Investigations into the nature of the HRM in general, and a PMHRM-design in particular, were undertaken. Two separate simulation models, based on a magnetic equivalent circuit and FEM-model of the machine were built and used to find inductances, induced voltages and torque potential. The process of building a prototype was started and ran parallell to the theoretical studies. A CAD-model of the machine and it's parts was built up, and stator laminations and magnets were ordered. Plans for the assembly were made, and the assembly work started. Some main conclusions are:The PMHRM has a very high torque potential. Individual control of the coils increases makes it higher than for the traditional HRM-design. PM magnetizing of the proposed HRM-machine is most sensible for generator operation, since the actuator approach in combination with individual control removes the need for DC-magnetization.Ratio of rotor to stator radius should be close to unity to achieve a high torque densityCog wheels are not ideal mechanical bearings, as interlocking restricts the rotor ratio

Fred Olsen is currently testing their latest wave energy converter outside of Falmouth Bay in England, preparing it for commercial exploitation at the Wavehub-project. Previous studies have shown that this device has potential for increased power extraction using reactive control, but so far these investigations have focused on the hydrodynamics of the device and on reducing the peak-to-average power ratio while omitting the effect of the electric power take off system. This thesis shows the development of the hydrodynamic model of the device as well as a detailed model of the all-electric power take-off system consisting of a permanent magnet synchronous generator, inverter and DC-link. Vector Control is used to control the Permanent Magnet Synchronous Generator, and field weakening control of the generator is applied in order allow over-speed operation. Time domain wave-to-wire simulations are performed to evaluate the power take off capabilities of the modelled wave energy converter with different control parameters. When tuned according to approximate complex conjugate control the accumulated average generator losses become large, giving a very low overall system efficiency. Optimal control with respect to electrical output power is found to be with low added mass, and when compared to pure passive loading a 1% increase in annual energy production is achieved. The main factor that reduces the effect of reactive control is found to be the minimum load-force constraint of the device, which cannot be lower than 1 ton. Example simulations on a device with different force constraint are performed which verifies this characteristic.These results suggests that the Bolt2 has limited potential for increase in power extraction by implementing reactive control. The analysis in this thesis is nevertheless valuable, as it demonstrates how a wave-to-wire model can be used for power take-off investigations, annual energy production estimations and evaluations of different control techniques.

Formålet med masteroppgaven har vært å studere opp og utladninger knyttet til ulike HVDC kabel isolasjonsmaterialer. Eksperimentelle forsøk ble gjennomført for å beskrive disse fenomene for en 450 kV HVDC massekabel, og en 12 kV PEX kabel.Litteraturstudier tilsier at den dominerende polarisasjonsmekanismen for en massekabel er grenseflatepolarisasjon. Ledningsevnen til denne kabeltypen kan estimeres ved å bruke:σ = σ0e(αT+βE)Polarisasjonsstrømmen kan beskrives med:i(t) = Ke-t/τ + U/R0 Denne formelen kan brukes til å estimere relaksjonstiden, , til polarisasjonsmekanismene i massekabelen.Litteraturstudier for PEX kabelen viser at dipol polarisasjon, pga grenseflatepolarisasjon mellom to flater, er den dominerende polarisasjonsmekanismen. Det vil i denne rapporten bli brukt samme forenklinger for begge kablene for å finne ledningsevne og relaksjonstid.For å kunne gjennomføre de nødvendige forsøkene, ble det laget en datastyrt målekrets. Denne bestod i all enkelhet av en spenningskilde, en bryter, et måleinstrument og testobjektet. Kretsen ble styrt, og resultatene logget, av et dataprogram laget i LabVIEW.Resultatene fra forsøkene viste at polarisasjonsstrømmen i massekabelen kunne uttrykke som:i(t) = K1e-t/τ1 + K2e-t/τ2 + U/R0 Hvor er relaksjonstiden til oljen(10,388-56,163 s), mens (53,345-1114,1 s) er relaksjonstiden til papiret. Det siste leddet representerer DC-strømmen i isolasjonen.Polarisasjonsstrømmen til PEX: i(t) = Ke-t/τ + U/R0Hvor er relaksjonstiden i PEX(59,406-3,103 s). Ledningsevnen og relaksjontiden ble estimert som funksjon av elektrisk felt og temperatur, og ut fra disse resultatene kan man trekke følgende konklusjoner:•Ledningsevnen er størst i massekabelen (8*10-16-6,2*10-14 Ω-1m-1 mot 7,2*10-18-3,2*10-16 Ω-1m-1 for PEX)•Ledningsevnen er temperaturavhengig for begge kablene.•Ledningsevnen er kun felt avhengig ved de høyeste temperaturene, for PEX kabelen. •Relaksjonstiden reduseres med økende spenning og temperatur, for massekabelen.•Relaksjonstiden PEX lavere enn for massekabel.

With the ‘Agreement on Climate Policy’ (Klimaforliket) signed by the Norwegian government on January 17th 2008, Norway has set a goal to reduce emission caused by transportation with 2.5 – 4 million tons CO2 equivalents compared with the reference for 2020. To reach this goal, high penetration of electrical vehicles is essential, and new technologies and solutions for the infrastructure must be cleared early in the process. With the aim of triggering a discussion on the topic, this thesis presents a methodology for analysing the impact of large-scale adoption of EVs on the electrical grid. A specific portion of a real network was selected and two charging modalities for the electrical vehicles were investigated. The analysis will focus mainly on chargers located at residences, to then explore how the utility can put forward a system for smart charging strategies ("dumb" vs. "smart" charging). Data from a low voltage network was provided by NTE, located in Steinkjer in Nord-Trøndelag. Three different scenarios were analysed. Scenario 1 was given as the base scenario, were the share of EVs where 0%. This was simulated to get a proper comparison. In scenario 2, a share of 10% EVs was implemented in the grid. The share of EVs in scenario 3 was decided to be 60%. The result obtained in the analysis, verified that the smart charging approach causes less strain on the gird. The low voltage network was not capable to handle a large share of EVs (>60%) without any charging scheduling. The smart charge strategy did not cause any extra strain at the grid during peak hours. In addition, the smart charging can introduce the Vehicle-to-Home solution. The EVs can provide ancillary service and support the network with matching supply/demand and reactive power support. A simplified analysis of V2H and reactive compensation was carried out to demonstrate how the grid could benefit from an implementation of EVs.The second part of the analysis, a series of wind measurement was included into the simulation in order to see if wind power can supply the load of the entire residential area. A design for suitable energy storage was also proposed in order for the system to operate as a stand-alone system. Grid stability and power quality was not included in the analysis. The result from the wind integration shows that in order for the network to operate as a stand-alone system in the worst-case scenario, there is a need of an enormous storage. It is assumed based on the results, that the system is self-supplied most part of the year. This thesis proposes a storage consisting of 7 battery-packs from old vehicles, with the capacity of 50 kWh each. This will result in a 30% reduction of the peak demand from the grid, when wind power is integrated.The case study addressed in the thesis, present a methodology for analysis the impact of a large adoption of EVs on the distribution network. The results obtained from this analysis, is considered transferable to similar networks. In order to achieve smart charging, there is need for further research on scheduling algorithms.

This thesis describes the process of making a new engine for the car DNV Fuel Fighter 2 that participated in the Shell Eco-Marathon Europe 2012 in Rotterdam. The decision was made to construct a new engine based on an optimized design by Lubna Nasrin. The engine is an ironless axial flux permanent magnet machine with Hallbach array. As a contingency plan there was the possibility to modify the engine used in 2010 and 2011 built by André Dahl-Jacobsen. The result of the process is that the new engine is operational and it has displayed an efficiency of 68 %. However the old engine has displayed an efficiency of 86 %. Because of this, the car raced with the old engine and had an energy consumption of 163 km/kWh. It is still believed that the new engine can be a very successful engine given that it is modified. The new engine has room for improvement especially with regards to stator production and wiring connections. Suggestions are made for how to achieve a higher efficiency. This thesis describes some of the practical engineering challenges that arise when constructing a novel motor. The construction of the Hallbach array is well described here.This work has been a part of a larger project where the end goal was to participate in and win the Shell Eco Marathon Europe. This means that the engine must not only perform well on its own. The engine must also function well with the other components technically and it must be possible to make it with the time and resources available. The development of new technology is expensive and it has therefore been a priority to find funding for this project. This will also be described.

A power electronic switch-mode three-leg converter is a flexible converter and hence very useful for practical teaching of several disciplines within electric power engineering. It can be used as a half-bridge, full-bridge and three-phase converter, to mention a few, and enables the user to study many different power electronic topologies. The converter, controlled by a microcontroller, can also be used for teaching digital control of power electronics. Its output can be varied in frequency, magnitude and waveforms, and can also be measured by the microcontroller. The flexibility of the converter makes it possible to utilize it in drive circuits for a wide range of loads and can therefore also be used for teaching electric drive systems.This thesis shows a solution of how to design the switch-mode three-leg power electronic converter. The converter is designed and implemented on a printed circuit board (PCB) together with other necessary components. To meet the safety requirements of the problem description, the power rating is low, 12 A * 50 V, and the power circuit is isolated from the microcontroller on the PCB. The microcontroller chosen is the Texas Instruments PiccoloTM ControlCARD and pulse-width modulation and analog-to-digital conversion is implemented with real-time programming.This system developed is verified, except for the MOSFET drivers and measurement circuits. As time was limited, the laboratory work had to be ended in favor of writing the report. Unfortunately, this made it impossible to test the full system setup. A full description of the changes to be implemented for the whole system to be functioning and further tested is provided.A system for using the converter designed in a DC motor drive, by utilizing two of the bridge-legs as a full-bridge converter, is studied. The programming code is tailored for the specific purpose and speed measurements and control algorithms were added. Due to the converter not functioning, the testing of the DC motor drive could not be performed. However, full planning and controller implementation was done.

This Master's Thesis work deals with the analysis of the observed slow floating in the grid frequency of the Nordic power system. Measurements indicate that the frequency of this floating or oscillation is around 11-17 mHz. The reason for this floating is not known to this date, but it is suspected that it will increase the wear of the turbines that are providing primary regulation. In this work possible interaction between the hydraulic system and the power system has been emphasized and whether the governor settings may influence the slow floating, or not. The hydraulic models for the water conduit and turbine that are implemented in power system analysis tools are often simplified. In this Master's Thesis a hydraulic model that includes the effect of water hammer, surge tank and head loss has been implemented in the analysis tool SIMPOW. A three-machine equivalent of the Norwegian and Swedish power system has been established to study the response of the frequency after a severe disturbance.It has been demonstrated that for a severe disturbance a low-frequency mass oscillation will occur in the tunnel between the surge tank and the reservoir in the model. This will cause an oscillation in the pressure at the turbine which affects the grid frequency. The frequency of the mass oscillation depends on the construction of the tunnel and surge tank and will be different for every hydro power plant. Frequencies in the range of 5-11 mHz was found for the model used in this Thesis. It has been illustrated that with different characteristics for the two hydro power plants in the model, the floating in power system frequency will be the sum of the mass oscillations, which is assumed to be the case for the real power system. The governor parameters are found to have little influence on the damping of the low-frequency oscillations. For further work a study of the impact of several different hydro power plants in a more extensive equivalent of the Nordic power system is suggested. For this work emphasis should be on finding parameters for typical Norwegian and Swedish hydro power plants.

Det er innledningsvis i denne masteroppgaven gått inn på prosjekter som er pakket sammen med observasjoner og annen type informasjon som skal hjelpe nettutvikler med å prioritere og reinvestere prosjekter.Et hendelsestre for hver observasjon er så laget for å vise hvilke typer uønskede hendelser som kan inntreffe. Om en observasjon får en uønsket hendelse, er det ikke sikkert at det fører til et utfall. Det er derfor presentert en sannsynlighet for utfall i kapittelet om hendelsestre. Denne sannsynligheten for utfall brukes for å regne ut kostnadskonsekvensen sammen med varslet og ikke-varslet avbruddskostnad.Videre er det gått mer inn på observasjonene og hvilke typer som kan rapporteres fra et mastepunkt siden fokuset i denne rapporten er på mastepunkt. Observasjonene som er tatt med i denne delen er kun de som har degradering og som kan generere en levetidskurve. Fra levetidskurven er det mulig å få ut en kurve og en tabell over sviktsannsynligheten. Denne sviktsannsynligheten er basert på den forventede levetiden og 10%-kvantilen for hver observasjon. Forventet levetid og 10%-kvantilen brukes i en topunktsanalyse for å generere sviktsannsynligheten.Deretter er det brukt en formel for aggregering av sviktsannsynligheten. Der det er kun observasjoner av samme type som aggregeres sammen. Dette gjøres på grunn av at forskjellige observasjoner kan ha forskjellige reparasjons- og avbruddskostnader. I tillegg kommer sannsynlighet for utfall inn.I kapittel 5 blir regnearket som er laget for aggregering av observasjoner gått igjennom. I dette regnearket legges sannsynligheten for svikt inn sammen med avbrudds- og reparasjonskostnader. Det legges også inn antall observasjoner, sannsynlighet for utfall og kalkulasjonsrente.Når alle tallene for observasjonene er lagt inn, legges observasjonene sammen i det siste regnearket og man får ut en tabell og en kurve over de forventede kostnadene som er nåverdiberegnet ut i fra kalkulasjonsrenten.Ut i fra denne kurven og tabellen er det mulig å se når de største kostnadene kan inntreffe, og derfra er det mulig å si noe om når reinvesteringene kan gjøres. Man får en oversikt over om det er enkelte observasjoner som genererer store kostnader enkelte år og som bør reinvesteres før andre observasjoner.I regnearket er det også mulig å legge inn investeringskostnader, samt i hvilket år man ønsker å reinvestere. Alle tallene som presenteres er nåverdiberegnet og presenteres for hvert år i en analyseperiode på 30 år. Videre blir det gått inn på hvordan aggregeringen fungerer og hvordan det går an å aggregere forskjellige observasjoner. Om det aggregeres for forskjellige observasjoner sammen, kan det ikke legges inn kostnader siden disse er forskjellige for hver observasjon.Det blir også sett på hvordan topunktsanalysen fungerer og det er vist eksempel av når en type observasjon blir aggregert for 1 til 5 observasjoner. Det som vises da er at kurven for sviktsannsynligheten vokser og man får større sannsynlighet for svikt tidligere i tid. Dette er logisk siden man for flere observasjoner har større sannsynlighet for svikt tidligere i tid. Så for én observasjon kan kurven være slakk, men om man aggregerer inn flere observasjoner vil den bli spissere, altså større sannsynlighet for svikt, og flyttes frem i tid.Etter at det er sett på hvordan aggregeringen fungerer og hvordan den brukes med kostnader, er det sett på tre reelle prosjekter fra NTE. I disse prosjektene er det observasjoner som kan generere sviktsannsynligheteskurver. Hvert prosjekt er regnet på med kostnader. Når det er kjørt en analyse av prosjektet basert på aggregering av observasjoner, blir det satt opp to eksempler ut i fra hvordan de forventede kostnadene fordeler seg ut over analyseperioden.I alternativene er det gjort reinvesteringer der det reinvesteres for hele prosjektet i samme år, og der det reinvesteres noe i tidligere år og resten av ved et passende tidspunkt senere i analyseperioden.Ofte vil en svikt trigge hele reinvesteringen, men det regnes også på å reinvestere over flere år.I siste kapittel er det foretatt en følsomhetsanalyse som ser på faktorer som påvirker aggregeringen. Der er det gjort små forandringer på antall observasjoner, sannsynlighet for utfall og avbruddskostnaden. Det viser seg der at sannsynlighet for utfall og avbruddskostnaden har mye å si. Mens antall observasjoner har mest å si om det er en kritisk observasjon med høy sannsynlighet for svikt. Det vises også at investeringen er viktig om denne er høy sammenlignet med avbruddskostnadene.

Kryssbunden polyetylen (PEX) er den mest brukte polymerisolasjonen i Noreg. PEX blir brukt i kablar opp til 420 kV. Auka spenningsnivå i høgspenningskablar gjev auka termiske påkjenningar og PEX kablar har ein maksimal driftstemperatur på 90 °C. Framover vil det bli behov for kablar som kan tåle høgare driftstemperaturar. I denne oppgåva blir polymeren syndiotaktisk polypropylen (s-PP) undersøkt til bruk som isolasjon i høgspenningskablar, spesielt ved høge temperaturar. s-PP har ein smeltetemperatur på 130 °C, låge dielektriske tap og høg gjennomslagsstyrke cite{Yoshino-mekanisk}. For å halde ynskt driftsikkerheit på høgspenningskablar av polymerar er det viktig med kunnskap om nedbrytingsmekanismar som elektrisk trevekst. Hovudformålet i denne oppgåva har vore å studere samanhengen mellom danninga av elektriske tre og partielle utladningar i både PEX og s-PP. Eigne testobjekt av både PEX og s-PP har vorte produsert. I testobjekta er det brukt stålnåler som elektrode for å skape område med lokal feltforsterking. Elektrisk trevekst har blitt undersøkt ved 8 kV, 10 kV og 12 kV, med ein frekvens på 50 Hz. Resultat frå denne oppgåva viser at elektriske tre i PEX og s-PP har ulik form. I PEX endrar treet form frå greinstruktur ved 8 kV til buskstruktur ved 10 kV og 12 kV. Det vil sei at antall forgreiningar aukar med auka spenning. Tid fram til gjennomslag aukar frå 7 minutt og 12 sekund ved 8 kV til 34 minutt og 30 sekund ved 12kV. Elektriske tre i s-PP derimot har greinform både ved 8 kV, 10 kV og 12 kV. Greinstruktur i s-PP skil seg frå greinstruktur i PEX, den har færre forgreiningar fram til treet når jord. Tida treet brukar for å nå jordelektroda minkar med auka spenning, i motsetnad til elektriske tre i PEX, frå 9 minutt og 36 sekund ved 8 kV til 2 minutt og 54 sekund ved 12 kV. Oppsiktsvekkande resultat for elektriske tre i s-PP er at dei ikkje nødvendigvis fører til gjennomslag idet fyrste grein når jordelektrode. Nye greiner vil fortsetje å vekse parallelt fram til jordelektroden til det blir gjennomslag. Tid til gjennomslag har stor variasjon, men minkar i gjennomsnitt frå 3 timar og 47 minutt ved 8 kV til 5 minutt ved 12 kV. Det er observert ein korrelasjon mellom partielle utladningar og elektrisk trevekst. For elektriske tre i PEX, der det er flest forgreiningar og stor tettleik av greiner, er dei partielle utladningane større enn dei er for s-PP. Ved 10 kV er den maksimale PD-amplituden ti gonger større i PEX enn for s-PP. Ved auka spenning frå 8 kV til 12 kV i PEX aukar antall forgreiningar og maksimal PD-amplitude går frå omlag 350 pC til 900 pC. Elektriske tre i s-PP derimot har lita endring i antall forgreiningar og den maksimale PD-amplituden held seg rundt 70 pC ved auka spenning.

The properties of a Modular Multilevel Converter (MMC) are investigated. This is a new and promising converter type for High Voltage Direct Current (HVDC) applications. A case with a MMC connected to a stiff grid through a filter and a transformer is analysed both theoretically and in simulations, with focus on three main topics.The first topic is the available voltage at the converter terminal and modulation strategies. Theoretical evaluations are compared with simulation results. Two different methods are used to analytically investigate the relationship between modulation index and terminal voltage. The effect of third harmonic distortion is also considered. Simulations proved that the converter followed the theoretical modulation curve. Qualitative comparisons of theoretically developed voltage wave shapes with simulation results with different modulation indexes were successful, but quantitative comparisons proved challenging. Modulation with third harmonic distortion showed very convincing results in the simulations.The second topic is the power equations valid for the case circuit. Relations between active power, reactive power, grid angle, and converter voltage were investigated. A phasor approach was used in the deduction of the theoretical formulas, and these formulas were verified by the simulation results. The grid angle range giving stable operation was defined by the angle stability limit for power systems. The ratio between the resistance and the inductance in the connection was found to have a large influence on the characteristics of the connection, for instance by defining the stable operation conditions. In the theoretical analysis the magnetizing effects of the transformer were disregarded. Simulations showed that this approximation was appropriate. A surprising simulation result was that the terminal voltage of the converter was depending on the grid angle. Further investigations are needed to identify the cause and compensate for this effect.The third topic is the converter control system. The converter was modelled in the dq reference frame, and the model was used to derive and tune the control system. The control system was made with a cascaded structure, controlling either active and reactive power at the grid or active power at the gird and the rms value of the voltage at the converter terminal. The inner loops controlling the current were tuned with modulus optimum to achieve fastness. The outer loops for power were tuned to be somewhat slower than the inner loops, while the outer loop for rms value was tuned with symmetric optimum. All the control loops were successfully implemented into the simulation model. This shows that it is possible to control parameters at two different locations in the network at the same time. Simulations showed that it is possible to change the control mode between reactive power control and rms value control, without making the control system unstable. This demonstrated both successful decoupling and a robust control system.