Read Microsoft PowerPoint - forside.ppt text version

Sammendrag av

Masteroppgaver 2009

Masteroppgaver 2009

Institutt for elkraftteknikk

NTNU Norges teknisk-naturvitenskapelige universitet Fakultet for informasjonsteknologi, matematikk og elektroteknikk Institutt for elkraftteknikk

Example of studied case with a cluster offshore wind farms and oil rig connected with a Single VSC-HVDC transmission to the main grid

1

INNHOLDSFORTEGNELSE

KANDIDAT

P.

VEILEDER

TITTEL

4 Arvidsen, Nina Kristine Sogn Wangensteen, Ivar 6 Assefa, Hana Yohannes

9 Barfod, Emma 10 Boinne, Raphael 12 Breistein, Hallvard 14 Dalen, Ingar 15 Ditlefsen, Arne Marius 16 Dorji, Tempa 19 Eidsaune, Christian 21 Einervoll, Torger 23 Eldrup, Martin 25 Engebrethsen, Eivind 26 Faleide, Rolv Marius 27 Feilberg, Espen

Smarte nett med smarte målere Stability Investigation of an Advanced Electrical Rail Molinas, Marta Vehicle Energy Storage Solutions for Wave Energy in Undeland, Tore Marvin Stand Alone Systems Stability Studies of an Offshore Wind Farms Cluster Connected with VSC-HVDC Transmission Gjengedal, Terje to the NORDEL Grid Fixed Speed Electric Motor Drives for LNG Nestli, Tom F. Refrigeration Compressors. Wangensteen, Ivar Nestli, Tom F. Sand, Kjell Norum, Lars Einar Norum, Lars Einar Fosso, Olav B Economic Benefit of New Capacity in the Central Grid Wave Energy Conversion Reliability Assessment of Distribution Systems Power Supply for Down-hole Instrumentation and Actuators Impact on Wind Turbine Systems from Transient Fluctuations in Offshore Utility Grids OPTIMAL PRODUKSJON AV REAKTIV EFFEKT I REGIONALNETTET I TELEMARK & VESTFOLD

Runde, Magne Eystein Testing av superledende spoler Shell Eco Marathon Investigation of the Doubly Fed Permanent Magnet Undeland, Tore Marvin Synchronous Machine Opprustning av Svorkmo kraftverk Large scale integration of wind power in a transmission grid Configuration of large offshore wind farms Simulering av spenningskvalitet i kraftnett Nilssen, Robert

28 Finnanger, Idunn Gangaune Doorman, Gerard 30 Finstad, Hans Wigen 31 Flo, Randi Aardal 33 Fossen, Kjell Erik 34 Gellein, Anders 35 Gjerde, Sverre Skalleberg 37 Gleditsch, Morten 39 Gryttingslien, Geir 41 Hatefi, Seyed Behzad Fosso, Olav Bjarte Gjengedal, Terje Sand, Kjell Norum, Lars Einar

Framdriftssystem til Eco marathon kjøretøy STATCOM and Energy Storage in Grid Integration Undeland, Tore Marvin of Wind Farms Gjengedal, Terje Sand, Kjell Nilssen, Robert Balancing of Offshore Wind Power in Mid-Norway Spenningskvalitet i distribusjonsnettet Development of Procedures for Analysis of PMSM based on Simplified Permeance Network Offshore Wind Farm Layouts

43 Haugsten Hansen, Thomas Gjengedal, Terje

2

45 Holt, Øystein 47 Kvamme, Christina 49 Mensah, Kwaku Sarpong 51 Mikkelsen, Joar Hylland 53 Ose, Heidi Theresa 55 Ravnaas, Kristian Wiik 57 Renaudin, Fabien 58 Rotevatn, Njål 60 Rui, Øyvind August 62 Simonsen, Silje Odland 64 Singelstad, Håvard 66 Skaflestad, Stig 68 Skogsrud, Johan 69 Steinholt, Arnulv 71 Stray, Kristian 73 Stølan, Ronny 75 Tambatamba, Terence 79 Tommelstad, Einar 81 Ulsund, Ragnar

Undeland, Tore Marvin Pressure Tolerant Power Electronics Doorman, Gerard Fosso, Olav B Norum, Lars Einar Doorman, Gerard Doorman, Gerard Investering i vannkraftverk Improving Stability of Ghana's Power System Using Power System Stabiliser(PSS) Decision support from monitoring of hydro power stations Optimal use of the hydro resources in Albania

Optimale bud for en vindkraftpark Integration and Stability of a Large Offshore Wind Farm with HVDC Transmission in the Gjengedal, Terje Norwegian Power System Design and testing of Flux Switched Permanent Nilssen, Robert Magnet (FSPM) Machines Optimal Operation of a Stand-Alone Power Undeland, Tore Marvin Supply using Artificial Intelligence Development of a Grid Connected Norum, Lars Einar PV System for Laboratory Use Stabilitetsanalyse ved integrasjon av Midtfjellet Fosso, Olav B Vindpark i kraftsystemet Doorman, Gerard Hansen, Eilif Hugo Solvang, Eivind Fosso, Olav B Nysveen, Arne Wangensteen, Ivar Hansen, Eilif Hugo Balance costs for windpower Undersøkelse av beregningsverktøy for elektriske lavspenningsinstallasjoner Nettutviklingsplan for Tromsøya-Nord Ny HVDC-kapasitet og vindkraft i Sør-Norge Losses and Inductive Parameters in Subsea Power Cables Improvement of power supply reliability Energisparepotensiale ved bruk av intelligente veilyssystemer

Undeland, Tore Marvin Offshore Power Transmission Dielectric Spectroscopy of Bisphenol A Epoxy 82 Vaishampayan, Deep Runde, Magne Eystein Resin Aged in Wet and Dry Conditions Integrated Gas and Electricity Markets in 84 Westeng, Mari Røhmesmo Doorman, Gerard eTransport System Analysis of Large-Scale Wind Power 86 Øren, Lars Pedersen Gjengedal, Terje Integration in North-Western Europe

3

Smarte nett med Smarte målere

Student: Faglærer: Nina Kristine Sogn Arvidsen Ivar Wangensteen

De siste årene har fokuset på klimaendringene økt. Klimaspørsmålet blir antagelig den viktigste drivkraften i energi- og miljøpolitikken framover. Det er også den viktigste drivkraften for å innføre Avanserte Måle- og Styringssystemer (AMS). I mange land er det er stor aktivitet for å installere AMS i alle målepunkter. Ingen land er i dag ferdig. Sverige er antagelig kommet lengst. Italia var det landet i Europa som kom først i gang med en storskala innføring. AMS er første steg i utviklingen av Smarte nett. Smarte nett er basert på at sluttbruker er i sentrum, integrasjon av fornybare energikilder og distribuert produksjon. Hensikten er at det Smarte nettet skal samordne handlingene til alle aktørene i elektrisitetsnettet, så de sammen kan levere bærekraftig, økonomisk og sikker elektrisitetsforsyning. I oktober 2008 sendte NVE ut et høringsdokument med forslag til endring av FOR 1999-0311 nr 301: Forskrift om måling, avregning og samordnet opptreden ved kraftomsetning og fakturering av nettjenester. Avgjørelsen er ikke tatt, men NVE vil høyst sannsynlig pålegge nettselskapene å innføre AMS til alle sluttbrukere. Et revidert forslag til forskriften antas å komme på høring i løpet av sommeren 2009. 10 % av landets husstander har AMS. Årsaken til at ikke flere har det er at det har vært opp til nettselskapene om de har ønsket å installere AMS. Til nå har dette ikke vært bedriftsøkonomisk lønnsomt. I tillegg er det stor usikkerhet knyttet til kostnader, nytteverdier, teknologi og nye forskrifter. Innføring av AMS påvirker ikke bare nettselskapene. Sluttbrukere, kraftleverandører og regulerings- og tilsynsmyndigheter blir også påvirket. I en samfunnsøkonomisk nyttekostnadsanalyse må virkningen for disse og samfunnet som helhet tas med. Analysen inkluderer også hvordan AMS vil påvirke miljøet og hvilke muligheter AMS gir på sikt. De viktigste nytteverdiene ved AMS er: Sluttbruker: - Fakturering etter faktisk forbruk og ikke basert på stipulerte verdier - Økt konkurranse i sluttbrukermarkedet og dermed lavere kraftpriser - Avvikling av selvavlesning - Bedre informasjon om elektrisitetsforbruk og priser som gir grunnlag for effekt- og energireduksjon Kraftleverandør: - Mindre pris- og volumrisiko - Mer effektive leverandørbytter - Mulighet for bedre omdømme Nettselskap: - Reduserte måle- og avregningskostnader - Bedre datagrunnlag for nettplanlegging

4

- Færre klager fra sluttbrukere - Mer effektive leverandørbytter - Muligheter for tilleggstjenester

Mulighet for bedre omdømme Miljø - Redusert energiforbruk - Større fleksibilitet i etterspørselen - Mulighet til å integrere lokal fornybar energiproduksjon Nytte-kostnadsanalysen jeg har gjennomført viser at innføring av AMS til alle sluttbrukere er samfunnsøkonomisk lønnsomt. Om investeringen er bedriftsøkonomisk lønnsom er derimot usikkert. Nytte-kostnadsanalysen gir en samfunnsøkonomisk nytte på kr 1182 per måler per år. Nytten tilfaller hovedsakelig sluttbrukerne, men også nettselskap, kraftleverandør og regulerings- og tilsynsmyndigheter. Nettselskapets nytte blir kr 200 og samfunnets nytte blir kr 982. Dette gir et forhold mellom nettselskapets nytte og samfunnets nytte på 1:5. Hvis en antar økning i nettleien på kr 380, vil netto nytte for sluttbruker bli kr 472. Redusert energiforbruk gir stort utslag på nytteverdien. I nytte-kostnadsanalysen er det antatt 5 % reduksjon i energiforbruk. Hvis en isteden antar 10 %, har dette punktet alene en nytteverdi på kr 1300. Det gir en netto nytte på kr 1122 for sluttbruker. Hvis energiforbruket ikke endres ved innføring av AMS gir det sluttbruker et tap på kr 178. Selv om sluttbruker taper på innføringen er den fortsatt samfunnsøkonomisk lønnsom, med en netto nytte på kr 152. Den bedriftsøkonomiske nytte-kostnadsanalysen ga netto nytte på kr 15, forutsatt dagens inntektsrammeregulering. Investeringen vil antagelig finansieres gjennom inntektsrammereguleringen. Avgjørelsen er enda ikke tatt, men uttalelser fra NVE tyder på dette.

5

Stability Investigation of an advanced electrical rail vehicle

Student Supervisor Co-supervisor : Hana Yohannes Assefa : Prof. Molinas, Marta : Danielsen Steinar

Problem Investigation of the effect of non-linearity introduced by a switching model of an advanced electrical rail vehicle on the general performance and for the evaluation of the stability limits associated with low frequency power oscillations. Summary Modern electrical locomotives are equipped with a number of power electronic equipment and advanced digital control technology. This improves the performance of the locomotive, but also introduces a lot of new dynamical phenomena of interaction with the railway power supply that are not yet well understood. For the latest years, the topic of poorly damped low frequency oscillations has been in focus. There are reported cases of these oscillations resulting in instability of the traction power system, but a clear explanation of the phenomenon has not yet been reported. In order to study the interaction between the power electronic component and the rest of the railway traction power system, models closer to reality need to be developed. This thesis is focused on the development of one possible model that includes the switching action of the semiconductor devices of the converter. Figure 1 shows the single line diagram of the developed model.

AC voltage measurement Synchronous rotating controllers

uac1

SOGI

u 1 u 1

dq

ud 1

uq1

PLL uq1

ref iq 0

pll

AC current measurement

iac

SOGI

i i

dq

id iq

ud 1

CC

uq1

i

ref d

ud 1 ref udc 1

VC

idcm _ fil

udc _ fil

pll

comp

Transformer

ref uq

ref ud

+ +

dq

ref uac

LPF

LPF DC

Lt

uac1

Rt

udc

idc

iac uac

1~ PWM =

idcm C2 L2 R2

CPL

udc

Cd

Figure 1: Single line equivalent circuit and block diagrams of the model

6

In general power system modelling includes a number of simplifications in order to reduce the model complexity and the simulation time a typical simplification is to neglect the modelling of the switching action during dynamical stability studies. However for more realistic results, the presence of semiconductor switching which is the main source of harmonics and nonlinearity in the power system should be considered to evaluate possible impacts on stability. In this thesis work Voltage Source Converter (VSC) for traction power system with and without including the detailed PWM switching is modeled using the simulation software tool EMTDC/PSCAD. The effect of semiconductor switching on the low frequency behavior of the dc-link voltage when a 3.67MW step power is added at 60km line is investigated. Furthermore, the performance of a PWM time delay compensation technique is analyzed.

Result and Conclusion

Figure 2: Response in dc-link voltage when a 3.67MW step in motor power is added at 60km line for the two models. The result shown in Figure 2 compares the average model and the switching model of the DClink voltage when a 3.67MW step in motor power is added at 60km line. In both curves it is observed the typical second harmonic ripple twice the line frequency as expected in this case. In addition, in the switching model a ripple due to the semiconductor switching which is twice the switching frequency due to use of unipolar voltage switching is observed.

7

In the low frequency oscillation comparison, Figure 3, shows that the switching model is more damped and has lower overshoot compared to the average model. This result agrees with the stability limit investigated in this Thesis work which gives the switching model a longer stability limit.

3 2.95

Switching Model Av erage Model

DC link voltage [KV]

2.9 2.85 2.8 2.75 2.7 2.65 2.6 2.55 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time [s]

Figure 3: Low frequency oscillation of DC-link voltage compared for the two models. From the presented results, the average model gives less realistic representation of the converter dynamics concerning transient oscillations. Therefore the detailed model with the PWM switching gives an improved representation of the system from the low frequency behavior point of view. Moreover, the delay in DC- link voltage control loop caused by the switching dead time effect was improved by compensation of dead time in the inverse-park transformation block of the control loop.

Conclusion: In order to study the low frequency behavior for stability limit investigation of a traction system, a detailed semiconductor switching (PWM) - model, which takes into account the semiconductor non linearity needs to be modeled and well understood. The same investigation could also be done with the average model by including an equivalent representation of the effect of the switching action that can give the same low frequency response as in the detailed switching model. This will remain as the next subject of investigation for this work.

8

Energy Storage Solutions for Wave Energy in Stand Alone Systems

Student: Supervisor: Emma Barfod Tore Marvin Undeland

Problem description · · · Discuss the need of energy storage for intermittent renewable energy to mate it more effectively utilize Evaluate Stand Alone Systems Chose two case studies that illustrate the need of energy storage for wave energy in Stand Alone Systems

Energy Storage technology is a key to further development of renewable energy sources. The objective of this master thesis is wave energy in a Stand Alone System. Two point absorber systems developed by FOBOX, Fred Olsens wave energy project, are used to illustrate the challenges in the power production from wave energy. There are two main concerns considered in this thesis. The Stand Alone System must be able to deliver power whenever it is needed, not only when there are waves. The second concern is that it can deliver power with sufficient quality. Both concerns can be solved by energy storage systems, respectively long- and short-term. Long term energy storage solutions are for instance pumped hydro, CAES and some types of batteries. Short term energy storage systems could be super capacitors, flywheels, SMES and also some types of batteries. The focus on Stand Alone Systems in an isolated grid is chosen for its` beneficial environmental and economical potential. Alternatives for power supply on islands are uneconomical, such as laying cables out to an island. Use of fossil fuel in diesel aggregates is problematic in an environmental concern. In addition islands often have a unique potential for renewable energy with good natural resources of energy and benefits related to promotion of renewable energy worldwide. Both point absorber cases proposed in this thesis are built up respectively by a three phase generator, a rectifier, an energy storage system with appurtenant DC/DC converter and finally an inverter. The converter proposed, made by Guidi Guiseppe, is a half controlled cascade with bidirectional power flow. This converter is very attractive in terms of cost and efficiency. It is proven that the fluctuations from 7 wave energy platforms with a power production on several MW, require very expensive and huge energy storage solutions. Therefore alternatives to smooth the power is required. It can be done by spatial aggregation of the platforms, so the power is smoothed out to an acceptable level for storing. A super capacitor is used to achieve higher power quality in a short term period. Long term variations are taken care of by a NaS battery. The last case discussed is a smaller system of 3 floating wave energy buoys supplying 12 houses on an island. They produce an average power at 20 kW. The short term fluctuations are in this case also handled by super capacitors, but for the synchronization between consumption and production, cheap Lead-Acid batteries are proposed.

9

Stability Studies of an Offshore Wind Farms Cluster Connected with VSC-HVDC Transmission to the NORDEL Grid

Student: Supervisor: Co-supervisor: Raphael Boinne Terje Gjengedal René Feuillet, INPG

Offshore wind power has proven to be a renewable energy source with a high potential, especially in the North Sea, where an important development is going on. The location of the wind farms tends to move far from the coast to benefit stronger and more constant wind. In the same time, the power output of the wind farm is increasing to several hundreds of MW up to 1 GW. In the European liberalized electricity market, the interconnection of the countries become very important to facilitate the cross-border trade of electricity but also to improve the reliability of the grid. Combining this both aspects into one, a big offshore HVDC grid connecting countries and large wind farms spread all over the North Sea is currently being studied and developed. So in addition of the challenge given by a high penetration of the wind power production in the European power production scheme, new challenges are opened especially for the offshore transmission. This master thesis presents first an overview of the wind power state in the North Sea and in Norway: technologies, challenges, wind resources and future. Then, the existing HVDC technologies are presented and its use for offshore wind power is explained. The choice of the Point of Common connection is briefly explained. Finally, some stability studies are performed for a single 1 GW or a cluster of wind farms connected to an oil rig with different connection scheme based on HVDC transmission using the Voltage Source Converter (VSC) technology. The connection of the offshore wind farms is done either with a single HVDC transmission or two HVDC transmissions connected to the main grid at two different Points of Common Connection situated in the south-west of Norway. The wind farms are not represented in detail but by a single generator. They are equipped for the simulation with Double Fed Induction Generators (DFIG) to be representative of the reality, almost half of the wind turbines are today equipped with DFIG technology. Two disturbances are used to test the electrical stability of the system: a classical 150ms three fault phase in agreement with the grid code requirements on the ride fault through requirements and 100ms fault leading to the tripping of a line. The impact of using different types of generator is also investigated with the simulation of cluster wind farm where a wind farm is equipped with Fixed Speed Generator (FIG). The emphasis is put on the response of the VSCconverter and to a lesser extent on the behaviour of the wind turbine generator. It is demonstrated the capacity of the VSC-converter to stabilize a small grid alone and to "isolate" a disturbance. The voltage and the frequency offshore are practically unaffected by a fault onshore and vice versa. As expected, it is demonstrated that the multiplication of the VSC-HVDC converter in a grid improves the stability of the system. Finally, it has been noticed that there maybe some interactions if several different types of generators are used. The replacement of a generator by another type inside the wind farm cluster may change completely the dynamic behaviour after a disturbance. Simulations are performed with PSS/E.

10

Example of studied case with a cluster of offshore wind farms and oil rig connected with a single VSCHVDC transmission to the main grid

11

Fixed Speed Electric Motor Drives for LNG Refreigeration Compressers Back-to-Back Starting Methods and Grid Consequences

Student: Supervisor:

Hallvard Breistein Tom F. Nestli

Problem description 1. Perform a literature study into the subject 2. Investigate the feasibility of low-, partial frequency and soft starting of generator and compressor drive system connected Back-to-Back 3. Perform laboratory tests of a scaled version of at least one of the aforementioned starting schemes 4. Validate simulation model for Back-to-Back start-up using parameters from laboratory, and study the physical effects that could prevent starting 5. Perform simulations on full scale model using actual parameters 6. Investigate the consequences of fixed speed drives in LNG plant grid 7. Conclude on the technical feasibility and advantages/drawbacks of fixed speed LNG

12

Abstract

Experimental studies as well as simulations have been performed on the Back-to-Back starting schemes low frequency-, partial frequency-, and soft -start-up. A Back-to-Back configuration of two synchronous machines has been established in the laboratory, upon which parameter estimation and start-up experiments have been performed. Extensive parameter estimation was conducted in order to, as accurately as practically possible, replicate the laboratory machines in a simulation model that was constructed in the pre-project. This was done in order to verify the validity of the simulation model. Studies into the effects of inductance interconnecting the machines were made in the laboratory and in the simulation model. Effects of resistance and inertia were studied in the simulation model. It is concluded that the simulation model appears to be as reliable as is its input parameters. Discrepancies were found in line voltages, due to weaknesses in the implementation of field current replication. Full scale simulations using ABB Motorformer parameters obtained from StatoilHydro were performed in the simulation model, featuring low frequency- and soft -staring. The effects of an interconnecting cable were studied. It is concluded that low frequency starting appears to be most reliable and least violent starting method. However, it might be limited by the availability of a turbine. This is not the case for soft starting, which has a lower starting capability and is more violent to the motor damper- and field windings. Low frequency startup is hence the recommended starting method of the ones studied. Dynamic short circuit simulations were done on a fixed speed LNG-facility. The fixed speed alternative appears to be more stable when responding to a short circuit. This is because the motors contribute to upholding the voltage during a fault by delivering reactive power to the short circuit, and because the motors do not loose all torque as is the case for LCI drives when the voltage dip exceeds 20%. Further work is needed in up-scaling the experiments. A sophisticated simulation model should be established and its validity tested on the up-scaled experiments. Preliminary custom design of machines should be initiated depending on what starting scheme is chosen. Custom machine parameters should then be used in full scale simulation using the more sophisticated model.

ii

13

Economic Benefit of New Capacity in the Central Grid

Student: Ingar Dalen Supervisor: Ivar Wangensteen Contact: Anders Kringstad, Knut Hornnes, Arild Helseth Collaboration with: Powel, Statnett, Sintef Norway and the EU have in recent years established ambitious goals to increase the share of renewable energy in their consumption. On account of these goals, a large-scale wind power development can be expected in northern Norway and Sweden. This development may be financed both by Norway and by countries with less wind resources in order to meet the energy goals imposed upon them. An increased power surplus is dependent on TSOs' abilities to transmit increased amounts of power through the Nordic grid. A scenario of likely power market conditions in year 2025 is used as a basis. The scenario has a high expectancy of new wind power as well as strong grid investments compared to the level in 2009. This thesis assumes an additional increase in annual renewable power production of 22 TWh, divided into 16 TWh in northern Norway and 6 TWh in northern Sweden. Results show that this amount of new power cannot be implemented without large grid investments. The Energy and Power Flow model is utlized to simulate the Nordic power flow for different levels of grid investments. Two grid solutions are proposed that allow the production increase while maintaining an acceptable state of system operation. The first uses DC transmission from Rana to Oslo in order to control power flow through Norway. An additional AC line from Kobbelv to Ritsem allows import from Sweden to the DC line. The second grid solution uses AC line upgrades throughout Norway ensuring two 420 kV lines from Ofoten to Kristiansand. Due to lower impedances in the Swedish grid, a large amount of the Norwegian production flows into and through Sweden. This solution requires a new line from Kobbelv to Ritsem and Rätan to Borgvik in order to solve resulting Swedish transmission congestion. Both grid solutions require a new DC cable from southern Norway to Germany in order to export most of the new power production. These cables require a number of supporting line upgrades in the region. Power producers schedule according to the new market situation, allowing a very high export during daytime and a low export during night. The increased power production in northern Norway and Sweden replaces other production. A high amount of gas and coal power is replaced in continental Europe. No hydropower, wind power or nuclear power is replaced. The DC and AC grid solutions allow European reductions corresponding to 19,3 % and 16,6 %, respectively, of the expected Norwegian CO2 ­ emissions in year 2025. The cost of each grid solution is calculated to 22 760 MNOK and 19 310 MNOK. Annual system increases in valued socio-economic benefit outweigh the grid investment costs of each option by 3 300 MNOK and 3 370 MNOK per year of the period of analysis. The total cost of new power production must not exceed these values for such a decision to be socioeconomically beneficial. Due to the high increases in calculated socio-economic benefit, a recommendation for further analysis is made.

14

Wave Energy Conversion Simulation Verification and Linearization of Direct Drive Wave Energy Converter with Variable DC-link Voltage Control

Student: Arne Marius Ditlefsen Supervisor: Tom F. Nestli Co-supervisor: Jonas Bakken, Fred Olsen Ltd. The World first commercial wave farm was launched in Portugal in 2009. In other parts of the world WEC (Wave Energy Converter) devices are at R & D and prototype stages, many with promising prospective. All electric power take off systems are in focus and several possible solutions can be found in the literature. The task of this project will be to investigate one such solution. A Permanent Magnet Synchronous Generator feeding into a 6-pulse diode rectifier with variable DC link voltage. The task is divided in three parts. 1. Implement a model of the WEC system in Matlab and develop and demonstrate stable control that emulates passive loading of the generator 2. Verify simulation results by laboratory experiments 3. Make a linearization of the system that can easily be adapted to any scale and rating Lowering the cost of wave energy conversion is an essential task for it to succeed as a future energy resource. In this work a converter, assumed cheaper than the regular back to back converter setting, have been investigated for a electric direct drive point absorber. Both experimental work and simulations are used in the analysis. In the experimental work, a permanent magnet generator with a 6-pulse diode rectifier, a DClink and a DC/DC converter equivalent, was used. Steady state, dynamic and transient measurements were preformed and a simulation model was compared to the measurements. Good results were obtained and deviations were in general small, mostly ± 3% for voltage and current measurements and ± 8 % for torque measurements. Based on transient measurements and simulations a general linearization of the system was made in order to obtain useful information about the system. A step up converter was used in the simulation and it demonstrated stable passive loading control. By using the information obtained by the linearization, the performance of the simulation model was improved by decreasing the DC-link capacitance. The modified simulation model had significant less torque ripple than the initial. The linearization model also can been used to identify time delay represented by the power take off unit in a wave energy converter. This will be done for a commercial size wave energy converter summer 2009.

15

Reliability Assessment of Distribution System -Including a case study on Wangdue Distribution system in Bhutan.

Student : Tempa Dorji

Supervisor : Sand Kjell Problem description Bhutan Power Corporation Ltd's mission is to transmit, distribute and supply adequate electricity in a safe, reliable and efficient manner and this has to be accomplished. The main problem facing by electric power utilities in developing countries today is that the power demand is increasingly rapidly where supply growth is constrained by scarce resources, environmental problems and other societal concerns. This has resulted in a need for more extensive justifications of the new system facilities, and improvements in production and use of electricity. System planning and operation based on reliability cost/worth evaluation approach provides an opportunity to justify one of the scrutinized and vulnerable economic sectors in Bhutan. It is with this objective to conduct customer surveys to find out the outage cost of interruptions. The customer interruption in Bhutan is as high as 20-30hours per year. With the vision of electricity to all within 2017, the interruptions per customer may further deteriorate due to rapid expansion of the distribution systems. As of now, no sufficient technical research have been carried out in the distribution network, it may be due to lack of technical expertise in the Utility. Most of the interruption has been caused due to the failure in the distribution systems in Bhutan. Hence it is felt necessary to improve the reliability of the system in order to improve the utility's performance and to keep our valued customers satisfied. Intelligent placement of protection devices, sectionalizers and switches in the distribution feeders has significant impact in reliability improvement and this will be further evaluated along with other outage mitigation techniques for the distribution system in Bhutan. Task · Collect, compile and review of fault statistics which shall be used in reliability assessments. · Literature review on basic theory of reliability assessments and customer cost of interruptions. · Frame sample questionnaires which could be used in Bhutan to assess the cost of interruption (reliability worth) from customer's perspective. · Carry out reliability assessments of the existing system and predictive reliability analysis for the future system with the help of simulation software tool called NetBas/Levsik. Number of alternatives like placement of LBS, Auto reclosers shall be evaluated along with the mitigation techniques and the solution for the system shall be proposed. An alternative such as automation is not economically viable in the case study area due to low energy consumption on the feeders. · Find out the major source of causes of fault in the case study area and suggest mitigation techniques. · Review theory on outage mitigation techniques, i.e Electric and Non electric mitigation techniques.

16

Model/ measurements To evaluate cost benefit of reinforcements and measure to improve reliability, two additional indices are being used; the expected annual non-delivered power and energy (NDP and NDE.) The most common evaluation techniques, using a set of approximate equations, are failure mode analysis or minimum cut set analysis. To deal with the general distribution system, the Norwegian Electric Power Research Institute (EFI), Trondheim, Norway has developed a model called Reliability evaluation of Radial distribution network (RELRAD). The 33kV and 11kV system is simulated with the help of software tool called NetBas/Levsik which is modeled based on RELRAD model. Start Select network component Calculate contribution to reliability indices Accumulate indices to affected load More components Yes

?

No

Result presentation

Stop

Fig.: Flow chart for calculation of reliability indices The expected values of reliability indices like; Non Delivered Energy, Non delivered power and the cost of NDE is generated based upon the input data like failure rate, repair times and sectioning time of all the components. The algorithm then accumulates reliability indices for each load point from each component giving outages to the load point. Finally, when all the fault contribution is found, the total accumulated indices are available. The figure shows a flow chart of the algorithm. Results The reliability of present systems could be further improved by installation of LBS at different locations for both 33 and 11 kV system. The consequences if there is outage could be reduced by taking corrective measures. For predictive reliability analysis for 33kV, the number of alternatives are tried, it is found that for year 2012, Case 2 (Assume the fault rate could be reduced due to timely trimming of trees) gives minimum break even cost of Nu. 2,04/kWh, while case

17

4(Installation of ARCB + Case 2) gives minimum SAIFI and SAIDI with break even cost of Nu10,10/kWh meaning that any cost of Non Delivered Energy higher than Nu.10,10/kWh will be cost effective. From 2017, Case 3( ARCB at WACHE and LBS at DPSA-RIDA, DDSA-GAN, ZAM-DLO and reduce fault rate from 20-16) gives a break even cost of Nu.6,697/kWh where as Case 4 (Case 3+ connection of Trongsa line as reserve) gives minimum SAIDI among alternatives and break even cost of Nu.4,518/kWh. For 11kV line fed from Lobesa, predictive analysis for year 2012, shows that Case 2 gives minimum break even cost of Nu.26,96/kWh where the rest of the case gives break even cost more than Nu.44,27/kWh. Case 4 gives less SAIFI and SAIDI then the rest of the case. In this context, Case 2 is preferred over case 3&4 due to large difference in break even cost and small difference in value of SAIFI & SAIDI. As for 2017, the minimum break even cost if Nu. 22,17/kWh in case 2, where maximum break even cost is Nu. 66,48/kWh in case of case 4. The high break even cost indicates that the system is already in optimized condition and need high investment to have further improvement. Therefore, customer interruption cost is required to assess reliability worth of having the reliability improvement project. Recommendations The present data recording system should be revolutionized from manual to computer aided system. All the events should be specific and the step restorations made should be recorded accordingly so that true reliability indices are obtained. The failure of individual components in the system should be maintained so the probability of failure represents its true system. Its repair time and sectionalizing time should be separated since it has high affect on the reliability indices during predictive analysis. Reliability of 33kV system could be further improved by installing LBS at node at KUMCHE-CGP, DKHA-LKHA and shifting of LBS from LGKHA-SHA to SHARKH (Alt. 3) which gives minimum interrupted energy and the lowest SAIDI. For year 2012, Case 2 (Assume the fault rate could be reduced due to timely trimming of trees) which gives minimum break even cost of Nu. 2,04/kWh is considered while Case 4 (Case 3 + connection of Trongsa line as reserve) for 2017, which gives minimum SAIDI among alternatives with break even cost of Nu.4,518/kWh is being recommended. For 11kV system fed from Lobesa substation, it should run with the alternative 3 of year 2009, since the break even cost for year 2012 and 2017 is quite high. Further analysis has to be carried out to find the reliability cost and worth from customer's perspective (interruption cost) and compare with the break even cost to find the worth of having reliability improvement projects. BPC should further explore on this studies to find out the reliability worth assessments based upon customer's perspective. The sample questionnaires developed could be further improved by doing detail study after visiting customer's premises and learning about their activities and their use of electricity.

18

Power Supply for Down-hole Instrumentation and Actuators

Student: Supervisor:

Christian Eidsaune Lars Einar Norum

Problem description Wireless communication with associated down-hole power generator, has long been a missing link for the ultimate smart well system. In this project we will address the technical challenges that must be met when the well does not have cables for power supply. In the proposed project we will focus on local electrical power generation and storage for a down-hole instrumentation unit. The main objective of the project is to establish or develop reliable electrical power for a down-hole instrumentation unit, and develop the electronic for back up battery charging and voltage control. A generator based on the induction principle, converting energy from the kinetic energy in the well stream to electric power is developed and patented. The following main tasks will be initiated in the proposed project: 1. Design of power converters for voltage control and energy management system 2. Testing of the laboratory prototype electrical generator and high temperature electronics for storage and voltage control The project is initiated by industrial partners and a research project at SINTEF.

19

SUMMARY

temperatures is to low for a down-hole instrumentation unit. An alternative is to use s

To create the ultimate wireless instrumentation unit for down-hole applications high

Power Supply for Down-hole Instrumentation and Actuators

NTNU III

Silicon on Insulator process under the fabrication of the semiconductors. The SOI

temperature electronics with very high reliability is needed. It is possible to use ordinary bulk-CMOS devices at temperature up to 175 C, but the lifetime at these

process is a fabrication process where there is buried a oxide layer in the silicon wafer, and thus allowing higher breakdown voltage and/or lower current leakage. The low 5 years at 225 C and thus much lower at junction temperatures below 200 C. current leakage allows the semiconductors to be used at higher junction temperature.

SOI devices that are commercial available off-the-shelf as a expected lifetime for at least devices. A package like this gives a system with high reliability both toward high capacitors is 200 C. substrate as a replacement for organic PCB and thick-film technology for the passive

temperature operation and lifetime. The main limitation in the high temperature design is the availability off the larger capacitors; the limitation for high temperature stacked supplies. The converters are designed with current mode control; current mode control constant current charging for the battery. The converters designed are the standard step-up and step-down switch-mode power

The SOI technology can then be used together with hybrid circuits using ceramic

is used because of the advantage that comes with it. One off the advantages is the When designing the SOI devices for high temperature operation it is difficult to achieve high enough breakdown voltage. With this in mind, the high temperature converter is

possibility to limit the inductor current; another advantage is the possibility to use designed with series coupled transistors to achieve high enough breakdown voltage for perturbations in for example turn-off speed makes an uneven voltage sharing; this is

high voltage operation. The transistors have always some small perturbations in their voltage sharing.

specifications, this has to be considered when connecting transistors in series. This solved by connecting suitable capacitors in parallel with the switches to maintain even

20

Impact on Wind Turbine Systems from Transient Fluctuations in Offshore Utility Grids

Student: Torger Einervoll Supervisor: Lars Einar Norum Co-supervisor: Bjarne Idsøe Næss, Unitech Power Systems AS

Problem description Electrical utility grids at offshore installations in the North Sea are mostly powered by the use of local generators driven by gas turbines. These gas turbines contribute to a significant amount of the total C02 emissions in Norway. A method for reducing these emissions is to supply electrical power to the oil installations from offshore wind turbines which is located near the installations. This is an installation from offshore wind turbines which is located near the installations. This is an interesting solution for existing oil installation where the accessible gas turbines can be operated together with the wind turbines and also function as back-up system due to the variable wind. The Transient fluctuations in the electrical grid on a typical offshore installation differ from the onshore utility grid. One of the main characteristic of an electrical grid at an offshore installation is the start of large squirrel cage motors on a relatively small generator capacity. In addition transients due to short circuits and operation of the circuit breakers will occur. It is thus of interest to study the impact of such fluctuations on different types of wind turbine technologies by the use of a transient simulation tool like EMTDC/PSCAD.

21

Impact on Wind Turbine Systems from Transient Fluctuations in Offshore Utility Grids

Summary

None of the modelled generators experienced problems with the disturbances caused by the electromechanical transient fluctuations. Based on the behaviour of the DFIG's grid side converter, it is believed that the result would be the same for a wind turbine with full frequency conversion. Variable speed wind turbines are expected to remain controllable throughout electromechanical transient fluctuations such as for the simulated case. However, the controllers, converters and equipment have to be designed while bearing these fluctuations in mind. The controllability of the variable speed wind turbines could be used to contribute to voltage control by production and consumption of reactive power. A controller scheme with the purpose of stabilising the voltage at the gas turbine generators' terminal was developed, but had low impacts on the power system behaviour.

In this thesis work, the behaviours of different wind turbine technologies during transient fluctuations in an offshore utility grid have been studied. For this purpose, a dynamic model for an offshore oil platform was developed. Models of squirrel cage and doubly fed induction generator based turbines were developed as well.

Gas turbines in offshore power systems contribute to about 23% of Norway's total emissions of CO2. One method for reducing these emissions could be the addition of wind turbines to the offshore utility grids. Power from shore is another alternative, but has been proven costly due to long cables and expensive HVDC converter stations.

A stator flux feed forward term for the speed controller was developed. The term stabilised the power output of the doubly fed induction generator. However, the impact on the power system's frequency response was minimal, and there is probably no material value of such an addition to the control loop.

ii

22

Optimal produksjon av reaktiv effekt i regionalnettet i Telemark & Vestfold

Student: Faglærer: Martin Eldrup Olav B. Fosso

Skagerak Energi AS eier og drifter mesteparten av regionalnettet og kraftverkene i Telemark og Vestfold. Med oppgaven ønsker Skagerak Energi AS å bedre samdriften av kraftverkene og regionalnettet i området. Arbeidet skulle resultere i en oversikt over optimal produksjon av reaktiv effekt i kraftverkene i tillegg til en oversikt over spenningsprofilen i nettet. Optimaliseringsverktøyet som ble brukt i denne oppgaven var MATPower som inneholder forskjellige algoritmer for løsning av generelle lastflyter og optimale lastflyter. Året ble delt i to perioder, hvor tunglastperioden utgjør tidsrommet fra og med desember til og med februar, mens resten av året ble modellert med sommerbelastning. Optimaliseringen tar utgangspunkt i en minimalisering av taps- og produksjonskostnader, men i tunglastperioden ble det i tillegg inkludert kostnader forbundet med uttak av reaktiv effekt fra sentralnettet. Sammenligningsgrunnlaget er en optimalisering uten uttak av reaktiv effekt fra kraftverkene noe som er i nærheten av dagens drift. De andre optimaliseringene ble sammenlignet med denne og viste at en optimalisering av reaktiv uttak fra kraftverkene ga en stor kostnadsbesparelse. Kostnaden ved uttak av reaktiv effekt fra sentralnettet i den som reduseres mest ved optimalisering av reaktiv effektuttak ved kraftverkene. Kostnadene ble redusert ved optimalisering både med og uten kostnaden for uttak av reaktiv effekt fra sentralnettet. Tapsog produksjonskostnadene ble også redusert i forhold til sammenligningsgrunnlaget. Sammenligningsgrunnlaget ga de høyeste kostnadene uansett belastning og produksjonsnivå. I oppgaven er det gjort en sammenligning av kostnadene ved tre forskjellige optimaliseringer hvor O1 viser resultatene av optimaliseringen når det ikke er uttak av reaktiv effekt fra kraftverkene, mens O2 viser resultatene når sentralnettkostnaden ikke er tatt med i optimaliseringen. O3 viser resultatene av optimaliseringen som tar hensyn til sentralnettkostnaden i tillegg til tapskostnader i nett og kraftverk. Optimaliseringene er gjort for tre forskjellige produksjonssituasjoner i tunglast og for fire forskjellige produksjonssituasjoner ved sommerbelastning. 100% vil si at alle kraftverkene kjører med maksimal effektproduksjon, mens kraftverkene med 75% produksjonsnivå kjøres med en belastning tilsvarende 75% av installert ytelse. Det samme gjelder for 50% og 25%. Figuren under viser en sammenligning av de totale timekostnadene når det blir avregnet for reaktiv effekt ved gitt produksjonsnivå. Det blir kun avregnet for reaktivt uttak fra sentralnettet i tunglastperioden. O1 vil i alle tilfeller gi den høyeste timekostnaden, men det er mindre forskjell mellom O2 og O3. O3 gir de laveste kostnadene.

23

Kostnadene som er presentert er kun ment for å sammenligne resultatene fra de forskjellige resultatene og er ikke de virkelige driftskostnadene som vil inkludere andre kostnader i tillegg til sentralnettskostnaden og tapskostnader i nett og kraftverk.

Figuren under viser uttak av reaktiv effekt fra sentralnettet i tunglastperioden. Ved en optimalisering av reaktiv effektproduksjon i kraftverkene i regionalnettet i Telemark og Vestfold, vil uttaket reduseres betydelig fra sammenligningsgrunnlaget, O1.

24

Testing of superconducting coils

Student: Eivind Engebrethsen Supervisors: Magne Runde, Niklas Magnusson and Frode Sætre Collaboration with: SINTEF Energy Reasearch AS Problem description SINTEF Energy Research is building two superconducting coils for use in an aluminium billet induction heater. The coils will be made by joining together 32 sub-coils, each wound from 550 m of MgB2 superconducting tape. The complete coil assembly will be installed in a cryostate and operate at a temperature of around 20 K. The master thesis project is largely experimental in approach and is essentially to assemble the sub-coils into two complete coils, cool the coils down to operating temperature and test them. This includes moulding together the sub-coils, splicing the superconducting tapes, designing and building a system for monitoring voltage drops, temperature distribution and other parameters that show how the coils are operating. Crucial coil and cryostat properties such as critical current at different temperatures cool down profile and time, etc. should be measured. Model/ measurements One coil was put together by splicing together the sub-coils and installing it in the cryostat. 10 thermocouples and one RTD element were distributed on the coil at crucial positions in the cryostat. A system for monitoring the voltage over the joints and over the sub-coils was made.

Figure 1: One sub-coil

Figure 2 The complete coil (diameter is 1,1m) Figure 3: Installation of the coil in the cryostat

Result and Conclusion The coil was put together, cooled below the critical temperature for the superconductor and measurements were performed. It took the coil about 120 hours to reach a stable temperature of about 7K. Measurements showed that all the joints had a resistance below 60n. The inductance of the coil was measured to be 4.9H. One sub-coil showed a resistance of 66u at a current of 9.6A. This indicated damage to the sub-coil, which together with a malfunctioning current source limited the amount of tests that could be performed. The temperature profile revealed that the different parts of the system cooled down at different rates. When passing a current of 5.34A the field generated by the coil was measured to be 0.32mT at a distance of 2 meters from the center of the coil.

25

Shell Eco Marathon Electric Drive for World's Most Fuel Efficient Car

Student: Supervisor:

Rolv Marius Faleide Robert Nilssen

Problem description A motor for the Shell Eco Marathon car must be redesigned to match the onboard voltage. Also consider improvements in the design, such as individually controlling each winding.

26

Grid integration of renewable energy from hydro power.

Student: Espen Feilberg Supervisor: Prof. Tore Undeland, Dr.Ing. Sigurd Øvrebø, Dr. Ing Richard Lund Contact: [email protected] Collaboration with: SmartMotor AS Problem Description: The project treats the patenting and testing process of a novel generator with converter concept. This generator is named "Doubly-Fed Permanent Magnet Synchronous Machine", DF-PMSM, and is based on connecting two sets of three phases from the generator to increase controllability of the Permanent Magnet Synchronous Generator. The first winding set is named "control winding" and is routed through an active-front-end converter. It will be used to control the power angle output from the machine. The other winding set is named "power winding" and will carry most of the load, directly coupled and in sync with the voltages of the connected grid.

The Task: · · Patenting the DF-PMSM machine including search for prior patents as well as papers on the machine and similar designs. State-of-the-art study that includes doubly fed machines and regular fed machines to build knowledge about multiphase and especially doubly fed machines and the other alternatives in hydro power generation today. Design a machine/converter setup based on the grid regulations in Norway today Case study comparing this generator to the already existing alternatives. Development of machine models and tools for simulation in LTspice. Simulations of the DF-PMSM in different operations. Development of a lab with SmartMotor machines and industry standard converters. Laboratory exercise with the DF-PMSM in a back-to-back configuration.

· · · · · ·

27

Opprustning av Svorkmo kraftverk

Student: Veileder: Kontaktperson: Samarbeid med: Idunn Gangaune Finnanger Prof. Gerard Doorman Siv.ing Lars Olav Hoset TrønderEnergi AS

Problembeskrivelse: TrønderEnergi ønsker å undersøke om innsetting av et tredje aggregat i Svorkmo kraftverk vil gi økt produksjonsgevinst, bedret totalvirkningsgrad ved lave vannføringer og mer fleksibel drift av vassdraget. Svorkmo kraftverk har i perioder om sommeren for lav vannføring til å drifte de to eksisterende aggregatene på en optimal måte. Produksjon på lav virkningsgrad og periodevis behov for ekstra tapping av magasinvann om sommeren, gir tapte inntekter. Hypotesen er at et nytt mindre aggregat i Svorkmo kraftverk vil bedre denne situasjonen. Oppgaven: Den matematiske beskrivelsen av Svorkmo kraftverk kompliseres av store falltap og bekkeinntak som påvirker falltapene i stor grad. God modellering av virkningsgradsbeskrivelsen i dette kraftverket har vært helt essensielt, spesielt modellering av bekkeinntak. Vansimtap er brukt til simulering av forventet drift av vassdraget både med og uten et tredje aggregat i Svorkmo. Liten mulighet for detaljert beskrivelse av totalvirkningsgraden i Svorkmo kraftverk i Vansimtap, gav et behov for verifisering av resultater i SHOP hvor detaljgraden er vesentlig høyere. Metode: Inndata til Vansimtap er produksjonskurver som beskriver sammenhengen mellom vannføring og produsert effekt. Kurvene ble beregnet i et selvkomponert Matlabscript som inkluderer virkningen fra falltapene og bekkeinntakenes innvirkning på falltapene. 70 tilsigsår simuleres i Vansimtap for systembeskrivelsen både med og uten et tredje aggregat i Svorkmo kraftverk. Fem produksjonssituasjoner optimaliseres i SHOP, med økt detaljeringsgrad, for å verifisere modellbeskrivelsen til Vansimtap og om resultatene fra Vansimtap kan brukes til videre analyse.

Matlabscript beregner PQkurver basert på optimal lastfordeling

Vansimtap simulerer 70 tilsigsår

Exel-makro henter ut resultater fra Vansimtap og vises i "driftsimIdunn.xls"

Optimaliserer 5 situasjoner i SHOP basert på resultatdata fra Vansimtap

28

Resultater:

Årlig produksjon per modul [GWh] Litjfossen Grana Ulset Brattset Svorkmo Total produksjon for hele vassdraget [GWh] Årlig vanntap for hele vassdraget [GWh] Forventet årlig inntekt for hele vassdraget [M Euro]

Referansemodell 163,725 287,532 137,576 407,777 266,107 1262,717 165,121 48,873

Utbyggingsmodell 163,679 287,428 137,539 407,780 270,613 1267,039 157,007 49,037

Differanse -0,046 -0,104 -0,037 0,003 4,506 4,322 -8,114 0,164

Simuleringene i Vansimtap viste at en innsetting av et tredje aggregat i Svorkmo kraftverk gir en produksjonsgevinst på 4,322 GWh. En gjennomsnittlig markedspris på 39 Euro/MWh, gir en økning i produksjonsverdien på 0,164 M Euro. Det totale magasinnivået, minker ved innstetting av aggregat 3 pga mulighet for økt tapping av magasinvann med mindre risiko for overløp. Produksjonsgevinsten kommer i hovedsak fra økt kapasitet i Svorkmo kraftverk, og dermed bedre utnyttelse av høyt tilsig. På sommervannføring vil et tredje aggregat bedre totalvirkningsgraden til kraftverket. Verifisering av fem produksjonssituasjoner i SHOP viste at Vansimtap og SHOP får omtrent lik effekt av et tredje aggregat. Konklusjon: Modellbeskrivelsen i Vansimtap kan antas god nok til bruk i videre analyser, men usikkerhet i beregningsresultater og verifisering av flere situasjoner må vurderes før resultatene tas i bruk. Lønnsomheten i prosjektet med å sette inn et tredje aggregat i Svorkmo kraftverk, er avhengig av fremtidige priser på kraft og kostnadene relatert til utbyggingen. En følsomhetsanalyse bør gjennomføres for å se hvor stor risiko det er knyttet til avvikende verdi for produksjonsgevinst. Dette kan avgjøre hvor stor betydning usikkerhetene i beregningene vil ha for avgjørelsen om prosjektet settes til live eller ikke.

29

Large scale integration of wind power in a transmission grid

Student: Supervisor: Hans Wigen Finstad Olav Bjarte Fosso

Nord-Trøndelag Elektrisitetsverk is planning to build a 250 MW wind farm at Ytre Vikna in the middle parts of Norway. As the introduction of large amounts of wind power in this relatively weak grid makes it necessary to upgrade parts of the grid surrounding, this thesis focuses on identifying the upgrades necessary and to give insight into the problems that may occur. In this report, only stationary analyses have been performed. The results and analyses performed are based on the model as described in the system description. Assumptions made and deviations from the "real life" grid may have impacts on the results, but the main observations in this report give indications on the problem areas which may need to be addressed. First, the loading values for branches in the 66 kV and 132 kV parts of the grid are investigated. By varying the production level at Ytre Vikna from 0 to 250 MW, key upgrades necessary has been identified. Secondly, the suggested upgrades have been modelled and investigated with regards to overloading. Bus voltages have also been monitored in order to search for indices of nearby voltage instability. Results from these analyses show that the branches between ABELVÆR ­ JØA-T ­ DALTRØA has to be upgraded in order to be able to have an acceptable loading situation in the NTE 66 kV grid. When producing 250 MW at Ytre Vikna, the transformer in Kolsvik has to be replaced in order to avoid overloading, when all the power from Ytre Vikna is fed towards Kolsvik. When investigating the n-1 criterion it was found that when one of the branches between RV130 and ARSANDY1, all the production from Ytre Vikna is fed into the NTE 66 kV grid. The maximum amount of power that can be fed into the 66 kV grid in RØRVIK was found to be 62.9 MW. When the branch from ARSANDY1 to KOLSVIK1 was set to be out of service, the wind generation at Ytre Vikna fed some power towards the load in ARSANDOY in addition to feeding into the NTE 66 kV grid. In this case, the maximum production at Ytre Vikna was 94 MW. In this case 62.9 MW was fed into the NTE 66 kV grid, while the rest was fed towards the load in ARSANDOY. In the case of 250 MW at Ytre Vikna and the transformer between RV130 and RØRVIK out of service, some kind of reactive power compensation is needed. Power flow analyses show that the buses ARSANDY1 and KOLSVIK1 have voltages below the specified minimum limit of 0.955 p.u.

30

Configuration of offshore wind farms Student: Supervisor: Collaboration with: Randi Aardal Flo Terje Gjengedal Statkraft

Problem description The European Commission proposal for 20% renewable energy by 2020 paves the way for a massive expansion of wind energy and a new energy future for Europe. To reach the goal wind energy is a key technology and large scale integration is required both onshore and offshore. This represents heavy challenges to the power system requiring new ways of designing and operating the system. Especially large scale offshore wind power will require attention to new focus areas. The wind may be more stable offshore, but there will be less geographical smoothing effect, so wind variations will still be a key issue. Power transmission and grid connection represent other main challenges for realization of large scale wind power, and especially for offshore wind farms. The scope of the project is defined as how to find an efficient and secure design and operation of the overall system. The main tasks in this thesis is to study different possible configurations of a 1000 MW offshore wind farm, located 75 km from shore, and select two configurations for further study. PSS/E will be used to establish a load flow and a dynamical model of each system. Summary This master thesis is written at the Department of Electric Power Engineering at the Norwegian University of Science and Technology. The work has been carried out at NTNU in Trondheim. The thesis deals with configuration of large offshore wind farms and transmission systems, and is a continuation of the project written during the autumn 2008. Today several plans on 1000 MW offshore wind farms exists. The size of the wind farms has led to a challenge of how to find an efficient and secure design of the overall system. The system has to be cost-effective in order to compete with other forms of power generation. In this study, costs are not considered. The purpose of this thesis was to study different transmission systems and configuration of an 1000 MW wind farm located 75 km from shore. The optimal distance between the turbines is a compromise between wake effects, wind farm are and cable lengths. To perform a detailed study of wake effects and optimal spacing, computer programs like WindSim would be necessary. Three common wind farm configurations is radial, star and ring layout. The selection of layout depends on costs, wind data and the wind farm area. Various wind turbine systems have been developed and different wind generators have been built. According to the survey of different wind generator system and considering the grid connection requirements on wind turbines, the developing trends of wind turbine generator systems shows that variable speed is very attractive and concepts with full-scale power converters will become more attractive. In this thesis two wind farm configurations with different transmission system were further studied. AC/AC, AC/DC and DC/DC are possible transmission systems. In this thesis AC/AC

31

and AC/DC were compared. The selected layout of the wind farm was the radial layout. Number of strings was 35, with eight turbines in each string. Each wind turbine could produce 3.6 MW, which gives a total generation of 1008 MW. The two configurations were modeled in PSS/E. Siemens has made a model called WT3 that was developed to simulate performance of a wind turbine employing a doubly fed induction generator (DFIG). The model was developed in close cooperation with the GE Energy modeling team. This model was used in this thesis. For the dc transmission the HVDC Light from ABB was used. Two different disturbances were applied. One at the connection point at shore, and one at the connection point for all the radials. The load flow results shows that the losses are 5.8$\%$ higher in the AC/DC system. The dynamical result shows that both of the systems were stable, and fulfill the grid code requirements. The results indicates that the short-circuit MVA is higher in the ac system than in the dc system. After a fault the voltage recovery was smoother in the dc system, and the voltage recovery time were shorter.

32

Simulering av spenningskvalitet i kraftnett

Student: Kjell Erik Fossen Veileder: Kjell Sand Utføres i samarbeid med: SINTEF Energiforskning AS Spenningskvalitet i kraftnett er et område som har fått økende oppmerksomhet de siste tiår. Dette skyldes bl.a. økte spenningsforstyrrelser i nettet som følge av øket innslag av kraftelektronikk i elektriske apparater og prosesser samt redusert immunitet overfor forstyrrelser i apparater pga øket bruk av elektronikk. For å kunne planlegge spenningskvalitet i kraftnett er det behov for simuleringsverktøy. Slike verktøy vil kunne gi verdifull beslutningsstøtte i planlegging og prosjektering av nett og installasjoner samt være til nytte ved problemløsning. Ved å kunne simulere ulike alternativ vil beslutningstakere kunne vurdere hvilket som tilfredsstiller best de krav som stilles (forskrifter, normer osv). Simuleringer vil også kunne gi raskere svar på forekomsten av feil med en lav hyppighet, sammenlignet med målinger ute i nettet. For å få en høy nøyaktighet må det kanskje måles over flere år. Ved simuleringer er det mulig å få svar innen minutter. For finne ut hvordan situasjonen er i dag med tanke på simulering av leveringskvaliteten, ble det sendt ut en spørreundersøkelse til 123 FASIT ansvarlige i norske nettselskap. Resultatet fra denne undersøkelsen viser at 57 % av de 35 selskapene som svarte på undersøkelsen utførte simuleringer på en eller flere av de fenomenene som er omtalt i FoL. Årsaken til simuleringen er i hovedsak nye tilkoblinger eller ny produksjon. De fenomenene som flest simulerer på er langsomme variasjoner i spenningens RMS verdi, kortvarige over- og underspenninger og spenningssprang. Når det gjelder hvilken type programvare som brukes er det NetBas som er det dominerende programmet, fulgt av PSS/E. NetBas blir brukt i nesten 50 % av alle simuleringer som utføres av selskapene. PSS/E har kun en andel på litt over 10 %. Det er valgt å studere fenomenet spenningsdipp i nærmere detalj. Dette er et fenomen som er årsak til noen av de største kostnadene for et nettselskap etter avbruddskostnader. Det er utviklet et enkelt program i Matlab som gir en stokastisk simulering av spenningsdipp. Programmet baserer seg på Monte Carlo simulering, programmet bruker tre stokastiske variable: feilposisjon, overgangsmotstand og feiltype. Dette gir en fordeling av størrelsen på spenningsdipp ved to samleskinner på en radial. Brukeren taster inn data for linjen og kortsluttningsytelsen til det bakenforliggende nettet. Når det kommer til valg av simuleringsprogram for nettselskap, er det mange faktorer som er viktig å tenke på. Det kan være hvor ofte og omfattende selskapet skal simulere. Små selskaper kan være tjent med å leie inn konsulenttjenester for å gjøre jobben. Siden det vil kreve god kompetanse og stadig oppdatering for å kunne utføre slike simuleringer effektivt. Dette er også tilbakemeldinger som kom i forbindelse med spørreundersøkelsen. Et større selskap bør kunne se på muligheter for integrering mot eksisterende NIS systemer der alle data for nettet ligger lagret. Dette vil gjøre det enklere å få tilgang på inngangsdata. For å studere enkelte transiente fenomen vil det kreves en detaljert elektrisk modell av enkelte komponenter. Dette kan medføre en høy brukerterskel på mange programmer, særlig der modellen må bygges opp fra grunnen av.

33

Framdriftssystem til Eco Marathon kjøretøy

Student: Faglærer:

Anders Gellein Lars Einar Norum

I denne masteroppgaven har det blitt arbeidet med NTNUs deltagelse i Shell Eco-marathon våren 2009. Med utgangspunkt i bilen fra 2008 og simuleringer og analyser fra forprosjektet høsten 2008, har arbeidet vært fokusert på å bygge opp et elektrisk framdriftssystem med hovedfokus på effektivitet.

9-fase motordrift

Det ble først utredet og startet som smått på arbeidet med å bygge en veldig eksperimentell motorstyring for å teste ut et nytt motorkonsept. Etter en stund måtte dette arbeidet avbrytes på grunn av at det trengtes ressurser til å bygge opp et styresystem til en ny brenseslcellestack, som hadde et mye større potensial til å bedre energieffektiviteten i systemet.

Brenselcellestyring

En brenselcellestack som hadde potensial til å bli en del av et ekstremt effektivt brenselcellesystem, var utgangspunktet for dette arbeidet. Det ble bygd opp hydrogenforsyning, viftesystem og elektronisk styring for brenselcellestacken. Systemet ble først testet grundig på lab med blant annet lastsimuleringer, slik at hydrogenforsyning og viftestyring kunne optimeres for bilen. Deretter ble systemet implementert i bilen. Resultatet var et brenselcellestystem med en virkningsgrad på rundt 60%. Med dette som utgangspunkt for resten av framdriftsystemet, ble Shell Eco-marathon 2009 en stor suksess for NTNU.

34

STATCOM with energy storage system for offshore wind integration

Student: Supervisor: Contact: Sverre Skalleberg Gjerde Tore M. Undeland Roy Nilssen

Problem description Offshore wind power is one of the power generation methods for the future. The floating wind farms proposed in the North Sea are promising regarding their potential for electrical power generation, but several technical challenges need to be solved. One of these is grid connection to the on-shore grid. While HVDC or HVDC light has been proposed for transmission to the coast, the interconnections between the turbines are too short for this technology. Here ACcables are more economical. In the internal grid of a wind farm, there is a need for voltage and power stabilisation. For this purpose, a STAtic synchronous COMpensator (STATCOM) with energy storage system is proposed. The STATCOM should be able to both keep the voltage steady, and smoothen out instantaneous active power fluctuations. The task In the master work, the focus was mainly on the converter topologies and the control system. The STATCOM part was implemented using a 3-level, PWM-switched, neutral point clamped (NPC) voltage source converter (VSC). The control system for the VSC was implemented using vector control and decoupling of the direct and quadrature axis. When tuning the PIcontrollers in the control loop, the optus modulus criterion and the symmetrical criterion were applied to the system. Going from to two three levels in the converter adds some complexity to the controller. A simple voltage balancing technique was added to avoid improper balanced DC-bus voltages. To interface the super capacitor bank with the STATCOM, a half-bridge buck-boost DC/DC-converter was implemented; this to control the current flowing from the super capacitor to the DC-side of the STATCOM. The time constants of the super capacitor controller were tuned based on the wind power fluctuations. Since this implies mechanical time constants are these noticeably larger than those of an electrical system. Due to this difference in time, the two control systems were disconnecting, assuming that the STATCOM system settles immediately compared to the SCESS. The full control system is shown in the figure below.

35

The super capacitor bank size was estimated roughly, based on wind power measurements and the control strategy for the active power compensation. The final step was implementing the STATCOM in the laboratory, using DSP and digital control techniques. Model/ measurements The full STATCOM/SCESS simulation model is shown below. The device is connected in shunt in the connection point of a wind turbine connected to a weak grid. The system was simulated in EMTDC/PSCAD.

Results/Conclusion The simulation results showed that the STATCOM/SCESS is capable of compensating reactive as well as active power in a grid with wind power integrated. The voltage stabilization ability of the STATCOM proved to work satisfactory. During small variations, the voltage was kept stable, while the implemented droop control gave a stable, lower voltage, when the reactive power capacity of the converter was exceeded. The compensation of the active power was functioning as assumed, and resulted in a stable, constant active power, compared to the output of the wind turbine. However, more attention should be payed to the dynamic reference of the active power controller, as this ended up either being to high or to low for the present power level in the grid. Hence, the super capacitor storage ended up going into high or low saturation, and this affected the active power level in the grid. The laboratory work showed the same general patterns as the simulations. However, due to limited time available for practical implementation, only the STATCOM part was implented, and only some steady state measurements were carried out.

36

Balancing of Offshore Wind Power in Mid-Norway

Student: Morten Gleditsch Supervisor: Terje Gjengedal Contact: Terje Gjengedal Collaboration with: Statkraft and the Technical University of Denmark Problem description In order to comply with governmentally announced greenhouse gas emission reductions goals and to consolidate an independent and stable electric power and energy supply, Norway must increase its installed renewable energy based power generation capacity. Profitability estimations, today's available technical solutions and regulations concerning preservation of natural resources leave construction of new small hydro power plants behind as the most plausible alternative together with construction of wind farms. Global trends such as technologic development and progress and the public opinion indicate that future wind farms in Norway will be located offshore. The assumption is supported by the recent handing out of a concession to an offshore wind farm project for the first time in Norwegian history. The projects name is Havsul 1 and the licence involves construction of 350 MW offshore wind power. Havsul 1 will be located in Mid-Norway, which is the region in Norway where the Norwegian Transmission System Operator (TSO) Statnett is most concerned about their ability to execute their task of assuring safety of power supply in the future. The concern owes to lack of generation capacity and transmission constraints. The task Experience show that commissioning of large offshore wind farms will impose power balance associated challenges on the TSO. Grid connection of two offshore wind farms of 350 MW and 1000 MW respectively shall be investigated, targeting reduction of frequency excursions. A model of of the Nordel synchronous power system shall be applied and the offshore wind farm models shall be connected to a bus bar representing Mid-Norway. Model/ measurements A slightly modified model developed by Sintef of the Nordel synchronous system in the power simulation tool DIgSILENT PowerFactory was applied. To execute the reduction task, a so-called centralised Load Frequency Control (LFC) scheme was implemented and four hydro power plants were designated to provide regulating power pursuant to a priority key that used their response times as input. To simulate power fluctuations in the time span of hours, real time wind data acquired from the Danish offshore wind farm Horns Rev 1 was used as input in the offshore wind farm model. These data were kindly provided by the Swedish power company Vattenfall. Conclusion The power fluctuations simulations showed that LFC is a well-fitted tool for bridling frequency excursions in the Nordel synchronous system caused by fluctuating power generation in an offshore wind farm. During the power fluctuations, which were of a particularly challenging kind, the system frequency complied with Statnett's normal operation requirements of 50 ±0.1 Hz. The results weren't too surprising since LFC has been used successfully in Europe for many years. They did however show that the amounts of the socalled frequency controlled normal operation reserves in Nordel may need to be expanded in case of a massive expansion in wind power in Norway.

37

Fault Ride Through (FRT) investigations were also conducted by introducing 3-phase short circuit faults at selected bus bars. The simulations showed that the FRT requirements in Norway were not violated even in the worst case simulations. Some choices regarding the setup of the model may have exalted the simulation results.

38

Spenningskvalitet i distribusjonsnettet

Student: Geir Gryttingslien Veileder: Kjell Sand Utføres i samarbeid med: Gudbrandsdal Energi AS Problemstilling I Gudbrandsdal Energis nett har det de senere år blitt tilknyttet flere likeretterlaster i distribusjonsnettet, som har gjort at spørsmålene rundt spenningskvalitet har fått økt fokus. Mange store alpinanlegg i området bruker tyristorstyrte likeretteranlegg og DC-motorer til å kjøre skiheisene. GE ser derfor behov for å utføre målinger på disse installasjonene og dokumentere i hvor stor grad disse påvirker leveringskvaliteten i området. Det største problemet med likeretting er at det oppstår overharmoniske strømmer som må kontrolleres slik at de ikke utgjør en fare for kraftsystemet. Oppgaven Målet med denne oppgaven er å dokumentere spenningskvaliteten i distribusjonsnettet til Gudbrandsdal Energi. Ved hjelp av målinger skal det undersøkes om heisanlegg med store likerettere produserer for mye støy i henhold til forskriftene. Der målingene viser at noen av parameterne er utenfor FoL skal det vurderes ulike tiltak for å opprettholde leveringskvaliteten. Videre skal det utarbeides planleggingsgrenser og fastsettes krav som skal stilles til eventuelle nye ulineære laster. Dette skal implementeres i Retningslinjer for nettilknytning, som gjelder for alle nettkunder hos Gudbrandsdal Energi. I tillegg vil en del av oppgaven bestå av å bli kjent med måleutstyr og tilhørende dataprogram, tolke data og heve generell kompetanse om leveringskvalitet. Resultater Lasten på Mosetertoppen gondol skaper overharmoniske spenninger som er utenfor forskriftskravene. THD er målt til 9,44 %. Forskriftskravet er 8 %. 5. harmonisk spenning er også utenfor grensene i FoL. Det er denne som gir det største bidraget til THD, men også 11. harmonisk ligger på grensen til hva som er tillatt. Simuleringer med Simulink viser at høy kortslutningsytelse og demping gjennom transformatoren gjør at de overharmoniske spenningene på høyspentsiden ikke er utenfor FoL. Dette gjør at lasten ikke skaper problemer for andre kunder i distribusjonsnettet. Konklusjon Grenseverdiene i FoL forholder seg til THD i spenningen. Som nettselskap er det derfor viktig å vite hva som ligger bak denne beregningen. Størrelsesforholdet mellom ohmsk og ulineær last og resulterende impedans i nettet er viktige parametere for å beregne THD. Det er derfor hensiktsmessig å sette planleggingsgrenser ut i fra kortslutningsytelse. Grenseverdiene bør også evalueres ved PCC, da hovedpoenget med å analysere disse forstyrrelsene er at de ikke skal skape problemer for andre kunder. Det er de overharmoniske strøminjeksjonene i kraftnettet som må reduseres for å dempe de overharmoniske spenningene. Ved å sette riktige krav til størrelsen på disse strøminjeksjonene avhengig av kortslutningsytelsen vil kravene til støynivået i FoL overholdes. Det anbefales å følge grenseverdiene i tabell 5-2 gitt av IEEE. I tillegg må det utføres frekvensanalyser for å undersøke om det kan være fare for resonans i nettet når det tilkobles ulineære laster. Det er kun på Mosetertoppen gondol det er avdekket forhold der de overharmoniske spenningene er utenfor FoL. Denne transformatoren er dimensjonert for å forsyne kun

39

heisanlegget og vil aldri forsyne andre kunder på lavspentsiden. Støynivået på høyspentsiden er såpass lavt at det ikke skaper problemer slik høyspentnettet ligger i normaldele. Dette skyldes at impedansen i linjene er lav. Allikevel er verdiene utenfor planleggingsgrensene gitt av IEEE. Siden denne lasten ble installert uten klare retningslinjer fra GE sin side, og parallelt med at de nye forskriftskravene ble innført, er det allikevel mulig å akseptere lasten slik den ligger. Denne installasjonen alene vil aldri kunne bidra til at spenningen på høyspentsiden kan skape problemer for omkringliggende tilknytningspunkt. Det anbefales derfor å godta at lasten kjøres uten filter, men lage klare retningslinjer for fremtidige heisanlegg og sikre at disse ikke bidrar til å øke problemene. Det bør lages en avtale der begge parter aksepterer at spenningskvaliteten er utenfor forskriftskravene. Det er viktig at GE i en slik avtale tar forbehold om endringer i kraftsystemet, slik at de på et senere tidspunkt kan kreve at det installeres filter hvis det skulle oppstå problemer på høyspentnettet. Dette kan bli en aktuell problemstilling hvis delingspunktene i nettet flyttes, eller radialen blir forlenget til nye utbyggingsområder. 6-puls tyristorlikerettere skaper store strøminjeksjoner av 5. og 7. harmonisk orden. Når flere slike anlegg kobles til samme høyspentradial kan summen av forstyrrelsene bli så store at de skaper problemer i nettet. For fremtidige installasjoner bør Gudbrandsdal Energi kreve at det installeres filter før slike anlegg over 500 KVA settes i drift. Dette bidrar til å holde %THD på høyspentnettet på et lavt nivå. Anlegg under 500 KVA må vurderes opp mot kortslutningsytelsen på stedet. 12-puls likerettere reduserer 5. og 7. harmonisk strøm med ca. 90 %, men de kan allikevel skape betydelige strøminjeksjoner av høyere harmonisk orden. Det er derfor viktig å ha gode modeller av nettet for å kunne simulere hvor mye støy som oppstår. Innkobling av 12-puls likerettere bør vurderes individuelt i hvert enkelt tilfelle.

40

Development of Procedures for Analysis of PMSM based on Simplified Permeance Network

Student: Supervisor: Contact: Collaboration with: Seyed Behzad Hatefi Professor Robert Nilssen Dr.Ing. Alexey Matveev, Dr.Ing. Sigurd Øvrebø SmartMotor AS

Problem description Permeance Network Method was developed in former works (TET5500-Specialization Project) to create a rapid model for an electrical machine which allows calculating its different characteristics. The main goal of this thesis will be to develop some analytical models for the analysis of Permanent Magnet Synchronous Machines (PMSM) based on Permeance Network Method. In these models, a presentation for the magnet will be introduced and magnet operating point will be obtained. A magnetic equivalent circuit will be developed to calculate different machine parameters. The optimized material permeability will be calculated considering high-loaded and saturated conditions. The magnet and iron losses will be analyzed and obtained. Finally, the comparisons will be made between analytical, Finite Element Method (FEM), and measurement results to define the accuracy of analytical calculations. The task Today, Finite Element Method is the most popular way for modeling and analyzing the behavior of electric machines. However, this method cannot be used when there is a need for fast calculations during the sizing and, sometimes, optimization stages of design process. At these stages analytical methods are preferable. The main goal of this thesis is to develop new analytical models for rapid analysis of one specific class of machines - radial-flux PMSM with concentrated windings and surface mounted magnets. Permeance Network Method is used to model the magnetic circuit. It is possible to define a general layout for the design of permanent magnet machines. There are different stages in design procedure. In one of the stages, by using inputs and analytical calculations, some parameters like air gap flux, back emf, magnet losses, and iron losses can be calculated. The calculation of these data is not accurate, and is done just for getting a general overview of design procedure. In the next steps and after modeling the machine by FEM, these data are calculated more accurately. Here, the main goals are to calculate these four parameters by developing analytical procedures (without using FEM) and to increase the accuracy of former mathematical models. The work is divided to four main parts. First, a general model for presenting the magnet in magnetic circuit is introduced. The magnet is analyzed and the operating point is calculated analytically. The variation of flux density in the magnet is analyzed and the comparison with FEM results is done. The limits in magnet flux density variation are obtained without FEM. Second, the equivalent magnetic model for the machine is introduced, taking into account the overloaded and saturation conditions. The same magnet presentation, as the first part, is used in the equivalent. Different parameters including magnetic fluxes are calculated analytically based on the data from Permeance Network Method. The optimized material permeability is introduced for any types of machine operation (normal, saturated, etc). The mathematical data is compared with FEM results, and the accuracies are defined. A simplified presentation of the magnet and air gap for Permeance Network Method is introduced.

41

Third, the losses in the permanent magnets are analyzed and different expressions are mapped for analytical calculations. The magnet losses are calculated from different equations based on the results obtained from the mathematical models, developed in previous tasks. The best equation is defined by comparing the results with FEM results. The importance of harmonics in magnet loss calculations is shown. Forth, the iron losses are analyzed and some discussions are made. One method for the calculation of iron losses is introduced. The method is based on the flux density variation curves in different parts of PMSM. According to different type of materials (e.g. steel) in sample machines, the iron losses in both rotor and stator are calculated from analytical methods. The comparisons with FEM are made to define the accuracy of mathematical calculations. Finally, by using the value of magnet and iron losses, the total losses are calculated from analytical method. The results are compared with the measurement results and the accuracy of analytical calculations is considered in an acceptable range due to measurement conditions. Model/ measurements As some examples from analytical works, two of the results are given here. Figure 1 shows the magnet operation point in a sample machine, including maximum and minimum limit for flux density variation. Figure 2 shows the iron losses in the stator and rotor for a PMSM.

1.4

A

1.2

70 60

Iron losses in rotor and stator

A

B

B

50

0.947 0.774

Iron losses (W)

0.857

40 30 20 10 0 0

B

0.6 0.4 0.2 0 -1.2

-1

-0.8

mu0 * H

-0.6

-0.4

-0.2

0

0.01

Time

0.02

0.03

0.04

Figure 1: Magnet operation point and magnet flux density variation limits for a PMSM

Figure 2: Iron losses in rotor and stator for a PMSM

Conclusion Some of the main results from the thesis are given here. As an important fact, it is seen that in high temperature applications and at overloaded and saturated conditions, the magnet cannot be presented as linear demagnetization curve below the knee area. It is shown that, the effect of using the optimized permeability dominates when the machine is in high loaded and saturated condition. Based on the accurate analytical method, the accuracy for the calculation of air gap flux and back emf with analytical methods is increased compared to previous calculations. In magnet loss calculations, it is seen that the magnet losses in the machine with sectioned magnets is much smaller than the one in machine with complete solid magnets. Both magnet and iron losses are calculated analytically and the accuracy of the calculations is defined in an acceptable range. The efficiency of the machines is calculated from analytical methods and the accuracy of analytical calculation is defined the comparison with measurement results.

42

Offshore Wind Farm Layouts Performance Comparison for a 540 MW Offshore Wind Farm

Student: Faglærer:

Thomas Haugsten Hansen Terje Gjengedal

Problem Description

With the recently accepted goals set by the European Union, stating that 20 % of all energy production shall come from renewable energy sources in 2020, and the Norwegian government's ambitious goal of no net CO2 emission in 2050, the development of wind power production is a more current issue than ever. Especially large scale offshore wind power will require attention to new focus areas. The wind may be more stable offshore, but there will be less geographical smoothing effect, so wind variations will still be a key issue. Power transmission, grid connection and the internal grid structure represent other main challenges for realization of large scale wind power, and especially for offshore wind farms. Off the Norwegian coastline, the large water depths of the Norwegian trench make it impossible to utilize today's technology, where the wind turbines are mounted to the seabed. To reach shallow enough areas, the wind turbines must be placed at a minimum distance of approximately 100 km from shore. At these distances, AC cable connection to shore is challenging, due to the high capacitance of the cables, and HVDC technology may be a competitive option. Presently, there is also a debate among the offshore wind community regarding the value of redundancy required in the offshore grid to maximize the energy yield, and the impact it may impose on the capital costs of the wind farm. The major concern is related to the cost of supplementary subsea cabling, either in terms of extra length or higher ratings, versus the value of the decreased losses during normal operation and contingency operation due to the redundancy provided to the wind farm. For small wind farms, the power loss due to a fault is relatively small, and redundancy has not been considered a profitable option. With planned wind farms in the 1GW range, the power and, in consequence, income lost due to a fault may be large enough for redundancy to be profitable. With this as the background, this thesis will focus on the effect of the offshore wind farm layout on the wind farm's performance. Different options for the design of the offshore grid and the transmission to shore will be studied. The wind farm investigated will have an installed capacity of 540 MW, and will be modeled in detail, meaning that all turbines are modeled separately.

43

Abstract

This master thesis has been written at the Department of Electric Power Engineering at the Norwegian University of Science and Technology. The work has been carried out at the Royal Institute of Technology in Stockholm, where the author spent the last year of his studies as an exchange student. In the thesis, six different designs of the electrical grid of a 540 MW offshore wind farm, placed 100km off the Norwegian coast, have been studied and compared. At this distance, AC cable transmission might be difficult because of the reactive power production in the cables. Taking this into consideration, two options for the transmission system to shore have been studied. In addition to the AC cable transmission, voltage source converter based HVDC transmission, in the form of HVDC Light, has been studied, giving a total of 12 models. The main scope of the thesis was to study the load flow situation and power system performance of the different offshore wind farm layouts. Two load flow cases were run for each model; the first studying the model when the active power transmission to shore was maximized, the second studying the model under a contingency situation. The reliability of the six designs was compared by calculating the expected number of cable failures during the life time of the wind farm for each design, and what consequence the disconnection of any cable would have on the power losses. In order to study the effect of the offshore grid design and transmission system design on the offshore power system stability, dynamic simulations have also been executed, and the voltage response and rotor speed response following a fault have been studied. All simulations have been executed in version 31 of the program PSS/E. The wind farm was modeled full scale, consisting of 108 wind turbines rated at 5MW. The wind turbines were modeled as doubly fed induction generators, using the generic wind model that comes with the program. The load flow simulations showed that an AC cable connection to shore gave lower total system losses than a DC connection for all designs. The lowest losses were found at the n-sided ring design in the AC/AC system, and the highest losses were found for the star design in the AC/DC system. These losses were 2.33% and 8.19% of the total installed capacity, respectively. In the dynamic simulations, a three phase short circuit fault, lasting 150ms, was applied at three different places in the system. The simulations showed that except from at the wind turbines that were islanded as a result of a fault, all dynamic responses were stable. The HVDC Light transmission to shore gave the highest voltage drops and the lowest voltage peaks offshore. Also, the maximum speed deviation was found to be larger when using HVDC Light transmission compared to using AC cables, with two exceptions; the radial and star designs when a fault was applied to the transmission system. A comparison of the six different grid designs showed that the results were varying. Based on the results in this thesis it has not been concluded that one of the offshore designs have better dynamic qualities than the other. The simulation results indicated that this is case specific, and more dependent on where in the offshore grid the fault occurs rather than the design of the offshore grid.

44

Pressure Tolerant Power Electronics

Student: Øystein Holt Supervisor: Professor Tore Undeland Contact: Magnar Hernes, SINTEF Energy Research Collaboration with: SINTEF Energy Research This project deals mostly with the thermal aspects of immersing an IGBT module in dielectric liquid. The module is tested in the active mode of operation to look at the thermal behaviour. Temperature sensors and thermocamera is used in order to document the temperature conditions. Attempts at making simple equivalents representing the thermal behaviour have been made. In addition, the module has been immersed fully in oil during double-pulse and normal switching in order to compare with the switching in air that was performed in the autumn specialization project. The testing in oil represents a reference to possible future testing in a pressurized environment. Simple models for some of the thermal behaviour were developed as thermal equivalent networks in a circuit simulator. Basic thermal behaviour was also studied using finite element methods in the software package COMSOL Multiphysics. A brief literature study regarding the inductance considerations for the bus bar connections in the test setup was performed. Thermal testing of the IGBT module in the active mode of operation was performed. As expected, the module without heatsink has a very low heat dissipation capability, with a dissipated power in the range of 10W giving chip temperatures in excess of 80 degrees Celsius. When a synthetic ester oil, Midel 7131, is added inside the open module, the time to heat up the module becomes significantly longer. Comparisons of the different temperature measurements for the two cases indicated that more of the heat is transported upwards from the chip when the cooling medium is Midel. Thermocamera measurements showed significant temperature gradients at the top of the substrate during testing. This was to be expected due to only one of the IGBT's of the module being heated. When using a heatsink, it was seen that the heatsink attains a quite uniform temperature. The temperature in the substrate away from the chips will attain approximately the temperature of the heatsink. The temperature will thus drop of more quickly when moving away from the chip area when using a heatsink. A sample inductance calculation by finite elements method was performed in COMSOL Multiphysics for two different busbar configurations. As was expected, the configuration with two bus bars over each other was significantly better than the configuration with two bus bars side by side. Components relevant to the test setup which had been subjected to passive pressure testing at 300 bars were investigated. Even though some of the components had been affected mechanically by the pressure, all components functioned properly during the electrical testing. Double-pulse switching of an open module in oil was performed. Waveform comparison with a test performed immediately before immersion in oil showed no significant differences before and after immersion. Comparison with the standard module that was tested in air during the autumn showed the transistor current waveform to be less noisy for the open module in oil.

45

Water was added to the Midel 7131, giving a relative water content of the Midel during testing of approximately 78%. When double-pulse switching the IXYS module in the Midel/water mixture at 600V and 200A no breakdown phenomena were observed. The work described in this report is a part of an ongoing research effort by SINTEF Energy Research to build knowledge about the pressurization of power electronics.

Picture of test setup for testing the open IGBT module in oil

46

Investering i vannkraftverk

Student: Veileder: Utføres i samarbeid med: Christina Kvamme Gerard Doorman BKK produksjon

Problemstilling Utvidelse og opprusting av vannkraftanlegg har de siste årene blitt mer aktuelt. I den forbindelse ønsker BKK produksjon å se nærmere på lønnsomheten av et slikt prosjekt for en av sine kraftstasjoner. Oppgavens formål har vært å utføre en investeringsanalyse for investering i vannkraftverk med hensyn på usikkerhet i fremtidige kraftpriser. Oppgaven BKK produksjon vurderer investering i økt produksjonskapasitet i Fosse kraftstasjon. I oppgaven skal det gjennomføres en investeringsanalyse av dette prosjektet med vekt på vurdering av usikkerhet. Rapporten analyserer lønnsomheten i et utvidelsesprosjekt på Fosse kraftverk som er en del av Bergsdalsvassdraget. Slik situasjonen er i dag er Fosse kraftverk en flaskehals i serievassdraget. Derfor er lønnsomheten av et tilleggsaggregat på Fosse kraftverk vurdert for tre ulike aggregatstørrelser. Separat er lønnsomheten av å bygge to nye bekkeinntak på Fosse kraftverk vurdert. I tillegg er det sett på lønnsomheten av å kombinere et nytt aggregat med bekkeinntak. Optimal størrelse for et nytt aggregat er også vurdert. Modell/målinger Et kjennetegn på vannkraftinvesteringer er store investeringsbeløp med lang levetid. Inntjeningen av investeringen er avhengig av fremtidige inntekter som investeringen utløser. På bakgrunn av usikkerhet i fremtidige kraftpriser er det utviklet scenarier som tar for seg ulike prisutviklinger. Scenariene tar utgangspunkt i Statnetts scenarier fra Nettutviklingsplan for sentralnettet 2008-2025. Simuleringsverktøyet Vansimtap er brukt til å simulere forventet fremtidig produksjon og inntekter. Simuleringene ble utført med SimtapEffekt som er et tilleggsprogram i Vansimtap. SimtapEffekt optimaliserer produksjonen innenfor uken, samtidig som den langsiktige produksjonsstrategien oppfylles. Beregninger Sammenlikning av simuleringsresultater i Vansimtap med SimtapEffekt viste at inntekter og produksjon økte i Bergsdalsvassdraget når produksjon ble optimalisert innenfor uken. Simuleringene i SimtapEffekt ble utført for alle scenariene. Simuleringene i scenario "Basis" viste at inntektene øker mest for alternativ 4, kun utbygging av bekkeinntakene. Med økte prisforskjeller over døgnet endret produksjonen seg slik at det ble mer produksjon ved høye priser, mens total produksjon over året ble mindre samtidig som flomtapet økte. Det ble videre utviklet en investeringsmodell i Excel. Inputdata for modellen er inntekter og produksjon fra simuleringene, og investeringskostnader. I investeringsmodellen inkluderes effekten av skatter og avgifter, da dette kan påvirke lønnsomheten. Den økonomiske analysen ga forskjellige resultater ut i fra hvilket scenario som er lagt til grunn. Konklusjon Med utgangspunkt i Scenario "Basis" viste lønnsomhetsanalysen at det bør investeres i bekkeinntak i Skårdalselvi og Fossegjelet. Investering i mer effekt på Fosse viste seg ulønnsom i scenario "Basis" og scenario "Fornybar". Hvis imidlertid fremtiden gir høye kraftpriser og store prisforskjeller innenfor døgnet som i scenario "Eksport" vil det lønne seg

47

å investere i mer effekt på Fosse kraftverk, i tillegg til bekkeinntak. Anbefalingen til BKK er at mulighetene for bekkeinntak bør vurderes nærmere. Utbygningen av de to bekkeinntakene vil ha få miljøvirkninger, men kan møte motstand blant lokalbefolkningen. Eksterne effekter som omdømme må derfor vurderes opp mot lønnsomheten i prosjektet. Lønnsomheten av bekkeinntakene bør også vurderes hver for seg.

48

Improving Stability of Ghana's Power System Using Power System Stabiliser (PSS)

Student: Supervisor: Kwaku Sarpong Mensah Olav Bjarte Fosso

Problem Description

The power system of Ghana has total generation capacity of 2030MW consisting 60% hydro and 40% thermal. The transmission system is made up of 45 substations and approximately 4,000 circuit kilometres of transmission lines with operating voltage of 161KV. Ghana's power system is interconnected with three neighbouring countries Togo, Benin and La Côte d'Ivoir. With increasing system demand, the power system has been made to operate close to its stability limit because system expansion has not proceeded as planned as a result of financial constraint. This has introduced a lot of stability problems, resulting in reduced damping of disturbances, voltage instability and sometime total system collapse, Two of the power stations have their generators equipped with PSS, but none of PSS is activated. It is believed that activating the PSS on some of the generators will improve system stability.

Task

· · · Assess the effectiveness of the PSS in improving stability of the Ghana power system and determine the most appropriate location for the PSS to optimise its performance. Determine the appropriate settings for the PSS necessary to avoid system oscillatory instability and improve both small signal and transient stabilities. Identify improvements that can be made to the power system in order to optimise the performance of the PSS.

Methodology

Data of Ghana and the interconnected systems were first collected, reviewed and modelled using PSSE program. Steady state stability studies were then performed to identify the inadequacy in the system during steady state operation. Dynamic stability studies were also carried out by selecting appropriate dynamic models for generators, exciters and governors that best fit the dynamic behaviour of the generating units in the PSSE program. Appropriate PSS models were selected for units equipped with PSS based on manufacturers' recommendation. Series of dynamic simulations were carried out to identify the best location and parameter settings for the PSS. Small signal stability studies were also carried out to complement the results obtained from the transient studies using NEVA program.

Results

PSS improves stability by adding damping to the generator rotor oscillation. The damping is achieved by modulating the generator excitation to develop a component of electrical torque in phase with the rotor speed deviation as shown in fig 1.1. Results from transient stability studies show that PSS at Akosombo GS damped effectively oscillations due large and small disturbance as shown in fig 1.2. Results from small signal stability studies shown in fig 1.3 shows the distribution of the system eigenvalues on the S-plane. The damping ratios of all the modes were found acceptable after the application of the PSS at Akosombo.

49

Fig. 1.1 Block diagram showing how PSS signal improves stability using the excitation system

Conclusions

The power system of Ghana has low steady state stability margin during peak condition. There is a high risk of voltage instability following a transmission outage on some critical lines. High generation of reactive power from generating units were observed due to inadequate reactive power compensation at the local level. The dynamic performance of Ghana's power system is greatly enhanced by the application of PSS on Akosombo generating units. Small signal and transient stability results showed that unstable modes were effectively damped. Inter-area oscillation of 0.5Hz between Ghana and Ivory Coast system, local area oscillation of 0.8Hz between the Akosombo units and Aboadze units in Ghana, were effectively damped within 7sec. PSS at Akosombo was found to be capable of preventing partial system collapse.

Recommendation

PSS on the Akosombo generating units be activated with their appropriate control parameter settings since their application has a positive impact on the dynamic performance of the Ghana's power system. Steady state stability operation be improved by installation of power factor correction devices at the local level .This will reduce the VAr generation from generating units and improve system voltages during peak load condition.

50

Intelligent software for monitoring of hydro power stations

Student: Joar Hylland Mikkelsen Supervisor: Lars Norum Contact: Øyvind Holm Collaboration with: Voith Siemens Hydro Power Generation Problem description The goal is to study how modern computer software can be used in advanced monitoring of hydro power stations. Today monitoring systems for hydro power stations mainly reacts on single information given on a binary level. To get better grounds for decision making modern computer software can be used to automatically compare different information making it possible to detect unwanted operating conditions and faults that are developing at an early stage. The task The task is to investigate two different software products containing features needed to perform advanced monitoring of hydro power stations. One of the software products is developed by a company named Volve the other is a software product developed by a company called SKF. In the report the different features of the two software products is shown. How the two software products works and how they detect faults are explained and the advantages and disadvantages of the two software products are discussed. In addition to investigation of these two software products two new types of sensors is investigated. Smoke sensors are investigated since they can contribute with information to the monitoring system that can make the monitoring system capable of detecting faults that produce smoke. It is investigated which types of smoke sensors that are available today. Ultrasonic detection sensors are investigated since they can give a lot of information about faults in bearing systems in addition to information about electrical discharges. Model/ measurements Results from monitoring of Swedish hydro power plants with the software developed by SKF were investigated. The results were investigated to find out in which way the software helped the user in detecting unwanted operating conditions and faults at an early stage. The program uses vibration analysis to give information to the user about faults. The figure under is a figure of a vibration spectrum showing a possible electric fault. In this vibration spectrum it is peaks visible at 100 Hz and the multiples of this frequency. This is a clear indication of electric generated vibrations and the fault generating these vibrations must be electrical. The software finds patterns of faults like this one and uses it to detect faults developing and present an alarm to the user. In this way the user is made observant of this fault at an early stage and he can investigate the fault closer and prevent it from developing further.

51

Conclusion This report shows that the two software products are quite different. The software from Volve is software meant to construct an expert system capable of recognising faults from previous cases of faults. The software from SKF is software that gives intelligent machine diagnostics from analysis of vibration measurements in addition to measuring and trending of other variables. It also gives the user tools for analysing the root cause of faults influencing the bearing system of different industry machinery. This means that the software from SKF demands some involvement from the user to produce the best and most precise results. The expert system developed from the Volve software on the other hand is meant to present only results and advice to the user.

52

Optimal use of the hydro resources in Albania

Student: Heidi Theresa Ose Supervisor: Gerard Doorman Co-supervisor:Erik Høstmark, Statkraft

Problem Description

Since fall 2007 Statkraft has had an engagement in Albania, a country which is totally dependent on hydro power. More than 90% of the electricity comes from hydro power, especially the Drin river in North Albania (Black Drin). The main objective with this study is to analyze the utilization of the hydro resources in Albania and to look at potential improvements in the short term (the next few years) and the long term (after Albania joins the regional market). The study includes the following: A. Describe the relevant hydro power projects in Albania under the present and future market conditions. Problems regarding rationing should also be analyzed. B. Collect data and build a model in the EOPS model. The model shall include all relevant constraints of the present market conditions to get a reliable result. This includes the hydro power system, additional production, demand and supply and import and export. C. Run simulations with the EOPS model and compare the results with historical data from the existing power plants. Analyze and explain the deviations. D. Run simulations under a future scenario with a deregulated regional market. Analyze the changes in the supply situation in Albania and the utilization of the hydro power.

53

| ii

Summary

This Master thesis analyzes the optimal use of the hydro resources in Albania. Albania is a country totally dependent on hydro power. More than 90% of the electricity today comes from hydro power, mainly from the Drin river system. There are three hydro power plants located in the Drin river system: Fierze (500 MW), Koman (600 MW) and Vau Dejes (250 MW). Only one third of Albania's hydro power potential is today exploited, and Albania is a net importer of energy. The main objective with this study is to analyze the utilization of the hydro resources in Albania and look at potential improvements in the short term (next years) and the long term (after Albania joins the regional market). Two scenarios were worked out. The first scenario focuses on the present market situation in Albania. Investigations are done through simulations with the EOPS model. The results were analyzed and compared with historical data to discover potential upgrades of the utilization of water in Fierze, Koman and Vau Dejes. In the simulations the production in Drin river system is increased with 1.3 TWh in an average year. Fierze power plant has the highest potential with 25% more production in the simulation than what is shown through historical data. Under the process towards a liberalized market, the optimizing problem regarding the production planning will change. Today the main task is cost minimization given an expected demand. In a free market it will be profit maximizing given a price expectation. A second scenario dealing with the potential market situation in Albania in 2020 was worked out. In addition to the new market situation four new power plants were included in the EOPS model. With new plants in the Drin and a functioning market it is possible to achieve 1 TWh more production during an average year compared with the simulation for the present market situation in Albania. If the implementation of the market, new power plants and transmission lines are accomplished, the supply situation in Albania will improve substantially through more secure power delivery. However a participation in a regional market forces the production company to plan each day like the participants in the Nordic market, both in the long and short the term, to be able to exploit the technical and financial opportunities and develop their country.

54

Optimal bidding for wind power

Student: Kristian Wiik Ravnaas Supervisor: Gerard Doorman Contact: Hossein Farahmand Collaboration with: Trønder Energi AS Problem description In 2009 the Norwegian TSO Statnett launched a new price system for the balancing market. This new system removes the possibility for extra revenue from trading in the balancing market compared to the spot market. There is still the possibility that a producer will lose revenue by settlements in the balancing market. This change in rules is expected to have bigger influence on passive producers. The balancing market is divided in two price systems, 1-price and 2-price. In 1-price the balancing price is used for all settlements in the balancing market. This system is applied for all active regulations. For passive participation there are two prices used when settling, 2price. A producer who supports the regulation state of the system receives spot price for the imbalance. This could be when a producer has a deficit and the system is down-regulated. A producer with a deficit in a up-regulation hour is contributing to the system imbalance and is punished for it by being settled with balancing price. In short this means that in 1-price there will be both additional income and additional cost in conjunction with settlement in the balancing market. In 2-price there will only be additional costs. For a passive producer it is evident that minimization of the cost in 2-price is important. The task The task of this thesis is to develop, implement and test an algorithm for optimal bids in Elspot for a given wind power park. The algorithm should be based on a description of wind power production as a function of wind speed, and balancing prices. The uncertainty of wind power and balancing prices should be included. Model/ measurements The wind power and balancing price models creates different scenarios, each with an appurtenant probability to occur. The revenue for each combination of scenarios and bid size is calculated and the bid size which has the highest revenue is the optimal bid. The optimization algorithm is differ for each price system. The algorithm shown here is for 2price. Pbud are bid size, Pi are power in wind scenario i and RKp j are balancing price in scenario j. pj j and pi are probability for balancing price scenario and wind power scenario, respectively. The spot price is represented by sp . py are a binary variable which indicates regulation state for a scenario combination.

55

max REV Pbud sp n m py y 1 i bal 1 j 1 n m py y 2 i bal 1 j 1 n m py y 0 i bal 1 j 1 Pi Pbud pi sp pj j Pi pi RKp j pj j i 1 j 1 i bal 1 m Pi Pbud pi RKp j pj j Pbud Pi pi sp pj j i 1 j 1 i bal 1 m Pi Pbud pi sp pj j Pbud Pi pi sp pj j i 1 j 1

i bal 1 m

P

bud

Calculation The calculations show that the optimal bid for a producer in 1-price is either 0 % or 100 % of allowed volume. This depends solely of expected balancing price in the particular hour. For 2-price the bid will be in the same order of magnitude as forecasted wind power. The difference between expected wind power and optimal bid is due to the probabilities for the different regulation states.

For a random range of days the optimal bid will result in a higher net cash flow than the reference bid when actual production and price data is applied to the different bids. This applies to both 1-price and 2-price.

Conclusion An optimal bidding strategy is found for both 1-price and 2-price. This shows that the change in the rules of the balancing market works as intended.

56

"Integration and Stability of a Large Offshore Wind Farm with HVDC Transmission in the Norwegian Power System"

Student: Supervisor: Summary:

Fabien Renaudin Terje Gjengedal

In the last decades, due to the environmental concerns and the increase of energy demand, wind power has strongly penetrated the field of electricity generation. Today, because of the lack of onshore sites and visual and noise nuisances, the development of wind farms turns more and more to offshore and Norway has a great potential of offshore wind power. This thesis investigates the impact of the integration of an offshore 1000MW wind farm on the Norwegian power system. Two different transmissions are used, one HVAC transmission system and one HVDC transmission system. They are installed in four different configurations which represent the possible cases of wind farm integration regarding the distance from the shore. Two different connection points have been chosen regarding the load flow simulations. The first one is situated in the region of Bergen in the West Norway and the other one is situated between Kristiansand and Stavanger in the south Norway. In order to investigate the power stability and the behaviour of the system, simulations are performed under both steady-state and dynamic conditions by using PSSTME. Disturbances are applied in different locations on the system both near the connection point and on the offshore wind farm. The results show that the power system with large offshore wind power remains stable after the different faults. The requirements of the Norwegian Transmission System Operator, Statnett, are respected after the integration of a large offshore wind farm in the Norwegian power system.

57

Design and testing of Flux Switched Permanent Magnet (FSPM) Machines

Student: Supervisor: Njål Rotevatn Robert Nilssen

Problem Brushless permanent magnet machines are usually designed with magnets in the rotor. In recent years there has however been much research on machines where the magnets are mounted in the stator. The FSPM belongs to these new machines and such a machine was to be designed, built and tested. Theory

The figure above show the operating principle of the FSPM machine, each stator coil encircles one magnet and two stator teeth. The steel rotor determines the path of the magnet flux, and a sinusoidal back emf is produced with the correct combination of stator and rotor teeth. Model/measurements

Two machines with a length of 50mm and a diameter of 210mm have been built. One with 12/10 (Stator teeth/ Rotor teeth) as pictured above and one with a 12/14 layout, a rotor change is the only difference between the two designs. The machines have been simulated in COMSOL, where inductances, back emf and cogging have been found and compared with the measurements.

58

A screen shot from the 10 pole FEM simulation. Conclusion The machine runs well with both rotor configurations. The 10- and 14 pole rotor have a measured 400 rpm phase back emf, of 136.3V and 149.7V rms respectively, giving the 14 pole a torque constant approximately 10% higher than that of the 10 pole. The load tests of both machines show a peak electrical power at 400rpm of ~1150W with a power density of ~540kW/m3 at ~85% efficiency. The 14 pole rotor seems to be better than the 10 pole rotor, but comparison of the loaded machine performance is difficult without better control the dand q axis currents, especially as saturation issues causes the machine inductances to vary much with machine loading. The machine stator where made with a slight eccentricity causing the air gap to vary from 0.8mm to 1.2mm, This is believed to have increased cogging, but does not seem to have effected the machine phase- voltages and currents much, as these keep within 2-3% each other at the tested loads.

59

Use of artificial intelligence for optimal operation of a stand alone power plant

Student: Supervisor: Contact: Øyvind August Rui Prof. Tore Undeland Amin Hajizadeh

Problem description To operate a stand alone power plant several things must be taken into account. On an energy management level the power produced by the renewable energy sources should be predicted to have an optimal state of charge of the battery as well as an optimal on/off strategy for the diesel engine at all time. On a level below the converters must be controlled to execute the energy flow given by the energy management algorithms and to maintain the stability in the grid at all time. The task Artificial intelligence is a collective term including several computing methods that can be used for modelling, estimation, prediction and control. They are based on how nature behaves, especially how the human brain reasons and take decisions. These techniques are often suitable for non-linear processes and they are quite rapid. A disadvantage is that they often have no design procedure. The parameters must be tuned based on experience and heavily depends on the rest of the system. Artificial intelligence has a variety of applications within electrical engineering. In this report the possibilities for using it for energy management and control have been investigated. Model/ measurements The theory behind the artificial intelligence techniques fuzzy logic, neural networks and genetic algorithms are briefly presented. The use of these techniques for energy management and control of converters are presented based on a literature study of what has been done up until now. In the second part of the project simulations on a DCDC Boost converter are performed. This part will be further presented here. A boost converter (Figure 1) is a highly non-linear system and is difficult to control. In the simulations a fuzzy controller is compared with a classical PI controller. The goal is to keep the output voltage constant when the input Figure 1: Boost converter voltage is changing. This perturbation has applications for PV panels, fuel cells and super-capacitors. A resistor on the output side models an inverter operating at constant power. There are several topologies for fuzzy controllers. The most suitable topology for the boost converter was found by simulations and is shown in Figure 2. It is a fuzzy controller with an integrator in parallel. This controller was found in the literature; however an anti wind-up was added. This gave better response.

60

Figure 2: Fuzzy controller with integrator and anti wind-up

Calculation The perturbation in input voltage as well as the response of both the PI and fuzzy controller with integrator is shown in Figure 3. The simulation with the fuzzy controller shows significantly less deviation from the output voltage reference (200 V). There are oscillations in the output of the fuzzy controller. The reason for these oscillations is not understood and should be further investigated. The fuzzy controller demands more of the DSP. Measurements with a dSPACE and a simple fuzzy controller showed that the DSP became the limit on the bandwidth of the controller and could only update the 10 kHz PWM signal every fourth period.

Figure 3: Excellent response of fuzzy controller

Conclusion For prediction algorithms artificial intelligence perform as good as classical methods. The AI methods can be faster in some cases. It also often has the advantage of not being site dependent, such as for example wind power prediction based on flow modelling. For control fuzzy logic can be very suitable for non-linear processes. AI methods demand more computing power than a classical PI and should therefore only be used when needed. An other disadvantage is that there is no design procedure for the controller.

61

Development of a Grid connected PV System for laboratory use

Student: Supervisor: Contact: Collaboration with: Silje Odland Simonsen Lars Einar Norum Fritz Schimpf Bose Research, India

Problem description A laboratory setup of a grid connected PV inverter system is currently under development at NTNU. The system will consist of a PV panel input, a power converter stage consisting of a DC-DC converter, DC link and DC-AC converter, and a transformer stage connecting the system to the grid. The finished system is intended for implementation at the University of Dar Es Salaam in Tanzania for teaching purposes in Power Electronics and Digital Control in PV Systems. The focus of this master thesis is development of a control design for a DC-DC converter implemented in the PV system. This includes consideration of the input from the PV panels and the DC-DC converter stage. A DC-DC converter design is already available. The focus will be on hardware testing and evaluation, software development and interconnection of the hardware and software modules. The task For this particular master thesis the input stage comprising the PV panel and the DC-DC converter will be of main focus. A control design will be developed, comprising voltage mode control (with feedback from the input of the converter) and Maximum Power Point Tracking (MPPT). The DC link voltage level is set to be 48 V, while the input voltage will vary from 0 to 45 V. In the experiments the setup will consist of DC source simulating the PV-panel, a DC-DC converter and an electronic load representing the grid connection through an inverter and a transformer. A setup of the system is shown below.

62

Conclusion The DC-DC converter was built and tested in a previous master project and can be configured as a buck, boost or buck-boost converter. For this thesis the boost topology was chosen, as this topology is the one most frequently used in PV systems. The control was implemented through C code programming. A regular voltage mode controller was developed and tuned through utilization of Ziegler-Nichols ultimate sensitivity method. At first a P-controller was implemented, but it was not able to cancel out the error between the reference voltage and the input voltage. This was expected, and an integral part was added to form a PI-controller. Now the closed loop control of the system turned out to be rather good for the whole range of the input voltage. The MPPT algorithm Perturb & Observe was chosen to track the maximum power point of operation. The MPPT was tested for both step changes in irradiance and temperature levels. When varying the irradiance levels the current was the parameter most affected. Even though the MPP was tracked rather well there was uncertainty regarding the MPPT algorithm capability since the voltage was only exposed to minor changes. When the temperature was changed, the voltage was affected in higher degree. The MPPT was able to track the MPP rather well, and tracking in the wrong direction only happened right after a step change. In real life the temperature will normally not change in steps, hence this test was said to be done under extreme conditions.

63

Stabilitetsanalyse ved integrasjon av Midtfjellet Vindpark i kraftsystemet

Student: Håvard Singelstad Supervisor: Olav B. Fosso Supervisor: Trond Toftevaag Contact: Truls Drange v/ SKL Nett AS Collaboration with: SKL Nett AS Problemstilling Sunnhordland Kraftlag (SKL) ynskjer å undersøkja integrasjon av Midtfjellet Vindpark (MVP) på 150 MW i kraftsystemet, spesielt integreringa i 66 kV-nettet og endringa som er pålagd i nettet. Det er i konsesjonen beskrive sanering av line mellom Årskog- og Stord sekundærstasjon i 66 kV-nettet, seinast 10 år etter at vindparken er sett i drift. Alle analysar er utført med grunnlag i at denne saneringa er utført, i tillegg er det lagt til grunn ei opprusting i 66 kV-nettet på ulike overføringar som er planlagde i SKL regi. Det er utført lastflytanalyse for to case, eksportsituasjon (case1) og importsituasjon (case2). Den viktigaste delen av arbeidet som er utført, er dynamiske analyser av nettet ved feilsituasjonar og utkopling av overføringar. Ein annan viktig analyse er små-signal stabiliteten i nettdelen. Oppgåva SKL ynskjer i hovudsak å påvisa eventuelle dynamiske problemstillingar ved integrasjon av vindparken. For å undersøkja dette må det etablerast ein grunnmodell i SIMPOW med både statisk modell for lastflyten og dynamisk modell for dynamiskeanalysar og små-signal stabilitetsanalysar. Den statiske modellen vart etablert under forprosjek hausten 2008 medan den dynamiske modellen er modellert i dette arbeidet. Dei sentrale delane i denne hovudoppgåva er: · · · · Etablering av dynamisk-modell av nettet i Sunnhordland/Haugalandet i simuleringsprogrammet SIMPOW. Kort teoridel, inkludert ein presentasjon av krava til nettilkopling frå FIKS. Dynamiske stabilitetsanalysar i området for tilknyting av vindparken, for å påvisa eventuelle problem med integrasjon av Midtfjellet Vindpark. Påvisa eigenverdiane i nettet som er modellert, å vurdera dempinga for desse eigenverdiane.

64

Oppsummering og konklusjon Analysane viser at vindparken ikkje vil skapa problem i ein normal driftsituasjon i nettet. Lastflytanalysen viser at effekten frå MVP i all hovudsak vil flyta ut på 300 kV-nettet i normaldrift. Spenninga og belastninga er innanfor grenser som er definert i dei ulike nettdelane. Det kan oppstå problem med overbelastning i 66 kV-nettet dersom vindparken er drifta med tilkopling berre til 66 kV-nettet. Dette problemet kan bli større ved integrasjon av nye vasskraftgeneratorar i dette området. Dersom MVP er tilkopla til 66 kV-nettet, kan spenningsregulering her, samt effektproduksjon oppretthalda ein god driftsituasjon, dette visast spesielt ved tilkopling til 66 kV-nettet på ende. Små-signalanalysar av eksportsituasjonen (case1) viser at alle eigenverdiane er godt dempa, dei har alle eit dempeforhold over 5 %. Analysane viser eit stabilt system som er godt dempa, både ovanfor lastendring og topologiendring i 66 kV-nettet i det området MVP skal integrerast. Analysane viser at eigenverdiane er meir sensitive for topologiendring enn lastendring.

Systemets eigenverdiar

65

Balance costs for wind power

Student: Stig Skaflestad Supervisor: Gerard Doorman Contact: Stein Danielsen (Powel), Michael Belsnes (Sintef) Collaboration with: Powel ASA Problem description The variability of wind power implies an increased risk of high balancing costs for responsible participants. A new market scheme, the two-price system, strengthens the incentive for maintaining production balance. The task Study balance costs of wind power in the new market scheme and look into a method for combined scheduling of wind and hydro power. Experiments on combined scheduling, or joint operation, are carried out using state-of-the-art optimization tool SHOP to indicate potential cost reductions. Model/ measurements The figure below illustrates the main principles of the method of study, known as the Balance costs model. Given a feasible set of in data, SHOP provides optimal hydro schedules. The optimal schedules are altered corresponding to the wind farm's deviation from spot, also referred to as balancing demand. The altered schedule is optimized again for feasibility.

The difference in economic results determines the benefit of joint operation. For separate operation, balancing costs are a result of wind power deviations, while for joint operation, balancing costs are a result of infeasible schedules. Test subjects are hydro plants Øvre and Nedre Røssåga (combined) of Statkraft with installed capacity of 487MW, and wind farm Smøla with installed capacity of 150MW, also property of Statkraft. Calculation Calculations/analysis are based on actual data from the year of 2007. A total of 46 cases was run, each with a duration of three days, one week apart. Results from each run include total income and balancing costs for one- and two-price systems, costs of additional water use in joint operation and accumulated production and deviation.

66

The average of differences between separate and joint economic result is interpreted as the overall benefit of joint operation. For the spring cases, the relative increase in income was 5%, while the fall cases showed a decrease of 4%. Aggregated for the full year, the overall relative increase in income between separate and joint operation is 1%. Calculations are made deterministically with access to perfect information. Conclusion The setup suggested in the Balance costs model do not appear to be beneficial, based on the very small increase in revenue of 1% in analysis environment with access to perfect information. It is probable that a setup with a stochastic approach, better would illustrate the consequences of forecast uncertainty for wind power participants. Such a setup should model both the uncertainty of the balance market, and the uncertainty of wind power forecasts.

67

Undersøkelse av beregningsverktøy for elektriske lavspenningsinstallasjoner

Student: Hovedveileder: Biveiledere: Johan Skogsrud Eilif Hugo Hansen Arne Nysveen, Elkraft og Øystein Trulsen, Norconsult

Av dagens beregningsverktøy for elektrotekniske installasjoner har Febdok den største markedsandelen. Febdok benyttes av elektroentreprenører, industri, rådgivende ingeniører, tavlebyggere og flere andre aktører i elektrobransjen. I denne oppgaven vurderes Nettdok, som er et nyere og mer avansert dataverktøy for elektroteknisk dokumentasjon, opp imot manuelle beregninger og Febdok. Hovedfokuset i denne oppgaven har vært å kontrollere noen av de matematiske beregningene som utføres av programmene. De manuelle beregningene er utført etter standarder der formelverk for elektrotekniske beregninger er presentert. Riktigheten til beregningene av kortslutningsstrøm, støtstrøm, spenningsfall, generatorkortslutning, motorkortslutning, UPS, selektivitet, kaskade, valg av overbelastnings- og kortslutningsvern er sett nærmere på. Resultatene viser at Febdok og Nettdok har en rekke feil, disse er av varierende omfang og alvorlighet. Kortslutningsberegninger viser at ingen av programmene tar for seg feiltypen topolet kortslutning med samtidig jordslutning. Dette fører til at den høyeste kortslutningsstrømmen og støtstrømmen ikke blir beregnet i flere tilfeller. Nettdok oppgir alle feilstrømmene som kmaks I og kmin I . Ved TN-nett er disse beregnet ved henholdsvis trepolet og enpolet kortslutning. De andre feiltypene som er dimensjonerende i flere tilfeller, blir ikke beregnet. Ved valg av vern kontrollerer ikke Nettdok krav i NEK 400 til kortslutningsvern ved utkobling under 0,1 sekund. Dette fører til at kabelen ikke er beskyttet mot høyeste kortslutningsstrøm. Selektivitet mellom vern kontrolleres etter selektivitetstabeller i begge programmene. Brukes vern fra ulike leverandører kontrollerer Febdok selektiviteten automatisk, forutsatt at vernene er innlagt med tilstrekkelig data. I Nettdok må denne kontrollen utføres manuelt. Kaskade mellom vern kan bare dokumenteres i Febdok, denne kontrollen må gjennomføres uavhengig av programmet om Nettdok benyttes. Beregninger på UPS viser at Febdok har noen fortrinn som kontroll av maksimal termisk overlast, tåleevnen til statisk switch i bypass krets og mulighet for separat innmating til bypass kretsen. Febdok beregner imidlertid ikke enfase UPS, noe som er tilgjengelig i Nettdok. Det er også utført beregning på kortslutningsbidrag fra generator og motor i Nettdok. Dette er noen av de mulighetene som gjør Nettdok til et mer avansert verktøy. Disse beregningene gjøres etter standarden IEC61363-1 som bare tar for seg trepolet kortslutning. Dette sammen med at verdiene beregnet ved generatoren har forholdsvis store avvik sammenlignet med leverandørdata, gjør at manuelle betraktninger må gjøres uavhengig av programmet. Skal Febdok og Nettdok vurderes opp mot hverandre som rene beregningsverktøy er Febdok foreløpig å foretrekke grunnet mindre feil sammenlignet med Nettdok. Samtidig så har Nettdok et bra konsept og en del funksjonalitet som tegning, mulighet for flere innmatinger, generatorberegninger osv. Dette fører til at programmet er fordelaktig ved større og mer komplekse anlegg.

68

Nettutviklingsplan for Tromsøya-Nord

Student: Hovedveileder: Biveileder: Arnulv Steinholt Eivind Solvang Fredd Arnesen, Troms Kraft Nett AS

Tromsø by er under konstant utvikling, både når det gjelder økning av innbyggertall og etablering av ny industri. Formålet med oppgaven er å utvikle en nettutviklingsplan for 11 kV distribusjonsnett forsynt fra Gimle transformatorstasjon for perioden 2010 til 2030 (analys-erperioden). Gimle transformatorstasjon er lokalisert på Tromsøya Nord og er en svært viktig stasjon på Tromsøya da den blant annet forsyner Universitetssykehuset og Universitetsområdet for øvrig samt Breivika Havn. Kraftnettet tilhører Troms Kraft Nett (TKN). I dag er transformatorstasjonen nesten full-lastet og det trenges mer effekt og energi i området. For å analysere hva morgendagens behov medfører, er det foretatt en teknisk /økonomisk analyse av nettet og fremtidige belastninger i området. Det er foretatt lastflytanalyser, tapsanalyser, og analyser av ikke levert energi (ILE). Drift og vedlikehold er ikke tatt med i analysen. Det er antatt full forsyningskapasitet fra det overliggende nettet i analyseperioden. Nettanalysene er utført i NetBas og benyttes videre for beregning av investerings-, taps-, og avbruddskostnader for alternativene. Det er summen av disse kostnadene for hvert alternativ som minimeres og alternativene rangeres etter totale kostnader i analyserperioden. Det foreligger med dagens prognoser en forventet lastøkning i størrelsesorden 15,5 MW i tillegg til en generell lastøkning i nettet som forsynes av Gimle transformatorstasjon. For å imøtekomme belastningsøkningen må det iverksettes tiltak. Det er gjort analyser på tre hovedalternativer for å se hvilket som er best egnet for å møte belastningsøkningen: 1. Reinvestere og utvide Gimle transformatorstasjon 2. Ny transformatorstasjon i Breivika havn 3. Ny transformatorstasjon nord på Tromsøya I hvert alternativ er det også tatt med kostnader som er felles for alternativene og som kommer uavhengig av hvilket alternativ som blir gjennomført. Det er reinvestering av kabler og fordelingstransformatorer, samt reinvestering av kontrollanlegg i Gimle transformatorstasjon. Disse har eller vil i løpet av analyseperioden overstige antatt teknisk levetid. Resultatene er delt opp i to forskjellige kategorier. Teknisk/økonomisk beste alternativ og teknisk beste alternativ. Det er hovedalternativ 2, Ny transformatorstasjon i Breivika havn, med 1 krafttransformator og 2 kabler til eksisterende nett og 1 kabel til Solneset i 2016 som er teknisk/økonomisk optimal i forhold til totale kostnader i analyseperioden. Kostnaden er beregnet til 99 650 kkr. Det må bemerkes at dersom det monterer 2 krafttransformatorer med en gang vil dette føre til en marginal økning av totale kostnader på ca 690 kkr. Det er usikkerhet om beregning av

69

KILE kostnader er utført på optimal måte og det anbefales dermed å montere 2 transformatorer med en gang siden dette er en teknisk bedre løsning. Differansen i totale kostnader mellom det å ha én kontra to krafttransformatorer er mindre enn differansen i investeringskostnadene, Dette er på grunn av mindre taps- og forventede avbruddskostnader for alternativet med to krafttransformatorer. Av de tekniske løsningene er det hovedalternativ 1, Reinvesterer og utvide Gimle transformatorstasjon, med 4 kabler til Breivika Havn samt 2 kabler til Solneset Boligfelt og ny ekstra krafttransformator T3 i Gimle, som er rimeligst løsning. Totale kostnader for alternativet er beregnet til 103 900 kkr. Det er vanskelig å komme med en entydig anbefaling av hvilket hovedalternativ som bør gjennomføres. Alternativene er tilnærmet like gode og differansen i totale kostnader mellom de forskjellige alternativene er liten. Det er derfor ingen alternativ som skiller seg direkte ut. Det anbefales i tillegg at det utføres mer inngående analyse av pålitelighet og teknisk tilstand for de forskjellige alternativene. Belastningsutviklingen må følges opp kontinuerlig og før en iverksetter investeringer må det undersøkes om forutsetningene har endret seg. Det er også viktig i en slik sammenheng å se nærmere på lokalisering, tomtekostnader, adkomst m.v. Dette inngår bare i svært begrenset grad i oppgaven.

70

Ny HVDC-kapasitet og vindkraft i Sør-Norge

Student: Kristian Stray Supervisor: Prof. Olav Bjarte Fosso Contact: Klaus-Ole Vogstad Collaboration with: Agder Energi Produksjon AS Problemstilling Det er flere nye likestrømsforbindelser under planlegging mellom Norge / Nordel og det kontinentale Europa / UCTE. Foruten NorGer-prosjektet har også Statnett SF om en ny forbindelse fra Kristiansand til Danmark. Stor importkapasitet på HVDC-forbindelsene setter spesielle krav til tilstanden i området hvor eksisterende konverterstasjonene er etablert. Eventuelle nye forbindelser gir opphav til økte utfordringer, spesielt i situasjoner hvor man har høy import samtidig med et lavt forbruk. En slik situasjon kan innebære lav produksjon i Sør-Norge, slik at systemtilstanden med hensyn på kortslutningsytelse kan karakteriseres svak. Kortslutningsytelsen sett i forhold til total importkapasitet kan si noe om forventninger til driften av likestrømsforbindelsene. Situasjonen som beskrevet i forrige avsnitt kan også være kritisk med tanke på stabilitet i kraftsystemet. Bortfall av import kan oppstå som følge av lav spenning på konverterterminalene. Det er således en sammenheng mellom stabilitet i AC-kraftsystemet, spenningsstabilitet og polhjulsvinkelstabilitet, og forutsetningene for feilfri drift av HVDC-systemene. Fremgangsmåte Det utarbeidet et analysegrunnlag på bakgrunn av antagelser vedrørende forventet last- og produksjonsutvikling, samt kapasitetsøkninger som kan forventes frem mot en eventuell idriftsettelse av NorGer. Det er tatt utgangspunkt i konsesjonssøknader og meldinger tilgjengelig via NVE og konklusjoner fra en studie mottatt fra Statnett SF vedrørende nettutvikling forutsatt nye likestrømsforbindelser. Import og vindkraft er balansert ved å ta utgangspunkt i en historisk situasjon for Sør-Norge med lav kortslutningsytelse. Likestrømsforbindelsene NorGer, samt Skagerrak12, Skagerrak4 og NorNed, er modellert og etablert med tilhørende reaktive komponenter i kraftsystemmodellen over Norge. NorGer er tilknyttet 420 kV-nett i Tonstad sammen med et vindkraftanlegg. Analysegrunnlaget har vært utgangspunkt for lastflytberegninger og dynamiske simuleringer for å undersøke konsekvensen av ytterligere HVDC-kapasitet og vekselvirkning HVDC-Vindpark. Det er utført kortslutningsberegninger for å identifisere linjeutfall i 420/300 kV-nettet i SørNorge som kan ha konsekvens for driften av konverterstasjonene. Beregningene er gjort med og uten vindpark. Bortsett fra dobbeltutfall av forbindelsen Tonstad-Feda, hvor en mulig kritisk systemtilstand er identifisert for NorNed, viser resultatene at systemet kan karakteriseres moderat sterkt for alle gjennomførte (N-1) utfall med gjennomførte forsterkningstiltak. En dynamisk stabilitetsanalyse er gjennomført med PSS/E, der simuleringer er gjort på bakgrunn av lastflytresultatene. Systemet er transient stabilt for alle gjennomførte simuleringer, men det er identifisert svakheter i modelleringen av enkelte statiske reaktive komponenter.

71

Konklusjon Med de nettforsterkningstiltak som er lagt til er systemet transient stabilt for alle simuleringer. Vindparkens konsekvens for NorGer har vist seg vanskelig å identifisere. En fordel synes å ligge i at et vindkraftanlegg kan være i drift på tross av lave kraftpriser, ettersom vind ikke kan magasineres. For videre arbeid anbefales en studie av prinsipiell karakter for å lettere kunne fastslå konsekvenser.

72

Losses and Inductive Parameters in Sub Sea Power Cables

Student: Supervisors: Collaboration with: Ronny Stølan Arne Nysveen (NTNU) Jarle Bremnes (Nexans Norway AS) Nexans Norway AS

Problem description Detailed knowledge of the magnetic field and its effects are essential for design and analysis of subsea cables. In recent years the market for subsea cables supplying electric power to subsea process facilities such as booster- and water injection pumps has been growing steadily. Tailor made cables for such application is a focus area for Nexans Norway AS. The umbilical supplying power to these units may comprises steel tubes and signal cables in addition to the power cable itself and detailed analysis is needed for calculating the different cable parameters. The task · Measurements on fabricated sub sea cables. · Finite Element Analysis of fabricated sub sea cables. · Establish a test set-up for measurement of armouring losses in test cables. · Measurements of permeability for applied armour steel qualities. · Calculations of cable parameters and loss according to IEC standard. · Comparison of results from measured-, calculated- and analysed parameters. Model/ measurements Measurements on two fabricated sub sea power cables were conducted. Figure 1 shows the circuit diagram for the measurements of a sub sea power umbilical. The cable consists of two three-phase circuits in two layers, helically twisted opposite of each other. In addition to the power cables, the inner circuit also has two steel tubes and a centre copper wire. The armour consists of two layers with a total 130 galvanized steel wires.

Figure 1: Circuit diagram for measurements on sub sea power umbilical.

73

Calculation Figure 2 shows surface- and contour plots for the power umbilical.

(a) Inner circuit

(b) Outer circuit

Figur 2: Surface- and contour plot of magnetic flux density and magnetic potential

Conclusion The armour's magnetic permeability was found to have an insignificant effect on the loss in the armour wires, within the region of expected magnetic flux density. Inductance values increased slightly with a relatively large increase in magnetic permeability. Calculated parameters based on measured values compare well with the computed values using finite element analysis. Despite problems with grounding and an unstable neutral point during the tests, the measured parameters from the frequency with the least amount of total harmonic distortion, do not deviate with more then 5% compared with computed values.

74

IMPROVEMENT OF POWER SUPPLY RELIABILITY CASE STUDY: ZAMBIA Student: Terence Tambatamba Supervisor: Ivar Wangensteen Problem Description The reliability of the power supply system in Zambia is not satisfactory. Two recent power blackouts have been serious. The purpose of this Master Thesis is to investigate the system by means of stability analysis, contingency analysis and review of protection settings with the aim of improving the reliability of the system. The work includes data collection from Zesco Ltd, Zambia The following tasks are included: 1. Dynamic stability studies of Zesco generators by use of SIMPOW 2. Contingency analysis for critical components in the transmission grid by use of NETBAS. 3. Investigation of protection settings for transmission, generation and consumption. 4. Work out recommendations based on the above mentioned investigations. Results The simulated results in SIMPOW are shown below

Fig 1 Transient response of Kafue gorge generators to fault at Leopards Hill

75

Fig 2 Transient response of Victoria Falls generators showing loss of synchronism Conclusions Transient Stability Study Transient stability of the generators in the Zesco system was studied. The system was modelled in SIMPOW by writing a script called an `optpow file' for load flow simulations and the `dynpow file' for transient stability simulations (see appendix D in the thesis report). In order to study the dynamic behaviour of the generators when subjected to a large disturbance, the generators were modelled as type 2A (i.e. model with one field winding, one damper winding in d-axis and one damper winding in q-axis. Saturation excluded). Generator parameters obtained from the Zesco catalogue were used. The simulations focussed on the dynamic behaviour of Zesco generators when subjected to a three phase short circuit applied at three selected buses namely Kariba North, Leopards Hill and Kafue gorge. These buses were selected because they are considered critical to the system. The duration of the short circuit in each case was varied until stability in at least one set of generators was lost. Results show that with the fault duration of 100 milliseconds, all generators in the system were able to regain synchronism after the fault was cleared but with the duration extended to 200 milliseconds the generators at Victoria Falls power station, which is the smallest of the three, were the first to lose synchronism. The simulations were carried out at peak load condition. It

76

can therefore be concluded that at the current loading level, the critical clearing time should be less than 200 milliseconds. This guarantees that no generator will lose synchronism. Contingency Study This study was implemented using software called NETBAS. Results show that although certain constraints are violated when a contingency is applied (because the analysis is done at peak load condition), the Zesco system can operate under (N-1) criteria. The violated constraints can be corrected by load-shedding. The power blackout of 21st January 2008 (see section 2.4.2 of the thesis report) was caused by an N-2 contingency. Appendix C of the thesis report gives the details of the consequences of each contingency. Protection System The study proposed that when calculating the distance relay protection settings, the effect of zero sequence mutual coupling that exists between parallel circuits should be considered. As seen in section 6.5.1 of the thesis report the performance of conventional distance relays on parallel lines is negatively impacted by the zero sequence mutual coupling effect between lines. Using settings based upon the worst case scenario results in sub-optimal performance of the relays under other operating conditions. It has been shown that it is feasible to achieve an optimal distance relay performance on parallel lines by accessing multiple additional locally available signals and adapt relay operation based on the signals availability and the line status. The adaptive protection scheme could provide an enhanced distance protection for parallel lines as no remote signals are required for the scheme. The adaptation to the signal availability provides a built-in fallback scheme, which ensures the reliable operation of the relay under all conditions. The application of the new adaptive scheme would enhance the performance of distance protection on parallel lines. The settings for under/over frequency protection should be implemented as described in section 6.5.2 of the thesis report. Other Considerations: Automatic Under Frequency Load Shedding The absence of an effective automatic under-frequency load-shedding scheme contributed significantly to the failure to contain the disturbances. Implementation of such a scheme would enable Zesco respond to the loss of generation by automatically switching off appropriate loads, thereby balancing demand with the available generation. Zesco should consider implementing such a scheme to improve reliability of power supply.

77

System Capacity and Spinning Reserve

The other most important factor for the building up of faults into total blackouts has to do with the failure to maintain a generation spinning reserve (spare capacity at generation). While it's understood that Zesco is currently under pressure to minimize load-shedding, the absence of a spinning reserve contributes to the inability of the system to contain the large disturbances. Zesco should review system operation and ensure that a reasonable reserve capacity is always maintained. New Generation Capacity Until significant new generation capacity is developed, there will be difficulties and compromises in running the Zambian system. However plans are currently underway to build a new power station at Itezhi tezhi and Kafue lower. This will greatly improve power supply reliability and reduce load shedding. System Monitoring It is currently difficult to reconstruct some of the events that occur during the disturbances because the recorders and protective relays on the system are not time-synchronized. It is therefore, recommended that Zesco takes immediate steps to acquire equipment that is needed to synchronise the so-called `time stamping' on all event recorders.

78

Energisparepotensiale ved bruk av intelligente veilyssystemer

Student: Veileder: Einar Tommelstad Eilif Hugo Hansen

Problemstilling Drift og vedlikehold av veilys utgjør en stor kostnad for norske kommuner og andre veilysholdere. Med ny styringsteknologi kan en i større grad regulere lysnivået etter behov, og således gi reduserte energikostnader. Systemene kan også benyttes for å redusere effekttopper, forenkle vedlikehold, samt å gi mulighet for avregning av anleggene. Systemer basert på LonWorks på powerline (PLT22-transceiver) er installert som testanlegg både på Gløshaugen i Trondheim og ulike steder i Oslo-området, og disse kan brukes til uttesting av ulike styrestrategier. Kandidaten skal: · redegjøre for dagens krav til veilys i Norge, og for muligheter for energisparing i ulike typer anlegg. Oppgaven Rapporten dokumenterer gjennom Statens vegvesens håndbøker hvilke veier, klassifisert etter dimensjonersklasser, som bør ha belysning og hvilken belysning som utgjør minstekravene. Gjennom normtall og eksempler fra prosjektering av veilysanlegg kommer energiforbruket frem og disse tallene gir et utgangspunkt for å beregne besparelsene av ulike ENØK-tiltak. Energisparepotensial ved ulike ENØK-tiltak er i rapporten eksemplifisert gjennom kategorisering av driftsløsninger. De fem kategoriene er: gammelt konvensjonelt anlegg, nytt konvensjonelt, nattslukking, trinnvis dimming og dynamisk dimming. De to siste kategoriene representerer såkalt intelligent veibelysning der styringssentralen nytter toveiskommunikasjon med armaturene for overvåkning og dimming. Eksisterende veilysanlegg er imidlertid svært ulike både i forhold til hvordan de er utstyrt og tilknyttet kraftnettet. Det er flere leverandører av styringsystemer med toveiskommunikasjon på det norske markedet. De ulike systemene er beskrevet og sammenstilt i rapporten, og de er teoretisk ikke så forskjellige. Modell/målinger

Gammelt Gammel armatur Nytt Ny armatur Natteslukking Ny armatur Trinnvis dimming Ny armatur Trinnvis dimming Elektronisk forkobling Sentral styring og regulering Dynamisk dimming Ny armatur Dynamisk dimming Elektronisk forkobling Sentral styring og regulering

Én inn- og utkobling Én inn- og utkobling Flere inn- og utkoblinger Konvensjonell forkobling Kun fotocelle Konvensjonell forkobling Kun fotocelle Konvensjonell forkobling Sentral styring

Oppsummering styrings- og reguleringskategorier.

1

79

Beregninger

Konvensjonelt anlegg, astrour medio februar

Ingen natts luk k ing eller dimming

Konvensjonelt anlegg

Nattslukking

100,00% 75,00%

Lysnivå

100,00% 75,00%

Ly s niv å

50,00% 25,00% 0,00% 12:00:00 00:00:00

Klokkeslett

50,00% 25,00% 0,00% 12:00:00 00:00:00

Klok k es lett

12:00:00

12:00:00

Anlegg med elektroniske armaturer

Trinnvis neddimming

Anlegg med elektroniske armaturer

Dynamisk dimming etter luminansmeter

100,00% 75,00% 50,00%

Lysnivå

100,00% 50,00% 0,00% 00:00:00

Klokkeslett

25,00% 0,00% 12:00:00 12:00:00

Lysnivå

12:00:00

Klokkeslett

00:00:00

12:00:00

Eksempel på effektnivået i anleggene avhengig av kategori.

Gml.konv. Anlegg uavhengig geografi Nytt konv. ­ 25-40 % Anlegg Oslo Anlegg Tromsø ­ 15-30 % ­ 15-30 % ­ 15-35 % ­ 15-35 % ­ 20-30 % ? ­ 20-40 % ? Nattslukking Dimming, trinn Dimming, dyn.

Energisparepotensial i prosent for kategoriene. Konklusjon · Utskifting av gamle konvensjonelle armaturer til moderne armaturer utgjør en besparelse på mellom 25 og 40 % avhengig av effektstyrker og valg av lyskilder. Årsaken er forbedret reflektoroptikk og høyere lysutbytte. · Besparelser fra nattslukking er kun aktuelt for kommunale og private veier, da offentlige veier i henhold til veinormalen ikke skal slukkes. Nattslukking kan imidlertid innføres i anlegg med konvensjonelle armaturer. Besparelsen er direkte avhengig av tidsperioden en velger å slukke veilyset, noe som innebærer 15-30 % reduksjon. · Med elektroniske armaturer kaller en veilysanlegget for intelligent i den forstand at armaturen kan dimme lyskilden samt kommunisere med en styringssentral. I tillegg til redusert energiforbruk ved hjelp av dimming kan en få store effektiviseringsresultater i vedlikeholdet fra overvåkingen av armaturene. Besparelsen fra dimmingen av lyskildene vil avhenge av effektstyrke og dimmeplan, og potensialet utgjør 15-35 %. · Leverandørene av intelligent veibelysning i Norge tilbyr styringssystemer som teoretisk sammenlignet ikke er så forskjellige. Dermed blir det prosjektgjennomføring og kompetansen hos leverandøren som blir avgjørende for hvor vellykket anlegg med intelligent belysning blir. Den tydeligste forskjellen er valg av powerline- eller trådløs kommunikasjon mellom områdenode og lampenoder. Per i dag er det Luminext og Datek markedsledende i Norge for levering av styringssystem for veibelysning. 2

80

Offshore Power Transmission

Student: Ragnar Ulsund Supervisor: Tore M. Undeland Contact: Tor-Eivind Moen Collaboration with: ABB Problem description

Offshore power transmission involves transmission of power over long distances using cables. Typical applications are power transmissions from shore to offshore installations such as oil platforms or power from an offshore wind park to shore. Offshore wind parks are a highly interesting alternative to onshore wind parks. The challenge, however, transmitting the power to shore in a cost-effective and reliable manner. Initially the project set out to look at the challenges related to offshore power transmission using alternating (AC) or direct current (DC) solutions. Here the possibilities were vast and could include: investigate theoretical and practical/economical outer limits of AC cable lengths, investigate the criteria involved in deciding whether AC or DC transmission is the best solution, develop a guideline for selection of design topology and technology solutions for various cases.

Summary Offshore power transmission is becoming an increasingly important issue. To moderate climate change, world leaders have set environmental goals that will be very difficult to reach without renewable power production and the removal of production units with high emissions. Wind power and electrification have been the focus in this report. Plans for the expensive wind power are already moving offshore. This report has made an attempt at suggesting a guideline for well-suited transmission systems, for wind power projects located at a distance in order to make them more economically attractive. Another emphasis has been to find the most suitable transmission system for gas turbines at offshore installations. As expected, the use of alternating current is best suited at shorter distances. At longer distances this system is still feasible up to 350 km, but losses will be high and there will be limited power available. A conventional thyristor-based direct current system will therefore be an attractive option for high power ratings and long distances. On the other hand, direct current based on voltage-source converters is considered more expensive, but has an improved control of reactive power and is therefore preferable to the conventional direct current system. To determine which system has the best design, one has to consider each case individually.

81

Title: Dielectric Spectroscopy of Bisphenol-A Epoxy Resin aged

in Wet and Dry conditions.

Name of student: Deep Vaishampayan Supervisors: Magne Runde Sverre Hvidsten Company: SINTEF Energy Research

Background

Epoxy is a commonly used material for offshore oil and gas applications. Epoxy resins are also widely used as matrices for structural composite materials, adhesives and organic coatings due to their good mechanical properties. For subsea applications it is important to study the effect of water on the dielectric properties of the material. Sensitivity of epoxy to humidity is indeed a serious matter of concern, when subjected to humid environments, these materials can absorb up to a few weight percent of water, leading to an overall degradation of the dielectric properties (lowering of resistivity, increase of permittivity and loss, decrease in breakdown voltage and affecting mechanical strength).

Problem description

The main objective with the work is to examine the effect of the combined action of high temperature and water on the long term performance of a Bisphenol A epoxy resin. This is characterized by measurements of dielectric response in frequency domain and glass transition temperature measurements by using Differential Scanning Calorimetry (DSC). More specific, the master thesis includes: Literature review of mechanical and electrical properties of epoxies subjected to humid environment. Description of experimental setups, procedures and test sample materials. Ageing of samples in wet and dry conditions. Examine the effect of the ageing on the glass transition temperature (Tg) of the epoxies and measure the impact on the complex permittivity. Measurement of dielectric response in frequency domain of new and aged samples (dry and wet conditions). To determine mechanical strength (stress-strain curve) of the epoxies.

82

Abstract

This thesis presents the laboratory test data on Bisphenol A epoxy insulation.This thesis work deals with electrical, mechanical and thermal analysis of Bisphenol A epoxy resin. The main aim of this thesis work was to examine if dry and wet aging changes the glass transition temperature (Tg) of the epoxies and measure the impact on the complex permittivity under different ageing conditions namely dry and wet. During ageing the samples (epoxy discs and dog bones) were kept in water at 20°C, 45°C and 80°C both in dry and wet conditions for a period of one month. After the samples were removed from ageing they were conditioned in a vacuum oven for one week. The effect of temperature and relative humidity on unaged epoxy i.e. dry characterization was determined by keeping the samples in climate chamber with 15%RH (Relative Humidity) and temperatures 20°C, 45°C and 80°C. The wet characterization was carried out with 90%RH and temperatures 20°C, 45°C and 80°C. The surface of these samples was painted with silver paint (electrodes). Two circular discs were used for dielectric response measurement and 2 rectangular pieces for water sorption measurement. The dielectric response was measured when equilibrium/saturation condition was achieved. The dielectric response was measured in the frequency range from 0.01 Hz to 1000 Hz at 200 volts (peak). The wet characterization showed increases with aging temperature. The dielectric loss was also increasing with the temperature. It can be deduced that the water uptake by the epoxy increases with increase in humidity and temperature. The dry characterization showed and has steady growth with aging temperature. Therefore it can be deduced that characterization done in dry condition didn't significantly affect the complex permittivity as compared with wet characterization. The glass transition temperature (Tg) of the samples were found using DSC (Differential Scanning Calorimetry) with a heating rate 20°C/min. The Tg was measured in the samples in dry condition before the water absorption process, then in samples after the water absorption process with moisture in the sample, and then in sample after the desorption. The effect of water on the Tg of the epoxy polymer was studied. Tg was increasing with aging temperature, for both dry as well as wet samples. The increase in the value can be mainly attributed to post curing process of the epoxy. The mechanical strength of the epoxy was studied by applying a tensile force to the dog bone shaped samples till breakdown and the stress versus strain curve was detected. This test was also performed on the dry aged sample before water absorption, then in sample which was kept under water at 20°C ,45°C and 80°C for absorption, and in sample which has undergone absorption and desorption at 20°C ,45°C and 80°C. The difference between the stress-strain curves was documented and discussed. The ageing temperature plays a significant role in reducing the value of stress and percentage strain at max. For dry aged epoxy, stress reduces around 14% from 200C to 800C. However for wet aged epoxy samples tensile strength reduces around 25%. For dry aged samples % strain reduces around 0,3 % and for wet aged samples it reduces around 0,5%.

.

83

Integrated electricity and gas markets in eTransport

Student: Supervisor: Co-supervisor: Collaboration with: Problem description The liberation of gas and electricity markets in Europe has created a need for a better understanding on how the two energy systems interact and depend on each other. A declining domestic gas production has motivated Denmark, which has an extensive electricity production based on natural gas, to further investigate these mutual dependencies. Traditionally, the two energy systems in question have been developed and operated separately, partly due to the use of different analyzing tools. For several reasons, analyzing tools designed for appliance on electrical systems are not well suited for calculations on gas systems, and the other way around. Model eTransport is an optimization tool for energy systems, developed by SINTEF Energy Research. The model includes multiple energy carriers and a variety of options for energy conversion and transport. The scope of this master thesis is to build a model of the integrated gas and electricity systems in Denmark, and carry out an operational analysis involving new gas supplies to the Danish system. The two supply options evaluated include a new gas pipe connection from Norway via Sweden, and an import of gas from Germany through existing pipeline constructions. Possibilities for exchanging electricity with Norway, Sweden and Germany are included, as well as the option of exporting natural gas to Holland, Sweden and Germany. Mari Røhmesmo Westeng Gerhard Doorman Bjørn Harald Bakken SINTEF Energiforskning

84

Results Two optimal operational states were identified, differentiated by the amount of gas imported and exported, as well as the amount utilized for electricity production. The Norwegian gas supply came out as favourable to the German, yielding larger imports, continued export to Germany and higher electricity production from natural gas. The results from the analysis should be further investigated. Due to the structure of eTransport as of today, patterns of change in energy prices or demand over the year or analyzing period can not be differentiated for different parts of the system. Modelling accruing costs in the gas supply chain also proved challenging, and did most likely result in the costs of electricity production from natural gas becoming too low. This gave a considerable export of electricity from Denmark to all adjacent countries that is not thought likely to become reality. Modelling gas flow in eTransport with the existing models creates large optimization problems, which on several occasions had to be stopped manually due to long computation time and failure of calculating optimal solution. A simplification of the model formulations might be an option to consider in order to secure that optimal solution is found when analyzing extensive gas systems.

85

System analysis of large-scale wind power integration in North-Western Europe

Student: Supervisor: Collaboration with: Lars Pedersen Øren Terje Gjengedal, Magnus Korpås Statkraft, SINTEF Energy Research

Problem description Wind resources in North-Western Europe have a vast potential, and the wind conditions along the Norwegian coast are considered to be among the best in Europe. All of the countries neighbouring the North Sea have ambitious plans to build wind farms both on- and offshore. It is of interest to investigate the impact such a massive expansion of wind power will have on the power system of North-Western Europe, with respect to other energy forms and to energy prices. Such an amount of offshore wind power may necessitate the creation of an offshore grid, allowing for interconnections between wind farms. An investigation into the effects of such an offshore grid will therefore also be in order. The project will cover the following issues: - Simulations of simplified power system scenarios set in the years 2005, 2020 and 2030. - Study how an increasing amount of installed wind power will affect energy prices, power production distribution, and power transmission flows. - Investigate how an offshore grid consisting of interconnections between offshore wind farms will affect the system. The task The simulations in this project were performed using an updated version of the power market model developed by Thomas Trøtscher for his master thesis "Large-scale Wind Power integration in a Hydro-Thermal Power Market". The model included 6 price areas: Denmark West, Denmark East, Norway, Sweden/Finland, Germany and UCTE/Others. The existing market model was modified in the following manner: - Split Norway into three price areas: Norway North, Middle and South - Add the Netherlands - Add the United Kingdom - Add corresponding offshore price areas for areas neighbouring the North Sea. Wind series were generated for each wind generator using reanalysis data. Scenarios were created for the years 2005, 2020 and 2030. In these scenarios, wind power capacities are increasing as time progresses. The 2020 and 2030 scenarios have been simulated with two alternative grid configurations: one where the offshore areas are connected only to their respective onshore areas and one where the offshore areas are also interconnected in an offshore grid. In total 7 different scenarios were simulated: 1. The 2005 scenario 2. The 2020 scenario without an offshore grid 3. The 2020 scenario with an offshore grid 4. The 2030 scenario without an offshore grid 5. The 2030 scenario with an offshore grid 6. The 2030 Mark 2 scenario without an offshore grid (features significant transmission line upgrades in Norway) 7. The 2030 Mark 2 scenario with an offshore grid Since the objective of this project was to study the concrete effects of wind power, load growth and changes in other generators have been omitted.

86

Results Figure 1 shows that wind power is able to supplant a large share of energy originally produced by conventional thermal generators.

2005

0% 3% 6% 22 % 60 % 79 % 6% 6% 29 % 1% 11 % 0%

2020

1% 11 % 0% 10 % 1%

2030

Conventional Thermal Hydro Wind

54 %

DHCP Nuclear Other

Figure 1: Percentage of yearly energy production by source.

The presence of an offshore grid does not have any dramatic effects on energy production for the system, though it is possible to conclude that the presence of an offshore grid may contribute to slightly shift the power system in favour of renewable energy sources. Wind power will cause a significant reduction in energy prices in all areas, resulting in reduced energy costs for the entire system. Adding the costs of the offshore grid connections to the total energy costs of each scenario gives an indication of the total social benefit of these wind power expansions and an eventual offshore grid. Analysis of lost wind and hydro power reveals the importance of sufficient transmission capacity when large quantities of wind power are added to the system. Scenario 4 features enormous quantities of lost hydro power in the North and Middle of Norway due to transmission limitations. Analysis of power transmissions reveal that the offshore grid used in the simulations overdimensioned. Rationalizing by reducing transmission capacities to more realistic levels will give a more cost-effective solution. This was demonstrated by performing a quick simulation and analysis of a scenario featuring such a rationalized grid. Wind power will cause more frequent variations in hydro power generation, due to balancing needs. Parts of the increased variability in the hydro generators can be explained by the increasing amount of wind power in the system, while other parts are most likely caused by limitations in the simulation model itself. Conclusion Given the number of assumptions made in the grid, in cost calculations and in the model at large, it more important to focus on general trends than on concrete numerical values. However, it is clear that increasing the amount of on- and offshore wind power in the European power system will have a beneficial impact to society's energy costs. It is also clear that wind power has the potential to dramatically reduce CO2-emissions caused by power generation. The offshore grid seems to be more beneficial to the power producers than to consumers since it causes slightly higher energy prices and providing a measure of flexibility as to where offshore wind power production is sent. Wind power will present challenges, especially regarding transmission grid development. A sufficiently dimensioned grid will be paramount to the successful implementation of such amounts of wind power, both with respect to profitability and in order to avoid waste of potential wind or hydro energy.

87

88

Information

Microsoft PowerPoint - forside.ppt

90 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

27053


Notice: fwrite(): send of 197 bytes failed with errno=104 Connection reset by peer in /home/readbag.com/web/sphinxapi.php on line 531