, the measuring instruments and other important considerations in this thesis. The next part presents an analysis of the SiC module performance through simulations in LTspice IV. The simulation circuit design and decisions are justified. The simulations are used as basis to investigate the switching characteristics of the SiC module, in addition to testing the impact of changes in the laboratory circuit. It is found through simulations that the combination of a DC snubber and a turn-off snubber could help improve the switching characteristics. Finally, results from the laboratory experiments are presented. Firstly, it is shown that the bandwidth and stray inductance of the measuring instruments influence the switching characteristics. This is followed by a discussion on the selection of gate resistance and its influence on the switching speed. A short-circuit protection (SCP) is added to the gate driver of the SiC MOSFET, and its practical operation is proven successful. The switching characteristics and switching losses are obtained through double-pulse tests of the SiC MOSFET, and are presented for different drain-to-source voltages and drain currents. The switching characteristics show high switching stresses on the SiC module. Thus, snubber circuits are added to the laboratory circuit. The addition of a DC snubber and a turn-off snubber results in 40 % reduction in voltage overshoot and 85 % reduction in ringing duration. This improvement is achieved with an increase in total switching losses of 24 %. It is found that an Si IGBT switching at 600 V 120 A and a switching frequency of 50 kHz dissipates four times more energy than a SiC MOSFET including snubbers at equal conditions.

This thesis concerns the development of a single-phase synchronous machine model in time-domain, with instantaneous value output. The model is meant to represent the generator part of the rotary converter used in the Norwegian traction power system (supplying railway). Its function is to convert the 50 Hz frequency of the main grid into 16 2/3 Hz which is the standard frequency of the traction grid in Norway. The chosen method of modelling is inspired by previous work on this matter. The solution is to model the single-phase machine by using a standard three-phase machine model, with unsymmetrical loading on two of its phases. The third phase is left idle. The equations of the standard synchronous machine model are used. An adjustment of the parameters is questioned, followed by a suggested set of parameter values on the basis of analysis and mathematical deduction. A comparative approach is used in the validation of the model. Jernbaneverket has done some measurements of the actual rotary converter, which have been compared to simulations of the model. According to these comparisons, the model has proven to reflect the general behavior of the rotary converter. High correlation can be achieved by adjusting the parameters. Nevertheless, no conclusion has been made upon the adjustment of the parameters.

Accurate estimation of losses is one of the major challenges in electromagnetic machinery modelling. Where permanent magnet synchronous machines are attractive for high performance applications, making them relevant for investigation with regard to optimal design. This thesis concerns the investigation of loss calculation methods for permanent magnet synchronous machines. As well as implementation of loss functionality in the further development of a permanent magnet synchronous machine, modelled in the finite element method program COMSOL Multiphysics. The model is developed in a previous master thesis, were losses were neglected for simplicity. A process of evaluation and testing of the developed model will also naturally take place. The method chosen to estimate the core losses separates the loss into hysteresis and eddy-current components. This is a well established model and has been shown to give satisfactory results, coupled with finite element method modelling, in previous studies. The method was deemed feasible to implement and should in theory give an adequate evaluation of the core losses. However mismatch between the time and frequency domain solutions, as well as an imbalance in the flux distribution, due to conflicting interactions within the model, puts the validity of the implemented functionality into question. Losses in the windings and the permanent magnets are accounted for by modelling the resistive losses for their corresponding domains, as the magnets can be assumed resistance limited. The eddy-current losses in the magnets are affected by their size and structure, where sectioning could be implemented to decrease losses. In the case of the windings problematic integration of their properties in the model results in unrealistic loss values.

Superconducting rotor windings are considered for large offshore wind turbines to reduce weight, size and cost. The European project INNWIND.EU utilize various national projects in different European countries to accelerate the development of innovations that can help realize larger wind turbines. SINTEF Energy Research is one of the organizations that are involved in this project by designing, building and testing a superconducting rotor field winding. This could potentially reduce weight, volume and cost of future wind turbines. This master thesis describes the method and steps that were done to complete a superconducting field winding and how it was tested. Ten MgB2 double pancake coils were glued together with a thermally conductive epoxy on a thermal interface. The coils were joined electrically by a validated procedure that makes low resistance joints by using a custom made soldering tool. BSCCO high temperature superconductors were soldered at each end of the field winding to lead the current from the high temperature stage of the cryostat to the field winding. A mechanical support placed around the field winding was designed to withstand the magnetic forces acting on the field winding during testing. The field winding was placed in a cryostat and cooled down to 17.7K for testing. Seven out of ten coils were proved to be superconductive. Coil 1, 6, 10 and the BSCCO HTS were not superconductive as a voltage was measured when a current was flowing through the field winding. The quality of the joints was validated by measuring the voltage over each joint and calculating the resistance. The average resistance of the joints was 40nOhms where the highest resistance was 110nOhms. 15mW of heat would be dissipated in the joints when testing with a current of 200A. The inductance of the field winding was found to be 4.6H.

Offshore wind energy has become a major part of the European energy mix, and new and improved technology enables the construction of wind farms with larger installed capacity. One of these large scale projects that has gotten granted consent in the UK is the Dogger Bank Creyke Beck A with an installed capacity of 1.2 GW. From previous work a base case of the collection grid for this project called the Dogger Bank Reference Wind Farm, DRW, has been developed by Kirkeby at SINTEF. This thesis has evaluated six configurations based on a section of this base case with different design of redundancy to minimize the production losses. The RELRAD methodology, power flow analysis and economic evaluations have been used to determine which is the optimum configuration for the DRW. Factors such as time varying power production, different levels of redundancy, and the costs of power loss in cables has been evaluate to determine their influence on the reliability and economic decision for the collection grid. When analysing the grid designs, only components that changed in number or placement has been evaluated. This included the 66 kV XLPE cables, circuit breakers, and disconnectors. Two of the configurations, which only offered some redundancy or none at all, was found to have much larger annual outage time than the rest, indicating that these would not be the optimum solution. However, for all configurations the expected interruption costs, cost of power loss, and the cables costs all have had huge contributions to the costs, which made the configurations with long cables much more expensive. The optimum configuration for the Dogger Bank Reference Wind Farm was found to be the multi ring configuration with two times three strings connected to a redundancy cable, and four strings connected to a redundancy cable, offering full redundancy. For all sensitivity analysis performed, the multi ring configuration was still found to be the best solution. To use this result further, a full reliability analysis of the DRW should be performed to find the actual costs of the DRW. Additionally it would beneficial to dimension the cables for short circuit events to meet the security demands for offshore wind farms.

Today, induction machines dominate the market for electrical machines in industrial pumping applications. This is also the case for subsea pumping applications. Permanent magnet machines are showing great promise for taking over this market, but in order for permanent magnet machines to succeed, their stability during load transients must be greatly improved. In this project, the aim has been to analyze the behaviour of a Permanent Magnet Synchronous Motor with damper windings during transients. In addition to stability studies, a main topic if interest has been to develop a method to comprehensively determine the individual direct-quadrant axis parameters of the machine. A particular special case of damper windings, where a metallic sleeve around the rotor does the job of both the damper winding and the magnetic retention sleeve, has been the main topic of investigation. The machine has been tested using three different materials in the damper winding. To perform these simulations, a Finite Element Method software called Ansys Maxwell has been used. A method for determining the dq-axis parameters has been developed, results using this method, in combination with the stability studies, suggest there is a direct relationship between the damper winding resistance and the damping abilities of the machine.

An active front end converter along with its associated control system has been the scope of this thesis. Relavant literature has been considered for identifying the operation and possible system implementations of this system with emphasis on its control. Different current- and voltage controllers have been studied and designed for simulation purposes. A discretized simulation model has been used to aid controller design. This simulation model has been used to investigate and compare the performance of the different controllers when employed in the stationary reference frame.

The models for hydraulic systems found in standard Power System Analysis tools are often based on simplified mathematical models. These simplifications may lead to inaccurate representations of the dynamic behaviour of hydro power plants. The objective of this Master's thesis was to study hydraulic effects and their impact on the performance of a hydro power plant from an electrical engineering point of view. A model of Riksheim power plant in Sykkylven, Møre og Romsdal has been in implemented in the hydraulic simulation tool LVTrans, in order to ensure that all hydraulic effects are included. The dynamic performance of the power plant in isolated operation was tested and compared to measurements from the real power plant. It was discovered that the hydraulic properties of Riksheim makes it almost impossible to maintain stability in all operating conditions and at the same time satisfy all specifications made by the Transmission System Operator. It is especially the effect of water hammer that has a negative impact on dynamic performance. It was also discovered that careless selection of governor settings may cause resonance between control system and the travelling waves.

Denne oppgaven er en videreføring av prosjektoppgaven skrevet av undertegnede høsten 2015. Under prosjektoppgaven ble det implementert en modell av en generator i analyseprogrammet Ansys Maxwell. Dette er et feltberegningsprogram som gjør feltberegninger ved hjelp av \textit{finite element metoden} (FEM). Modellen ble utviklet med utgangspunkt i en av Statkraft sine generatorer. Det ble implementert både en 2D- og en 3D-modell av generatoren. Videre i prosjektoppgaven ble modellen verifisert for tomgangskjøring. Både den induserte spenningen, magnetfeltet i generatoren og tapssimuleringer ble gjennomført og sammenlignet med reelle verdier fra Statkraft. Resultatet fra prosjektoppgaven var altså en modell av en generator som var verifisert i tomgang. I denne oppgaven ble modellen testet for ulike driftssituasjoner. Dette medførte at fasestrømmen i statorviklingene måtte implementeres. To forskjellige metoder ble utprøvd før det ble bestemt å tilknytte generatormodellen til en ekstern krets. Underveis i arbeidet med oppgaven ble det aksille feltet som penetrerer overflaten av stator analysert for forskjellige driftssituasjoner. Driftssituasjonene som ble valgt ut var: tomgang, undermagnetisert og overmagnetisert drift av generatoren. Gjennom simuleringsresultatene ble det konkludert med at det høyeste aksielle feltet oppstår når generatoren opererer i undermagnetisert tilstand. I simuleringene ble det plassert et målepunkt 30 mm fra luftgapet på statoroverflaten, her ble RMS-verdien til feltet målt for en periode. Her ble det målt et aksielt felt på 180-190 mT for undermagnetisert, 90-100 mT for overmagnetisert, mens feltet for tomgangskjørin var i området 80-90 mT. Disse analysene viste at det aksielle feltet, altså lekkfluksen, er størst når generatoren opererer undermagnetisert. Grunnen til dette er at magnetfeltene fra rotor og stator ved overmagnetisert kjøring av generatoren til en viss grad nøytraliserer hverandre. Denne nøytraliseringen er mye lavere ved undermagnetisert kjørin av generatoren. Videre ble det gjort 2D-simuleringer for et lite utsnitt av enderegionen til generatoren. Her ble det konkludert med at lamineringen i stator fører til at lekkfluksen kun går i de øverste statorlagene, noe som gjør at dette området er særlig utsatt for virvelstrømstap. Virvelstrømmene som oppstår i enderegionen kan føre til termsike skader på generatoren.