UNIVERSIDAD POLITECNICA´ DE MADRID

ESCUELA TECNICA´ SUPERIOR DE INGENIEROS INDUSTRIALES

A voltage disturbances prototype for testing electrical generators connected to microgrids

TESIS DOCTORAL

Julia Merino Fern´andez Ingeniera Industrial

2015

DEPARTAMENTO DE AUTOMATICA,´ INGENIER´IA ELECTRICA´ Y ELECTRONICA´ EINFORMATICA´ INDUSTRIAL

ESCUELA TECNICA´ SUPERIOR DE INGENIEROS INDUSTRIALES

A voltage disturbances prototype for testing electrical generators connected to microgrids

Autora: Julia Merino Fern´andez

Ingeniera Industrial

Director: Dr. Carlos Veganzones Nicol´as

Dr. Ingeniero Industrial

Director: Dr. Francisco Bl´azquez Garc´ıa

Dr. Ingeniero Industrial

2015

Abstract

A VOLTAGE DISTURBANCES PROTOTYPE FOR TESTING ELECTRICAL GENERATORS CONNECTED TO MICROGRIDS

By

Julia Merino Fern´andez

The search for new energy models arises as a necessity to have a sustainable power supply. The inclusion of distributed generation sources (DG) allows to reduce the cost of facilities, increase the security of the grid or alleviate problems of congestion through the redistribution of power flows. In remote microgrids it is needed in a particular way a safe and reliable supply, which can cover the demand for a low cost; due to this, distributed generation is an alternative that is being widely introduced in these grids.

But the remote microgrids are especially weak grids because of their small size, low voltage level, reduced network mesh and distribution lines with a high ratio R/X. This ratio affects the coupling between grid voltages and phase shifts, and stability becomes an issue of greater importance than in interconnected systems. To ensure the appropriate behavior of generation sources inserted in remote microgrids -and, in general, any electrical equipment-, it is essential to have devices for testing and certification. These devices must, not only faithfully reproduce disturbances occurring in remote microgrids, but also to behave against the equipment under test (EUT) as a real weak grid. This also makes the device commercially competitive. To meet these objectives and based on the aforementioned, it has been designed, built and tested a voltage disturbances generator, in order to provide a simple, versatile, full and easily scalable device to manufacturers and laboratories in the sector.

Resumen

A VOLTAGE DISTURBANCES PROTOTYPE FOR TESTING ELECTRICAL GENERATORS CONNECTED TO MICROGRIDS

Por

Julia Merino Fern´andez

La b´usqueda de nuevos modelos energ´eticos surge como una necesidad para conseguir un abastecimiento el´ectrico sostenible. La inclusi´on de generaci´on distribuida (GD) permite una reducci´on del coste de las instalaciones, incrementando los niveles de seguridad de la red y aliviando problemas de congesti´on a trav´es de la redistribuci´on de los flujos de potencia.

En las microrredes remotas es especialmente necesario conseguir un abastecimiento fiable y seguro, que permita cubrir la demanda con un coste bajo; por ello, la generaci´on distribuida se presenta como una alternativa que, de manera masiva, se est´a introduciendo en este tipo de redes.

Pero las microrredes remotas son redes particularmente d´ebiles debido a su peque˜no tama˜no, sus bajos niveles de tensi´on, el escaso mallado y la presencia de l´ıneas de distribuci´on con un alto ratio R/X. Este ratio afecta al acomplamiento entre las tensiones de las red y susangulos ´ y la estabilidad se convierte en un problema de mayor importancia que en los grandes sistemas interconectados. Para asegurar el comportamiento apropiado de las fuentes de generaci´on antes de su inserci´on en microrredes remotas y, en general, de cualquier equipo el´ectrico es necesario disponer de equipos de ensayo y certificaci´on. Estos equipos deben, no s´olo reproducir fielmente las perturbaciones que ocurren en las microrredes remotas, sino tambi´en comportarte frente al equipo a ensayar como una red d´ebil real. Para cumplir estos objetivos y basado en lo anterior, se ha dise˜nado, construido y validado un generador de perturbaciones de red, con el objetivo de proveer de un equipo simple, vers´atil, completo y f´acilmente escalable para los fabricantes y laboratorios del sector.

Let the future tell the truth and evaluate each one according to his work and accomplish- ments. The present is theirs; the future, for which I really worked, is mine.

NIKOLA TESLA

A mis padres Antonio y Mary Luz y a mi hermano Antonio. A mi t´ıa Julia, que hubiera presumido tanto de m´ı.

vi

ACKNOWLEDGMENTS

Y por fin lleg´oeld´ıa. Al volver la vista atr´as, no puedo evitar sentir cierta satisfacci´on de haber llegado a esta cima, cuyo ascenso se me ha hecho a veces tan empinado, y es justo dar gracias a los que me han ayudado a conseguirlo. Hoy tambi´en son ellos parte de este triunfo.

A mis directores. A Carlos Veganzones, por todo el tiempo que me ha dedicado y por las ense˜nanzas que me ha transmitido estos a˜nos. Por prestarme su idea para el desarrollo de este trabajo. Por la ayuda en la configuraci´on y revisi´on final del documento de tesis. A

Francisco Bl´azquez por sus sugerencias para el dise˜no de las m´aquinas el´ectricas.

A Sergio Mart´ınez, por su implicaci´on desinteresada en mi formaci´on de posgrado. Por mar- car la diferencia y ser el que siempre me consider´o una compa˜nera y no una alumna.

To the professors and colleagues I was lucky to meet during my experiences as visiting scholar in UW-Madison and MSU. To Prof. Giri Venkataramanan and Patricio. To Dr. Strangas,

Dr. Foster and the Spartans (Andrew, Cristi´an, Reemon and Jorge). Thank you for your warmth and your friendship. Thank you very much for making me feel, at least for some months, part of a research group.

A otros profesores de la UPM, que me permitieron colaborar y aprender con ellos en sus

Departamentos durante esta etapa doctoral y que me trataron siempre con tanto afecto.

Gracias J. Angel´ S´anchez y gracias, Pablo Reina.

Al Departamento de Ingenier´ıa El´ectrica de la E.T.S.I. Industriales y al profesorado y per- sonal que lo componen. A los t´ecnicos de M´aquinas El´ectricas, por su soporte (bancada arriba, bancada abajo...). A Elena, por compartir conmigo losultimos ´ meses de laboratorio.

A mis compa˜neros de oficina, que en esta recta final me han aguantado con paciencia y comprensi´on mientras hablaba, casi obsesivamente, de esta tesis.

viii A mis padres, mi hermano y mis amigos, por su incondicional apoyo. Por soportar mi des´animo cuando, por primera vez en mi vida, me vi obligada a “tirar la toalla” y desistir de seguir persiguiendo mi sue˜no universitario. Gracias por compartir la alegr´ıa de verme hoy aqu´ı, m´as feliz, m´as fuerte y con nuevas metas que conquistar.

Julia

2015

ix x CONTENTS

LIST OF TABLES ...... xvii

LIST OF FIGURES ...... xix

LIST OF SYMBOLS ...... xxiii

Chapter 1 Introduction ...... 1 1.1Motivation...... 1 1.2 Objectives and contributions ...... 3 1.3 Outline of the thesis structure ...... 4 1.4 Dissemination ...... 5

Chapter 2 State of the Art ...... 7 2.1Microgrids...... 7 2.1.1 Definition and structure ...... 7 2.1.2 Classification...... 10 2.2 Remote microgrids with distributed generation ...... 13 2.2.1 Singularities ...... 13 2.2.2 Characterizationofdisturbances...... 17 2.2.2.1 Voltage dips ...... 17 2.2.2.2 Frequencydisturbances...... 18 2.2.2.3 Frequency disturbances associated with voltage dips .... 19 2.2.3 Worldwidesituationofremotemicrogrids...... 21 2.3 Regulations applicable to distributed generation sources connected to electri- calsystems.Gridcodes...... 23 2.3.1 Introduction...... 23 2.3.2 Gridcodesstructure...... 24 2.3.3 Requirementsingridcodes.Basicconcepts...... 25 2.3.4 Codesreview...... 28 2.3.4.1 Gridcodesofinterconnectedsystems...... 28 2.3.4.1.1 Introduction...... 28 2.3.4.1.2 Voltagerequirements...... 29 2.3.4.1.3 Frequencyrequirements...... 31 2.3.4.1.4 Phaseshift...... 32 2.3.4.2 Gridcodesofremotemicrogrids...... 33 2.3.4.2.1 Introduction...... 33 2.3.4.2.2 Major international codes applicable to remote sys- tems...... 33 2.3.4.3 Towardsharmonizedstandards...... 36 2.3.4.3.1 Introduction...... 36 2.3.4.3.2 ENTSO-Egridcode...... 37 2.3.4.3.3 Highlights of ENTSO-E grid code ...... 38 2.3.4.3.4 ENTSO-Eandremotemicrogrids...... 41 2.4Griddisturbancesgeneratorstotestelectricaldevices...... 42 2.4.1 Impedance-based...... 43 2.4.2 Electrical machines based ...... 44 2.4.3 Full-converter based ...... 45 2.4.4 Comparative between commercial topologies and the proposed device 46 2.5Conclusions...... 47

Chapter 3 General description and operating model of the voltage distur- bances generator prototype ...... 49 3.1 General description of the device ...... 49 3.1.1 The electrical machines ...... 51 3.1.1.1 Variablefrequencytransformer(VFT)...... 51 3.1.1.2 The induction regulator ...... 51 3.1.2 Thecontrolprogram...... 55 3.1.2.1 Cascadecontroldesign...... 55 3.1.2.2 ProgramI/O...... 56 3.1.2.3 Implementation in MATLAB/SimulinkR ...... 57 3.2 Mathematical expressions ...... 58 3.2.1 Variablefrequencytransformer-EM1...... 59 3.2.1.1 Steady-stateequationsandequivalentcircuit...... 59 3.2.2 Inductionregulator-EM2...... 62 3.2.2.1 Steady-stateequationsandequivalentcircuit...... 62 3.2.3 Theveninequivalents...... 64 3.2.3.1 Voltagefrequencytransformer-EM1...... 64 3.2.3.2 Induction regulator - EM2 ...... 65 3.2.4 Simulationmodel...... 66

xii 3.2.5 Transientanalysis...... 71 3.3Conclusions...... 75

Chapter 4 Experimental setup, operational procedure and performance evaluation ...... 77 4.1Experimentalsetup...... 77 4.1.1 The electrical machines ...... 77 4.1.2 Controlandmonitoringsystem...... 78 4.1.3 The electric cabinet ...... 82 4.2Operationalprocedure...... 83 4.2.1 Setupofthevoltagedisturbancesgenerator...... 83 4.2.2 Preparationofthetest...... 84 4.2.2.1 Calibration ...... 84 4.2.2.2 PI controller tuning ...... 85 4.2.3 Testprocedure...... 86 4.3Performanceevaluationoftheprototype...... 87 4.3.1 Voltagedisturbances...... 87 4.3.2 Frequencydisturbances...... 89 4.3.3 Combined disturbances ...... 90 4.3.4 Harmonicbehaviour...... 91 4.4Conclusions...... 92

Chapter 5 Guidelines for the design of new prototypes ...... 93 5.1Problemapproach...... 93 5.1.1 Overview...... 93 5.1.2 Voltage drop evaluation of the preliminary design ...... 95 5.2 Minimization of the voltage drop in the induction regulator EM2 ...... 98 5.3Criteriadesignfornewprototypes...... 100 5.3.1 Prototype 2: Leakage reduction by redesigning the magnetic circuit . 100 5.3.1.1 Mutual inductance ...... 100 5.3.1.2 Leakage inductances ...... 101 5.3.1.2.1 Flux leakage components in wound rotor induction machines ...... 101 5.3.1.2.2 Mathematical expressions of the leakage inductances inEM2...... 103 5.3.1.2.3 Validation of leakage inductances in the prototype 1 106 5.3.1.3 Constantsandvariablesinthemagneticdesign...... 108 5.3.2 Prototype 3: Leakage reduction by redesigning the electric circuit . . 112

xiii 5.3.2.1 Optimization of the ampere-turns ratio ...... 112 5.4Conclusions...... 113

Chapter 6 Sizing, construction and assembly of the new prototypes .... 115 6.1 The finite-element methods (FEM) for electrical machines design ...... 115 6.2 Definition of new prototypes sizing by finite element analysis ...... 117 6.2.1 1.Slotdesign...... 117 6.2.2 2.Airgap...... 120 6.2.3 3. Transformation ratio & double-layer rotor winding ...... 122 6.3SimulationoftransientresponsebyFEManalysis...... 124 6.4 Construction and assembly ...... 126 6.5Conclusions...... 129

Chapter 7 Experimental results and comparative analysis of the prototypes131 7.1 Determination of the equivalent circuit parameters for new prototypes .... 131 7.2Experimentalresults.Comparativeanalysis...... 133 7.2.1 No-loadtest...... 133 7.2.2 Loadedtests...... 135 7.2.2.1 Testagainstapassiveload...... 135 7.2.2.2 Testagainstanactiveload...... 136 7.2.2.3 Testagainstgenerators...... 138 7.2.2.3.1 Asynchronousgenerator...... 138 7.2.2.3.2 Synchronousgenerator...... 140 7.3Conclusions...... 142

Chapter 8 Final conclusions and future work ...... 145

Appendix A Technical Data of the Experimental Setup Equipment ..... 149 A.1ServoMotors...... 149 A.1.1 Servo motor for EM1 machine ...... 150 A.1.2 Servo motor for EM2 machine ...... 150 A.2Drives...... 151 A.2.1AngledriveforEM1...... 151 A.2.2SpeeddriveforEM2...... 152 A.3Encoders...... 153 A.3.1AbsolutmagneticencoderforEM1...... 153 A.3.2IncrementalencoderforEM2...... 153 A.4DSPsdSPACE...... 154

xiv Appendix B Thevenin equivalent for the EM1+EM2 case ...... 157 B.1Theveninsource...... 158 B.2Theveninimpedance...... 158

Appendix C Calculation of Laplace residuals for transient regime solutions 161 C.1 Rotor current ir(s) ...... 161 C.2 Stator current is(s)...... 162

Appendix D Analytical calculation of slot leakage components in EM2 ... 165 D.1 Slot calculation ...... 165 D.1.1Rotorslotleakagecalculation...... 165 D.1.2Statorslotleakagecalculation...... 173 D.1.2.1Slotpermeanceduetothelowercoil...... 173 D.1.2.2Slotpermeanceduetotheuppercoil...... 173 D.1.2.3 Slot permeance due to the mutual effects between coils . . . 174

Appendix E Windings and steel sheets drawings ...... 177 E.1Magneticmaterial...... 177 E.2 Windings ...... 178 E.2.1 Rotor Windings ...... 178 E.2.2 Stator Winding ...... 178 E.3Electricsteelsheetsdrawings...... 179

BIBLIOGRAPHY ...... 182

xv xvi LIST OF TABLES

Table 2.1 PGM installed capacity limits according to ENTSO-E regions [1] . . 38

Table 2.2 Operating ranges of voltages. a) Voltages from 110 kV to 300 kV. b) Voltagesfrom300kVto400kV...... 40

Table2.3 PGMsfrequencyrequirements...... 41

Table 3.1 Electrical machine parameters ...... 66

Table4.1 Maincharacteristicsofprototype1...... 77

Table4.2 Basicdataofprototype1magneticcircuit...... 78

Table4.3 Basicdataofprototype1electricalcircuit...... 78

Table 5.1 Electrical machine parameters ...... 107

Table 6.1 Comparison between leakage reactances in prototype 1 and prototype 2 according to changes in the slot openings ...... 119

Table 6.2 Comparison between leakage reactances in original and improved pro- totype 2 according to the combined effect of the increment of the air-gapwidthplustheslotrefitting...... 122

Table6.3 Dataofprototype3electricalcircuit...... 123

Table 6.4 Comparison between leakage reactances in original and improved pro- totypes because of the change in the transformation ratio and the double-layer rotor winding design ...... 123

Table A.1 Data of Baum¨uller DS 56 S-3-R-K ...... 150

Table A.2 Motor data from Pujol Muntal´a IPCM 128/90L-4/148 ...... 151

Table A.3 Gearbox data from Pujol Muntal´a IPCM 128/90L-4/148 ...... 151

Table A.4 Main technical characteristics of Baum¨uller b maXX 4413 ...... 152

Table A.5 Main technical characteristics of Altivar 71 ...... 152 TableA.6 DataofRM36absolutencoder...... 153

Table A.7 Data of encoder Schneider XCC1506PS50X ...... 154

xviii LIST OF FIGURES

Figure2.1 Generalschemeofamicrogrid...... 8 Figure 2.2 Classification of microgrids according to the market segment . . . . 10 Figure 2.3 Distribution of microgrids due to the market segments [2] ...... 12 Figure 2.4 Requirements of energy reserves in a remote grid ...... 14 Figure 2.5 Number of unscheduled events recorded in La Palma island during 2008...... 16 Figure 2.6 Present and expected wind power generation in the isolated system ofGranCanaria...... 16 Figure 2.7 a) Voltage dip in the remote microgrid of El Hierro island; b) Voltage dipspreadinthegridofGranCanariaisland[3]...... 18 Figure2.8 Recordofafrequencydisturbanceinaremotemicrogrid[4]..... 19 Figure 2.9 Record of main electrical magnitudes in a dip in a 400 kV node . . . 20 Figure 2.10 Record of main electrical magnitudes in a dip in a spanish nonmain- landnode...... 21 Figure2.11Consequencesoftheuseofgridcodesinelectricalsystems...... 24 Figure 2.12 Classification of voltage disturbances according to IEEE Std. 1159-1995 26 Figure 2.13 Profile types of voltage dips. a) Rectangular profile. b) Profiles with recoveryramp...... 27 Figure2.14Voltagedipprofilesrequiredinmaininternationalcodes...... 29 Figure 2.15 Over/undervoltage and voltage swells requirements in main intercon- nectedgridcodes...... 31 Figure2.16Operatingfrequencyranges...... 32 Figure 2.17 a) Voltage dip profiles for microgrids up to 50MW. b) Voltage dip profiles for microgrids above 50MW ...... 34 Figure2.18Over/Undervoltagerequirementsinremotemicrogrids...... 35 Figure2.19Frequencyrequirements...... 36 Figure 2.20 Voltage dip profiles in ENTSO-E code. a) SPGMs type B and C; b) PPMs type B and C; c) SPGMs type D; d) PPM type D ...... 39 Figure2.21Impedance-basedvoltagedipgenerator...... 44 Figure2.22Transformer-basedvoltagedipgenerator...... 45 Figure2.23Voltagedipgeneratorbasedonfullpowerconverter...... 46

Figure3.1 Generalschemeofthedevice...... 50 Figure 3.2 Single-phase scheme of the wound rotor induction machine connection 52 Figure 3.3 Single-phase scheme of the induction regulator connection ...... 53 Figure3.4 Voltagephasordiagram...... 54 Figure3.5 ControlloopsforEM1andEM2...... 55 Figure3.6 ModelI/O...... 56

Figure 3.7 MATLAB/SimulinkR control...... 57 Figure 3.8 Single-phase representation of the EM1 machine ...... 59

Figure 3.9 Single-phase representation of the EM1 machine reduced to fr frequency 60 Figure 3.10 Single-phase representation of the EM1 machine reduced to rotor winding ...... 61 Figure3.11Single-phasecircuitofEM2...... 62 Figure3.12EquivalentcircuitofEM2...... 63 Figure3.13Simulationmodelofthevoltagedisturbancesgenerator...... 67 Figure 3.14 Steady-state comparative between simulation and experimental results 67 Figure 3.15 Steady-state comparative between simulation and experimental results 68 Figure 3.16 Percentage errors between simulation and experimentation results . 69 Figure 3.17 Comparative between the simulation model and the experimental re- sultsintherecoveryramp...... 70 Figure 3.18 Equivalent circuit representation for transient analysis. a) Time- domain.b)Laplacedomain...... 72

Figure4.1 Schemeoftheelectricdrivesystem...... 79 Figure4.2 HMIofthevoltagedisturbancesprototype...... 81

xx Figure 4.3 Voltage disturbances generator cabinet ...... 82 Figure 4.4 Voltage magnitude output during calibration ...... 85 Figure 4.5 Reproduction of voltage swells and dips ...... 87 Figure4.6 Anglesetpointandtracking...... 89 Figure 4.7 Voltage dips profiles ...... 89 Figure4.8 Frequencydisturbances...... 90 Figure 4.9 Combined voltage, frequency and phase jump disturbance ...... 91 Figure4.10Harmonicspectraofvoltagewaveforms...... 91

Figure 5.1 No-load and full load voltage dips with the prototype 1 ...... 95 Figure5.2 Phasordiagramoftheprototype1...... 96 Figure 5.3 Output voltage at the induction regulator for a generator with cosφ =1 97 Figure5.4 Slotleakageflux...... 102 Figure 5.5 Leakage flux paths in the end-windings ...... 102 Figure5.6 Zigzagleakagefluxpaths...... 102 Figure 5.7 End-winding main dimensions ...... 104 Figure 5.8 Influence of the geometric parameters over the slot permeances . . . 111

Figure6.1 RotorLeakageReactancecontourplot(Ω)...... 118 Figure6.2 StatorLeakageReactancecontourplot(Ω)...... 119 Figure6.3 Cartercoefficientvariationduetorotorslotting...... 120 Figure 6.4 Variation of main parameters affected by air gap width ...... 121 Figure 6.5 Magnetic flux density in prototype 3 before and during voltage dip reproduction...... 125 Figure 6.6 Rotor and stator laminations for new prototype designs ...... 126 Figure 6.7 Electrical machines with taps to be used as EM1 ...... 127 Figure6.8 Mainstepsinprototypesconstruction...... 128

Figure7.1 Comparisonofparametersbetweentheprototypes...... 132

xxi Figure 7.2 Comparison of voltage dips for the prototypes at no-load ...... 134 Figure 7.3 Comparison of voltage dips against resistive load at full rated power 135 Figure7.4 Asynchronousmotor/generatortestbench...... 136 Figure7.5 Comparativebehaviouroftestsagainstasynchronousmotor..... 137 Figure7.6 Comparativebehaviouragainstasynchronousgenerator...... 139 Figure 7.7 DC machine and synchronous generator test bench ...... 140 Figure 7.8 Comparative of the behaviour of prototypes connected to synchronous generatorswithstaticexcitation...... 141 Figure 7.9 Comparative of the behaviour of prototypes connected to synchronous generatorswithstaticexcitation...... 142

Figure A.1 Servo motors involved in the disturbances generator prototype . . . 149 Figure A.2 Block diagram of dSPACE DS1104 R&D ...... 155

Figure B.1 Single-phase representation of the device ...... 157 Figure B.2 Single-phase representation of the device ...... 159

FigureD.1 Rotorslot...... 166 FigureD.2 Fractionofrotorslot...... 168 FigureD.3 Changeoftheturnsratiodependingontheslotdepth...... 169 FigureD.4 Region3oftherotorslot...... 170 Figure D.5 Stator slot geometric dimensions ...... 174

Figure E.1 BH curve of V600-50A magnetic steel ...... 177 Figure E.2 Rotor winding configuration for prototypes 0 and 1...... 178 Figure E.3 Double-layer lap winding design for stators ...... 179 FigureE.4 Designforprototype0 ...... 180 FigureE.5 Designforprototypes1&2 ...... 181

xxii LIST OF SYMBOLS

α Electrical angle between rotor and stator windings (deg)

ω Speed (p.u.)

φm Mutual flux σ Factor for harmonic leakage reactance calculation

τ Slot pitch

τp Pole pitch

Dor Rotor outer diameter

Ir Rotor current

Is Stator current K Transformation ratio

Uo Output voltage phasor

Ur Rotor voltage phasor

Us Stator voltage phasor

Xσ Leakage reactance

Xm Mutual reactance between rotor and stator referred to rotor side Z Impedance b Slots widths few Permeance factor associated to axial length fW Permeance factor associated to the coil span g Air gap width ge Effective gap considering effects of fringing and saturation H Inertia constant (s) h Slots heights Lew End-winding leakage inductance per unit of length Lh Harmonic leakage inductance per unit of length Lslot Slot leakage inductance per unit of length Lzz Zig-zag leakage inductance

Lew End-winding length P Poles p Permeance

Pe Sum of the power of the loads (p.u.)

Pm Sum of the power of the generators (p.u.) q Slots per pole and phase in rotor

R Resistance

S Number of slots s Slip

Wew End-winding width ξ Winding factor th ξh Winding factor for the h harmonic kb Back-calculation constant of the PI controller ki Integral constant of the PI controller kp Proportional constant of the PI controller kw Weighting constant of the PI controller m Number of phases

N Number of turns per phase n Number of turns per coil

Note: Additional subscripts r and s along the content refer to rotor and stator values

xxiv Chapter 1

Introduction

1.1 Motivation

The current power consumption involves the emission to the atmosphere of great amounts of

CO2 contributing to accelerated climate change. During the year 2012, 31.6 Gt of CO2 were emitted, of which 60% belonged to countries outside the OCDE (mainly to the emergent economies of China, India and Brazil). Due mainly to the growth of these countries, it is expected that by 2020 the world electricity consumption will be almost 30% higher than at present [5]. This evidences a crisis in the traditional energy model, posing serious problems of sustainability in the medium and long term. If we add that fossil fuels are finite and scarce, which in turn has untied a price war on the markets, it is clear the need to find new models for energy supply.

On the approach of new models two main objectives are pursued: the replacement of the conventional generation by renewable energy sources (RES) and the improvement of overall efficiency. Distributed generation (DG) is presented as the most appropriate alternative in order to achieve these goals. DG favors the inclusion of non-conventional generation, allowing the installation of groups located close to the natural resources, at distribution voltage levels.

It also allows an efficiency rise by bringing the generation towards the consumption points to reduce losses in the transportation of electrical energy.

The massive increasing of wind power plants in electrical systems has forced the deve- lopment of a more complete and rigorous regulation in terms of connection and operation requirements for electrical generators. From this new perspective, traditional electrical sys- tems are gradually fragmenting into smaller ones, with a high percentage of renewable en- ergy installed. Decentralized generation helps to improve grid security, because the supply no longer depends on a few critical nodes. The ultimate goal is have a grid formed by a set of smaller systems called microgrids, connected to each other, self-supplied and with the capacity of operating in isolated mode in the event they are required.

The needed electricity supply to remote areas, as islands or faraway locations in devel- oping countries provides a very particular configuration for their electrical systems. They are a subgroup of special microgrids -remote microgrids- that have a very weak link inter- connection or even a totally lacking link. In remote microgrids, low levels of short-circuit power, the weakness of the network, low inertia of the groups and accused variations in the load profile as main characteristics, make that any abnormal situation ocuring at a point can affect the whole of the network and can seriously compromise the security and stability of the system.

It is worth remarking that the availability of a strong link between a microgrid and the local electric distribution system allows continuity of supply in the microgrid in case of disconnection of several generators due to maintenance or fault. The absence of this option requires to establish higher levels of reserve groups to deal with contingencies and to adopt additional procedures in the management and operation protocols.

The remote microgrids are especially suitable for the inclusion of renewable energy but they are more vulnerable than any other type of microgrid. Hence the technical requirements in regulations applicable to these systems are generally more demanding. It is essential for manufacturers, electrical distributors and certifiers to have flexible equipment, useful to validate the behavior of their devices according to new grid codes requirements.

2 This thesis discusses the design and construction of a new voltage disturbances prototype, in order to be employed in the validation and certifying compliance with the grid codes. The requirements will be those currently in use or those that will foreseeably appear on new standards. In the previous literature search of this work, detailed in cha. 2, no device has been found that shares the characteristics and objectives of the one developed, so this new equipment covers a technological gap.

1.2 Objectives and contributions

The main objective pursued by this thesis involves the design, construction, programming, testing and validation of a voltage disturbances prototype. This equipment allows the cre- ation of frequency and voltage disturbances as well as phase jumps, so it can be used for the certification of electrical equipment prior to its inclusion in particularly weak electrical networks, as remote microgrids.

This main objetive can be furthered splitted into three intermediate targets that are listed below:

• Design of two electrical machines, improved from a commercial wound rotor induction

motor intended for operation as part of this prototype, with the aim of increasing

the performance of the device. These machines have modifications on some design

parameters in order to reduce, to a certain extent, the effect of current flow through

the prototype that can have on the disturbance to be created or on the equipment to

be tested.

• Programming of the real-time control system that allows the accurate reproduction of

the desired disturbances and whose main features are precision, speed of response and

flexibility.

3 • Implementation of a data acquisition system for control and monitoring of the variables

of interest, like the development of a human machine interface (HMI) to allow the user

an easy and intuitive handling of the test equipment.

The result of the fulfillment of these goals involves a contribution of knowledge to the research area that frames this work.

1.3 Outline of the thesis structure

Chapter 2 gathers the information concerning to the state of the art in the main topics constituting a framework of the work. It has been divided in three blocks of contents. First, main characteristics defining microgrids and definition of disturbances in these systems are studied. Second, it has been collected a wide study about grid codes: structure, requirements and main regulations for isolated grids as well as interconnected networks. And third, a literature search of commercial devices to reproduce voltage dips as test equipment have been compiled.

In Chapter 3 the conceptual idea underlying the device and the operational procedure of the equipment are drafted. The first performance results for the prototype built from commercial machines are also collected.

Chapter 4 develops the mathematical expressions, the Thevenin equivalents and the simulation model to represent the steady-state behaviour of the prototype and the equations to define the transient regime have been explored.

Chapter 5 clusters the main information concerning to the electrical machines design. The weak points of the commercial setup are analyzed and some improvements for the design of future prototypes are proposed.

4 Chapter 6 focuses on the final definition of new electrical machines. It also shows the transient model developed by finite elements and describes the process for construction and assembly of enhanced prototypes.

Chapter 7 shows the characterization of new prototypes by means of testing, the experi- mental results reached and the comparison between original prototypes and enhanced ones, in order to quantify the effectiveness of design changes applied to the electrical machines.

Chapter 8 collects the final conclusions of the dissertation and settles the basis for future improvements.

1.4 Dissemination

Papers in JCR journals:

• Merino, J.; Mendoza-Araya, P.; Venkataramanan, G.; Baysal, M.; Islanding Detection

in Microgrids using Ambient Harmonics. Power Delivery, IEEE Transactions on,vol.

PP, no. 99, pp.1,1, Dec. 2014.

• Merino, J.; Mendoza-Araya, P.; Veganzones, C. State of the Art and Future Trends in

Grid Codes Applicable to Isolated Electrical Systems. Energies, no. 7, pp. 7936-7954.

Nov. 2014.

• Merino, J.; Veganzones, C.; Sanchez, J.A.; Martinez, S.; Platero, C.A. Power System

Stability of a Small Sized Isolated Network Supplied by a Combined Wind-Pumped

Storage Generation System: A Case Study in the Canary Islands. Energies, no. 5, pp.

2351-2369. July 2012.

• Veganzones, C.; Sanchez,J.A,; Martinez, S.; Platero, C.A.; Blazquez, F.; Ramirez, D.:

Rodriguez, J.; Merino, J.; Herrero, J.; Gordillo. F. Voltage Dip Generator for Testing

5 Wind Turbines Connected to Electrical Networks. Renewable Energy, Vol. 36, no. 5,

pp. 1588-1594. May 2011.

Patent application:

• Rodr´ıguez, J.; Herrero, N.; Merino, J.; Platero, C.A.; Veganzones, C.; Bl´azquez, F.;

Mart´ınez, S.; S´anchez, J.A. “Generador de perturbaciones de tensi´on para ensayo de

equipos el´ectricos y su procedimiento de operaci´on en redes con generaci´on distribuida”.

WO Patent App. PCT/ES2014/070.099. Feb 25, 2015.

6 Chapter 2

State of the Art

This chapter collects a literature review of the topics on which this thesis focuses. First of all, microgrids are analyzed as a distinct structure within the electric power systems. Then, the particularities that define remote microgrids and the characteristics of disturbances occurring in them are studied. Later, it is examined the legislation that currently must meet generators

-especially wind turbines -before its insertion in electrical systems, and the tendency which is expected to be followed by new regulations. Eventually, is extensively analyzed the specific bibliography on voltage disturbances generators employed in the study or certification of DG systems. These devices are evaluated with the purpose of highlighting the contributions of the new device over other existing equipment.

2.1 Microgrids

2.1.1 Definition and structure

According to the definition given by the United States Department of Energy (DOE), a microgrid is an electrical system consisting of a set of generators, storage devices, loads and interconnection elements that has the ability to operate autonomously or grid-connected

[6]. Researchers at the Consortium for Electric Reliability Technology Solutions (CERTS) added, moreover, that a microgrid must be able to provide combined electrical and thermal energy to users [7]. The ability to work stably and safely in islanded and grid-connected modes, make mi- crogrids a highly flexible structures. They appear as the best alternative to promote the integration of DG sources in power systems. The microgrids have regulation systems over their generation units and loads so they provide, at every moment, a full and complete power flow control. From the utility grid side, a microgrid is an active electrical system which behaves like an individual generator or load at the point of common coupling (PCC).

In Fig. 2.1 the general scheme of a microgrid can be observed with the elements that typically comprise it, although it is not necessary for all of them to be present simultaneously.

The generators can be rotating electrical machines directly connected to the grid, thermal or hydraulic synchronous units, asynchronous machines or renewable energy sources connected to the utility grid through electronic converters -wind generators, PV generators, etc-. The storage systems such as batteries, flywheels or supercapacitors, help stabilize the grid in case of transient disturbances. They also contribute to restore the equilibrium when generation- load imbalances take place. The advantages provided by storage are especially needed when the microgrid is working in islanded mode.

GRID

PCC MICROSWITCH

M MOTORS

LOADS

G STORAGE DEVICES G CONVENTIONAL GENERATOR

MICROGRID Figure 2.1: General scheme of a microgrid

8 According to the standard IEEE P1547 [8], the power limit of a source to be considered as a DG source is 10 MVA and it has to be connected to a distribution grid. This unique generation/storage unit with their associated controls would already form a microgrid. This value must be taken as an indicative manner since, in practice, the inclusion of a grid in the microgrid group is based on topology and functional criteria.

As a summary of the different definitions in the literature, it could be concluded that there are four common characteristics that a grid must meet to be classified as a microgrid:

• Ability to supply combined electricity and heat.

• Physical proximity between generation and consumption centers.

• Stability in its operation in islanded and grid-connected modes and in the transitions

between them.

• Real-time control and management of power flows.

A microgrid, from the operational point of view, must fulfill two conditions. On the one hand, any component in the assembly can be critical, i.e., the stability and security of the grid has to be maintained regardless of the failure of any element of the system. This shows a similarity with the N-1 safety criterion applicable to the conventional electrical systems.

This first condition means necessarily that each generation/storage unit must have its own power-frequency controls and its associated protections. On the other hand, within the microgrid, any two units can be exchangeable without making necessary the redesign of the grid [9].

The isolated power systems can be considered as a particular case of microgrids, which do not have the possibility of connection to the mains. These microgrids are known as remote microgrids.

9 There are unresolved aspects that pose barriers to the bulk introduction of microgrids in electrical systems. From the technical perspective, those are linked to complex control of multiple generating sources, detection of abnormal islanding events and the subsequent re- connection process. In terms of regulations, there are shortcomings in the specific legislation in a large number of countries, such as Spain. There are difficulties in defining the connec- tion requirements, who are the agents responsible for the grid management and the trade of surplus power -if connected to the utility grid - and what will be the suitable remuneration regime.

2.1.2 Classification

The microgrids can be classified according to several criteria. The most common is to establish the organization in terms of demand characteristics established by the needs of the users or according to the market segment, as seen in Fig. 2.2:

Figure 2.2: Classification of microgrids according to the market segment

The institutional microgrids assemble a set of public buildings, and are intended to supply areas of administrative, medical or educational services. They have a specific design accor- ding to the required function and are State-owned. This aspect facilitates the technical

10 management of the grid, makes it easily centralized and dependent from a single adminis- trator in generation and demand. This also happens with economic management, since the resulting benefits have a direct impact on citizens. A fundamental aspect of their operation is the need to ensure the continuity of supply especially in buildings, such as, for example, hospitals.

The industrial or commercial microgrids are, in the essence of their structure, similar to the institutional, so that in many articles appear as a single group [10, 11, 12]. Technical management of the grid is more complicated since various stakeholders are involved, for example, in the case of a commercial complex where there are several types of businesses with different administration models.

The military microgrids tend to have a strong support of renewable generation sources, since they are in remote areas and they often operate isolated from the mains. Most of them are temporary facilities. In these grids the reliability factor is a priority over the quality of the service.

The community microgrids are intended to supply residential areas. These microgrids will be an important part of its operation time connected to a larger grid. Being completely private properties, its development will generally depend on the existence of advantageous specific legislation for owners. They can serve to palliate congestion problems that exist in the low- or medium-voltage grids. A planned disconnection of these microgrids from the utility grid can avoid load shedding in case of an abnormal situation.

And the last group are the remote microgrids, of special interest in this thesis. Its structure is similar to the community or commercial microgrids. They present the main difference that unless further developments of the infrastructures, they will never operate connected to the mains. Compared to the military microgrids, that also operate in islanded

11 mode, the remote microgrids are designed to fulfill additional requirements of reliability, efficiency and quality of service, because they are intended for permanent installations.

Reduced Off-grid energy 2012: (5%, 110 MW) cost 2020: 16% CAGR

Commercial & Industrial 2012: (25%, 510 MW) 2020: 17% CAGR Communies & Ulity 2012: (20%, 420 MW) 2020: 12% CAGR

Instuonal & Campus 2012: (42%, 870 MW) 2020: 11% CAGR Military 2012: (9%, 180 MW) 2020: 17% CAGR

Reduced emissions Improvedd PowerP Quality / Reliability Figure 2.3: Distribution of microgrids due to the market segments [2]

The idea of remote microgrid is not novel itself. Many rural areas had for years a electri-

fication grid that according to the the current standards would come within the definition of microgrid. It is relatively recent the discovery of the technical and economic advantages they offer and the establishment of common criteria that are the basis for its development. Com- pared to very extended models of microgrids, the remote microgrids have a wide progression ahead.

In the bubble chart in Fig. 2.3 it is observed the current (2012) and planned distribution

(2020) of microgrids according to the segment market. In the figure it can be appreciated that remote microgrids are the least polluting, by the high proportion of renewable energy installed, and those which get a cheaper energy supply. At the other end they are the military microgrids. In them the most important is the security of supply in exchange for high

12 investment costs which would be unacceptable in other contexts. For the remote microgrids it is expected a compound annual growth rate (CAGR) of 16% up to 2020. Institutional and commercial microgrids are the most implemented (almost 70% of them belong to these two groups). They use a smaller proportion of renewable energy than the community or remote microgrids, because the security of supply required forces, in some cases, to the use of back-up diesel units. This reserve of energy required to face up with potential failures of the microgrid increases the final cost of the installation.

2.2 Remote microgrids with distributed generation

2.2.1 Singularities

Remote microgrids are electrically isolated systems, weak networks that are in geographically far-off locations -especially on islands or small populations of developing countries-. They currently remain largely dependent on the use of fossil fuels. Until recently, only diesel generators were able to ensure a secure and reliable supply in exchange for very high prices in the transportation of fuel and in the operation of the system. They are in areas of high ecological interest, which forces to search for the trade-off between energy efficiency and environmental friendliness. Existing solutions involve a mixed supply, maintaining the support of conventional generation groups with the advantages provided by RES.

As remote microgrids do not have connection to the mains, the system has to be designed to autonomously meet the criteria of reliability and quality of service. In addition, a high percentage of the supply depends on renewable energy, i.e. intermittent energy sources is necessary to establish new protocols for management of such microgrids. Below, three areas that make clear the need of operation procedures that gather the incidences derived from the management of these grids are highlighted:

13 • The reserves managment

• The the security criteria

• The generation and demand forecasting

It becomes clear that the operation is strongly influenced by the impact that contingen- cies have on the system. This impact is much higher in remote microgrids compared with interconnected systems, making necessary the establishment of higher reserve levels. The conventional groups are forced to work below their rated power, leading to cost overruns.

This reserve value tends to be the minimum allowed by the country’s legislation and it is often insufficient. Due to this, in isolated systems or remote microgrids, it is usual that in case of an unscheduled failure, load has to be shed in order to retrieve the normal operating condition.

To emphasize the importance of the need for a proper management of the energy reserves in an isolated system it is described, as an example, a contingency registered on the Spanish island of Gran Canaria on November 7, 2006 (Fig. 2.4). Electricity Demand (MW)

+ 75 MW Wind power (MW) Wind power - 40 MW

Source: REE Figure 2.4: Requirements of energy reserves in a remote grid

14 It can be observed that in the period between 16.48h and 18.40h -less than two hours- there was a simultaneous increase in demand for 75 MW and a sudden decrease of wind power generation of about 40 MW. The wind power lost had to be compensated by the reserve planned in the system, assigned to the back-up conventional generation.

In remote microgrids it is difficult to foresee any incidents as they possess a high degree of uncertainty due to the grid variability of voltage levels. This makes difficult the compliance regards to security criteria. The lines offer a high impedance which causes a voltage drop highly dependent on the degree of load, seriously influencing the power quality in the PCC.

In systems that have not yet implemented DG, there is a risk of operation because the electrical generation is condensed in a few nodes. They are also poorly meshed grids with low voltage level (< 66kV ). For all the above, in these grids it is usually registered a high number of breaches of the N-1 safety criteria. As an example, in the Fig. 2.5 data of the unscheduled events which were recorded during the year 2008 in the Canary Island of La

Palma are shown [13]. The variability in the number of failures that occur in the system is very high, which complicates the management of the microgrid.

For a proper operation of the remote microgrids it is needed some improvements in the forecasting systems of the production of the non-dispatchable resources and the demand. It is also required a good estimation of the ratio between conventional and renewable generation necessary to ensure stability.

In each graph shown in Fig. 2.6 appears, simultaneously, the value of wind power gene- rated (in MW) respect to the predicted value by the estimation model and for different time horizons. The gap between the forecast and the actual generation reaches maximum errors of up to 80% with 24h in advance.

15 Unscheduled events (2008) 45 40 39 35 30 25 20 15 14 9 10 8 6 5 2 2 2 2 2 0 0 0

Figure 2.5: Number of unscheduled events recorded in La Palma island during 2008.

Real Max. 30.73 MW at 16.00 p.m. Real Min. 1.13 MW at 10.00 a.m. Est. Max. 17.78 MW at 13.00 p.m. Est. Min. 3.56 MW at 1.00 a.m. (24h before)

Est. Max. 13.87 MW at 15.00 p.m. Est. Min. 3.56 MW at 1.00 a.m. (6h before)

Est. Max. 37 MW at 15.00 p.m. Est. Min. 1.32 MW at 9.00 a.m. (1h before)

Source: REE

Figure 2.6: Present and expected wind power generation in the isolated system of Gran Canaria

16 2.2.2 Characterization of disturbances

2.2.2.1 Voltage dips

The contingency that most affects the power quality is the voltage dip since it represents almost 80% of short-term disturbances in the grid [14].

A voltage dip, as defined in the reference [15], is a sudden decrease in the voltage at a grid node followed by their subsequent recovery and whose duration is between 10 ms and 1 min.

There are many causes that can lead to the appearance of a voltage dip. These causes are related to the connection or disconnection of elements in the system, either for operational reasons or as a result of short-circuits. The remote microgrids are usually reduced power, nodes are physically close together, connected through very short distribution lines and are electrically equivalent points. In addition, the relative power of the groups of generation/load with respect to the total power of the system is important.

These circumstances determine the characteristics that usually present voltage dips in remote microgrids: they are very deep and their propagation area is very broad. As an example, in Fig. 2.7a, it is shown the temporal profile of a voltage dip characteristic of those produced in the distribution grid of 20 kV in El Hierro island. It is noted that the voltage in several nodes of the grid is almost the same because of the small size of the microgrid.

In addition, in Fig. 2.7b it is displayed the spread of a voltage dip in the isolated electrical system of Gran Canaria island. Even the latter is a relatively large electrical system within the remote microgrids, a short circuit at the output of one of the major power plants affects the entire electrical system (of 66 kV). In these conditions no node is able to stay within the allowable voltage limits of the system.

17 1.25 MUELLE GRANDE

0.8 GUIA GUANARTEME ARUCAS 1 BUENAVISTA LOMO APOLINARIO 070.7

C.T. JINAMAR BARRANCO SECO 0.75 0.6 MARZAGAN

SAN MATEO TELDE 0.5 0.5 CINSA 0.4

Voltage (p.u) Voltage CARRIZAL 0.25 0.3

0.2 0 ALDEA BLANCA 0 2.5 5 7.5 10 MATORRAL Time (s) 0.1

C.T. BCO. TIRAJANA ARGUINEGUIN SAN AGUSTIN 0

LOMO MASPALOMAS CEMENTOS ESPECIALES

(a) (b)

Figure 2.7: a) Voltage dip in the remote microgrid of El Hierro island; b) Voltage dip spread in the grid of Gran Canaria island [3]

2.2.2.2 Frequency disturbances

In any electrical system, often frequency disturbances appear as the result of imbalances between generation and load. The swing equation of the rotor of electrical machines, in

(p.u.) over the base power is shown in Eq. 2.1:

dω P − P =2· H · (2.1) m e dt

where

Pm = Sum of the power of the generators (p.u.)

Pe = Sum of the power of the loads (p.u.) H = Inertia constant (s)

ω =Speed(p.u.)

From the previous equation it follows that the change in the rotational speed of the generator, and therefore the frequency of the network, is directly proportional to the in- stantaneous imbalance between generation and demand and inversely proportional to the sum of the inertia provided by the groups. In the remote microgrids there are mostly diesel

18 groups of low inertia. This means that in case of a disturbance, the frequency deviations are noticeably greater than those that would occur in an interconnected system.

f (Hz)

50

49.75

49.50

49.25

0 50 100 150 200 250 300 t/s

Source: REE Figure 2.8: Record of a frequency disturbance in a remote microgrid [4]

Figure 2.8 shows a typical frequency disturbance pattern in a remote microgrid, which corresponds to an event on the island of Tenerife on April 7, 2011. The trip of one of the larger generation units caused a sudden drop in frequency. Since the frequency kept outside of normal operation ranges after an extended time (150 s), the System Operator shed interruptible load, in order to accelerate grid recovery.

2.2.2.3 Frequency disturbances associated with voltage dips

The changes that happen on the voltage and frequency magnitudes during a disturbance are not independent from each other.

In interconnected grids a sudden voltage dip creates small frequency deviations. As an example, in Fig. 2.9 it is displayed the records of main electrical magnitudes in a 400 kV node of the Spanish grid when a severe fault takes place (a three-phase short circuit).

19 Trip 07/05/2009 22:27:05.253

I/ A K1:Current phase A Baixas K1:K1:Current phase B Baixas K1:Current phase C Baixas

0 -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 t/s

-10000

U/ V K1:Voltage phase A Baixas K1:Voltage phase B Baixas K1:Voltage phase C Baixas 300000 200000 100000 t/s -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 -100000 -200000 -300000

I/ A K2:Current phase A Pierola K2:Current phase B Pierola K2:Current phase C Pierola

5000

t/s -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0

-5000

U/ V K2:Voltage phase A Pierola K2:Voltage phase B Pierola K2:Voltage phase C Pierola 300000 200000 100000 t/s -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 -100000 -200000 -300000

K1:Frequency bar 1A / Hz

50.5

50.0

49.5

t/s -0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 49.0 Figure 2.9: Record of main electrical magnitudes in a dip in a 400 kV node

The frequency disturbance associated with a harsh voltage dip in the Spanish mainland grid does not reach the value of 0.1 Hz. Moreover, in Fig. 2.10 it can be observed a voltage dip recorded in a node of the Spanish nonmainland territory. The fault provokes a very deep voltage dip (4% of residual voltage) and a frequency change in the node of about 2 Hz. The comparison of these logs allows to check the statements of the preceding Subsections 2.2.2.1 and 2.2.2.2, i.e., voltage dips are deeper in remote microgrids and frequency deviations are more pronounced than in interconnected systems.

20 In the interest of this thesis concerns,it has to be highlighted the interdependence between voltage and frequency which occurs in remote microgrids. This needs the establishment of devices that can jointly represent both disturbances, such as the one developed in this work.

T1: -0:00.15 T2: 0:02.86 TD:0:03.01 min:seg 51.820 x43 Frec: Ten.UST Bar. A1 49.832 51.816 1.984 Hz 47.571 17184 x6 Ten.URS Bar. A1 14978 15027 48.07 V 3869.9 17165 x6 Ten.UST Bar. A1 14972 15027 54.93 V 3961.9 17140 x6 Ten.UTR Bar. A1 14991 15031 39.83 V 3867.2

Pot Act G11 8783.6 5644.0 -3139.6- KW

Pot React G11

-2448.3- -4385.3- -1937.0- KVA

-0:00.50 0:00.00 0:00.50 0:01.00 0:01.50 0:02.00 0:02.50 0:03.00 0:03.50 0:04.00 0:04.50 0:05.00 min:seg Figure 2.10: Record of main electrical magnitudes in a dip in a spanish nonmainland node

2.2.3 Worldwide situation of remote microgrids

The progress in the development of submarine interconnections using HVDC/HVAC links is making it possible that many of the systems that were remote -islands- have been progres- sively connected to other grids with similar characteristics or the Mainland. As an example of the first cases it may be mentioned the North and South islands of New Zealand, Lan- zarote and Fuerteventura or the main islands of Hawaii. In this sense, there is also projected

21 an interconnection between the islands of Guadeloupe, Martinique and Dominica [16]. As an example of the second situation, as islands connected to Mainland, it is intended a connec- tion of the Italian Islands (Island Project) to the continental Italy [17] as it already occurs in the interconnection of Sri Lanka to India [18].

Also, within this second case, it is important to draw attention to two projects currently being implemented due to its scale and the technological challenge they pose. The first one is the Eurasia Interconnector Project, which is developing a link of almost 1000 km long, with a capacity of 2000 MW, which will join the microgrids of Cyprus and Crete to the continental territory of Greece and Israel. Its entry into service is scheduled for 2016 [19].

The second project, not yet started, will connect Iceland to the United Kingdom. Iceland has a very important geothermal, hydraulic and wind potentials, so it is able to generate nearly five times the energy it consumes at a very low price. The United Kingdom would take the economic advantage of importing cheap energy from Iceland and, consequently, it would not need to expand its offshore wind generation capacity [20].

For other remote microgrids their interconnection cannot be planned either because of the high cost of the construction of the link (not affordable for the remote places in developing countries) or because of the geographical location. This occurs, for example, in the Canary

Islands: the distance between the islands and the Mainland and the depth of ocean floor do not make viable the efficient connection at an affordable cost. As solutions for the future, although unlikely, have been raised inter-connections of islands by means of offshore wind farms with HVDC links [21]. In this type of grids, without the possibility of later connection, it is essential the development of codes to allow the replacement of the current energy model for another, based on RES. It is also in this type of systems in which the work developed in this thesis is focused.

22 2.3 Regulations applicable to distributed generation

sources connected to electrical systems. Grid codes

2.3.1 Introduction

The increasing integration of non-conventional sources in power systems -mainly wind power

- has forced the transmission and distribution system operators (TSOs and DSOs) to update and redesign its grid codes. The grid codes are, essentially, sets of rules governing the connection and behavior that must meet the generators connected in electrical systems. The regulations are different in each country and the corresponding operator is responsible for the establishment of such conditions and verification of compliance.

The grid codes take as a reference the electrical characteristics and the design of the network itself. Their degree of demand is directly linked to the unmanageable power present in the system and the expected penetration rate. With the new policy it is pursued, as an end goal, to equate the behavior of renewable generation to the conventional groups already in service, and to ensure that the replacement of generation units in the system by others, means no additional risks in reliability.

There is a close relationship between standards, and the consequences they establish for manufacturers. The graphical summary of this idea can be seen in Fig. 2.11.

Increasing demands on the grid codes requires manufacturers to make improvements on their devices, thus resulting in an evolution of technology generation system and con- trol drives. With these improvements, an effective contribution to the maintenance of the equipment in the network is achieved and the inclusion of new sources of non-conventional generation is favored.

23 Higher requirements in grid codes

Renewable Technology energy evoluon increase

Figure 2.11: Consequences of the use of grid codes in electrical systems

2.3.2 Grid codes structure

All grid codes are structured in a similar way. In them it can be found three groups of standards, according to the aspects they regulate [22]:

• Connection requirements: These standards establish the conditions and connection

procedures for generators and consumers.

• Operation and safety procedures: These standards fix the action schemes for the elec-

trical system in normal operation and emergency (disconnection of generators, load

shedding plans, etc.). In addition, the regulation and managment of energy reserves is

included.

• Regulation of the electricity market: They settle the mechanisms of the day-ahead,

intraday and balancing electricity markets, as well as the conditions to access to the

market.

The distinction between the first two groups of standards, which have the most technical characteristics and are interrelated, is difficult in many regulations. For example, in the

24 Spanish case, all regulations are gathered in the so-called “Operation procedures”, that include the norms of the three previous groups.

In every country, the regulation framework is particular, complex and changing. In some cases, the corresponding grid code is equally applicable to the whole generation, as it happens in India [23]. In others, it is only defined the required response to the wind installations, as in the P.O.12.3 [15] in Spain, in China [24] o in the procedure still in draft in India [25]. The

German code [26] distinguishes between synchronous generators and the other technologies that do not employ a synchronous generator directly coupled to the grid. In the same way, it will be done in the draft Spanish code P.O.12.2 [27]. The code P.O.12.2 includes other generation sources such as photovoltaics, whenever they are connected to the transport grid, or to the distribution grids if they are greater than 10 MW. In other relevant grid codes, such as the Danish, the requirements are different not only in terms of the technology used, but also depending on the voltage level and the power of the generating plant [28, 29, 30].

2.3.3 Requirements in grid codes. Basic concepts.

This section addresses the general aspects that relate the requirements in grid codes. These criteria establish the disturbances that the developed test equipment must be able to repro- duce accurately. Therefore they have been left out of the analysis the obligations of active or reactive power contribution from generators and, in general, all aspects that depend on the generator control.

- Requirements facing voltage disturbances

For the proper functioning of the electrical system with DG, system operators (SOs) must keep the voltage levels within acceptable limits under normal operating conditions in all grid nodes. Moreover, the generation groups are responsible for helping SOs to meet those

25 requirements, staying connected if unusual circumstances in the system (failures that can cause voltage dips and over/under voltages) and restoring its previous operating condition as quickly as possible after clearance of the fault. The standard IEEE Std. 1159-1995

[31] establishes the classification of voltage disturbances that can occur in a power system by magnitude and duration, as shown in Fig. 2.12. The voltage dip -as defined in [15]- and voltage sag terms are usually used interchangeably [31, 32, 33]. They will also be equivalent along this thesis. Care must be taken because in some contexts, sag and dip refer, respectively, to the voltage drop or the remaining voltage after the fault according to the pre-fault value [34].

Figure 2.12: Classification of voltage disturbances according to IEEE Std. 1159-1995

Usually, the voltage ride-through requirements (VRT) are defined as patterns, which relate the admissible voltage levels and the associated duration times. In Fig.2.13 it can be observed the two types of existing profiles in the codes:

1. - Rectangular profiles: they are set according to the tripping steps of the protection

systems.

26 Figure 2.13: Profile types of voltage dips. a) Rectangular profile. b) Profiles with recovery ramp

2. - Profiles with recovery ramp: they represent the most severe envelopes obtained by

statistical analysis of the grid failures.

For over/undervoltage and transient overvoltage situations (voltage swells), the require- ments are always fixed with reference to the different steps and times set for protection systems.

- Requirements facing frequency disturbances

The frequency of the system is an indicator of the degree of stability, since any instan- taneous imbalance between the generation and the demand results in a variation of this magnitude. Furthermore, it is directly related to the obligations of active power input to the control systems of generation. The amplitude and period of the frequency oscillations depends on the characteristics of loads connected and the transient response of the gener- ation groups regulators. Both conventional and renewable generation sources must be able to stay connected and in operation if disturbances in steady-state or transient regime do appear. In some regulations, additional rate of change of frequency (ROCOF) requirements are requested to the groups.

27 - Requirements facing phase shifts

The phase shifts naturally happen in electrical power systems after a fault in the grid and as a result of a change in the impedance of the equivalent circuit. A phase jump is a displacement that occur in a voltage wave with respect to a reference that has the same frequency and harmonic content which is usually the voltage existing in the PCC prior to disturbance. The phase shift requirements are very specific and are only included in the most advanced grid codes.

2.3.4 Codes review

In this section several grid codes are analyzed. Firstly, an analysis of the grid codes applicable to interconnected electrical systems will be made, since they are the most complete and point the way to future developments in regulations. Later, the inquiry will focus on the codes in use nowdays in remote microgrids. They are, in number, much smaller. This is due to the fact that small-sized electrical systems located, sometimes, in developing countries, have not specific regulations or the reduced number of renewable energy sources installed does not justify the creation of a code that regulates its integration into the electrical system.

2.3.4.1 Grid codes of interconnected systems

2.3.4.1.1 Introduction

In the review of the literature relevant to the interconnected systems, the grid codes con- sidered have been those in countries with more wind power installed by the end of 2012 (by order: China = 75.324 MW; E.E.U.U. = 60.007 MW; Germany = 31.315 MW and India =

18.421 MW) [35]. For the United States it has been selected the code of the North Amer- ican Electrical Reliability Corporation (NERC) [36]. NERC brings together the users and operators not only of virtually the whole of the country, but also Canada and a small part

28 of the North of Mexico. For Spain, which remains by its wind power installed in 4th position worldwide (22.796 MW) has been compared the existing legislation P.O.12.3 [15] and the one pending approval P.O.12.2. [27]. Eventually, it has been considered of interest the inclusion of the Danish code. Denmark’s installed wind power is greater in comparison with the total system generating capacity (30% over the total, 4162 MW is wind power [35]).

2.3.4.1.2 Voltage requirements

- Voltage ride-through requirements (VRT)

In Fig.2.14 it is shown the voltage dip profiles required in the main international codes. Voltage_RMS (p.u.) Voltage_RMS

Time (s) Voltage_RMS (p.u.)

Time (s) P.O.12.2 Spain China India Denmark P.O.12.3 Spain USA_NERC Germany

Figure 2.14: Voltage dip profiles required in main international codes

29 Only some regulations include the zero ride-through (ZRT) requirement. Sources co- nnected through electronic converters are less able to contribute to the short circuit current compared to conventional units. This causes a more severe requirement regarding the mi- nimum depth of the voltage dip appearing in electrical systems when conventional sources are massively replaced by other technologies. To validate this statement it is compared the current and proposed profiles for the Spanish Peninsular territory which are represented in

Fig. 2.14:

1. The P.O.12.3 came into operation in the year 2006. The profile was the envelope of

voltage dips of a system with a total installed capacity of 83.198 MW. The 13.9%

corresponded to wind power generation [37, 38]. The depth of the voltage dip was set

for the given circumstances in 20% (0.2 p.u.).

2. The profile of the draft PO12.2 was proposed for the new situation in Spain, which in

the late 2012 had an installed capacity of 107,615 MW. Out of these, 25% corresponded

to technologies that do not use a synchronous generator directly connected to the grid

(22.785 MW of wind power and 4.298 MW of photovoltaic solar energy [39]). The

forecasts of the Government set the growth of these energies as values to be reached

the 35,000 MW of wind power and 7.250 MW of photovoltaic solar energy in 2020 [40].

According to this scenario the profile has been modified to include a ZRT requirement

of 150 ms.

- Over/undervoltage and voltage swells

For the other over/undervoltage and voltage swells, depending on their magnitude and duration and corresponding to the standard IEEE Std. 1159-1995 (Fig. 2.12) the require- ments are displayed in Fig. 2.15.

30 Dra P.O.12.2- Spain P.O.12.3 - Spain Germany USA_NERC Denmark China India

P > 11 kW y P < 25 kW P > 25 kW y P < 25 MW P > 25 MW

1,2 200 ms <100 ms 1,175 Defined in figures 8.2.1.b 500 ms <100 ms 1,15 2 s 1 h 1 s 1,115 30 min* 100 ms 1,1 200 ms 1,07 60 s 60 s 1,06 60 s Always 1 Not defined Always if U<400kV. If U =400kV, 0,97 Always Always Always Always Umax= 1.05.p.u. 0,95

0,9 1 s 10 s - 60 s 3 h 0,875 30 min 0,85 *Depending on system frequency up to 30 min Figure 2.15: Over/undervoltage and voltage swells requirements in main interconnected grid codes

2.3.4.1.3 Frequency requirements

- Allowable frequency ranges

In Fig. 2.16 it is shown the conditions required to generation sources in several grid generation codes to remain connected when frequencies outside the normal operating limits appear in the system.

- ROCOF requirements

In some grid codes the generators are required to remain connected to the mains not only when the frequency is outside normal ranges for a given time but also when facing to

ROCOF events. Thus it occurs, for example, with the new P.O.12.2.It will oblige the plants with synchronous generators not directly connected to the grid to withstand variations in the frequency for up to ±2Hz/s without disconnection. The Danish legislation sets a higher value of ±2.5Hz/s but only for those wind farms with a rated power between 11 kW and 25 kW. The current code in India [23] makes no reference to this requirement, but nevertheless,

31 Dra P.O.12.2- Spain P.O.12.3 - Spain Germany USA_NERC Dermark China India Eastern Western Quebec Ercot

66 5s 63 62,5 62 61,8 90 s 30 s 61,7 30 s 61,6 61,5 180 s 540 s 660 s 61 60,6 60,5 60 Always Always Always Always 59,5 59,4 59 180 s 660 s 540 s 58,5 58,4 30 s 30 s 58 90 s 57,8 2 s 7,5 s 57,5 57,3 0,75 s 2 s 57 56,5 0,35 s 55,5

53 52,5 f > 52 Hz - 200 ms 52 51,5 2 min 51 Not defined. 30 min 50,5 0.85 p.u. < U < 0,875 p.u. - 30 m 50,3 P.O.1.6 establishes Always 50 0,875 p.u. < U < 0,9 p.u. - 3 h the disconnecon Always 49,5 Always Always 49,2 0,9 p.u. < U < 1.115 p.u. - Siempre condion if frequency 49 drops to 48 Hz for 3s. 1.115 p.u. < U < 1.5 p.u. - 1 h 48,5 30 min 10 min 48 20 min 47,5 10 min 47 f < 47 Hz - 200 ms Figure 2.16: Operating frequency ranges wind farms will be asked to support ROCOF events when it the new regulation currently in draft [25] enters into force.

2.3.4.1.4 Phase shift

The most advanced code up to the date, the Danish, is the only one that includes the requirement of withstanding with an instantaneous 20o phase shift. The P.O. 12.2 will be even more strict, establishing a phase shift criterion of up to 30o. Note also that in this code, the requirement for the DG is more severe than for conventional generators, which are forced to remain connected to phase jumps of 20o and only occasionally to 30o, as a result of transport lines switches closures.

32 2.3.4.2 Grid codes of remote microgrids

2.3.4.2.1 Introduction

The current regulations in remote microgrids are scattered and incomplete. It has been identified two countries which host a significant number of isolated grids in their electrical system, Spain and France. This has forced them to develop specific grid codes for these remote microgrids.

In the island territories of Spain, very small systems can be identified as, for example, El

Hierro island with just 11.180 MW [41]. In the nonmainland territories of France some islands have installed capacities between 27 MW and 435 MW -C´orcega, Guadaloupe, Martinica,

Reuni´on, St. Pierre et Miquelon y St. Martin & St. Barthelemy- [42].

In all these remote microgrids it is essential to maintain a high proportion of conventional generation to ensure security. The French SO can order the shedding of RES when they reach a 30% instant penetration[43]. Under Spanish law the limit depends on each territory, but in no case may exceed 40% of the instantaneous production [44]. By the year 2015 in Canary islands it is expected a rate of renewable energy penetration both in peak and off-peak hours that can reach 100% of demand [45] and that will oblige, either a search of efficient storage solutions, or significant wind energy spills.

2.3.4.2.2 Major international codes applicable to remote systems

- Voltage ride-through

Figure 2.17 shows different profiles of voltage ride-through required by the different SOs of the aforementioned power systems. In the upper and lower part of the figure it is distin- guished depending on the size of the grid between systems up to and below 50 MW.

33 Figure 2.17: a) Voltage dip profiles for microgrids up to 50MW. b) Voltage dip profiles for microgrids above 50MW

In general it can be concluded that the voltage dip profiles are demanding as smaller the size of the microgrid is since virtually all include the ZRT capability. In contrast, the interconnected systems only have to withstand a zero voltage dip in those systems with a higher proportion of wind power installed, such as us (NERC), Germany or Spain (P.O.12.2).

In the EDF-SEI code in service, the required voltage dip depth is 0.05 p.u. [46]. To allow for the increase of RES, it is in process a modification of the profile for future regulations requiring the VRT with a depth of 0.01 p.u [47]. It is also expected the next approval

34 of a new profile in an amendment to the standard IEEE P1547 (IEEE P1547a/D2). In the

P.O.12.2-SEIE, applied to the Spanish Canary islands, it is demanded a pattern of 0 p.u. and lasting 500 ms. For the rest of the island territories, the profile is the same as in Mainland.

- Overvoltage/Undervoltage/Voltage swell

Ranges of operation at over/undervoltage regimes required on remote microgrids are shown in Fig. 2.18. The wider operation limits applicable to an interconnected system are collected in the new Spanish procedure P.O.12.2. According to this standard, the generating source must remain connected during 30 min when the voltage level drops to 0.8 p. u. and

50 ms if the voltage is maintained at 1.20 p.u. Up to 1.40 p.u. during 1s is required for the

Puerto Rico island. It can be observed that, in general, the required operating limits are greater in nonmainland territories.

EDF- SEI Spain - SEIE IEEE P1547 Hawaii - HECO Iceland Cyprus Jamaica PREPA NZ_North NZ-South

1.4 P< 100 kVA P>100 kVA 1.25 1.23 1 s 1.21 1.2 5 s U=1.23-0.125t 1.18 Instantaneous trip U=1.21-0.167t 0.16 s 1.175 2 s 1 s 1.16 1.15 0.4 s 1 s 0.2 s 1.115 3 s 1.1 Inst.Trip 1.5 s 1.07 Always 1.06 60 min 1.05 Always 1 Always Always Always Always 0.97 Always Always UNE 50160 Always Always 0.95 0.9 0.875 60 min 0.85 60 s Voltage dip 0.8 Instantaneous trip Voltage dip Voltage dip 5 s Figure 2.18: Over/Undervoltage requirements in remote microgrids

- Frequency requirements

Figure 2.19 shows the different bands of frequency operation due to different SOs req- uisites. In some countries it is appreciable the variations in frequency that must withstand

35 the generators in steady-state regime. This is the case, for example, of the South Island of

New Zealand, where the steady-state frequency may vary up to 8 Hz.

EDF- SEI Spain - SEIE NZ - North NZ - South Jamaica Iceland Cyprus IEEE 1547 IEEE 1547a Hawaii - HECO PREPA

55 53 P ≤ 30 kW P > 30 kW 52.5 5 s 52 63 3 min 51.5 Abnormal 62.5 51 60 min 62 0.16 s 6 s (P condions) 30 s 50.5 61.5 30 min 50.3 61 20 s 50 Always A Always 60.5 0.16 s 0.16 s Always * Always Always 49.5 l 60 ***Always Always 49.2 59.8 Always Always 49 59.5 Always Abnormal Always 48.5 (P condions) 60 min 59.3 0.16 s Adjustable 48 30 min 59 0.16 s to 20 s 47.5 **120 s 58.5 300 s 47.33 min 3 s **20 s 5 s 58 47.1 **5 s** Intermediate values 57.5 0.16 s 0.16 s 10 s 47 60 s **0.1 sobtained by interpolaon 57 46 0.4 s30 s ***If 0.9

Figure 2.19: Frequency requirements

- Derivatives of frequency

The requirement for generation sources to remain connected under derivatives of fre- quency is not common to all analyzed codes, but it is reflected in some of them. For example,

ROCOF requirements up to 0.37 Hz/s in Hawaii, 0.75 Hz/s in New Zealand, and up to 1.3

Hz/s in Cyprus are obliged.

2.3.4.3 Towards harmonized standards

2.3.4.3.1 Introduction

The dependence of the legislation in each country forces to manufacturers to certify their equipment according to every grid code of the territories in which their products are installed.

36 That is why it seems more necessary to harmonize the different regulations. The creation of a common European code will increase the levels of security and the stability in the grid and will facilitate the insertion of new RES in the system.

2.3.4.3.2 ENTSO-E grid code

The European Network of Transmission System Operators (ENTSO-E) brings together the technical operators of the European electrical systems and it is responsible for both coordi- nation between different TSOs and control of energy exchanges through European borders.

This organization, in collaboration with the Agency for the Cooperation of Energy Reg- ulators (ACER) and the operators of the electricity markets, have been working on the development of a joint code [48].

By means of adopting a common grid code among the members of the European Union it is intended to achieve several benefits [49]:

• To assure a few requirements that must be specified in all the codes of the member

states.

• To unify terminology, parameters and conditions between the different grid codes.

• To establish compliance obligations and exceptions equal to all European countries.

In the section on Requirements for Grid connection (NC RfG) the new pan-european code defines what are called “not exhaustive requirements”. Not exhaustive requirements

flexible operating limits that each national TSO must adapt and supplement. The ENTSO-E grid code will prevail over the national regulations and it will be only applicable to the new generators that want to get connected to the European transport system. The obligation of compliance or not by the existing wind power plants will depend on each national TSO.

37 The code is complex, and consists of several regulations that will be adopted with a temporal sequence still not defined, so the final schedule is not fixed. It is expected the NC

RfG entry into force throughout 2015. Since then, there will be a three-year period for the adaptation of the national codes to the new common framework.

2.3.4.3.3 Highlights of ENTSO-E grid code

It table 2.1 the classification of power plants, named Power Generation Modules (PGMs), in four categories according to the ENTSO-E code is shown. PGMs go from A to D taking into account several factors like power installation, location and the generation technology -power plants with conventional synchronous generators (SPGMs) or power plants with generators connected to the grid through electronic converters (PPMs).

Synchronous area Maximum capacity Maximum capacity Maximum capacity threshold from which threshold from which threshold from which whichonaPGM whichonaPGM whichonaPGM is of Type B is of Type C is of Type D Continental Europe 1MW 50 MW 75 MW Nordic 1.5 MW 10 MW 30 MW Great Britain 1MW 10 MW 30 MW Ireland 0.1 MW 5MW 10 MW Baltic 0.5 MW 10 MW 15 MW

Table 2.1: PGM installed capacity limits according to ENTSO-E regions [1]

It can also be noticed how the requirements increase when the installation power incre- ments. The PGMs type A, with a power over 0.8 kW, must remain connected within the normal operating limits of voltage and frequency in steady state. For the D-type, it is also demanded controls over active and reactive power, response against disturbances, etc., as the loss of a PGM would be a potential risk for grid security.

38 The following paragraphs show the margin requirements that establish the limits inside every TSO must define the voltage and frequency requirements.

- Requirements related to voltage disturbances

As one of the not exhaustive requirements, it is defined a feasible region of operating

RMS voltage-time, in which each SO must fix its own profile according to the particular conditions of its power system. The maximum and minimum limits to frame the profiles, depending on the technology and the power of the plant are depicted in Fig. 2.20:

Figure 2.20: Voltage dip profiles in ENTSO-E code. a) SPGMs type B and C; b) PPMs type B and C; c) SPGMs type D; d) PPM type D

For the PGMs type A and B it is not specified normal operating ranges against voltage variations. These limits will be set directly by the grid aggregator and for type C, the levels shall be fixed by agreement between the grid aggregator and the national SO. As an example, allowable ranges for voltages in type D PGMs are shown in tab.2.2:

39 Synchronous area Voltage Range Time period for operation 0.85 p.u. — 0.9 p.u. 60 min 0.9 p.u. — 1.118 p.u. Unlimited Continental Europe 1.118 p.u. — 1.15 p.u. To be decided by each TSO while respecting the provisions of Art. 4(3) but not less than 20 min

a) Synchronous area Voltage Range Time period for operation 0.85 p.u. — 0.9 p.u. 60 min 0.9 p.u. — 1.05 p.u. Unlimited Continental Europe 1.05 p.u. — 1.0876 p.u. To be decided by each TSO while respecting the provisions of Art. 4(3) but not less than 60 min 1.0875 p.u. — 1.10 p.u. 60 min b) Table 2.2: Operating ranges of voltages. a) Voltages from 110 kV to 300 kV. b) Voltages from 300 kV to 400 kV

- Requirements facing to frequency disturbances

In Table 2.3 the minimum times that the PGMs must remain connected depending on the frequency in the continental region are collected:

It can be seen that the code only establishes the minimum times of operation at several frequencies in which the system should continue working. It is required a continuous oper- ation of the wind farm for frequencies between 49 Hz to 51 Hz and they have to withstand during 30 min for an overfrequency of up to 51.5 Hz before disconnection. According to the location, these requirements will be different. It remains for the Regulatory Authorities the responsibility for setting the bottom time limits for unassigned frequency steps. These values must be labeled with sufficient time prior to the entry into force of the code to make the necessary investments in order to ensure the system reliability.

40 Synchronous area Frequency range Time period for operation 47.5 Hz — 48.5 Hz To be decided by each TSO while respecting the provisions of Art. 4(3) but not less than 30 min Continental Europe 48.5 Hz — 49.0 Hz To be decided by each TSO while respecting the provisions of Art. 4(3) but not less than the period for 47.5 Hz — 48.5 Hz 49.0 Hz — 51.0 Hz Unlimited 51.0 Hz — 51.5 Hz 30 min

Table 2.3: PGMs frequency requirements

It is also a condition in the code for the generators to stay connected and operating when facing frequency ramps. It is the responsibility of the National Operator the setting of the limit values.

2.3.4.3.4 ENTSO-E and remote microgrids

According to the European directive 2009/72/EC a small isolated network had a power con- sumption of less than 3000 GWh in the year 1996 and obtained by means of interconnections, less than 5% of annual consumption [50].

There are various European territories that meet the characteristics of isolated systems such as the Canary Islands, Cyprus, Malta, Madeira or the Aland Islands. ENTSO-E code aims at the harmonization of the grid codes for the future construction of a unified European electrical system. The isolated power systems would presuppose a very small impact on the global system (if they could lately be connected to the Mainland) or no impact if they would keep isolated because of geographical circumstances. In either case, the effects of electrical systems will be minimal and it is excluded in this initial version of the ENTSO-E code.

41 Nevertheless the importance of the harmonization between the different territories has forced the creation of a group within the structure of ENTSO-E, focused on isolated systems: the Regional Voluntary Group Isolated Systems (VRG IS). It is composed by four large

European operators: REE (Spain), Terna (Italy), Landsnet (Iceland) and CTSO (Cyprus).

Its mission is to ensure the proper management and operation of the isolated power systems and the provision of its expertise in the development and updating of the grid codes. The experience provided by the VRG IS is expected to be included in future versions of the code. This is also intended to also advance the harmonization status in grid codes applied to isolated systems [51].

2.4 Grid disturbances generators to test electrical de-

vices

In Subsection 2.2.2.1 the differential characteristics have been described that define the disturbances that appear in remote microgrids respect to other electrical power systems of larger size and strongly meshed. It is essential to have devices to test and predict the behavior of the generating plants against combined voltage and frequency disturbances. This will be needed to certify its operation in accordance with the requirements set by the relevant SO before being installed.

The aspects related to the certification of distributed generation sources have been broadly addressed in the scientific sphere, where many examples of technical papers on this topic can be found.

Most of the papers and patents found in literature search raise the physical constitution and the development of equipment that exclusively generate voltage dips. Only some of the latest devices allows the reproduction of more than one disturbance type. The development

42 validation and certification devices have always been closely linked to the evolution of the requirements demanded by the grid codes. Voltage dips are the most severe disturbances that can occur in the system. In the past they were responsible for massive disconnection of wind power plants. Hence the SOs modified the grid codes to establish new requirements for the response of the wind generation plants when facing voltage dips.

This section is a review of the current state of the art in the existing solutions for certification of DG sources. Three are the common topologies of voltage dip generators: those based on impedance, those using an electric machine, static () or rotating and those which use electronic converters.

2.4.1 Impedance-based

The insertion of a ground impedance allows an alternative path to the current and generates a voltage variation on the point of common coupling (PCC).These generators are the simplest and, therefore, they are widely used. The adjustment by using electronic devices of the impedance value, determines the voltage dip depth in the experiment. They are easily scalable up to the usual power of the devices to be tested. By using a precise control over the electromechanical or static contactors, they can generate single-phase, two-phase and three-phase voltage dips. In [52] it is displayed the most common setup of this voltage dip generator type that schematically can be seen in Fig. 2.21. In [53] the impedances are the ones of a transformer, but the equipment maintains the usual T-configuration. Because of this, it has been considered as belonging to the first group.

In Spain, this voltage dip configuration is recommended by the Asociaci´on Empresarial

E´olica (AEE) in their document titled “Procedimientos de verificaci´on, validaci´on y certi-

ficaci´on de los requisitos del P.O.12.3 sobre la respuesta de las instalaciones e´olicas y foto-

43 Zs EUT

Grid Zg

Figure 2.21: Impedance-based voltage dip generator voltaicas ante huecos de tensi´on” [54] for testing wind turbines. In [55] it is specified for this topology that if the short-circuit power in the PCC is less than five times the rated power of the EUT, there are effects in the dynamic response of the device due to the series impedance. This is its main drawback and therefore the operation code of the AEE requires the adjustment of the certification equipment at no-load or loaded according to relationship between the PCC and the EUT powers.

2.4.2 Electrical machines based

This voltage dip generator type uses as main element an electric machine, either static or rotating. There has only been found two references [56, 57] in which voltage dip generators with synchronous machines, which reproduce the voltage dip profile by modification of the excitation voltage are mentioned. The implementation at a larger scale of voltage dip gene- rators with synchronous machine involves a very high cost and, as a consequence, it has not been a widespread topology.

Most of voltage dip generators based on electrical machine use a transformer in combi- nation with any kind of electronic switches (SCRs or IGBTs) [58, 59] as observed in Fig.

2.22.

44 Stac switches

EUT

Tapped transformer Grid Figure 2.22: Transformer-based voltage dip generator

These systems are easy to scale, because the component elements exist commercially up to the required levels of voltage and power to test the available wind turbines. Its main disadvantage is that the voltage dip depth is fixed by the value of the inductance. It is common to install tapped transformers, which allow the testing of various voltage dip patterns. Within this topology it is convenient to emphasize the model proposed in cite [60, 61]. In it, the authors described a voltage dip generator oriented to microgrids.

Through simulation, it is checked the analysis of the interaction between a conventional voltage dip generator based on transformers and the SCRs with the SCRs of the microswitch that connects the microgrid to the mains. Reference [62] shows a device consisting of a tapped transformer with also tapped inductances, so both voltage dips and phase jumps disturbances are defined in various steps depending on the number of connected inductances. This device is versatile, and it allows to testing in high voltage and power of the EUT up to 10 MW. It reproduces voltage sags, frequency disturbances and phase jumps, but never simultaneously.

It is perhaps the device found throughout the literature revision that better approaches to the objectives defined for this thesis device.

2.4.3 Full-converter based

This is up to the present the type on which has been done a greater number of developments

[63, 64, 65, 66, 67, 68] The scheme of a device with this topology can be seen in Fig.

45 2.23. It has two back to back converters between the grid and the EUT which permits a complete control over the dip to be generated. However, the full-converter based voltage disturbances generators distort the wave and the devices are more expensive than the other classes abovementioned. The electronic components themselves have a high price and they also require a complex control system. Therefore, the implementation of such equipment remains in low voltage levels, and there are no commercial devices based on this technology.

EUT

Figure 2.23: Voltage dip generator based on full power converter

2.4.4 Comparative between commercial topologies and the pro-

posed device

The prototype developed in this thesis fits into the second class (voltage dip generators with electrical machines), but it is based on an induction regulator and is the only one that uses this type of machine. By making the voltage changes through an induction regulator, continuous adjustment is achieved. This gives the device the possibility of reproducing any profile, any depth and any recovery ramp settled with independence of the grid code, eliminating the main disadvantage linked to disturbances generators based on transformers.

It is also the first device whose design has been specifically oriented to be used in the certification of equipment before its connection to remote microgrids. The use of a second electric machine in cascade allows simultaneous reproduction of more than one disturbance.

In remote microgrids, as described in Subsection 2.2.2.3, disturbances are characterized by

46 synchronous changes in voltage, frequency and phase. As the device can synchronously reproduce all these disturbances, and no other commercial equipment with this characteristic has been found, it is specially suitable to emulate the real behavior of remote microgrids and, therefore, to test devices before its connection to such grids.

2.5 Conclusions

In this chapter three main points of interest have been analyzed in regard to this thesis. The progressive integration of DG in power systems has led to two main consequences, addressed in the first two parts of the chapter. On one hand, the evolution of the topology of power systems, from large electric systems interconnected to smaller, decentralized systems, which can operate in isolated or grid-connected mode, called microgrids. On the other hand, increased requirements in the grid codes, derived from the replacement of the conventional groups by intermittent generation. In particular, the microgrid is a very suitable structure for isolated areas (such as islands or populations of developing countries). But in dealing with especially weak grids, there are also greater requirements on the grid codes for remote microgrids. The need to certify DG sources before its insertion into the power systems has led to the search of different technological solutions, collected in the third part of the chapter.

47 48 Chapter 3

General description and operating model of the voltage disturbances generator prototype

3.1 General description of the device

The device proposed along this dissertation is mainly composed by two electrical wound rotor machines, their associated electrical drives, the set of electromagnetic contactors and the control & monitoring system. The two major advantages of this equipment are the absence of electronic power components (so there is no additional distortion of the voltage waveform) and its capacity to reproduce any voltage dip profile existing in the grid codes, including their recovery ramps.

A general scheme of the device is displayed in figure 3.1.

The voltage disturbances prototype is located between the undisturbed grid (input of contactor KM1) and the EUT (output of contactor KM7). Throughout this section, the operational information, the control system description and the mathematical models for static and dynamic regimes are gathered. Wind turbine Electrical grid

1 3 5 1 3 5 135 135

K1 KM7 K1 KM1 246 246

2 4 6 2 4 6 KM4 KM6 – KMT1 K1 K1 5 6 5 6 1 2 1 2 4 3 4 3 3 4 3 4 1 2 1 2 5 6 5 6 1 3 5 135

1 3 5 KM5 K1 135 246

KM2 K1 2 4 6 246 KM3 K1 2 4 6 5 6 1 2 3 4 3 4 1 2 5 6 ω MAGNITUDE (EM2) ROTOR

ROTOR STATOR α STATOR FREQUENCY Rlim KMT2 K1

PHASE 5 6 (EM1) 1 2 3 4 3 4 1 2 5 6

KMT3 KMT4 Rcc K1 K1 5 6 5 6 1 2 1 2 3 4 3 4 3 4 3 4 1 2 1 2 5 6 5 6

Figure 3.1: General scheme of the device

50 3.1.1 The electrical machines

3.1.1.1 Variable frequency transformer (VFT)

Identified as EM1 in Fig. 3.1, this machine is powered by the mains with a voltage of a settled frequency. The output is connected to the PCC of EM2, as presented in Fig. 3.2.

The variable frequency transformer it is a relative new electrical machine. The first one in service started its operation in 2004 to interconnect the USA and Canadian power systems, with different frequencies [69, 70]. By means of a continuous and controlled phase shift, the

VFT can regulate the power flows between asynchronous grids.

Even if EM1 has the same design that a regular VFT, in the voltage disturbances genera- tor it has two other functions not directly related with the power flow control. The first one is to modify the frequency of the input voltage and the second is to set up an initial phase to obtain the controlled phase jump. The modifications of the frequency and the phase are achieved through the control system, which adjusts the initial angle between the rotor and the stator axis as well as the rotational speed.

3.1.1.2 The induction regulator

The induction regulator, identified as EM2 in Fig. 3.1 is the electrical machine responsible for creating voltage magnitude disturbances. Because of the importance that voltage dips have in power systems, justified in Subsection 2.2.2.1, the production of voltage dips will be the main task of the voltage disturbances generator and in which primarily will focus subsequent design targets.

The induction regulator is constructed like a wound rotor induction machine. Due to this, it is an economical, robust and easily scalable device. Despite these advantages, it has never been a industrially relevant machine. In its nascency, at the first half of the 20th century, it

51 Ug EM2 Ui ω Ur Us

WOUND ROTOR INDUCTION MACHINE

Figure 3.2: Single-phase scheme of the wound rotor induction machine connection was used to control voltages in railway traction, but the development of tapped transformers definitely relegated it to the background [71, 72]. With this dissertation it is intended to recover the induction regulator as a suitable machine with a commercial purpose.

EM2 is placed between the grid and the EUT. The grid can be directly the PCC of the distribution network or the output of the machine responsible of the frequency and phase changes, located upstream. To use the induction regulator in this device, windings have to be connected in parallel and to the PCC, as observed in Fig. 3.3. The rotor is also linked to a servo motor controlled by a servo drive. The servo drive takes action over the relative angle position between the windings (α). Voltage applied to the EUT is always the difference of rotor voltage phasor (Ur), which is equal to the input voltage (Ui), and stator voltage phasor (Us). Equation 3.1 shows this relationship.

Uo = Ur − Us (3.1)

52 Us

EUT Uo Ui Ur α

INDUCTION REGULATOR

Figure 3.3: Single-phase scheme of the induction regulator connection

With the use of an induction regulator can be achieved a simple buck-boost voltage device. A graphical image of the machine phasor diagram is represented in Fig. 3.4 for two shaft angles. It can be observed that depending on the physical position, it is obtained a different output voltage in both magnitude and angle [73].

Using the Law of Cosines applied to the phasor diagram it can be easily deduced the Eq.

3.2:

2 2 2 Uo = Ur + Us − 2 · Ur · Us · cos (α) (3.2)

Magnitudes of rotor and stator voltages are related through the module of the transfor- mation ratio, expressed in terms of the number of phases in rotor (mr) and stator (ms)and the corresponding effective turns en each winding (kws · Ns for the stator and kwr · Nr for the rotor), as shown in Eq. 3.3:

U m · ξ · N s = s s s = k (3.3) Ur mr · ξr · Nr

53 Ur α Us Uo

Figure 3.4: Voltage phasor diagram

With this expression, above equation 3.2 it can be rewritten as Eq. 3.4 to present the output voltage magnitude of the prototype as a function that depends on the input magnitude value.

2 2 2 Uo = Ur · (1 + k − 2 · k · cos (α)) (3.4)

The transformation ratio is close to unity, so without committing an excessive error, the

Eq. 3.4 can approach the more simple expression 3.5. It can be ascertained that depending on the mechanical angle turned by the shaft, output voltage can vary approximately from zero to double of the input value.

Uo = Ur · 2(1 − cos(α)) (3.5)

54 3.1.2 The control program

3.1.2.1 Cascade control design

The control program developed in MATLAB/SimulinkR was carried out by setting as its main objectives: flexibility, precision, high performance and extensiblity. It has also been searched the simplicity in the design. The structure of the control is equal for both sets of servo motors plus electrical machines, as observed in Fig. 3.5. They are based on a cascade control scheme composed by two nested loops.

f f* n* PI c c PI c SM1 EM1 n

MATLAB/Simulink CONTROL LOOP ALTIVAR CONTROL LOOP

U U* α* PI c c PI c SM2 EM2 α

MATLAB/Simulink CONTROL LOOP BAUMÜLLER CONTROL LOOP

Figure 3.5: Control loops for EM1 and EM2

The use of the cascade structure grants a better control of the primary variable —me- chanical angle or mechanical speed—, which is less affected by external disturbances. It also enables a faster operation and it improves the system dynamic response. Due to this, the cascade control helps to fulfill the objetives previously defined for the control system. The external PI controllers developed in MATLAB/SimulinkR for the dSPACE DSPs, act as

55 the master controllers while the commercial drives regulate the internal loops, behaving as slaves.

In the control of EM1, the error between the reference frequency (f*) and the measured frequency (f) sets the speed control for the servo motor of EM1. The speed measurement read by the encoder establishes the feedback for the secondary loop. Similarly, in the machine control EM2, the comparison between the magnitudes of the reference voltage (U*) and the measured value (U) results in the generation of a set-point angle (α). The feedback through the encoder installed on the actuator shaft, closes the internal control loop.

3.1.2.2 Program I/O

In Fig. 3.6 it is presented a graphical scheme of the control program inputs and outputs.

UrEM1, UsEM1, UiEM1

UrEM2, UsEM2 α* IrEM1, IsEM1 Analog I/O n* IrEM2, IsEM2 α, f

DR

D1 Digital I/O D2

D3

Figure 3.6: Model I/O

Analog variables UiEM2 y f are the values of the voltage magnitude at the output of the induction regulator and the frequency measured at the terminals of the machine EM1. These two quantities are used as control variables. The remaining analog inputs are measurements for logging and monitoring purposes.

56 In terms of the digital outputs, DR Controls the sequence of actions of the contactors that allows the creation of voltage dips. D1, D2 y D3 are digital signals that configure the device contactors according to the type of disturbance to be produced. They are equivalent to the operation of the buttons on the front panel of the cabinet.

3.1.2.3 Implementation in MATLAB/SimulinkR

The program has been accomplished following a block structure which general scheme as showninFig.3.7.

Figure 3.7: MATLAB/SimulinkR control

57 Blocks from the A to D correspond to the generation of over/undervoltage disturbances and voltage dips, namely corresponding to the electric machine control EM1. Block A acts on the initial calibration of the equipment, controlling the complete rotation of EM1 in both directions necessary for the establishment of maximum and minimum reference voltages. The process of searching for maximum and minimum voltages, the corresponding values of angle where they are reached, and the storage of these values for the following reproduction of the disturbance, are made in subsystem C. Once the initial calibration of the test equipment is

finished, voltage dips or over/undervoltage disturbances are generated with the programming in block B. D block encloses the PI controller for voltage magnitude disturbances.

The subsystem E includes the necessary algorithms for the reproduction and control of the frequency disturbances and phase jumps, that is, the whole regulation system corresponding to the electrical machine EM1.

Blocks F and G are general program blocks. Block F centralizes the entire system of measurements acquisition as well as sensors calibration and adjustment. The preset of the contactors of the equipment according to the type of disturbance can be done both manually and automatically, managed by the MATLAB/SimulinkR program. G block allows the direct action of the program on the power cabinet, replacing the manual panel switches.

3.2 Mathematical expressions

In this section the analysis of the mathematical expressions that govern the disturbances generator behavior will be presented. Separately, the equations are studied corresponding to each of the two electrical machines that compose the device; on one side, the variable frequency transformer (VFT) used to produce the frequency oscillations and controlled phase jumps (EM1), and on the other side, the induction regulator (EM2). In order to simplify the

58 nomenclature, the use of EM1 and EM2 subscripts are avoided in reference to the magnitudes of the rotor and stator of the machines.

3.2.1 Variable frequency transformer - EM1

3.2.1.1 Steady-state equations and equivalent circuit

The process of obtaining expressions of EM1 electric machine is similar to that used for the analysis of an . There are two major differences between the operation of the machine as asynchronous motor and the operation of the VFT (EM1) as part of the voltage disturbances equipment. The first one is the rotor motion it does not occur free but controlled by a variable speed servodrive, so the slip and the frequency disturbance associated is imposed by the operator; the second is the magnetization of the machine is conducted from the rotor side.

The original single-phase equivalent of the VFT is shown in Fig. 3.8. The winding

Rr jXσr Ir Rs Is jXσs

Ur ω Er Es Us

Figure 3.8: Single-phase representation of the EM1 machine

powered with the mains at a rated frequency —primary winding— is the rotor. The spinning of rotor coupled with the servomotor shaft induces an emf on the stator whose frequency depends on the slip. To study the behavior of the machine EM1 under different operating conditions, it is desirable to obtain an equivalent to simplify the analysis that does not directly involve the transformation ratio between windings. Since frequency is different in both windings, the circuit should be reduced to an equivalent which is referred to the rotor

59 frequency and that remains constant. It is known from the general theory of electrical machines, the expression of the slip, Eq. 3.6, representing the relationship between the difference of speeds of synchronism and the shaft turn as a fraction of the synchronous speed, and thus between the frequencies of the rotor and the stator.

ωs − ω ωs s · ωr s = ,fs = = = s · fr (3.6) ωs 2 · π 2 · π

Applying the definition of slip to the equation that relates frequency stator with emf in that winding the Eq. 3.23 is obtained. In this formula, ξs is the stator winding factor and

φm the common flux that goes through the air gap.

Ess =4.44 · ξs · Ns · fs · φm =4.44 · ξs · Ns · s · fr · φm = s · Es (3.7)

Es is the electromotive force in the stator if the frequency was the same as in the rotor, i.e. if there is no relative motion between the windings (stationary rotor). On the other hand, the frequency has a direct influence over the value of the stator reactance (Xσss = s · Xσs). The mesh equation in the stator winding is then expressed as shown in Eq. 3.8

Us = s · Es +(Rs + j · s · Xσs · Is) (3.8)

And divided by the slip it can be attained the final equivalent circuit that has been represented in Fig. 3.9.

jXσr Rr Ir Rs/s Is jXσs

US Ur Er Es s

Figure 3.9: Single-phase representation of the EM1 machine reduced to fr frequency

60 To simplify the analysis of the original stator circuit, it is reduced to an equivalent stator with the same number of turns as the rotor and physically aligned with it. The variables in this new circuit are denoted with ’. Voltages and emfs of the original stator and the equivalent are related through transformation ratio K as displayed in Eqs. 3.9 and 3.10:

E s = K · Es (3.9) U s = K · Us (3.10)

When replacing the secondary circuit by an equivalent, it must be accomplished that the apparent power has to be equal on both stators. The current ratio is then expressed as shown in Eq. 3.11:

∗ 1 S = U · I = U · I −→ I = · I (3.11) s s s s s s K∗ s

An analogous development is used to obtain the relationships between the resistances and the leakage inductances in both stators. The dissipation of active and reactive power in the original winding must be identical to the equivalent winding, leading to the Eqs. 3.12 and 3.13 and to the circuit referred to rotor-side —Fig. (3.10)—:

2 2 2 2 2 Rs · Is = Rs · Is −→ Rs = Rs · K (3.12) 2 2 2 2 2 Xs · Is = Xs · Is −→ Xs = Xs · K (3.13)

jXσr Rr Ir R’s/s I’s jX’σs

U’S Ur jXm s

Figure 3.10: Single-phase representation of the EM1 machine reduced to rotor winding

61 3.2.2 Induction regulator - EM2

3.2.2.1 Steady-state equations and equivalent circuit

In Fig. 3.11 is shown the single-phase equivalent of the induction regulator, EM2, where

• Ur and Us are the rotor and stator voltages

• Rr and Rs are the rotor and stator resistances

• Xσr and Xσs symbolize the leakage inductance in the windings

• Xm is the mutual inductance between windings, associated to the rotor-side.

• Er and Es are emfs in the rotor and stator windings

• Ir and Is are the currents in the rotor and the stator

jXσr Rr Ir Rs Is jXσs

Ur jXm Er Es Us

Uo B A Figure 3.11: Single-phase circuit of EM2

The rotor voltage phasor is taken as reference for angle measurements and α is the electrical angle between the electromotive forces of the rotor and the stator. This angle is positive when the stator voltage is in delay with regard to the rotor. Therefore the transformation ratio K is defined as in Eq. 3.14:

1 K = · ejα (3.14) k

62 The writing of the equations from the electrical circuit of the induction regulator leads to the Eq. 3.15: ⎛ ⎞ ⎛ ⎞ · ⎛ ⎞ · · j Xm ⎜ Rr + j Xσr + j Xm ∗ ⎟ ⎜ Ur ⎟ ⎜ K ⎟ ⎜ Ir ⎟ ⎝ ⎠ = ⎝ ⎠ · ⎝ ⎠ = j · Xm j · Xm Us Rs + j · Xσs + Is K K2 ⎛ ⎞ ⎛ ⎞ jα ⎜ Zr j · k · Xm · e ⎟ ⎜ Ir ⎟ = ⎝ ⎠ · ⎝ ⎠ (3.15) −jα j · k · Xm · e Zs Is

Zr y Zs are the total impedances of both rotor and stator windings. If changing the stator to a another one, equivalent, and reduced to rotor side, the circuit of the induction regulator is depicted on figure 3.12 and the glean of the equations, previously shown, that govern its operation is immediate from the mesh equations.

jXσr Rr Ir R’s I’s jX’σs

Ur jXm U’s

Figure 3.12: Equivalent circuit of EM2

⎛ ⎞ ⎛ ⎞ ⎛ ⎞

⎜ Ur ⎟ ⎜ Rr + j · Xσr + j · Xm j · Xm ⎟ ⎜ Ir ⎟ ⎝ ⎠ = ⎝ ⎠ · ⎝ ⎠ (3.16) Us j · Xm Rs + j · Xσs + j · Xm Is

This equivalent circuit will be later used as starting point for the development of the transient regime equations.

63 3.2.3 Thevenin equivalents

Based on the equivalent circuits developed in Subsections 3.2.1.1 and 3.2.2.1 it is prompt the deduction of the Thevenin equivalent for both machines. The equivalents will allow to work in a more comfortable way with the circuits and will be used for the simulation and validation of the mathematical models.

3.2.3.1 Voltage frequency transformer - EM1

• Thevenin source

Let be Ug the voltage of the mains supply. The calculation of the no-load terminal voltage in the stator leads to the following expression 3.17:

· · Us j Ug Xm Uo1 = = (3.17) s Rr + j · Xr

• Thevenin impedance

The Thevenin impedance calculated as the ratio between the no-load voltage and the short-circuit current(Icc = −Is) then fulfills the Eq. 3.18:

2 Uo Rs Xm Zth1 = = + j · Xσs + j · Xm + (3.18) Icc s Rr + j · Xr

Expressing the equivalent in the real stator magnitudes, final equations of Thevenin source and impedance are displayed in Eqs. 3.19 and 3.20:

· · · · · · −jα Us j Ug Xm j Ug Xm k e Uo1 = = = (3.19) s K · (Rr + j · Xr) (Rr + j · Xr) · 2 Uo Rs j Xσs · Xm Zth1 = = 2 + 2 + j Xm + (3.20) Icc k · s k Rr + j · Xr

64 3.2.3.2 Induction regulator - EM2

• Thevenin source

Again Ug represents the rotor voltage supply. In Eq. 3.21 is presented the no-load voltage: −jα j · k · Xm · e Uo2 = Ug − U s = Ug · 1 − (3.21) Rr + j · Xr

• Thevenin impedance

The induction regulator impedance is, then, displayed in Eq. 3.22:

2 Uo Xm Zth2 = = Rs + j · Xs + (3.22) Icc Rr + j · Xr

If the equivalent is expressed in stator real variables, but referred to rotor-side the final formulation is shown in Eqs. 3.23 and 3.24: −jα j · k · Xm · e Uo2 = Ug − U s = Ug · 1 − (3.23) Rr + j · Xr

Uo Rs j · Xσs j · Xm · (Rr + j · Xσr) Zth2 = = 2 + 2 + (3.24) Icc k k Rr + j · Xr

It can be seen the similarity of the equations that govern the behavior of the two electrical machines. In the first case, the angle α offset is a constant, but the slip is a variable.

Meanwhile, in EM2 the slip will be very close to the unit throughout all the operation regime but the angle value varies in a controlled way to adapt the output to the disturbances profile set by the user.

The development of a global Thevenin equivalent, involving both machines, supposes a more complex mathemathical approach. Since it has not been directly used in the simulation

65 model and in order to keep the clearness of this document, this circuit and its equations have been obtained and embodied in Appendix B.

3.2.4 Simulation model

This section demonstrates the validity of the simulation model developed in the SimulinkR environment for the verification of the equivalent circuits deducted in the previous section.

To have a model is important for two reasons: firstly, if the model is valid, the preceding reckoning of the induction regulator parameters is also well-grounded; and secondly, a tested model constitutes a powerful tool to analyze later, if necessary, the effect of connecting different loads to the device.

The verification of the simulation model in steady-state regime is shown for the base prototype, being equally valid for the other designs by the substitution of the equivalent circuit parameters according to the convenience. The general scheme, whose image is shown in Fig. 3.13 incorporates the two electrical machines. The validation is displayed with independent disturbances carried out by EM1 and EM2, being trivial the extension to the general case with combined disturbances.

Through preliminary lab experiments, the induction regulator parameters have been determined, and are shown in Table 3.1.

Rr Rs Xr Xs Xmrs Lab measurements (Ω) 2.42 3.06 192.7 228.8 181.05

Table 3.1: Electrical machine parameters

66 Figure 3.13: Simulation model of the voltage disturbances generator

(a) Induction regulator output (simulation) (b) Induction regulator output (experimental)

Figure 3.14: Steady-state comparative between simulation and experimental results

In Figs. 3.14a and 3.14b, the validation of the operating model of the first machine, EM1, is observed. The data displayed has been contrasted for two different types of disturbances at two load levels. The one in Fig. 3.14a compares the frequency set-point and the response in the machine terminals for an overfrequency disturbance with a 20% load level. The second,

Fig. 3.14b, represents an underfrequency ramp with a 2.5 Hz/s drop and the machine operating at full load.

Figure 3.15 shows a comparison of the results in permanent regime for different load steps tested (from no-load to full load) and for diverse angles. In Fig. 3.15a it has been represented

67 Simulaon results Experimental measurements 250 250

200 200

150 150

100 100 Uo (V) Uo (V) 50 50

0 0 0 102030405060 0 102030405060 α(º) α(º)

No load Is=1A Is=2A Is=3A No load Is=1A Is=2A Is=3A

Is=4A Is=5A Is=6A Is=7A Is=4A Is=5A Is=6A Is=7A (a) Induction regulator output (simulation) (b) Induction regulator output (experimental)

Figure 3.15: Steady-state comparative between simulation and experimental results the simulation results while in Fig. 3.15b the experimental measurements are displayed. The angles between 0◦ and electrical degrees 60◦ are, approximately, the range of angles swept by the induction regulator for the reproduction of different dip profiles discussed in this thesis and, therefore, it has been considered an ambit of interest to study.

From the values shown in Figs. 3.15a and 3.15b and in order to quantify the differences between the simulations and the experimental results, errors have been evaluated depending on the motion angle and the load level. Collected results appear in Fig. 3.16.

The percentage errors are much greater in absolute value for low angles (< 10◦)and higher currents. As the values for very low voltage values (low angles) are very small, any imprecision can contaminate the results, leading to important estimated errors in the comparison between the model and the real output voltage of the regulator. For angles above

(> 10◦), and output voltage values over 20V errors between simulations and measurements are always below 6%. As expected, for a given value of angle, an increase in the circulation of current through the stator means higher voltage drops in the machine, that brings on a lower voltage at the induction regulator terminals. Dut to this, the Thevenin impedance

68 16

14

12

10

8 10 Error (%) 6 20 4 30 40 ) ° 2 50 α( 60 0 1234567 Load (A)

Figure 3.16: Percentage errors between simulation and experimentation results estimation becomes more important and there is a greater mismatch of the model. By examining no-load situation, where there is only influence of the Thevenin voltage source, the fitness between model simulation and measurements is very precise.

Next the model has been proved for a voltage dip disturbance at several load steps.

This is intended to determine whether or not a specific transient model is necessary for the study of the behavior of the generator prototype. In the operation of the equipment for the production of a severe disturbance two facts can be pointed out:

• The changes in frequency and angle produced in the machines are, most of the time,

very slow. Frequency disturbances in remote microgrids have either a long period or

they have not very pronunced ramps as it has been shown in Chapter 2. In the voltage

dip recovery ramp, the angle modification is also gradual.

• In the deep part of the dip, or the upper value of the overvoltage disturbance, the

induction regulator remains at a constant angular position, i.e. stopped. It can be

69 considered during the greater part of the disturbance, EM1 is operating at a permanent

regime.

Figures 3.17 shows the output of the simulation model in comparison to the output of the induction regulator for two extreme situations previously analyzed —for a determined load and angle—. The first, Fig. 3.17a is in the situation where the machine running with low load (20%) and the second, Fig. 3.17b, with the machine fully loaded (100%). In the

first case the transient at the connection created by the contactors has been also included.

Voltage dip regulator output 300 Uo meas Uo sim 200 ang ref

100

0 Uo(V)

−100

−200

−300 0 0.5 1 1.5 Time(s) (a) a) Induction regulator output of a 20% load case

Voltage dip regulator output 300 Uo meas Uo sim 200 ang ref

100

0 Uo(V)

−100

−200

−300 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 Time(s) (b) b) Induction regulator output of a 100% load case

Figure 3.17: Comparative between the simulation model and the experimental results in the recovery ramp

The adjustment of the response of the model with respect to the actual behavior is very accurate throughout the whole profile of voltage dip except at the first 2-3 connection cycles

70 (≈ 40 ms over a total dip length of several seconds). It can be concluded that the steady- state model is sufficient in almost all the voltage dip time for accurate representation of the behavior of EM2. Despite the availability of the mathematical model in transient regime for the machine EM2, which will be shown in the following section, its implementation in the simulation model for fine tuning in those first few cycles present no interest for the purpose of this thesis. A subsequent model will consider the analysis of the transient effects in the machine by its representation through the finite element method.

3.2.5 Transient analysis

The induction regulator, working within the voltage disturbances generator, suffers a sudden transient at the beginning of the over/under voltage disturbance. At that moment, the stator of the EM2, which was open circuited, becomes connected to the EUT. The EUT can be a voltage source or a Thevenin equivalent circuit in case of a conventional generator, or a current source if a non-conventional source is connected by a converter.

The transient event, created at the induction regulator connection, cannot be eliminated or even mitigated by a device like the one proposed, which lacks power converters. The significance of this work is not rooted in the transient model of the electrical machine EM2.

In spite of this, and due to the fact that the induction regulator is a not a well-known electrical machine, it has been considered of interest the inclusion of the transient model as an additional contribution of this thesis.

In Fig. 3.18a the final circuit to study, according to its representation in the time domain is presented. S is the equivalent switch that simulates the connection of the stator through the contactors. In t<0s there is no current flowing through the stator and the regulator is in

71 Rr jXσr Ir S jX’σs I’s R´s t=0

Ur jXm U’s

a)

Lms i´s Lms ir

r Lr·sL’s·s s

R ı + R´ ir + ı i´s

- - ur UM1(0 ) UM2(0 ) u´s

b) Figure 3.18: Equivalent circuit representation for transient analysis. a) Time-domain. b) Laplace domain a stand-by position. In t=0s it happens the sudden closure of the switch, which will already stay in that position for every t>0s.

An analysis of the transient regime is carried out through the approach and resolution, using the Laplace transform, of the differential equations that model the circuit’s behavior.

Each of the windings can be represented as an inductance initially discharged t=0 in series with a voltage source. Its value is the initial charge of the inductance plus a voltage source dependent upon the current flowing through the other winding. This source symbolizes the existing mutual coupling between the two windings. In Fig. 3.18b it is displayed the equivalent circuit in the Laplace domain, where the initial values of the voltage sources UM1 and UM2 will be as shown in Eq. 3.26: − − UM1(0 )=− (Lσr + Lm) · ir(0 )=−Lr · ir0 (3.25) − UM2(0 )=−Lm · ir(0) = −Lm · ir0 (3.26)

At the beginning, the value of the rotor current is obtained from the resolution in the time domain of circuit Eqs. 3.27.

72 U U i = r = r ∠90 (3.27) r0 · · j Xσr + j Xm t=0− Xr

The rotor will be powered by a source of sinusoidal voltage with frequency ω. This ω will depend on the operating mode of the device, and the supply to EM2 -either from the mains or from EM1-. With this consideration, the instantaneous value of the rotor current for t<0 is expressed according to Eq. 3.28.

Ur ir(t)=− · cos(ωt) (3.28) Xm

With these results, the system of equations in transient regime are shown in 3.29

⎛ ⎞ ⎛ ⎞ ⎛ ⎞

⎜ ur + Lσr · ir0 ⎟ ⎜ Rr + LrsLms ⎟ ⎜ ir ⎟ ⎝ ⎠ = ⎝ ⎠ · ⎝ ⎠ (3.29) us + Lm · ir0 LmsRs + Lss is

The resolution of this system leads to the expressions of the rotor and the stator currents in the Laplace domain, Eqs. 3.30 and 3.31.

(ur + Lr · ir0) · Rs + Lss −·Lms us + Lm · ir0 ir = 2 2 (3.30) (Rr + Lrs) · (Rs + Lss) − Lm · s (Rr + Lrs) · us + Lm · ir0 − Lms · (ur + Lr · ir0) is = 2 2 (3.31) (Rr + Lrs) · (Rs + Lss) − Lm · s

The input voltage ur is considered as sinusoidal and establishes the phase origin. The voltage us it will be also a sinusoidal source with a determined phase lag φs. Laplace transforms for these two power supplies are displayed in Eq. 3.32:

73 Ur · ω Us · [s · sin(φs)+ω · cos(φs)] L [ur]= L u = (3.32) s2 + ω2 s s2 + ω2

Substituting the Eq. 3.32 in Eqs. 3.30 and 3.31 it can be reached the characteristic polynomial shown in the formula 3.33:

p(s)= s2 + ω2 · A · s2 + B · s + C (3.33)

where 2 A=Lr · Ls − Lm B=Rr · Ls + Lr · Rs C=Rs · R s

The roots of this equation are listed in Eq. 3.34, where r1 and r2 are the forced poles, depending on the nature of the supply to EM2, and r3 y r4 are the natural poles of the system.

√ −B ± B2 − 4 · A · C s = ±jω = r1,r2; s = = r3,r4 (3.34) 2 · A

Breaking down the functions ir(s)eis(s) into simple fractions and applying the inverse Laplace transform, two analogous expressions can be achieved for both the rotor —Eq.

3.35— and the stator —Eq. 3.36— currents:

74 r t r t ir(t)=(N1 + N2) · cos (ωt)+j · (N1 − N2) sin (ωt)+N3 · e 3 + N4 · e 4 (3.35)

r t r t is(t)=(N5 + N6) · cos (ωt)+j · (N5 − N6) sin (ωt)+N7 · e 7 + N8 · e 8 (3.36)

N3 y N4 are complex conjugate residues, as well as N7 y N8. In Appendix C it is gathered the detailed calculation of these residues from the parameters of the equivalent circuit of the induction regulator. The real value is(s) would be obtained by simply returning to the original circuit through the transformation ratio K.

3.3 Conclusions

In this chapter the experimental setup used along this dissertation has been described, by analyzing the main commercial components involved, as well as the electric machines that are part of the initial proof-of-concept accomplished with prototype 1, that will serve as a basis for subsequent improvements. It has also been deduced the mathematical models and the equivalent circuits that represent the behavior of the two electrical machines involved in the voltage disturbances generator. This mathematical analysis has been complemented, in case of EM2, with the transient model that could be used in future developments, out of this thesis framework. Subsequently, a simulation model has been established for the operation in permanent regime of the device; it allows the checking of the electromagnetic parameters of the machines, calculated from lab tests. This model also serves as a basis for further study of the device when connecting different types of generators or loads. It has been observed, a high degree of fitness of the simulation model in steady-state operation during the whole frequency disturbance and almost all the dip —except the first cycles—. Due to this, it has

75 been justified that, for the purpose of this thesis, is not necessary to develop a simulation model that specifically represents the transient regime of the full device.

76 Chapter 4

Experimental setup, operational procedure and performance evaluation

4.1 Experimental setup

4.1.1 The electrical machines

A commercial induction regulator has been used for the prototype 1 of the voltage distur- bances generator. With the purpose of establishing the starting point for future designs improvements, in Tables 4.1, 4.2 and 4.3 it is gathered the main information about the magnetic configuration and electrical circuits, used for the leakage inductances evaluation detailed in next section. Extra information and steel and winding diagrams are collected in

Appendix E.

Magnitude Value Description Ur(V ) 400Δ / 690Υ Rated rotor voltage Us(V ) 400Δ / 690Υ Rated stator voltage Ir(A) 7.5 Rated rotor current Is(A) 7.5 Rated stator current P (kW) 5 Rated power

Table 4.1: Main characteristics of prototype 1 Magnitude Value Description Steel V600-50A Non-oriented electrical steel (DIN) l(mm) 120 Core length ks 0.97 sn 1 Rotor skew in slots g(mm) 0.5 Air gap width

Table 4.2: Basic data of prototype 1 magnetic circuit

Magnitude Description Value Rotor Stator P Number of poles 4 4 nr|ns Number of turns per coil 68 29 Nr · ξr|Ns · ξs Turns per winding factor 262.2 289.5 ξr|ξs Winding factors 0.964 0.832 qr|qs Slots/pole/phase 2 3 k Transformatio ratio 1.1

Table 4.3: Basic data of prototype 1 electrical circuit

A second commercial electrical machine has been also used in order to get the first proof- of-concept of the complete device. However, it will be later demonstrated that, due to the operating mode of the device, it is the induction regulator the machine to be improved by subsequent design so the information about the features of the second machine lacks interest and they are not collected in this thesis document.

4.1.2 Control and monitoring system

The experimental setup is based on the employment of DSP-based electric systems. The dSPACE products integrate the MATLAB/SimulinkR environment and provide high per- formance for real-time applications.

The main components of the DSP-based electric systems engaged for the voltage distur- bances generator are the servo motors, the drives, the dSPACE controller boards and the

78 software packages ControlDesk Next Generation and MATLAB/SimulinkR . A schematic view of the control and monitoring system is shown in figure 4.1. The green arrows repre- sent the relationships between components through control signals and the blue ones, the test points where signals are collected for monitoring. In Appendix A is clustered further information about the commercial devices involved in this prototype and their technical fea- tures. The electric machines were used as primitive component of the device and which were subsequently replaced by new designs that will be employed as EM1 and EM2. The charac- teristics and manufacturing drawings of these electrical machines are shown in appendix E of this thesis.

PC dSPACE rti 1104 (2 units) dSPACE connector pannels

Angle drive EM2 Baumüller DS 56 Baumüller b maxx 4400 Speed drive Schneider – Altivar 71

EM1 Geared motor – Pujol Muntalá IPCMIPCM 128/90L-4/148128/90L 4/148

Figure 4.1: Scheme of the electric drive system

• Servo Motors and related equipment

The voltage disturbances generator has two actuators, each of them attached to the

rotor of machines EM1 and EM2. The servo motor corresponding to EM1 is a per-

79 manent magnet synchronous motor. It is equipped with a brake to avoid accidental

movement due to the power provided to EM2. On the machine’s shaft it also placed

an absolute magnetic encoder, to provide feedback about the angle position, used with

control purposes.

An asynchronous geared motor has been chosen for EM1. It allows the control over

speed rotation needed for the frequency disturbances. The control loop is closed by

means of an incremental encoder and its corresponding encoder interface card. By

regulating the initial shaft position, it is achieved a controlled phase jump shift.

The PWM voltage for both servo motors is supplied by the associated drives.

• Angle and Speed Drives

Two independent drives are used to provide the required PWM voltage. The EM2

control drive belongs to the Baum¨uller company. It is a vector controlled based con-

verter which allows an accurate and fast response. The internal PLC is responsible

for the angle control loop. The speed drive is a commercial equipment from Schneider

company.

Both setpoints needed for the drives are provided by two dSPACE DS1104 controller

boards. The setpoints are given as analog signals 0 - 10 VDC, proportional to the

magnitude of the required disturbance.

• dSPACE DS1104 R&D Controller Boards and dSPACE pannels

The dSPACE DS1104 R&D Controller Boards are used to establish the analog setpoint

signals required to create the disturbances. They are also responsible of the control

of the set of electromagnetic contactors through its digital ports. The action that the

controller boards have to take is programmed with the use of the MATLAB/SimulinkR

interface. Between the setup and the controller boards the dSPACE pannels are lo-

80 cated. Analogs signals are adapted by means of sensor boxes and connected to the

pannels via BNC connectors. Sub-D connectors allows the I/O of the digital signals.

• Related software: MATLAB/SimulinkR and ControlDeskR Next Generation

I/O digital and analog ports of the dSPACE panels are accessible from a special

Simulink library. MATLAB/SimulinkR environment allows an easy way to define the

control structure for the controller boards. The SimulinkR program, once compiled,

is transformed in a real-time application implemented in the dSPACEs DS1104. The

software ControlDeskR Next Generation is used to permit the entry in real time to

variables. The development of the HMI has been rooted on a look and feel requirement,

in order to facilitate the interaction of the user with the device interface. This HMI

allows the user to communicate with the boards, to select the desired perturbation, to

change the set-points and to have a complete access to the variables for monitoring.

In Fig. 4.2a) it is shown the general view of the HMI developed for the interaction of

the user with the voltage disturbances prototype.

Figure 4.2: HMI of the voltage disturbances prototype

81 A

D C G B H I E J

F K

L

(a) Outer aspect of the cabinet (b) Inner aspect of the cabinet

Figure 4.3: Voltage disturbances generator cabinet

4.1.3 The electric cabinet

An electrical cabinet to integrate the DSPs, the switchgear and the controlgear devices was built to tie all the elements and compact the final device. As the voltage disturbances generator is intended to be used as a field test equipment that would be transported by a truck if scaled with commercial pursposes, the compactness is a valuable variable that was considered in this equipment. In Figs. 4.3a and 4.3b are displayed the inner and outer aspect of the electric cabinet of the voltage disturbaces generator.

The following pictures show the main elements placed in the front pannel of the cabinet.

The synchronoscope (A) used to synchronize generators to be tested. The power analyzers

(B) and the LEDS indicators (D) permit a quick overview by the user of the main magnitudes.

In the case that any abnormal event happens during the testing procedure, the emergency stop (C) blocks the supply and disconnects the device to test. The selection of the disturbance

82 to be applied can be selected both automatically or manually, by using the selection buttons

(E).

Concerning to the inner distribution of the cabinet, (F) it encloses the switching circuit for voltage disturbances’ relays. The mechanical contactors (G) and (J) are associated to frequency and voltage disturbances respectively. The relays that can substitute the use of the manual selection buttons are (I). The current transformers (H) adapt the measurements to the values allowed by the sensors located in the sensor boxes (L). The output of the sensor boxes is then introduced to the DSP by using dSPACE connector pannels (K).

4.2 Operational procedure

As described in Section 3.1, the voltage disturbances generator prototype is placed between the undisturbed grid and the EUT. The electrical supply to the device is received from the utility grid and at the output, it is created the required disturbance to be applied to the

EUT. The logic associated with the set of electromechanical contactors which are shown in

Fig. 3.1 allows the application of the desired disturbance pattern to the EUT. With the contactors KM1, KM4, KM6-KMT1 and KM7 closed and the remaining opened, a total independence is maintained between the grid and the EUT with respect to the equipment.

4.2.1 Setup of the voltage disturbances generator

The first step to conduct a test with the device is to select, from the HMI, the desired type of disturbance and its characteristic parameters. In case of a voltage disturbance, selectable parameters are the step magnitude, the span as well as the recovery ramp, for both dips and over/undervoltages. If it is a frequency disturbance, multiple patterns as steps, ramps or swing disturbances can be elected. Additionaly, it is also feasible the selection of the initial phase shift.

83 If the test to be performed is a disturbance which requires a modification in the voltage magnitude -and the employment of machine EM2-, the contactor KM6-KMT1 must be initially closed. This is due to the obligation for, at first, separate the induction regulator operation from the EUT, since the voltage at regulator terminals can reach up to 2 p.u. The grid to supply the rotor voltage can be directly the the mains without disturbing, through

KM4 and KM5, or the grid with modified frequency and phase (with KM2 and KM3 closed) and KM4 opened.

If the user selects any disturbance that involves the use of the machine EM1, the contactor

KM4 switches its position to grant the supply to the EUT via the grid with variable frequency and phase. In turn, contactors KM2, KM3 y KM5 also closed. If the EUT is a synchronous generator, KM7 will be automatically closed by a synchronouscope when needed conditions meet. Otherwise, it will be closed with KM2, KM3 y KM5, all at once.

4.2.2 Preparation of the test

4.2.2.1 Calibration

With the preset position of the contactors, the voltage disturbances generator is supplied from the utility grid. At laboratory scale, the undisturbed grid is an with a transformation ratio of 400V/400V.

Prior to the performance of the experiment, the EUT must be calibrated. From Eq. 3.5 can be deduced that a full turn of the induction regulator, i.e., a variation of mechanical angle between 0 ◦ and 360 ◦, the module of the output voltage follows a curve like the one showninFig.4.4.

Two maximum and minimum values are reached in each turn, because both the original machine and the new designs have two pole pairs. Ideally, the minimum voltage value would

84 2.5

Umax 2

1.5 ) . u . p (

e g a t

l 1 o V

0.5

Umin 0 0 50 100 150 200 250 300 350 α(°)

Figure 4.4: Voltage magnitude output during calibration be 0 p.u. and the maximum, 2 p.u. This would happen on the occasion that the emfs of rotor and stator should be fully compensated and the machine should lack internal voltage drop. Faced with the impossibility of that occurring in a practical way, it is indispensable to perform a calibration to calculate the real value of the maximum and minimum voltages, and to store the angles where the extrema are found. These values are the foundations for the entire process of control and regulation of the voltage disturbance. Once the calibration process is completed, the regulator remains in standby, waiting for the control program triggering action. EM1 angle is such that, at the time of connection, it gets in the device output the voltage value settled at the beginning of the disturbance.

4.2.2.2 PI controller tuning

Once the calibration stage is completed, the PI controller comes into operation and places the induction regulator in the angle corresponding to the initial voltage of the disturbance to be created. For example, if it is a voltage dip, it will correspond to the value of the dip depth. When the order to create the voltage dip is given, it takes place a very sharpen decrease of the voltage in terminals of the EUT until the preset voltage in the regulator output is reached.

85 The global system is highly non-linear. The reproduction of the different patterns de- mands the system the operation in wide ranges. The PI controller was designed by adding an anti-windup scheme to avoid the saturation of the integral term, thus reducing sudden transients due to this effect. Among the different anti-windup classic schemes collected in the literature [74] the back-calculation and tracking method was selected. If there was a sat- uration effect, the normal calculation path would break, and the additional feedback signal, multiplied by a constant kb, would prevent the system to work in open loop. The distur- bances occur in a sudden step in reference, which seriously affects the performance of the

PI controller. The step in the reference leads to a boost in the control signal. To avoid this effect the reference signal was weighted with a constant kw. As it is a proportional action it only affects to a fraction of the reference so the overshoot in the system response is highly reduced.

The adjustment of the proportional (kp) and integral (ki) terms of the controller should be done according to the current that is going to circulate between the EUT going through the voltage disturbances generator. The device introduces an impedance between the grid and the EUT -the stator impedance of the induction regulator- and its power is similar to the power of the EUT. If it cannot be guaranteed that the short-cirtuit power in the PCC

(KM1) is 5-times bigger than the power of the EUT, the PI controller adjustment has to be done with the rated load for the specific test. This requirement is actually prescribed in [54] for type 1 voltage dip generators. An impedance also remains in this topology between the grid and the EUT so the rule may be extrapolated to this new design.

4.2.3 Test procedure

Once it has been configured the device for a voltage dip or an over/undervoltage disturbance, the user gives the order for the test execution. At that moment, contactor KM6-KMT1 opens and simultaneously KMT2 closes. The resistance Rlim prevents the short-circuit associated

86 (a) Voltage Swell (b) Voltage Dip

Figure 4.5: Reproduction of voltage swells and dips to the open-close transient. An moment later KMT4 closes and opens, with the aim of provoking a very brief short-circuit with a resistance Rcc. This results in an initial sharp drop of the RMS voltage value if the disturbance is a dip. When contactor KMT4 opens,

KMT3 closes and bypasses the Rlim. Since that moment, the EUT is connected to the output of the induction regulator and the angle control will result in a continuous change of the voltage applied to the EUT, and, as a result, the selected voltage dip pattern will be created. If the EM1 machine is connected in cascade, it will also be applied the corresponding frequency and phase jump disturbances.

4.3 Performance evaluation of the prototype

4.3.1 Voltage disturbances

The proposed device is capable of emulating all voltage disturbances described in IEEE Std.

1159-1995 and appearing in the Fig. 2.12 of the previous chapter. Therefore it serves as certification equipment for compliance with the requirements of both LVRT and HVRT. As an example, Figs. 4.5a and 4.5b show, respectively, a voltage swell and a voltage dip with rectangular profiles.

87 In both figures it has been displayed the voltage set-point and tracking curves followed by the EM2 machine. It has also been included, in Fig. 4.5a and in a demonstrative way, the value of the voltage in the induction regulator output, plotted with a red line. The induction regulator, after the calibration process finishes, is positioned to reach the initial value of the disturbance set by the user (in this case 1.1 pu). According to the operation procedure at the time the electromechanical contactors act, the device to be tested changes from being supplied from the mains to be connected through the regulator. Hence the blue line represents the resulting voltage applied in EUT terminals.

Regardless to the rise time and the overshoot, it is observed that there is a delay in the reference tracking of approximately 100 ms. This delay is due to the lag in the response of the commercial angle controller (internal EM1 control loop). It appears as a systematic error independent of the control programming developed in this thesis. To validate this claim, the Fig. 4.6 shows the response of the inner control loop EM2 where the angle set-point generated and the absolute magnetic encoder measurement are compared. The delay, consequence of the performance of commercial regulators, it will also appear in the reproduction of frequency disturbances. As it is a device for testing and certification to be used offline, speed of response is not a priority requirement. Note that this systematic error occurs in both regulators and so it does not affect the final disturbance applied to the EUT, which will have the extent set by the user as defined in each grid code.

One advantage that has been highlighted in this device is that it is capable of reproducing different patterns -and ramps- in diverse grids code due to the possibility of regulating the voltage continuously. In Fig. 4.7 it is shown the profiles used in this dissertation as reproduced by the preliminary device. As a complement of the patterns found in the grid codes, shown in Chapter 2, it has also been defined an additional profile called ISOL,

88 Figure 4.6: Angle setpoint and tracking representing the envelope of all isolated systems profiles analyzed, with a depth of 0 during

650 ms followed by a recovery time of 3s to reach 0.75 p.u. of the pre-fault value.

Figure 4.7: Voltage dips profiles

4.3.2 Frequency disturbances

The machine EM1 allows the initial positioning of the phase in the device and the reproduc- tion of the frequency disturbances, with different shapes and up to a value of ±5 Hz. This

89 range is sufficient for testing according to all the grid codes, for both interconnected systems and isolated. Three general profiles, corresponding to the most common frequency distur- bances types have initially been preprogrammed and sample set-points and trackings are shown in Fig. 4.8. The steps of different value and duration can appear due to unscheduled events, ramps allow EUTs validation against the derivates of frequency established in the codes, and over and under frequency oscillations could represent the frequency disturbances resulting from voltage sags. Any other profile could be easily programmed if needed.

Figure 4.8: Frequency disturbances

4.3.3 Combined disturbances

A combined disturbance reproduced by the voltage disturbances generator is shown in Fig.

4.9. In the lower subfigure it can be seen the phase jump naturally occurred as a result of the fault, by the change in the circuit impedances. This angular difference coming from the operation of the induction regulator must be combined with the EM1 machine offset to get the phase output required for testing in compliance with grid codes.

90 Figure 4.9: Combined voltage, frequency and phase jump disturbance

4.3.4 Harmonic behaviour

Apart from the simultaneous reproduction of disturbances, one of the highlighted features of this device is that it does not introduce additional harmonic distortion on the EUT. To check and verify this, it has been represented in Fig. 4.10 the spectral analysis of the input and output voltage waveforms for harmonics up to order 20th. It is easily seen that very similar magnitude peaks appear at harmonics of the same order.

(a) Harmonic spectra of the input voltage (b) Harmonic spectra of the voltage applied to the EUT

Figure 4.10: Harmonic spectra of voltage waveforms

91 4.4 Conclusions

In this chapter has been described the main topological features of the voltage grid distur- bances generator. These aspects include the physical description of each one of the main components in the equipment: the electromagnetic devices and the essential control system elements. Subsequently, it has been depicted the operation practice and testing procedure with the device. In the last part of the chapter, the preliminary device has been proved and verified as a certification equipment capable of reproducing voltage disturbances, frequency changes and phase jumps required by the codes. It has also been checked that the device introduces no additional harmonic distortion over the existing pollution in the grid. Ap- pendix A complete the information gathered in the chapter, detailing the characteristics of the commercial equipment involved.

92 Chapter 5

Guidelines for the design of new prototypes

5.1 Problem approach

5.1.1 Overview

In Chapter 3.1 the composition and operation procedure of the voltage disturbances gene- rator proposed as a global target of this thesis has been described. The use in the original prototype of commercial machines has allowed a first conceptual validation of the device.

In accordance with the EM2 electrical machine operation, its output is the difference between the rotor and stator phasor voltages. Theorically, if those voltages were perfectly balanced for an angle, 0◦, it would be possible, if desired, to get a zero voltage in the EUT’s point of interconnection .

The EUT connected in the voltage dip generator will impose a fixed current. This current corresponds to the load level established by the verification procedures linked to the grid codes. In the Spanish regulation currently in use for mainland territory, the PVVC 12.3

[54], the load levels for testing the electrical devices are settled to 10% − 30% of the nominal power of the EUT —partial load— and over 80% of nominal power —full load—. In other territories, like Great Britain [75], the tests have to be done with the EUT working at full load.

In the operators procedures for the verification of distributed energy sources connected to grids, the suggested testing equipment belongs to the impedance-based type. These are rooted on the primary model as defined in IEC61400-21 [76]. The document PVVC is not so specific, referred to the possibility of other devices to be used in the certification process.

The only condition is that the behavior of both devices must be comparable.

The flow of the generator current demanded for grid code compliance by the Thevenin equivalent impedance of the EM2 machine causes a voltage drop. This voltage drop estab- lishes a minimum voltage dip depth that can be reached, with independence of the profile defined by the user. In agreement with PVVC, if the short-circuit power in the PCC is below

5 times the power injected by the EUT, the voltage dip profile has to be achieved with the

EUT operating at full load. This is the situation applicable to remote microgrids, which are characterized by a low short-circuit power in the PCC. Not specific verification standards have been found for remote microgrid codes.

The problem can be that the lower value of the voltage level in EM2 machine working at full load could be higher than the required by the code. As an example it is shown in

Fig. 5.1a a voltage dip profile created by the device with a zero voltage output reference at a no-load regime. A minimum value of voltage remains in the output due to the current circulation by the rotor winding.

Figure 5.1b shows the test of a synchronous generator under a voltage dip pattern accor- ding to the exising P.O.12.3 procedure. It can be observed that the current flowing for both the inductances of the synchronous generator and the induction regulator stator effectively

94 Ref = 0%

Ref = 20%

(a) Voltage dip at no load and at zero voltage setpoint(b) Voltage dip with synchronous generator at rated power

Figure 5.1: No-load and full load voltage dips with the prototype 1 avoids the voltage to reach the 20% imposed by the user. Even with the test of a synchronous generator it cannot be separated the effect coming from each inductance, the improvements in new designs will be compared for the same EUT.

In order to regulate effectively the voltage at the output of the device regardless of the load that is connected, it is necessary to minimize the internal drop in EM2. This voltage drop is intended to be reduced by design considerations, developing a more appropriate machine for the proposed operation. This will favor the independence of the voltage disturbances generator operation from the EUT.

5.1.2 Voltage drop evaluation of the preliminary design

The internal voltage drop in an electrical machine is defined as the difference between the voltage output at rated power with respect to the voltage in the no-load case. It matches with the input voltage minus the drops in the resistances and leakage inductances of the windings. The circumstances are sligthly different in the induction regulator. The voltage drop of the machine is coincident with the drop in the Thevenin impedance which is a huge value, in magnitude, in comparison with the drops in, for example, a transformer.

95 One phasor diagram extracted from experimental measurements has been represented in 5.2. The black axis apply to the voltages while the blue axis correspond to the current magnitudes. The variables with the subscript “nl” designate the magnitudes associated to the no-load case. It shows the voltage drop in case of a synchronous generator operating at a rated power for a steady angle between the rotor and stator phasors.

Figure 5.2: Phasor diagram of the prototype 1

It can be clearly observed from Fig. 5.2 how the current circulating by the stator decreases the stator phasor voltage, according to the reference system defined in Chapter 4. Thus, the output voltage is higher than the one at no-load.

A numerical computation has been carried out for the loaded and non-loaded cases pre- viously shown.

Uo = 167 58.64 ; Uo−nl = 179 68.3

And expressed in percentage, the internal voltage drop value is, from Eq. 5.1:

96 |U − U − | o o nl · 100 = 7.18% (5.1) Uo

The voltage drop depends on the angle α between the rotor and stator voltages and the current circulating due to the EUT. As an example, it can be observed in Fig. 5.3. In there, a spectrum of output voltages for several angles and currents ranged from no-load to full load is shown. It has been evaluated from the simulation model and allows to check the behavior of the induction regulator while testing converter-based generating sources, working at cosφ = 1, as the usual power factor requirement for sources connected to the grid through electronic converters.

Figure 5.3: Output voltage at the induction regulator for a generator with cosφ =1

This confirms the hypothesis that the limit is determined by the internal voltage drop of the machine, and depending on the operating point, sets a settle depth that cannot be surpassed. It limits the capacity of the voltage disturbances prototype to fulfill perfectly the deeper voltage dip profiles required in some grid codes.

97 5.2 Minimization of the voltage drop in the induction

regulator EM2

In agreement with the equations obtained for the Thevenin equivalent of the induction regulator (see Eq. 3.21), it is clear that the reduction of the voltage drop in the machine can be achieved as a result of the simultaneous Thevenin source and equivalent impedance minimization. The expressions will be deduced without considering the resistances of the windings. Even if this is not real, it will allow to draw approximate conclusions.

The minimization of Thevenin voltage in these circumstances is obtained in fulfillment of the condition —Eq. 5.2—:

j · k2 · X · ejα m ≈ 1 (5.2) j · Xm + j · Xσr

which results in a relationship between the rotor leakage inductance, the transformation ratio and the mutual inductance, shown in Eq. 5.3:

Xσr ≈ Xm · (k − 1) (5.3)

On the other hand, and due to the need of a voltage drop reduction in the series Thevenin impedance, the objetive function to be minimized in the new induction regulator designs is

Eq. 5.4:

2 Xth = Xσr · Xσs + Xm · (k · Xσr + Xσs) = 0 (5.4)

98 From here it can be reached the expression of the stator leakage reactance 5.5:

2 −Xm · k · Xσr Xσs ≈ ≈ Xm · k · (1 − k) (5.5) Xσr + Xm

All the terms involved in Eq. 5.5 are strictly positive so the only feasible solution to assure the accomplishment of the equation is by making null the rotor leakage reactance and, as a result, the stator leakage reactance too. The Eqs. 5.3 to 5.5 obtained from this simplified analysis suggest that a transformation ratio equal to unity would turn in a minimum value of the rotor and stator leakage reactances.

If rolling back to the Thevenin impedance and using Eq. 5.3, it is explicit than the

Thevenin reactance can be expressed as a function of the stator and rotor ones, as displayed in Eq. 5.6, thus the minimization of both leakage reactances implies the global Thevenin reactance minimization.

Xth ≈ Xσs + k · Xσr (5.6)

Conforming to the previous, the solution adopted for the new designs is based on a two-step improvement conducted in as many prototypes. It has to be noticed that one of the design targets has been to make the required improvements in the prototype 1 trying to keep the design as close as possible to commercial machines, not introducing additional complexities for future large-scale manufacturing.

• In the second prototype (Prototype 2) the magnetic circuit in the machine will be designed in order to reduce the leakage inductance by acting over the diverse leakage com- ponents. It will also have a drastic impact over the magnetizing inductance.

99 • In the third prototype (Prototype 3) the magnetic circuit will be kept as in the previous but a change in the design of the electric circuit will be done to reduce the leakage inductance through the transformation ratio.

5.3 Criteria design for new prototypes

5.3.1 Prototype 2: Leakage reduction by redesigning the magnetic

circuit

All the inductances along this section are specified per unit of effective length (le)and pointed out with ’. According to the basic sizing equations of electrical machines design, 2 the power of EM2 is directly proportional to the product of the rotor inner diameter (Dor) by the effective length. The power of the original prototype (prototype 1) is going to be maintainedas a design criteria in the new induction regulators constructions, to make the machines easily comparable.

5.3.1.1 Mutual inductance

The mutual inductance for the machine EM2, magnetized from the rotor winding is presented in 5.7:

48 2 2 μ0 · Dor Lm = · ξr · Nr · (5.7) π ge

100 Nr is the number of series-connected turns in the rotor winding, Dor is the rotor outer diameter and ge is the effective gap, which takes into account the fringing effect and the core saturation.

5.3.1.2 Leakage inductances

5.3.1.2.1 Flux leakage components in wound rotor induction machines

The leakage inductance in induction machines represents the portion of the flux created by the primary winding which is not linked by the secondary circuit.

The calculation of leakage flux components in electrical machines suppose a complex problem. To estimate the values of leakage flux reactances it is required to know the magnetic permeance, which depends on the circuit geometry. The existence of analitycal expressions or tabulated parameters grant a numerical approach.

In induction machines, several leakage flux components can be found [77]. The total leakage will be the sum of all these items gathered in the following list.

• Slot: It is produced by the flux that goes from one tooth to the next through the slot

but without crossing the airgap. This leakage flux appears inside every slot but also

between contiguous slots. In Fig. 5.4 it is shown the slot leakage flux lines associated

to this leakage component flux:

• End-winding leakage flux: it corresponds to the portion of leakage flux derived from

the circulation current in the end-winding, as seen in Fig. 5.5. This leakage flux runs

almost exclusively by a circuit of air. As the computation of this component is difficult

due to the geometry, it is usually calculated by applying empirical correction factors.

101 Figure 5.4: Slot leakage flux

Figure 5.5: Leakage flux paths in the end-windings

• Zigzag: It is the leakage flux that passes from one tooth to another through the airgap,

as observed in Fig. 5.6:

Figure 5.6: Zigzag leakage flux paths

• Harmonic leakage flux: The difference between a quasi-sinusoidal emf, due to the

distribution of the winding, with respect to a perfect sinusoidal waveform, favors the

102 production of harmonic fluxes which are not linked in the same way by stator and

rotor. In squirrel-cage induction machines, the rotor harmonic component would be

null.

• Skew leakage: It appears when there is a skew in the rotor or stator slots. This skew

is intended to reduce the high order harmonics that appear due to slotting. The main

difference in comparison with the other components is that it does not appear as a

natural phenomenon but it is caused by the effort to compensate the pulsating torques

linked to the harmonics components of the mmf.

5.3.1.2.2 Mathematical expressions of the leakage inductances in EM2

For both the rotor and stator windings, the total leakage inductance is the sum of the slot, end-winding, zigzag, harmonic and skew leakage components. The skew effect is directly computed through the winding factor.

• Slot leakage inductance

For the single-layer winding in the rotor, the leakage inductance in terms of the number of slots (Sr), the number of turns in the winding (Nr) and the permeance of the magnetic circuit (pr) is shown in Eq. 5.8. The deduction of these equations is based on the theory of [77, Ch. 4].The mathematical equations for the permeance of the slots for the shape of interest has been developed and shown in Appendix D, as well as the geometric dimensions required for its evaluation.

· 2 · pr Lrslot =12 Nr (5.8) Sr

103 In the double-layer winding of the stator, the total permeance is the sum of several components, one due to the lower coil (pL), another due to upper coil (pU ) and a third one 2 that takes into account the mutual effects between both coils (pLU ). For a pitch of p = 3.

· 2 · (pU + pL + pLU ) Lsslot =3 Ns (5.9) Ss

• End-winding leakage inductance

The end-winding leakage inductance is figured out through the use of permeance factors.

These factors vary depending on the winding type. For a three-phase two plane distributed winding, the permeance factor associated to the axial length is few =0.55 and the factor related with the coil span is fW =0.35 [78, Ch. 4].

Figure 5.7: End-winding main dimensions

Conforming to the end-winding dimensions shown in Fig. 5.7, the end-winding induc- tances for the rotor and stator can be written as in Eqs. 5.10 and 5.11, where qr and qs are the slots per pole per phase in the windings.

2 12 · qr · Nr · μ0 · lw · fw Lewr = (5.10) Sr

104 2 12 · qs · Ns · μ0 · lw · fw Lews = (5.11) Ss

The product of the average length in the winding and the permeance factor is obtained through the Eq. 5.12:

lw · fw =2· Lew · few + Wew · fW (5.12)

• Zigzag leakage inductance

The zigzag leakages inductances have a direct relationship with the mutual inductance as displayed in Eqs. 5.13 and 5.14 [79, Ch. 4]. It is associated with the slot harmonics so it is also a function of the number of slots and the number of poles (P ). A second comma has been included in the notation for the stator. The first one, represents, as usual along this chapter, the mutual inductance per core length unit and the second one, refers, as in he previous chapter to the evaluation of the mutual inductance as seen from stator-side.

2 2 · π · Pr Lzzr = Lm 2 (5.13) 12 Nr

2 2 · π · Ps Lzzs = Lm 2 (5.14) 12 Ns

• Harmonic leakage inductance

Eventually it is shown the expressions for the leakage inductances due to the harmonic

fluxes created as a consequence of the distributed windings. It considers the different dis-

105 tribution of the current circulating in the diverse coils, so it reckons on the coil pitch. In a simpler formulation, the harmonic leakage inductance can be evaluated depending on the mutual inductance and a tabulated factor (σ) based on the number of slots per pole and phase and the phase belt of the winding. Both equations for the rotor and stator are displayed in

Eqs. 5.15 and 5.16 [77, Ch. 4].

∞ 2 τ 2 3 · 8 · Nr · · pr · ξhr · Lhr = μ0 = Lm σr (5.15) 2 π2 Pr g h e h=2

∞ 2 τ 2 3 · 8 · Ns · · ps · ξhs · Lhs = μ0 = Lm σs (5.16) 2 π2 Ps g h e h=2

In these equations, ξhr and ξhs represent the winding factors for the h harmonic and τpr and τps, the pole pitches.

5.3.1.2.3 Validation of leakage inductances in the prototype 1

The validation of the leakage reactances for the EM2 machines has been accomplished through the use of a commercial finite-element analysis software, since in lab tests it is not possible to separate the different components.

Three components are shown in Table 6.1 concerning the results in simulations: slot leak- age reactance, end-winding leakage reactance and spread harmonic reactance. The spread harmonic reactance —named as differential in most literature references— includes the two leakage reactances affected by harmonic components. The zigzag leakage reactance, pro- duced by the harmonics linked to the slotting and the harmonic leakage —or belt leakage— associated to the space harmonics. The evaluation by means of analytical formulas is usually

106 less exact than the evaluation by finite element software, but it allows a first and approximate appoach to the problem. Additionally, by applying the mathematical formulas, the different leakage components can be estimated separately. In the third column it is registered the leakage reactance value derived from the lab tests where the different leakage contribution terms cannot be distinguished so it can only be compared in terms of total leakage fluxes.

Simulation (Ω) Calculated (Ω) Lab tests (Ω) slot 3.031 2.702 end-winding 3.458 3.811 Stator 1.816 1.762 differential harmonic 1.759 zigzag 0.003 Total 8.305 8.275 8.093 slot 4.494 3.808 end-winding 2.994 2.536 Rotor 4.315 4.176 differential harmonic 4.171 zigzag 0.004 Total 11.803 10.520 11.650

Table 5.1: Electrical machine parameters

Several points can be highlighted from the analysis of the results shown in table 6.1

• The harmonic inductance in the rotor is higher in comparison to the stator. When increasing the slots per pole and phase, the mmf in the air gap becomes closer to a sinusoidal waveform, reducing the harmonic content.

• Most of the harmonic flux produced in the machine comes from the space harmonics

(pitching) and it is very reduced the harmonic flux component due to the slotting.

107 • It has also been accomplished the calculation for the slot leakage reactances from the mathematical expressions developed for the slot shape and shown in Appendix D. It has to be taken into account the permeances estimation from the formulas trust on the fraction of conductors distributed in every portion of the slot, which is obviously unknown. The difference between the real placing and the foreseen distribution of the conductors in the slots can lead to the variations observed.

• The differences between the calculated parameters and the results obtained by finite element algorithms for the end-winding leakage flux were expected due to the influence of the geometric parameters, which makes the estimation complicated. An exact evaluation of the end-winding leakage components, would imply the use of 3D finite elements models, since the distribution of the end-turn leakage fluxes implies an analysis in three axes. Due to the computational effort to develop 3D models for the end windings, it is commonly accepted the precision provided by the tabulated coefficients.

• It can be seen that the global reactance evaluated by the finite element model matches properly with the lab experiments.

5.3.1.3 Constants and variables in the magnetic design

In the second prototype it will be only modified the parameters linked to the magnetic circuit, leaving aside the transformation ratio. From the equations shown along the section, several variables can be selected to reduce the leakage inductance associated to the prototype.

The suitable parameters are listed below.

• Effective length: le

108 • Rotor inner diameter: Dor

• Slot permeances: p

• Effective airgap: ge

• Slots per pole and phase: q

• Poles: P

• Number of slots S

From those shown in the some of them it will be kept as constants regarding the original prototype. The effective length and the rotor outer diameter determine the power of the device designed, because the machine is magnetized with the rotor which acts as primary winding. Since the power of the new devices has to be maintained because it is linked to the power of the EUT, these two parameters will also be kept as constants. But the inverse relationship between the mutual inductance, and, as a consequence, with the harmonic leakages and the zigzag leakage inductances suggest the need to increase the airgap. This will be done by extending the stator inner diameter.

The increment of the airgap supposes a novelty and a contradiction over the conventional electrical machines designs, that try to maintain the airgap as small as possible. An increment in the airgap acts over the leakage flux components by reducing them, opposing an extra obstacle to the closing of the leakage flux paths between the rotor and the stator. It has also to be remarked that an increase of the airgap will upturn the reactive current required to magnetize the machine, inflates the Joule losses and, from that point of view, has a negative effect. It will be essential to establish a design compromise between the airgap enlargement and the rotor current in the new designs.

109 In the prototype 1 the relationship between the number of stator and rotor slots is one of the classical combinations established for motor design to avoid vibrations. A greater number of slots in prototype 2 would reduce the slot leakage inductance and would shorten the end windings, minimizing the effect of the end-turn inductances. A bigger number of slots would also make possible to increase the number of slots per pole and phase, which is positive in order to reduce harmonic leakage fluxes. The combinations 36/36 and 48/36 where tested during design attemps by the finite element model. Since these combinations would be prohibited for standard motor designs due to the cogging during start up, the movement of EM2 thanks to a servomotor avoids the uncertainty. The unacceptable saturations levels reached in the machine teeth made recommendable to keep also the number of slots and the q parameter as in the prototype 1.

Eventually two parameters where selected to improve the original design: the air gap and the slot permeances. In the slot permeances, several geometric parameters are involved. In a first approach, the effect of changing the different parameters was studied. This is only an approach since the restrictions such as saturation limits, that will be taken into account in the design process developed in next chapter, were not considered. In Figs. 5.8a and 5.8b it is shown the values of the slot permeances in the rotor and stator when changing any geometric dimension from its rated value to another one 50% higher.

Figure 5.8 shows the influence of the tooth and slot design in the slot permeances values where it can be observed that the widths parameters decrease the slot permeances while the heights increase them. The slot opening is the most influential variable if trying to reduce the stator permeance. An increase in the parameters b1 and the b0 would also suppose a noticeable reduction in the rotor slot permeances, with slight differences between the effect that every one can produce. It is common in motor design processes to establish a maximum value of the slot opening to keep the windings correctly placed inside the slots while the

110 b 1 b 0

d0 d1

d2

d3

2ro=b3

(a) a) Variation of rotor slot permeance due to changes in geometric magnitudes

b1

b0 d0

d1

d2

b2

b3 d3

d4

b4

(b) b) Variation of rotor stator permeance due to changes in geometric magnitudes

Figure 5.8: Influence of the geometric parameters over the slot permeances machine is spinning at high speed. An abnormal value of slot openings in the new prototypes will be the second unusual characteristic in these designs —apart from the wider air gap—.

Since this electrical machine is going to be stopped or spinning at a very low speed, this will not represent a handicap.

111 5.3.2 Prototype 3: Leakage reduction by redesigning the electric

circuit

5.3.2.1 Optimization of the ampere-turns ratio

As deduced in the first part of the chapter a transformation ratio in the machine equal to unity is theoretically required to minimize the voltage drop in the machine.

Since the number of turns is bigger in the stator winding, the change in the turns will be done over this electrical circuit. A reduction in the number of series connected turns is positive to reduce the leakage components associated to:

• slot

• harmonics

• end-leakage

The zigzag leakage flux will be bigger in the prototype 3 with respect to the prototype 2, but due to the drastic reduction in the mutual inductance, it will be in both cases, noticeably smaller that in the original prototype. If considering the weight of the zigzag over the global leakage reactance, this can be considered negligible.

The second rotor winding will be also designed as a double-layer winding. Since q pa- rameters and slots will be kept as constants -not affecting the end-leakage and the harmonic components-, it will suppose and additional cut back in the slot leakage component.

112 5.4 Conclusions

In this chapter the problems associated to the operation of the original voltage disturbances generator have been described. The circulation of current by the stator inductance joined to the magnetic couplings in the induction regulator establishes a minimum voltage level that cannot be exceeded. This level fixes the dip profiles that can be reproduced.

It is required for testing devices being independent in its operation from the EUTs.

In every voltage disturbances generator commercially available this affection between the testing device and the EUT can be detected. It also happens in this new topology.

Along the chapter it has been analyzed and quantized the effects of some magnetic and electrical magnitudes over the equivalent Thevenin impedance of the induction regulator and thus, over the behavior of the device. It has been deduced the dependence of this voltage drop with the leakage inductances and the number of turns in the windings. Eventually, and rooted on the previous, the guidelines for new designs have been settled according to a detailed study of the leakage components.

113 114 Chapter 6

Sizing, construction and assembly of the new prototypes

6.1 The finite-element methods (FEM) for electrical

machines design

The finite element method is currently very used in electrical machines design. It allows a precise real-time study of the electromagnetic behavior, considering the non-linearities and saturations. It is also widely flexible with independence of the geometry, the materials or the supplying sources. The method is based on a discretization process where the solution of the Maxwell equations in a complex domain is simplified until it is transformed in the addition of multiple simple solutions.

The objective of the application of the FEM techniques concerning to the interest of this dissertation is the definition of the specifications for the electrical machines that are part of the voltage disturbances generator. The modeling with this method allows the simulation of the expected behaviour of the machines in their usual operating conditions in order to define the final electromagnetic designs. In the literature it has been extensively gathered a number of examples that prove the results obtained by FEM software faithfully representing the reality. Every finite element process can be considered as to be done in three steps:

• Pre-processing • Analysis • Post-processing

The first phase is the definition of the geometry from the ad-hoc design toolbox included in the commercial FEM packages or imported from any other design program. The application of symmetries can noticeable reduce the problem sizing. Then it is assigned the materials, the excitations, from internal sources o through an external connection circuit —if required— and the motion. For running the analysis, it is required to define the mesh. The generation of the mesh is the most important step when applying the FEM methodology. It has to be defined over the basis of a trade-off between accurate enough results for the solution of the problem and the computation time required. Finally, and depending on the study purposes, it is selected the type of study (Electrostatic, Magnetostatic, Transient...). The results of computations, including parameters, field distributions, vector plots, flux lines, etc, can be further explored in the post-processing step.

To study the voltage generator disturbances by means of the FEM software, it is repre- sented the device with the help of an additional software package. This package allows the connection of external components to the electromagnetic design model of the machine. It adds adaptability to the model by permitting multi-technology co-simulations. The connec- tion of the induction regulator with a common point requires the use of this external tool due the need to have accesibility to the inputs and outputs of the windings. This model will be used for the steady-state analysis as well as for transient regimes —voltage dips—.

116 6.2 Definition of new prototypes sizing by finite ele-

ment analysis

The design process was accomplished by subsequent improvements. They were listed in

Chapter 5 and repeated here to focus the attention over them. The starting point was the redefinition of the slot shapes, for both rotor and stator magnetic circuits. Once the new slot shape was defined, it was studied the effect of the air gap modification. These two changes in the magnetic circuit were applied to the design of the second prototype. Eventually, the changes over the electrical circuit were introduced in the third prototype. These changes were the modification of the ampere turns ratio to make it as close as possible to the unity, as well as the transformation of the rotor winding in a double-layer configuration.

6.2.1 1. Slot design

The slot shape was finally defined by considering the expected changes in leakage inductances due to slot opening values.

The contour plots shown in Fig. 6.1 and Fig. 6.2, based on FEM analysis, display the ranges of values for the total leakage reactances according to the slots openings, from the starting width to the selected geometric value. Over that maximum, the teeth of the machine are highly saturated introducing an unnecessary inflation of the saturation levels that would boost the iron losses.

Even the study about the influence of the different parameters over the slot permeance shown in Chapter 5 suggests that an increment over the magnitude b1 in rotor would imply a higher effect over the slot leakage reduction —see Chapter 5, Fig. 5.8a—, this is only an

117 approach that does not consider the combined effect due to multiple changes. The more profound analysis accomplished by finite elements confirms that an additional increase of b1 magnitude with regards to the rotor slot would have an stronger effect in slot permeance reduction. The quantification of the effect in rotor slot leakage reduction —if the magnitude was increased in 1 mm— hints an additional reduction of around 4%. In terms of the new design, that increment would be translated into a narrowing of the teeth at b1 height that would generate points where the magnetic flux would concentrate, creating hot spots in the machine. As a consequence, the b1 modification has been rejected in terms of width parallelism maintenance.

By comparing the original design, in the simulation environment, with regards to the new prototype, and if only considering the modifications leakage inductances in this first step, it is expected to achieve a reduction over the slot leakage reactances of 20.716% in the rotor and 28.07% in the stator.

7

8.249

6.5

6 10.179

5.5 7.863

10.565

5 stator (mm) 0 10.9 b 4.5

51

4 9.793 9.407 9.02 8.635

11.337 1

3.5

3 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 b rotor (mm) 0

Figure 6.1: Rotor Leakage Reactance contour plot(Ω)

Table 6.1 summarizes the expected improvement in leakage reactances for the new pro- totypes due to the refitting of the slots.

118 7 5.9773 5.9773 6.5 6.2195

6.2195 6 6.4618

6.4618 5.5 6.7041

5 6.9464 6.7041 stator (mm) 0

b 4.5 7.1886 6.9464

7.4309 4 7.1886 7.6732 7.4309 3.5 8.1577 7.6732 7.9155 3 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 b rotor (mm) 0

Figure 6.2: Stator Leakage Reactance contour plot (Ω)

Rotor Stator

b0 (mm) b0 (mm) b0 (mm) b0 (mm) 3.6937 slot 4.494 3.563 3.031 1.842 Leakage end-winding 2.994 2.994 3.458 3.458 differential 4.315 1.578 1.816 0.68

Table 6.1: Comparison between leakage reactances in prototype 1 and prototype 2 according to changes in the slot openings

It has to be noticed that the increment of the slot opening widths (b0) has an expected effect over the slot leakage, but also it has a noticeable consequence in the harmonic leakage reactances, which would be reduced in a 63.03% in the rotor and in a 62.51% in the stator.

This is as a consequence of the increment in the effective air gap.

The effective air gap takes into account the slotting in the machine through the multipli- cation of the mechanical air gap by the Carter factor, where τr is the slot pitch and g,the air gap width. An increment of b0 parameter in the slots leads to a cut of the Carter factor. The modification of the Carter reduces the mutual inductance and the magnetic flux in the air gap so in consequence, the harmonic fluxes responsible of the harmonic leakage reactance

119 too —see Chapter 5, Eqs. 5.15 and 5.16—. The variation of the Carter factor as a function of the rotor opening is plotted in Fig.6.3.

Figure 6.3: Carter coefficient variation due to rotor slotting

As a summary of the proposed changes for the prototype 2 design, the expected effect over the total leakage reactance is a decrement of 31.96% in the rotor and a 28% in the stator.

6.2.2 2. Air gap

According to the conclusions of Chapter 5, an increment in the air gap has a positive effect over the harmonic leakage reactance. Even the action over the slot openings has demon- strated an appreciable impact on the harmonic leakage, an extra cutback can be achieved by increasing the air gap length.

This modification will be slight in order to find a compromise between the reduction of the harmonic leakage inductance while avoiding an excessive increment of the rotor current consumption for magnetizing the machine. With this purpose, a parametric analysis was ac- complished for steady-state conditions. In Fig. 6.4 it is shown the variations of the harmonic

120 2 90 5

1.8 4.8 85

1.6 4.6 80 1.4 4.4 ) Ω 75 ( 1.2 4.2 hs ) X Ω (A) ( ) 70 4 1 m m I Ω X (

hr 3.8 X 0.8 65

0.6 3.6 60 0.4 3.4

55 0.2 3.2

0 50 3 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 g (mm)

Figure 6.4: Variation of main parameters affected by air gap width reactances for the rotor and the stator, the magnetizing reactance and the magnetization current as functions of the air gap width.

It can be deduced from Fig. 6.4 that the decrease of the harmonic leakage reactances is smaller in comparison with the magnetizing reactance, that is clearly reduced in a drastic way. A final air gap width of 0.8 mm was defined in order not to overpass the magnetizing current value in more than a 50%.

With this second step, the results for the calculated leakage reactances are shown in table

6.2. With the proposed air gap width, a 22% of improvement was achieved for the rotor and a 28% for the stator.

121 Rotor Stator g (mm) g (mm) g (mm) g (mm) 0.5 0.8 0.5 0.8 slot 3.563 3.563 1.842 1.842 Leakage end-winding 2.994 2.994 3.458 3.458 differential 1.578 1.23 0.68 0.49

Table 6.2: Comparison between leakage reactances in original and improved prototype 2 according to the combined effect of the increment of the air-gap width plus the slot refitting

6.2.3 3. Transformation ratio & double-layer rotor winding

It was previously demonstrated in Chapter 5 that if an induction regulator prototype had a transformation ratio equal to unity, the leakage inductance would be zero. This approach is obviously not realistic and thus the definition of a theorical transformation ratio equal to unity will involve an improvement, but far away from the zero value. Due to the effect of the number of rotor turns in the magnetizing reactance, which had been previously undermined, the adjustment of the number of turns was done over the stator circuit.

In the design of the electrical circuit for the new prototypes, it has also been substituted the single-layer rotor winding by a double-layer winding short pitched by 2 slots. The use of double-layer windings with short-pitch is a traditional solution among the electrical machines since the short-pitching helps to the cancellation of undesireable harmonic fluxes.

Main characteristics of new electrical circuit in prototype 3 are summarized in Table 6.3.

And as a result of the new winding design and according to the FEM method, the improvement over the leakage reactances due to the new electrical circuit design for prototype

3 and the comparison with the protoypes 2 and 1 is summarized in Table 6.4.

122 Magnitude Description Value Rotor Stator P Number of poles 4 4 nr|ns Number of turns per coil 34 25 Nr · ξr|Ns · ξs Turns per winding factor 246.5 249.6 ξr|ξs Winding factors 0.964 0.832 qr|qs Slots/pole/phase 2 3 qr|qs Transformation ratio 1.03 Table 6.3: Data of prototype 3 electrical circuit

Rotor Stator Prot 0 Prot 1 Prot 2 Prot 0 Prot 1 Prot 2 slot 4.494 3.563 2.63 3.031 1.842 1.458 Leakage end-winding 2.994 2.994 1.4056 3.458 3.458 2.831 differential 4.315 1.23 0.566 1.816 0.49 0.3

Table 6.4: Comparison between leakage reactances in original and improved prototypes because of the change in the transformation ratio and the double-layer rotor winding design

The stator leakage reactances are reduced due to the decrease of the number of turns. For the rotor, the change from a single to double-layer winding has a contribution to the cutback of the slot leakage reactance. If the double-layer winding is short-pitch, an additional effect crops up when the two coils inside the slot belong to different phases [79, ?]. The effect of the short-pitching is also reflected in the reduction of the harmonic leakage component, as can be observed from Eq. 5.15. The new distribution of the rotor winding and the shorten of the end-turn causes a drop in its end-leakage reactance. The final quantification of the expected effect in leakage reactances reductions due to modifications in electric machines design are a 61% in the rotor and a 44.74% in the stator.

123 6.3 Simulation of transient response by FEM analysis

As a preliminary step to the final manufacturing of the new prototypes, the transient FEM model was developed. By the dint of the co-simulation tool of the FEM software package the induction regulator with the parallel connection of the windings was modeled.

The transient FEM model reproduces the whole voltage dip reproduction cycle. In every simulation step a full snapshot of the variables of interest: voltages, currents, dip depth, electromagnetic fields, fluxes, etc. can be taken. In addition, it is also used to estimate the operating temperature of the induction regulator or the losses in the different components, such as Joule, iron losses or parasitic current losses. It was previously mentioned that due to the purpose of this machine, it is going to be used for very brief periods of time. Due to this, the thermal study did not suppose any constraint for the design and it is not shown in this thesis document.

Several virtual tests were accomplished:

• No-load and short-circuit tests for figuring out the parameters of the equivalent circuit

and the subsequent Thevenin impedance

• loaded tests for evaluating the induction regulator response in normal operating con-

ditions

The results from Thevenin reactance calculation due to FEM model are compared with experimental results in the following chapter. In the comparison of the virtual tests with regards to the lab results for voltage dip reproduction and for all the cases of interest, discrepancies between real and simulated values of voltage in the PCC, in the deep part

124 of the dip, were below 10%. Apart from the analysis of voltages and currents, it was also checked that in none of the reproduced use cases, the magnetic flux density was high enough to invalidate the calculated inductances due to saturation.

a) Magnetic flux density for α =29.8o

b) Magnetic flux density for α =0.19o Figure 6.5: Magnetic flux density in prototype 3 before and during voltage dip reproduction

The model is also valid for checking the behaviour of the prototypes when facing to power converters. The power converters are main parts in RES not composed by synchronous generators directly coupled to the mains. The converters can be represented, at their simplest way, as current sources with controlled angle. These virtual tests could not be reproduced

125 in the lab due to the lack of commercial equipment with the required power and being impossible to withstand the voltage dips without disconnecting.

6.4 Construction and assembly

The electrical machines were built with demostrative purposes. Under this perspective some considerations were taken into account when manufacturing these ad-hoc prototypes.

In order to facilitate the access to the rotor, the machine structure was defined to be opened and without shields. The magnetic core was inserted into a metallic case realized from a steel pipe. The absence of shields implied the use of bearings with very small tolerances to avoid misalignments in the shaft. The laminations were laser cut. One example of the laminations with the new slot design can be observed in Figs. 6.6a and 6.6b.

(a) Rotor laminations (b) Stator laminations

Figure 6.6: Rotor and stator laminations for new prototype designs

The machines has been thought to be interchangeable, to allow the comparison of two different variants over the design. Even if the targets were mainly oriented to the use as an

126 Figure 6.7: Electrical machines with taps to be used as EM1 induction regulator, the two constructed machines can be inserted in the prototype for their operation as VFT or induction regulator. For this reason, a special coupling was used, in order to be adaptable for multiple servo motor models. Also the elastic coupling placed at the end of the shaft grants the permutation between the absolute magnetic and the incremental encoders depending if the use of the electrical machine is as EM1 or EM2.

The design of the machine with a bigger air gap entails a higher current consumption when used as variable frequency transformer and, as a consequence, a bigger voltage drop.

To compensate this effect, the electrical machines were provided with intermediate taps in the winding to adjust the transformation ratio if needed, as observed in Fig. 6.7.

The main steps leading to the construction and assembly of the electrical machines are gathered in the following pictures:

• In Fig. 6.8a the stacking of the stator laminations is shown. An analogous procedure

was followed to conform the rotor.

127 (a) (b)

(c) (d)

Figure 6.8: Main steps in prototypes construction

• In Fig. 6.8b the magnetic package is inserted inside the case and fixed due to the

dovetails. Simultaneously, the stator winding was inserted in its place and held with

the wedges introduced in the slots.

• In Fig. 6.8c, the bench is configured, over a reinforced metallic basement, with high

precision bearings and the supports previously grinded.

• The last step of the prototype construction process is the final mounting and the

installation of ancillary parts, such as the rings, the brushes, the brushes yokes or the

encoder, as displayed in Fig. 6.8d.

128 6.5 Conclusions

Along this chapter, it has been accomplished the sizing of the new prototypes with the finite- elements tools. This has been done by taking into consideration the expected positive effects in the leakage reactances reduction analyzed in Chapter 5. Later it has been shown, also within the FEM environment, the model for the study of the behaviour of the new designs during a voltage dip reproduction. Eventually, a summary of the process for the construction and the assembly of the new prototypes has been gathered in order to show the specifics on physical implementation at a laboratory scale.

129 130 Chapter 7

Experimental results and comparative analysis of the prototypes

7.1 Determination of the equivalent circuit parameters

for new prototypes

In this section, the equivalent circuit parameters of new prototypes are evaluated. Once the parameters are determined they can be applied for the evaluation of the prototypes in steady- state regime through the simulation models developed in Matlab/SimulinkR . These models, detailed in Chapter 4, grant a precise study of the prototypes behavior without the need of long computation times required by the FEM models in order to fulfill the accomplishment of the design targets.

The Thevenin impedance is evaluated through the equivalent circuit parameters. The reduction of that Thevenin impedance has been considered as the key performance indica- tor to predict a better behaviour of the prototype in its operation as part of the voltage disturbances generator prototype.

The characterization is attained by the no-load and short-circuit tests. In Fig. 7.1 the magnetizing and leakages reactances of the three prototypes are gathered as well as the value of the Thevenin reactances . The differences between simulation and comparative results for the resistances have been neglected, due to their scarce modification between prototypes.

For every variable, the predicted values according to the FEM model are displayed together with the experimental results.

Xσr Xσs 14 9 12 8 7 10 6 8 5 6 4 3 4 2 2 1 0 0 Prot 1 Prot 2 Prot 3 Prot 1 Prot 2 Prot 3 Sim 11,803 7,787 4,607 Sim 8,305 5,790 4,589 Exp 11,652 7,257 4,481 Exp 8,093 5,616 4,418

Xm Xth 200 30 25 150 20 100 15 10 50 5 0 0 Prot 1 Prot 2 Prot 3 Prot 1 Prot 2 Prot 3 Sim 180,248 64,512 52,842 Sim 21,780 14,305 8,906 Exp 181,051 63,481 51,920 Exp 23,825 13,624 8,556

Figure 7.1: Comparison of parameters between the prototypes

Differences can be appreciated in the comparison between the magnetizing reactances in the second and third prototypes, where the magnetic circuit has remained unchanged in its design. Apart from possible inaccuracies that could derive from manufacturing, the substitution of the single layer by a short-pitched double layer winding introduces variations over the winding factor (mainly over the harmonic factor). This provokes a cut over the harmonic content of the field waveform that favours the machines magnetization, causing an effective reduction in the magnetizing reactance Xm.

With the design changes proposed in this work and according the the results shown in

Fig. 7.1 it can be concluded that a reduction of 61.56% was achieved in the rotor leakage

132 reactance and a 45.41% over the stator leakage reactance. The expected percentages of the decrease according to FEM simulations were 61.01% and 44.74% for the rotor and the stator respectively.

These results in the leakage reactances lead to the fulfillment of the design target of decreasing the Thevenin reactance. In the analysis of the experimental values from the lab tests, it can be concluded that a reduction of 42.81% over the original value was reached with the first improvement and, reaching at the end, a total cutback of 64.09% over the original value.

7.2 Experimental results. Comparative analysis

In the following sections it is displayed the experimental results of the comparison between the three different prototypes. This comparison has been established based on the laboratory tests for diverse load status (passive and active) and generators operating in different points of the PQ plane. This dissertation has put the attention on the electromagnetic performance of the various prototypes used as induction regulators but not in the control systems of the generating sources. Due to this and for contrasting purposes, the tests of the voltage distur- bances generator with the three electrical machines that can be used as EM1 independently of the type of source connected —but working under the same conditions—proves to be sufficient.

7.2.1 No-load test

The no-load tests grant an initial validation. Since there is no current circulating by the stator, the effect of the mutual coupling is cancelled. In Fig. 7.2 it is shown the compari-

133 son between the voltages at the output of the voltage disturbances generator for the three induction regulators.

Prototype 1

X: -0.412 X: 0.351 Y: 16.37 Y: 15.74

Prototype 2

X: -0.412 X: 0.351 Y: 15.12 Y: 17.17

Prototype 3

X: -0.412 X: 0.351 Y: 12.33 Y: 13.61

Figure 7.2: Comparison of voltage dips for the prototypes at no-load

Prototypes 2 and 3 present a reduced Thevenin impedance with respect to the original prototype, as was presented in Fig. 7.1. Although the reduction of the Thevenin impedance is evident, it is also clear that the current consumption required to magnetize the machine increases due to the air gap widening. As there is no other effect involved due to the absence of coupling with the stator winding, the voltage drop does not evidence this impedance reduction since it is partially balanced with the magnetizing current enlargement. The

60.09% of Thevenin reactance reduction between the prototype 1 and the prototype 3 is later translated into a 24% of voltage drop cutback. Slight differences can be appreciated between the voltages in steady-state condition prior to the disturbance. In that situation, the induction regulator has reached its theoretical minimum value achievable in the regulator at the no-load condition.

134 7.2.2 Loaded tests

7.2.2.1 Test against a passive load

The first trials of the voltage disturbances generator working at diverse load regimes were attained with passive loads. Fig. 7.3 shows the limiting case where EM1 is operating at full-power. The voltages in the output of the prototypes are plotted, where the ENTSOE

D4 profile has been programmed (see Chapter 2).

400 Prototype 1 200 X: 0.379 Y: 20.68 0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 400 Prototype 2 200 X:X: 0.379 0.379 Y:Y: 15.23 15.23 0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 400 Prototype 3 200 X: 0.379 1 Y: 12.13

Voltage at the induction regulator output (V) 0

−200

−400 −0.5 0 0.5 1.5 2 2.5 3 3.5 4 Time (s) Figure 7.3: Comparison of voltage dips against resistive load at full rated power

As expected, bigger differences can be observed while there is current circulating along the whole dip profile with respect to the no-load case. To focus only on the electromagnetic behavior of the prototypes, the comparison is established for the deep part of the voltage dip when reached the steady-state condition. From the prototypes 1 and prototype 2 responses it can be deduced there is a reduction in the remnant voltage of 26%. An additional 20.35% can be achieved with prototype 3. The voltages in the deep part of the voltage dip are obviously greater with respect to the ones in the no-load case because of the circulation of current through the stator during the dip.

135 The tests with resistive load represent the advantage of not introducing additional tran- sients coming from the load behaviour. However, due to the resistance nature, the sudden voltage drop during the disturbance is translated into a similar reduction in the current.

Even if it is clear that the new prototypes present a reduced voltage drop with respect to the original electrical machine, the impossibility of maintaining a noticeable current circulating during the dip implies that an accurate quantification of the effect cannot be achieved by means of resistive tests.

7.2.2.2 Test against an active load

In Fig. 7.4 it is shown the bench used for the tests with the asynchronous machine, operating as motor or generator. It is composed by a high-efficiency asynchronous motor mechanically coupled to a synchronous machine. For the operation as a motor, the asynchronous machine is connected to the output of the voltage disturbances generator while the synchronous machine connected to a resistor bank allows the variation of motor loading.

Figure 7.4: Asynchronous motor/generator test bench

136 Testing with the asynchronous motor makes it possible a more direct check of the differ- ences between the Thevenin reactances of the three prototypes. But following considerations must be taken into account:

• When a small-sized motor, like the one in the lab test bench is subjected to a voltage

dip, it suddenly stops. If the voltage recovers, the motor starts, consuming a current

several times bigger than the rated.

• If the starting current is limited by a higher short-circuit impedance, considering the

series Thevenin impedance of the voltage disturbances generator, the voltage recovery

time slows down.

In Fig. 7.5 it can be easily observed the differences between the responses of the three prototypes for the tests conducted at rated power. From the comparison it can be observed that the recovery time after the dip for prototype 2 is reduced in almost half with regard to the original prototype and an additional third decrement can be achieved with prototype 3, as a direct effect of the contraction of the Thevenin reactances in similar proportions.

400 Prototype 1 200

0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 400 Prototype 2 200

0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 400 Prototype 3 200

Voltage at the induction regulator output (V) 0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 Time (s)

Figure 7.5: Comparative behaviour of tests against asynchronous motor

137 7.2.2.3 Test against generators

Once it has been concluded the loaded tests with active and passive loads, the behaviour of the device when connecting generators has to be checked. These generators should represent the diverse generating technologies commercially available suitable to be certified by the disturbances generator prototype. Three types of suitable sources can be found:

• Asynchronous generator: which represents the simplest wind power technology cur-

rently in service.

• Synchronous generator: which is the usual machine used in conventional energy sources,

present at microgrid voltage levels mainly in diesel units and hydraulic groups.

• Generators coupled to the grid through converters: existing in all the technologies not

employing a generator directly connected to the grid. It is a widely extended topology

in distributed generation.

Due to avoid the interference of the control systems of the converters over the voltage disturbances prototype, affecting the response of the device against disturbances, the third technology has been rejected for the validation of the new electrical machines design. It has also to be remarked that the non-existence of commercial full-power converters with the required power able to withstand when facing to voltage dips, would have made impossible to carry tests with this technology at lab scale.

7.2.2.3.1 Asynchronous generator

These machines cannot naturally stand up voltage dips. The voltage drop in machine termi- nals leads to a decrease of the electric torque that causes an overspeed issue. In a real loca-

138 tion, this technology, used in primary wind power turbines, usually tripped by the overspeed protections performance. Due to this, it was progressively substituted by other topologies, such as doubly-fed induction generators or generators coupled through full-power converters

(synchronous or asynchronous). At a lab scale and with the testing purposes related to this work, the asynchronous machine allows to keep some current circulating through the stator impedance during the whole voltage disturbance, with the generator sign criteria and adding the extra transient regime of the electrical machine to the voltage disturbances generator.

This is carried out with the aim of validating the design target of the internal voltage drop reduction.

Prototype 1 X: 0.435 Y: 67.61

Prototype 2

X: 0.435 Y: 47.13

Prototype 3

X: 0.435 Y: 28.34

Figure 7.6: Comparative behaviour against asynchronous generator

The bench used was the one previously presented in Fig. 7.4. The synchronous generator drove by the converter, is used as turbine emulator for the asynchronous machine. It can be observed that when using the prototype 2 a reduction of 30.29% in the voltage dip depth can be achieved over the prototype 1 whereas prototype 3 shows a decrement of 58.08%. At the same time, these results also demonstrate that the cutback in the Thevenin reactance

139 has also a noticeable effect in the transient regime drop time at the first beginning of the voltage dip.

7.2.2.3.2 Synchronous generator

The tests with synchronous generator machines were conducted with the bench that can be observed in picture 7.7. It is composed by the joint of a DC machine and a synchronous generator with the same rated power than the voltage disturbances generator. The DC machine acts as the turbine by means of the control of the active power injected while the regulation of the voltage over the excitation system allows the management of the reactive power flow.

Figure 7.7: DC machine and synchronous generator test bench

Several cases were tested with the synchronous generator. They were carried out with two different power factors, meeting the requirements in [54], 0.9 inductive and 0.95 capacitive.

These power factors in current regulations are required for wind turbines, and it was used as guidance. Even the type of generating source used is different, from the point of view of the disturbances generator prototype, the importance roots on the current injected by the generating source, since the goal is not the testing of the associated control systems. Four

140 loading situations in steady-state were selected (≈ 10%, ≈ 35%, ≈ 60% and ≈ 85%) covering the scenarios required in the regulations for the testing procedure (between 10% − 20% and over 80% of rated active power).

The field current is provided by an independent power source which is directly supplied by the output of the generator when rectifying the AC voltage. Due to the low price, it is the widespread solution in medium to large conventional power plants. When facing to a voltage dip, the field current is affected by the voltage drop in the machine terminals. The non-existence of control over the excitation allows a better appreciation of the improvement in the response of the ad-hoc prototypes over the original model.

An example of the whole profile for one of the cases, with a 10% of loading is displayed in Fig. 7.8. It can be observed that during the transient mode, the prototype 2 also reaches a very low value of remnant voltage but the system achieves the steady-state regime faster with prototype 3, where the effect of the Thevenin reactance reduction becomes more visible.

400 Prototype 1 200

0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 400 Prototype 2 200

0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 400 Prototype 3 200

Voltage at the induction regulator output (V) 0

−200

−400 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 Time (s)

Figure 7.8: Comparative of the behaviour of prototypes connected to synchronous generators with static excitation

141 The whole range of tests accomplished are summarized in Fig. 7.9 where it can be easily observed the reduction in the voltage dip generator disturbances output when facing a voltage dip with a depth reference of 10%. In all the cases, the remaining voltage is higher if compared with any other type of loads/generator because the transient reactance of the synchronous generator is high enough (around 25Ω) to limit the level of the voltage drop in the connection node. As expected, a smaller decrease is achieved with the prototype

2 with regards to prototype 3. The improvement with prototype 2 can be estimated in a range between 55% − 60% depending on loading condition while 70% − 75% of enhancement is obtained with the modified prototype 3, both of them considered over the commercial electrical machine response.

100 Prototype 1 90 40 Prototype 2 80 Prototype 3 PF=0.9 ind 70 30 % dip depth PF=0.95 cap 60 50 20 Uo(V) 40 30 20 10 10 0 10 35 60 85 % Load

Figure 7.9: Comparative of the behaviour of prototypes connected to synchronous generators with static excitation

7.3 Conclusions

Thorough this chapter the experimental verification of new prototypes behaviour have been accomplished. With that purpose, the electrical machines specifically designed in this work

142 have been characterized, in order to find the new parameters of their equivalent circuits.

As a result, the new Thevenin reactances have been calculated. Once has been checked the effective curtailment of the Thevenin reactance, the new prototypes have been validated by their testing from no-load conditions to fully-loaded with loads/generators with diverse characteristics. The comparative analysis of the three prototypes evidences the better suit- ability of the new designs (mainly prototype 3) to reduce the affection of the EUT current injection over the testing device.

143 144 Chapter 8

Final conclusions and future work

The expected growing in the insertion of distributed generation in electrical systems has forced the tightening of requirements in grid codes. For isolated microgrids and due to their specificities, this trend is more pronounced. On the other hand, the deployment of regulations is linked to the need for experimental equipment to certify the devices according to those regulations.

Among all the diverse topologies existing from a theoretical perspective just a few of them are commercially available. To this small group it is added the voltage disturbanes generator developed in this dissertation. It presents the advantages of being a complete equipment for reproduction of the whole voltage dip pattern, including the recovery ramp, as well as the frequency disturbances and the controlled phase jumps. In this thesis, an initial small-scale prototype was built for its study, in order to setting the basis for a further development at a bigger scale.

Thorough this work a complete study from different perspectives was accomplished. It included the assembly and the proof-of-concept tests, the study in permanent regime —ma- thematical expressions, simulations and testing— and the transient regime —mathematical expressions and FEM—. A nuclear part of this dissertation was focused on the enhacement of the induction regulator design to be used for the voltage dip reproduction. In the vali- dation of the new prototypes it could be verified the predicted reduction of the Thevenin reactance, around 65%, which is translated into a lower value of remnant voltage dip depth depending on the loading condition.

As a future work for validation, the study of the prototype behaviour against a source of constant current is suggested. Since the current injected to the prototype would not be affected by the disturbance, it would be the best option to quantify, precisely, the effect of the design over the remaining voltage. This possibility was not available during this work due to the absence of the required equipment in the lab testing site.

Some other actions have also been identified, with the purpose of going one step further towards a scalable prototype up to the power of suitable EUTs (up to 5 MW).

With regard to the construction and assembly of the new voltage disturbances generator:

• Substitution of the electromechanical contactors by static contactors in order to reduce

the transients peaks in the voltage dips reproduction.

• Centralization of the servo motors controls in a single drive, to simplify the equipment

and reduce, to the possible extent, the areas for potential failure.

Concerning the electrical machines:

• Once checked the guidelines proposed for the design of the new prototypes, these are

aligned with the goal of improving the voltage disturbances prototype response, it

could be possible to get an optimized electrical machine by applying any optimization

algorithm during the design process.

146 APPENDIX

147 148 Appendix A

Technical Data of the Experimental

Setup Equipment

A.1 Servo Motors

Two commercial servo motors for its use in the voltage disturbances generator were selected.

Traditionally, the basic reasons for using servo systems include the demand of improved transient response times, the reduction of the steady state errors and the decrease of the sensitivity to load parameters. In Fig. A.1 the two servo motors in the experimental setup can be observed.

Figure A.1: Servo motors involved in the disturbances generator prototype A.1.1 Servo motor for EM1 machine

EM1 servo motor is a permanent magnet synchronous motor. This type of machine is identified by its high dynamic response, big torque density and accurate position and control.

The main characteristics of the Baum¨uller servo motor chosen for this application are shown in Table A.1 [80].

Manufacturer : Baum¨uller

Model : DS 56 S-3-R-K

Magnitude Value Description P (kW) 1.4 Electrical Power nn(rpm) 3000 Rated Speed Mn(Nm) 3.8 Rated Torque In(A) 1.8 Rated Current ke(v/1000rpm) 95.2 Back e.m.f at 1000 rpm kt(Nm/A) 1.57 Torque constant kd(Nm/1000rpm) 0.030 Torque loss referred to 1000 rpm Mr(Nm) 0.154 Bearing friction w(kg) 6.6 Weight J(kgcm2) 5.7 Rotor inertia

Table A.1: Data of Baum¨uller DS 56 S-3-R-K

A.1.2 Servo motor for EM2 machine

An asynchronous geared motor for the frequency disturbances generation was selected. In the application of interest it is not required a very high performance. Due to this, a geared motor is convenient because of its simplicity and low cost. The fundamental parameters of the motor and the gearbox are shown in Tables A.2 and A.3 [81].

150 Manufacturer : Pujol Muntal´a

Model : IPCM 128/90L-4/148

- Motor

Magnitude Value Description P (kW) 1.5 Electrical Power nn(rpm) 1400 Nominal speed V (V ) 210-230Δ / 370-400Υ Rated voltage I(A) 6.02/3.48 Rated Current f(Hz) 50 Frequency cosφ 0.79 Power factor

Table A.2: Motor data from Pujol Muntal´a IPCM 128/90L-4/148

-Gearbox

Magnitude Value Description n2(rpm) 148 Rated low speed M2(Nm) 93 Rated torque at low-speed shaft i 9.43 Reduction ratio

Table A.3: Gearbox data from Pujol Muntal´a IPCM 128/90L-4/148

A.2 Drives

A.2.1 Angle drive for EM1

Baum¨uller b maXX 4400 [82], which data appears in A.4, is a modular servo controller. As it is a modular and scalable structure it can be adopted for multiple applications. It is suitable for its usage in the voltage disturbances prototype due to its speed response in positioning control, required for the accurate reproduction of voltage disturbances profiles.

151 Manufacturer : Baum¨uller

Model : b maXX 4413-STO-01200-03

Magnitude Value Description Pm(kW) 2 Motor power Vs(V ) 400 (3 phases) Power supply voltage In(A) 4.5 Rated current Imax(A) 9 Maximum current F 2 Overload factor

Table A.4: Main technical characteristics of Baum¨uller b maXX 4413

A.2.2 Speed drive for EM2

Altivar 71 (Table A.5) is a variable speed drive of Schneider Electric company [83]. It is a cost effective device that provides an easy and complete performance.

Manufacturer : Schneider Electric

Model : Altivar 71

Magnitude Value Description Pm(kW) 7.5 Motor power Vs(V ) 380V...480V (3 phases) Power supply voltage I380V /I480V (A) 27/22.2 Line current Pa(kV A) 17.8 Apparent power Ic380/Ic480V (A) 17.6/14 Maximum continuous current at 380V/480V It60/It2(A) 26.4/29 Maximum transient current for 60s/2s fo(Hz) 0.1...599 Speed output frequency w(kg) 5.5 Weigth

Table A.5: Main technical characteristics of Altivar 71

152 A.3 Encoders

A.3.1 Absolut magnetic encoder for EM1

An absolute encoder was installed in the shaft of the machine EM1. As the control loop of angle is closed through Baum¨uller drive, the encoder is used just with information purposes.

The main characteristics of the selected RM 36 are gathered in Table A.6 [84]. The RM 36 encoder is based on the magnetic principle. A magnet rotates jointly to the shaft, activating, without direct contact, the Hall sensors placed in the non-rotating part. The rotation angle between 0◦-90◦gives a proportional analog signal varying between 0-10 VDC.

Manufacturer : Elap

Model : RM 36

Magnitude Value Description Vs(VDC) 20-30 Voltage supply Vo(Nm) 0-10 Output voltage I(mA) 10 Maximum output loading

Table A.6: Data of RM 36 absolut encoder

A.3.2 Incremental encoder for EM2

An incremental encoder translates rotary motion into a group of digital pulses. The three signals collected are: A for the rotational speed, B for the direction of rotation and Z for the absolute position of the zero. They are processed and decodified by the encoder card

VW3A3401, installed in the Altivar 71 [85] drive.

153 Manufacturer : Schneider Electric

Model : XCC1506PS50X

Magnitude Value Description D(mm) 58 Diameter Ds(mm) 6 Shaft diameter Resolution 5000 Points per revolution Us(V ) 4.75..40 VDC Rated Supply Voltage Table A.7: Data of encoder Schneider XCC1506PS50X

A.4 DSPs dSPACE

The dSPACE DS1104 R&D [86] control boards are responsible for the control of the operation of the voltage disturbances prototype. The simplicity of its use, due to the human machine interface (HMI) that can be used with the MATLAB/SimulinkR environment, makes it a powerful tool to be used in rapid control prototyping experiments. The signals are collected through Hall sensors boxes. These sensors adapt the measured voltage and current levels to corresponding 0-10 VDC required for the DSPs. In Fig. A.2 can be observed the internal structure of the board.

Each board integrates an A/D converter with 8 channels, that will be used for amass the measurements. Four channels are multiplexed and have a conversion time of 2 μs. The other four are parallel, and have a faster conversion time of 0.8 μs. The D/A transforms the signals from the RT interface in analog setpoints to command the drives. The digital outputs will control the opening and closing of selected switches of the voltage disturbances prototype.

154 Figure A.2: Block diagram of dSPACE DS1104 R&D

155 156 Appendix B

Thevenin equivalent for the

EM1+EM2 case

This annex collets the mathematical development from the equivalent circuit that allows the achivement of the Thevenin equivalent expressions of the global equipment. In figure B.1 it is shown a single-phase scheme of the device. Subscripts r and s are used for rotor and stator windings and 1 and 2, for EM1 y EM2 machines.

s·R´r1 jX´σr1 Rs1 jXσs1 Is1 Is2 Rs2 jXσs2

Ir2 E22 I´r1 I´m Rr2

jX´m1 jXσr2 Uo s·U´r1

jXm2 E21

Figure B.1: Single-phase representation of the device

This scheme has been considered the simplest representation to find the equations of interest. Two relevant considerations concerning to this scheme must be highlighted: • The VFT is represented through its equivalent as seen from stator-side. It means that the parameters of the circuit are referred to the variable frequency f2, related with the frequency of the grid U1 through the slip.

• The induction regulator is shown as its real equivalent circuit, so the voltage input to the rotor is fed from the variable frequency supply. The parameters of EM2 machine are referred to rotor or stator depending on the corresponding winding.

B.1 Thevenin source

The calculation of the Thevenin source leads to the following expression B.1

Z · j · X · s · U U = r2 m1 1 (B.1) th · · · (s Rr1 + jXσr1) (Zs1 + Zr2)+jXm1 Zr2

where the impedances Zs1 and Zr2 are: Zs1 = Rs1 + jXσs1 + jXm1

Zr2 = Rr2 + jXσr2 + jXm2

B.2 Thevenin impedance

The circuit to obtain the Thevenin impedance by short-circuiting the independent sources is displayed in B.2:

The equivalent impedance corresponding to the VFT, Zeq is then B.2:

158 Rs2 jXσs2

E22 Rr2

Zeq jXσr2 Icc

jXm2 E21

Figure B.2: Single-phase representation of the device

(s · R + jX ) · jX Z =(R + jX )+ r1 σr1 m1 (B.2) eq s1 σs1 · s Rr1 + jXσr1 + jXm1

And the Thevenin impedance is displayed in B.3. It can be easily observed than a particular case where Zeq = 0 leads to the previously deduced equations for EM2 shown in the main body of this thesis document.

Z jX jX X2 Z = eq · m2 − Z · m2 − 1 + m2 + Z (B.3) th ∗ r2 · · 2 s2 Zr2 + Zeq Km2 Zr2 Km2 Zr2 Km2

159 160 Appendix C

Calculation of Laplace residuals for transient regime solutions

C.1 Rotor current ir(s)

The equation in Laplace domain of the current circulating by the rotor of EM1 and obtained from 3.2.5 is shown in C.1:

(ur + Lr · ir0) · Rs + Lss − Lms us + Lm · ir0 ir(s)= 2 2 (C.1) (Rr + Lrs) · (Rs + Lss) − Lm · s

After replacing in the previous expression the value of the rotor and stator voltages, the resulting equation is presented in C.2:

2 2 Ur · ω + s + ω · Lr · i R + L s i (s)= 0 s s − r 2 2 2 2 s + ω · Lr · Ls − Lm · s +(Rr · Ls − Lr · Rs) · s + Rr · Rs (C.2) L · s · Us s · sin (φ )+ω · cos (φ )+ s2 + ω2 L · i − m s s m 0 2 2 2 2 s + ω · Lr · Ls − Lm · s +(Rr · Ls − Lr · Rs) · s + Rr · Rs

The equation C.2 can be splitted in simple fractions attending to the poles number: N1 N2 N3 N4 ir(s)= + + + (C.3) s − jω s + jω s − r3 s − r4

with the subsequent values for the residuals:

j · Ur · Rs + jωLs + ω · Lm · Us · (j · sin(φs)+cos(φs)) N =(s − jω) · ir| = 1 s=jω 2 · [−A · ω2 + jω · B + C] (C.4)

−j · Ur · Rs − jωLs + ω · Lm · Us · (−j · sin(φs)+cos(φs)) ∗ N =(s + jω) · ir| − = = N 2 s= jω 2 · [−A · ω2 + jω · B + C] 1 (C.5)

2 2 2 s + r · [Lr · i · R + L · r − L · i · r ] N =(s − r ) · i | = 3 0 s s 3 m 0 1 + 3 3 r s=r3 · 2 2 · − A (s + r3) (r3 r4) (C.6) U · ω · (R + L · r ) − L · r · U · (r · sin(φ )+ω · cos(φ )) + r s s 3 m 3 s 3 s s · 2 2 · − A (s + r3) (r3 r4)

2 2 2 s + r · [Lr · i · R + L · r · L · i · r N =(s − r ) · i | = 4 0 s s 4 m 0 4 + 4 4 r s=r4 · 2 2 · − A (s + r3) (r3 r4) (C.7) U · ω · (R + L · r ) − L · r · U · (r · sin(φ )+ω · cos(φ )) + r s s 2 m 2 s 2 s s · 2 2 · − A (s + r3) (r3 r4)

C.2 Stator current is(s)

Similarly, the residuals calculation for is(s) breeds the following formulas:

162 j · Us (Rr + jωLr) · (j · sin(φs)+cos(φs)) + j · Lmω · Ur N =(s − jω) · ir| = (C.8) 5 s=jω 2 · [−A · ω2 + jω · B + C]

−j · Us (Rr + jωLr) · (−j · sin(φs)+cos(φs)) − j · Lmω · Ur ∗ N =(s + jω) · ir| − = = N 6 s= jω 2 · [−A · ω2 + jω · B + C] 5 (C.9)

2 2 s + r · [Lm · i · (Rr + Lr · r ) − Lm · Lr · i · r ] N =(s − r ) · i | = 7 0 7 0 7 + 7 7 r s=r7 · 2 2 · − A (s + r7) (r7 r8) (C.10) U · ω(R + L · r ) · (r · sin(φ )+ω · cos(φ )) − L · ωr · U + s r r 7 7 s s m 7 s · 2 2 · − A (s + r7) (r7 r8)

2 2 s + r · [Lm · i · (Rr + Lr · r ) − Lm · Lr · i · r ] N =(s − r ) · i | = 8 0 7 0 8 + 8 8 r s=r8 · 2 2 · − A (s + r8) (r8 r7) (C.11) U · ω(R + L · r ) · (r · sin(φ )+ω · cos(φ )) − L · ωr · U + s r r 4 4 s s m 8 s · 2 2 · − A (s + r8) (r8 r7)

Once calculated all residuals, it can be defined the transient regime of the electric ma- chines that act as a induction regulator according to its equivalent circuit parameters.

163 164 Appendix D

Analytical calculation of slot leakage components in EM2

D.1 Slot leakage inductance calculation

D.1.1 Rotor slot leakage calculation

In every electrical machine, the value of the differential inductance that provoke the slot leakage flux is directly related with the magnetic permeance derived from the geometry, P .

It is also linked with the winding distribution and the number of conductors inserted in the slot, ns, according to D.1:

2 dL = n s · P (D.1)

The inductance can be determined in terms of the flux linkages created by the magnetic

field (λ) with regard to the current which created it, (I),asshowninD.2:

λ L = (D.2) I The rotor slot has a shape like the one shown in D.1:

Figure D.1: Rotor slot

where n(x) is the number of conductors included in a stip with differential height in the x axes. According to Amp`ere Law D.3:

2 H · dl = n(x) · I (D.3) 1

For a generic region containing conductors and assuming the magnetic field in every section with differential height is constant, the following expression D.4 is satisfied.

Hy · y = n(x) · Ix (D.4)

166 μ · n2(x) · I · l · dx dλ = n(x) · dΦ= o e (D.5) y

μ · n2(x) · l · dx dL = n(x) · dΦ= o e (D.6) y

Matching the expressions D.1 and D.5 and dividing for the specific length of the machine le, the equation relating the slot specific permeance is shown in D.7

μ · n2(x) · dx dp = n(x) · dΦ= o (D.7) y

The function n(x) has different values depending on the slot region where is placed.

Therefore the value of this function has to be calculated for each of the four zones marked on the figureD.1.

• Regions 1 y 2

The leakage flux that close their path in this area link no conductors. This supposes that n(x) = 0 and, as a consequence, the slot specific permeance in this area is also null.

• Region 3

For region 3, the mathematical calculation is more complex. In this case, n(x)varies within the section. The starting point to calculate the permeance in this region is also

D.7. To solve the integral that leads to the permeance value, it is required to establish a relationship between x and y. From figure D.2 can be determined by double similatiry, the relation D.9.

167 Figure D.2: Fraction of rotor slot

ν ν + x ν + x − r − d = = 0 3 (D.8) b2/2 y/2 y/2 ν ν + d = 2 (D.9) b2/2 b1/2

Through the matching of these equations can be finally established the relationship bet- ween the variables x e y, as shown in figure D.10:

b1 − b2 y = b2 + · (x − r0 − d3) (D.10) d2

Then it is necessary to establish the relationship between the coils that encompasses a line of leakage flux passing through the slot to a certain height x with respect to the number of total coils into the slot. From figure D.3 can be deduced D.11.

168 Figure D.3: Change of the turns ratio depending on the slot depth

n(x) 1 = · (x − (r0 + d3)) (D.11) ns d2

With all the known terms, specific permeance in slot region 3 is calculated by solving the integral D.12

r +d +d 0 2 3 1 2 dx p3 = μ0 · · (x − r0 − d3) · (D.12) d2 b1−b2 r0+d3 2 b2 + · (x − r0 − d3) d2

Performing the following change of variable

b1 − b2 t − b2 b2 + · (x − r0 − d3)=t −→ x − r0 − d3 = · d2 (D.13) d2 b1 − b2 b − b d · dt 1 2 · dx = dt −→ dx = 2 (D.14) d2 b1 − b2

The equation D.12 is expressed in terms of the t variable as:

μ b1 (t − b )2 · d3 p = 0 2 2 · dt (D.15) 3 2 − 3 · d2 b2 (b1 b2) t

169 The final expression for the permeance in the region 3 is displayed in D.16:

b2 − b2 μ0 · 1 2 2 · b1 − · · − p3 = 3 d2 + b2 ln 2 b2 (b1 b2) (D.16) (b1 − b2) 2 b2

• Region 4

In region 3, the factor n(x)/ns is the unit due to every leakage flux path will include all the conductors. According to the geometry D.4:

Figure D.4: Region 3 of the rotor slot

The coordinates of a generic point P placed over the edge of the slot can be expressed according to D.18:

x = x − d1 − r0 = r · cos θ (D.17) y = y =2· r · sin θ (D.18)

The area on which the calculation of permeance must be done is the difference between the areas marked as A1 y A2 in D.4, and that will be named A3:

170 r2 b 2 A = π · = π · 1 (D.19) 1 2 8

And the area of the circular segment A2 can be expressed as the difference between the area of the circular sector that sweeps a 2θ degree angle less two times the triangle which

b0 base is d1 and height 2 :

2 2 2θ b0 b1 b0 b0 · d1 A2 = π · r · − b1 · = arcsin − (D.20) 2 · π 2 4 b1 2

where the angle θ can be expressed according to the geometrical dimensions of the slot as D.21:

b θ = 2 arcsin 0 (D.21) b1

The area A3 is then D.22:

2 2 b1 b1 b0 b0 · d1 A3 = A1 − A2 = π · − arcsin + (D.22) 8 4 b1 2

The permeance of the semicircular region is shown in the equation D.23:

dx 0 −r · sin θ · dθ 1 p = μ · = μ · = · μ · π (D.23) A1 0 y 0 π 2 · r · sin θ 4 0 2

Applying the proportionality of areas between sections A1 and A3, A3 permeance asso- ciated to the area is expressed as D.24:

171 A π 1 b b · d p = 3 · p = μ · − arcsin 0 + 0 1 (D.24) A3 A1 0 2 A1 4 2 b1 b1

• Region 5

In this region, in contrast to what happened in region 1, all conductors would be within the leakage flux path which closed in this area, n(x)=ns, and the integral expression that allows the calculation of specific permeance in that area is very simplified:

dx r0+d1+d2 dx d dp = μ · −→ p = μ = μ 0 (D.25) 5 0 y 4 0 b 0 b r0+d0+d1+d2 0 0

And the total permeance in the rotor slot, pr, is the addition of all the permeances in every slot region.

4 b2 − b2 μ0 · · 1 2 · b1 − · · − pr = pi = 3 d2 + b2 ln 2 b2 (b1 b2) + (b − b ) 2 b2 i=1 1 2 · · π − 1 b0 b0 d1 d0 +μ0 arcsin + 2 + μ0 (D.26) 4 2 b1 b1 b0

Multiplying by the total effective length and the number of turns connected in series in the slot, can be deduced the final expression for the rotor leakage inductance for one slot

D.35:

2 Lrotor−slot = nc · le · pr (D.27)

172 D.1.2 Stator slot leakage calculation

The stator winding is a double-layer. The value of the inductance of the slot must take into account the effect of the leakage flux of each of windings and the interaction among the two.

The leakage inductance equation for the rotor slot is displayed in D.28.

2 Lstator−slot = nc · le · (pL + pU + pLU ) (D.28)

D.1.2.1 Slot permeance due to the lower coil

For the lower coil placed in the slot and assuming there is no upper coil, the slot permeance expression is gathered in D.29, with the geometry parameters that appear in D.5.

4 μ b2 − b2 b p = p = 0 · d · 4 3 + b2 · ln 4 − 2 · b · (b − b ) + L i − 3 4 2 3 b 3 4 3 i=1 (b4 b3) 3 π 1 b b · d d d + d b +μ · − arcsin 0 + 0 1 + μ 0 + μ · 2 3 ln 1 (D.29) 0 2 0 0 − 4 2 b1 b1 b0 b1 b3 b3

D.1.2.2 Slot permeance due to the upper coil

Exactly the same procedure is followed for the upper coil. The value of the permeance due to the presence of the upper coil and taking out the lower one, allows the calculation of the expression D.30:

173 b1

b0 d0

d1

d2

b2

b3 d3

d4

b4

Figure D.5: Stator slot geometric dimensions

4 · d0 · π − 1 b0 b0 d1 pU = pi = μ0 + μ0 arcsin + 2 + b0 4 2 b1 b i=1 1 b2 − b2 μ0 · · 2 1 · b2 − · · − + 3 d2 + b1 ln 2 b1 (b2 b1) (D.30) (b2 − b1) 2 b1

D.1.2.3 Slot permeance due to the mutual effects between coils

The methodology for the calculation of the mutual inductance is based on the integration of the effects one strip of the upper coil has over the lower coil. The distance x is measured as a height with the origin in the lower side of the upper coil. The differential of flux produced in a band of height dx has the equation D.31:

x dx dΦ=F (x) · le · dp (x)=I · nc · · μ0 · le · (D.31) d2 y

And, thus, the specific permeance is D.32:

174 x dx dp(x)=μ0 · · (D.32) d2 y

The permeance will be obtained by integrating D.32, where, y is a function of x that can be written due to geometric considerations as D.33.

b1 − b2 y = b2 + · x (D.33) d2

The solution to the integral give the final expression of the permeance associated to the mutual flux created between coils D.34.

b3 b2 p2 =2· μ0 · −1 − · ln (D.34) b3 − b2 b3

The total specific mutual permeance, taking into account the regions b0 − d0 and b1 − d1 is D.35:

d0 1 b2 − b0 b3 b2 pLU = p0 + p1 + p2 = μ0 · + · arcsin +2· −1 − · ln b0 2 b2 b3 − b2 b3 (D.35)

175 176 Appendix E

Windings and steel sheets drawings

In this annex additional information about the prototypes is clustered. Some information is common to all the electrical machines, since the magnetic material or the stator winding distribution is the same for all of them. It is also gathered the designs of the electric steel sheets to compare the differences in the slot design between the original prototype and the new ones.

E.1 Magnetic material

The magnetic material is a V600-50A according to the specification of the DIN 46400 stan- dard. The BH curve is shown in figure E.1:

V600-50A 1,8 1,6 1,4 1,2 1

B (T) 0,8 0,6 0,4 0,2 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 H (A/m)

Figure E.1: BH curve of V600-50A magnetic steel E.2 Windings

E.2.1 Rotor Windings

Prototypes 0 and 1 are designed as single layer windings, with two slots per pole and phase, as shown in Fig. E.2.

A C B A B C

Figure E.2: Rotor winding configuration for prototypes 0 and 1.

In prototype 2 the rotor winding was changed to a double-layer winding with q=2 and a pitch of 1-5 (short pitch of 2 slots).

E.2.2 Stator Winding

In Fig. E.3 is displayed the double-layer lap winding in the stator. It is the same winding distribution for the stators of the three machines, with three slots per pole and phase and the pitch shorted in three slots (1-7). In order to make the figure easier to comprehension, only

178 phase A end windings have been drawn. The other phases have been plotted maintaining the code color and the direction of the current flow.

A B C B A C Figure E.3: Double-layer lap winding design for stators

E.3 Electric steel sheets drawings

Next are detailed the electric steel sheets drawings for the prototypes, where can be clearly observed the increment in the slot openings for both stator and rotor magnetic circuits.

179 Figure E.4: Design for prototype 0

180 Figure E.5: Design for prototypes 1 & 2

181 BIBLIOGRAPHY

182 BIBLIOGRAPHY

[1] European Network of Transmission System Operators for Electricity (ENTSO-E). (2013). [Online]. Available: https://www.entsoe.eu/about-entso-e/system-operations/ regional-groups/, [Accessed: May 8, 2015]

[2] SBI Energy, “The World Market for Microgrids,” 2011.

[3] Red El´ectrica de Espa˜na. (2010) Retos de la Penetraci´on de Energ´ıas Renovables en los Sistemas El´ectricos Canarios. [Online]. Available: http: //www.catedraendesared.ulpgc.es/index.php/descargas/doc download/9-ponencia- retos-de-la-penetracion-de-energias-renovables-en-los-sistemas-electricos-canario

[4] Rup´erez, J. (2012) Integraci´on de Energ´ıas Renovables en Sistemas Aislados y Estabilidad. [Online]. Available: http://proyectotres.itccanarias.org/es/documentacion

[5] International Enegy Agency (IEA), “CO2 emissions from fuel consumption. Highlights,” 2012.

[6] Department of Energy (DOE). United States Government, “Summary Report: 2012 DOE Microgrid Workshop,” 2012. [Online]. Availa- ble: http://energy.gov/oe/downloads/2012-doe-microgrid-workshop-summary-report- september-2012, [Accessed: May 8, 2015]

[7] Consortium for Electric Reliability Technology Solutions (CERTS), “Integration of Distributed Energy Resources. The CERTS microgrid concept,” 2003. [Online]. Available: http://certs.lbl.gov/pdf/50829.pdf, [Accessed: May 8, 2015]

[8] IEEE, “IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems,” 2008.

[9] J. Lee, “Islanding Detection Methods for Microgrids,” Master’s thesis, University of Wisconsin-Madison, USA, 2010.

[10] , “Microgrids. White paper,” 2011.

[11] J. Driesen and F. Katiraei, “Design for distributed energy resources,” IEEE Power and Energy magazine, no. 8, pp. 30 – 39, 2008. [12] J. A. Pecas, A. Guimaraes, and C. C. L. Monteiro, “A view of microgrids,” WIREs Energy Environ, no. 2, pp. 86 – 103, 2013.

[13] Gobierno de Canarias. Consejer´ıa de Industria y Comercio. (2008) Estad´ıstica de incidencias en 2008. [Online]. Available: http://www.gobcan.es/energia/temas/ energiaelectrica/estadisticas/

[14] Electric Power Research Institute (EPRI). (2005) Voltage sags, Swells and Interruptions Characterized in DPQ Phase II Project.

[15] Red El´ectrica de Espa˜na, “Requisitos de respuesta frente a huecos de tensi´on de las instalaciones e´olicas,” 2006.

[16] Guadaloupe Energy. [Online]. Available: http://www.guadeloupe-energie.gp/wp- content/uploads/Plaquette Geothermie English.pdf, [Accessed: May 8, 2015]

[17] TERNA. [Online]. Available: http://www.terna.it/LinkClick.aspx?fileticket= 88244, [Accessed: May 8, 2015]

[18] A. J. M. I. Jowsick, A. Arulampalam, and H. M. Wijekoon, “HVDC transmission line for interconnecting power grids in india and sri lanka,” in Industrial and Information Systems (ICIIS), 2009 International Conference on, Dec 2009, pp. 419–424.

[19] Eurasia Interconnector Project. [Online]. Available: http://www.euroasia- interconnector.com/, [Accessed: May 8, 2015]

[20] T. Hammons, E. Hreinsson, and P. Kacejko, “Proposed iceland/uk (peterhead) 1.2 gw hvdc cable,” in Universities Power Engineering Conference (UPEC), 2010 45th International, Aug 2010, pp. 1–9.

[21] L. M. Lobato, “Offshore Wind Power Potential: Opportunity for the Development of Submarine Interconnections,” in Jornadas: Integraci´on en red: soluciones para altas penetraciones e´olicas en sistemas insulares, Las Palmas de Gran Canaria. November 2010.

[22] European Network of Transmission System Operators for Electricity (ETNSO-E). (2008) Network Codes for Electricity. [Online]. Available: http://networkcodes.entsoe. eu/

[23] Central Electricity Regulatory Commission, “Indian Electricity Grid Code,” 2010.

184 [24] China Electric Power Research Institute (CEPRI), “Technical rule for connecting wind farm to power system GB/T 19963-2011,” 2011.

[25] Power Research and Development Consultant Private Limited, “Indian Wind Grid Code - draft,” 2009.

[26] E.On, “Grid code. High and Extra High Voltage,” 2006.

[27] Red El´ectrica de Espa˜na, “Borrador. Instalaciones conectadas a la red de transporte y equipo generador: requisitos m´ınimos de dise˜no, equipamiento, funcionamiento, puesta en servicio y seguridad,” 2009.

[28] Energinet, “Technical regulation for thermal power station units of 1.5 MW and higher. Regulation for grid connection TF 3.2.3,” 2008.

[29] Energinet, “Technical regulation for thermal power station units larger than 11 kW and smaller than 1.5 MW. Regulation for grid connection TF 3.2.4,” 2008.

[30] Energinet, “Technical regulation 3.2.5 for wind power plants with a power output greater than 11 kW,” 2010.

[31] IEEE, “Recommended Practice for Monitoring Electric Power Quality,” 1995.

[32] L. Zhang and M. H. J. Bollen, “Characteristic of voltage dips (sags) in power systems,” in Harmonics and Quality of Power Proceedings, 1998. Proceedings. 8th International Conference On, vol. 1, Oct 1998, pp. 555–560 vol.1.

[33] J. Morren and S. W. H. De Haan, “Ridethrough of wind turbines with doubly-fed induction generator during a voltage dip,” Energy Conversion, IEEE Transactions on, vol. 20, no. 2, pp. 435–441, June 2005.

[34] R. K. Sinha, R. Kumar, M. Venmathi, and L. Ramesh, “Analysis of voltage sag with different dg for various faulty conditions,” International Journal of Computer Commu- nication and Information System, vol. 2, no. 1, pp. 189–193, 2010.

[35] International Energy Agency - IEAWind , “2012 Annual Report,” 2013. [Online]. Availa- ble: http://www.ieawind.org/annual reports PDF/2012.html, [Accessed: May 8, 2015]

[36] North American Electric Reliability Corporation NERC, “Standard PRC-024-1 — ge- nerator frequency and voltage protective relay settings,” 2013.

185 [37] Comisi´on Nacional de Energ´ıa. (2007) Informaci´on b´asica de los sectores de la energ´ıa -2007-. [Online]. Available: http://www.cne.es/cne/doc/publicaciones/PA002 07- anexo.pdf, [Accessed: May 8, 2015]

[38] The European Wind Energy Association (EWEA). (2006) Wind Energy Map 2006. [Online]. Available: http://www.ewea.org/fileadmin/ewea documents/documents/ publications/statistics/070129 Wind map 2006.pdf, [Accessed: May 8, 2015]

[39] Red El´ectrica de Espa˜na. (2012) Informe Anual sobre la Operaci´on del Sistema El´ectrico (2012). [Online]. Available: http://www.ree.es/sites/default/files/informe anual sobre la operacion del sistema electrico.pdf, [Accessed: May 8, 2015]

[40] Ministerio de Industria, Turismo y Comercio. (2011) Plan de Energ´ıas Renovables 2011-2020. [Online]. Available: http://www.idae.es/index.php/id.670/relmenu.303/ mod.pags/mem.detalle, [Accessed: May 8, 2015]

[41] Gobierno De Canarias.Consejer´ıa De Empleo, Industria y Comercio. (2011) Anuario Energ´etico de Canarias. [Online]. Available: http://www.gobiernodecanarias.org/ industria/publicaciones/Anuario2011.pdf, [Accessed: May 8, 2015]

[42] U.S. Energy Information Administration. (2013) International Energy Statistics. [Online]. Available: http://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=2& pid=2&aid=12, [Accessed: May 8, 2015]

[43] French Ministerial Order of April 23 2008, “Design and Operating Requirements for the connection of Power Generation Facilities to MV or LV Distribution Grids,” ref.DEVE0808815A.

[44] Canary Green Energy. (2013). [Online]. Available: http://www.canarygreenenergy. com/Servicios/Serviciosenergticos/Energaelica.aspx, [Accessed: May 8, 2015]

[45] Red El´ectrica de Espa˜na. (2010) Retos de la Penetraci´on de Energ´ıas Renovables en los Sistemas El´ectricos Canarios. [Online]. Available: http://jornadasenergeticas.files. wordpress.com/2010/02/retos integracion renovables v5.pdf, [Accessed: May 8, 2015]

[46] EDF. (2010) Protection de d´ecouplage pour le raccordement d’une production d´ecentralis´ee en HTA et en BT dans les zones non interconnect´ees. SEI REF 04 (in french). [Online]. Available: http://saint-pierre-et-miquelon.edf.com/fichiers/fckeditor/ Commun/SEI/corp/sei ref 04 protection decouplagev5.pdf, [Accessed: May 8, 2015]

186 [47] EDF. (2012) How to manage intermittency on islands? [Online]. Available: http://chercheurs.edf.com/fichiers/fckeditor/Commun/R et D/7nov EnR Intermittence/Presentations 7nov12/10 Barlier EDF SEI.pdf, [Accessed: May 8, 2015]

[48] European Network of Transmission System Operators for Electricity (ENTSO-E). (2013) Network code overview. [Online]. Available: https://www.entsoe.eu/major- projects/network-code-development/, [Accessed: May 8, 2015]

[49] European Network of Transmission System Operators for Electricity (ENTSO-E). (2013) NC RfG Implementation Guidelines. [Online]. Availa- ble: https://www.entsoe.eu/major-projects/network-code-development/requirements- for-generators/,[Accessed: May 8, 2015]

[50] European Parliament and The Council of European Union. (2009) Directive 2009/72/CE. [Online]. Available: http://eur-lex.europa.eu/LexUriServ/LexUriServ. do?uri=OJ:L:2009:211:0055:0093:EN:PDF, [Accessed: May 8, 2015]

[51] European Network of Transmission System Operators for Electricity (ENTSO- e). (2014) Voluntary Regional Group Isolated Systems. [Online]. Avai- lable: https://www.entsoe.eu/about-entso-e/system-operations/regional-groups/vrg- isolated-systems/, [Accessed: May 8, 2015]

[52] J. C. Ausin, D. N. Gevers, and B. Andresen, “Fault ride-through capability test unit for wind turbines,” Wind Energy, vol. 11, no. 1, pp. 3–12, 2008.

[53] J. I. Llorente and M. Linares, “Dispositivo generador de huecos de tensi´on,” Patent ES 2 263 375, 11 16, 2007.

[54] A. E. E. AEE, “Procedimiento de Verificaci´on, Validaci´on y Certificaci´on de los requi- sitos del P.O.12.3. sobre la Respuesta de las Instalaciones E´olicas y Fotovoltaicas ante Huecos de Tensi´on,” 2011.

[55] I. J. Gabe, H. Grundling, and H. Pinheiro, “Design of a voltage sag generator based on impedance switching,” in IECON 2011 - 37th Annual Conference on IEEE Industrial Electronics Society, 2011, pp. 3140–3145.

[56] E. Collins and R. L. Morgan, “A three-phase sag generator for testing industrial equip- ment,” Power Delivery, IEEE Transactions on, vol. 11, no. 1, pp. 526–532, 1996.

[57] C. A. Platero, C. Veganzones, F. Bl´azquez, D. Ram´ırez, S. Mart´ınez, J. A. S´anchez, J. Rodr´ıguez, and N. Herrero, “Banco de ensayo de equipos el´ectricos, generadores o consumidores frente a huecos de tensi´on,” Patent ES 2 325 902, 09 23, 2009.

187 [58] C. Wessels, R. Lohde, and F. Fuchs, “Transformer based voltage sag generator to per- form lvrt and hvrt tests in the laboratory,” in Power Electronics and Motion Control Conference (EPE/PEMC), 2010 14th International. IEEE, 2010, pp. T11–8.

[59] L. Dongyu, Z. Honglin, X. Shuai, and Y. Geng, “A new voltage sag generator base on power electronic devices,” in Power Electronics for Distributed Generation Systems (PEDG), 2010 2nd IEEE International Symposium on, 2010, pp. 584–588.

[60] E.-C. Nho, J.-H. Jung, I.-D. Kim, T.-W. Chun, H.-G. Kim, N.-S. Choi, and J. Choi, “Voltage disturbance generator with phase jump for the test of microgrid,” in Power Electronics Conference (IPEC), 2010 International, 2010, pp. 487–491.

[61] E.-C. Nho, Y.-H. Lee, J.-K. Seok, I.-D. Kim, N.-S. Choi, T.-W. Chun, and H.-G. Kim, “Characteristics of a power quality disturbance generator for the test of microgrid with sts,” in Telecommunications Energy Conference, 2009. INTELEC 2009. 31st Interna- tional, 2009, pp. 1–4.

[62] M. Garc´ıa, M. Mart´ınez, and D. L´opez, “Device that generates electrical disturbances,” Patent WO 2012/113 951 A1, 08 30, 2012.

[63] A. Uphues, K. Notzold, R. Wegener, K. Fink, M. Bragard, R. Griessel, and S. Soter, “Inverter based test setup for LVRT verification of a full-scale 2 MW wind power con- verter,” in Power Electronics and Applications (EPE), 2013 15th European Conference on, 2013, pp. 1–5.

[64] R. Lohde and F. Fuchs, “Laboratory type pwm grid emulator for generating disturbed voltages for testing grid connected devices,” in Power Electronics and Applications, 2009. EPE ’09. 13th European Conference on, 2009, pp. 1–9.

[65] K. Oranpiroj, S. Premrudeepreechacharn, M. Ngoudech, W. Mungjai, K. Yingkayan, and T. Boonsai, “The 3-phase 4-wire voltage sag generator based on abc algorithm,” in Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology, 2009. ECTI-CON 2009. 6th International Conference on, vol. 01, 2009, pp. 82–85.

[66] X. Yan, B. Zhang, S. Wang et al., “Three-phase voltage sag generator,” Patent CN 10 887 074 A, 11 17, 2010.

[67] J. Chen, X. Zhang, J. Li et al., “Voltage sag generator,” Patent CN 102 244 466 A, 11 16, 2011.

188 [68] C. Saniter, J. Janning, and A. Bocquel, “Test bench for grid code simulations for multi- MW wind turbines,” in Power Electronics and Applications, 2007 European Conference on, 2007, pp. 1–10.

[69] R. Piwko, E. Larsen, and C. Wegner, “Variable frequency transformer - a new alternative for asynchronous power transfer,” in Power Engineering Society Inaugural Conference and Exposition in Africa, 2005 IEEE, July 2005, pp. 393–398.

[70] A. Merkhouf, P. Doyon, and S. Upadhyay, “Variable frequency transformer;concept and electromagnetic design evaluation,” Energy Conversion, IEEE Transactions on, vol. 23, no. 4, pp. 989–996, Dec 2008.

[71] W. E. M. Ayres, “The application of the induction voltage regulator,” Electrical Engi- neers, Journal of the Institution of, vol. 69, no. 418, pp. 1208–1218, October 1931.

[72] R. Tabb, “The induction regulator,” Students’ Quarterly Journal, vol. 3, no. 11, pp. 158–159, February 1933.

[73] C. Veganzones, J. Sanchez, S. Martinez, C. Platero, F. Blazquez, D. Ramirez, J. Arribas, J. Merino, N. Herrero, and F. Gordillo, “Voltage dip generator for testing wind turbines connected to electrical networks,” Renewable Energy, vol. 36, no. 5, pp. 1588 – 1594, 2011. [Online]. Available: http://www.sciencedirect.com/science/article/ pii/S0960148110004933

[74] C. Bohn and D. Atherton, “An analysis package comparing pid anti-windup strategies,” Control Systems, IEEE, vol. 15, no. 2, pp. 34–40, 1995.

[75] IEC, “Iec 61400-21. wind turbines – part 21. measurement and assessment of power quality characteristics of grid connected wind turbines.” 2008.

[76] National Grid, “Guidance notes. power park modules. technical report. issue 3.” 2008.

[77] T. Lipo, Introduction to AC machines design. Wisconsin Power Electronics Research Center. University of Wisconsin, 2011.

[78] J. Pyrhonen, T. Jokinen, and V. Hrabovcov´a, Design of rotating electrical machines. Wiley, 2009.

[79] H. Toliyat and G. Kliman, Handbook of Electric Motors, ser. Electrical and computer engineering. Taylor & Francis, 2004. [Online]. Available: http: //books.google.es/books?id=4-Kkj53fWTIC

189 [80] Baum¨uller DS servo motors. [Online]. Available: http://www.baumueller.de/ e servo motors dsda.htm, [Accessed: May 8, 2015]

[81] Pujol Muntal´a Series IPC. [Online]. Available: http://www.pujolmuntala.eu/es/index. php?doc=is&ext=php, [Accessed: May 8, 2015]

[82] Baum¨uller. [Online]. Available: http://www.baumueller.de/e servo controller bmaxx4400.htm, [Accessed: May 8, 2015]

[83] Schneider Electric, “Variable speed drives Altivar 71 and Altivar 71 Plus.” [Online]. Available: http://download.schneider-electric.com/files?p File Id=28891731&p File Name=Altivar-71-Catalogue EN Ed-2011-09-(web).pdfp, [Accessed: May 8, 2015]

[84] Elap, “Encoder RM36 specifications.” [Online]. Available: http://www.elap.it/eng/ Download/Absolute-Encoders/RM36 en.pdf, [Accessed: May 8, 2015]

[85] Schneider Electric, “Encoder XCC1506PS50X specifications.” [Online]. Availa- ble: http://pdf.schneider-electric.nu//files/partnumbers/XCC1506PS50X document. pdf, [Accessed: May 8, 2015]

[86] dSPACE, “dSPACE DS1104 r&d Controller Board.” [Online]. Available: https://www.dspace.com/en/pub/home/products/hw/singbord/ds1104. cfm, [Accessed: May 8, 2015]

190