This thesis deals with the development of self-sustained sensors based on non-conventional measurement techniques. The motivation behind this topic is to help utility companies reduce their fault handling time and improve the way faults are handled. As of today, the distribution system does not have widespread surveillance and automation systems in place. The thesis therefore investigates a sensor design based on Rogowski coils (RC), with both measurement and power supply capabilities.The concept of printed-circuit-board Rogowski coils (PCB RC) and conventional RCs for current measurement is presented. A theoretical, dimensional study with respect to feasibility and output signal magnitude is then performed. On basis of this analysis, a PCB RC prototype is manufactured. For the sensor to be implemented in the resonance grounded distribution system, it must be able to measure currents of low magnitude, as the fault current is compensated by Petersen coils. The prototype is therefore tested for common distribution system current values. Due to limitations during construction and design flaws, the PCB RC performance was poor. The induced voltage error, with respect to the theoretical response was 64.3 - 69.1 %.As the prototype's ability to supply enough power to the sensor circuitry (demand found to be > 10 mW) proved so limited, a more thorough analysis concerning the RC is presented. Obtaining this amount of power is, theoretically possible and easier with a conventional RC. Its maximum average power supply capability was found to be 10.4 mW during ideal conditions. Even better results can probably be obtained as well. Based on this RC, the complete sensor system is discussed. This include rectification, digitizing, integration and boosting of the RC voltage to 3 V for battery application. A survey of possible fault localization methods for resonance grounded distribution systems is presented and an implementation is suggested.Finally, the prototype is tested with short circuit currents of 30 kA amplitude. The performance was better than for the load current test, but the magnitude error was still substantial. It varied from 20 % to 40 %. However, the lab's signal integration showed that the prototype reproduced the applied current waveform accurately.The simulations, testing and analysis performed indicate that a RC can be used for both supply of power and measurements. However, a prototype should be manufactured in order to assess this thesis' theoretical results. Combined with the proposed auxiliary circuitry and implementation strategy, it can improve the utility's fault handling.

In this thesis work, different proposed methods for detecting and locating short circuit faults in Multi Terminal HVDC grids have been evaluated by implemenation and transient simulations in PSCAD. The research has been limited to cable based systems. HVDC grids have seen increasing interest in recent years, but have yet to be fully realised. Suitable converter technology was introduced in 1997 and has been further developed since, while switches able to interrupt large DC currents were introduced in June 2013. One of the major issues left is fault localisation. Due to the low impedance in HVDC cable systems, fault currents rise to severe magnitudes system wide in a matter of a few milliseconds. This may cause damage to the converter diodes if not dealt with quickly.In order to obtain a better understanding of the fault propagation, research into the subject is presented. It is found that the capacitors in the converters is a main source of large fault currents, and fast fault detection is essential for protection of converter components. Time between first detection of fault until current interruption should be within a few milliseconds. Different methods have been proposed for localisation of faults in recent years. These are presented together with traditional fault localisation methods, and briefly discussed with the intent of deciding which to implement and evaluate in PSCAD.Protection based on current derivative and wavelet transformation, as well as travelling wave protection is chosen and implemented in a three converter VSC system. Different fault types are applied at various locations with varying system capacitance and up to 16 $Omega$ fault impedance. The results indicate that none of the three methods are able to detect and locate all impedance faults on their own. The travelling wave protection is suitable for short lines, but fails when exposed to high fault impedances and distances. By using derivative polarity to determine direction of fault and wavelet magnitude to determine distance, faults are successfully located in a high capacitance system within a respectable time. It is concluded that all the three tested methods should be considered for implementation when designing a future HVDC protection system.

In this study, existing work, opportunities, standards and essential safety aspects on development of LVDC distribution systems have been considered. It can be summarised that the standard requirements for AC systems also applies to DC systems, and that one does not come apart DC-DC regulators in equipment due to the galvanic separation which is needed to fulfil the safety requirements. One of the main challenges of introducing an LVDC system is the non-existence of building codes and standards intended for an LVDC system in buildings.The critical step of developing a DC distribution system is selecting the voltage level. Earlier studies conclude that:•A standard voltage level of 230 V DC is proven to be sufficient for normal office loads, as long as the cable length does not exceed 80 m•For higher power loads up to 6.5 kW the voltage level has to be increased to 326 V DC for systems with a cable length of maximum 47 m and 2.5 mm2This study proposes an LVDC system supplied by converted power from the main grid (AC-DC), where the size of the LVDC system is decided by performed measurements at an example building. In order to compare the losses in an AC and DC system, it was chosen to perform simulations on component level. It can be concluded that:•An LVDC system supplied by a central VSC converting power from the AC grid is an energy efficient system solution compared to the existing AC system solution, largely depending on the performance of the VSCSimulations and calculations resulted in a requirement of performance > 97.7 % of the VSC in order for it to be more energy efficient with an DC system instead of the existing AC system solution with AC-DC converters in each link with a performance of approximately 97 %. Experts believe that it is possible to gain this performance in the future based on performance for smaller converters developed by leading manufactures.Based on the different aspects considered regarding introduction of an LVDC distribution system, simulated models, and performed measurements, it can be concluded that:•From earlier studies it can be concluded that on the economic side, an LVDC distribution system seems benefitual•Original DC loads will benefit from having a separate DC system in terms of power loss at a voltage level of 230 V DC•AC loads require high power delivered and it will be most energy efficient to keep the connection to the existing AC system•An LVDC system with a size of approximate 20 kW is realistic in the future, supplying a floor in a building with DC loadsFrom the material presented in this report, it can be summarised that it is a realistic possibility for future distribution systems in buildings to have an energy efficient mixed supply system, AC and DC.

Particle Swarm Optimization (PSO) and Genetic Algorithms (GA) are usedto reduce the cost of a permanent magnet synchronous generator with concentratedwindings for tidal power applications. Reducing the cost of the electricalmachine is one way of making tidal energy more competitive compared to traditionalsources of electricity.Hybrid optimization combining PSO or GA with gradient based algorithmsseems to be suited for design of electrical machines. Results from optimizationwith Matlab indicate that hybrid GA performs better than Hybrid PSO forthis kind of optimization problems. Hybrid GA shows better convergence, lessvariance and shorter computation time than hybrid PSO.Hybrid GA managed to converge to an average cost of the generator that is 5.2% lower than what was reached by the hybrid PSO. Optimization results showa variance that is 98.6 % lower for hybrid GA than it is for hybrid PSO. Movingfrom a pure GA optimization to the hybrid version reduced the average cost31.2 %.Parallel processing features are able to reduce the computation time of eachoptimization up to 97 % for large problems. The time it took to compute aGA problem with 2500 individuals was reduced from 12 hours to 21 minutesby switching from a single-processor computer to a computer with 48 processorcores. The run time for PSO with 400 particles and 100 iterations went from18.5 hours to 74 minutes, a 93 % reduction.

This Master’s thesis proposes a method for implementing an enhanced hydropower planning formulation in a long-term expansion planning model. The motivation for this work is the important role that hydropower plays in a generation investment environment. In a time where penetration of intermittent renewable energy sources like wind and solar power is heavily increasing, new challenges in the continuous balancing of supply and demand are also introduced. Hydropower and its use of reservoirs as power batteries can respond more or less immediately to such fluctuations. As such, a detailed framework for hydropower scheduling is highly relevant.The presented implementation is carried out in an already-existing expansion planning model for Europe called EMPIRE, which is written in Mosel Xpress. This is a two-stage stochastic optimization model whose objective function is to minimize the total net present value of expected operational costs and investment costs for generation and transmission capacities.The main feature of the proposed framework involves penalization of hydropower through water values. This necessitates a complete hydropower scheduling representation where each reservoir is divided into segments which are assigned a fictitious marginal cost. The inclusion of water values enables comparability with the short-run marginal cost for competitive technologies and introduces the important aspect of conserving water for other periods of the year. Data from SINTEF Energy Research has been used for this purpose.Results from optimization runs in the time span from 2010 to 2060 for an EU 20-20-20 like policy scenario show that the original hydropower availability is too relaxed, thereby causing an overvaluation of this technology. The revamped cost representation by means of water values leads to a lower utilization of hydropower relative to the original model. An earlier deployment of solar power is carried out to replace the lower generation, with a capacity difference between the final and original models peaking at 45% in 2040. Total costs in the system are therefore increased. For both models extensive investments in intermittent renewables are taking place, amounting to 47% of the total capacity in 2060.

This project was done in cooperation with the TE-EPC-LPC section at CERN. They were experiencing failures in one of their power converter from the rectifying power diodes, and were interested in finding the reason for the failure. They were observing a high leakage current in some of the diodes. The purpose of this project was therefor to investigate the reliability of a diode, and its different failure mechanisms. Diodes can fail from multiple mechanisms some of which can be detected under operation of the diode, in this project the possibility of predicting the lifetime of the diode from its reverse leakage current was investigated. CERN suspected that the failure was either due to avalanche currents in the device or because thermal cycling in the device. Therefor was a non-repetitive avalanche current test, repetitive avalanche current test and a power cycling test performed on the diode. The non- repetitive avalanche current test gave no failures, proving that single event avalanches were not the reason for failure. In the repetitive avalanche current test the diodes was crashed, but no connection between leakage current and time before failure could be observed. In the power cycle test it was observed that failure was induced much faster in a diode with a much higher reverse leakage current than another.

On the way to meet the internationally sanctioned climate targets, zero emission buildings / zero energy buildings (ZEB) will be an important step. Research is ongoing on what a reasonable definition of ZEB will contain. In Norway, it is decided that the building code should be nearly zero energy buildings from the year 2020. In this master’s thesis, an optimization model for finding cost-optimal investment and operational strategies for ZEB is developed. The building modelled, is a passive school with a hydronic heat distribution system. Possible investments include photovoltaic solar cells (PV), solar collectors, heat pumps, biomass boilers, electric boiler, heat storage and connection to the district heating grid. The model is designed as a dynamic mixed integer programming model, and implemented in Mosel Xpress. The model minimizes the total discounted costs of operations and investments over the lifetime of the building. Different restrictions of zero CO2 emissions, zero primary energy consumption and level of grid burden can be applied. The analysis shows that if a zero CO2 restriction with Norwegian CO2 factors are applied, the least expensive way to reach ZEB is by investing in PV in combination with pellet biomass boiler as base load and district heating to cover peak demand. To reach the zero balance for the school with Norwegian CO2 factors, the highest hourly value for export of electricity per hour exceeds the maximum hourly value of imports by about 120%. If European factors for CO2 is applied, it will be more reasonable to reach ZEB than with Norwegian factors. If asymmetric primary energy factors are used instead of symmetric factors, investment in PV becomes higher, and the peak export values increases. The model is developed as a deterministic model, and does not take into account uncertainties in input data. To compensate for this, various sensitivity analyses are conducted. Future work includes testing the model with load profiles for other types of buildings.

The objectives of this master thesis have been to model and simulate oxy-combustion CO2 capture in a cement plant. The model developed is a process simulation of the calcination process with varying degree of air in-leakage, where heat is supplied by combustion in an oxygen rich environment, followed by capture of the CO2. The further gas separation after H2O condensation to achieve the required CO2 quality was evaluated. In addition to the process simulations, a review of literature related to oxy-combustion CO2 capture and cement production was performed, and an engineering evaluation of the necessary modifications to the cement plant conducted.A simulation model was built in Aspen HYSYS, and student Jelmer de Winter’s project work was utilized as a starting point. The model was developed with the aim to achieve results comparable to a process model constructed by the European Cement Research Academy (ECRA) in 2009. The goal was to capture as much of the CO2 as possible, and to achieve a CO2 purity of minimum 95 mol-% after the CO2 Compression and Purification Unit (CPU).CO2 purity in the dry flue gas of ~85 mol % was achieved, with a CO2 capture rate up to 96.4 %. Five different air in-leakages (2, 4, 6, 8 and 10 % of total flue gas flow) were tested. The results showed that the CO2 concentration in the flue gas decreased with increasing degree of air in-leakage. The decrease in CO2 concentration causes an increase of the power consumption of the CO2 CPU of ~2.6 % per percentage point of air in-leakage, and the CO2 capture rate was also reduced when the air in-leakage increased. These results agree well with results from previous oxy-combustion studies, and show the importance of minimizing air in-leakages in the cement plant.If oxy-combustion capture is to be utilized at a cement plant, some process modifications and additional equipment is required. An Air Separation Unit (ASU) is needed to provide almost pure oxygen for the combustion process. A Compression and Purification Unit (CPU) is also required, in order achieve the necessary CO2 purity and transport conditions. When using oxy-combustion technology, both the material conversion in the cement kiln system and the operational specifications of the overall process are different from those in conventional kiln operation. However, research made by ECRA in 2012 showed that the negative impacts of oxy-combustion on the product quality seem to be negligible.Other necessary process modifications when retrofitting with oxy-combustion are news design of the kiln burner and the clinker cooler in the cement plant. In addition, prevention of excessive air in-leakage by improving sealing locations at the cement plant is necessary, as the simulation results show. This is possible e.g. by waste gas flushed systems, or by an improved maintenance of inspection doors and similar devices. The CPU is up to a certain point capable of handling changes in the flue gas composition at short-term inspections; however it limits its efficiency.

The operational properties of a bipolar configuration composed of a Line-Commutated Converter (LCC) and a Voltage Source Converter (VSC) in a multilevel topology (MMC) are investigated. Simulation models are made in the PSCAD software in order to represent the two bipolar HVDC-links between Norway and Denmark, named Skagerrak 3 and Skagerrak 4. The topology combines two fundamentally different converter technologies with an intricate switching scheme for the function of power reversal.The objective of this thesis is to validate the three simulation models; a monopolar LCC model, a monopolar Modular Multilevel Converter (MMC) model, and an LCC-MMC model in a bipolar configuration. By investigating the interaction of the combined LCC model and MMC model, a better understanding of the challenges of the bipolar link is achieved. The steady state behaviors of the three systems are as expected based on the theory. The LCC consumes reactive power, and require reactive power compensation and filtering. When combined with the MMC, the MMC is able to provide the reactive power consumption by the system. There is an interaction between the MMC and the passive filters of the LCC, resulting in the need to redesign the filters in the bipolar configuration. Also, the DC voltages in the combined model are found to be divided somewhat unevenly between the HVDC-links.When a three-phase to ground fault is applied to the inverter side of the models, the LCC suffer from failure of commutation. This is due to the reduction in the AC voltage, and is to be expected. A DC chopper is placed at the MMC DC-link and reduces the DC voltage overshoot significantly during the fault. When the LCC and MMC systems are combined, they operate in an independent manner, and as a result, no additional protection arrangements are necessary.A single-phase to ground fault at the inverter side of the models, gives rise to harmonic disturbances due to the unbalance of the fault. The reduction in AC voltage results in failure of commutation at the LCC inverter side. The LCC and MMC systems are still almost unaffected by each other during the fault, and so no additional protection systems are required for the combined model.Finally, a DC pole to ground fault is applied to the three simulation models. The LCC model does not recover from the fault, because of loss of controllability in the PI controllers. There are solutions for this problem, for example anti-windup PI controller. An interesting result is observed when the two models are combined into the bipolar configuration. The MMC is able to support the LCC in such a way that it recovers. DC breakers are required for high voltage and current spikes during the DC fault at the MMC link. The LCC and MMC systems are no longer operating independently, and different protection strategies must be implemented when compared to the monopolar models.

Det nye Sparebankkvartalet i Trondheim ble tatt i bruk høsten 2010 og er satt opp av Sparebank 1 SMN (Sparebank Midt-Norge). I tillegg til å ha et energisparende bygg, var målet å oppnå meget godt inneklima i lokalene for å fremme trivsel og helse blant de ansatte. HR-avdeling har hatt løpende kontakt med brukerne av bygget for å registrere hvordan opplevelsen i nytt bygg har vært. Mange av brukerne er fornøyde med de nye omgivelsene, men en del har problemer med at temperaturen er for lav, som arbeidsdagen negativt. NTNU har i samarbeid med Sparebank 1 SMN foretatt beregninger og bekreftet at løsningen med eksponert betong, datagulv og fortrengningsventilasjon var mer energieffektiv enn en konvensjonell ventilasjonsløsning i himling med omrøring. Denne masteroppgaven dokumenterer inneklimaet og kartlegger hvordor brukerne synes det er for kjølig, i tillegg til at et uttrykk for den termiske massens påvirkning av lufttemperaturen er funnet. Mye av arbeidet ligger i valg av måleparameter, måleutstyr,samt finne ut hvilke parameter som er representative for evaluering av inneklimaet. Måleparametrene som ble valgt ut som representative for inneklima, var operativ temperatur, lufthastighet, lufttemperatur, luftmenge og CO2-nivå.På grunn av manglende måleutstyr ble det ikke gjennomført målinger av den operative temperaturen. Det ble heller ikke gjennomført målinger av luftmengden. Det ble besluttet å bruke verdier på luftmengden fra sentral driftskontroll(SD)-anlegget ved evaluering av inneklimaet.Første måling bidro til etablering av et uttrykk for den termiske massens påvirkning på lufttemperaturen pr. meter. Termoelementer ble benyttetfor å måle overflatetemperaturen til betongen i plenumet, samtidig som lufttemperaturen inn og ut av plenumet ble registrert med utstyr fra TinyTag. Differansen mellom temperaturen inn i plenumet og temperturen ut av plenumet ble estimert til 3,5°C, og temperaturøkningen på tilluften i plenumet pr. meter ble estimert til 0,22 °C /m. Den store temperaturdifferansen kan føre til variasjoner av tilluftstemperaturen over kontorlandskapet.En sammenligning av CO2-nivået fra SD-anlegget ble gjort med resultat fra målinger gjennomført med en datalogger fra Kimo, tatt i arbeidssone E3. Deretter ble plassering av givere i landskapet undersøkt, for å se om de var representative for CO2-konsentrasjonen i kontorlandskapene. Plasseringen viste seg å være tilfredstillende og giverne er ikke i veien for brukerne. De måler rett verdi på CO2-konsentrasjonen i rommet, og de får med seg variasjonene i løpet av dagen. Evalueringen av inneklimaet ble basert på lufttemperaturen istedetfor den operative temperaturen, hvor det ble antatt at overflatetemperaturene i landskapet var normale.Gjennomsnittsverdien på lufttemperaturen i kontorlandskapet er innenfor akseptabelt område, 21-23,5°C, i løpet av dagen. Resultater fra spørreundersøkelsen viser at brukerne er mer fornøyde med temperaturen etter lunsj, men det er likevel et ønske om økt temperatur i lokalet gjennom hele dagen. Mindre nattkjøling av den termiske massen, kan gi en høyere tilluftstemperatur til kontoret på begynnelsen av dagen og dermed imøtekomme ønsket om en høyere tempertur i landskapet.Den maksimale lufthastigheten målt under første måling under et skrivebord i landskapet, var til tider helt oppe i 0,25 m/s i høyden 0,6 m. Det maksimalt forventede antall misfornøyde på 14,25 %, ved målinger i høyden 0,6 m over gulvet under et skrivebord. Turbulensintensiteten var ikke målt, men ble satt til 40 %. For en fullverdig konklusjon burde man måle strålingen fra overflater i landskapene, for å se om det er noe mer enn trekk som kan resultere i at brukerne synes at det er kaldt. Noen av ventilene er plassert relativt nærtbrukerne, og en ny plassering av ventilene bør vurderes. Kommentarer om trekk fra avvtrekksventilen ovenfra og ned på brukerne har ikke blitt undersøkt nærmere, men dette kan være en faktor som øker ubehaget hos brukerne. Luftmengden er innenfor tillatt grense gitt av tek10, det samme gjelder CO2-konsentrasjonen, som holder seg under 900 ppm.