design of a power electronic assisted series …

DESIGN OF A POWER ELECTRONIC ASSISTED SERIES

COMPENSATOR FOR GRID VOLTAGE REGULATION

MASTER OF SCIENCE THESIS

GAUTHAM RAM CHANDRA MOULI

JULY 2013

DELFT UNIVERSITY OF TECHNOLOGY

FACULTY OF ELECTRICAL ENGINEERING, MATHEMATICS AND COMPUTER SCIENCE

ELECTRICAL POWER PROCESSING GROUP

MA

ST

ER

OF

SC

IEN

CE

TH

ES

IS



MASTER OF SCIENCE THESIS


JULY 2013

DELFT UNIVERSITY OF TECHNOLOGY

FACULTY OF ELECTRICAL ENGINEERING, MATHEMATICS AND COMPUTER SCIENCE

ELECTRICAL POWER PROCESSING GROUP

iii



THESIS

submitted in partial fulfillment of the

requirements for the degree of

MASTER OF SCIENCE in

ELECTRICAL ENGINEERING

by


க ௌதம ராம சநதிரக ௌலி

Electrical Power Processing Group

Department of Electrical Sustainable Energy

Faculty EEMCS

Delft University of Technology

Delft, the Netherlands

www.ewi.tudelft.nl

GE Global Research - Europe

Freisinger Landstrasse 50

D-85748 Garching bei München

Germany

www.ge.com

http://www.ewi.tudelft.nl/en/the-faculty/departments/electrical-sustainable-energy/

v



AUTHOR : GAUTHAM RAM CHANDRA MOULI

STUDENT ID : 4180496

EMAIL ID : [email protected]

DATE OF DEFENSE : 30TH

JULY 2013

THESIS COMMITTEE:

Chair : prof. dr. eng. J.A. Ferreira, Faculty EEMCS, TU Delft

University supervisor : prof.dr.ir. P. Bauer, Faculty EEMCS, TU Delft

Committee Member : prof.dr. M. Popov, Faculty EEMCS, TU Delft

Advisor : ir.V. Prasanth, Faculty EEMCS, TU Delft

COMPANY SUPERVISORS:

dr.T.Wijekoon , GE Global Research Center, Munich

dr.A.Panosyan, GE Global Research Center, Munich

ir.Eva-Maria Baerthlein, GE Global Research Center, Munich

mailto:[email protected]

vii

ABSTRACT

In recent years, high penetration of distributed generation (DG) driven by PV panels and

heavy load insertion in the distribution network has led to frequent voltage fluctuations in

the form of undervoltage and overvoltage. Voltage control using traditional voltage

regulators are unable to cope with this situation as frequent tap changes reduce the lifetime

of the mechanical taps due to arcing. Further, the nature of European distribution network

in general, makes voltage control through shunt compensation methods typically

ineffective and expensive. Series compensation through centralized on-load tap changing

(OLTC) distribution transformers or feeder-specific compensators is hence a suitable

strategy for voltage regulation in Europe.

The thesis work describes the design of a novel, power electronic assisted OLTC

transformer that provides voltage regulation in the European distribution network through

series compensation. The aim is to ensure that the tap changes occur in an arc free manner,

thus providing for long lifetime of the OLTC. Different topologies for OLTC are examined

in detail and shortlisted based on the steady state operation and power ratings of

transformer and switches required. A novel design of a partially rated autotransformer is

proposed which has taps developed through a combination of no-load switches and a single

hybrid switch. The hybrid switch is composed of a mechanical switch and two

semiconductor switches which are used for steady state and tap change operation

respectively. The mechanical switch ensures low steady state losses and the semiconductor

switches are used for arc-free tap changing. Back-back series connected IGBT with anti-

parallel diodes are used for the two semiconductor switches and voltage polarity based 4-

step commutation is used for commutation between the taps.

The novel design of the OLTC autotransformer is cost effective, efficient and has long

lifetime. The complete system is simulated in the PLECS and the steady state and transient

operation of the system have been investigated. The OLTC has customized for application

in both MV and LV three phase distribution network. Detailed study of the failure

mechanisms owing to internal and external faults is performed and possible protection

mechanisms are suggested. A low level control mechanism is also developed, thus

providing a holistic design for building a prototype.

viii

ACKNOWLEDGEMENTS

The research work was carried out at the Electrical Power Processing Group, TU Delft in

joint collaboration with GE Global Research, Munich. I take this opportunity to extend a

warm hand of gratitude to all the people who have supported me in this endeavor.

Firstly, I am immensely grateful to my supervisor Prof. Pavol Bauer from the Electrical

Power Processing Group, TU Delft. His constant guidance throughout the phase of the

thesis work and his experience in the area of tap changing transformers has been of great

value. I express my sincere gratitude to my supervisor from GE Global Research Munich

namely Dr.Thiwanka Wijekoon, Dr.Ara Panosyan and Ir.Eva-Maria Baerthlein. I have

known them from my internship days at GRC Munich and I am thankful to them for

providing me this opportunity. My interaction with them provided a practical perspective

to the design of the voltage regulator and motivated me to be innovative in my approach.

Eva especially helped me a lot with the critical aspects of the modeling of the line and

transformer and patiently answered all my queries over the months. I also thank

Prof.Marjan Popov and Prof.Braham Ferreira for consenting to be a part of my thesis

committee.

I would like to thank my advisor, Ir.V.Prasanth who has been a good friend and guide

since the day I got my admission into the Msc program at TU Delft. He made sure that he

was always available for a discussion and guided me to reach a consensus with regard to

various problems faced in the course of the thesis. He meticulously read my thesis and

helped me improve it in all aspects. I sincerely acknowledge Ir.Todor Todorcevic for the

lively discussions on various aspects of power electronics. The brainstorming sessions

helped me clear doubts and provided me with new ideas. I also thank my longtime buddy

Nithya Anand for reviewing my thesis and giving me valuable feedback.

I am grateful to the Nuffic for supporting my studies at TU Delft through the Huygens

Scholarship program.

Finally, I wish to acknowledge the support of my parents Mrs.M.Bharathi and

Mr.S.Chandra Mouli, my sister Manasa, my grandparents and Kamakshi for their constant

support and encouragement during the execution of this thesis work. I thank all my friends

who made my time at TU Delft enjoyable and memorable.

I am indebted to the Almighty for his blessings! जय ह िद !

ix

LIST OF ABBREVIATION

MV – Medium voltage

LV – Low voltage

DSO – Distribution system operator

DG – Distributed generation

PV – Photovoltaic

UPFC – Unity power flow controller

pf – Power factor

rms – Root mean square

pu – Per unit

OLTC – On load tap changers

LDC – Line drop Compensation

FACTS – Flexible AC transmission system

SVC – Static VAr Compensators

TCR – Thyristor–controlled reactor

TSC – Thyristor–switched capacitor

TCSC – Thyristor–controlled series

TCSR – Thyristor–controlled series reactor

STATCOM – Static synchronous compensator

DVR – Dynamic voltage restorers

UPFC – Unified power flow controllers

PWM – Pulse Width Modulation

IGBT – Insulated gate bipolar transistor

MOSFET – Metal oxide semiconductor field effect transistor

MV – Medium voltage

LV – Low voltage

NL – No load

M – Mechanical switch

BS – Bidirectional electronic switch

x

LIST OF SYMBOLS

Chapter 2:

Rs – Resistance of feeder [Ω]

X s – Reactance of feeder [Ω]

V – Load voltage at end of feeder [V]

E – Supply voltage or sending end voltage at start of feeder [V]

Sload – Apparent power drawn by the load [VA]

P1 – Real power drawn by the load [W]

Q1 – Reactive power drawn by the load [VAr]

I1 – Current drawn by the load [A]

Δ V – Voltage drop along feeder feeder [V]

Chapter 4:

V1 – Voltage across input terminals of transformer [V]

V1 MAX – Maximum voltage across the input of transformer [V]

V2 – Voltage across the output terminals of the transformer [V]

V1W – Voltage across the secondary winding of transformer [V]

V2W – Voltage across the primary winding of transformer [V]

E1 – Voltage at input of transformer considering drop across leakage impedance [V]

E2 – Voltage at output of transformer considering drop across leakage impedance [V]

I1W – Current through primary winding [A]

I2W – Current through secondary winding [A]

I1 – Current through the input terminals of transformer [V]

I2 – Current through the output terminals of the transformer [V]

N1 – Number of turns of primary

N2 – Number of turns of secondary

x – Tap position on OLTC

Vb – Blocking rating of switches [V]

Vsource – Source voltage [V]

Vload – Voltage delivered to load [V]

Isource – Current drawn from the source [A]

Iload – Current supplied to the load [A]

S – Apparent power [VA]

Fc – Capacity multiplication factor

r – Ratio of the output voltage to the input voltage of autotransformer

Swinding1– Apparent power through the primary winding of the transformer [VA]

xi

Swinding2– Apparent power through the secondary winding of the transformer [VA]

Chapter 5:

Vx –Line–neutral phase voltage of phase x [V]

Vxy –Line–line voltage between phase x & y [V]

Vm – Amplitude of phase voltage [V]

Δ Vx – Series compensation voltage injected by the OLTC in Phase x [V]

C% – Tap position expressed as a fraction (0 to 1)

Vx’ – Line–neutral phase voltage of phase x at the output of OLTC [V]

Vxy’ – Line–line voltage between phase x and y at the output of OLTC [V]

Rs – Self resistance of the line per unit length [/km]

Rm – Mutual resistance of the line per unit length [/km]

Ls – Self inductance of the line per unit length [mH/km]

Lm – Mutual inductance between the phases per unit length [mH/km]

Cp – Capacitance per unit length between line & point of coupling of all phases

[uF/km]

Cg – Capacitance per unit length between common point of coupling of all phases &

ground [uF/km]

Chapter 6:

Iload – Load current [A]

Zline – Line impedance [Ω]

Vb – Blocking voltage rating of switch [V]

Lleak – Leakage inductance of tap [H]

Rleak – Resistance of the tap winding [Ω]

Vtap – Voltage of one tap of OLTC [V]

Rsnub – Snubber resistance [Ω]

Csnub – Snubber capacitance [F]

Idis – Capacitor discharge current [A]

Iosc – LC oscillation current through tap leakage impedance & snubber [A]

fosc – Frequency of LC oscillation between tap leakage impedance & snubber [Hz]

w0 – Angular frequency corresponding to fosc [rad/s]

Chapter 7:

xl1 – Leakage inductance of common winding [H]

r1 – Winding resistance of the common winding [Ω]

xl2 – Leakage inductance of series winding [H]

r2 – Winding resistance of the series winding [Ω]

Lm – Magnetizing inductance of transformer [H]

Rm – Magnetizing resistance of transformer [H]

xii

Vx –Line–neutral phase voltage of phase x [V]

Vxy –Line–line voltage between phase x & y [V]

Vx’ – Line–neutral phase voltage of phase x at the output of OLTC [V]

Vxy’ – Line–line voltage between phase x and y at the output of OLTC [V]

Vload_xy – Line–line voltage between phase x and y at the load [V]

Vload_x – Line–neutral phase voltage of phase x at the load [V]

VBS1 – Voltage across the electronic switch BS1 of the hybrid switch [V]

VBS2 – Voltage across the electronic switch BS2 of the hybrid switch [V]

I BS1 – Current through the electronic switch BS1 of the hybrid switch [A]

I BS2 – Current through the electronic switch BS2 of the hybrid switch [A]

Vcap – Voltage across the snubber capacitor [V]

Icap – Current through the snubber capacitor [A]

Ileak – Current through the tap leakage inductance [A]

Chapter 8:

Vref – Voltage set point reference of the control loop [A]

BW – Voltage bandwidth [A]

ΔU – Voltage error in the control loop [A]

xiii

LIST OF FIGURES

Figure Page

Chapter 2

Fig. 2.1 – Conventional downstream power flow with minimum voltage at the

end of line giving rise to undervoltage 6

Fig. 2.2 – Occurrence of undervoltage due to lagging current drawn by the load 7

Fig.2.3 – Fluctuations in power produced from a group of 100 PV panels

distributed all over Germany over a period of three days 7

Fig. 2.4 – Present & future scenario with downstream and upstream power flow

giving rise to both undervoltage and overvoltage 8

Fig 2.5 – Schematic of feeder system during low levels of PV injection 9

Fig 2.6 – Schematic of feeder system during high levels of PV injection 9

Fig. 2.7 – Evaluation of load voltage as a function of load power & line

impedance 10

Chapter 3

Fig.3.1 – OLTC with taps on primary (a) and secondary (b) side 15

Fig.3.2 – OLTC auto transformer providing series compensation with taps on load

side that can both step–up and step–down load voltage 16

Fig 3.3 – Evolution of FACTS technology over the years 17

Fig. 3.4 – Shunt (left) & series (right) compensation using FACTS devices 18

Fig. 3.5 – Topology of a Unified Power Flow Controller (UPFC) 19

Fig. 3.6 – Distribution system layout in North America & Europe – a comparison 20

Fig.3.7 – Voltage regulation using reactive power compensation 21

Fig.3.8 – Voltage regulation using centralized full transformer 22

Fig.3.9 – Voltage regulation using feeder specific decentralized series

compensator that are partially rated for the compensating power 22

Fig.3.10 – Selective compensation using partially rated series voltage

compensator 23

Fig.3.11a – PV generation profile as a function of time 24

Fig.3.11b – Voltage profile at the head of feeder as a function of time 25

Fig.3.11c – Voltage profile at the end of feeder as a function of time 25

Chapter 4

Fig 4.1a – Tap changing in on–load tap changer 29

Fig 4.1b – ‘Break and make’ mechanism resulting in loss of load 30

Fig 4.1c – ‘Make and break’ mechanism using an impedance during tap change 30

Fig 4.2a – Ideal two winding transformer with turn ratio a=N1/N2 32

Fig 4.2b – Autotransformer is boost (left) and buck (right) operation 33

xiv

I. Topology 1 & 1a 36

II. Topology 2 39

III. Topology 3 & 3a 40

IV. Topology 4 & 4a 44

V. Topology 5 & 5a 47

VI. Topology 6 49

VII. Topology 7 51

Fig 4.3a – Autotransformer voltages & currents for (a) buck & (b) boost

operations 42

Fig 4.3b – Direction of currents in autotransformer in topology 3 43

Fig 4.4 – Schematic to estimate the voltage rating of switches in topology 6 50

Chapter 5

Fig 5.1a – Y connection of OLTC transformer to a three phase four wire network 57

Fig 5.1b – Topology 3 with connection points S, L and SL 57

Fig 5.1c – Phasor diagram for four wire network with Y connected transformers 58

Fig 5.1d – Simulation of Y connection of OLTC to a three phase four wire

network 58

Fig 5.2a – Simulation of Δ connection of OLTC transformers to three phase

network 59

Fig 5.2b – Phasor diagram for three wire network with Δ connected transformers 60

Fig 5.2c – Simulation of Δ connection of OLTC transformers to three phase

network 61

Fig 5.3a – Y connection of OLTC transformers in a three wire network 63

Fig 5.3b – Phasor diagram for three wire network with Y connected transformers 63

Fig 5.3c – Simulation of Y connection of OLTC transformers in a three wire

network 64

Fig 5.4a – Open Δ connection of OLTC transformers in three wire network 65

Fig 5.4b – Phasor diagram for Δ distribution network with open Δ connected

transformers 66

Fig 5.4c – Simulation of open Δ connection of OLTC transformers in three phase

network 66

Fig 5.5 – T section model of distribution network 69

Fig 5.6a – Single phase PLECS model of Topology 1 and 1a 70

Fig 5.6b – Single phase PLECS model of Topology 2 71

Fig 5.7a – PLECS model of Topology 3 & 3a connected to a three phase

distribution network 72

Fig 5.7b – PLECS model of Topology 3 72

Fig 5.7c – PLECS model of Topology 3a 73

Fig 5.7d – Voltage waveforms of OLTC 74



Fig 5.8b – PLECS model of OLTC based on Topology 4 (left) & 4a (right) 76

xv



Fig 5.9b – PLECS model of OLTC based on Topology 5 78

Fig 5.9c – PLECS model of OLTC based on Topology 5a 78

Fig 5.10a – PLECS model of Topology 6 connected to a three phase distribution

network 79

Fig 5.10b – PLECS model of OLTC based on Topology 6 80

Fig 5.11a – PLECS model of Topology 7 connected to a three phase distribution

network 81

Fig 5.11b– PLECS model of OLTC based on Topology 7 81

Chapter 6

Fig 6.1 – Schematic of a simple hybrid tap changer 87

Fig 6.2 – Schematic of no–load switch in series with an electronic switch 88

Fig 6.3a – Diverter switch type voltage regulator using resistors [33] 90

Fig 6.3b – Diverter switch type voltage regulator using inductors [32] 90

Fig 6.4 – Novel realization of topology 3a using no–load and hybrid switch 91

Fig 6.5a – Steady state operation of novel topology when any green–tap is ON 91

Fig 6.5b – Steady state operation of novel topology when any red–tap is ON 92

Fig 6.6.1 – Step0 of the tap changing process 93








Fig 6.7 – Novel realization of topology 4 using no–load and hybrid switch 98

Fig 6.8 – Bidirectional electronic switch for use in hybrid switch 99

Fig 6.9 – Bidirectional electronic switch for use in LV OLTC 100

Fig 6.10a – Voltage polarity based 4–step commutation 102

Fig 6.10b – (Clockwise) Step1 to Step4 for voltage polarity based commutation

when V12>0 102

Fig 6.11a – Current polarity based 4–step commutation 103

Fig 6.11b – (Clockwise) Step1 to Step4 for current polarity based commutation

when Iload>0 103

Fig 6.12– Effect of tap leakage inductance of tap commutation 106

Fig 6.13a – RC Snubber connected across the electronic switch 107

Fig 6.13b – Overview of currents when RC snubber is connected across BS1 and

BS2 108

Fig 6.14a – RC Snubber connected across the electronic switch 109

Fig 6.14b – Overview of currents when RC Snubber connected across BS1 &BS2

110

xvi

Chapter 7

Fig 7.1 – Schematic of a topology 3a 114

Fig 7.2 – Autotransformer of topology 3a during boost and buck operation 114

Fig 7.3 – Model of autotransformer corresponding to topology 3a 115

Fig 7.3a – OLTC system for MV having two units connected in open–delta 117

Fig 7.3b – OLTC unit with no–load switches and single hybrid switch 117

Fig 7.3c – Load block of Fig 7.3a 118

Fig 7.3d – IGt & IGb block of Fig 7.3b 118

Fig 7.4a – OLTC system for LV having three units connected in Wye 118

Fig 7.4b – OLTC unit with no–load switches and single hybrid switch 119

Fig 7.4c – Load block of Fig 7.4a 119

Fig 7.4d – MFt & MFb block of Fig 7.4b 119

Fig7.5a – Line-line voltages [V] at input Vxy & output of OLTC Vxy’ as a function

of time [s] when both OLTC units are set for 4% positive compensation for MV

scenario

120

Fig7.5b – Line-line voltages [V] at input Vxy & output of OLTC Vxy’ as a function

of time [s] when each OLTC unit is set for 4% and 8% positive compensation for

MV scenario

121

Fig7.5c – Line-line voltages [V] at input Vab & output of OLTC Vab’ and at load

Vload_ab as a function of time[s] for line a-b when the OLTC provides 8% positive

compensation for MV

121

Fig7.5d – Phase voltages [V] at input Vx & output of OLTC Vx’ as a function of

time[s] when the three OLTC units are set for -2%, 6% and 8% compensation for

LV

122

Fig7.6a – Line-line voltages [V] at input Vxy & output Vxy’ of OLTC as a function

of time [s] when both units are set for 4% negative compensation for MV scenario 123

Fig7.6b – Phase voltages [V] at input Va & output Va’ of OLTC as a function of

time[s] when an active load is connected at the load end Vload_a of the LV OLTC

for -8% compensation

124

Fig7.7a – Line-Line voltage [V] at input Vab & output of OLTC Vab’ as a function

of time[s] for phase ab when it goes from 0% to 2% compensation in positive half

cycle

125

Fig7.7b – Phase voltage [V] at input & output of OLTC as a function of time[s]

for phase a when it goes from 0% to 2% compensation in negative half cycle 126

Fig7.8 – Voltage [V] and current [A] of the electronic switch IGt and IGb during

4-step commutation from IGt to IGb switch in presence of snubber 127

Fig7.9a – Voltage [V] & current [A] of electronic switches during 4-step

commutation from switch IGt to IGb using snubber of 100 μF (top) and 10 μF

(bottom)

129

Fig7.9b – Voltage [V] & current [A] of electronic switches during 4-step

commutation from switch IGb to IGt in the absence of overvoltage snubber

(Parasitic capacitance of 1 nF)

130

Fig7.10 – Voltage [V] & current [A] of electronic switches during 4-step

commutation from switch IGt to IGb when current at commutation instant is

900A (top) and 400A (bottom)

131

xvii

Fig7.11 – Voltage [V] & current [A] of snubber capacitor during 4-step

commutation depicting the LC oscillation with leakage impedance 132

Chapter 8

Fig 8.1a – Disconnectors connected to LV series compensator 137

Fig 8.1b – Disconnectors connected to MV series compensator 137

Fig 8.2– Short circuit through two no–load switches 138

Fig 8.3a – Bypass switch connected to LV series compensator 139

Fig 8.3b – Bypass switch connected to MV series compensator 140

Fig 8.4 – Bypass switch connected to novel design of topology 3a 142

Fig 8.5a – Bypass switch connected to LV series compensator 143

Fig 8.5b – Bypass switch connected to MV series compensator 143

Fig 8.6 – Block diagram of the control mechanism for the series compensator 145

Fig 8.7a – Tap change initiated after voltage exceeds the BW limit for a period of

time > TD 147

Fig 8.7b – Dynamic time delay based on voltage error using inverse time delay

curve 148

Chapter 9

Fig 9.1 – Feeder specific compensation through partially rated series compensator 152

Fig 9.2 – Novel realization of topology 3a using no–load and hybrid switch 154

xviii

LIST OF TABLES

Table Page

Chapter 2

Table 2.1 – Voltage variation in grid with and without presence of DG 8

Chapter 3

Table 3.1 – Analysis of the critical aspects of series and shunt compensation 14

Table 3.2 – Comparison of series compensation using OLTC & full power

electronic power based FACTS devices 26

Chapter 4

Table 4.1a – Sizing of transformer and switch for topology 1 and 1a 38

Table 4.1b – Sizing of transformer and switch for topology 1 and 1a 38

Table 4.2 – Sizing of transformer and switch for topology 2 40

Table 4.3a – Winding, source and load currents in autotransformer during

different modes 42

Table 4.3b – Sizing of components for topology 3 43

Table 4.3c – Sizing of transformer and switch for topology 3 & 3a 44

Table 4.4 – Sizing of transformer and switch for topology 4 & 4a 46

Table 4.5 – Sizing of transformer and switch for topology 5 & 5a 48



Table 4.8a – Summary of sizing of transformer for all topologies 53

Table 4.8b – Summary of sizing of switches for all topologies 54

Chapter 5

Table 5.1 – Parameters of three phase distribution network 56

Table 5.2: Connection scheme and rating of transformer for MV and LV 67

Table 5.3: Parameters of Overhead line/cable used by EnBW in German utilities 69

Chapter 6

Table 6.1 – Comparison of characteristics of mechanical & electronic switches 86

Table 6.2 – Two step method to create/interrupt current through NL switch 89

Table 6.3 – Reduction in number of switch & voltage ratings using novel design 92

Table 6.4 – Switch states for Voltage polarity based 4–step commutation 101

Table 6.5 – Switch states for current polarity based 4–step commutation 103

Table 6.6 – 8 different firing sequence for the gating circuit of the bidirectional

switch 105

Chapter 7

Table 7.1 – MV and LV transformer parameters 115

Table 7.2 – System parameters for MV and LV 116

xix

CONTENTS

Chapter Page

Chapter 1 : Introduction and motivation for research 1

Chapter 2 : Voltage fluctuation in the distribution network 5

2.1 Causes for voltage fluctuation 5

2.1.1 The scenario in the past 5

2.1.2 The scenario in the present and for the future 7

2.2 Magnitude of voltage variation at load end 10

2.3 Effect of voltage fluctuation on utilities and load 11

2.4 Other voltage issues 12

2.5 Summary and conclusion 12

Chapter 3 : Voltage compensation methodologies 13

3.1 On load and off circuit tap changing transformers 14

3.2 Reactive power compensation through inductors/capacitors and

synchronous condensers 16

3.3 FACTS and power electronic based voltage compensators 17

3.4 Voltage control through grid connected PV inverters 19

3.5 Which method suits the European distribution network? 20

3.6 Comparison of voltage compensation through self-commutated switch

based series FACTS devices and fractionally rated tap changers 26


Chapter 4 : OLTC topologies – Design and sizing of components 29

4.1 Conventional two winding transformer 31

4.2 Autotransformer 32

4.2.1 Boost autotransformer 33

4.2.2 Buck autotransformer 34

4.3 OLTC Topologies 35

I. Topology 1 and 1a 35

II. Topology 2 38

III. Topology 3 and 3a 40

IV. Topology 4 and 4a 44

V. Topology 5 and 5a 47

VI. Topology 6 49

VII. Topology 7 51

4.4 Rating of components for different topology 53

4.5 Summary and conclusion

54

xx

Chapter 5: Connection schemes for three phase network and simulation of

OLTC topologies 55

5.1 Mathematical analysis for connection of OLTC to a three phase network 55

5.1.1 Three phase four wire network with Y connected OLTC 56

5.1.2 Three phase network with Δ connected OLTC 59

5.1.3 Three phase three wire network with Y connected OLTC 62

5.1.4 Three phase three wire network with open-Δ connected OLTC 64

5.1.5 Connection scheme for MV and LV distribution network 67

5.2 Modeling of distribution network as T-section 68

5.3 Modeling and simulation of OLTC topologies 70

I. Topology 1, 1a and 2 70

II. Topology 3 and 3a 71

III. Topology 4 and 4a 75

IV. Topology 5 and 5a 77

V. Topology 6 79

VI. Topology 7 80

5.4 Topology selection 82


Chapter 6: Optimization of design of OLTC and Hybrid switch 85

6.1 Electronic, mechanical and hybrid switches for OLTC taps 85

6.2 Novel design of OLTC using hybrid switch 88

6.2.1 Main issues in realizing hybrid switches for OLTC topologies 88

6.2.2 Concept of no-load switches 88

6.2.3 Diverter switch type voltage regulators 89

6.2.4 Design of OLTC using no-load switches and single hybrid switch 90

6.2.5 7-step tap changing methodology 93

6.2.6 Advantages of the proposed design 97

6.2.7 Novel design for topology 4 and short listing of topologies 97

6.3 Bidirectional electronic switch for hybrid tap changer 98

6.3.1 Different types of bidirectional electronic switch 98

6.3.2 Choice of bidirectional electronic switch for MV and LV 100

6.4 4-step commutation based on voltage and current polarity 101

6.4.1 4-step commutation based on voltage polarity 101

6.4.2 4-step commutation based on current polarity 102

6.4.3 Choice of commutation strategy 104

6.5 Overvoltage snubber and Effect of tap leakage inductance 105

6.5.1 Single overvoltage snubber connected between BS1 and BS2 106

6.5.2 Two overvoltage snubber with each across BS1 and BS2 109

6.5.3 Choice of snubber design 110


111

xxi

Chapter 7: Simulation of OLTC system for MV and LV distribution

network 113

7.1 Modeling of autotransformer with taps 113

7.2 Simulation model of OLTC system for MV and LV 116

7.2.1 Simulation model of OLTC system for MV 116

7.2.2 Simulation model of OLTC system for LV 118

7.3 Simulation of OLTC system for steady state and transient operation 119

7.3.1 Steady state operation of OLTC system 120

7.3.2 Transient operation of OLTC system 124


Chapter 8: Protection and control of series compensator 135

8.1 Internal and external fault conditions and protection mechanism 136

8.1.1 Internal fault conditions and protection mechanism 136

8.1.2 External fault conditions and protection mechanism 141

8.2 Control of series compensator 144


Chapter 9: Conclusion and scope for future work 151

9.1 Overview of thesis work 151

9.2 Results and conclusion 152

9.3 Scope for future work 154

References 157

Appendix - IEEE paper 161

Chapter 1: Introduction & motivation for research

1

Chapter 1

Introduction and

Motivation for research

Voltage fluctuation is a usual phenomenon that happens in the distribution grid. Traditional

downstream power flow results in a voltage drop along the feeder causing undervoltage at

feeder end. To counteract this effect tap changing mechanism in voltage regulators and in

distribution transformers are used to set the voltage at the feeder head at a higher value to

compensate for the line drops. However in recent years there has been a high penetration of

distributed generation (DG) in the LV distribution network mainly driven by solar and the

situation is only expected to increase in the future. This combined with heavy load insertion to

the grid such as electric vehicle charging has made the voltage control more complicated than

before. Large variation in DG power owing to short and long term fluctuations in wind and

sunshine results in large amplitude (up to ±10%) and frequent variations in load voltage.

Moreover, feeders experience upstream flow of power during high DG production causing

overvoltage at feeder ends. Traditional voltage regulators using transformer with mechanical

switches for taps are unable to cope up with such frequent fluctuations and get worn out due

to frequent tap changes.

Voltage compensation can be achieved through shunt and series compensation. Owing to the

nature of the distribution network in Europe in general, voltage control through shunt

compensation is not an efficient and cost-effective method. This is because of the long lengths

and low X/R ratio of the distribution feeders in Europe. A suitable solution is through series

compensation where a compensating voltage is injected in series with the grid voltage, such

that the load voltage remains a constant at all times. Examples of such devices are self-

commutated switch based FACTS devices and solid-state/hybrid switch based on-load tap

changing transformers (OLTC). The series compensator can be either centralized compensator

close to the distribution transformer or a feeder-specific compensator unit. The latter is a cost


2

effective solution in the case of non-uniform distribution of DG, loads and uneven feeder

lengths, which are often encountered in the European scenario.

The thesis describes the design of a partially rated, feeder specific series compensator that

uses an OLTC autotransformer with taps on the load side. The aim was to devise an OLTC

that has low steady state losses and changes taps without the occurrence of an arc,

guaranteeing long lifetime of the tap changer. An innovative design using a combination of

no-load switches and a hybrid switch to realize the taps has been proposed. The design saves

on cost, has reduced steady state losses and provides for arc-free tap changing. Positive and

negative compensation of the grid voltage can be achieved on specific hot-spot feeders that

have high DG power injection, to ensure optimal voltage level throughout the network. The

system has been customized for application in both MV and LV distribution networks.

Protection issues and control of the voltage compensator have been addressed providing a

holistic design for the future development of a prototype.

The thesis work was done in collaboration with GE Global Research, Munich.

Overview of chapters:

An overview of the content in the following chapters of the thesis is presented below:

In chapter 2, the voltage fluctuation issue in the grid is investigated. Detailed system level

study is carried out to understand the voltage fluctuation due to renewable energy integration

and heavy loads

In chapter 3, a study of the past research work in the area of voltage compensation is

presented. Shunt and series compensation methods are examined, which are the two methods

for mitigating voltage fluctuation in the grid. Potential compensation techniques such as

voltage regulators, FACTS devices and OLTC are described. The different methods are

compared on their applicability to the European distribution network and on the basis of

simplicity, ease of control, efficiency and cost. It is reasoned out that partially rated, feeder

specific on-load tap changing transformer is a suitable choice for providing voltage regulation

in the European scenario. The design of such device will be undertaken in this thesis work and

discussed in the further chapters.

In chapter 4, different topologies for the tap changing transformers are proposed and analyzed

for detailed operation. The key is to provide both positive and negative compensation to the

grid and compensate frequent voltage fluctuations. The topologies are compared and

theoretical estimations of the voltage and current ratings of the transformer and the switches

are done.


3

In chapter 5, the topologies proposed in chapter 4 are simulated in the PLECS simulation

environment to study their steady state operation. The analytically calculated transformer and

switch ratings are verified in simulation. Few topologies are then shortlisted based on their

operation, component ratings and cost. Different possible connection schemes for single phase

autotransformers to a three phase network are presented. Based on the voltage regulation

needs for the MV and LV network, an open-delta and Y connection of transformers are

chosen as most effective connection scheme respectively.

In chapter 6, an innovative design of a tap changer using no-load switches and a hybrid

switch has been proposed. The design is cost-effective, has low steady state losses and has no

arcing problems during tap change. A scientific elucidation of the choice of power electronic

switches for the hybrid switch and commutation scheme is presented. The design of an

overvoltage snubber to protect the switches during tap change is examined and analytical

derivation is presented for the sizing of the snubber and switches.

In chapter 7, the complete system with the no-load switches and hybrid switch is customized

for a three phase connection scheme for application in both MV and LV. Detailed component

level modeling and simulation of the system in PLECS is performed. Test-benches are run to

verify its transient and steady-state operation.

In chapter 8, research into aspects of fault scenarios and protection mechanism for the system

is performed. Detailed study of the failure mechanisms and different possible internal and

external faults of the system are investigated. The effects of component failure to the grid and

vice versa are examined to ensure that reliability of the grid and of the operation of OLTC is

not affected. Possible protection mechanisms are suggested to ensure interruption-free

operation of the compensator. A low level control mechanism is developed and presented.

In the final chapter 9, the important observations and conclusions from the design of the

compensator are presented and the scope for future work is examined.


4

Chapter 2: Voltage fluctuation in the distribution network

5

Chapter 2

Voltage fluctuation in the distribution network

In this chapter a study of the voltage fluctuation issue in the distribution network is carried out.

In the first part, the causes for voltage variations in the grid are probed and the effect of

distributed generation such as PV injection on the distribution network voltage is studied.

While in the past, undervoltage was the major issue for DSO due to heavy loads being

connected to the network, the introduction of large scale PV injection has led to overvoltage

becoming a frequent phenomenon in the grid [2-5,9,10]. In the second part of the chapter, a

mathematical analysis to quantify the load voltage as a function of load current and power

factor is performed. In the third section, the effects of undervoltage and overvoltage on

devices connected to the network are elucidated to highlight the seriousness of the problem.

2.1 Causes for voltage fluctuation

2.1.1 The scenario in the past

Voltage variation is a repetitive phenomenon in the distribution network. In the past, the

major source of this variation was on the account of time-varying loads from industrial and

domestic consumers. Conventionally power grids consider a downstream flow of power from

the distribution transformer to the load as shown in Fig. 2.1 [1]. This will cause a gradual fall

in voltage magnitude as we move along the line towards the load. The voltage at the

transformer is maintained above the nominal voltage to compensate for the voltage drop along

the line. Thus the maximum voltage in the line is seen at the transformer and it decreases

along the line and minimum is seen at the load end. In reality, the situation is much more

complex than what is shown in the figure, owing to the non-uniform load distribution along

the feeder.


6

Fig. 2.1 - Conventional downstream power flow with minimum voltage at the end of line

giving rise to undervoltage [1]

Generally the loading on the grid reaches a maximum during the morning and evening hours

and a minimum during the afternoon and night. If a fixed voltage is maintained at the

distribution transformer, it can be expected that the voltage at end of the line to be time-

varying, depending on the load. This is because the line has finite impedance and hence a

finite load dependent voltage drop occurs across the line. In this case the fluctuations are

typically undervoltage i.e. voltage is lower than its rated value, if it is assumed that the

majority are inductive loads. For the German grid which operates nominally at 230V rms,

undervoltage means that the load voltage is lower than 230V rms.

Fig.2.2 shows a lumped element representation of distribution side with the source being the

distribution transformer and the load drawing a lagging current at a certain power factor

pf [1]. This current causes a voltage drop ΔV along the line due to the effective impedance of

the line represented by (Rs+jXs). Load voltage V in hence lower than the supply voltage E. It

is important to realize that voltage difference due to drawing reactive current has a more

adverse effect than due to real component of current.

V s l s l s l s lR P X Q X P R Qj

V V

(Refer section 2.2)

This undervoltage depends mainly on the following factors –

1. Distance between distribution transformer and the load/ Impedance of the line

2. Load current & distribution of loads along feeder

3. Proximity to voltage regulating equipment

4. Power factor of the connected load

5. Reactive power sources present in the vicinity.

In order to manage the undervoltage, utility operators typically use transformers with on-load

or off-load tap changing feature using mechanical switches. MV to LV transformers typically

have off-load tap change feature while HV to MV transformers change taps on-load. The

voltage at the transformer is usually set above the rated voltage in order to compensate for the

line drops. The tap position is set based on the load and varied to ensure nominal load voltage.


7

Fig. 2.2 - Occurrence of undervoltage due to lagging current drawn by the load [14]

2.1.2 The scenario in the present and for the future

Compared to the past, the present scenario of the distribution network has seen a rapid

increase in the integration of renewables in the recent years and it is envisaged that this

number will only go up in the future. The best example is that of Photovoltaic (PV) panels

connected through inverters to the distribution network. In [2] it has been estimated that the

major cause for voltage fluctuation in low voltage grid is due to residential photovoltaic

systems which are in the power range of 1kVA to 50 kVA. High concentration of Distributed

Generation (DG) will lead to unpredictable variations in power production and line voltages,

owing to short-term and long term variations in weather [2-4]. The long-term variations can

be seasonal variations in wind and sunshine between summer and winter while short term

variations include phenomenon like wind speed variations within a single day or variation of

solar irradiance due a passing cloud.

In Fig. 2.3, the fluctuations in power produced in Germany over three days from a single

panel and an ensemble of 100 panels with a combined capacity of 243 kWp is shown [5]; it

can be inferred that these fluctuations can adversely affect power flows and voltages in the

Fig.2.3 - Fluctuations in power produced from a group of 100 PV panels distributed all over

Germany over a period of three days [5]


8

Fig. 2.4 – Present and future scenario with downstream and upstream power flow giving rise

to both undervoltage and overvoltage [1]

grid. The power produced from DG will lead to a reduction in the power drawn from the

mains. This means that the load voltage will be dependent on DG power output. More

importantly DG may result in reverse power flow i.e. power will flow from the load end of the

line in an upstream direction towards the distribution transformer during times of high power

output. This reverse flow of power will cause the voltage at the end of the line to be higher

than at the transformer as shown in Fig. 2.4. This situation is further worsened if the set

voltage at the transformer is by itself at a value higher than the nominal in order to

compensate for line drops. This results in an overvoltage, when the load voltage is greater

than rated value. It was listed earlier that five major factors are responsible for undervoltage;

the same factors affect overvoltage as well. Thus depending on the power production from

DG, the load voltage can vary over a wide range from values below to above the rated voltage.

A comparison of the two scenarios with and without DG is presented in Table 2.1.

Without DG With DG

Power flow From ‘generation’ end to ‘load’

end

From ‘load’ end to ‘generation’ end

and vice versa

Effect of

(Rs+jXs) of line Voltage drop along feeder Voltage gain and drop along feeder

Voltage

variation

Undervoltage only

(Load voltage < Nominal value)

Under and over voltage

(Load voltage >or< Nominal value)

Frequency of

fluctuations Less frequent Very frequent

Table 2.1 – Voltage variation in grid with and without presence of DG

If a rooftop PV system is considered, one extreme situation is when the PV system is

switched off or produces small amount of power, lower than the consumption of the home.

Then the house will draw significant power from the line resulting in undervoltage.

Alternatively, the PV production increases on a sunny afternoon and it can provide for most

of the house loads thereby reducing the current drawn from the line. It could also be that the


9

PV production exceeds the house loads and power is fed back to the grid. The other extreme

situation then occurs when the house load is minimal or zero and all the PV power produced

is fed back to the line, giving rise to an overvoltage.

Fig 2.5 shows a distribution network having large PV penetration where a number of feeders

of equal capacity but unequal length emanate from three distribution transformers. Two

situations are shown in Fig 2.5 and Fig 2.6 for 5% low level and 95% high level of PV

production respectively, assuming that the loading on the feeders is 5% of nominal load. It is

observed that during low PV injection, the taps of transformers at the head of the feeders are

set to 1pu, so that voltage at all points in the distribution network is close to 1pu. At high

levels of PV injection, overvoltage is observed along the feeder as we move away from the

transformer. The situation worsens at the end of long feeders where the overvoltage is beyond

0.05pu. This overvoltage will exist for hours together till either the load at the feeder increases

or the PV injection reduces. In this special case where uneven feeder lengths are encountered,

Fig 2.5 – Schematic of feeder system during low levels of PV injection [26]

Fig 2.6 – Schematic of feeder system during high levels of PV injection [26]


10

the overvoltage cannot be regulated by the main transformer as setting a very low voltage at

the transformer will result in undervoltage for consumer near the transformer and close to the

feeder head. Thus this case calls for a strategy to regulate voltage along each long feeder

instead of controlling using a fixed central transformer where all feeders are regulated in the

same fashion; without taking into account the reverse power flow or variable feeder length.

In the future it is expected that connection of high power loads like electric vehicle charging

will only increase and add to the demand on the distribution network. This coupled with the

occurrence of reverse power flow means that load voltages will fluctuate over a large range

thus becoming a formidable problem for both utility operators and consumers. It is hence vital

to look at the effects of such a scenario and discover possible solutions to avert it.

2.2 Magnitude of voltage variation at load end

In this section, an analysis of the varying voltage at the load end is made as a function of load

power assuming a constant sending end voltage. It assumed that the load draws a power

Sload=Pl+jQl at a voltage V when the sending end voltage at the start of the feeder is E, as

shown in Fig.2.7. The line is assumed to have an impedance of Rs+jXs and carries the load

current Il.

The difference in voltage at the start and end of the feeder is given by:

(Eq2.1)

The relationship between the load powers and its voltage and current is: *

load l(I ) l lS V P jQ (Eq2.2)

lI l lP jQ

V

(Eq2.3)

Substituting (Eq2.2) & (Eq2.3) in (Eq2.1),

s lV E V Z I ( )( )l ls s

P jQR jX

V

Fig. 2.7 - Evaluation of load voltage as a function of load power & line impedance [23]


11


V V

(Eq2.4)

The voltage drop can be simplified and written as made up of two components, one in phase

with the load voltage ΔVR and another in quadrature to it ΔVX

V V V

V

V

R X

s l s lR

s l s lX

j

R P X Q

V

X P R Q

V

From the above expression, the voltage drop is dependent on the feeder impedance and

magnitude and phase of the load current. Thus the power factor of the load is an important

factor affecting the load voltage. The load current in turn can be such that active and reactive

power are drawn from the grid i.e. P and Q are positive. The above equations are equally

valid when real and reactive power is fed back to the grid through DG, where P and Q are

hence negative.

2.3 Effect of voltage fluctuation on utilities and load

Voltage fluctuations (both under and over) are undesirable for loads and affect their operation

and lifetime. There are two types of voltage issues namely short term and long term voltage

fluctuation. The short term voltage problem is usually caused by voltage sag or swells which

is defined as a drop/rise in voltage over any time frame between one half-cycle and sixty

seconds [22]. It is generally caused by a fault in the power system [22]. In contrast, long term

voltage fluctuations are those that could last for minutes or even hours. Overvoltage and

undervoltage are considered as a long term voltage problems and they lead to a more serious

problem in power system operation compared to voltage sags/swells.

When the voltage exceeds the tolerance level of ±10% at customer utilization point, it affects

the operation of the loads connected to it [6, 57]. Extended overvoltage can decrease the

lifetime of most equipment. In an extreme case, breakdown of the insulation can occur and

permanently damage appliances when voltage reaches very high values. On the other hand,

undervoltage can cause malfunctioning of devices in the form of dimming of lights referred to

as ‘brownouts’, heating up of motors and inability to power some devices like air conditioners.

Secondly, the power converters that connect the DG to the grid will de-synchronize from the

grid during periods of high overvoltage effectively bringing the power production to zero.

This can have negative equity effects on customers who own DG that are connected close to

the end of the feeders compared to those connected at the start [6]. In a worst case scenario,

extreme voltage fluctuations can cause all the DG to de-synchronize irrespective of their

location on the feeder; the situation getting further deteriorate during undervoltage even


12

leading to a blackout. Thirdly, the power fluctuations due to DG cause varying currents along

the cables causing excessive thermal expansion and contraction of the copper which affects

their lifespan.

Finally and most importantly, frequent voltage fluctuations will cause the conventional

voltage regulators like on-load tap changing transformers to change taps very frequently. This

will reduce their lifetime and lead to increased maintenance requirements. This calls for new

design of voltage regulators that can work in a future scenario of distribution grid with high

levels of DG penetration. An alternate strategy is to investigate methods to modify existing

voltage regulators to work in this new scenario [24].

2.4 Other voltage issues

Besides voltage fluctuations, other important repercussions of an increased penetration of DG

exist and are of importance to utility operators and consumers [6]. Some of these are:

1. Non-sinusoidal load voltage waveform at the point of common coupling due injection of

harmonics from the grid-connected power converters of DG systems.

2. Unbalanced connection of loads and DG in the distribution network across the three

phases can lead to unbalanced three phase voltages at the load end.

3. Frequency regulation becomes a complicated issue as centralized generation has to follow

the de-centralized generation to ensure a constant 50/60 Hz grid frequency

4. Increased fault levels & subsequent protection mechanisms


Voltage fluctuation in the distribution network is a major problem facing DSO. The problem

has been aggravated by the large scale penetration of DG. Undervoltage and overvoltage are

detrimental to devices connected to the grid and there are strict regulations that restrict the

maximum deviation that can occur from the nominal voltage. Conventional tap changers

cannot cope up with the frequent voltage fluctuations due to their use of mechanical switches.

New design of voltage regulators is required to ensure constant load voltage for consumers in

the distributed grid.

Chapter 3: Voltage compensation methodologies

13

Chapter 3

Voltage compensation methodologies

Voltage compensation is the method of ensuring that the grid voltage remains within

permissible limits at all times. This must take into the account the variations in customer loads

and de-centralized generation and effectively manage the centralized generation and voltage

regulation mechanism in the utility. This can be achieved by a number of methods of which a

few are briefly discussed in the first section of the chapter. In the next section, a study of

nature of the European distribution network is made to see which compensation method will

be most effective. It is seen for Europe that series compensation is a more effective solution

than shunt compensation in general. In the final section, series compensation using tap

changers and power electronic based FACTS devices are compared to point out the inherent

advantages of the former.

The most widely used methods for voltage compensation in the distribution network can be

categorized into two - shunt and series compensation.

1- Shunt compensation where a lagging/leading current is injected into the grid to control

the voltage.

2- Series compensation where a voltage is injected in series to the existing grid voltage or

a reactive element in connected in series to line to modify the line impedance.

Example of shunt compensation is capacitor/inductor banks or FACTS devices like Thyristor-

controlled reactor (TCR) and Thyristor-switched capacitor (TSC) [14]. Modern PV inverters

are equipped with the ability to inject/absorb reactive power and hence control the voltage at

point of injection [6, 9, 18]. Series compensator can be realized using tap changing

transformer or FACTS devices like Unity power flow controller (UPFC). The advantages and

disadvantages of the two methods and challenges in their implementation for future

distribution networks are discussed in Table 3.1


14

Advantage Disadvantage Challenge

Shunt

Low investment

costs.

Ineffective for networks

with low X/R ratio.

Increases the network

losses.

Reduces the capacity and

efficiency of DG.

Feasible methods of

controlling the reactive

power source(s).

DSO might not have full

control on voltage regulation

(E.g. DG providing reactive

power).

Series

Effective for all

types of networks

(e.g. also LV cable

networks).

Does not increase

network losses.

DSO has full

control over

voltage regulation.

Mechanical series

compensation devices

(e.g. OLTC, Voltage

Regulator) cannot cope

with the high variability

of DGs (e.g. PV).

Higher investment costs.

Advanced control methods,

also in combination with

reactive power

compensation.

New flexible series

compensation devices to

cope with high variability of

renewable DGs.

Table 3.1 - Analysis of the critical aspects of series and shunt compensation

A number of different methods are present to realize these two basic compensation techniques.

These methods are analyzed in the following section.

3.1 On load and off circuit tap changing transformers

On-load (OLTC) and off circuit tap changing transformers provide compensation by using a

transformer with multiple taps [7,32,36]. They are widespread in the electricity networks and

are likely to remain in service for many years in the future. They come in the category of

series compensating devices that inject a voltage by inclusion of additional taps. The taps can

be present either on the high voltage or the low voltage side of the transformer. When the tap

is moved from one position to another, the turn ratio of the transformer is changed as shown

in Fig3.1. For Fig3.1(a) when there is undervoltage, the tap is moved to a higher position and

vice versa during an overvoltage. The OLTC have the advantage over off-circuit tap changers

that the taps can be changed without the interruption of the load.

The OLTC can also be used with autotransformers with taps on the secondary side providing

voltage regulation but no electrical isolation [9]. Such a technique has been implemented in

conventional voltage regulators present in the MV network as shown in Fig.3.2, which

generally provide for upto ±10% compensation. The switch position S determines whether the

transformer operates in a step-up or step-down mode [10]. Autotransformer based OLTC are

more economical, efficient and compact compared to a conventional tap changer shown in

Fig 3.1 when the primary and secondary voltages are close in ratio (~3:1).


15

Fig.3.1 – Tap changing transformer with taps on primary (a) and secondary (b) side [7]

Each tap usually represents a change of 0.625% to 1.25% of voltage on the load side and can

provide up to a total of 10% compensation [10,33,58]. Traditional control strategies for

voltage regulators are based on the assumption that power flow occurs in a unidirectional

manner and occurs from HV generation end to LV load end. The Line drop Compensation

(LDC) method works on this assumption and it tries to set a higher voltage at the distribution

transformer to compensate for the voltage drop along the distribution line [52]. It uses a

constant voltage set point and does not take into account the presence of active feeders

resulting from DG feeding power. Thus the traditional LDC method will fail when the partial

power of the loads comes from the overhead PV panels and during times of reserve power

flow during excessive PV generation. Thus a dynamic voltage set point approach is required

for their operation so that if the voltage along the line increases due to DG, the voltage set

point will be reduced at the transformer to prevent over-voltage.

The tap changing was traditionally done with the use of mechanical switches, with each tap

having a switch [32]. These switches undergo wear and tear due to the arcing that occurs

when taps are changed on load in OLTC, similar to the arc produced in a circuit breaker. Thus

frequent tap changes are not suitable for such a system as it reduces the lifetime of the

switches. This is the reason that these switches are not being favored for current and future

use as voltage fluctuations and requirement for tap changes become more frequent with

introduction of DG.

The solution is through the use of power electronic switches – the advantage being that they

can be switched a very high number of times with no effect on its lifetime and operation

[8,32]. They have faster response times and practically need no maintenance compared to the

mechanical switches. However power electronic switches have a serious disadvantage that the

switches have higher losses compared to the mechanical switches; losses stemming from the

conduction and switching losses occurring in the semiconductor. Further they have low

overload and short-circuit capacity compared to mechanical switches. This has been the

reason why researchers have begun to see the potential in combining the advantages of both

the electronic and mechanical switch by using hybrid switches [37-40]. Such a hybrid switch

will use electronic switch during a tap change and a mechanical switch during steady state.


16

Fig.3.2 - OLTC auto transformer providing series compensation with taps on load side that

can both step-up and step-down load voltage [10]

3.2 Reactive power compensation through inductors/capacitors and

synchronous condensers

Reactive power compensation involves the use of capacitor and inductor banks to provide

shunt compensation. Feeding reactive power through capacitors increases the grid voltage at

that point while the voltage decreases when reactive power is drawn through inductors. The

effectiveness of this method depends largely on the R/X ratio of the line - if ratio is high the

method is less effective and results in high losses [12, 14]. With the addition of a capacitor in

series with the transmission line, the transfer impedance of the line is hence reduced. A

capacitor can also be connected in shunt to the line and it draws a leading current. This

provision of reactive power reduces the reactive current drawn from the main lines and

improves the voltage profile along the line. Alternatively an inductor can also be connected in

shunt which will draw reactive power and reduce the voltage along the line, a method quite

effective during overvoltage. The reactive power inductor/capacitor banks are switched in and

out of the system through a mechanical switch.

This method has the additional advantage that the capacitor bank can provide for inductor

power requirements of local loads like induction machines, effectively reducing the current

flows in the lines, increasing grid capacity and reducing losses. The principal disadvantage is

the large sized capacitors and inductors that are required for this technique. The effectiveness

depends on the number of capacitor/inductors units present; a smooth control requires a large

number of small capacitor/inductor units while a cost effective solution requires a small

number of large capacitors/inductors. In some cases this method becomes impractical when

the amount of reactive power required to be injected can be too large.

Synchronous condensers are a proven technique that have been in use for several years in

which a synchronous machine is used to provide or consume reactive power [14] and improve


17

power factor. The machine is simply connected to the power system and the excitation is

varied after synchronization to operate the machine in leading or lagging pf mode. It can

provide continuous reactive power control through the use of a controlled excitor circuit.

However they are rarely used today because of the high cost of machine and protective

equipment, sluggish operation, high losses compared to modern FACTS devices and the fact

that they contribute to the short-circuit current of the power system during faults [14].

3.3 FACTS and power electronic based voltage compensators

FACTS (Flexible AC transmission system) devices are an advanced version of reactive power

compensation which provides high level of flexibility in operation through the use of power

electronic switches [13]. Example of this includes electronically switched reactive banks or

self-commutated switch based converters [15]. Such a technique was envisaged in [11] as

general method for improving power quality by the use of power electronics.

Both series and shunt compensation can be achieved using FACTS devices. In series

compensation, the FACTS device either acts as a controllable voltage source or as a variable

series reactive element that modifies the equivalent impedance of the line. In shunt

compensation, the FACTS device acts as a variable current source and injects a

leading/lagging current to the grid. Both techniques are explained in the Fig3.4 [14]. The load

draws both real and reactive power from the grid and as pointed out earlier, it is the reactive

component of current that has the most adverse effect on voltage fluctuation. The control

strategy of the FACTS device tries to supply this reactive power to the load so that the current

drawn from the grid is reduced.

Fig 3.3 - Evolution of FACTS technology over the years [15]


18

Fig. 3.4 – Shunt (left) & series (right) compensation using FACTS devices [14]

This is achieved by supplying the reactive current IQ in shunt to the load or by adding a

leading compensating voltage VCOMP in series to the grid voltage so that the grid only supplies

the real current IR. After compensation it can be observed that the load continues to draw real

current from the grid, but the voltage drop is much lower and the load voltage is closer to that

of the supply.

Different types of FACTS devices exist, the simplest being the Static VAr Compensators

(SVCs) that can be grouped mainly into the Thyristor-controlled reactor (TCR) and the

Thyristor-switched capacitor (TSC) for shunt configuration and Thyristor-controlled series

capacitor (TCSC) and Thyristor-controlled series reactor (TCSR) for series compensation [14,

16]. Here a fixed inductor and capacitor bank can provide a range of VAr compensation

respectively, by the use of a thyristor switch operating at different firing angles. This provides

more flexibility when compared to fixed switching of banks as discussed in the previous

section but still poses a problem with regard to need for bulky passive elements for their

operation.

In recent years, the evolution of self-commutated devices has opened up a new chapter in

FACTS. Devices like IGBT and MOSFETs have been used in the design of compensators

such as static synchronous compensators (STATCOMs), unified power flow controllers

(UPFCs) and dynamic voltage restorers (DVRs) [13, 17]. The self-commutated VAr

compensators have the ability to both generate and absorb reactive power without requiring

large banks of capacitors or reactors. They are compact in size and can provide very fast

response to grid fluctuations. They have no problems of inrush current that affects the

operation of compensators with passive elements. They can provide both series and shunt

compensation and can be built in multi-level configuration. They can be controlled using

different methods like Pulse Width Modulation (PWM) or Space Vector Modulation.

The STATCOM is based on a solid-state voltage source implemented through an inverter and

connected in shunt to the line [17, 19]. The Dynamic Voltage Restorer (DVR) and the Unified

Power Flow Controller (UPFC) generates a series compensation voltage with an inverter and

feed it in series to the grid voltage through a transformer [14, 20, 21]. The voltage can be fed


19

Fig. 3.5 - Topology of a Unified Power Flow Controller (UPFC) [14]

in-phase or with a phase shift providing for reactive power compensation as well. UPFC has

active converters for both the rectifier and inverter and can control the flow of both real and

reactive power at both the AC terminals of the device. The UPFC power circuit topology [14]

is shown in Fig.3.5 where the rectifier input is connected in shunt to the line while the

inverter output is connected in series providing a compensating voltage that is controllable

both in phase and magnitude.

3.4 Voltage control through grid connected PV inverters

The grid connected inverters are an integral part of a PV system. They convert the DC power

of the panels into AC. The PV inverters can also be controlled to provide voltage regulation.

This can be done by two methods namely reactive power control and power curtailment [6, 9,

18]. For reactive power control, the inverter operates in what is referred to as a ‘voltage

regulating mode’. This means that the inverters besides feeding real power to the grid will

feed reactive power during undervoltage and absorb reactive power during overvoltage. This

can be achieved by overrating the inverters above the nominal power of the panels to handle

the additional reactive power and by implementing proper control. This method has the

disadvantage of higher cost of the inverters, increasing the losses in the inverter, causing

additional current flow and losses in the line and increasing the probability of islanding as

described in [6,9].

In case of power curtailment, the output of the PV inverters is limited so as to ensure that

overvoltage is within the set limits. This is achieved by moving the operating point of the

inverter away from the maximum power point towards the open circuit voltage of the panel,


20

thereby reducing the real power fed to the grid. This technique has the drawback of wasting

available PV power and can have negative customer equity impacts [6, 9].

The strategies for PV inverters can be equally applied to other converters that connect DG to

the grid. The basic idea is to provide a voltage control from the source of the fluctuation itself,

i.e. at the point from where the DG feeds power to the grid.

3.5 Which method suits the European distribution network?

In European distribution network, PV penetration is a major cause for power fluctuation. Low

voltage feeders in Europe can reach four to five times the length of those in North America

and are predominantly underground cable networks with a high R/X ratio as shown in Fig 3.6

[25]. In German LV grids R/X values of 2 are typical for overhead ines and R/X values of 2.5

are typical for cable networks [59]. This means that the voltage variation is much higher in the

European grid compared to the North American grid. Secondly and more importantly reactive

power compensation using leading/lagging current injection (shunt compensation) is less

effective due to low X/R ratio in the European system and will result in large current through

feeders and increased losses [12, 14].

Fig. 3.6 - Distribution system layout in North America & Europe – a comparison [25]


21

The impact of voltage compensation through shunt reactive power compensation, series

compensation using a full transformer and partially rated autotransformer are analyzed for

schematic of Fig.2.5 where radial feeders emanate from multiple sending end distribution

transformers.

a. Shunt reactive power compensation

In the first case, voltage compensation is done through shunt compensation where a

leading/lagging reactive current is injected to the grid to regulate the voltage. The aim is to

ensure 1pu voltage throughout the network and prevent the red hot spots as seen in Fig 2.5.

The method is certainly effective as seen from Fig 3.7 where the red hot spots have

disappeared and a green area corresponding to nominal 1pu voltage can be seen throughout

the region. However it is neither a cost effective nor efficient solution to the problem. Since

the feeders in the European networks are long and have high R/X ratio, large reactive currents

have to be injected to the network to ensure constant grid voltage of 1pu. This results in

feeders requiring thick copper conductors to handle this additional reactive current injection.

There will be added costs for increasing the current capacity of the feeders and increased

losses along the line. The cost involved in setting up the power electronic controlled inductor

and capacitor banks have to be accounted for as well. Thus it can be seen that this is not the

best solution for the voltage problem.

b. Tap changing on centralized full rated transformer

The second solution is through the implementation of voltage control though the main

distribution transformer present at the head of the feeder. This is shown in Fig 3.8 and can

provide compensation for different feeders emanating from the distribution transformer.

Unfortunately the drawback is that all the feeders are provided with the same compensation

Fig.3.7 - Voltage regulation using reactive power compensation [26]


22

Fig.3.8 - Voltage regulation using centralized full transformer [26]

irrespective of the load in each feeder. Because of the long lengths of line in European

distribution network, if the taps on the central transformer are adjusted to maintain nominal

voltage at the feeder ends, the voltage set point will have to be low at around 0.95pu so as to

ensure that overvoltage during high PV injection is within limits. This will in turn result in

undervoltage for customer close to the feeder head. Thus while solving the overvoltage

problem for customers at the feeder end; it recreates undervoltage for customers at the feeder

head. Thus a comprehensive solution for all customers cannot be obtained by this method. It

must be mentioned that this is a very cost effective method as it uses existing transformer

infrastructure to provide a solution.

c. Tap changing on feeder specific partially rated series compensator

Fig.3.9- Voltage regulation using feeder specific decentralized series compensator that is

partially rated for the compensating power [26]


23

The last and most comprehensive solution is series compensation through the use of a

partially rated transformer or FACTS device that would do voltage injection on selective

feeders to treat the areas of red hot spots. Such a scenario is shown in Fig 3.9 where a series

compensation device rated for the compensating power of 10% of the total power will be used

to selectively treat overvoltage along the long feeder. Such a device will be compact and cost

effective as it is fractionally rated and does not need major changes in the network

infrastructure as was the case with reactive power compensation. Further this method does not

cause voltage variations in other parts of the network as was observed with a full transformer.

Thus the units can be independently connected to the each hot spot feeder that come from the

distribution transformer and has high levels of PV power injection- providing flexibility and

accurate voltage across the system.

Implementation of such a device can be as shown in Fig 3.10 and categorized into three types

listed below. In reality, the connection can be a combination of two or all three types.

1. Selective compensation on MV feeders which connect to regions with high PV injection

and voltage variation

2. Selective compensation on LV feeders with high PV injection

3. Selective compensation on LV feeders and/or busbar when feeders have unequal loading

Fig.3.10 – Selective compensation using partially rated series voltage compensator

1


24

Fig.3.10 – Selective compensation using partially rated series voltage compensator

A comparative analysis of the voltage at the head and end of the feeder for different voltage

compensation strategies discussed above is presented in Fig 3.11a,b,c, for the case of variable

PV injection [26]. Fig 3.11a gives the PV generation profile as a function of time. Fig 3.11b

and Fig 3.11c shows the voltage profile at the feeder head and end under the following

conditions:

Fig.3.11a – PV generation profile [pu] as a function of time [hr] [26]

11:45 11:50 11:55 12:00 12:05 12:10 12:150

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PV Generation Profile

Time [hrs]

Act

ive

Po

we

r [p

u]

2

3


25

1. No PV power injection

PV power injection with:

2. No voltage compensation

3. Full compensation - voltage compensation provided through reactive power

compensation or through PV inverter ensuring the load voltage is maintained at 1pu

4. Tap change on the main distribution transformer

5. Tap change on the feeder specific transformer

Fig.3.11b – Voltage profile [V] at the head of feeder as a function of time [hr] [26]

Fig.3.11c – Voltage profile [V] at the end of feeder as a function of time [hr] [26]


26

It can be observed that tap changing on feeder and reactive power compensation provides the

best results for ensuring optimum voltage levels at both feeder head and end. This specially

applies to networks having uneven feeder length and non-uniform distribution of DG when

tap change on distribution transformer leads to undervoltage at feeder head. Comparing these

voltage profile and cost effectiveness of different methods, it can be concluded that series

compensation on selective feeders using series compensation is a comprehensive solution to

tackle the voltage fluctuation problem in the European distribution network.

3.6 Comparison of voltage compensation through self-commutated switch

based series FACTS devices and fractionally rated tap changers

Fractionally rated series compensators can be realized using tap changing transformers or the

self-commutated switch based FACTS devices like UPFCs. The FACTS devices are compact,

have quick response and are flexible in their operation. Tap changing transformers are less

flexible and much larger in size but have simple control methodology and are more cost

effective. A comparison of the two is summarized below in Table 3.2.

Self-commutated switch

based FACTS devices On Load Tap Changers

Flexibility

Provides high level of flexibility in

operation and the compensation can

be smooth

OLTC which operates with discrete

taps has lesser flexibility

The compensation voltage/current

can be fed in phase or with a finite

phase shift with the grid voltage;

effectively it can provide reactive

power support as well

The change of taps can affect only the

magnitude of the load voltage and can’t

provide a control on the phase of the

injected voltage.

Control

Require complicated closed loop

control algorithms for their

operations

Have simpler control strategies

Speed of

operation

Provides a much faster speed of

operation to grid changes

Comparatively slower speed of

response even if the taps are made of

electronic switches

Harmonic

By using suitable control, the power

electronic converter can be designed

to do harmonic compensation as

well.

No such feature exists


27

Need for filter to eliminate the

harmonics generated by the converter

adding to cost and space

Does not require filters

Cost High cost depending on nature of

power electronic switches and filter

The cost depends on the nature of

switches and transformer and ranges

from moderate to high

Losses

The steady state losses are high due

to the use of IGBTs and MOSFETs

The use of hybrid electro-mechanical

switches & purely-mechanical switches

gives very low losses in steady state

The switching losses are very high as

the IGBTs or MOSFETs are

switched in the kHz range; to easily

remove the harmonics with smaller

filters

The switching losses are minimal as

the commutation of electronic switches

occurs at line frequency and the

mechanical switches change state only

during tap changes and never in steady

state.

Table 3.2 – Comparison of series compensation using OLTC & full power electronic power

based FACTS devices

Both methods described above have their own advantages and disadvantages and it is not

really possible to propose either of them as a universal solution for all applications. In this

thesis work, the aim will be to design a cost effective and simple on-load tap changing

transformer that can provide selective feeder specific compensation on hot spot feeders

having high PV injection. With the advent of taps made of hybrid switches with low steady

state losses they pose a worthy competitor to FACTS devices. The tap changer will have the

ability to provide both positive and negative compensation upto a magnitude of 10%. By the

use of autotransformer, throughput power will be nominal power, but only 10% of this will

need to be transformed; making it compact and cost effective. The steps in the realization of

such a design will be looked at in the following chapters.


Shunt and series compensation are the two chief methods for compensating voltage

fluctuations in the distribution network. Various voltage compensation methods include on-

load tap changing transformers, reactive power compensation and FACTS devices. For the

European scenario in general, series compensation is more effective than shunt compensation,

owing to the long lengths of feeders and the low X/R ratio of the lines.


28

In many cases, the use of a partially rated, feeder specific series compensator is the most

efficient and cost effective solution for Europe. This applies specially to distribution networks

that have uneven feeder lengths and non-uniform distribution of loads and DG. The series

compensators can be realized using full power electronic based FACTS devices or OLTC

autotransformers. Lower steady losses and cost make OLTC an excellent candidate over

FACTS devices. The thesis will be dedicated to realize a novel design of such an OLTC for

grid voltage regulation in the European distribution network.

Chapter 4: OLTC topologies – Design and sizing of components

29

Chapter 4

OLTC topologies – Design and sizing of components

From the conclusions of the previous chapters it is seen that a tap changing transformer is a competent strategy to solve the problem of voltage fluctuation. Tap-changers are categorized into two main groups [31, 32] – ‘Off-circuit or no-load tap-changers’ and ‘On-load or under-load tap-changers’. Off circuit tap changers are those that can change their tap position only when they are disconnected from the load and there is no current through the taps at moment of tap change. On the other hand on-load tap changers (OLTC) can change their taps even when they are supplying the load. Thus the integrity of uninterrupted power supply is maintained. This makes on-load tap changer more attractive for implementation in the grid but at the same time they are more expensive and complex in their construction. Operation of an OLTC is shown in Fig 4.1a. When switch S1 is closed , the load voltage is the voltage at Tap1. When a tap change is required from S1 to S2, ideally tap S1 should open and tap S2 should close at the same time. This is to ensure that there is no short circuit and load current is not interrupted [30, 32]. In reality, implementing such an ideal tap change is practically impossible.

Fig 4.1a – Tap changing in on-load tap changer


30

So what is normally done in practice is:

1. Break and make – where S1 opens first and then S2 closes. This will result in a momentary loss of load which is unacceptable as shown in Fig 4.1b. The situation is similar to what happens in a circuit breaker where an arc is formed when the current is interrupted and large overvoltage will be observed due to the inductivity of the load and line. Each tap has to be designed like a circuit breaker to interrupt large currents at high voltages, which is impractical and expensive. Thus this is not an option.

2. Make and break - where S2 closes first and then S1 opens. A momentary short circuit

will be observed between the taps. A mechanism to prevent this has to be identified. For example, an impedance inserted between the taps to reduce the short circuit current. Usually an inductor or resistor is used to limit the short circuit current as shown in Fig 4.1c. No load interruption will be caused, hence making this the suitable strategy for tap changing.

Conventional on-load tap changer systems were implemented on the sub-transmission transformer and in voltage regulators. They have mechanical switches that are moved with the help of a motor for tap changing. These conventional tap changers are designed on the strong assumption of unidirectional power flows from MV to LV networks and from the head of the distribution feeders to its ends. With bidirectional power flow and frequent power fluctuations in distribution grid due to high levels of solar penetration, the mechanical tap changers have yielded poor performance and exhibited reduced lifetime due to an increased arcing phenomenon resulting from frequent tap changes.

Fig 4.1b – ‘Break and make’ mechanism resulting in loss of load [30, 33]

Fig 4.1c – ‘Make and break’ mechanism using an impedance during tap change [30, 33]


31

Hence the requirement for the future is to build tap changers that:

1. Provide feeder specific positive and negative compensation 2. Are low cost and have low losses 3. Have long lifetimes under conditions of frequent tap changes with minimum

maintenance requirements 4. Withstand severe grid faults and not compromise on the reliability of the grid

OLTC can be built using conventional two winding transformers or using auto transformers in which the primary and secondary windings are electrically and magnetically connected to each other. The autotransformers have two windings namely the ‘shunt’ and ‘series’ winding - the shunt winding is connected to the source side and the series winding provides the compensation voltage. Analysis of conventional two winding transformers and autotransformers will be discussed in the first section. In the second section various tap changer topologies are proposed using both types of transformers. Subsequently for each topology, the ratings of transformer windings and switches are analyzed analytically and through simulation using PLECS® software [56]. In the final section, specific topologies are shortlisted based on cost, number and ratings of components to build a more comprehensive design.

4.1 Conventional two winding transformer Fig 4.2a shows an ideal two winding transformer where the primary high voltage winding HV and the secondary low voltage winding LV are magnetically coupled to each other through the iron core [27-29]. The turn ratio of the transformer is a=N1/N2. Currents and voltages of the primary and secondary are given by I1, I2, V1, and V2 respectively. It can be shown for an ideal transformer with no leakage impedance and magnetizing branch that

2 2 1

1 1 2

V NV

IN I

= =

(Eq 4.1)

If taps were present on the secondary side and tap position is given by x, then N2 can be replaced by x in Eq 4.1 to give Eq 4.2:

1 22 1

1 1

x I xN NV IV

= =

(Eq 4.2)

Thus by varying the tap position, it can be observed that a variable voltage at secondary can be obtained by using a fixed voltage at the primary side. It should be noted that the current dawn from the source I1 is the same as the current flowing through the primary winding I1W

and same applies for the secondary side for current I2 = I2W. This property is an important distinction between conventional transformers and autotransformer, as will be seen in the next section.

http://www.plexim.com/plecs


32

Fig 4.2a – Ideal two winding transformer with turn ratio a=N1/N2

4.2 Autotransformer

A more compact and cost effective solution to offer a variable secondary voltage is through the use of an autotransformer [27-30,32,33]. Fig 4.2b shows the schematic of an autotransformer operating in two modes of operation namely the buck mode and boost mode of operation. In this the case the two windings of a conventional transformer HV and LV as described earlier, are electrically connected to each other in a series fashion as shown. HV winding is considered as the primary/shunt winding and the LV is considered as the secondary/series winding. By varying the connection of the source and the load across the two windings, the two different modes of operation can be obtained – buck mode and boost mode. It is important to realise that for autotransformer the power at the terminals of the transformers S(Through put) is much higher than the power transformed S(Transformed) through the windings by magnetic action. This is because of the presence of an electrical connection between the primary and secondary because of which majority of the power is directly transmitted. The capacity multiplication factor Fc is defined as the ratio of power transmitted through the terminals of the autotransformer to power the power that is magnetically transformed through the windings and transformer core. Fc is a function of the ratio of the output voltage to the input voltage of the autotransformer given by r [29].

𝐹𝑐 = 𝑟𝑟−1

(Boost mode) (Eq 4.3a)

𝐹𝑐 = 𝑟1−𝑟

(Buck mode) (Eq 4.3b)

𝑟 = 𝑉2𝑉1

(Buck & boost mode) (Eq 4.4)


33

AC

I1

I2

I1 -I2

I2

V2

V1

LV(N2)

HV(N1)

It is assumed that the HV primary winding is rated for 1pu voltage and the LV secondary winding is rated for 0.1pu voltage. The load is rated for drawing a current of 1pu. For such a scenario the value of Fc would be approximately 10! This means that ten times more power can be transferred across the terminals of a conventional transformer if it were operated as an autotransformer.

Fig 4.2b – Autotransformer is boost (left) and buck (right) operation

4.2.1 Boost autotransformer: Input source is connected across the HV winding Output load is connected across the series connection of HV and LV winding Output voltage is 10% higher than input voltage

1crF

r=

− (From Eq 4.3a)

2 2 1

1 1

1.1V N NrV N

+= = = (From Eq 4.4)

V1=1pu; V2=1.1pu I2=1pu Sin=V1*I1= V2*I2 = Sout = 1.1pu (Conservation of energy) So, I1=1.1pu I1W= I1 - I2 = 0.1pu I2W= I2 =1pu Swinding1= (V2-V1)*I2=0.1*1=0.1pu Swinding2= V1*(I1-I2) =1*0.1=0.1pu S (Through put) = Sin = Sout = 1.1pu S (Transformed) = Swinding1 = Swinding2 = 0.1pu

𝐹𝑐 =S(Through put)

S (Transformed)=

𝑟𝑟 − 1

= 11

AC

I1

I2

I2 -I1 V2

V1

I1LV(N2)

HV(N1)


34

4.2.2 Buck autotransformer: Input source is connected across the series connection of HV and LV winding Output load is connected across the HV winding Output voltage is 10% lower than input voltage

1crF

r=

− (From Eq 4.3b)

2 1

1 2 1

0.9V NrV N N

= = =+

(From Eq 4.4)

V1=1pu; V2=0.9pu I2=1pu Sin=V1*I1= V2*I2 = Sout = 1.1pu (Conservation of energy) So, I1=0.9pu I1W= I2 – I1 = 0.1pu I2W= I1=0.9pu Swinding1= (V1-V2)*I1=0.1*1=0.1pu Swinding2= V2*(I2-I1) =1*0.1=0.1pu S (Through put) = Sin = Sout = 0.9pu S (Transformed) = Swinding1 = Swinding2 = 0.1pu

𝐹𝑐 =S(Through put)S(Transformed)

=𝑟

1 − 𝑟= 9

The advantage of autotransformer can thus be summarized as:

1. Majority of power being transmitted from source to load is transmitted electrically and only a fraction of the power is magnetically transferred through windings.

2. Part of the primary winding is used as secondary winding as well. So separate secondary winding is not required resulting in copper savings.

3. Current through the HV winding which is common to both the input and output is equal to the difference between input and output currents. This in turn reduces the current rating of the winding used, leading to further copper savings. This is an important distinction when compared to a conventional two winding transformer.

4. The effective copper used in an autotransformer compared to a conventional two winding transformer is given by Cr, where Cr = (Copper used in autotransformer / Copper used in two winding transformer) [27, 28].

For boost mode, 𝐶𝑟 = |1 − 𝑁1+𝑁2𝑁1

| (Eq 4.5a)

For buck mode, 𝐶𝑟 = 1 − 𝑁1𝑁1+𝑁2

(Eq 4.5b)


35

So it can be seen that for small values of N2, the savings are maximum. For small compensation for upto 10%, N2 = 0.1 N1 and Cr= 10%. So the autotransformer will require only 10% of the copper as required by a full transformer for same S(throughput)!

4.3 OLTC Topologies Through the use of conventional two winding transformers and autotransformers different configurations of OLTC can be achieved. These are presented in this section and analyzed. The voltage and current ratings of the switches and transformer are evaluated for steady state operation and the values are verified using a simulation package. This will form the basis for shortlisting topologies that will eventually be used for the final design of the tap changer. The OLTC models are made for a single phase system. The source side of the network is modeled using an AC source with a constant voltage magnitude of 1pu. The input of the OLTC is connected to this AC source and its output is connected to the load through a finite impedance line/cable. It is required that voltage at load is constant. This is achieved by the OLTC by feeding a series compensating voltage of upto 0.1pu that compensates for the voltage drop/gain along the line. When power is drawn by the load, there is voltage drop along the line and OLTC provides a voltage boost to compensate for this; vice versa happens when the load feeds power to the grid and the line current is negative. The assumptions made during the design of topologies are:

1. 1pu is set as the rated load current and rated source voltage. Thus the rated load power will be 1pu as well. With 10% positive compensation the load voltage will hence be 1.1pu and load power also 1.1pu.

2. Tap switches can block bidirectional voltages and conduct bidirectional currents 3. The OLTC device is rated to provide for upto ±10% voltage compensation. 4. The positive or negative compensation can individually be done in N steps each. 5. For simplicity it is assumed that each tap provides 2% compensation. For providing

full 10% positive compensation, a total of (10% / 2%)=5 taps will be required (N=5) I. Topology 1 and 1a Topology 1 and 1a uses a conventional two winding transformer of 1.1pu power rating with the taps positioned either on the primary or the secondary windings as shown in the figure. The turn ratio is 10:11 and 11:10 for topology 1 and 1a respectively. The topology provides complete isolation between the source and the load and can compensate for both undervoltage and overvoltage. For providing positive and negative compensation in N steps, (2N+1) taps will be required. When the taps are set to provide 10% positive compensation it results in load


36

Topology 1

Turn ratio = 10 : 11

Topology 1a

Turn ratio = 11 : 10

Topology Pros Cons

1

Does not require an additional series transformer

The load/grid fault current on secondary side directly flows through the switches and can damage them

Grid and load are isolated through the main transformer

More cost, material and space when main transformer is not an autotransformer and it is rated at 1.1pu

Requires (2N+1) number of switches

1a

Does not require an additional series connected transformer

More cost, material and space when main transformer is not an autotransformer and it is rated at 1.1pu



The switches are isolated and hence protected from load/grid fault current on secondary side


37

voltage of 1.1pu. If the load current is 1pu, then 1.1 pu power flows through the system. In 1, the major disadvantage is that the switches are not isolated from the load and hence during load/grid faults on the secondary side, the fault current will directly flow through the switches and can permanently damage them if they are of electronic or hybrid type. In contrast, in topology 1a the switches are safely isolated; however the switch and transformer ratings necessary are much higher than 1 as shown in Table 4.1a. Sizing of components for topology 1:

Eq 4.1 shows the relation between voltage, currents and number of turns for an ideal transformer. If x is used to represent the tap position on secondary, then N2 can be replaced by x. Further V1=1pu and maximum voltage on primary winding, V1 MAX=1pu as well as there are no taps on primary. For all cases the load current remains at I2=1pu assuming a fixed current load.

2 2 1

1 1 2

V NV

IN I

= =

(From Eq 4.1)

1 22 1

1 1

x I xN NV IV

= =

(From Eq 4.2)

Depending on the tap position in secondary, x can range from x=0.9N1 to x=1.1N1. When the one of the switch is conducting, the other switches will have to block up to 0.2pu voltage across them depending on switch position. Further if a worst case scenario is considerd in which all switches are open, that which might occur during say a fault condition - the forward blocking voltage required for the switches will range from 0.9pu to 1.1pu for the S3 to S1. These are summarized in the Table 4.1a and Table 4.1b.

Sizing of components for topology 1a:

2 1 2 22 1 V II

xV

xN N

= = (Eq 4.5)

Eq 4.1 holds good here as well. If x is used to represent the tap position on primary, then N1

can be replaced by x as shown in Eq 4.5 and x can range from x=1.1 N2 to x=0.9 N2. Further E1=1pu and load current I2=1pu. But the maximum voltage on primary winding E1 MAX will vary depending on the tap position and will take maximum value of 1.22pu when S3 is ON. When the one of the switch is conducting the other switches will have to block upto 0.2pu. When all the switches are in OFF condition they must be able to withstand a forward blocking voltage of 1pu as they are connected to same source voltage. These are shown in the Table 4.1a and Table 4.1b. The primary winding for topology 1a has to be designed for 1.22pu voltage; this occurs when the tap is at its S3 position providing 10% positive compensation. Presence of the windings above position S3, when S3 is in close position causes the total voltage on the primary to be 1.22pu.


38

Topology 1 Topology 1a x=1.1 N1 x=0.9 N1 x=1.1 N2 x=0.9 N2

I2(pu) 1 1 1 1 V2(pu) 1.1 0.9 0.9 1.1 Il(pu) 1.1 0.9 0.9 1.1

V1(pu) 1 1 1 1 V1 MAX (pu) 1 1 1 1.22

S (pu) 1.1 0.9 0.9 1.1

Condition Vb (pu)

Topology 1 Topology 1a All taps open 1.1 to 0.9 1

One tap closed Upto 0.2 Upto 0.2

Table 4.1a: Sizing of transformer and switch for topology 1 and 1a

(Refer to section 4.4 for †)

Topology

Rating of main transformer (p.u.) V †3 I

S 1’’ 2’’ 1’’ 2’’

1 1 1.1 1.1 1 1.1 1a 1.22/1 1.1 1.1 1 1.1

Topology No. of switches

Tap switch rating (p.u.)

I†6 Vb (One tap close)†4 Vb

†8 (All taps open) Fwd Rev

1 11 1 (0.2, 0) (0, 0.2) (1.1, 0.9) 1a 11 1 (0, 0.2) (0.2, 0) 1

Table 4.1b: Sizing of transformer and switch for topology 1 and 1a

II. Topology 2

Topology 2 is derived from 1 taking into consideration the need for isolation of switches. Here the compensating voltage is fed in series to the grid voltage using a series transformer of 1:1 turn ratio and 0.1pu power rating. The main transformer is a three winding transformer of 1.1pu rating with a turn ratio of 10:10:2 and the tertiary windings are used to produce the compensating voltage. The use of two transformers makes the design very expensive but


39

Topology 2

Turn ratio = 10:10:2 (Main) & 1:1 (Series)

Pros Cons The switches are isolated and hence protected from load/grid fault current

Requires additional series transformer of 0.1pu rating


More cost, material and space when main transformer is not an autotransformer and is rated for 1.1pu

Requires (2N+1) number of switches Main transformer is a three winding transformer which is costly and bulky

ensures that the load, source and switches are all isolated from each other. This paves way for reducing the blocking voltage ratings of the switches to about 10% of the previous case as they are fully isolated from source and load. At any given point of time either the top or bottom set of switches are activated to provide for positive or negative compensation respectively. There are a total of (2N+1) switches. Sizing of components for topology 2: The use of a center tap on the tertiary of the main transformer makes the maximum voltage on the series transformer primary to equal ±0.1pu when either S1 or S4 switch is conducting. For calculating switch ratings, when one of the switches is conducting say S1, then voltage that needs to be blocked by the other switches is much higher at 0.2pu for S4 and 0.15pu for S3. So the switches have to be rated for blocking 0.2pu to 0.1pu voltage depending on their position. The load current is 1pu and use of a 1:1 turn ratio series transformer means that the current through the switches is 1pu as well. The results are summarized in Table 4.2.


40


Topology Rating of main transformer (p.u.) Rating of series transformer (p.u.)

V †3 I S

V I S

1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 2†2 1 0.2/0.1 1.1 1 1.1 0.1 0.1 1 1 0.1





2 11 1 (0.2, 0) (0, 0.2) (1.1, 0.9)

Table 4.2: Sizing of transformer and switch for topology 2 III. Topology 3 and 3a Topology 3 corresponds to the conventional design of voltage regulators that are utilized in the grid where a mechanical arm connects the load point to the different taps through a motor operation [30]. It consists of an autotransformer that has a series and a shunt winding that are connected in series and shunt to the source voltage respectively. The taps are present on the series winding and a selector switch S connects either the top or the bottom of the series winding to the shunt winding so as to provide negative or positive compensation respectively. When S is connected to bottom position, the voltage of series winding adds to the voltage of the shunt winding providing positive compensation. When S is connected to the top, the dot point of both windings is connected together, resulting in the voltages opposing each other and negative compensation hence results.

Topology 3

Turn ratio = 10 : 1


41

The mechanical arm arrangement which is used generally in conventional OLTC to realize S1 to S3 can be replaced by separate electronic or hybrid switches. The benefit of this topology is the reduction in the number of switches required by half when compared to the topologies 1,1a and 2.

Topology 3a

Turn ratio = 9 : 2

Topology Pros Cons

3



Gains on cost, material and space when main transformer is an autotransformer of 0.1pu rating

Grid and load are not isolated

Requires (N+1+2) number of switches, N+1 for the taps and 2 for the selector switch

3a







42

In contrast, topology 3a does not have a selector switch S. Taps are present on the LV series winding and the input by itself is connected across the HV common winding and mid-point of the LV series winding. This design has the drawback of requiring twice the number of taps as compared to topology 3 but the benefit of not requiring a selector switch. The major advantage of these two topologies is the use of an autotransformer instead of a conventional transformer which saves on material, cost and space requirements. The main downside is the lack of isolation for the switches and between source and load resulting in high ratings for the switch of 1.1 to 0.9pu as explained below. Sizing of components for topology 3: The shunt winding (also referred as primary) is connected across the supply and is rated at 1pu voltage. The series winding (referred to as secondary) is on the same core and is electrically connected to the primary to give an autotransformer design. The output voltage is taken across the series and shunt winding as shown above. An important point to be noted here about an autotransformer is that the primary or secondary winding current is not the same as the source or the load current respectively. This is discussed in section 4.2 [27] and further explained here using Fig 4.3a and Table 4.3a. LV winding is connected in series to the HV winding. The current I2W through LV is equal to the input (source) current during buck operation and to the output (load) current during boost operation. For both boost and buck operation, the HV winding is common to both the input and output and carries a current I1W which is the difference of the input (source) current and output (load) current. This can be observed in Fig 4.3a. The HV common winding thus carries only a reduced current of |Iload - Isource| during the step up or down operation, so thinner copper wires can be used. The summary of the connections of LV and HV winding and the currents are provided in Table 4.3a.

Boost operation Buck operation Input=HV Input=LV+HV

Output=LV+HV Output=HV I2W=Iload I2W=Isource

I1W=|Isource - Iload| I1W= |Iload - Isource|

Table 4.3a: Winding, source and load currents in autotransformer during different modes

Fig 4.3a: Autotransformer voltages & currents for (a) buck & (b) boost operations


43

All values in pu

S=up & S3=On

S=down & S1=On

I2=Iseries 1 1 V2W -0.1 0.1 Vload 0.9 1.1

Il 0.9 1.1 Ishunt=|I2–I1| 0.1 0.1

V1 1 1 S 0.1 0.1 Vb 0.1 to 0 0 to -0.1

For providing 10% compensation, the turn ratio of HV:LV will be 10:1 and the transformer windings will be rated as 1pu voltage and 0.1pu current for HV and LV for 1pu current and 0.1pu voltage! To evaluate the ratings of the switches, Fig 4.3b is used as a basis. The input source voltage is assumed to be 1pu, the load current I2 =1pu and turn ratio between HV and LV is 10:1. It is observed that the rated load current of 1pu flows directly through all the switches and hence the switches are rated for 1pu current. When one of the taps is ON then the other switches have to block up to 0.1pu voltage; e.g. If S1, S2 and S3 provide 0%, 5%, 10% compensation respectively then when S3 is ON, voltage across S1 is 0.1pu and that across S2 is 0.05pu. On the other hand if S1 is ON, the switches have to block a negative voltage across them. Thus the taps have to block bidirectional voltages. In the worst case scenario when all taps are OFF during a fault condition, the switches experience a range of voltages that they must be capable of blocking depending on the selector switch position. When S is connected down, S1 experiences 1.1pu voltage while S3 blocks 1pu voltage. In case that S is connected up, then S1 block 1pu and S3 blocks 0.9pu. Even though the selector switch is shown as one switch, it is actual realized with help of two electronic or hybrid switches. When the input is connected to one of the output terminals, then the selector switch must block 0.1pu voltage between the input and unconnected output terminal. This is a vital consideration in the design of the selector switch. Sizing of components for topology 3a: With the absence of the selector switch the ratings of components becomes straightforward in topology 3a. If it is assumed that the switches S1 to S5 provide output voltage of 1.1pu to 0.9pu in steps of 0.05pu, then the switches will have to block upto 0.2pu voltage in both forward and reverse direction depending on which switch is ON. If S1 is ON, S2 will block a

Table 4.3b: Sizing of components for topology 3

Fig 4.3b: Direction of currents in autotransformer in topology 3


44

reverse voltage of 0.05pu, S3 of 0.1pu, S4 of 0.15pu and S5 of 0.2pu. On the other hand, if S5 in ON all the other switches will have to block forward voltages, the maximum being that for S1=0.2pu. All switches have to be rated for 1pu load current. The transformer design is similar to 3, the difference being that turn ratio between HV: LV ratio is 9:2 and the voltage rating of the LV is 0.2pu and HV is 0.9pu. Twice as many taps are there on the LV winding in topology 3a as compared to topology 3.


Topology Rating of main transformer (p.u.)

V †3 I S

1’’ 2’’ 1’’ 2’’ 3 1 0.1 0.1 1 0.1 3a 0.9 0.2 0.1 1 0.1


Tap switch rating (p.u.) Selector switch rating (p.u)


†8 (All taps open)

I†6 Vb †7 Fwd Rev

3 6+2†5 1 (0.1, 0) (0, 0.1) (1.1, 1) 1 -0.1, 0.1 3a 11 1 (0.2, 0) (0, 0.2) (1.1, 0.9) - -

Table 4.3c: Sizing of transformer and switch for topology 3 and 3a

IV. Topology 4 and 4a

Topology 4

Turn ratio = 8:2 (Main) & 1:1 (Series)


45

Topology 4a


Topology 3 and 3a have the drawback of the switches not being isolated from the load. Topology 4 combines the benefits of using an autotransformer and need to have isolation for switches. Here the compensation voltage is derived from the grid voltage through the use of an autotransformer tapped at 0% and 20% points and a center tap at 10%. (2N+1) switches are required for the operation, half for positive and rest for negative compensation. A series transformer injects the compensating voltage to the grid. However, the switches are not isolated from the source side in 4. When isolation is actually a requirement, then topology 4a can be used. It must be noted that in both 4 and 4a, the load is never isolated from the source. The switches used can be electronic or hybrid switches combining mechanical and electronic switches. A major downside of the design is the need for an additional series transformer which is rated for 0.1pu power rating besides the 0.1pu power rated main autotransformer. The autotransformer always operates in buck mode.

Topology Pros Cons

4



The switches are isolated and hence protected from load/grid fault current



4a

The switches are protected from load/grid fault current because of isolation


The switches are isolated from both source and load side

More cost, material and space when main transformer is not an autotransformer



46

Sizing of components for topology 4 and 4a: The sizing of components is very similar to topology 2, the main difference being that instead of a conventional three winding transformer an autotransformer is being used in 4 and a two-winding transformer in 4a. The transformers have a reduced power rating of 0.1pu as it handles only the compensating power. In 4, the design is similar to Fig4.1(b) of an autotransformer with step-down operation. The common winding is the section of winding from 0% to 20% with the taps on it. The primary is rated at 1pu voltage while the secondary is rated at 0.2pu voltage. The series transformer is rated for 0.1pu voltage i.e. for ±10% compensating voltage and for 1pu load current. Since it is a 1:1 series transformer, the current through the switches is equal to the 1pu load current. The corresponding current drawn by autotransformer from source is 0.1pu and this flows through the upper part of the transformer. From the previous discussion, the current through the common winding is the difference of source and load currents and hence it is 0.9pu. For 4a, the design of the switches is same as topology 4. The main transformer however has a distinct primary and secondary. The current in the secondary winding is the same as the load current and is 1pu. So thicker copper wire is required and a separate secondary winding has to be made in the case of 4a, which is disadvantage when compared to 4. For the blocking voltage of the switches in both cases, it can be observed that when one of the switches is conducting, the maximum voltage that other switches have to block is upto 0.2pu. For example, when S1 is ON, 0.2pu reverse voltage must be blocked by for S4. Both positive and negative voltages must be withstood depending on the switch position. If all the switches are in blocking state and no compensating voltage is being fed to the grid, the series transformer operates in a reverse fashion and used the grid voltage as input and injects voltage from the grid onto the switches. In such a scenario, voltage of upto 1.1pu must be blocked by the switches. The results are summarized in Table 4.4.



V †3 I S

V I S

1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 4 0.8 0.2/0.1 0.1 0.9 0.1 0.1 0.1 1 1 0.1 4a 1 0.2/0.1 0.1 1 0.1 0.1 0.1 1 1 0.1

Topology No. of

switches




4 & 4a 11 1 (0.2, 0) (0, 0.2) (-0.9, -1.1)

Table 4.4: Sizing of transformer and switch for topology 4 and 4a


47

V. Topology 5 and 5a

Topology 5


Topology 5a


The use of a selector switch was shown to reduce the total number of switches by half as in topology 3. The same technique is implemented in case 4 and 4a, to give topology 5 and 5a. The position of the selector switch decides the sign of compensation. The total number of switches is reduced from (2N+1) as in the previous case to (N+1). For topology 5, the taps are present on 0% to 10% section of the autotransformer windings. When S is connected in UP position, positive compensation is obtained and when S is connected DOWN, negative compensation results. Both main and series transformers are rated for 0.1pu power. All the switches that constitute the taps have to be bidirectional in nature. The transformer turn ratio is 9:1 where the primary is considered as that part with no taps and the secondary is 10% of windings having taps. The series transformer isolates the switches from the load side. In case of topology 5a, a conventional two winding transformer is used with turn ratio of 10:1 and taps are present on the secondary side along with the selector switch.


48

Topology Pros Cons

5



Gains on cost, material and space when main transformer is an autotransformer of 0.1 pu rating


Requires (N+1+2) number of switch, N+1 for taps and 2 for selector

5a

The switches are isolated from both source and load side.


Requires (N+1+2) number of switch, N+1 for taps and 2 for selector

More cost, material and space when main transformer is not an autotransformer

Sizing of components for topology 5 and 5a: Both these topologies 5 and 5a have similar design requirements as topologies 4 and 4a. The main difference comes from the voltage ratings and turns of the transformers and the blocking ratings of the switches. In 5, the secondary of the autotransformer has taps in the 0% to 10% point. The upper HV section will be rated for 0.9pu voltage and the lower LV section with taps will be rated for 0.1pu voltage, thus the total voltage across the series connection of HV and LV will be 1pu. The maximum switch blocking voltage when one of the tap is ON is 0.1pu; this value is half of what was observed in the previous topology. The selector switch blocking rating is also 0.1pu. All the switches carry the load current of 1pu so their current rating is 1pu. The series transformer will be rated for the load current of 1pu and 0.1pu voltage for 10% compensation.



V †3 I S

V I S

1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 5 0.9 0.1 0.1 0.9 0.1 0.1 0.1 1 1 0.1 5a 1 0.1 0.1 1 0.1 0.1 0.1 1 1 0.1

Topology No. of

switches



†8 (All taps open)


5 & 5a 6+2 1 (0.1, 0) (0, 0.1) (-1, -1.1) 1 0.1, -0.1

Table 4.5: Sizing of transformer and switch for topology 5 and 5a


49

The main distinction in 5a is that the autotransformer is replaced by a two winding transformer of turn ratio 10:1. The primary will be rated for the full nominal voltage and 10% of the nominal current. The switch and series transformer ratings will be the same as topology 5 discussed above .

VI. Topology 6 In an attempt to completely do away with the main transformer and have only a single series transformer, topology 6 and 7 are obtained. The selector switch is directly connected to the main supply and has a maximum voltage of 1pu across it. The series transformer has a center tap design with taps present on both the upper and lower half. When the top switches are ON, negative compensation is obtained and positive compensation is obtained when the bottom switches are ON. It is important to note here that unlike the previous designs, the series transformer does not have a 1:1 turn ratio but is of step-down type. It has the advantage of isolation of switches from load as seen before when series transformers are used.

Topology 6

Turn ratio = 100:1

Pros Cons


Requires series transformer of 0.1pu rating

Gains on cost, material and space when main transformer is absent


Voltage rating of switches required are very high


50

Sizing of components for topology 6: Since source voltage has to be stepped down to provide upto 10% compensation, the series transformer is a step down type with turn ratio of 100:1. Fig 4.4 explains the sizing of switches for one half of the taps for negative compensation; where S1 is the tap for 2% compensation and S2 for 10% compensation. If N2 is the number of turns of the series transformer secondary, as expected the taps S1 and S2 are positioned at 50N2 and 10N2

respectively based on Eq 4.1. The taps for 8%, 6% and 4% compensation will be present in between at 12.5 N2, 16.66 N2 and 25 N2 respectively. Now lets us look at the case when S1 is conducting. Voltage across S2 is given by the difference of 1pu and the primary winding voltage at 10N2 of 0.2pu; which equals 0.8pu. It must be considered that the taps S3 and S4 for positive compensation are also connected to the same transformer and due to transformer action they will experience a voltage of upto 2pu. However on the other hand if S2 tap is ON providing 10% compensation, then the voltage at 50N2 will then be 5pu (boost operation) and hence voltage across the switch S1 will be -4pu. In such a case S4 will experience 6pu voltage across it. Since there are a total of 11 switches, depending on which switch is ON, the other switches will have to block different set of voltages ranging between 6pu and -4pu. The worst case scenario for blocking voltages occurs when the taps are in OFF condition and do not feed a compensating voltage. In such a case the transformer acts in a reverse fashion and takes in voltage from the tap-less side and boosts it by the turn ratio of 50:1 and very large voltages are implicated on the switches. This can be upto 51pu and -49pu as indicated in Table 4.6.

Fig 4.4: Schematic to estimate the voltage rating of switches in topology 6

When taps are varied, largely different turn ratio are obtained. S1 is at 50 N2 and S2 is at 10 N2. From Eq 4.1 it is seen that this will result in largely different current through the switches and through the different tap sections of the transformer. The salient gain is hence that the switch S1 can have a lower current rating of 0.01pu as compared to 0.1pu for S2.


51

It can thus be seen that even though this topology does not require a main transformer, it requires switches of high blocking voltage, as high as 6pu. Further the series transformer will be also have to be rated for 10pu voltage on the primary and 0.1pu on the secondary. This makes the transformer and taps extremely costly to build, posing as a major drawback.


Topology Rating of series transformer (p.u.)

V I S

1’’ 2’’ 1’’ 2’’ 6 10/1 0.1 (0.1,0.02) 1 0.1

Topology No. of switch



(All taps open) Fwd Rev 6 11 (0.02, 0.1) Upto 6 Upto 4 (-49, 51)

Table 4.6: Sizing of transformer and switch for topology 6

VII. Topology 7

Topology 7

Turn ratio = 20:1

An alternate strategy to topology 6, would be to move the taps to the secondary of the series transformer and thereby preventing the boost operation. The major hindrance in doing this is that the switches now carry the load current directly and are not isolated. The benefit is that the switch voltage ratings are reduced and no main transformer is required as in the previous case. The primary side of the transformer is of a center tap design and the transformer has


52

Pros Cons Requires (N+1) number of switches Requires series transformer of 0.1pu rating

Gains on cost, material and space when main transformer is absent


20:1 turn ratio. A selector switch is present to connect the transformer in such a way so to be able to provide both polarities of compensation. Sizing of components for topology 7: The series transformer primary and secondary are rated for 2pu and 0.1pu voltage respectively. The rated currents for primary and secondary will be 0.1pu and 1pu. To evaluate the ratings of the switches, when one tap is ON for say S1 providing 10% compensation, the other switches will have to block upto 0.1pu voltage. Depending on the position of the selector switch, both polarities of compensation can be achieved. When S is connected DOWN, positive compensation is obtained and vice versa for UP connection. The selector switch by itself has to be blocking a voltage at the unconnected terminal and this voltage will equal the 2pu voltage at the primary of the transformer. Both selector switch and tap switch have to be of bidirectional nature. When all switches are blocking, the maximum voltage to be blocked is 1pu, as the entire source voltage comes across the switches. The switches will be rated for 1pu current as they directly carry the load current. The major advantage of the topology is its ability to reduce the transformer and switch ratings as compared to topology 6 but still managing to use only a single transformer.


Topology Rating of series transformer (p.u.)

V I S

1’’ 2’’ 1’’ 2’’ 7 2/1 0.1 0.1 1 0.1



I†6 Vb (One tap close) †4 Vb

†8 (All taps

open) I†6 Vb †7

Fwd Rev 7 6+2 1 (0.1,0.04) (0.1,0.04) (-0.9, -1) 0.1 2, 2

Table 4.7: Sizing of transformer and switch for topology 7


53

4.4 Rating of components for different topology

Note (indicated by †):

1. The nominal load current is 1pu and nominal source voltage is 1 pu. 1” and 2” refers to the primary and secondary winding respectively.

2. For topology 2, the secondary winding of the main transformer that provides for the load power will be rated at 1pu voltage, 1pu current and 1pu power rating. 2” corresponds to the ratings of the tertiary winding of topology 2.

3. The ratings that are given as ‘a/b’ corresponds to passive and active ratings of the component. Active rating of the component is defined by that part of the component that is involved in active power conversion, while passive rating is actual rating of the component but not necessary that all of it carries power. E.g. Topology 1a, the primary winding of transformer has a rated voltage of 1.22pu / 1pu. When S3 is closed, it means even though the winding is rated for 1.22pu voltage, only part of it having 1pu across it is involved in active power conversion and carries the 1pu current.

4. The switch ratings are indicated as (x, y). It indicates that, depending on the switch position the switch is rated for values in the range of ‘x to y’.

5. When the number of switches is indicated as 6+2, the +2 corresponds to number of components required to realize the selector switch

6. The switches should be capable of conducting bidirectional current of the specified rating and block bidirectional voltage.

7. In ON condition, the selector switch is connected to one of the two output terminals and has to block the voltage between the input and unconnected output terminal.


V †3 I S

V I S

1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 1’’ 2’’ 1 1 1.1 1.1 1 1.1 - - - - - 1a 1.22/1 1.1 1.1 1 1.1 - - - - - 2†2 1 0.2/0.1 1.1 1 1.1 0.1 0.1 1 1 0.1 3 1 0.1 0.1 1 0.1 - - - - - 3a 0.9 0.2 0.1 1 0.1 - - - - - 4 0.8 0.2/0.1 0.1 0.9 0.1 0.1 0.1 1 1 0.1 4a 1 0.2/0.1 0.1 1 0.1 0.1 0.1 1 1 0.1 5 0.9 0.1 0.1 0.9 0.1 0.1 0.1 1 1 0.1 5a 1 0.1 0.1 1 0.1 0.1 0.1 1 1 0.1 6 - - - - - 10/1 0.1 (0.1,0.02) 1 0.1 7 - - - - - 2/1 0.1 0.1 1 0.1

Table 4.8a: Summary of sizing of transformer for all topologies


54




(All taps open)


1 11 1 (0.2, 0) (0, 0.2) (1.1, 0.9) - - 1a 11 1 (0, 0.2) (0.2, 0) 1 - - 2 11 1 (0.2, 0) (0, 0.2) (1.1, 0.9) - - 3 6+2†5 1 (0.1, 0) (0, 0.1) (1.1, 1) 1 -0.1, 0.1 3a 11 1 (0.2, 0) (0, 0.2) (1.1, 0.9) - -

4 & 4a 11 1 (0.2, 0) (0, 0.2) (-0.9, -1.1) - - 5 & 5a 6+2 1 (0.1, 0) (0, 0.1) (-1.1, -1) 1 0.2, 0.2

6 11 (0.02, 0.1)

Upto 6 Upto 4 (-49, 51) - -

7 6+2 1 (0.1,0.04) (0.1,0.04) (-0.9, -1) 0.1 2, 2

Table 4.8b: Summary of sizing of switches for all topologies

4.5 Summary and conclusion In this chapter, the theory of the ‘make and break’ tap changing phenomenon is examined and the fundamentals of an autotransformer are explained. It is shown that an autotransformer has several advantages over a conventional two winding transformer with the same rating of the windings – the autotransformer uses much lesser copper and more power can be transferred across the terminals. Different topologies are proposed for the design of an on-load tap changing transformer using both autotransformers and conventional transformers. The topologies are analyzed with respect to the voltage and current ratings of the transformer and tap switches. This forms the basis for the short listing of the topologies and for simulating their steady state performance which is performed in the next chapter.

Chapter 5: Connection schemes for three phase network and simulation of OLTC topologies

55

Chapter 5

Connection schemes for three phase network and simulation of OLTC topologies

The aim of this chapter is to make a simulation model for the analysis of various OLTC topologies and its connection to a distribution network. It is important to realize that the distribution network is a three phase network operating at two voltage levels namely the medium voltage and low voltage level. As a first step, different connection schemes for single phase tap changing transformers to a three phase network are analyzed. In the next step, simulation models for overhead lines and cables are made based on parameters obtained from GE. Using the line model and three phase connection scheme, the single phase OLTC topologies discussed in chapter 4 are then simulated in the PLECS simulation package. The aim of the simulation is to:

1. Understand the steady state operation of the circuit 2. Look into the requirements for a bidirectional switch with respect to blocking capabilities

for both forward and reverse voltage and conducting bidirectional currents 3. Verify the different ratings for the switches and transformers as they have been estimated

theoretically in chapter 4. 4. Build a foundation for developing the controller for the compensator and for performing

transient and fault analysis of the circuit for the shortlisted topologies. 5.1 Mathematical analysis for connection of OLTC to a three phase network In this section, different possible connection schemes for single phase transformers to a three phase distribution network are looked into [34]. The network parameters as obtained from GE for the three phase distribution network are mentioned in Table 5.1.


56

MV LV Voltage level (line-line) 20 kV 400 V Maximum rated load current 577A 577A Maximum length of line (km) 30 1 Problems due to unbalanced loading No Yes Independent regulation of phases required No Yes Neutral available No Yes

Table 5.1 – Parameters of three phase distribution network An important difference between the medium voltage (MV) and low voltage (LV) network is the presence of the neutral and the need to have independent regulation of the phases. This means that the MV is a three phase three wire network while the LV is a three phase four wire network. The objective in the MV is to control the line-line voltage while in the LV; the aim is to control the phase voltage. Different possible connection can be [33, 34]:

1. Three phase four wire network with Y connected transformers 2. Three phase network with Δ connected transformers 3. Three phase three wire network with Y connected transformers 4. Three phase network with open Δ connected transformers

The 4 connections schemes are analyzed mathematically. Then simulations of above 4 connection schemes are carried out for an ideal series compensator based on topology 3. In all cases the magnitude and phase of injected series voltage is analyzed assuming a constant load current. 5.1.1 Three phase four wire network with Y connected OLTC (for LV): It is typical for LV European distribution network to have a neutral available [25, 26]. Thus for three phase four wire network, the three single phase OLTC transformers can be connected between the phase and neutral in Y formation as shown in Fig 5.1a. The points S, L and SL shown in Fig 5.1a are indicated on Fig 5.1b for topology 3a where S-SL corresponds to that part of the winding that is connected to the input and S-L is that part of the winding that injects the series voltage to the grid. The start point of the transformer connection is connected to the neutral of the network. The compensating voltage is derived from the phase voltage and injected in/out phase with the phase voltage for positive and negative compensation respectively. The main feature of this type of connection is that three OLTC units can achieve independent regulation of each phase voltage and the injected voltage will always be in phase with the grid voltage. This is mathematically derived and explained using the phasor diagram in Fig5.1c


57

Fig 5.1a: Y connection of OLTC transformer to a three phase four wire network

Fig 5.1b: Topology 3 with connection points S, L and SL

For a three phase line, it can be written (Eq 5.1):

0

240 ( 0.5 3 / 2)

120 ( 0.5 3 / 2)

a m

b m m

c m m

V V

V V V i

V V V i

= ∠ °

= ∠ ° = − −

= ∠ ° = − +

(Eq 5.1)

Since the OLTC is connected across the phase and neutral, the compensation voltage ΔVx is hence derived from the phase voltage. The output voltage at the OLTC, Vx’ can be written as (Eq 5.2), where x can be any phase a,b or c and C% is tap position of OLTC in fraction (0 to 1)

%

%

)''

( (1 )

x x

x x

xx x

V C VV V VV C V

∆ == + ∆= +

(Eq 5.2)


58

The different phasors explained in the equation (Eq 5.1-Eq 5.3) are indicated in the phasor diagram Fig. 5.1c. It can be seen that the injected compensation voltage ΔVx (red) is in-phase with the phase voltages Vx (black).

Fig 5.1c: Phasor diagram for four wire network with Y connected transformers Conclusion for a four wire distribution network with Y connected transformers:

• The transformer should be rated for the phase voltage of the network • The injected series voltage is in phase with the corresponding phase voltage • Independent regulation of phase voltages can be achieved by setting different values of C%

for each phase. • Fluctuations in source voltage and load currents in a particular phase, will only affect tap

position of transformer connected to that phase

Fig 5.1d: Simulation of Y connection of OLTC to a three phase four wire network

ΔVc Vc

120° N Vc ΔVa

Vb

ΔVb

Va Va’ ΔVa


59

The connection was verified using a simulation model of topology 3 in PLECS and it is shown in Fig 5.1d. The modeling of the line is explained in section 5.2 of this chapter. 5.1.2 Three phase network with Δ connected OLTC (for MV and LV): For three phase network having three or four wire, the OLTC transformers can be connected in a Δ fashion as shown in Fig 5.2a. The transformers are connected between the phases and compensating voltage is extracted from the line-line voltage. Since the compensating voltage is derived from the line-line voltage and injected into the phase voltage, there is phase shift between the injected voltage and the corresponding grid voltage. It can be shown that the compensation voltage is injected in/out phase with the phase voltage by 30°. The mathematical analysis of the connection scheme is given below and the phasor diagram is shown in Fig5.2b. Fig5.2c shows the simulation model of the system. For such a system, the line-line voltage can be written as (Eq 5.4):

3 ( 3 / 2 / 2) 3 30

3 ( ) 3 90

3 ( 3 / 2 / 2) 3 150

ab a b m m

bc m m

ca c m

b c

a m

V V V V i V

V V V V i V

V V V V i V

= + = ∠ °

= − = ∠ − °

− =

− =

= − + = ∠− = °

(Eq 5.4)

The compensation voltage is extracted from the line-line voltage Vxy and injected in series with the phase voltage Vx, where C% is the tap position of the OLTC unit:

% ( )

' x xy

xyxx

V C VV V V∆ =

= + ∆ (Eq 5.5)

Fig 5.2a: Simulation of Δ connection of OLTC transformers to three phase network


60

The output phase voltage of the OLTC Vx’ is given as:

%

%

%

' 0 3 30

' 240 3 90

' 120 3 150

a m m

b m m

c m m

V V C V

V V C V

V V C V

= ∠ °+ ∠ °

= ∠ °+ ∠− °

= ∠ °+ ∠ °

(Eq 5.6)

The corresponding line-line voltages at output are Vxy’:

%

%

%

3 30 3 ( 3 60 )

' 3 90 3 ( 3 60 )

' 3 150 3 ( 3 180

'

)

ab m m

bc m m

ca m m

V V C V

V V C V

V V C V

= ∠ °+ ∠ °

= ∠− °+ ∠− °

= ∠ °+ ∠ °

(Eq 5.7)

Some important observations that can be made from the equations are:

• In (Eq5.6) the injected voltage leads the corresponding phase voltage by 30°, so the compensation is not in-phase. The output voltage of the OLTC Vx’ leads the phase voltage by upto 5 ° depending on the tap position C% of each of the transformers.

• The injected voltage has a magnitude that is √3 times greater than that observed for a Y network for the same tap position i.e. C%. This can be observed in equations (Eq 5.6) and (Eq 5.7) by the extra √3 term. Thus Δ connection has the ability to provide for compensation greater than 10%. Calculations show that this can go upto 15% when all transformers are set for positive compensation with C%=0.1 [33,34].

Fig 5.2b: Phasor diagram for three wire network with Δ connected transformers

ΔVc Vca V’c VC Vab 120° V’a

ΔVa

N 30° Va Upto 5° Vb

ΔVb V’b

Vbc


61

Fig 5.2c: Simulation of Δ connection of OLTC transformers to three phase network

In the case for the line-line voltage (Eq 5.7), the output voltage of the OLTC Vx’ leads the input Vx by a finite angle but much less than 30°. This is because the injected voltage magnitude is only upto 10% of the line voltage, this causes small but definite phase shift between the voltages at the input and output of the compensator, upto 5°.

Conclusion for a three phase network with the Δ connected transformer:

• The transformer should be rated for the line-line voltage of the line, which means a √3 times higher voltage ratings for the transformer compared to the Y connection. Higher voltage of operation for the transformer results in √3 times increase in voltage ratings for the switches

• The compensated line-line voltage Vxy’ is affected by the compensating voltage injected in phase x and y. Thus fluctuations in source voltage and load currents in a particular phase, will affect tap position of transformer connected to all three phases. So independent regulation of phases cannot be achieved in a straight forward fashion.

• It should be noted here that in the above analysis all three phases are assumed to have same sign of compensation and tap position. If this is not the case, then the magnitude and phase shift in the output voltage will be affected for each phase.

Vab Va Vab’

ΔVc

Va’


62

5.1.3 Three phase three wire network with Y connected OLTC (for MV): For a three phase three wire network, the OLTC transformers can also be connected in a Y connection with a floating neutral as shown in Fig 5.3a. The system will work only under the strong assumption that the grid voltages are balanced and all three phases are provided with the same magnitude of compensation voltage. If that is not the case, it will result in shift of the start point of the Wye connection resulting in overstressing of the transformer insulation and erratic compensator operation [33]. This is because Y connected compensators in a four wire network have their Y point connected to the neutral of the system. They can thus regulate the phase voltages independently and if there is an unbalance in the regulation between the phases, it results in an unbalance current that flows through the neutral wire and prevents the Y point from shifting. In a three wire network, such a neutral connection is not possible and during unbalanced operation the Y point will shift continuously resulting in erratic operation. Thus independent regulation of phase or line voltages cannot be realized through a Y connection in a three phase three wire distribution network. The solution to this problem would be to ground the start point by some means [33]. One method is to connect SL point back to the grounded secondary neutral of a substation transformer that is located nearby. Alternatively, a small grounding bank consisting of three transformers, each from one-third to two-thirds the kVA rating of the individual regulators can be installed to ground the neutral. The mathematical analysis of the connection scheme is similar to that of the Y connected transformers in a four wire network, assuming balanced regulation. From (Eq5.2) it can be written

%

%

)''

( (1 )

x x

x x

xx x

V C VV V VV C V

∆ == + ∆= +

(Eq5.8)

The compensated phase voltages can be written as below, similar to (Eq5.3):

%

%

%

0 0' 240 240' 120 12

'

0

a m m

b m m

c m m

V V C VV V C VV V C V

= ∠ ° + ∠ °= ∠ ° + ∠ °= ∠ ° + ∠ °

(Eq5.9)

The corresponding line-line voltages at output are:

%

%

%

3 30 3 30

' 3 90 3 90

' 3 150 3 150

'ab m m

bc m m

ca m m

V V C V

V V C V

V V C V

= ∠ ° + ∠ °

= ∠ − ° + ∠ − °

= ∠ ° + ∠ °

(Eq5.10)


63

ΔVc ΔVca ΔVab Vc Vab Vca 120°

30° N Va ΔVa

Vb

ΔVb

Vbc ΔVbc

Fig 5.3a: Y connection of OLTC transformers in a three wire network

Thus the line-line voltage at output of OLTC Vxy’ are in phase to the input line voltages Vxy, thus in-phase compensation is obtained. The different phasors explained in the equation (Eq5.8-Eq5.10) are indicated in the phasor diagram Fig. 5.3b. It can be seen that the injected voltage ΔVx (red) is in-phase with the phase voltages Vx (black) and so is the case for the line voltages ΔVxy (pink), Vx (blue). Fig. 5.3c shows the simulation model for the connection scheme

Fig 5.3b: Phasor diagram for three wire network with Y connected transformers


64

Conclusion for a three wire network with the transformer connected in Y:

• The transformer should be rated for the phase voltage of the network • The injected series voltage ΔVx is in phase with the corresponding phase voltage Vx. The

compensating voltage as seen between the lines ΔVxy is also in phase with the corresponding line-line voltage Vxy.

• Because of the floating star-point problem, independent regulation of phases cannot be achieved. The system cannot be implemented when the voltages are unbalanced, unless and until additional methods are implemented to ground the star-point.

Fig 5.3c: Simulation of Y connection of OLTC transformers in a three wire network

5.1.4 Three phase three wire network with open-Δ connected OLTC (MV): An innovative method for controlling the line-line voltage in a three wire network using only two OLTC units is through the use of an open-delta connection as shown in Fig 5.4a. The two units are connected between phase a-b and phase c-b using phase b as the common connection point. The injected voltages ΔVa and ΔVc are thus derived from the line-line voltages. Direct and independent regulation of the line-line voltages Vab and Vbc results from this connection scheme, while the compensation in phase a-c, is average of the compensation voltages ΔVab and ΔVcb. During balanced operation (ΔVa= ΔVc), in-phase compensation of all three line-line voltages occurs and during unbalanced operation, Vab and Vbc experience in-phase compensation while Vac alone experiences a phase shift of upto 5 . Since unbalanced

Vab Va

ΔVc

Vab’


65

Fig 5.4a: Open Δ connection of OLTC transformers in three wire network

loading of phases is a situation that does not occur in the MV as given in Table 5.1, this would be a cost-effective connection scheme for MV network as construction of a third MV transformer can be avoided. The mathematical analysis of the connection scheme is presented below and the phasor diagram is shown in Fig 5.4b. The phase and line voltages can be defined using Eq.5.1 and Eq. 5.4. The voltage injected into phase a and c can be given by (Eq. 5.11). It can be seen that the injected voltages are in phase with the corresponding phase voltages. ΔVa leads Va by 30 while ΔVc lags Vc by 30

%

% % ( )0

a ab

c cb bc

b

V C VV C V C VV

∆ =∆ = = −∆ =

(Eq. 5.11)

The corresponding line-line voltages at output of OLTC are: '

'%

( ) ( )

(1 )ab a a b b ab a

ab ab

V V V V V V VV C V

= + ∆ − + ∆ = +∆

= +

( )'

'%

( )

(1 )bc b b c c bc c

bc bc

V V V V V V V

V C V

= + ∆ − + ∆ = −∆

= + (Eq. 5.12)

( )( ) ( )

'

'% % %

'%

( )

(1 )

ca c c a a ca a c

ca ca ab bc ca ab bc

ca ca

V V V V V V V V

V V C V C V V C V V

V C V

= + ∆ − + ∆ = −∆ + ∆

= − + − = − +

= +

Thus it can be observed that the line-line voltages at the OLTC output are in-phase with the corresponding grid voltages. Here it is assumed that both units have same value of C%. If that were not the case, Vab’ and Vbc’ will be in phase with Vab and Vbc, while Vca’ will be phase


66

shifted from Vca by upto 5 . The magnitude of the phase shift is proportional to difference in C% of the two units. The phasors corresponding to (Eq5.11-Eq5.12) are indicated in the phasor diagram Fig. 5.4b. The connection scheme was verified in simulation and the model is in Fig. 5.4c.

Fig 5.4b: Phasor diagram for Δ distribution network with open Δ connected transformers

Fig 5.4c: Simulation of open Δ connection of OLTC transformers in three phase network

Vab Vab’

ΔVc


67

Conclusion for a three wire network with the open Δ connected transformer:

• The transformer should be rated for the line-line voltage of the line, which means a √3 times higher voltage ratings for the transformer compared to the Y connection.

• Two units are sufficient to provided regulation of all three line-line voltages. If the two units are injecting voltage in phase a and c, then independent regulation of line-line voltage Vab and Vbc can be achieved. The compensation in phase a-c ΔVac will be the average of other two injected voltages ΔVab and ΔVcb.

• Since all three phases are usually equally loaded in the MV (refer Table 5.1), both the units in open-delta connection are set for providing the same magnitude of compensation C%. As shown earlier, this results in in-phase compensation meaning that the injected line-line voltages are in phase to the grid voltage. 5.1.5 Connection scheme for MV and LV distribution network: From the comparison of the three types of connection, it can be concluded that the best way of connecting the transformers in three phase four wire LV network is in Wye fashion with star-point connected to neutral. The presence of neutral in LV and the ability to have independent in-phase compensation in all three phases drives this choice. For MV no neutral wire is available and the need is to regulate the line-line voltages. The open-delta connection is the most economical connection in such a scenario requiring just two units. The three phase connections and voltage-currents ratings of the transformer for both MV and LV are summarized below in Table 5.2 based on information from Table 5.1.

Connect Rated voltage

(V)

Rated current

(A)

3 Phase load

power

Single phase transformer Voltage

(V) Current

(A) Power

[A] [B] [C] [D=0.1*A] [E=B] [F=D*E]

MV Open delta

20,000 (Line voltage) 577 20 MVA 2000 V 577 1.15

MVA

LV Wye 230 (Phase voltage) 577 400 kVA 23 V 577 13.3

kVA

Table 5.2: Connection scheme and rating of transformer for MV and LV


68

5.2 Modeling of distribution network as T-section: The OLTC will be connected to the distribution network to compensate for voltage drop/gain along the feeder. It is hence vital that a model of the distribution network be made. The three phase distribution network is modeled using a lumped parameter T-network. The T-network is chosen over a simpler series RL network as it takes into account the influence of the line/cable capacitance on the voltage fluctuations in the line. It is assumed that the MV line/cable is rated at 20kV and can go upto a maximum possible length of 30km and LV rated at 1kV can go upto 1km as given in Table 5.1. Since these distances are relatively short, the long transmission model using distributed parameters is hence not used. As a first step, the data about the different line types used by the electric utility company EnBW AG in Germany were obtained. In particular 8 different types of lines/cables are chosen, half for MV and rest for LV. For each voltage level two types are used, one with maximum R/X ratio and other minimum R/X ratio. The line parameters are in sequence components and are shown in Table 5.3. For the purpose of modeling a three phase system, the sequence components are converted to abc components. This is done by using the equations listed below:

(2 ' 0 ') / 3(2 ' 0 ') / 3( 0 ' ') / 3( 0 ' ') /

'

'

3

3( 0 ')( ) / ( ' 0 ')p

s

s

m

m

g

RLRL

R R

C CC C C

R

C

L LR

L L

C

=

−

+= += −= −=

=

(Eq5.13)

Rs – Self resistance of the line per unit length [Ω/km] Rm – Mutual resistance of the line per unit length [Ω/km] Ls – Self inductance of the line per unit length [mH/km] Lm – Mutual inductance between the phases per unit length [mH/km] Cp – Capacitance per unit length between the line and common point of coupling of all phases as shown in Fig 5.5 [uF/km] Cg – Capacitance per unit length between the common point of coupling of all phases and ground as shown in Fig 5.5 [uF/km] The values obtained from Eq 5.13 are for unit length of the line. The actual R, C, L values are obtained by multiplying the above with the appropriate length of the line. The three phase network is obtained through a series connection of two lumped parameter T-network as shown in Fig5.4.


69

Name Cable/OHL Rated VLL

Rated ILL

Rated ILL (air)

Nominal Freq.

kV kA kA Hz Al150 OHL 20 0.47 0.47 50 Al_St35_6 OHL 20 0.47 0.47 50 N2XS2Y3x1x300 Cable 20 0.599 0.724 50 NAEKBA3x25 Cable 20 0.599 0.724 50 Al70 OHL 1 0.27 0.27 50 Al25 OHL 1 0.27 0.27 50 NAYYJ4x150SE Cable 1 0.275 0.246 50 NYY4x10RE Cable 1 0.275 0.246 50

Name R' (20°C) C' L' R0' C0' L0'

Ω/km uF/km mH/km Ω/km uF/km mH/km Al150 0.192 0.01 1.1364 0.336 0.0043 5.5582 Al_St35_6 0.835 0.0091 1.2796 0.979 0.0042 5.8065 N2XS2Y3x1x300 0.059 0.33 0.3469 0.698 0.3374 0.9199 NAEKBA3x25 1.2 0.24 0.4902 2.061 0.2389 0.2674 Al70 0.436 0.012 0.9836 0.872 0.006 2.9507 Al25 1.18 0.01 1.0854 2.36 0.005 3.2563 NAYYJ4x150SE 0.206 0.83 0.2547 0.824 0.3859 1.024 NYY4x10RE 1.83 0.24 0.2992 2.69 0.13 6.0511

Table 5.3: Parameters of Overhead line/cable used by EnBW in German utilities

Fig 5.5: T section model of distribution network


70

5.3 Modeling and simulation of OLTC topologies In this section the OLTC topologies described in chapter 4 are modeled in PLECS and simulated. The blockset version of the simulation package PLECS will be used to simulate topologies 3 to 7. The models are made for a three phase four wire network assuming Y connected OLTC units with star pointed connected to line neutral. The simulations are made in pu domain and the line is modeled using the overhead line type Al150 with a length of 10km. The main aim of the simulation is to verify the different ratings for the switches and transformers for the different topologies as estimated theoretically in chapter 4. Summary of the assumptions made in the simulation model:

• The transformer and switches are considered to be ideal. The aim is to first observe the steady state behavior of the topologies and verify the switch and transformer ratings that have been calculated by hand in the previous chapter.

• The switches are lossless and can conduct bidirectional currents when ON. During OFF state they can block bidirectional voltages. The transformers are lossless and have an ideal core with unity coupling between the windings.

• The three phase network is assumed to have a neutral. • The load is assumed to be a three phase unity pf resistive load

I. Topology 1, 1a and 2

Fig 5.6a: Single phase PLECS model of Topology 1 and 1a


71

Fig 5.6b: Single phase PLECS model of Topology 2

Topologies 1,1a and 2 require a costly and bulky 1.1pu full transformer for their operation. Since the aim is to build a low cost, partially rated and feeder specific OLTC, topologies 1,1a and 2 are not carried forward for the next stage of design. They are modeled for a simple single phase system to verify the switch ratings and shown in Fig.5.6a,b. They are included in the simulation nevertheless for providing a quantitative and qualitative comparison of different designs.

II. Topology 3 and 3a Simulation model of topology 3 and 3a are shown in Fig.5.7a,b,c. Three units regulate the voltage of each phase. Salient observations from simulation:

• Load current directly flows through the tap switches and hence vulnerable to damage during faults on the line. The switches must be opened/OFF in such a situation and would have to block voltages in the range of 1 to 1.1pu. Separate protection mechanism required to carry/break fault current if such a situation arises

• The tap switches experience both positive and negative voltages and currents during operation. So they should be able to block and conduct in both directions. The selector switch in reality will be composed of two separate switches.

• Power handled by the autotransformer is 0.1pu and no series transformer is present to provide isolation.


72

Fig 5.7a: PLECS model of Topology 3 & 3a connected to a three phase distribution network

Fig 5.7b: PLECS model of Topology 3

Va Va’

Vla


73

Fig 5.7c: PLECS model of Topology 3a

The voltage waveforms at the input (green) and output (red) of the OLTC and at the load (blue) as indicated in Fig.5.7a are shown in Fig.5.7d for phase a. At the input of the OLTC is the rated 1pu rms phase voltage Va of the grid (1.414pu peak, green). The OLTC units are connected between the phase and neutral and set for providing 8% positive compensation. Thus the output voltage Va’ is at 1.08pu rms (1.53pu peak, red). It can be observed that the compensation is in phase so the peaks of the voltages at input and output of OLTC align themselves. This voltage of 1.08pu is at the head of the three phase transmission line and the load is connected at its end. There is a finite voltage drop along the line depending on the magnitude of load current and line impedance. The voltage at the load end Vla is hence lowered and is at 0.98pu rms (1.39pu peak, blue). Thus by using the compensator the voltage drop along the feeder is compensated and the load voltage is within 2% of nominal voltage of the grid. Similar waveforms at the input of the OLTC and at the load can be observed for the phases b and c in Fig.5.7d. During reverse current flow due to high PV injection, the selector switch can change its position to provide negative compensation so that overvoltage at the load end is hence prevented.


74

Fig 5.7d: Voltage waveforms of OLTC

Legend: Va Vla Va’ Vb Vlb Vc Vlc


75

III. Topology 4 and 4a Simulation model of topology 4 and 4a are shown in Fig.5.8a,b. Salient observations from simulation:

• Bottom set of switches S1 to S5 shown in Fig.5.8b are used for positive compensation and current drawn by the autotransformer from source is positive. For negative compensation, switches S6 to S10 are used and autotransformer current is negative. Power handled by the transformers is 0.1pu

• Load current does not directly flow through the tap switches and hence tap switches protected from the line fault currents though the 1:1 isolation transformer.

• When the OLTC is in OFF condition and not feeding any compensating voltage but is connected to the network, then the series transformer behaves in a reverse fashion and feeds voltage from the grid and into the OLTC. The switches observe a voltage in the range of 0.9 to 1.1 pu which would damage them permanently. So switches have to be oversized to handle such a high voltage across them.



76

Fig 5.8b: PLECS model of OLTC based on Topology 4 (left) & 4a (right)


77

IV. Topology 5 and 5a Simulation model of topology 5 and 5a are shown in Fig.5.9a,b,c. Salient observations are:

• Reduction of the number of switches and switch ratings by half makes topologies 5 and 5a, a better design than 4 and 4a.

• During fault conditions or in OFF condition of the OLTC, the selector switch should be opened and one of taps should be kept closed. Such a control strategy will optimize the voltage ratings of the tap switches.



78

Fig 5.9b: PLECS model of OLTC based on Topology 5

Fig 5.9c: PLECS model of OLTC based on Topology 5a


79

V. Topology 6 Simulation model of topology 6 are shown in Fig.5.10a,b. Salient observations from simulation:

• The tap switch ratings depend on the lowest compensating voltage that will be fed by the OLTC. In the current case, it is 2% and corresponding switch ratings are upto 6pu. If the lowest compensation is reduced to 1%, the required switch rating will go upto 11pu. This is a major hindrance in the flexibility of this design.

• The forward and reverse blocking ratings of individual switches are not equally spread out in steps of 0.02. This is unlike the other topologies.

• When OLTC is in OFF condition, then the transformer acts in reverse fashion and the voltage on the switch side of the series transformer is 100pu and this will burn the transformer windings and the switches. So the topology must always feed voltage to the grid and never be connected in OFF condition without additional isolation and protective equipment.

Fig 5.10a: PLECS model of Topology 6 connected to a three phase distribution network


80


VI. Topology 7

Simulation model of topology 7 are shown in Fig.5.11a,b. Salient observations from simulation:

• The flux in the main transformer gets completely reversed when sign of compensation changes through the selector switch

• Load current directly flows through the tap switches and hence vulnerable to damage during faults on the line. The switches must be opened/OFF in such a situation and would have to block voltages in the range of 0.9 to 1pu. Separate protection mechanism required to carry/break fault current if such a situation arises

• In comparison topology 3, which also has only a single transformer and switches directly carrying load current, 3 has the advantage that the transformer has to be rated only for half the voltage on its primary which reduces the size and cost.


81

Fig 5.11a: PLECS model of Topology 7 connected to a three phase distribution network



82

5.4 Topology selection For the further design of the voltage compensator, the 11 topologies are analyzed to shortlist a few topologies. From the perspective of requirements of the characteristics of the compensator, the following criteria are used for shortlisting topologies:

1. The aim is to build partially rated feeder specific compensator. Hence the grid and load need not be isolated for such an application and transformer need not be rated for 1pu. Topologies 1,1a and 2 require a costly and bulky 1.1pu transformer providing grid and load isolation. So these topologies are not considered.

2. An autotransformer is a more cost effective option saving both material and space when compared to a conventional two winding transformer.

3. Blocking voltage requirements of the switches determines how many switches have to be put in series and overall cost. Lower the blocking voltage, lesser number of switches and lower cost. Topology 6 poses a major disadvantage in this regard as it requires switches with voltage rating of 6pu. So topology 6 is not considered.

4. The topology using (N+1) switches will cost lesser than the topology using (2N+1) switches. However the topologies with (N+1) switches also require an additional selector switch whose design and ratings must be considered. Thus a combination of the two decides which topology is more cost effective.

5. Voltage and current ratings of the transformer decided their size and cost. When design 3 and 7 are compared even though both require same rating and number of switches, the transformer voltage ratings is twice in topology 7. So design 3 is preferred over 7. Further the conventional two winding transformer design in 7 requires more copper than autotransformer.

6. A series transformer is a requirement as it provides isolation for the switches of the tap changers. This becomes important from a protection perspective when using electronic/hybrid switches as they do not have high overload capability. During a short circuit in the load, the series transformer can be designed so that it gets saturated and prevents a large current from flowing through the switches. However on the other hand, for topologies without a series transformer, there is significant savings on cost of the series transformer and this can be used in alternate methods to protect against short circuit. E.g. Design 3 saves on cost and component as it does not require an additional series transformer; however the switches need to be oversized for protecting them against faults. The cost comparison of the two strategies will drive a more effective topological chose. From the above discussion topology 3 & 3a, topology 5 and topology 4 are most suitable for the design of OLTC.


83

5.5 Summary and conclusion The key agenda of this chapter was to examine the three phase connection schemes for single phase OLTC to three phase network and to simulate the various topologies of chapter 4 and shortlist the most appropriate designs. With unbalanced loading and the need to regulate individual phases, the wye connection of OLTC transformers with the start point connected to the line neutral is the most suitable connection scheme for the LV network. On the other hand the MV network doesn’t have an issue with unbalanced loading and the necessity is to regulate the line-line voltages. The open delta connection is hence the most suitable choice requiring only two OLTC units. The simulation of the various topologies in PLECS verified their operation and the ratings of the components as was theoretically estimated in chapter 4. Based on a number of criteria such as ratings of switches, sizing of transformer and need for isolation, topologies 3, 3a ,4 and 5 were shortlisted as suitable topology designs.


84

Chapter 6: Optimization of design of OLTC and Hybrid switch

85

Chapter 6

Optimization of design of OLTC and Hybrid switch

The aim of this chapter is to optimize the design of the switches that make the taps of the OLTC transformer to reduce the overall cost and realize a robust tap changing mechanism for the OLTC. In the first part, different types of electronic, mechanical and hybrid switches that have been used in past for OLTC will be looked at. In the second part, a novel method of switch realization using a combination of no-load switches and a single hybrid switch will be proposed which will significantly reduce both the cost of the OLTC and ratings of semiconductor components. In the third section, different types of bidirectional switches that can be used in the hybrid switch will be investigated and their corresponding commutation strategies will be analyzed. It will be reasoned that a series connection of two back to back IGBT with anti-parallel diodes is a suitable bidirectional electronic switch for the OLTC. It can do a ‘make and break’ tap change without the occurrence of a short/open circuit using a 4-step commutation strategy. 6.1 Electronic, mechanical and hybrid switches for OLTC taps In the previous chapters, different single phase topologies for OLTC were proposed where the switches were assumed to be ideal. In this section, the motive will be to look into how these switches can be realized in practice and what are the pros and cons of the different technologies used. Traditionally voltage regulators made use of mechanical switches for realizing the taps [30, 32, 36]. The main advantage of these mechanical taps is their high overload capacity and very low on-state resistance resulting in low steady state losses. The main issue with these switches is twofold – one is the occurrence of an arc when the mechanical switch opens and


86

the current is interrupted. This arcing phenomenon reduces the lifetime of the switches and requires regular maintenance operation to keep the regulator working. This problem has been further aggravated in recent times due to large scale integration of DG in the distribution network. Frequent fluctuations in power and voltage resulting from DG power injection leads to frequent tap changes and quicker deteriorations of the mechanical taps. The second problem is the requirement for a tap changer to operate in a ‘make and break’ fashion as discussed in the introduction of chapter 4. This results in a short circuit period between the taps which necessitates the placement of an inductor/resistor between the taps during tap change to limit the short circuit current. These components are lossy and bulky. If a resistor is used it cannot be retained during steady state as it leads to high losses; therefore a mechanism to remove the resistor has to be realized as well! The OLTC taps can also be realized using electronic switches such as back to back thyristors [35] or two series connected IGBT with anti-parallel diodes [7,8]. The chief advantage of electronic switches over mechanical switches is the absence of any arcing during the switching operation leading to long lifetime and maintenance free operation. They are very fast and flexible in operation. A full electronic tap changer can be used as a custom power control device [8, 42]. At the same time, electronic switches suffer from higher cost and higher on-state voltage drop giving rise to larger amount of steady state losses. The cost of the switches are largely determined by the number of taps and the fault conditions of the system, as the semiconductor switches have low overload capacity. A comparative overview of mechanical and semiconductor taps is presented in Table 6.1.

Feature Mechanical switch Semiconductor switch Switching mechanism Metallic contact and arc

quenching PN Junction and gate activation

Response time Order of ms Order of µs

Lifetime of switch Limited due to contact erosion Theoretically unlimited

Conduction losses Very small Relatively higher Galvanic isolation Yes No Overload capacity Very high Limited Switching frequency Low Very high Overvoltage and overcurrent protection Not necessary Required (E.g. Snubber circuit)

Size and volume Relatively big depending on interrupting current & voltage

Relatively compact depending on cooling necessities

Maintenance Necessary Not necessary Cost Relatively low Relatively high

Table 6.1 – Comparison of characteristics of mechanical and electronic switches


87

By combining both mechanical and electronic switches, hybrid switches are obtained that can be used to realize transformer taps, shown in Fig 6.1. The basic idea of hybrid switches is to use the mechanical switches in steady state and electronic switches during tap change. The use of mechanical switches results in low losses in steady state while the electronic switches assist in the tap changing process, to prevent the arcing problem. They thus combine the advantages of both types of switches but the drawback being that they are much more costly owing to the increased number of components. Many novel ideas have been proposed using hybrid switches [37-40]. Since the major problems in the present day tap changers is lifetime reduction due to the increased arcing in mechanical switches; the solution for the future tap changers will be through the use of hybrid tap changers which have maintenance free operation as well as low steady state losses. The aim of the thesis chapter would be to make a novel proposal for a hybrid switch which can be incorporated in the topologies that were shortlisted in chapter 5. For the hybrid switch shown in Fig 6.1, the following properties can be listed:

1. During steady state, the mechanical switch will carry the load current. 2. The electronic switch across which the conducting mechanical switch is connected

will have zero voltage across it (Electronic1) while the other switch will block the voltage across the taps (Vtap1-Vtap2)

3. The electronic switches will have to be of bidirectional nature and will solely be used during tap change operation

`

Fig 6.1 – Schematic of a simple hybrid tap changer


88

6.2 Novel design of OLTC using hybrid switch 6.2.1 Main issues in realizing hybrid switches for OLTC topologies For the six shortlisted topologies of chapter 5, the required ratings for the switches are upto 0.2pu voltage and 1pu current (Refer Table 4.8a and 4.8b). The switches have to be bidirectional and upto 11 such switches will be required depending on the topology, if ±10% compensation is done in taps of 2% each. For the MV scenario, this translates to tap switch rating of 4kV rms and 577A rms. If this switch has to be realized using an hybrid switch, they require bidirectional electronic switch of such high rating besides a mechanical switch. Semiconductor switches of 4kV ratings are extremely expensive and series connection of smaller rated switches will be costly as well besides the complexity involved in their operation. Further requiring 11 switches for the entire OLTC will make the overall cost of the OLTC very high. Hence it is required that the design of the OLTC be optimized. The major goals of the optimization will be to:

1. Reduce the number of active hybrid switches 2. Reduce the voltage and current ratings of switches 3. Realize short circuit free ‘make and break’ commutation during tap changes

6.2.2 Concept of no-load switches

A no-load switch (NL) is a mechanical switch that opens or closes under no-load. In other words, it is operated in such a way that it doesn’t have to create or break a current and hence has no arcing problems! Such an operation is realized by placing an electronic switch in series with the no-load switch as shown in Fig. 6.2. The current creation or interruption process is done in two steps using the electronic as stated in Table 6.2. In simple words, change of state of the no-load switch must always be done with the series electronic switch in OFF state. During turn ON, the no-load switch is first tuned ON and then the electronic switch. Also during turn OFF, the electronic switch is first turned OFF and then the no-load switch. As can be seen in the table, the two step process cleverly ensures that the current creation or interruption is solely done by the electronic switch (BS) and hence protects the mechanical switch from interrupting a current and experiencing an arc.

Fig 6.2 – Schematic of no-load switch in series with an electronic switch


89

Initial state Step 1 Step 2 NL switch BS switch NL switch BS switch NL switch BS switch

Create a current OFF OFF ON OFF ON ON Break a current ON ON ON OFF OFF OFF

Table 6.2 – Two step method to create/interrupt current through NL switch

The major advantage in using no-load switches for a tap changer is their low cost resulting from simple operation under no-load and low steady state losses (neglecting the cost and losses in the electronic switch). In the next section it shall be shown on how a number of no-load switches can be smartly connected to a tap changer and use just two electronic switches to realize their operation. 6.2.3 Diverter switch type voltage regulators Fig 6.3a shows a class of voltage regulators referred to as ‘diverter switch’ type voltage regulators [30, 32, 33]. It makes use of two movable no-load switches referred to as ‘selector switch’ which are used to select the taps. A mechanical ‘diverter’ switch is used for the tap changing process and for carrying the load current in steady state. During normal operation, the tap selector switch is connected to a specified tap and the diverter switch is moved in position to supply the load power. To make a tap change, the following steps are followed:

0. Initially the diverter switch is in position 1 and tap-selector1 supplies the load. 1. To make a tap change, tap-selector2 moves to the tap to which a change is required. 2. The diverter switch moves from position 1 to 2 resulting in a short circuit of the two

taps; the short circuit current being limited by the leakage impedance of the taps and by the two resistors.

3. Diverter switch reaches position 2 and tap-selector2 supplies the load An alternative design of the diverter switch type regulators is shown in Fig 6.3b [32]. Here the diverter switch is realized using two separate mechanical/electronic switches and an inductor is used to limit the short circuit current. The inductor being a lossless element can be used in steady state as well when selectro1 and selector2 are in two different taps – it will result in the load voltage being equal to the average of the two tap voltages. The advantage of this design is the use of no-load switches and the minimum requirement for active switches that make/break currents. An interesting contrast can be observed in the two diverter switch layout – in Fig 6.3a the tap selectors can individually only connect to odd taps or even taps but not both. While in Fig 6.3b, the selector taps can connect to any tap of the transformer and by using the inductor, the load voltage can be an average voltage of any possible tap combinations.


90

Fig 6.3a – Diverter switch type voltage regulator using resistors [33]

Fig 6.3b – Diverter switch type voltage regulator using inductors [32]

6.2.4 Design of OLTC using no-load switches and single hybrid switch A novel design of the tap changer based on the ‘diverter’ switch type regulators is being proposed in this section which satisfies the goals of the optimization listed in section 6.2.1. Fig 6.4 shows the proposed realization of topology 3a where a combination of no-load switches and a single hybrid switch is used to realize OLTC. Each tap of the transformer is connected to a no-load switch and alternate no-load switches are connected to each other (marked in red and green). If it is assumed that there are ten taps of 2% compensation each,


91

Fig 6.4 – Novel realization of topology 3a using no-load and hybrid switch

then 5 alternate taps are connected together. The two groups of five taps (green and red) are in turn connected to a single hybrid switch as shown in Fig 6.4. No inductors or resistors are used in the design to limit the short circuit current during tap change because of a smart commutation strategy of electronic switches that will be discussed in section 6.4. During steady state operation the switches will be connected as shown in Fig 6.5a and Fig 6.5b for the case of green-tap and red-tap supplying the load respectively. The yellow arrow marks the current path through the switches. As can be seen the mechanical switch M conducts the load current in steady state and this ensures low steady state losses. The bidirectional electronic switches BS1 and BS2 which are part of the hybrid switch are used for the tap changing process. The impedance Zline marked in the figure indicates the impedance of the line ahead of the OLTC and Load indicated in the figure can be a passive load drawing power or an active load injecting power in reverse fashion.

Fig 6.5a – Steady state operation of proposed topology when any green-tap is ON


92

Fig 6.5b – Steady state operation of proposed topology when any red-tap is ON

Certain conditions are imposed during the operation of the tap changer to ensure safe operation and optimal switch ratings. These are listed below:

1. One of the transformer taps will remain in closed condition at all times. This will ensure uninterrupted power flow and that the switches will not have to block voltages in the range of 0.9pu to 1.1pu as indicated in Table 4.8b for the condition of all taps open.

2. At any point of time only one no-load switch amongst green or red will be closed. This is to prevent the occurrence of a short circuit between the taps

3. Tap changes are always made in steps of one. This means that if tap 2 is ON, then a tap change can be made only to tap 3 or tap1. This has two advantages:

a. It ensures that tap change are always made from a green-tap to a red-tap or vice versa. Tap changes between two red taps or two green-taps is anyway impossible to be realized using this design as highlighted in the first point.

b. The maximum voltage across the hybrid switch will be equal to the voltage of one tap i.e. 0.02pu. Thus the switches have to be rated only for 0.02pu which is a reduction by 90% when compared to 0.2pu, which was initially the switch voltage rating for topology 3a; see Table 6.3.

Topology Number of

switches Active switch rating (p.u.)

No-load switch rating (p.u.)

No Load Active I Vb I Vb 3a - 11 1 0.2 1 0.2

New 3a 11 2 1 0.02 1 0.2

Table 6.3 – Reduction in number of switches and voltage ratings using proposed design


93

6.2.5 7-step tap changing methodology The tap changing process from one tap to another can be carried out through a 7-step procedure. Through this method, the voltage regulator moves from the state described in Fig 6.5a to that in Fig 6.5b or vice versa. The 7 steps are elaborated using Fig 6.6.1 to Fig 6.6.8. Step 0: • In the initial state of tap changer, one of the no-load switches NL1 is ON and the

mechanical switch M of the hybrid switch conducts the load current. • The bidirectional electronic switch BS1 and BS2 are OFF. BS1 has zero voltage across it

owing to closed state of M. Zero voltage is present across BS2 as well, since all the red no-load switches are open and they block the tap voltages.

Fig 6.6.1 – Step0 of the tap changing process

Step 1: • In step1, the no-load switch corresponding to the tap to which a change is required is

closed i.e. NL2 is closed. Mechanical switch M of the hybrid switch continues to conduct the load current.

• BS1 and BS2 are OFF. BS1 has zero voltage across owing to closed state of M. However owing to the closed state of NL2, BS2 now blocks the voltage of one tap winding i.e. VtapNL2 -VtapNL1=0.02pu



94

Step 2:

• In step2, BS1 is switched ON and load current is shared between the BS1 and M • BS2 is OFF and blocks the voltage of one tap winding i.e. VtapNL2 -VtapNL1


Step 3:

• In step3, M is switched OFF with no subsequent arcing phenomenon owning to the presence of load current path through BS1

• BS2 is OFF and blocks the voltage of one tap winding i.e. VtapNL2 -VtapNL1



95

Step 4:

• In step4, through a ‘make and break’ mechanism commutation of load current from BS1 to BS2 is carried out. A number of methods to do the current commutation between electronic switches BS1 and BS2 exist, based on the type of switch used. This is analyzed in depth in section 6.3

• BS1 is OFF at end of step 4 and blocks the voltage of one tap winding i.e. VtapNL1 -VtapNL2. This voltage is opposite in polarity to what BS2 was blocking in step 2 and 3.


Step 5:

• In step5, mechanical switch M is switched ON and load current is shared between the BS2 and M

• BS1 is OFF and blocks the voltage of one tap winding



96

Step 6:

• In step6, BS2 is switched OFF and M carries the full load current • BS1 is OFF. BS2 has zero voltage across owing to closed state of M. However owing to

the closed state of NL1, BS1 blocks the voltage of one tap winding.


Step 7:

• In the last step7, NL1 is open and M conducts the load current through NL2. This is done to facilitate a future tap change to say NL3

• BS1 and BS2 are OFF. BS2 has zero voltage across owing to closed state of M. Zero voltage is present across BS1 as well, since all the green no-load switches are open and they block the tap voltages.



97

6.2.6 Advantages of the proposed design The advantages of using the proposed topology are:

1. Reduction in number of active hybrid switches from eleven to just one 2. Reduction in the voltage ratings of the switch by 90% from 0.2pu to 0.02pu 3. Low cost – replacement of 11 hybrid switches of 0.2pu voltage rating with a combination

of 11 no-load switch and a single hybrid of 0.02pu rating reduces cost 4. Steady state losses are lowered due to use of hybrid switch 5. Long and maintenance free operation due to absence of arcing problem 6. Modular nature of design – By increasing the number of taps either the total regulation

range can be increased to 15% or even 20%; or finer regulation can be obtained by reducing the voltage per tap to 1.5% or 1%. This can be realized by simply adding more no-load switches and still using only a single hybrid switch. It should be realized that the cost of no-load switches does not increase with increased voltage ratings unlike semiconductor switches. Increased voltage ratings only requires increased air gap between the ends of the no-load mechanical switch in OFF state to ensure breakdown-free air gap in OFF state.

The major disadvantages of the proposed design are:

1. If a tap change of upto 4 taps is required, it can only be carried out in steps of 1. Thus tap jumping is not possible.

2. Additional sensors required for current and voltage polarity detection (see section 6.4) 3. Complexity of control 6.2.7 Novel design for topology 4 and short listing of topologies

The novel design described for topology 3a however cannot be implemented for the other topologies namely 3, 5 and 5a but only for topology 4 and 4a. This is due to presence of a selector switch in 3, 5 and 5a. The realization of the selector switch requires an additional hybrid switch of 0.1pu rating which is 5 times the rating of the existing hybrid switch. Compared to this, it is more cost effective to use design 3a or 4 where only one hybrid switch of 0.02pu will be required and rest are low cost no-load switches. The proposed design which has been described above for topology 3a is now implemented using topology 4 and shown in Fig 6.7. When comparing the two designs of topology 3a and 4, it is observed that the additional cost of the series transformer in topology 4 poses a major drawback and offsets the cost savings obtained through the proposed design. However the series transformer also provides isolation for the switches from grid faults which can be quite


98

advantageous from a practical perspective. A more in-depth look into the fault modes and protection of the OLTC will be analyzed in chapter 8.

Fig 6.7 – Novel realization of topology 4 using no-load and hybrid switch

6.3 Bidirectional electronic switch for hybrid tap changer 6.3.1 Different types of bidirectional electronic switch In this section, the aim is to choose a suitable bidirectional electronic switch that can be used in the hybrid switch. A bidirectional switch is one that can block both positive and negative voltages in OFF condition and can conduct current of both polarities when ON. It should be such that both direction of current can be individually controlled. A number of semiconductor components can be combined together to create a bidirectional switch as shown in Fig 6.8. The switch should be such that the load current can commutate from one switch to another through the ‘make and break’ mechanism without causing a short/open circuit. The first switch indicated in Fig 6.8(1) is a back to back parallel connection of two thyristors. Each of the thyristor can block bidirectional voltage and conduct unidirectional current. When such a switch is used as an electronic tap an additional inductor/resistor is required to limit the short circuit current during tap change [32,35]. The losses in the switch arise from on-state voltage drop across the conducting switch. Further the thyristor cannot be turned OFF through the gate which results in lesser flexibility and longer duration of the tap change process. An additional forced commutation strategy must be implemented to turn it OFF. The second switch is an IGBT connected to a combination of 4 diodes as shown in Fig 6.8(2). A single IGBT is used for both forward and reverse currents making the design cost effective


99

Fig 6.8 – Bidirectional electronic switch for use in hybrid switch

and the diodes provide blocking capability for the forward and reverse voltages. The drawback is that the direction of current cannot be individually controlled. As the case for the back-back thyristor, an additional inductor/resistor is required to limit the short circuit current during ‘make and break’ tap commutation process [43]. Due to the fast switching speed of IGBT the tap commutation time is low. Three devices namely two diodes and the IGBT will conduct the load current at any given point of time. The switch hence has high steady state losses resulting from the losses of three devices. A MOSFET or IGCT can be used in the place of the IGBT and the operation will be the same. The third and fourth switches, Fig 6.8(3) and Fig 6.8(4) are obtained using the series connection of two back-to-back IGBT having an anti-parallel diode [7,8,32]. The current conduction path is through one IGBT and the anti-parallel diode of the other IGBT. Thus distinct current conduction path for both forward and reverse currents exist and they can be controlled through the gates of the two IGBT. Unlike the other switches discussed earlier, no additional inductors/resistors are required for the tap changing process. A 4-step commutation strategy based on current or voltage polarity can be used for the commutation of taps without the occurrence of a short circuit [32, 41, 45-47]. The tap change time is extremely fast in the order of μs. The difference between Fig 6.8(3) and Fig 6.8(4) is that, the former has a common emitter configuration while the latter has a common collector configuration. The common emitter configuration is preferred as it requires a lower number of isolated gate drive circuits. A bidirectional switch with the same properties as the switch in Fig 6.8(3) or Fig 6.8(4) can be realized using a back to back parallel connection of two non-punch through IGBT shown in Fig 6.8(5) [44]. The non-punch through type IGBT has reverse voltage blocking capabilities unlike the usual punch through type IGBT used in Fig 6.8(3) and Fig 6.8(4). 4-step commutation can be implemented here as well for the tap changing process. The major drawback is that these switches are not widely available in the market, making them less cost competitive.


100

6.3.2 Choice of bidirectional electronic switch for MV and LV To make a choice for the electronic switch, it should meet the following criteria:

1. Low cost & high reliability 2. Minimum number of components (switch/diode) 3. Short commutation time for quick tap changes 4. No requirement for inductors/resistors to limit the short circuit during tap change

Based on the above criteria the preferred choice of switch is the series connection of two back-to-back IGBT having an anti-parallel diode in the common collector configuration as shown in Fig 6.8(4). For the MV scenario, based on Table 5.1, the switches will require a blocking voltage rating of 400V rms (2% of 20kV line-line voltage assuming open delta connection of two OLTC units) and current rating of 577A rms. Since IGBT in the market that are rated for 400V rms are not usually rated for such high currents of 577A rms, parallel connection of IGBT is suggested. For the LV case the switches have to be rated for 577A rms and a voltage of 4.6V rms (2% of 230V phase voltage assuming a Y connection of 3 OLTC units). IGBT of such low voltages and high currents are virtually non-existent in the market. In such a case, an excellent alternative is to use a back to back series connection of two MOSFET as shown in Fig 6.9. Low-cost MOSFET of low voltage and high current rating upto 150A used for automotive and industrial applications are widely available in the market. The body diode of the MOSFET will serve the purpose of current conduction path and parallel connection of MOSFET can be done to reach higher current requirements. It is important to realize that the electronic switches are used solely during the tap changing process. So the issue of additional losses in MOSFET compared to IGBT does not pose a problem. Further one might pose a question on why MOSFET cannot be used for the MV bidirectional switch – the answer is that high voltage and high current MOSFET of the order of 400V rms and 577A rms are virtually non-existent in the market owing to high switch losses due to device characterization.

Fig 6.9 – Bidirectional electronic switch for use in LV OLTC


101

6.4 4-step bidirectional electronic switch commutation based on voltage and current polarity

The commutation between the two electronic switches occurs in step 4 of the 7-step tap changing methodology described in section 6.2.5. This commutation can be achieved using a 4-step current commutation technique based on current direction or voltage polarity [32, 41, 45-47]. By the use of a more complicated control, the commutation can be shortened to a 2-step commutation which is highly advantageous for high frequency switching applications such as matrix converters [48, 49]. But for the tap changer application, such high operating frequencies are not encountered so there is no inherent advantage of going for a 2 step commutation. Hence the focus here will be on the use of the 4-step commutation strategy for the electronic switch. The basic idea of the 4-step commutation is to ensure that:

1. Short circuit of taps does not occur 2. Open circuit must not occur i.e. No loss of load must occur

6.4.1 4-step commutation based on voltage polarity Fig 6.10a shows the schematic of the IGBT based bidirectional switch connected between two taps of the OLTC. The commutation strategy is based on the polarity of voltage V12 between the taps and independent of the direction of the current Iload [32,41,45,46]. This means that at all points of time, a path for both the forward and reverse current must exist to ensure there is no open circuit. Further to prevent a short circuit when V12>0, S1f and S2r should not be simultaneously ON. Alternatively when V12<0, S2f and S1r should not be simultaneously ON. Keeping these conditions in mind, the switching steps to move from Tap1 to Tap2 are given in Table 6.4 for the case of both V12<0 and V12>0 referred to as Sequence A and B respectively. 1 indicates the switch is ON while 0 is represents OFF condition. The sequence for moving from Tap2 to Tap1 is exactly the reverse of that shown in the table. Fig 6.10b shows the switching sequence for Tap1 to Tap2 when V12>0. It is important to note that at all instants of time, one forward and one reverse conducting switch is always ON!

V12>0 (Sequence A) V12<0 (Sequence B)

S1f S1r S2f S2r Spacinggg S1f S1r S2f S2r Initial state 1 1 0 0 1 1 0 0 Initial state Step 1 1 1 1 0 1 1 0 1 Step 1 Step 2 0 1 1 0 1 0 0 1 Step 2 Step 3 0 1 1 1 1 0 1 1 Step 3 Step 4 0 0 1 1 0 0 1 1 Step 4


102

Table 6.4 – Switch states for Voltage polarity based 4-step commutation

Fig 6.10a – Voltage polarity based 4-step commutation

Fig 6.10b – (Clockwise) Step1 to Step4 for voltage polarity based commutation when V12>0

6.4.2 4-step commutation based on current polarity Fig 6.11a shows the schematic for 4-step commutation based on current polarity of the load current Iload. Knowing the current polarity, it must always be ensured that a switch corresponding to that polarity is always ON to prevent an open circuit [32,47]. The commutation strategy is independent of the polarity of voltage V12 between the taps. Unlike the voltage polarity based method where during step 1 and step 3 three switches were ON, in this method only one switch will be ON during step 1 and step 3.

1 2

4 3


103

Iload>0 (Sequence A) Iload <0 (Sequence B)

S1f S1r S2f S2r Spacinggg S1f S1r S2f S2r

Initial state 1 1 0 0 1 1 0 0 Initial state

Step 1 1 0 0 0 0 1 0 0 Step 1

Step 2 1 0 1 0 0 1 0 1 Step 2

Step 3 0 0 1 0 0 0 0 1 Step 3

Step 4 0 0 1 1 0 0 1 1 Step 4

Table 6.5 – Switch states for current polarity based 4-step commutation

Fig 6.11a – Current polarity based 4-step commutation

Fig 6.11b – (Clockwise) Step1 to Step4 for current polarity based commutation when Iload>0

1 2

4 3


104

The switching steps to move from Tap1 to Tap2 are given in Table 6.5 for the case of both Iload>0 and Iload<0 referred to as Sequence A and B respectively. The sequence for moving from Tap2 to Tap1 is exactly the reverse of that shown in the table. Fig 6.11b shows the switching sequence for Tap1 to Tap2 when Iload>0. Just like in voltage polarity method, there is a short time lag in the order of µs between each step. The total commutation can be concluded within 10 µs which is very fast. A current direction detectotr for the judgment of the current polarity is required for the operation. Alternatively the collector-emitter voltage of the IGBT (or gate source voltage of MOSFET) can be measured to determine which switch is ON and hence the current direction. A failure to judge the right current polarity will result in an open circuit and a loss of load current. This calls for additional overvoltage protection for the switches. 6.4.3 Choice of commutation strategy To make a comparison and choose one method of the voltage or current polarity based commutation, it is important to analyze the severity of the hazard if a failure in judging the current/voltage polarity were to occur. In case of the voltage polarity based method, the wrong judgment of the voltage will result in a short circuit between the taps. Such a mistake can occur around the zero crossing of the voltage where an erroneous judgment will fortunately only short circuit the taps at small voltages! The corresponding short circuit current will be small, limited by the tap leakage impedance - hence no damage to the switches will occur. Further, the OLTC will usually have a voltage sensor to determine the grid voltage and the magnitude of compensation required. Similarly voltagesign detection circuitry can be used for judging the polarity as well. On the other hand, a wrong judgment of the current in the current commutation method will result in an open circuit between the taps and a subsequent overvoltage will be observed due to the loss of load. This overvoltage can be given by L*(di/dt) where L corresponds to the combined inductance of the tap leakage, feeder and load inductance and the (di/dt)corresponds to the slope of the current turn off transient. Such an erroneous judgment of current polarity will occur during the zero crossing of the current where (di/dt) can have a large value even if the current is close to zero. Since L can also be sufficiently high, large overvoltage will be observed by the switch which can permanently damage it. This calls for additional overvoltage protection for the switches. Further, this method will require an additional current direction detector of high reliability besides the voltage sensor thus adding to the cost of the system. From the above comparison based on failure hazard and number of components required, it can be concluded that 4-step voltage based commutation is the suitable choice for commutation between the electronic switches!


105

During a tap change through the bidirectional switches, 8 different tap change sequences for voltage polarity based commutation can be envisaged. This is shown in Table 6.6. Three parameters determine which tap change sequence from 1 to 8 will be used. These three parameters are mentioned below:

1. The initial tap position is odd or even positioned tap (i.e. Green-tap or red-tap) 2. Tap change to higher or lower tap is required 3. Tap change is required to be done in positive or negative voltage polarity

Tap change Sequence

Initial tap Tap change

to Voltage polarity

Firing sequence

1 Odd

(Green-tap)

Higher tap Positive Sequence A 2 Higher tap Negative Sequence B 3 Lower tap Positive Sequence B 4 Lower tap Negative Sequence A 5

Even (Red-tap)

Higher tap Positive Sequence A 6 Higher tap Negative Sequence B 7 Lower tap Positive Sequence B 8 Lower tap Negative Sequence A

Table 6.6 – 8 different firing sequence for the gating circuit of the bidirectional switch 6.5 Overvoltage snubber and Effect of tap leakage inductance

In the analysis till now the effect of the tap leakage inductance in the operation of the OLTC has not been addressed. Normally during a change of tap, the current through the tap leakage inductance is interrupted, leading to an overvoltage [32]. Fig 6.12 can be used to analyze this effect where the schematic of a single tap of an OLTC is shown. The effect of the capacitance between the taps is neglected for simplification of the analysis. In the figure, Lleak is the leakage inductance between the taps, Rleak is resistance of the tap winding, BS1 and BS2 are the two bidirectional electronic switches and Iload is the load current represented as a current source. Initially let BS1 be ON and the load voltage and voltage at point C is hence (V1+Vtap). Now a tap change is made from BS1 to BS2 through the ‘make and break’ mechanism so that the load voltage and voltage at point C is V1. During this process, let BS1 interrupt the load current with a slope α=(di/dt) as the current commutates to switch BS2, then a overvoltage is experienced by the turning-off switch BS1 given by

𝑉𝐵𝑆1 = 𝑉𝑡𝑎𝑝 + 𝐿𝑙𝑒𝑎𝑘 𝑑𝑖𝑑𝑡 (Eq 6.1)


106

Thus a large overvoltage will be experienced by the switch depending on the leakage inductance of the tap and the current being turned off. Usually a damped LC oscillation will be observed due to the small but finite value of tap capacitance and a damping due to the resistance of the tap windings. This overvoltage can very high and can permanently damage the switch. This calls for the need for using an overvoltage snubber to protect the switches.

Fig 6.12– Effect of tap leakage inductance of tap commutation

6.5.1 Single overvoltage snubber connected between BS1 and BS2 To reduce the overvoltage due to current interruption in the leakage impedance, a RC snubber can be used across the switches as shown in Fig 6.13a [32]. Rsnub and Csnub are the resistance and capacitance of the snubber. The operation of this snubber is analyzed in a step by step manner from the state where switch BS2 is supplying the load to the state where BS1 supplies the load after a tap change process. The current direction and voltage between the taps Vtap is assumed to be positive and voltage polarity based commutation is used. a. BS2 is ON (Forward and Reverse) and BS1 is OFF [Steady state]

• BS2 carries the load current Iload • Load voltage = V1 • The RC snubber is connected across switch BS1 as nodes B and C are equipotential • Voltage across the snubber Vsnub is approximately equal to Vtap neglecting the voltage

drop across the leakage impedance of the tap

b. BS2 is ON (Forward) and BS1 is ON (Forward and Reverse) [Step3 of 4-step comm.]

• During this step, BS2 continues to supply the load • Load voltage = V1


107

Fig 6.13a – RC Snubber connected across the electronic switch

• The RC snubber connected across the BS1 discharges due to turning ON of BS1 and the

discharge current Idis flows as shown in Fig 6.13b through BS1 and BS2. The magnitude of Idis depends on the voltage at which commutation occurs and on the value of Rsnub

𝐼𝑑𝑖𝑠 = 𝑉𝑠𝑛𝑢𝑏|𝑠𝑡𝑒𝑝3

𝑅𝑠𝑛𝑢𝑏 (Eq 6.2)

c. BS2 is OFF and BS1 is ON (Forward and Reverse) [Step4 of 4-step commutation]

• During this step current commutates from BS2 to BS1 and BS1 supplies the load

• Load voltage = V1+ Vtap

• The current injected into the leakage inductance due to the turning ON of BS1 and the current flows through the snubber capacitance resulting in damped LC oscillation indicated by current Iosc as shown in Fig 6.13b. The current through the inductor and the voltage across the capacitor oscillate at high frequency. If the damping due to the resistances in the loop is neglected, the worst case overvoltage can be given by equating the energy stored in the inductor to the energy of capacitor. Higher the value of Csnub and lower the value of Lleak, lower is the overvoltage on the switch.

𝐼𝑙𝑜𝑎𝑑|𝑠𝑡𝑒𝑝4 = 𝐼1 12𝐿𝑙𝑒𝑎𝑘(𝐼1)2 = 1

2𝐶𝑠𝑛𝑢𝑏(∆𝑉)2 (Eq 6.3)

∆𝑉 = 𝐿𝑙𝑒𝑎𝑘𝐶𝑠𝑛𝑢𝑏

𝐼1 (Eq 6.4a)

𝑉𝐵𝑆1 = 𝑉𝑡𝑎𝑝|𝑠𝑡𝑒𝑝4 + 𝐿𝑙𝑒𝑎𝑘𝐶𝑠𝑛𝑢𝑏

𝐼1 (Eq 6.4b)


108

• The frequency of the oscillations is given by

𝑓𝑜𝑠𝑐 = 12𝜋𝐿𝑙𝑒𝑎𝑘𝐶𝑠𝑛𝑢𝑏

(Eq 6.5)

• It is important to note that Iosc does not flow through the switches, shown in Fig 6.13b.

• The resistance of the windings and that of the snubber will damp out the oscillations. It should be ensured that the value of Rsnub is optimally chosen. A large value of Rsnub will lead to reduction in the current peak through the inductor while a small value of Rsnub will reduce the voltage peak across the capacitor and switch. The transient response of the system can be obtained by solving the differential equation:

𝑅𝑙𝑒𝑎𝑘 + 𝑅𝑠𝑛𝑢𝑏 = 𝑅𝑠

𝑑2𝑖(𝑡)𝑑𝑡2

+𝑅𝑠𝐿𝑙𝑒𝑎𝑘

𝑑𝑖(𝑡)𝑑𝑡

+1

𝐿𝑙𝑒𝑎𝑘𝐶𝑠𝑛𝑢𝑏𝑖(𝑡) = 0

𝑑2𝑖(𝑡)𝑑𝑡2

+ 2𝛼 𝑑𝑖(𝑡)𝑑𝑡

+ 𝜔02𝑖(𝑡) = 0 (Eq 6.6)

𝛼 = 𝑅𝑠2𝐿𝑙𝑒𝑎𝑘

𝜔0 = 1𝐿𝑙𝑒𝑎𝑘𝐶𝑠𝑛𝑢𝑏

(Eq 6.7)

Fig 6.13b – Overview of currents when RC Snubber connected across BS1 and BS2

Note – Steps 1 and 2 of the commutation are not considered as they are relevant for a negative load current. For the given configuration the snubber will always discharge in step 3 due to the closed loop formed by the two switches.


109

d. BS1 is ON (Forward and Reverse) and BS2 is OFF [Steady state]

• BS1 carries the load current Iload • Load voltage = V1+ Vtap • The RC snubber is connected across the BS2 as node A and C are equipotential • Voltage across the snubber Vsnub is approximately equal to -Vtap neglecting the voltage

drop across the leakage impedance of the tap 6.5.2 Two overvoltage snubber with each across BS1 and BS2 An alternate way of connecting the snubber is to use two snubbers with one across each of the two bidirectional switches as shown in Fig 6.14a. The operation of the snubbers is exactly the same as discussed in section 6.5.1 with respect to the discharging of the snubber and the LC oscillation between the snubber and leakage inductance. The difference comes in the nature of the current path for LC oscillations namely Iosc and for the snubber discharge current Idis as shown in Fig 6.14b for the case of tap change from BS2 to BS1.

Fig 6.14a – RC Snubber connected across the electronic switch

• Initially when BS2 is ON, snubber1 connected across BS1 has a finite voltage across it

approximately equal to that of Vtap. No voltage exists across snubber2 as BS2 in ON • During commutation from BS2 to BS1, snubber1 discharges in step3 but the discharge

current Idis solely flows through BS1 only as shown in Fig 6.14b.


110

Fig 6.14b – Overview of currents when RC Snubber connected across BS1 and BS2

• In step4, when current commutates from BS2 to BS1, snubber2 protects the switch BS2

from overvoltage which results in damped LC oscillations indicated by the Iosc in Fig 6.14b. It can be observed that the oscillating current flows through both the switches BS1 and BS2 which does not occur in for the single snubber design of section 6.5.1.

• All the equations used in section 6.5.1 are valid here equally here as well 6.5.3 Choice of snubber design The single snubber design has two advantages which make it a better choice for the snubber:

1. Simulations in PLECS have shown that for the same values of Rsnub and Csnub, the overvoltage observed on the switches are exactly the same for both types of snubber (See simulation results in chapter 7). Therefore lesser number of components required in the single snubber makes it a suitable choice over the two snubbers.

2. The two designs have a main difference in the nature of current through the switches. The large currents in the LC oscillations Iosc directly flow through the switches in double snubber design while they bypass the switches and only flow through the snubber in the single snubber design. This prevents the overheating/over ratings of the switches for the single snubber design.

Besides the snubber, the overvoltage can be reduced through the control of the turn off time of the IGBT through the gate resistance. Larger turn off time will lead to a lower value of (di/dt) during turn OFF. This in-turn will result in lesser magnitude of overvoltage on the leakage inductance as VL=L*(di/dt)and subsequently on the switch.


111

From Eq.6.4a it can be seen that the current through the switches at the commutation instant has a direct impact on the overvoltage produced.

∆𝑉 = 𝐿𝑙𝑒𝑎𝑘𝐶𝑠𝑛𝑢𝑏

𝐼1 (Eq 6.4a)

When voltage polarity based commutation is used, the controller of the OLTC system should be designed such that at the instant of commutation, the current is minimal. Commutation around current zero must be preferred and commutation near current peaks must hence be avoided. The strategy is very easily implementable for load having a non-unity power factor. For unity power factor loads, the preferred commutation instant would be with 1ms to 4ms after the zero crossing of voltage and current so as to ensure constant voltage polarity and minimal load current at commutation instant.

6.6 Summary and conclusion The chapter examines the realization of the taps of an on-load tap changing transformer through a novel design using a combination of no-load switches and a single hybrid switch. The proposed design reduces the number of active switches required and the blocking rating of the semiconductor switches, resulting in cost savings. Prevention of arcing and reduced steady losses are the vital benefits in using a hybrid switch. A 7-step tap changing methodology has been proposed that provides an effective method of changing from one tap to another. Voltage polarity based 4-step current commutation using back to back IGBT or MOSFETs as bidirectional electronic switches provides a cost effective method of changing between two taps without creating a short circuit or an open circuit. The tap changing process in a transformer results in overvoltage on the hybrid switch due to the interruption of current in the leakage inductance. This calls for the need for an overvoltage snubber to be connected across the switches. Further, through the control of the turn off times of the switches and by performing the commutation close to the zero crossing of the load current, the overvoltage can be further reduced.


112

Chapter 7: Simulation of OLTC system for MV and LV distribution network

113

Chapter 7

Simulation of OLTC system for MV and LV distribution network

In this chapter, the complete OLTC system customized for MV and LV is simulated for a three phase distribution network using PLECS. Different aspects of the design examined in the previous chapters are put together and implemented in the simulation stage. In the first part, the modeling of the autotransformer including the leakage impedance and the magnetizing branch is presented. In the second part, the modeling of the OLTC system in PLECS for the MV and LV scenario is described. In the third part, simulations are carried out to investigate various aspects of the OLTC system such as steady state operation, transient response during tap change and snubber operation. 7.1 Modeling of autotransformer with taps Topology 3a is used as a basis for the modeling of the autotransformer. It is shown in Fig 7.1 where the taps are present on the series winding. The input is connected across the common winding and mid-point of the series winding and the output is taken at the taps on the series winding. The voltages and nature of connection are indicated in Fig 7.2. The important parameters for the transformer are the magnetizing branch and the leakage impedance of the windings. The leakage impedance of the common winding (xl1, r1) and the series winding (xl2, r2) are indicated in Fig 7.2 where xl1, xl2 are the leakage inductances and r1, r2 are the winding resistances. If it is assumed that there are 10 taps on the series winding, each providing 2% compensation, then the leakage impedance in each tap section is 0.1*(xl2, r2). The transformer parameters as obtained from GE are indicated in Table 7.1 [26]. For the LV case, it is assumed that all the leakage impedance of the transformer is present on series winding. This will be a worst case assumption as higher the leakage impedance of the taps,


114

Fig 7.1 – Schematic of a topology 3a larger are the overvoltage during tap change process (Refer Eq.6.4). The short circuit and open circuit losses are assumed to be 270W and 200W respectively and Rm/Xm=10. For the MV scenario it is assumed that the leakage impedance is equally divided between the common and series winding. Since the magnetizing branch values were not available for MV, they have been neglected. Fig 7.3 shows the simulation model of the autotransformer. It is has 11 magnetically coupled windings, the winding on the left corresponds to the common winding rated for 0.9pu of the voltage and the ten windings on the right connected in series represents the series winding with each rated for 0.02pu voltage. The autotransformer is obtained by electrically connecting the windings on the left and right.

Fig 7.2 – Autotransformer of topology 3a during boost and buck operation


115

MV LV Rated input voltage [V] 20000 230 Nominal input current [A] 577 577 Compensation range of OLTC ±10% ±10% Number of taps 10 10 Compensation per tap 2% 2% Resistance of common winding [Ω] 2.34 0 Inductance of common winding [H] 0.033 0 Resistance of tap [Ω] 4.68e-3 1.62e-4 Inductance of tap [H] 6.56e-5 1.1641e-6 Magnetizing resistance [Ω] - 266.7 Magnetizing inductance [H] - 0.085

Table 7.1 – MV and LV transformer parameters

Fig 7.3 – Model of autotransformer corresponding to topology 3a

The ten windings on the right correspond to the ten tap sections and the tap points are indicated by (0.9, 0.92, 0.94 … 1.06, 1.08, 1). (Li, Ri) where i ranges from 1 to 11 corresponds to the leakage impedance of each section of the winding. (Lm, Rm) corresponds to the magnetizing branch. Input indicated by (Ph, N) is connected between one end of the


116

common winding and at the midpoint of the series winding indicated by 1. Output is taken at any of the tap points indicated by (0.9, 0.92, 0.94 … 1.06, 1.08, 1) where the ten no-load switches are connected. 7.2 Simulation model of OLTC system for MV and LV For the simulation of the OLTC system, the PLECS simulation platform is used. The OLTC is fully customized for the application in the three phase MV and LV networks and the system parameters for the two are summarized in Table 7.2. The main difference comes from the nature of three phase connection and the type of electronic switch used. The tap changing mechanism as explained in section 6.2.5 and 6.4 remain the same for both designs.

MV LV Voltage level (line-line) 20 kV 400 V Maximum rated load current 577A 577A Maximum length of line (km) 30 1 Neutral available No Yes Connection scheme Open delta Wye Number of single phase OLTC units required 2 3 OLTC autotransformer transformed power 1.15 MVA 13.3 kVA OLTC autotransformer throughput power 12.65 MVA 146.3 kVA Percentage impedance 2.5% 6% Bidirectional electronic switch IGBT based MOSFET based Reference electronic switch used for model IR GP4066DPbF IR LB8743PbF Line type (Refer Table 5.3) Al150,10Km Al70, 500m

Table 7.2 – System parameters for MV and LV

7.2.1 Simulation model of OLTC system for MV

Fig 7.3a,b,c,d shows the PLECS model for the OLTC system for MV. In Fig 7.3a, the two OLTC units are connected between the lines in an open-delta fashion in a three phase three wire network. The operation of the open-delta connection scheme and the line model are as examined in chapter 5. The Auto1 and Auto3 blocks of Fig 7.3a are indicated in Fig 7.3b where the autotransformer is shown with taps realized through a combination of no-load switches and a single hybrid switch. The snubber can be seen connected across the hybrid switch. The Load block of Fig 7.3a is a delta connected load as shown in Fig 7.3c, whose pf can be adjusted to realize different load combinations. The IGt and IGb blocks correspond to a series connection of two back to back IGBT in a common collector configuration as seen in Fig 7.3d.


117

Fig 7.3a – OLTC system for MV having two units connected in open-delta

Fig 7.3b – OLTC unit with no-load switches and single hybrid switch

Hybrid switch

Vab Vab’

Vload ab

VIGb

VIGt


118

Fig 7.3c – Load block of Fig 7.3a Fig 7.3d – IGt and IGb block of Fig 7.3b

7.2.2 Simulation model of OLTC system for LV Fig 7.4a,b,c,d shows the PLECS model of the OLTC system for LV. The figures are ordered in a fashion similar to that described above for MV. The Load block has variable pf loads connected in a Y fashion. In LV, individual control of each of the three phase voltages is achievable unlike the MV model where control of each line-line voltage can be realized.

Fig 7.4a – OLTC system for LV having three units connected in Wye

Va Va’ Va’

Vload_a


119

Fig 7.4b – OLTC unit with no-load switches and single hybrid switch

Fig 7.4c – Load block of Fig 7.4a Fig 7.4d – MFt and MFb block of Fig 7.4b

7.3 Simulation of OLTC system for steady state and transient operation In this section different test benches are simulated using the PLECS model for MV and LV. The aim of the simulation is to study the steady state and transient operation of the OLTC units in a comprehensive manner. The list of test bench is shown below:

Hybrid switch

VIGb

VIGt


120

1. Steady state operation of OLTC system a. Positive compensation b. Negative compensation

2. Transient operation of OLTC system a. Tap change phenomenon during positive and negative polarity b. Turn off transient of hybrid switch c. Effect of snubber capacitor on tap change phenomenon d. Effect of load current at commutation instant on switch overvoltage

7.3.1 Steady state operation of OLTC system

a. Positive compensation The steady state operation of the OLTC system is studied for positive compensation. With the MV model, the three phase line-line voltages are observed at the input Vxy and output of the OLTC units Vxy’ and at the load Vload_xy, taking into account the drop across the line. The test is performed for different types of loads namely - loads of 0.8 pf leading, 0.8 pf lagging and unity pf load; assuming they are drawing the maximum rated current. The operation is further investigated when the two units of the MV OLTC system are set for the same and for different magnitudes of compensation. Further for the LV model, the phase voltages are analyzed when each of the three units of the LV OLTC are set for different compensation magnitudes. Some of the selected waveforms from the analysis are reported in Fig 7.5a,b,c,d.

Fig7.5a – Line-line voltages [V] at input Vxy & output of OLTC Vxy’ as a function of time [s] when both OLTC units are set for 4% positive compensation for MV scenario

Legend Vab, Vab’ Vbc, Vbc’ Vca, Vca’


121

Fig7.5b – Line-line voltages [V] at input Vxy & output of OLTC Vxy’ as a function of time [s] when each OLTC unit is set for 4% and 8% positive compensation for MV scenario

Fig7.5c – Line-line voltages [V] at input Vab & output of OLTC Vab’ and at load Vload_ab as a function of time[s] for line a-b when the OLTC provides 8% positive compensation for MV

Legend Vab, Vab’ Vload_ab



122

In Fig7.5a the line-line voltages [V] at the input (Vab,Vbc,Vca) and at the output of the MV OLTC (Vab’,Vbc’,Vca’) are plotted as a function of time[s]. Both units of the open-delta connected MV system are set to provide 4% positive compensation. It can be seen that the peaks of the waveforms (Vab’,Vbc’,Vca’) are above those of (Vab,Vbc,Vca). Fig7.5b shows the zoomed in voltage waveforms of line-line voltages[V] as a function of time[s] when the OLTC unit across phase a,b is set at 4% compensation and the unit across b,c is set at 8% compensation. It can thus be verified that the line voltage Vca has a compensation of 6% which is the average of the other two; thus proving the open-delta operation.

Fig7.5d – Phase voltages [V] at input Vx & output of OLTC Vx’ as a function of time[s] when the three OLTC units are set for -2%, 6% and 8% compensation for LV

Legend Va, Va’, Vb,Vb’, Vc, Vc’


123

Fig7.5c depicts the line-line voltage for phase a,b at the point before (Vab) and after (Vab’) the MV OLTC unit is connected and the corresponding load voltage (Vload_ab) at the end of the line. In this case the OLTC unit across phase a,b is set at 8%. The peaks of Vab and Vab coincide indicating the occurrence of in-phase compensation. Vload_ab is phase shifted with respect to Vab’ due to the reactance of the line. For LV case, Fig7.5d depicts the waveforms of phase voltages [V] as a function of time [s] for phase a,b,c. Voltage at input (Va,Vb,Vc) and output of the LB OLTC units (Va’,Vb’,Vc’) are shown. The OLTC units are set for -8%, 2% and 6% compensation for phase a,b,c respectively. b. Negative compensation For the negative compensation, a similar test bench is set as for the case of positive compensation. The main criterion is to provide satisfactory operation when reverse power flow occurs and to operate under conditions of load with various pf. For the MV scenario, Fig7.6a shows the voltage waveforms of line-line voltages [V] for phase a,b,c as a function of time[s] at the input (Vab,Vbc,Vca) and output (Vab’,Vbc’,Vca’) of the OLTC units. The two open-delta connected units are set to provide 4% negative compensation. Fig7.5a and Fig7.6a can be distinctly compared to see that the peaks for the waveforms for (Vab’,Vbc’,Vca’) that were above peaks of (Vab,Vbc,Vca) in Fig7.5a have now moved below in Fig7.6a.

Fig7.6a – Line-line voltages [V] at input Vxy & output Vxy’ of OLTC as a function of time [s] when both units are set for 4% negative compensation for MV scenario



124

In Fig7.6b, the load is an active load modeled in the form of a current source that feeds in power into the line for the LV scenario. There will hence be a voltage gain along the distribution line when looking from the OLTC end to the load. The OLTC connected to phase a is hence set for 8% negative compensation so as to balance this line voltage gain so that the voltage at the load end is within safe limits of the nominal voltage. It can be observed that the voltage at the output of OLTC Va’ is lower than the voltage at input Va, while the voltage at the output is Vload_a is higher than Va and Va’.

Fig7.6b – Phase voltages [V] at input Va & output Va’ of OLTC as a function of time[s] when an active load is connected at the load end Vload_a of the LV OLTC for -8% compensation

7.3.2 Transient operation of OLTC system In this section, the transient operation of the OLTC during a tap change is investigated. In the first part, voltage polarity based 4-step commutation is observed for commutation in positive and negative half of the voltage cycle. As discussed in chapter 6, overvoltage occurs during a tap change owing to interruption of current through the leakage inductance (refer Eq.6.1). It is vital to examine this overvoltage and study the effect of leakage inductance, snubber capacitance and the commutation current on the overvoltage mechanism. This is analyzed in the second and third part. In the fourth section, the LC oscillation through the snubber capacitance and leakage inductance and the corresponding oscillating current and voltages are reported.

Legend Va, Va’, Vload_a’


125

a. Tap change phenomenon during positive and negative polarity

Fig7.7a – Line-Line voltage [V] at input Vab & output of OLTC Vab’ as a function of time[s] for phase ab when it goes from 0% to 2% compensation in positive half cycle

Legend Vab Vab’


126

Fig7.7b – Phase voltage [V] at input & output of OLTC as a function of time[s] for phase a when it goes from 0% to 2% compensation in negative half cycle

Legend Vab Vab’


127

Fig 7.7a shows the voltage polarity based 4-step commutation (Section 6.4.1) being carried out in the positive polarity. Line-Line voltage [V] at input (Vab) and output (Vab’) of OLTC for phase ab as a function of time[s] is shown for the MV scenario. The unit moves from the tap for 0% compensation to that for 2% positive compensation. It can be seen in the zoomed-in figures from left to right that the output and input voltages overlap when no compensating voltage is injected (leftmost). As soon as a tap change is made (middle), the output voltage has a high frequency oscillation superimposed on it due to the effect of the LC oscillation between the snubber and leakage inductance. Due to the resistance of the snubber and that of the windings the oscillations get damped out so that after one cycle (rightmost), minimum oscillations can be seen and output voltage is 2% higher than the input of the OLTC. Fig 7.7b shows a similar tap transition when a tap change is made during the negative half cycle. In this case, the oscillation in the output voltage damps out in about three to four cycles. In reality the actual duration depends on the resistance of the snubber and of the tap windings, leakage inductance of the tap, snubber capacitance and on the load current at commutation instant.

b. Turn off transient of hybrid switch

Fig7.8 – Voltage [V] and current [A] of the electronic switch IGt and IGb during 4-step commutation from IGt to IGb switch in presence of snubber

Legend: VIGb IIGb VIGt IIGt


128

A tap change is made from the tap connected to the bidirectional switch IGt to the tap connected to the switch IGb during the positive cycle, for the MV scenario (see Fig 7.3b). The voltage (VIGt and VIGb) and current (IIGt and IIGb) through the bidirectional electronic switches IGb and IGt are examined in Fig 7.8. The snubber is set at 100µF and 0.01Ω and load current at commutation instant is about 400A. Oscillating voltage VIGt can be observed across the turning-off switch IGt due to the LC oscillations between the leakage inductance of the off-going tap and snubber capacitance. This verifies the estimations made in section 6.5. The oscillations get damped out in about four cycles. The oscillation can be described by the following equations from chapter 6:

𝑉𝐵𝑆1 = 𝑉𝑡𝑎𝑝|𝑠𝑡𝑒𝑝4 + 𝐿𝑙𝑒𝑎𝑘𝐶𝑠𝑛𝑢𝑏

𝐼1 (Eq 6.4b)


(Eq 6.5)

c. Effect of snubber capacitor on tap change phenomenon The effect of the snubber capacitance value on the LC oscillations is studied in this next section. According to Eq.6.4b and Eq.6.5, if the snubber capacitance is high, the magnitude of overvoltage across the switch and frequency of LC oscillations is low. By using two different value of snubber capacitance namely 100µF and 10µF, the switching transient as described in the previous section is examined in Fig7.9a for the MV scenario (VIGt, VIGb, IIGt

and IIGb). It can be seen that when the lower value of snubber capacitance of 10µF is used (Fig7.9a bottom figure) the peak overshoot in the voltage VIGt of the turning off switch IGt is extremely high. The blue patches indicate very high frequency oscillations which due to repeated cycles in an observed period appear as patches. These oscillations will stress both the switch and the transformer windings. The high amplitude and high frequency of the LC oscillations highlights the need for either increasing the snubber capacitance or overrating the switch voltage rating, the latter being a costly option. When the snubber capacitance is increased to 100µF (Fig7.9a top figure), it results in reduced voltage overshoot in VIGt and lower frequency of oscillations. Further the voltage oscillations get damped out quicker to safe limits. In Fig7.9b, an extreme scenario with no snubber capacitance is simulated for a transition from the IGb to IGt. Only the stray winding capacitance across the tap is modeled using a capacitance of 1nF. A sharp voltage spike in the tuning off switch VIGb in the order of thousands of volts can be seen in the switch voltage waveform. Such a spike in practice will permanently damage the switch. This confirms the necessity of a snubber for overvoltage protection.


129

Fig7.9a – Voltage [V] & current [A] of electronic switches during 4-step commutation from switch IGt to IGb using snubber of 100 μF (top) and 10 μF (bottom)



130

Fig7.9b – Voltage [V] & current [A] of electronic switches during 4-step commutation from switch IGb to IGt in the absence of overvoltage snubber (Parasitic capacitance of 1 nF)

d. Effect of load current at commutation instant on switch overvoltage The overvoltage and LC oscillation during tap commutation are directly dependent on the load current at instant of commutation as shown in Eq.6.4b. This is simulated for the MV case and shown in Fig7.10. In the positive half cycle of voltage, two scenarios are simulated where the commutation is done at two different instants of time. In one case the load current at commutation instant is 900A and the other case it is 400 A respectively. In Fig7.10, the voltage (VIGt and VIGb) and current (IIGt and IIGb) through the bidirectional electronic switch IGt and IGb during a tap change from IGt to IGb is observed. It can be seen that more energy is stored in the leakage inductance at the commutation instant when the current is 900A. All the inductor energy gets exchanged with the snubber capacitor leading to high overvoltage as can be examined in Fig7.10 (see Eq.6.3). The overvoltage is reduced when the load current at commutation is 400A. Thus the control strategy must be such that during the positive or negative cycle of the voltage between the taps, the commutation occurs close to the zero crossing of the load current. This will ensure minimum overvoltage across the switches and the snubber capacitor.

12𝐿𝑙𝑒𝑎𝑘(𝐼1)2 = 1

2𝐶𝑠𝑛𝑢𝑏(∆𝑉)2 (Eq 6.3)


(Eq 6.5)



131

Fig7.10 – Voltage [V] & current [A] of electronic switches during 4-step commutation from switch IGt to IGb when current at commutation instant is 900A (top) and 400A (bottom)



132

e. Voltage and current of snubber capacitor during tap change process

Fig7.11 – Voltage [V] & current [A] of snubber capacitor during 4-step commutation depicting the LC oscillation with leakage impedance

In the last set of analysis, the snubber capacitor voltage Vcap and current Icap are observed during the tap change phenomenon when the load current at commutation instant is 400A and snubber capacitor is 100 µF. For the MV scenario a tap change is made from tap 9 to tap 8 on the OLTC unit connected across phase a,b. At the commutation instant, the current through the leakage inductance Ileak of the tap section between tap 8 and tap 9 is transferred to the snubber capacitance when the bidirectional electronic switch for tap 9 turns off. This results in LC oscillations between the leakage inductance and the snubber capacitance as can be seen in Fig7.11. The peak current through the capacitor corresponds to 400A, which is the load current at the commutation instant. The oscillating voltage across the capacitor can be described by the equation Eq.6.4b and Eq.6.5. The oscillations are damped by the snubber resistance and by the tap winding resistance; in this case it takes about 4 cycles for the oscillations to damp out.

Legend: Vcap Icap Ileak


133

7.4 Summary and conclusion The series compensator was modeled in the PLECS environment and the model was customized for the MV and LV scenario. The steady state operation of the system and the transient behavior during a tap change through the electronic switches was verified and it matched the theoretical estimation. The distinct nature of operation in the MV and LV OLTC system was validated where the motive was to control the line-line voltages and the phase voltages respectively. The wye connection in the LV scenario ensured an independent regulation of all three phase voltages. LC oscillations occur between the snubber capacitance and the tap leakage inductance during a tap change. The mathematical analysis in chapter 6 and the simulations described in this chapter could prove as a useful tool for the snubber design and sizing of the switches.


134

Chapter 8: Protection and control of series compensator

135

Chapter 8

Protection and control of series compensator The single phase OLTC autotransformers will be directly connected to the grid. They will be subjected to severe grid conditions namely those of high over current due to short circuit faults and large overvoltage owing to switching and lightning surges. Further during the startup of the transformer, there will be an in-rush current flowing through the transformer windings that can be several times the nominal current, with a DC component. The series compensator must have the capability to withstand such conditions. Besides these external grid conditions, there can be system generated faults from within the OLTC transformer that can affect both the reliability of its operation and in the worst case, even the reliability of the grid. The first part of this chapter will hence be focused on the fault conditions and protection aspects of the series compensator. The second part of the chapter will be dedicated to the development of a low-level control mechanism for the OLTC. The operation of the series compensator as described in previous chapters will be monitored by a centralized controller. The controller depends on information about the grid which is obtained through voltage and current measurements for the load and source. The coordination of several aspects of the compensator system like processing of voltage and current measurements, generating pulses for firing circuits of the hybrid and no-load switches and protection will be under the eyes of the controller. Further if several OLTC are present, it would beneficial to have communication between these units for a coordinated framework for voltage control in the distribution network. These aspects will be analyzed in detail in this chapter. The key aspects of the instrumentation, protection and control of the series compensator can be summarized as follows:

• Instrumentation for measurement of grid and load voltage, load current and load power • Control of the compensator for accurate setting of tap position and for supervising the tap

changing process


136

• Reliable and rapid detection of OLTC system fault and action initiated to protect the grid, the transformer and switches with the priority: Grid> Transformer> Switches

• Ability to withstand lightning and switching surge and short circuits in the line in a safe manner. Additionally it should withstand the in-rush current and associated transients when the transformer is first energized.

• Remote monitoring and communication among different OLTC units 8.1 Internal and external fault conditions and protection mechanism In this section, various potential internal and external faults of the OLTC system are investigated and possible protection schemes are enumerated. An internal fault is defined an abnormal condition that originates from within the series compensator that may/may not result in an abnormal condition to the external grid. An external fault on the other has its origin outside the series compensator and affects its operation either directly or indirectly. 8.1.1 Internal fault conditions and protection mechanism The major internal faults/abnormal conditions that can occur in the series compensator are:

a. Inrush current during transformer energization b. Short circuit of taps through no-load switches c. Faulty commutation sequence ↔ Short circuit of taps through hybrid switch d. Hybrid switch or no-load blows out permanently e. Transformer windings burn or insulation failure occurs a. Inrush current during transformer energization Though the in-rush current phenomenon during transformer energization cannot be called a ‘fault’ in conventional terms, it is certainly an abnormal condition where the current through the transformer windings are several times the nominal current of the transformer and have a DC component as well. Transformer windings are usually designed to carry this in-rush current safely. What is then important is to ensure that the switches are protected from the high in-rush current. A close look at topology 3a will show that the switches are present on the load side of the autotransformer. When an autotransformer is energized with an active load connected, the inrush current flows from both the source side and load side (Active load is a load that can draw power as well as inject power, e.g. a house with overhead PV panels). Disconnectors should be incorporated into the design of the compensators [33] that are connected to the points S, L, and SL (Refer chapter 5 for S,L,SL) as shown in Fig 8.1a,b. In such a case, the load side of the autotransformer must be disconnected during energization


137

using a disconnector L in open position. By this way, an inrush current from the load side through the switches is prevented. The disconnectors double up as isolators for the system during periods of no-service such as regular maintenance. The ground connection of the metal body of the compensator for safety purpose is shown in the figure as well.

Fig 8.1a – Disconnectors connected to LV series compensator

Fig 8.1b – Disconnectors connected to MV series compensator


138

An alternative cheaper way would be to use the hybrid switch of the series compensator in ON condition (Mechanical switch is in ON position and electronic switches are OFF) during transformer energization. By this way the in-rush current flows through the mechanical switch of the hybrid switch and the no-load switches. Since these non-semiconductor switches have high overload capacity, they can bear the in-rush current for short period of time in a safe manner (see section 6.1). One might suggest using the hybrid switch as a pseudo disconnector and keeping it in OFF condition during the transformer energization to prevent a load-side in-rush current. This is not a solution as it will permanently damage the electronic switches in the hybrid switch. If the mechanical switch is in OFF condition, the electronic switches will have to block 1pu voltage across them when they are rated only for 0.02pu voltage! b. Short circuit of taps through no-load switches When two no-load switches in the same connected branch (green or red) are simultaneously in closed position, a short circuit of two taps occurs as shown in Fig 8.2. With no impedance except the leakage impendence of the taps to limit the short circuit current, the short circuit current can be large. The current is impossible to interrupt considering the nature of no-load switches. Further, the hybrid switch can have no influence in averting the situation. As a result, the no-load switch will mostly probably blow permanently or the transformer windings in that section will burn. This is a catastrophic fault that should never occur. The controller should always ensure that at any point of time only one no-load switch amongst the red and green connected branches is ON (see section 6.2.4). Additional logic gates can be implemented at the input of the gating circuit of the no-load switches to ensure that at no point of time are two no-load switches simultaneously gated in the same group (red or green).

Fig 8.2– Short circuit through two no-load switches


139

c. Faulty commutation sequence ↔ Short circuit of taps through hybrid switch Short circuit between taps can also occur due to a faulty commutation sequence of the voltage polarity based 4-step current commutation. As discussed in section 6.4.3, this is most likely to occur around the zero crossing of the voltage between the taps. The hazard of failure is minimal owing to the availability of a small short circuit voltage. The worst case scenario is when a faulty commutation sequence is implemented close to the voltage peak of the wrong polarity. This can occur when the controller makes a delay of upto ms in generating the gating signals for the bidirectional switch. This is however unlikely considering the high reliability of present day DSP and microcontroller technology and their advanced reliable application to high frequency converters; where gating signals are accurately given with delays of less than a μs. d. Hybrid switch or no-load blows out permanently Hybrid or no-load switch blowing can be as a result of short circuit between the taps or an unprecedented overvoltage occurring in the grid. If the switch gets damaged because of an overvoltage, it will result in the switch becoming a permanent short circuit path, while an overcurrent damaging the switch will make it a permanent open circuit. Several possibilities exist which are examined. First, if a no-load switch is damaged due to overcurrent with no damage to the transformer, the OLTC can still operate with the other no-load switches. Tap changes that involve the damaged switch however cannot be performed.

Fig 8.3a – Bypass switch connected to LV series compensator


140

Fig 8.3b – Bypass switch connected to MV series compensator Secondly, if the electronic switches in the hybrid switch are damaged due to overcurrent, no tap change can be made henceforth. The system can continue to serve the load with the existing tap through the mechanical switch until the unit is disconnected for maintenance. Thirdly and the most unlikely situation is that of the mechanical switch of the hybrid switch blowing out. This is unlikely owing to the high overload capacity of mechanical switches [32]. In such a condition, the electronic switch of the hybrid switch can be used to supply the load temporarily. Fourthly, if an overvoltage damages either a no-load switch or the hybrid switch it will result in a permanent conducting path through the switch which is disastrous and should be prevented under all circumstances! In all four cases, a bypass switch can be incorporated into the design so that the OLTC unit can be bypassed and disconnected using the disconnectors. The load power can be directly supplied through the bypass switch however without any voltage compensation [50]. Such a switch is shown in Fig 8.3a,b. The bypass switch can be a mechanical switch, back-back connected thyristor or a vacuum switch. e. Transformer windings burn or insulation failure occurs In the improbable occurrence of the burning of the transformer windings or an insulation failure, the series compensator operation is permanently inhibited. The bypass switch can be incorporated in such a scenario to supply the load and the disconnectors can be used to isolate the transformers so as to avert further damage to the transformer.


141

8.1.2 External fault conditions and protection mechanism

The major external faults that can affect the series compensator are:

a. Short circuit on line during steady state operation of the compensator b. Short circuit on line during tap change (transient) operation of the compensator c. Overvoltage due to switching and lightning surges

a. Short circuit on line during steady state operation of the compensator If a short circuit occurs on the distribution line due a fault (LG, LLG or LLL fault), a high current flows through the line and subsequently through the series winding of the autotransformer. In such a scenario if the OLTC unit is in steady state, the mechanical switch of hybrid switch and the corresponding no-load will be conducting the fault current. Assuming that circuit breakers are present on the line to interrupt the fault current in a maximum of a few ms, the mechanical switch and no-load switch must be able to withstand the fault current for that short duration. Owing to high overload capacity of these switches, this is certainly realizable. It is important that tap changes are temporarily inhibited in this fault condition as the electronic switches (of the hybrid switch) that are required for a tap change cannot withstand the fault current and will get damaged. b. Short circuit on line during tap change (transient) operation of the compensator On the other hand, if a short circuit on the line occurs during an on-going tap change process, the situation is very different compared to that discussed above. The electronic switches will be directly handling the fault current which could permanently damage them. To prevent such a scenario three possible solutions are proposed:

1. Overrating of the electronic switches and the cooling mechanism (heat sink) so that they can handle the fault current for a short duration till the circuit breaker interrupts the fault current. It would be expensive and complicated to realize high current capacity IGBT/MOSFET that can handle fault currents in the order of kA

2. Topology 4 which utilizes a series transformer can be used instead of topology 3a (see Fig 6.7). The series transformer will act like an isolation transformer that will saturate when a large fault current flows on the line side preventing a high current from flowing through the switches [55]. Two major drawbacks of this solution exist, one being the additional cost of the series transformer and the necessary design for fault current limiting capability. The other is that, there would be a finite value of transformer current Isat below which the transformer does not saturate and for practical reasons, Isat will have to be at least a few times that of the nominal current. The switches will hence have to be overrated to handle the fault current upto Isat.


142

3. The most cost effective solution would be to use the Bypass switch [32, 50] shown in Fig 8.3a,b. The bypass switch can be rapidly triggered as soon as a short circuit is detected and the hybrid switch can be turned OFF. The current will commutate from the series winding and the hybrid switch, to the bypass switch. Back-back thyristors can be used for the bypass switch so that they can be turned ON quickly under a fault condition. Fig 8.4 shows such a bypass switch connected to the proposed design of topology 3a.

Fig 8.4 – Bypass switch connected to proposed design of topology 3a c. Overvoltage due to switching and lightning surges Overvoltage on the lines is a typical phenomenon that happens due to the occurrence of lighting and switching surges. Both the transformer windings and the switches have to be protected against such overvoltage. The most widely used method for protection against overvoltage is by using surge protectors. Surge protectors are connected across the series and shunt winding of the autotransformer [33] as shown in Fig. 8.5a,b (marked as MOV) for the MV and LV OLTC units. Compared to the LV, the MV OLTC unit requires an additional surge protector at SL. This is because in the MV units, SL is connected to phase B while for the LV case SL is connected to the grounded neutral. If the neutral is not grounded, then a third surge arrestor will be required for the LV unit as well. The most suitable surge protector will be metal oxide type surge protectors or metal oxide varistor (MOV) as indicated in the figure. Thyristor surge protection device (TSPD) and crowbar technology using back to back thyristors are also an alternative for providing overvoltage protection [51].


143

AΔVa

A’

Series Compensator

LS

SL

MOV

N N

Disconnector S

Disconnector L

Disconnector SL

Bypass Switch

Fig 8.5a – Bypass switch connected to LV series compensator

AΔVab

A’

Series Compensator

LS

SL

MOV

B B’

Disconnector S

Disconnector L

Disconnector SL

MOV MOV

Bypass Switch

Fig 8.5b – Bypass switch connected to MV series compensator


144

8.2 Control of series compensator A low level control mechanism is developed for the overall operation of the series compensator. The key aspects of the control of the system for robust operation can be categorized into steady state, tap change operation and protection of system: Steady state and tap change operation:

• Determination of system voltage, load current, power factor and related quantities for control processing and user display.

• Algorithm to accurately fix the voltage reference point for the OLTC system based on the grid voltage, load voltages and magnitude of load current.

• Reverse power flow detection and altered operation based on such detection • Monitor tap position and ensure the short circuiting of taps through no-load switches is

prevented • Generate firing pulses for the gating circuit of the hybrid switch for tap change operation • Communication between OLTC units of different phases and at different locations for

coordinated voltage control throughout the distributed network • Ability to have remote operation of units from a centralized control center

Protection of system:

• Rapid and reliable detection of internal or external fault condition and take necessary protection/precautionary action. Operate disconnectors and bypass switches accordingly

• Transformer tap changing must be temporarily inhibited if there is an unsafe situation owing to internal or external fault condition

• Communicate to grid operator of fault condition Fig 8.6 depicts the block diagram of the control mechanism for the series compensator. The currents and voltages of the OLTC systems are measured and are processed through a series of blocks to determine the steady state and transient operation of the compensator. The operation of the overall control mechanism and of the different blocks is explained below. Voltage reference Vref* block Based on the grid voltage and load current, the voltage reference block calculates the reference voltage set point for the output of the series compensator. This voltage set point can be of fixed type or dynamic type [52]. In case of fixed voltage set point, the aim of the compensator is to set the voltage at its output to a constant value under conditions of a fluctuating input voltage and fluctuating load current. For a dynamic voltage set point, the aim is to regulate the voltage at a remote location say at the end of the feeder, by regulating the compensator output voltage.


145

Fig 8.6 - Block diagram

of the control mechanism

for the series compensator

Chapter 8: Protection & control of series compensator

146

The line drop compensation (LDC) method belongs to type of dynamic voltage set point [32,52]. Here the aim is provide a series voltage so as to compensate for the line drops/gain along the feeder due to the load current and line impedance. Recalling Eq 2.4,

V s l s l s l s lR P X Q X P R QjV V+ −

∆ = + (Eq 2.4)

Thus depending on the impedance of the line and power drawn by the load a finite voltage difference exists between the head of the feeder and its end. The aim of the voltage reference block will hence be to estimate this voltage difference and set the voltage reference as:

𝑉𝑟𝑒𝑓∗ = 𝑉𝑛𝑜𝑚 + ∆𝑉 (Eq 8.1)

Such a dynamic set point will be most suitable in environments of high distributed generation. Conditions of reverse power flow can be implemented by setting negative values for Pl which will result in negative values for ΔV. Vref will hence be set at a value less than Vnom through the implementation of negative compensation. Voltage Bandwidth BW block The on-load tap changer has a voltage step nature of operation. Due to this there must be an acceptable range of voltage error ΔU upto which there will be no tap change initiated. This is to reduce both the number of tap changes and also to prevent the occurrence of a hunting phenomenon [52, 53, 33]. For e.g. if each tap injects 2V and if the voltage error is ΔU=0.5V, changing to a higher tap must not be initiated as that will lead to a ΔU= -1.5V! This will subsequently cause a tap change to a lower tap and the process will repeat itself, a phenomenon called hunting. The bandwidth BW is defined in the ANSI=IEEE C57.15 standard as the total voltage range one-half of which is allowed above and one-half below the voltage set point Vref*, without initiating a tap change [52, 53, 33]. So modulus of error in voltage |ΔU| must exceed half of bandwidth BW in order for a tap change to be initiated. The minimum possible bandwidth is equal to the voltage step of one tap. When the voltage exceeds or goes below the voltage set point Vref* by margin of BW/2, a tap change command is send to the time delay block. For e.g. if each tap injects 2V, then the bandwidth BW>=2V. If the bandwidth is set at 3V, then the minimum value of |ΔU| for initiating a tap change is 1.5V. If Vref*=230V, then only if the output voltage Vout exceeds 231.5V or goes below 228.5V will a tap change be initiated. Time delay TD block The time delay block is used to introduce a premeditated time delay to the tap initiation command obtained from the band width block [52, 53, 33]. This is to ensure that a tap change


147

is not initiated in response to short duration voltage fluctuation like that caused due to the starting of a large induction motor. Undervoltage caused by starting of an induction motor will exist for upto 15 seconds due to the large starting current drawn by the machine but will disappear once the machine reaches it nominal speed. Making a step up and step down operation in a span of ten seconds is hence futile and will not help the induction motor starting in any positive way. Only when the output voltage Vout remains outside the bandwidth limits for a period greater than time delay TD, a tap change command is sent to the controller. The predetermined delay is usually set in the range of 20 to 60 seconds depending on the distribution system configuration. The concepts of band width and time delay are explained figuratively in Fig 8.7a [53]. When Vout remains outside for a period greater than TD, only then is a tap change initiated. The time delay can be a fixed time delay as mentioned earlier but can also be dynamically set based on the magnitude of voltage error |ΔU|. This is shown in Fig 8.7b [54]. An inverse time curve is used for determining the time delay - when the voltage error is high, quick remedial action is required so TD is low. While on the other hand when voltage error is small, the TD is set high. Further the time delay can be used for coordinated voltage control along a distribution having more than one series compensator. In such a case, the compensator at the head of the network will have more influence on the voltage in different feeders than a compensator in the middle of the network, assuming downstream power flow. Then the time delay for the middle

Fig 8.7a – Tap change initiated after Vb > BW limit for a period of time > TD [53]

Vout


148

Fig 8.7b – Dynamic time delay based on voltage error using inverse time delay curve

compensator can be set higher than the delay for the head compensator for the same voltage error. This will ensure that the head compensator will first operate and if the voltage problem is not yet fixed, action by the middle compensator will follow. To summarize the role of the time delay block, the following can be listed:

1. Prevent frequent tap changes and override short transient voltage fluctuation 2. Inverse curve ensures dynamic time delay based on magnitude of voltage error 3. Coordinated voltage control through multiple compensators in distribution network

Control block The control block is the brain of the compensator that provides the gating signals for the hybrid switch and the no-load switches based on inputs from the measurement, time delay and protection block. It monitors the tap position and when it gets a tap change command from the time delay block, it provides the necessary pulses to the gates of the hybrid switch to perform the 7-step tap change process. It processes signals from the measurement block that helps in determining the voltage polarity and the load current magnitude for initiating 4-step current commutation through the electronic switches. It has one another important function which is to determine the correct instant for the closure of no-load switches (Step 1 of 7-step of tap change process). The presence of a charged snubber capacitor connected across the hybrid switch means that step 1 must be performed at the right instant when the voltage between the taps matches the snubber capacitor voltage. A voltage measurement is thus required across the hybrid switch to realize this. A mismatch in the two voltages will lead to a current spike resulting from the charging/discharging of the snubber capacitor.


149

The control block has the ability to communicate with OLTC units of different phases and at different locations for coordinated voltage control throughout the distributed network. It enables remote operation of unit from a centralized control center. The output signals are sent to the backup control and protection block, where the compensator and grid are checked for abnormal conditions. If everything is fine, the control block signals are passed on to the series compensator switches.

Back-up control and Protection block The back-up control and protection block is responsible for the rapid and reliable detection of internal or external fault condition and to take necessary protection/precautionary action through the control block. Transformer tap changing is temporarily inhibited if there is an unsafe situation detected. This involves the operation of the disconnectors and turning ON of the bypass switches depending on the fault diagnosis. Additional measurements including the temperature of switches and transformer windings can be incorporated into the block to provide on-line monitoring of the complete series compensator system. It also communicates to grid operator and with other OLTC units during a fault condition. It is advisable to have independent hardware protection in addition to the protection enabled through the controller.

Measurement block The measurement block obtains parameters from the voltage and current measurement devices on the series compensator namely - grid/input voltage, load/output voltage and load current. The grid voltage polarity will directly reflect on voltage polarity between the transformer taps and hence will be used in the 4-step commutation of electronic switches. The load voltage and current measurements will be used in estimating the voltage drop along the line and for setting the voltage reference for the controller in the voltage reference block. The measurement will also be used for detecting a fault condition in the protection block and to initiate necessary remedial action in the compensator. Additionally, the voltage across the snubber capacitor in the hybrid switch will be measured for performing a tap change through the control block. Voltage across the IGBT/MOSFET switches and the mechanical switch can also be measured and incorporated to the controller to provide details of the conduction state of the switches.


150

8.3 Summary and conclusion The major internal and external fault conditions that can affect the operation of the series compensator were examined in this chapter. Through the use of a bypass switch, metal oxide surge protectors and disconnectors, the system can be protected to a large extent from these abnormal conditions. A deeper analysis of the protection is still required to choose the best alternatives among the protection schemes. A low-level control mechanism for the compensator was developed taking into account the need to control the voltage in the distribution network in a coordinated fashion and to override short term fluctuations in the order of a few seconds. The control will also detect an abnormal condition in the tap changer and in the grid to initiate necessary protective action.

Chapter 9: Conclusion and scope for future work

151

Chapter 9

Conclusion and scope for future work

This chapter gives an overview of the project’s contributions. It will reflect on the results

obtained from this thesis work and draw important conclusions. Finally, some ideas for future

work will also be discussed.

9.1 Overview of thesis work

A novel design for a series voltage compensator based on on-load tap changing transformer

has been proposed. The aim was to design an OLTC that has low steady state losses and arc-

free tap changing. Various topologies for an OLTC were proposed and an autotransformer

design having taps on the secondary side was concluded as being a simple and cost effective

solution. The OLTC taps were made from a combination of no-load switches and a single

hybrid switch and exhibited several advantages. The hybrid switch eliminated the arcing

problems during tap change and had low losses during steady state. The use of no-load

switches reduced the voltage rating and the number of active hybrid switches that were

required. The use of voltage polarity based 4-step current commutation on series connected

back to back IGBT/MOSFET provided a convenient method for performing a tap change

without the need for a current limiting impedance. A single overvoltage snubber connected

across the hybrid switch protects it from overvoltage resulting from interruption of current

through the tap leakage inductance. The design has been customized for application in both

LV and MV European distribution network. The entire system was simulated in the PLECS

environment and both the steady state and transient operation of the system was analyzed.

With investigation into the fault condition, protection methods and the control for the series

compensator, a holistic design is provided for building a prototype.


152

9.2 Results and conclusion

Voltage fluctuation in the distribution network

Frequent voltage fluctuations are observed in the distribution network owing to large

scale renewable energy integration such as PV. The fluctuations are dependent on the

impedance of the line, load current & power factor, power injected due to distributed

generation (DG) and the distribution of the loads & DG along the feeder

Undervoltage is observed at end of feeders during period of high loads and low PV

injection and overvoltage during periods of low loads and high PV injection.

Voltage compensation is of two types – series and shunt compensation. European

distribution network in general is characterized by long lines that have a low X/R ratio,

which makes shunt compensation inefficient and not cost-effective.

Under conditions of uneven feeder lengths and non-uniform distribution of loads and DG,

partially rated, feeder specific series compensation is a suitable voltage compensation

technique for European distribution network, as shown in Fig 9.1.

Fig 9.1 – Feeder specific compensation through partially rated series compensator


153

Series compensation using on-load tap changing transformer (OLTC)

‘Make and break’ tap change is preferred over ‘break and make’ tap change owing to loss

of load and resulting overvoltage.

Voltage regulators using mechanical switches have reduced lifetime owing to the arcing

phenomenon due to frequent tap changes. Electronic switch based tap changers have no

arcing problem but higher steady state losses. Hybrid switches combine the best of both –

no/minimal arcing and low steady state losses and hence are the suitable choice for future

design of OLTC taps.

OLTC implemented using partially rated autotransformer are used as the basis for

building the series compensator. They are more compact and cost effective compared to

conventional two winding transformers as they have reduced copper usage and much

higher throughput power.

11 OLTC topologies were presented and examined analytically. They were simulated in

the PLECS platform to study their steady state operation and verify the theoretically

estimated switch and transformer ratings. Topologies 3, 3a, 4 and 5 were shortlisted based

on the voltage-current ratings and number of the components required.

The goal is to obtain independent control of phase voltages in LV network and control of

line voltages in the MV network. It was demonstrated using PLECS simulation and

mathematical analysis that open-delta connection using two OLTC units in MV and Wye

connection using three OLTC units in LV were the most connection scheme

Design of series compensator

Novel design of a series compensator was proposed based on topology 3a, that uses a

combination of no-load switches and a single hybrid switch as shown in Fig 9.2. It has

arc-free tap change operation and low steady state losses. The bidirectional electronic

switch is realized using a back to back series connection of MOSFET for the LV design

and IGBT for the MV design.

7-setp tap change process is proposed for changing between transformer taps through the

no-load switches and single hybrid switch with minimum/no arcing.

4-step voltage polarity based current commutation as used in matrix converters is used to

change between the bidirectional electronic switches during a tap change. It is chosen

over current polarity based 4-step commutation due to less severe failure hazard.

Tap change in OLTC results in overvoltage due to the interruption of current through the

leakage inductance. A single RC overvoltage snubber connected across the hybrid is used

to protect the hybrid switch against overvoltage.

Slow turn off of switches through control of gate resistance and commutation around

current zero of the load current would also aid in reducing the overvoltage.

The transient and steady state operation of the series compensator were analyzed in

PLECS for both the MV and LV scenario. The simulations verified the robust operation

of the series compensator for both positive and negative compensation.


154

Fig 9.2 – Proposed realization of topology 3a using no-load and hybrid switch

Protection and control strategy for series compensator

Several conceivable internal and external faults that can affect the operation of the series

compensator were investigated. This includes overvoltage due to switching and lightning

surges, short circuit fault on the line and faulty commutation sequence among others.

Possible protection mechanisms were suggested through the implementation of a bypass

switch, surge arrestors and disconnectors.

A low-level control mechanism for the series compensator operation was developed. The

key aspects of the control were the use a voltage bandwidth and time delay blocks to

override short duration voltage fluctuation and prevent the occurrence of hunting.

The protection block monitors and ensures that the system is working fine at all times and

overrides all control commands in case of a fault condition.

Communication between OLTC units of different phases and at different locations will

provide a coordinated framework for voltage control throughout the distribution network

9.3 Scope for future work

The scope of future work includes and is not restricted to the following:

1. Use a unidirectional switch with current polarity based commutation

If the entire commutation 7-step commutation can be performed within 10ms (steps 2 to steps

6) using fast mechanical switches in the hybrid switch, the bidirectional electronic switch can

be replaced with a unidirectional electronic switch like an IGBT with a series connected


155

diode. 4-step current polarity based commutation can be used and commutation can be

performed in that polarity that matches the directionality of the electronic switch. The number

of components in the electronic switch gets reduced by half. But at the same time a failure in

commutation due to wrong polarity judgment can be catastrophic owing to the lack of a

current path in the reverse direction.

2. Realization of no-load switch and possibility of using movable no-load switch as in

diverter type voltage regulator

Simple no-load switches as described in the thesis are not available in the market off the shelf.

A dedicated design is required for building them. Alternatively, the no-load switch design can

be adapted from ‘diverter switch’ type conventional voltage regulator. Instead of using 10 no-

load switches, two movable no-load switches can be used, like ‘tap selector’ switches shown

in Fig 6.3a,b.

3. Protection mechanism and development of control

The protection mechanisms suggested in chapter 8 have to be investigated further to make a

scientific choice for the most suitable mechanism. A detailed comparison based on the

operation mechanism, reliability, lifetime and cost would be required. The protection scheme

has to be adapted to the specific environment in which the compensator will be connected.

The low-level control strategy has to be developed in more detail and implemented in a

FPGA or DSP controller. The time delay and bandwidth block must be adaptable to the

requirements of the DSO. A protocol for communication between OLTC units of different

phases and at different locations must be established for coordinated voltage control.

4. Building a prototype

The final and most important step is the building of a full scale prototype of the compensator

for the MV and LV scenario. It involves the setting up of the transformer and the necessary

power electronics and switches for the taps. The gating circuit for the switches and

instrumentation for the voltage and current measurement have to be incorporated. The

controller for the compensator and the necessary protection equipment has to be integrated.


156

157

REFERENCES

[1] Esslinger, P., Witzmann, R., “Improving grid transmission capacity and voltage quality in low-voltage grids with a high proportion of distributed power plants”, Energy Procedia, 12, pp. 294-302., 2012

[2] Woyte, A.; Van Thong, V.; Belmans, R.; Nijs, J.; , "Voltage fluctuations on distribution level introduced by photovoltaic systems," Energy Conversion, IEEE Transactions on , vol.21, no.1, pp. 202- 209, March 2006

[3] Jewell, Ward; Ramakumar, R.; , "The Effects of Moving Clouds on Electric Utilities with Dispersed Photovoltaic Generation," Power Engineering Review, IEEE , vol.PER-7, no.12, pp.31-32, Dec. 1987

[4] Jewell, W.T.; Ramakumar, R.; Hill, S.R.; , "A study of dispersed photovoltaic generation on the PSO system," IEEE Transactions on Energy Conversion, vol.3, pp.473-478, Sep 1988

[5] E. Wiemken, H.G. Beyer, W. Heydenreich, K. Kiefer,; “Power characteristics of PV ensembles: experiences from the combined power production of 100 grid connected PV systems distributed over the area of Germany”, Solar Energy, Volume 70, Issue 6, 2001, Pages 513-518

[6] R.Passey, T.Spooner, I. MacGill, M.Watt, K.Syngellakis,; “The potential impacts of grid-connected distributed generation and how to address them: A review of technical and non-technical factors”, Energy Policy, Volume 39, Issue 10, Pages 6280-6290, Oct. 2011

[7] Bauer, P.; De Haan, S.W.H.; , "Solid state tap changers for utility transformers," Africon, 1999 IEEE , vol.2, no., pp.897-902 vol.2, 1999

[8] Bauer, P.; de Haan, S.W.H.;, "Electronic tap changer for 500 kVA/10 kV distribution transformers: design, experimental results and impact in distribution networks," Industry Applications Conference, 1998. Thirty-Third IAS Annual Meeting. The 1998 IEEE , vol.2, no., pp.1530-1537 vol.2, 12-15 Oct. 1998

[9] Masters, C.L.; , "Voltage rise: the big issue when connecting embedded generation to long 11 kV overhead lines," Power Engineering Journal , vol.16, no.1, pp.5-12, Feb. 2002

[10] Kojovic, L.; , "Impact of DG on voltage regulation," Power Engineering Society Summer Meeting, 2002 IEEE , vol.1, no., pp.97-102 vol.1, 25-25 July 2002

[11] Hingorani, N.G.; "Introducing custom power," IEEE Spectrum, vol.32, pp.41-48, June 1995 [12] T. J. Miller, “Reactive Power Control in Electric Systems”. New York: Wiley, 1982 [13] N. Hingorani and L. Gyugyi, “Understanding FACTS, Concepts and Technology of Flexible

ac Transmission Systems”, New York: IEEE Press, 2000 [14] Dixon, J.; Moran, L.; Rodriguez, J.; Domke, R.; , "Reactive Power Compensation

Technologies: State-of-the-Art Review," Proceedings of the IEEE , vol.93, no.12, pp.2144-2164, Dec. 2005

158

[15] G.Beck, W.Breuer, D.Povh, D.Retzmann, E.Teltsch,;"Use of FACTS and HVDC for Power System Interconnection and Grid Enhancement"; POWER-GEN Middle East 2006, Jan 30-Feb 1, 2006

[16] S. Torseng, “Shunt-connected reactors and capacitors controlled by thyristors,” IEE Proc. Part C, vol. 128, no. 6, pp. 366–373, Nov. 1981.

[17] L. Morán, P. Ziogas, and G. Joos, “Analysis and design of a synchronous solid-state Var compensator,” IEEE Trans. Ind. Appl., vol. IA-25, no. 4, pp. 598–608, Jul./Aug. 1989.

[18] Demirok, E.; Sera, D.; Teodorescu, R.; Rodriguez, P.; Borup, U.; , "Clustered PV inverters in LV networks: An overview of impacts and comparison of voltage control strategies," Electrical Power & Energy Conference (EPEC), 2009 IEEE , vol., no., pp.1-6, 22-23 Oct. 2009

[19] R. Grünbaum, M. Noroozian, and B. Thorvaldsson, “FACTS—powerful systems for flexible power transmission,” ABB Rev., pp. 4–17, May 1999

[20] N. H. Woodley, “Field experience with dynamic voltage restorer systems,” IEEE Power Engineering Society Winter Meeting 2000, Singapore

[21] T. K. Saha and P. T. Nguyen, “Dynamic voltage restorer against balanced and unbalanced voltage sags: modeling and simulation,” IEEE Power Engineering Society General Meeting, Denver, CO, 2004

[22] IEEE Recommended Practice for Monitoring Electric Power Quality, IEEE Std 1159, 1995 [23] Dr. Mahesh Kumar, "Power Quality in Power Distribution Systems", NPTEL Course, Indian

Institute of Technology Madras, India [24] Divan, D.; Sastry, J., "Controllable Network Transformers," Power Electronics Specialists

Conference, 2008. PESC 2008. IEEE , vol., no., pp.2340,2345, 15-19 June 2008 [25] Tom Short, "Electric Power Distribution Equipment and Systems", Taylor & Francis, Nov

2005 [26] GE Global Research Center, Munich & EnBW AG, Germany. [27] “Transformer”, Module 7, NPTEL, EE, IIT Kharagpur, India [28] “Transformers”, Electrical Machines I, NPTEL,EE, IIT Madras, India [29] John J. Winders Jr., “Power Transformers: Principles And Applications”, Taylor & Francis,

2002 [30] “On-Load Tap-Changers for Power Transformers - A Technical Digest”, Maschinenfabrik

Reinhausen GmbH, 2009 [31] Bimbhra PS, “Generalized theory of electrical machines” Khanna Publishers, New-Delhi,

India, 1992 [32] J.Faiz, B.Siahkolah, "Electronic Tap-changer for Distribution Transformers", Springer, 2011 [33] James H.Harlow, “Electric Power Transformer Engineering”, 2nd Edition, Taylor & Francis

Group, 2006

159

[34] Bishop, M.T.; Foster, J.D.; Down, D. A., "The application of single-phase voltage regulators on three-phase distribution systems," Rural Electric Power Conference, 1994. Papers Presented at the 38th Annual Conference , vol., no., pp.C2/1,C2/7, 24-26 Apr 1994

[35] Zai FQY, O’Kelly D, “Solid-State on-load transformer tap-changer”, IEE Proc Electric Power Appl,143(6):481–491, 1996

[36] ABB (1997) On-load tap-changers, Type UBB-Technical guide, 1997 [37] Gao D, Lu Q, Luo JA, “New Scheme for on-load tap-changer of transformers”, International

Conference on Power System Technology, pp 1016–1020, 2002 [38] Robert ME, Ashman WG, “A thyristor assisted mechanical on-load tap-changer”, IEE Proc

Power Thyristors Their Appl, 185–192, 1969 [39] Cooke GH, Williams KT, “Thyristor assisted on-load tap-changers for transformers”, 4th

International conference on power electronics and variable-speed drives, pp 127–131, 17–19 July 1990

[40] Cooke GH, Williams KT, “New thyristor assisted diverter switch for on-load transformer tap-changers”, IEE Proc Electric Power Appl , pp 507–511, 1992

[41] Bauer P, Schoevaars R, “Bidirectional switch for a solid-state tap-changer”, IEEE PESC 03, pp 466–471, 2003

[42] Das, D.; Divan, D., "Power flow control in networks using controllable network transformers," Energy Conversion Congress and Exposition, 2009. ECCE 2009. IEEE , pp.2224,2231, 20-24 Sept. 2009

[43] M. Venturini, “A new sine wave in, sine wave out, conversion technique eliminates reactive elements,” Proceedings of Powercon 7, San Diego,CA, pp. E3-1-E3-15, 1980

[44] S. Bernet, T. Matsuo and T.A. Lipo, “A Matrix Converter Using Reverse Blocking NPT IGBT’s and Optimised Pulse Patterns,” IEEE PESC’96, Italy, pp. 107-113, June 1996

[45] N. Burany, “Safe Control of Four-Quadrant Switches,” IEEE-IAS Annual Meeting, 1989, pp. 1190-1194

[46] A. Alesina, M. Venturini, “Analysis and Design of Optimum-Amplitude Nine-Switch Direct AC-AC Converters”, IEEE Transactions on Power Electronics, Vol. 4, no. 1, pp. 101-112, January 1989

[47] C. Klumpner, P. Nielsen, I. Boldea, F. Blaabjerg, “A New Matrix Converter-Motor (MCM) for Industry Applications,” IEEE/IAS Annual Meeting 2000, vol. 3, pp. 1394-1402, 2000.

[48] M. Ziegler, W. Hoffman, “Semi Natural Two Steps Commutation Strategy for Matrix Converters,” IEEE PESC’98, vol. 1, pp. 727-731, 1998

[49] L. Empringham, P.W. Wheeler and J.C. Clare, “Intelligent Commutation of Matrix Converter Bi-directional Switch Cells using Novel Gate Drive Techniques,” IEEE PESC’98, pp. 707-713

[50] Yousef-Zai, F. Q.; O'Kelly, D., "Solid-state on-load transformer tap changer," Electric Power Applications, IEE Proceedings - , vol.143, no.6, pp.481,491, Nov 1996

160

[51] Bauer P, de Haan SWH, “Protective device for electronic tap-changer for distribution transformers”, EPE 97, pp 4/282–4/287, Sept 1997

[52] “Basic considerations for the application of LTC transformers and Associated controls”, Tap changer controls Application Note 17, Beckwith Electric co., USA

[53] J.H.Harlow, “Load tap change control”, National rural electric cooperation association, Houston, Texas, March 1996

[54] “Basic considerations for the application of LTC transformers and Associated controls”, Tap changer controls Application Note 29, Beckwith Electric co., USA

[55] Cvoric, D.; De Haan, S. W H; Ferreira, J.A., "New saturable-core Fault Current Limiter topology with reduced core size," Power Electronics and Motion Control Conference, 2009. IPEMC '09. IEEE 6th International , vol., no., pp.920,926, 17-20 May 2009

[56] PLECS® version 3.3, Plexim GmbH, Zurich, Switzerland, http://www.plexim.com [57] “Standard EN 50160 - Voltage characteristics of electricity supplied by public distribution

networks”, CENELEC, 2010 [58] J.H.Harlow, “Transformer Tap changing under load – A review of concepts and standards”,

64th Annual Engineering Conference, Missouri, April 1993 [59] Esslinger, Peter, “Studie Q(U) - Schlussbericht”, Kooperationsprojekt der EnBW Regional

AG, der E.ON Bayern AG, der SMA Solar Technology AG, der KACO new energy GmbH, der Siemens AG und des Fachgebiets Elektrische Energieversorgungsnetze der Technischen Universität Münche, p 18, 31 August 2012


http://www.plexim.com/

Novel Design of a Power Electronic Assisted

OLTC for Grid Voltage Regulation Gautham Ram

#, Pavol Bauer

#, Thiwanka Wijekoon*, Eva-Maria Bärthlein*, Ara Panosyan*

# Department of Electrical Sustainable Energy, Delft University of Technology,

Mekelweg 4, 2628 CD, Delft, The Netherlands

[email protected]

[email protected]

* GE Global Research – Europe

Freisinger Landstrasse 50, D-85748, Garching bei München, Germany

[email protected]

Abstract — The paper describes the design of a partially

rated, feeder specific series compensator that uses an on-load tap

changing (OLTC) autotransformer. Positive and negative

compensation of the grid voltage can be achieved on specific hot-

spot feeders that have high loading and/or distributed

generation. A novel design comprising no-load switches and solid

state switches has been proposed that saves on cost, has low

steady state losses and requires lower maintenance. The hybrid

switch ensures that there is no arcing during the tap change

process ensuring long life of the switches. The OLTC system has

been customized for both LV and MV three phase distribution

network. Simulations carried out in PLECS verify the steady

state and transient operation.

Keywords— Hybrid switch, four step commutation, series

compensation, voltage fluctuation, on-load tap change

I. INTRODUCTION

In recent years, high penetration of distributed generation

(DG) driven by PV panels and heavy load insertion in the

distribution network has led to frequent voltage fluctuations in

the form of undervoltage and overvoltage [1]-[5]. Voltage

control using traditional voltage regulators are unable to cope

with this situation as frequent tap changes reduce the lifetime

of the mechanical taps due to arcing [6]. Further, the nature of

European distribution network in general, makes voltage

control through shunt compensation methods ineffective and

expensive [10], [11]. Series compensation through centralized

on-load tap changing (OLTC) distribution transformers or

feeder-specific compensators is hence a suitable strategy for

voltage regulation in Europe.

This paper describes the design of a power electronic

assisted OLTC autotransformer that provides voltage

regulation in European distribution network through series

compensation. A novel design of a partially rated

autotransformer is proposed which has taps developed through

a combination of no-load switches and a single hybrid switch.

The hybrid switch is composed of a mechanical switch and

two semiconductor switches which are used for steady state

and tap change operation respectively. The mechanical switch

ensures low steady state losses and the semiconductor

switches provide arc-free tap changing.

The novel design of the OLTC autotransformer is cost

effective, efficient and has long lifetime. Back-back series

connected IGBTs with anti-parallel diodes are used for the

two electronic switches and voltage polarity based 4-step

current commutation is used for changing between the taps.

The complete system is simulated in the PLECS simulation

tool and customized for application in both MV and LV three

phase distribution networks. A low level control mechanism

and protection scheme is also developed, thus providing a

holistic design for building a prototype.

II. VOLTAGE FLUCTUATION AND COMPENSATION METHODS

A. Voltage fluctuation in distribution network

Voltage fluctuation is a usual phenomenon that happens in

the distribution grid. Traditionally power grids assume a

downstream power flow that results in a voltage drop along

the feeder causing undervoltage at feeder end [1]. To

counteract this effect OLTC mechanism in voltage regulators

and in sub-transmission transformers are used to set the

voltage at the feeder head at a higher value to compensate for

line drops.

However in recent years there has been a high penetration

of DG in the LV distribution network mainly driven by solar

and this is only expected to increase in the future [1]-[5]. This

combined with heavy load insertion to the grid such as electric

vehicle charging has made the voltage control more

complicated than before [2]. Large variation in DG power

owing to short and long term fluctuations in wind and

sunshine results in large amplitude (±10%) and frequent

variation in load voltage [2]-[4], [7]. Moreover, feeders

experience upstream power flow during high DG production

causing overvoltage at feeder ends [4], [5].

For the simple case of Fig. 1, the load voltage variation at

feeder end can be quantified by ΔV, which is the difference

between the voltage at the feeder head (E) and voltage at

feeder end (V): s lV=E V= Z I

*

load l(I ) l lS V P jQ (Eqn. 1.1)


V V

(Eqn. 1.2)

MV/LV

Feeder End

Feeder Head

Line imedance (Rs+jXs)

Feeder length

ΔV

Sload

Bidirectional power flow

Power flow

Power flow

Fig. 1 Voltage variation along feeder depending on power flow direction

The voltage variation thus depends on the effective

impedance of the line (Rs+jXs), apparent power

drawn/injected by the load (Sload=Pl+Ql) and voltage at feeder

end (V). The role of voltage regulating equipment will be to

control the above three parameters to ensure that ΔV is

minimum and the load voltage is within permissible limits.

In reality, the situation is much more complicated than

what is shown in Fig.1 due to:

1. Non-uniform distribution of loads/DG along the feeder

2. Time varying nature of the load/DG power

3. Uneven nature of feeder length emanating from bus bar

The negative impacts of overvoltage, undervoltage and the

large scale DG penetration have been addressed in [8], [9].

According to [9], the voltage at customer utilization point

cannot exceed the tolerance level of ±10%. Currently no

reliable and fast voltage control strategy is present for the

European distribution network. Off-load tap changing

mechanism is present on distribution transformers but they

cannot be operated without interrupting the supply. On the

other hand, conventional voltage regulators and OLTC sub-

transmission transformer use mechanical switches that can be

operated on-load. But under conditions of large scale voltage

fluctuation due to DG, they are unable to cope up with the

frequent tap changes as the switches undergo frequent wear

and tear due to the arcing phenomenon [6], [14].

Thus a comprehensive strategy is required for ensuring

permissible voltage levels throughout the distribution network.

It must be able to compensate frequent voltage fluctuations in

form of both over and undervoltage in a cost effective and

efficient manner. Some of these methods will be looked into

in the next section.

B. Voltage compensation methods- Series and shunt

Voltage compensation can be achieved through shunt and

series compensation [10], [11]:

1. Shunt compensation, where a lagging/leading current is

injected into the grid to control the voltage.

2. Series compensation, where a voltage is injected in series

to the existing grid voltage or a reactive element in

connected in series to line to modify the line impedance.

It can be seen from Eqn.1 that the reactive power Ql

flowing through the line has a direct correlation to the voltage

drop along the line. Shunt compensation is based on the fact

that injecting a leading current decreases the voltage at that

point while injecting a lagging current increases the voltage.

Example of shunt compensation is mechanically switched

capacitor/inductor banks or FACTS devices like thyristor-

controlled reactor (TCR) and thyristor-switched capacitor

(TSC) that act as a variable current source. The major

advantage is the low investment cost and the possibility to

provide power factor correction. The main drawback is that it

is ineffective for networks with low X/R ratio [10], [11]

resulting in increased losses and it requires a network revamp

to handle additional reactive current flows.

Fig. 2 shows a schematic of a European distribution

network having large PV penetration where a number of

feeders of equal capacity but unequal lengths emanate from

three distribution transformers. Assuming 5% loading and

100% PV injection, overvoltage can be observed at the ends

of long feeders due to reverse power flow. In Fig.3, shunt

compensation is implemented which requires increasing the

capacity of the long feeders to handle the reactive currents.

Fig. 2 Schematic of feeder system during high levels of PV injection

Fig. 3 Voltage regulation using reactive power compensation

Series compensation on the other hand works well for all

types of networks irrespective of the X/R ratio and does not

increase network losses; however the investment costs are

comparatively higher. Examples are self-commutated switch

based FACTS devices like Unity Power Flow Controller

(UPFC), on-load and off-load tap changing transformers. The

increased cost comes from the cost of power electronics and

required protection schemes in case of FACTS devices and

from the cost of transformer in case of tap changers.

C. Suitable strategy for European distribution network

The nature of the European distribution network in

generally characterised by feeders that can reach four to five

times the length of those in North America and are

predominantly underground cable networks with high R/X

ratio [13],[25]. In German LV grids R/X values of 2 are

typical for overhead lines and R/X values of 2.5 are typical for

cable networks [25]. This means that the voltage variation is

much higher in the European grid compared to the North

American grid. More importantly shunt compensation is less

effective due to the low X/R ratio. The best solution for

Europe is hence through series compensation.

The series compensation in the distribution network can be

implemented in two ways - one by centralized compensation

through OLTC distribution transformer or by a FACTS device

close to the distribution transformer. The second method is

through feeder specific decentralized compensation where a

series compensator is connected to those hot spot feeders that

have high PV power injection and/or long length. In the first

case, all feeders emanating from the transformer will have the

same amount of compensation. Because of the long lengths of

line in Europe, if the taps on the central transformer are set to

ensure that overvoltage at the feeder ends during high PV

injection is within limits, the voltage set point will be low at

around 0.95p.u. This will in turn result in undervoltage for

customer close to the feeder head, as shown in Fig. 4.

The second method through feeder specific compensation is

shown in Fig. 5, where a single compensator unit is connected

to the long feeder. It can be seen that this method is more

effective than the centralized compensation and permissible

Fig. 4. Voltage regulation using centralized full transformer

Fig. 5. Voltage regulation using feeder specific decentralized compensator

voltage level is ensured throughout the network. Further, the

compensator has to only be rated for the compensation power

of that feeder alone.

A comparative analysis of the voltage at the head and end

of the longest feeder for different voltage compensation

strategies is presented in Fig. 6, 7, 8 for the case of:

1. No PV power injection

PV power injection with:

2. No compensation

3. Full compensation (Shunt compensation)

4. Tap change on distribution transformer

5. Tap change on feeder

11:45 11:50 11:55 12:00 12:05 12:10 12:150

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PV Generation Profile

Time [hrs]

Act

ive

Po

we

r [p

u]

Fig. 6. PV generation profile as a function of time

Fig. 7. Voltage profile at the head of feeder as a function of time

Fig. 8. Voltage profile at the end of feeder as a function of time

Fig. 9. Connection of series compensator to MV and LV network

Fig. 7,8 shows that voltage at both feeder head and end

remain within limits of 0.02pu during shunt compensation and

tap change on feeder. Since shunt compensation requires a

network revamp (see Fig. 3), tap change on feeder is hence the

cost effective solution in the case of non-uniform distribution

of DG, loads and uneven feeder lengths; which are often

encountered in the European scenario.

Series compensation can be implemented using self-

commutated switch based FACTS devices or OLTC

transformer. Since FACTS have high losses and are expensive

owing to the use of filters and large number of semiconductor

switches, the aim will be to design an OLTC series

compensator. Fig. 9 shows different possibilities of

connecting such a device to the MV/LV distribution network.

III. NOVEL DESIGN OF OLTC AUTOTRANSFORMER

Traditional OLTC transformers and voltage regulators

using mechanical switches for taps are unable to cope up with

frequent voltage fluctuations and get worn out due to

repetitive tap changes. However the advantage of these

mechanical taps is their high overload capacity and very low

on-state voltage resulting in low steady state losses. On the

other hand tap changers using semiconductor switches do not

have any arcing problems and provide flexibility in operation

but they suffer from much higher steady state losses [6], [13].

By combining both types, hybrid switches are obtained [15],

[16] as shown in Fig. 10. The basic idea is to use the

mechanical switches in steady state and semiconductor

switches during tap change to ensure low steady state losses

and arc-free tap changing process. The high overload capacity

of mechanical switches can be made use if fault conditions

occur during steady state operation.

A. Novel design using no-load switch & single hybrid switch

A novel topology of OLTC is shown in Fig. 10 where an

autotransformer with taps is used on the load side is used. The

turn ratio of input to output is 10:11. If the rated input voltage

is 1p.u. then the taps on present on the section of windings

from 0.9p.u. to 1.1p.u. Ten taps are present each of 0.02p.u

and thus the OLTC can provide upto ±10% compensation.

The use of an autotransformer saves on material and cost of

and the throughput power is approx. ten times the transformed

power [24]. A combination of no-load switches and a single

hybrid switch is used to realize the OLTC mechanism. A no-

load switch is a mechanical switch that opens or closes under

no-load. By operating it under no-load, it does not have any

arcing phenomenon occurring. The idea is derived from

‘diverter switch’ type voltage regulators [6],[14],[17] where

two movable no-load switches referred to as ‘selector switch’

are used to select the taps and a mechanical ‘diverter’ switch

is used for the tap change process and for carrying the load

current in steady state.

Each tap of the transformer is connected to a no-load

switch and alternate no-load switches are connected to each

other (green and red). The hybrid switch consists of two

bidirectional semiconductor (electronic) switches connected to

the red & green taps and a single mechanical switch. For the

normal operation of the OLTC system, the following

conditions are imposed:

1. Mechanical switch M conducts the load current in steady

state and bidirectional electronic switches BS1 and BS2

are used for the tap changing process

2. At any point of time only one no-load switch amongst

green or red will be closed. This is to prevent the

occurrence of a short circuit between the taps and ensure

that the maximum voltage the no-load switches in OFF

condition will block is 0.2p.u.

3. Tap changes are always made in steps of one. This means

that if tap 2 is ON, then a tap change can be made only to

tap 3 or tap 1. This guarantees that the maximum voltage

across the hybrid switch will be equal to the voltage of

one tap i.e. 0.02pu.

Fig. 10. Novel series compensator using no-load and hybrid switch

The tap change process through the hybrid switch is done

in 7-step mechanism that is illustrated in Table 1, when we

move from tap corresponding to NL1 to NL2. In the table, ON

condition of switch is indicated by ‘1’ and OFF by ‘0’. In step

3, when the mechanical switch M is turned off, no arcing

occurs as the current automatically commutates to BS1. The

change of state of no-load switches always happens with the

series connected electronic switch (BS1 or BS2) in OFF

condition.

The OLTC system is customized for application in MV and

LV distribution network and Table 2 shows the corresponding

parameters. The no-load switches are rated for 1p.u. nominal

load current and 0.2p.u. voltage. The hybrid switch (M,

BS1,BS2) is rated for 0.02p.u. voltage and 1p.u. current. The

simple operation of no-load switches and the small voltage

TABLE I

7 STEP COMMUTATION BETWEEN TAPS

Step No load switch Hybrid switch

NL 1 NL 2 BS 1 M BS 2

0 1 0 0 1 0

1 1 1 0 1 0

2 1 1 1 1 0

3 1 1 1 0 0

4 1 1 0 0 1

5 1 1 0 1 1

6 1 1 0 1 0

7 0 1 0 1 0

TABLE II

PARAMETERS OF THE MV AND LV OLTC SYSTEM

MV LV

Voltage level (line-line, rms) 20 kV 400 V

Maximum rated load current (rms) 577A 577A

Neutral available No Yes

Connection scheme Open delta Wye

1-phase OLTC units required 2 3

Autotransformer transformed power 1.15 MVA 13.3 kVA

Autotransformer throughput power 12.65 MVA 146.3 kVA

Percentage impedance 2.5% 6%

Bidirectional electronic switch IGBT MOSFET

ratings of the electronic switches results in low overall cost of

OLTC.

A vital aspect of the novel design is its modular nature.

Modifying the total regulation range to ±15% or ±5%, or

changing the voltage per tap to 1.5% or 3%, can be realized by

simply adding/reducing the number of no-load switches and

still using only a single hybrid switch. It should be realized

that the cost of no-load switches does not increase with

increased voltage ratings unlike semiconductor switches. The

novel OLTC can be implemented on existing voltage

regulators like in [17].

B. Bidirectional electronic switch for hybrid tap changer

The electronic switch used in the hybrid switch must be

bidirectional - block both positive and negative voltages when

OFF and can conduct current of both polarities when ON.

Several possibilities exist for realising such a bidirectional

switch as discussed in [6]. In this design, series connection of

two back-to-back IGBT having an anti-parallel diode in

common emitter configuration is used for switch BS1 and

BS2. This is the because the commutation of current between

BS1 and BS2 (Step 4 of the 7-step tap change method) can be

done without the occurrence of a short/open circuit through

voltage polarity based four-step commutation [6], [18]-[20].

It takes upto few μs to commutate between the switches and

requires no current limiting impedances [6]. Voltage polarity

based commutation is preferred over the current polarity based

commutation because of lower failure hazard. In the latter

case, an erroneous current polarity judgement which typically

occurs around the zero crossing of the current will cause an

open circuit condition. The subsequent large overvoltages are

given by L*(di/dt), where L corresponds to the effective line

inductance and (di/dt) is the slope of the current turn off

transient. On the other hand for the voltage polarity method,

an error in judgement around the zero crossing of voltage will

lead to a short circuit of the tap. The failure is however less

hazardous due to the small voltage available for driving the

short circuit current.

For the LV OLTC, the electronic switch has to be rated for

577A current and 8V blocking capacity. Since commercially

available MOSFET are more economical than IGBT for such

ratings, back to back series connected MOSFETs will be used

for the LV scenario. The MOSFET body diode replaces the

anti-parallel diode used in case of IGBT.

C. Overvoltage snubber for hybrid switch

Normally during a change of tap, the current through the

transformer tap leakage inductance is interrupted in step4,

leading to an overvoltage [6], [21]. Fig. 11 can be used to

analyse this effect where the schematic of a single tap of an

OLTC is shown - Vtap, Lleak, Rleak are the voltage, the leakage

inductance and winding resistance of one tap with the load

modelled as a current source. Initially let BS1 be ON and the

voltage at C is (V1+Vtap). Now a tap change is made from BS1

to BS2 so that the load voltage at point C is V1. During this

process, let BS1 interrupt the load current with a slope

α=(di/dt) as the current commutates to switch BS2, then a

overvoltage is experienced by the turning-off switch BS1 due

to the tap leakage inductance given by

1BS tap leak

diV V L

dt

(Eqn. 2)

Fig. 11. Overvoltage snubber connected across the electronic switches

Thus a large overvoltage will be experienced by the switch

depending on the leakage inductance of the tap and the current

being turned off. Damped LC oscillations will be observed

due to the small but finite value of tap capacitance and a

damping due to the resistance of the tap windings. This

overvoltage can be very high and can permanently damage the

switch. This calls for the need for using an overvoltage

snubber to protect the switches. Alternatively if the tap change

is done close to zero current, then there will be minimum

overvoltage effect as given by Eqn 3.2.

A RC overvoltage snubber (Rsnub and Csnub) can be used

across the switches [6] as shown in Fig. 11. This results in the

energy stored in the inductor to be exchanged with the

snubber capacitor resulting in LC oscillations. If we neglect

the damping due to Rleak and Rsnub, we can write:

If 4 1|load stepI I ,

22

1

1 1( )

2 2leak snubL I C V

(Eqn. 3.1)

1/ leak snubV L C I (Eqn. 3.2)

1 14/ BS tap leak snubstep

V V L C I (Eqn. 3.3)

1/ 2 osc leak snubf L C (Eqn. 3.4)

Higher the value of Csnub and lower the value of Lleak, lower

is the overvoltage on the switch. Rsnub is designed mainly to

limit the capacitor discharge current at switch turn ON and to

damp the LC oscillations.

IV. CONNECTION SCHEME FOR THREE PHASE NETWORK

In this section, different possible connection schemes for

single phase OLTC transformers to a three phase distribution

network are examined. Based on the general nature of

European distribution network, the following assumptions are

made in the design of the connection scheme:

1. MV distribution network – three phase three wire

network where line-line voltage has to be regulated.

Unbalanced loading of phases is not an issue, therefore

independent regulation of phases is not required.

2. LV distribution network – three phase four wire network

with neutral present where phase voltage has to be

regulated. Unbalanced loading of phases is prevalent,

therefore independent regulation of each phase is required.

Two/three single phase OLTC can be connected to three

phase network in 4 ways [22], [23]:

1. Wye connection with star point floating

2. Closed-delta connection

3. Wye connection with star point connected to line

neutral (only for three phase four wire system)

4. Open-delta connection using two units

The first two methods suffer from the drawback that the

floating star point can lead to erratic operation of the tap

changer and overstress the winding insulation; while the

closed-delta connection does not result in in-phase

compensation and requires an extra unit compared to open-

delta connection [22].

A. Y connection for 3-phase 4-wire LV network

It is typical for LV European distribution network to have a

neutral available. Thus three single phase OLTC transformers

can be connected between the phase and neutral in grounded

Y formation with the start point of the transformer connection

connected to the neutral of the network [22], shown in Fig. 12.

The points S, SL and L correspond to those marked in Fig. 10.

The compensating voltage is derived from the phase voltage

and injected in/out phase for positive and negative

compensation. The main feature of the connection is that the

three OLTC units can achieve independent regulation of each

phase voltage. This is explained using the phasor diagram in

Fig. 14 where Vx, Vxy are the phase and line voltage at the

input of the transformer and ΔVx, ΔVxy are the corresponding

phase and line voltage that are series injected into the grid.

The phase voltage Vx’ after series compensation is given by: '

x x xV V V (Eqn. 4)

B. Open delta connection for 3-phase 3-wire MV network

An innovative method for controlling the line-line voltage

in a three wire network using only two OLTC units through an

open-delta connection is, shown in Fig. 13. The two units are

connected between phase a-b and phase c-b using phase b as

the common connection point [22]. The injected voltages ΔVa

and ΔVc are thus derived from the line-line voltages. Direct

and independent regulation of the line-line voltages Vab and

Vbc results, while the compensation in Vac i.e. ΔVac is average

of ΔVab and ΔVbc. During balanced operation (ΔVa = ΔVc), in-

phase compensation of all three line-line voltages occurs and

during unbalanced operation, Vab and Vbc experience in-phase

S LSeries

CompensatorSL

AB

C

ΔVa

ΔVb

ΔVc

N

S LSeries

CompensatorSL

S LSeries

CompensatorSL

A’

B’

C’

N

Fig. 12. Y connection of OLTC transformers to a 3-phase 4-wire network

S LSeries

CompensatorSL

AB

C

ΔVa

ΔVc

S LSeries

CompensatorSL

A’

B’

C’

Fig. 13. Open Δ connection of OLTC transformers to 3-phase 3-wire network

Fig. 14. Phasor diagram of series compensation in MV (left) and LV (right) distribution network using open-delta and Y connection respectively

compensation while Vac alone experiences a phase shift of

upto 5 . The phasor diagram is shown in Fig. 14. The line

voltage Vxy’ after series compensation is given by:

'

xy xy xyV V V (Eqn. 4)

V. SIMULATION OF PROPOSED OLTC SYSTEM

The MV and LV 3-phase OLTC system was modeled in

PLECS [26] based on the single line diagram shown in Fig. 15.

SourceFeeder

SeriesCompensator

Vx’

Load

Vx Vload_x

Fig. 15. Single line diagram of OLTC system simulated in PLECS

The feeder (line/cable) was modeled as a series connection

of two T-sections. The load for the MV scenario was a delta

connected impedance drawing nominal current at power

factors ranging between 0.8 lagging to 0.8 leading, while a

similar Y connected load were used for the LV case. DG

power injection was modeled using a current source as a load.

Steady state and transient operation of the system was

investigated and the waveform are shown in Fig. 16 and

described in Table 3. The 4-step commutation time step was

2µs. The impact of the snubber capacitance and current at

commutation instant as described by Eqn. 3.1-3.4 can be

examined through the waveforms in Fig. 16.4-16.7. A lower

snubber capacitance leads to large amplitude oscillations of

high frequency. The current at the commutation instant has a

direct correlation on the magnitude of the LC oscillations as

seen in Fig. 16.4 and 16.5.

TABLE III SIMULATION OF OLTC SYSTEM

Figure Description

16.1

Steady state line voltages [V] at input Vxy and output Vxy’

of OLTC as a function of time [s] for MV scenario. Each

OLTC unit is set for 4% and 8% positive compensation

respectively, resulting in 6% compensation in third line

voltage

16.2

Steady state phase voltages [V] at input Vx and output Vx’

of OLTC as a function of time[s] for LV case. The three

OLTC units are set for 8%, 6% and -2% compensation.

16.3

Steady state phase voltages [V] at input Vx and output Vx’

of OLTC as a function of time[s] when an active load is

connected at the load end of the LV OLTC system for -8%

compensation

16.4

IBS1, IBS2of bidirectional switch during 4-step

commutation from BS2 to BS1 when current at

commutation instant is 200A and snubber capacitance is

100 µF

16.5 IBS1, IBS2of electronic switch during 4-step commutation

from BS2 to BS1 when current at commutation instant is

600A and snubber capacitance is 100 µF

16.6

Transient voltage VBS1, VBS2 [V] and current IBS1, IBS2

[A] of bidirectional switch during 4-step commutation

from BS2 to BS1 when current at commutation instant is

200A and snubber capacitance is 10 µF

16.7

Transient voltage Vsnub [V] and current Isnub[A] of snubber

capacitor during 4-step commutation from BS2 to BS1

depicting the LC oscillation between leakage impedance

and snubber capacitance of 10 µF

VI. CONCLUSIONS

Frequent voltage fluctuations are observed in the

distribution network owing to large scale renewable energy

integration like PV. European distribution network is

characterized by long lines that have a low X/R ratio. The

most suitable voltage compensation solution for Europe under

conditions of uneven feeder length and DG distribution is

through the use of partially rated feeder specific series

compensation.

Voltage regulators using mechanical switches as OLTC

have reduced lifetime owing to the arcing phenomenon due to

frequent tap changes. Electronic switch based tap changers

have no arcing problem but higher steady state losses. Hybrid

switches combine the best of both – no/minimal arcing and

low steady state losses and hence are the suitable choice for

future design of OLTC taps.

A novel design for a series voltage compensator for the

European distribution network has been proposed. The design

has been customized for application in both LV and MV

distribution network. An autotransformer having taps on the

secondary side was used to build the OLTC. The taps were

made from a combination of no-load switches and a single

hybrid switch and exhibited several advantages. The hybrid

switch eliminated the arcing problems during tap change and

has low losses during steady state. The use of no-load

switches and the 7-step tap changing mechanism reduced the

voltage rating and the number of active hybrid switches. The

use of voltage polarity based 4-step commutation on back to

back connected IGBT/MOSFET provided a convenient

method for performing a tap change without the need for

current-limiting impedance.

A single overvoltage snubber connected across the hybrid

switch protects the switches from overvoltage resulting from

interruption of current through the tap leakage inductance.

Open-delta connection using two OLTC units and Y

connection using three OLTC units were most suitable for

MV and LV scenario respectively. The entire system was

simulated in the PLECS environment and both the steady state

and transient operation of the system was verified.

ACKNOWLEDGMENT

The authors would like to acknowledge the guidance and

support of Ir.V.Prasanth and Ir.T.Todorcevic, both PhD

students from the Electrical Power Processing (EPP) group,

TU Delft.

REFERENCES

[1] Esslinger, P., Witzmann, R., “Improving grid transmission capacity

and voltage quality in low-voltage grids with a high proportion of

distributed power plants”, Energy Procedia, 12, pp. 294-302., 2012 [2] Woyte, A.; Van Thong, V.; Belmans, R.; Nijs, J.; , "Voltage

fluctuations on distribution level introduced by photovoltaic systems," Energy Conversion, IEEE Transactions on , vol.21, no.1, pp.

202- 209, March 2006

[3] Demirok, E.; Sera, D.; Teodorescu, R.; Rodriguez, P.; Borup, U.; ,

"Clustered PV inverters in LV networks: An overview of impacts and

comparison of voltage control strategies," Electrical Power & Energy

Conference (EPEC), 2009 IEEE , vol., no., pp.1-6, 22-23 Oct. 2009 [4] Masters, C.L.; , "Voltage rise: the big issue when connecting embedded

generation to long 11 kV overhead lines," Power Engineering Journal ,

vol.16, no.1, pp.5-12, Feb. 2002 [5] Kojovic, L.; , "Impact of DG on voltage regulation," Power

Engineering Society Summer Meeting, 2002 IEEE , vol.1, no., pp.97-

102 vol.1, 25-25 July 2002 [6] J.Faiz, B.Siahkolah, "Electronic Tap-changer for Distribution

Transformers", Springer, 2011

[7] E. Wiemken, H.G. Beyer, W. Heydenreich, K. Kiefer,; “Power characteristics of PV ensembles: experiences from the combined power

production of 100 grid connected PV systems distributed over the area of Germany”, Solar Energy, Volume 70, Issue 6, 2001, Pages 513-518

[8] Standard EN 50160 - Voltage characteristics of electricity supplied by

public distribution networks, CENELEC, 2010 [9] R.Passey, T.Spooner, I. MacGill, M.Watt, K.Syngellakis,; “The

potential impacts of grid-connected distributed generation and how to

Fig. 16.1 [Vab, Vab’, Vbc, Vbc’, Vca, Vca’]

Fig 16.7 [Vsnub, Isnub]

Fig 16.6 [VBS1, IBS1, VBS2, IBS2]


Fig. 16.2 [Va, Va’, Vb, Vb’, Vc, Vc’] Fig. 16.3 [Va, Va’, Vloada ]


address them: A review of technical and non-technical factors”, Energy Policy, Volume 39, Issue 10, Pages 6280-6290, Oct. 2011

[10] T. J. Miller, “Reactive Power Control in Electric Systems”. New York:

Wiley, 1982 [11] Dixon, J.; Moran, L.; Rodriguez, J.; Domke, R.; , "Reactive Power

Compensation Technologies: State-of-the-Art Review," Proceedings of

the IEEE , vol.93, no.12, pp.2144-2164, Dec. 2005 [12] Tom Short, "Electric Power Distribution Equipment and Systems",

Taylor & Francis, Nov 2005

[13] Bauer, P.; de Haan, S.W.H.;, "Electronic tap changer for 500 kVA/10 kV distribution transformers: design, experimental results and impact

in distribution networks," Industry Applications Conference, 1998.

Thirty-Third IAS Annual Meeting. The 1998 IEEE , vol.2, no., pp.1530-1537 vol.2, 12-15 Oct. 1998

[14] “On-Load Tap-Changers for Power Transformers - A Technical

Digest”, Maschinenfabrik Reinhausen GmbH, 2009 [15] Gao D, Lu Q, Luo JA, “New Scheme for on-load tap-changer of

transformers”, International Conference on Power System Technology,

pp 1016–1020, 2002 [16] Cooke GH, Williams KT, “New thyristor assisted diverter switch for

on-load transformer tap-changers”, IEE Proc Electric Power Appl , pp

507–511, 1992 [17] Das, D.; Divan, D., "Power flow control in networks using controllable

network transformers," Energy Conversion Congress and Exposition,

2009. ECCE 2009. IEEE , pp.2224,2231, 20-24 Sept. 2009 [18] Bauer P, Schoevaars R, “Bidirectional switch for a solid-state tap-

changer”, IEEE PESC 03, pp 466–471, 2003

[19] A. Alesina, M. Venturini, “Analysis and Design of Optimum-Amplitude Nine-Switch Direct AC-AC Converters”, IEEE

Transactions on Power Electronics, Vol. 4, no. 1, pp. 101-112, January

1989 [20] J. Mahlein, J. Igney, M. Braun, O. Simon, “Robust Matrix Converter

Commutation without explicit Sign Measurement,” Proceedings of

EPE 2001 , pp. 1-7, 2001 [21] Zai FQY, O’Kelly D, “Solid-State on-load transformer tap-changer”,

IEE Proc Electric Power Appl,143(6):481–491, 1996

[22] James H.Harlow, “Electric Power Transformer Engineering”, 2nd Edition, Taylor & Francis Group, 2006

[23] Bishop, M.T.; Foster, J.D.; Down, D. A., "The application of single-phase voltage regulators on three-phase distribution systems," Rural

Electric Power Conference, 1994. Papers Presented at the 38th Annual

Conference , vol., no., pp.C2/1,C2/7, 24-26 Apr 1994 [24] John J. Winders Jr., “Power Transformers: Principles And

Applications”, Taylor & Francis, 2002

[25] Esslinger, Peter, “Studie Q(U) - Schlussbericht”, Kooperationsprojekt der EnBW Regional AG, der E.ON Bayern AG, der SMA Solar

Technology AG, der KACO new energy GmbH, der Siemens AG und

des Fachgebiets Elektrische Energieversorgungsnetze der Technischen Universität Münche, 31 August 2012

[26] PLECS® version 3.3, Plexim GmbH, Zurich, Switzerland,

http://www.plexim.com


http://www.plexim.com/

design of a power electronic assisted series …

Documents