Professional Documents
Culture Documents
Fault Diagnosis and Protection of Phase Shifting Transformers
Fault Diagnosis and Protection of Phase Shifting Transformers
University of Coimbra
Faculty of Sciences and Technology
Department of Electrical and Computer Engineering
Fault Diagnosis and Protection of Phase‐Shifting Transformers
Pedro Miguel Machado Apóstolo
A dissertation presented for the degree of
Master of Science in Electrical and Computer Engineering
Coimbra, October 2014
Fault Diagnosis and Protection of Phase‐Shifting Transformers
Supervisor:
Prof. Doutor Sérgio Manuel Ângelo da Cruz
Jury:
Prof. Doutor Jaime Batista Mendes
Prof. Doutor André Manuel dos Santos Mendes
Prof. Doutor Sérgio Manuel Ângelo da Cruz
Report written for the dissertation course, included in the
Electrical and Computer Engineering Course, submitted in partial fulfillment
for the degree of Master of Science in Electrical and Computer Engineering.
Coimbra, October 2014
Acknowledgments
I would like to express my sincere gratitude to Dr. Sérgio Manuel Ângelo da Cruz, my supervisor, for
his guidance and support throughout this thesis. I am also extremely grateful for all the opportunities that
he presented me and I would like to emphasize that his ideas and his support were remarkably crucial to
my work.
Many thanks to the resources and equipment provided by Instituto de Telecomunicações (Pólo de
Coimbra), and the Department of Electrical and Computer Engineering, as without them, this thesis would
not have been possible.
I also would like to thank my parents, my brother Luís, my girlfriend Rita and my friend Daniel for all
the support, motivation, patience and understanding, not just during this period but also during all my
years of studies.
Last but not least, I would like to thank my laboratory and faculty colleagues for the all the support
and companionship that helped me to carry on during the course of this journey.
Page | i
Abstract
This thesis summarizes the some of the existent differential protections for phase‐shifting
transformers, describes the adaptation and use of an unpublished simulation model on Simulink that
mimics an existent transformer and also the implementation of an offline differential protection system.
In this thesis we also approach the subject of fault diagnostics in phase‐shifting transformers due to the
fact that no previous information regarding this subject was found during the bibliography survey carried
out. In order to fulfill this gap, in this work we propose a new diagnostic tool based on the Extended Park’s
Vector Approach that uses the differential currents and differential active and reactive powers in order to
detect turn‐to‐turn faults in the windings of phase‐shifting transformers, more specifically on the delta‐
hexagonal type.
In chapter 1, an introduction to phase‐shifting transformers is given. It is also established a
comparison between standard and phase‐shifting transformers. Furthermore, an explanation of the basic
principles of phase‐shifting transformers is given, as well as their role and importance in the power
system.
In chapter 2 it is presented in more detail the application of phase‐shifting transformers, along with
their most common designs, types and characteristics. In this chapter it is also chosen the delta‐hexagonal
phase‐shifting transformer as the type our work is going to be based on.
In chapter 3 it is presented an overview of the different types of faults that may appear in power
transformers. Later on, the subject of protections is approached, highlighting the differential type and
presenting some of the available differential types of protection for phase‐shifting transformers. The
chapter finishes with the explanation of the methods available for diagnosis of faults in core and shell‐
type transformers and the introduction of a new diagnostic method for phase‐shifting transformers based
on its differential currents and the Extended Park’s Vector Approach. This method uses the differential
current and active and reactive power Park’s Vector components and the EPVA to determine the phase
affected by the fault as well as an indication of its extent regardless of the tap position.
Chapter 4 describes the use and adaptation of a new simulation model used to replicate the
experimental transformer. The simulation model is based on the duality between electric and magnetic
circuits and accounts for the core dimensions and hysteretic, electric and magnetic properties of an
existing 3‐phase transformer with standard winding connections. Also in this chapter, we further describe
and explain the necessary adaptations to simulate the delta‐hexagonal phase‐shifting transformer in
which this work is based on as well as detailed information regarding the segmentation of the windings
that was used to emulate the tap‐changers.
Chapter 5 starts by validating the simulation results by establishing a comparison between the
simulation data and the transformer in the laboratory. Furthermore, the obtained results in terms of
Page | ii
differential protection and fault diagnosis are presented. For this, several results are presented regarding
faults in different windings and under different working conditions and a comparison between them is
established.
In Chapter 6 the main conclusions of this work are presented and a number of topics which can be
addressed in future work is proposed.
Page | iii
Resumo
Esta tese faz um sumário de algumas das proteções diferenciais existentes para transformadores
desfasadores e descreve a adaptação e uso de um modelo de simulação em Simulink não publicado
baseado num transformador existente. Descreve também a implementação de um sistema de proteção
diferencial offline para o transformador. Nesta tese é também abordada a temática do diagnóstico de
avarias em transformadores desfasadores devido ao facto de nada ter sido encontrado sobre a mesma
durante a pesquisa bibliográfica realizada. Para suprir esta lacuna, neste trabalho propomos uma nova
ferramenta de diagnóstico baseada no Extended Park’s Vector Approach que usa as correntes diferenciais
bem como as potências activas e reactivas diferenciais a fim de detectar curto‐circuitos entre espiras nos
enrolamentos dos transformadores desfasadores, mais especificamente, nos do tipo delta‐hexagonal.
No capítulo 1, é feita uma introdução aos transformadores desfasadores. Neste capítulo é também
estabelecida uma comparação entre transformadores standard e transformadores desfasadores e são
explicados os princípios básicos dos transformadores desfasadores bem como o seu papel e importância
no sistema de energia.
No capítulo 2 apresenta‐se em mais detalhe o uso e a aplicação prática dos transformadores
desfasadores juntamente com os seus designs, tipos e características mais comuns. Neste capítulo é
também selecionado o transformador desfasador delta‐hexagonal como o tipo em que o nosso trabalho
vai ser baseado.
No capítulo 3 é apresentada uma visão geral dos diferentes tipos de avarias que podem surgir em
transformadores de potência. De seguida, é abordada a temática das proteções, com destaque para as do
tipo de diferencial e são apresentados alguns tipos de proteções diferenciais disponíveis para os
transformadores desfasadores. O capítulo termina com a explicação dos métodos disponíveis para o
diagnóstico de avarias em transformadores do tipo core e do tipo shell e com a introdução de um novo
método de diagnóstico para transformadores desfasadores baseado nas suas correntes diferenciais e no
Exteded Park’s Vector Approach. Este método utiliza as componentes do Vector de Park e o EPVA
aplicados às correntes diferenciais e às potências activas e reactivas diferenciais para determinar a fase
dos enrolamentos afetados pela avaria e providenciar uma indicação relativa da sua extensão,
independentemente da posição do comutador de tomadas.
O Capítulo 4 descreve o uso e adaptação de um novo modelo de simulação usado para replicar o
transformador experimental. O modelo de simulação baseia‐se na dualidade entre os circuitos eléctrico
e magnético e tem em conta as dimensões do núcleo e as propriedades histeréticas, eléctricas e
magnéticas de um transformador trifásico do tipo standard existente no laboratório. Também neste
capítulo são ainda descritas e explicadas as adaptações necessárias para simular o transformador
desfasador do tipo delta‐hexagonal em que este trabalho é baseado, e são fornecidas informações
detalhadas sobre a segmentação dos enrolamentos criada para emular os comutadores de tomadas.
Page | iv
O Capítulo 5 começa com uma validação dos resultados da simulação, estabelecendo a comparação
entre os dados da simulação e os resultados experimentais. De seguida, são apresentados os resultados
obtidos em termos de proteção diferencial e diagnóstico de avarias. Para isto, foram apresentadas vários
resultados relativos às avarias em diferentes enrolamentos, sob diferentes condições de carga e
desfasamento e foi estabelecida uma comparação entre eles.
No Capítulo 6 são apresentadas as principais conclusões deste trabalho e é também proposto um
conjunto de temas que poderão ser abordados em trabalhos futuros.
Page | v
Table of Contents
ACKNOWLEDGMENTS I
ABSTRACT II
RESUMO IV
TABLE OF FIGURES IX
ACRONYMS XII
LIST OF SYMBOLS XIV
CHAPTER 1 – INTRODUCTION 1
CHAPTER 2 – PHASE‐SHIFTING TRANSFORMERS 6
2.1 – APPLICATION PRINCIPLES OF PHASE‐SHIFTING TRANSFORMERS 6
2.2 – TYPES OF PHASE‐SHIFTING TRANSFORMERS 9
2.2.1 – Single‐Core Design 9
2.2.1.1 – The Asymmetrical Type 10
2.2.1.2 – The Symmetrical Type 11
2.2.1.3 – Delta‐Hexagonal 12
2.2.2 – Two‐Core Design 12
2.2.3 – Special Configurations (Quadrature Boosters) 14
2.3 – SELECTION OF THE DELTA‐HEXAGONAL PHASE‐SHIFTING TRANSFORMER 15
CHAPTER 3 – FAULTS, PROTECTIONS AND DIAGNOSTICS IN TRANSFORMERS 17
3.1 – INTRODUCTION 17
3.2 – TYPES OF FAULTS 18
3.2.1 – Overheating, Overpressure and Tank Faults 18
3.2.2– Phase‐to‐Ground Faults 19
3.2.3 – Phase‐to‐Phase Faults 19
3.2.4 – Turn‐to‐Turn Faults 19
3.2.5 – Iron Core Faults 20
3.2.6 – On‐Load Tap‐Changer Faults 21
3.3 – GENERAL CONSIDERATIONS ON PROTECTIONS 21
3.4 – DIFFERENTIAL PROTECTION OF TRANSFORMERS 22
3.5 – DIFFERENTIAL PROTECTION OF PHASE‐SHIFTING TRANSFORMERS 25
Page | vi
3.5.1 – Introduction 25
3.5.2 – Differential Protection for Delta‐Hexagonal Phase‐Shifting Transformers 25
3.5.2.1 – Differential Principle for Electrically Connected Windings 25
3.5.2.2 – Protection Methods for Delta‐Hexagonal Phase‐Shifting Transformer by Kastenny 26
3.5.2.3 – Universal Differential Protection Method for Arbitrary Three‐Phase Transformers 28
3.6 – DIAGNOSIS OF FAULTS IN CORE AND SHELL‐TYPE TRANSFORMERS 28
3.6.1 – Extended Park’s Vector Approach 29
3.6.2 – A New Differential Current Extended Park’s Vector Approach for Phase‐Shifting
Transformers 30
3.6.3 – Active and Reactive Differential Power Analysis 31
CHAPTER 4 – SIMULATION MODEL 32
4.1 – DEVELOPED MODEL 32
4.1.1 – Electric Circuit 32
4.1.2 – Magnetic Circuit 34
4.2 – MODEL PARAMETERIZATION AND ADJUSTMENTS 37
4.2.1 ‐ Primary Winding Segmentation 38
4.2.2 ‐ Secondary Winding Segmentation 40
4.2.3 – Testing of Segmentation 40
CHAPTER 5 – ANALYSIS OF RESULTS 42
5.1 – SIMULATION MODEL VALIDATION 42
5.1.1 – Phase Angle Shift at Full Tap 42
5.1.2 ‐ Experimental Results Comparison 42
5.1.2.1 ‐ Measurement nº1 43
5.1.2.2 – Measurement nº2 43
5.2 – PROTECTION AND FAULT DIAGNOSIS RESULTS 45
5.2.1 – Differential Protection Results 45
5.2.1.1 – Healthy Operation 46
5.2.1.2 – Turn‐to‐turn Fault in the Phase A of the Excitation Windings 47
5.2.1.3 – Turn‐to‐turn Fault in the Phase A of the Series Windings 48
5.2.2 – Diagnosis of Faults 49
5.2.2.1 – Healthy Conditions 49
5.2.2.2 – Turn‐to‐turn Fault in the Phase A of the Excitation Windings 54
5.2.2.3 – Turn‐to‐turn Fault in the Phase A of the Series Windings 56
5.2.2.4 – Turn‐to‐turn Fault on Phase A, B and C of the Excitation Windings 58
CHAPTER 6 – CONCLUSIONS 61
Page | vii
REFERENCES 63
APPENDIX A – SCHEMATIC REPRESENTATION OF THE QUADRATURE BOOSTER 66
APPENDIX B – EXPERIMENTAL SETUP 67
B.1 – TEST BENCH AND DATA ACQUISITION SYSTEM 67
B.2 – TRANSFORMER 68
B.3 ‐ CRIO 69
B.4 – TRANSDUCERS 70
APPENDIX C – SIMULATION WAVEFORMS FOR THE DHPST AT D=0 UNDER HEALTHY CONDITIONS
71
C.1 – SIMULATION OF THE DHPST FOR THE MIDDLE‐TAP POSITION (D=0), IN NO‐LOAD CONDITIONS 71
C.2 – SIMULATION OF THE DHPST FOR THE MIDDLE‐TAP POSITION (D=0), WITH A 42Ω/PHASE RESISTIVE LOAD 72
APPENDIX D ‐ COMPARISON BETWEEN SERIES AND EXCITATION FAULTS ON A TRANSFORMER
WITH EQUAL SIZED SERIES AND EXCITATION WINDINGS. 73
Page | viii
Table of Figures
Figure 1 – Schematic representation of a basic power network [3]. ................................................... 1
Figure 2 – Interconnection of the power networks of the Netherlands, Germany and Belgium [11]. 2
Figure 3 – Parallel connection between points A and B with a PST. .................................................... 3
Figure 4 – 450 MVA, 400/150/130 kV Phase‐shifting autotransformer, in a substation in Pedralva,
Portugal [26]. ........................................................................................................................................ 4
Figure 5 ‐ Parallel connection between two points (A and B). ............................................................ 6
Figure 6 ‐ An additional imposed voltage, ∆U, drives a current IX. ...................................................... 7
Figure 7 – Internal configuration of a phase‐shifting transformer (two‐core symmetrical type). ...... 8
Figure 8 – PST designs and types. ........................................................................................................ 9
Figure 9 – Schematic representation and phasor diagram of a single‐core asymmetrical type PST. 11
Figure 10 ‐ Schematic representation and phasor diagram of a single‐core symmetrical type PST. 11
Figure 11 – Schematic representation and phasor diagram of a delta‐hexagonal single‐core PST. .. 12
Figure 12 – Two‐core PST, 600 MVA with a phase angle shift of ±40⁰, for operation in Slovenia. ... 13
Figure 13 ‐ Phasor diagram of the two‐core symmetrical PST. .......................................................... 14
Figure 14 ‐ Detailed schematic representation and phasor diagram of the DHPST used in this
work……………………………………………………………………………………………………………………………………………...15
Figure 15 ‐ Transformer stoppage per specific component failure [20]. ........................................... 17
Figure 16 ‐ Illustration of turn‐to‐turn, phase‐to‐phase and phase‐to‐ground faults. ...................... 18
Figure 17 – Appearance of an hot‐spot due to the degradation of the insulation between the core
laminations. ........................................................................................................................................ 20
Figure 18 ‐ Differential protection scheme for a single‐phase transformer. ..................................... 23
Figure 19 – Differential relay tripping characteristic. ........................................................................ 24
Figure 20 – Energizing inrush current for a single‐phase transformer obtained from the simulation
model. ................................................................................................................................................ 24
Figure 21 ‐ Implementation of the primary circuit of the transformer. ............................................ 33
Figure 22 – Implementation of the secondary circuit of the transformer. ........................................ 33
Figure 23 – Simulink blocks of the primary and secondary: (a) CSB; (b) VSB. ................................... 34
Figure 24 – Representation of a brick of the core [28]. ..................................................................... 35
Figure 25 – Implementation of the model of the magnetic core of the transformer in Simulink [28]
……………………………………………………………………….…………..…………………………………………………………………36
Figure 26 – Equivalent magnetic circuit for the core type transformer [28]. .................................... 36
Figure 27 – Simulink model of the DHPST with a resistive load. ....................................................... 37
Page | ix
Figure 28 – Segmented primary winding schematic for: (a) middle tap position (D=0); (b) full retard
tap position (D=1); (c) advanced tap position (‐1<D<0). .................................................................... 38
Figure 29 – Detailed schematic of the series winding with 3 segments in the Simulink model. ....... 39
Figure 30 ‐ Series Winding segmentation with a turn‐to‐turn fault in the Simulink model. ............. 40
Figure 31 ‐ Segmented EW with a turn‐to‐turn fault at the bottom with nX short‐circuited turns. .. 40
Figure 32 – Current measurement for the testing of segmentation: (a) Excitation current, I0; (b) Post
fault winding current, I1, and current on the short‐circuit resistance, Icc. ......................................... 41
Figure 33 – Measurements of source and excitation currents, and load voltages for no‐load
conditions at full tap retard (D=1): (a) experimental results; (b) and simulation results. ................. 43
Figure 34 – Measurements of source, load and excitation currents, and load voltages for a 42Ω/phase
resistive load at an advance tap position (D=‐0,3): (a) experimental results; (b) and simulation results.
............................................................................................................................................................ 44
Figure 35 – Differential currents for the no‐load operation of the DHPST in healthy conditions: (a)
advanced tap position (D=‐0.3); (b) full retard tap position (D=1). ................................................... 46
Figure 36 – Differential currents for the on‐load operation, with a 42 Ω/phase resistive load, of the
DHPST in healthy conditions: (a) advanced tap position (D=‐0.3); (b) full retard tap position (D=1).46
Figure 37 – Differential currents for the no‐load operation of the DHPST with a turn‐to‐turn fault in
the excitation winding of phase A: (a) advanced tap position (D=‐0.3); (b) full retard tap position
(D=1). .................................................................................................................................................. 47
Figure 38 – Differential currents for the on‐load operation, with a 42 Ω/phase resistive load, of the
DHPST with a turn‐to‐turn fault in the excitation winding of phase A: (a) advanced tap position (D=‐
0.3); (b) full retard tap position (D=1). ............................................................................................... 47
Figure 39 ‐ Differential currents for the no‐load operation of the DHPST with a turn‐to‐turn fault in
the series winding of phase A: (a) advanced tap position (D=‐0.3); (b) full retard tap position (D=1).
............................................................................................................................................................ 48
Figure 40 ‐ Differential currents for the on‐load operation, with a 42 Ω/phase resistive load, of the
DHPST with a turn‐to‐turn fault in the series winding of phase A: (a) advanced tap position (D=‐0.3);
(b) full retard tap position (D=1). ....................................................................................................... 48
Figure 41 ‐ Experimental results for the Park’s Vector of the differential current regarding multiple
tap positions: a) no‐load; b) 42Ω resistive load per phase. ............................................................... 49
Figure 42 ‐ Simulation results for the Park’s Vector of the differential current regarding multiple tap
positions: a) no‐load; b) 42Ω resistive load per phase. ..................................................................... 50
Figure 43 ‐ EPVA, PDIFF and QDIFF spectrums for the experimental results obtained at no‐load, under
healthy conditions regarding D=‐0.3 and D=1 tap positions. ............................................................ 52
Figure 44 ‐ EPVA, PDIFF and QDIFF spectrums for the experimental results obtained with 42 Ω/phase
resistive load under healthy conditions regarding D=‐0.3 and D=1 tap positions. ............................ 53
Page | x
Figure 45 ‐ Experimental results for the Park’s Vector of the differential current with a 5% turn‐to‐
turn fault in the excitation winding of phase A for D=‐0.3 and D=1: a) no‐load; b) 42Ω resistive load
per phase. ........................................................................................................................................... 54
Figure 46 ‐ EPVA, PDIFF and QDIFF spectrums for the experimental results obtained at no‐load, with a
turn‐to‐turn fault of 5% on the EWA regarding D=‐0.3 and D=1 tap positions. ................................. 55
Figure 47 ‐ Experimental results for the Park’s Vector of the differential current with a 5% turn‐to‐
turn fault in the series windings of phase A for D=‐0.3 and D=1: a) no‐load; b) 42Ω resistive load per
phase. ................................................................................................................................................. 56
Figure 48 ‐ Simulation Results for the Park’s Vector of the differential current with a 5% turn‐to‐turn
fault in the excitations windings of phases A, B and C for D=‐0.3 and D=1 under no‐load and loaded
conditions. .......................................................................................................................................... 59
Figure 49 ‐ Schematic representation of the Quadrature Booster. ................................................... 66
Figure 50 – Experimental setup of the DHPST with measurement apparatus. ................................. 67
Figure 51 – Transformer used in the experimental setup and accessible tapped winding detail. .... 69
Figure 52 – Transducer (measurement box). ..................................................................................... 70
Figure 53 ‐ Simulation waveforms of the DHPST for the middle‐tap position (D=0) for no‐load
conditions under healthy conditions. ................................................................................................ 71
Figure 54 ‐ Simulation waveforms of the DHPST for the middle‐tap position (D=0) for on‐load
conditions, under healthy conditions. ............................................................................................... 72
Figure 55 – PV plot of a turn‐to‐turn fault on the SWs and EWs for a transformer with n1= n2=220.
………………………………………………………………………………………………………………………………………………………73
Page | xi
Acronyms
CB Circuit Breaker
CSB Current Source Block
CT Current Transformer
CTr Current Transducer (measurement box)
DGA Dissolved Gas Analysis
DHPST Delta‐Hexagonal PST
DP Differential Protection
DPMCW Differential Principle for Electrically Connected Windings
EDE Electromagnetic Differential Equations
EW Excitation Winding
FACTS Flexible AC Transmission System
FFT Fast‐Fourier Transform
FSx Faulted Segment X
HV High Voltage
L‐side Load Side
LCDP Linear Current Differential Principle
NLCDP Non‐Linear Current Differential Principle
NLTC No‐load Tap‐Changer
OLTC On‐Load Tap‐Changer
PAR Phase Angle Regulator
PCP Phase Comparison Principle
PDP Power Differential Principle
PF Power Factor
Ph‐g Phase‐to‐Ground
Ph‐ph Phase‐to‐Phase
PST Phase‐Shifting Transformer
PV Park’s Vector
QB Quadrature Booster
SDPST Standard Delta PST
SF Severity Factor
SLCDP Simplified Linear Current Differential Principle
S‐side Source Side
Page | xii
SU Series Unit
SW Series Winding (also known as Regulating or Tapped Winding)
TTT Turn‐to‐Turn
VSB Voltage Source Block
VT Voltage Transformer
Page | xiii
List of Symbols
D Tap position 1 D 1
FDIFF Differential signal
FRES Restraining signal
I0 Excitation current A
ICC Current on the short‐circuit resistance A
'
ICC Current on the short‐circuit resistance referred to the primary of the transformer A
ID ,IQ Park’s Vector components of the transformer’s currents A
IDIFF Differential current A
IEX Excitation current in phase x A
IF Fault current A
ILX Load current in phase x A
IP Pickup current/signal
IPI Pickup current/signal for iron losses
IRES Restraining current A
ISX Source current in phase x A
IX Superimposed current by the PST A
IZ Current of the line z A
I 'Z Sum of the superimposed current by the PST and the current in line z. A
Iph Phase current A
K Compensation factor of the restraining current
LLP Primary winding leakage inductance H
LLS Secondary winding leakage inductance H
LLX Total inductance of the series winding H
LLX 1 Inductance of the upper series winding segment in the 3 segment case (VSB Block) H
LLX 2 Inductance of the lower series winding segment in the 3 segment case (VSB Block) H
LLXD Inductance of the middle series winding segment in the 3 segment case (CSB Block) H
M( 0 ) Phase angle shift compensation matrix
N Series‐to‐excitation windings turn ratio
NP Number of turns of the primary winding
Page | xiv
NS Number of turns of the primary winding
P Active power W
PDIFF Differential active power given by PS PL W
PL 3‐phase active power on the L‐side W
PS 3‐phase active power on the S‐side W
PF Power factor
Q Reactive power VAr
QDIFF Differential reactive power given by QS QL VAr
QL 3‐phase reactive power on the L‐side VAr
QS 3‐phase reactive power on the S‐side VAr
RP Primary winding resistance Ω
RS Secondary winding resistance Ω
RX Total resistance of the series winding Ω
RX 1 Resistance of the upper series winding segment in the 3 segment case (VSB Block) Ω
RX 2 Resistance of the lower series winding segment in the 3 segment case (VSB Block) Ω
RXD Resistance of the middle series winding segment in the 3 segment case (CSB Block) Ω
SN Rated power VA
SFP Severity factor for the active power losses
SFQ Severity factor for the reactive power losses
UL Load‐side voltage V
ULX Load voltage in phase x V
UN Rated voltage V
US Source‐side voltage V
USX Source voltage in phase x V
U ph Phase voltage V
Z EX Impedance of the excitation windings Ω
ZL Line Impedance Ω
ZX Impedance of the total series winding of phase X Ω
Z X1 Impedance of the upper series winding segment of phase X Ω
ZX2 Impedance of the lower series winding segment of phase X Ω
Z XD Impedance of the series winding segment between the two tap‐changers of phase X Ω
n1 Total number of turns of the series windings
Page | xv
n2 Total number of turns of the excitation windings
nx Number of shorted turns in the winding
UPST Driving voltage imposed by the PST V
e Estimated angle of the major axis of the differential current Park’s Vector deg
B Base Park’s Vector angle for D=0 at no‐load deg
Phase angle shift between source and load deg
0Y Zero‐sequence of the magnetic flux in column Y of the transformer Wb
Y Magnetic fluxes in column Y of the core of the transformer Wb
Page | xvi
Chapter 1 – Introduction
Facing the increase of the world energy demand nowadays, there is a need to match the energy
production demand. This translates into a compulsory rise in electrical energy production and, hence, its
transportation through the power lines [1].
Unlike long ago, power generating plants, are often located based on availability of fuel and
connectivity to the high‐voltage grid, rather than on the location of the targeted load [2]. This change
occurred due to the development of very‐high voltage (VHV), extra‐high voltage (EHV) and even ultra‐
high voltage (UHV) connections and also due to the development of the grid both in terms of complexity,
connections and the constantly increasing number of small and large producers.
The power network can be roughly divided in generation, transmission, distribution and consumers.
A schematic representation of a basic power network is shown below.
Figure 1 – Schematic representation of a basic power network [3].
Generators are usually built in the range of 11‐25 kV. These are connected to the transmission grid,
which operates at voltages up to 765 kV, by the use of step‐up transformers. The connection between the
transmission grid and the distribution grid and between the distribution grid and consumers is commonly
1
carried out by the use of step‐down transformers [4]. The use of these high‐voltage connections allows
the reduction of losses through the reduction of the line currents [5]. Transformers are then the link
between apparatus or between lines of different voltage levels. These are highly efficient (nearly 100%)
and reliable [6]. Transformers of standard connections have a transformation ratio that reflects the ratio
between the input and output voltages. They have also a phase shift between input and output voltages,
which is a fixed value and multiple of 30 degrees, known as vector group.
Considering the higher voltage levels of the transmission network, other type of transformers have
been used in the late few years in order to provide power flow regulation. These are called phase‐shifting
transformers (PSTs) and unlike standard transformers, they provide power flow regulation by allowing
continuous variation of the phase‐shift voltage between their input and output. Their ability to advance
or delay the phase angle on the line they are inserted affects how power flows in the network [7, 8]. PSTs
offer a reliable, complete, and more economical solution for the control of power flow as compared to
FACTS devices [9]. Furthermore, they are among the most expensive types of transformers [10].
The transmission network can serve the purpose of connecting the producer A to consumer B, which
can be sometimes very far apart, and not necessarily in the same country, or establish an interconnection
between two or more separate systems. The interconnection between the Netherlands, Germany and
Belgium networks is an example of systems interconnection, as it is possible to observe in Figure 2.
Figure 2 – Interconnection of the power networks of the Netherlands, Germany and Belgium [11].
Consider the first situation, (Figure 3), where the transmission network connects points A and B. If
these two points are connected by two or more lines, uneven loading of the transmission lines may occur.
This may be caused by parallel lines with different impedances or variations in terms of generation output,
loads and load power factor [11, 12].
Page | 2
Point A Line 1
I1 I Point B
Z1
PST
UPA Line 2 UPB
I2
Z2
Figure 3 – Parallel connection between points A and B with a PST.
Supposing that line 1 has a smaller impedance relatively to line 2, without using a PST, line 1 would
carry the largest part of the load. In this case, one of the two lines could be operating below its nominal
rating because otherwise the parallel line would be overloaded [11]. To mitigate this effect, a PST may be
installed in the branch with higher or lower impedance.
The number of power lines has increased due to the increase of interconnections between power
networks [13] providing this way an increase on the overall efficiency and reliability. Several methods and
devices ensure that the transportation is done in order to reduce losses and costs and avoiding, if possible,
unnecessary spending in additional power lines, while maintaining system security as much as possible,
efficiency and controllability in order to maximize the power flow in the power lines [14].
The second situation, mentioned before, reflects the interconnection of systems such as in Figure 2.
If two independent systems are connected by two or more parallel lines, create a loop, the difference of
impedance between the lines will cause unbalanced line loading inside the systems. This happens if they
were not designed to support these parallel flows. As a consequence, some lines will be overloaded in the
process, thus limiting the power flows inside the system1. PSTs can be used in order to mitigate
uncontrolled parallel flows [15].
The process of installation and operation of PSTs in various locations and interconnections is
nowadays a well‐documented fact [7, 15‐17].
As with any other electrical machine, PSTs are subjected to stresses, power surges, overvoltages,
overtemperatures and ageing. These conditions may eventually lead to faults which will require the
shutdown of the transformer and therefore, major costs due to non‐operation and need for repair [18‐
20]. Hence, the importance of both protection and non‐invasive diagnostic procedures that are able to
detect the fault at an early stage and while the transformer is in operation.
In case of a fault, fast measures need to be taken in order to prevent aggravation of the fault and
minimization of dangerous situations [10, 21‐23]. Typically, by using overcurrent relays, current
transformers (CTs) and lower fault settings, faster operating times can be obtained [24]. In addition,
1
The phenomenon of loop power flows leads to power lines being operated closer to their stability limits and
results in a power system which may be operated far from its optimal state in terms of losses and security margins
[8].
Page | 3
differential protection provides fast tripping with absolute selectivity for internal high‐level shunt faults
and it is, therefore, often used as the main protection for the most important elements of the power
system, including transformers [25].
Fault diagnosis comprises the detection, identification and location of a fault in a machine. It may
also provide information regarding the severity of the fault [23]. Sometimes the fault may be incipient, as
such, instead of shutting down the transformer, it may be possible to keep it working at lower power level
than the rated one, providing this way enough time to find a replacement or reroute the electrical power
flow until a repair or a replacement has been set up. In case of a major fault, which may destroy or damage
not only the transformer but also the connected apparatus, power should be cutoff immediately, in order
to prevent larger damage or even dangerous situations. 2
Figure 4 – 450 MVA, 400/150/130 kV Phase‐shifting autotransformer, in a substation in Pedralva, Portugal [26].
As it is possible to visualize in Figure 4, transformers and PSTs in particular are electrical machines of
a considerable size, hence, their shutdown, transportation, dismantlement and invasive testing is
complicated and expensive, therefore, on‐site non‐invasive diagnose is of the uttermost importance in
order to reduce costs and keep the machine running as uninterruptedly as possible [20, 27].
The bibliography survey carried out during this work has highlighted that no previous studies were
conducted with regard to fault diagnosis in phase‐shifting transformers, an approach to this subject has
been chosen for this thesis theme. It should be noted that the operation of a PST differs quitely from a
normal transformer, therefore, some of the existing literature regarding fault diagnosis in regular
transformers does not apply.
2
Cut off should be done by proper devices such as circuit‐breakers or others of similar function.
Page | 4
An introduction to the working principles of PSTs will be given in chapter 2 as well as its particular
aspects, construction and characteristics. A focused approach to faults, protections and diagnostics as
well as our proposed diagnostic tool will be given in chapter 3. In chapter 4 it is proposed the adaptation
of an unpublished theoretical model of a standard transformer [28], in order to simulate and match the
working principles of a PST. The validation of the model and experimental results of the implemented
differential protection and our diagnostic tool will be presented in chapter 5. Conclusions and suggestions
for future work will be presented on chapter 6.
Page | 5
Chapter 2 – Phase‐Shifting Transformers
PSTs, also known as Phase Angle Regulators (PAR), are characterized by having a secondary voltage
with a controlled phase‐shift in relation to the primary voltage [16, 29].
2.1 – Application Principles of Phase‐shifting Transformers
In order to properly clarify the use of PSTs, the figure below, explains 3 possible situations regarding
the connection between two points A and B.
Figure 5 ‐ Parallel connection between two points (A and B).
Figure 5 (a) represents the connection between two points A and B through a single line. The basic
equation that governs the active power flow on the line between two points, is given by [11, 16]
US UL
P sin , (1)
ZL
the angle between U S and U L (measured from load to source) and Z L is the line impedance whose
resistance is approximately zero. It is possible to observe that the active power is directly proportional to
the voltages on the source and load sides and also to the sine of the angle between the two3.
Considering Figure 5(b), a case where points A and B are connected by two or more lines, one can
write the following relations:
Z2
I1 I (2)
Z1 Z 2
3
The power flow between two ends is established by a voltage drop and a phase‐angle shift between the
source and the load that dictates the direction of the power flow. These components are dependent on the
magnitude and power factor of the load current.
Page | 6
Z1
I2 I . (3)
Z1 Z 2
Under this situation, current will tend to flow through the line of least impedance causing
unbalanced line loading [12]. However, a different current distribution other than the one shown in (2)
and (3) may be required.
One of the approaches to regulate the power flow is to vary the source and/or load voltages but
doing that, influences to some extent the reactive power on the line [30]. PSTs create a stepwise variable
phase angle shift, advance or retard, between the source and load terminals, and, therefore, they present
themselves as a more feasible solution4.
By inserting a PST on one of the branches, as in Figure 5 (c), it is possible to control the power flow
distribution. A PST is able to increase the current in one of the lines by some quantity I X and decrease the
current on the other line by the same amount:
Z2
I '1 I I X (4)
Z1 Z 2
Z1
I '2 I I X . (5)
Z1 Z2
∆UPST ∆UPST
Point A I1’=I1‐IX Line 1 Point B
Z1
∆UPST UL US US UL
IX
α α
Advance Retard
Figure 6 ‐ An additional imposed voltage, ∆U, drives a current IX.
As seen in Figure 6, IX can be regarded as circulating current flowing around the system that overlaps
the load currents determined by the line impedances. For this current to circulate, a driving voltage,
U PST , must be created in such a way that:
U PST
Ix . (6)
Z1 Z2
4
Other Flexible AC Transmission Devices (FACTS) are available for power flow control such as the Unified Power
Flow Controller (UPFC) and the Dynamic Power Flow Controller (DFC), however, these are usually more expensive
and complicated than PSTs [31].
Page | 7
Depending upon whether the PST is installed in the branch with the higher or lower impedance, an
“advance” or a “retard” phase angle is needed [12]. On “advance mode” the load terminal voltage U L , at
the output of the PST, is leading the source terminal voltage U S . On “retard mode”, the voltage vector at
the output of the PST lags the input voltage U L [32]5. The vectors in Figure 6, translate this relation.
A possible configuration of a PST is illustrated in Figure 7.
SA LA
SB LB
SC LC
Series Unit
Excitation Unit
Figure 7 – Internal configuration of a phase‐shifting transformer (two‐core symmetrical type).
Since the resistance, in a PST, is fairly negligible, the impedance is largely reactive, hence, the voltage
∆U is required to be in nearly quadrature with the line current [24]. This can be obtained with a distinct
connection on the transformer, by connecting to a single phase the other phases in a distinct manner6.
This gives rise to a large number of configurations [33]. In advance mode a leading quadrature voltage is
added to the source voltage by the PST, whereas in retard mode the quadrature voltage is lagging the
current.
The control of the PST is done by the use of On‐Load Tap‐Changers (OLTC). Their number depends
on the type of PST, the desired function and the number of steps needed. On some PSTs, such as the ones
of design two‐core, the phase‐shifting capability can be combined with voltage magnitude control in the
same transformer, providing this way the ability to control the reactive power flow. This results in a more
complex unit, with two sets of tap‐changers [33].
5
In this thesis, the angle measurement convention is established to be from the load to the source.
6
As shown in Figure 7 in which the line A, in green, is fed from two voltage sources that are out of phase with
each other.
Page | 8
In a PST, one or more OLTCs is used to obtain the desired phase shift. In case a single OLTC is unable
to provide enough phase‐shifting capability or the number of steps is not enough, a serial connection of
OLTCs can be used. A table with the minimum number of OLTC per type of PST can be found in [12].
2.2 – Types of Phase‐Shifting Transformers
On the bibliography survey carried out during this work, the classification, representation schemes
and nomenclature of phase‐shifting transformers and their types changed significantly from source to
source. On this thesis the classification will comprehend firstly the number of cores and secondly the
symmetry of the PST.
PSTs are divided into two main types of design, being them single‐core and two‐core. This is a
relevant aspect as it dictates how the phase shift is accomplished.
The symmetry of the PSTs is also an important detail in their construction and operation since it
dictates how the load and no‐load condition will affect the phase shift of the PST and how the quadrature
injected voltage adjusts only the phase angle (symmetrical) or, besides the angle variation, it also adjusts
the voltage magnitude (asymmetrical) [2, 8].
Figure 8 summarizes the main PST designs and types.
Asymmetrical
Single‐Core Design Symmetrical
Delta‐Hexagonal
PSTs
Asymmetrical
Two‐Core Design Symmetrical
Quadrature Booster
Figure 8 – PST designs and types.
2.2.1 – Single‐Core Design
The single‐core design is also known as direct design due to the fact that the PST is manufactured
using only one core. It is by principle, the simplest type of PST. It is used at lower voltages, for smaller
phase shifts and smaller PST ratings [32]. In addition it has also fewer winding segments and it does not
need a separate excitation transformer. Among the disadvantages there is the fact that the OLTCs or the
regulating (tapped) windings are directly connected in series with the power system and, therefore, are
directly exposed to system disturbances and fault currents. Depending on the overvoltages and isolation
levels, the use of surge arresters may be convenient or required. The short‐circuit impedance of a PST of
single core design is very low at tap positions near 0 deg phase angle shift, hence the ratio between the
Page | 9
external fault currents passing through the PST and its rated current may become very high, especially in
systems with low fault current impedance. This has to be taken into account when selecting the OLTC
[12].
For this particular type of design some considerations have to be established. As the connection
scheme differs from the one of standard transformers, PSTs use a different terminology regarding the
currents and windings.
The primary windings are usually named as series, regulating or tapped windings and the secondary
windings are named excitation windings. For coherence purposes, the names Series Windings (SWs) and
Excitation Windings (EWs) were adopted in this work as these are the most common names found in the
bibliography.
Single‐core PSTs usually have one series winding and one excitation winding per phase. The OLTC
establishes the position of the source tap or source and load taps. The relation between the two taps
translates the relative tap position D ( 1 D 1 )7.
Excitation currents ( I E ) are considered the currents that flow in the EWs. Voltages of the source and
2.2.1.1 – The Asymmetrical Type
The asymmetrical single‐core PST is also known as half‐tap PST due to having just half of a full
regulating winding. When the PST is working under no‐load conditions, the voltage magnitude on the load
side is different from the one on the source side. Frequently, the difference increases as the phase angle
shift α increases. Figure 9, shows the internal configuration of this type of PST along with its phasor
diagram.
In this type of transformer, the commutation between the advance and retard modes is done by a
switch in each phase that connects the excitation windings to the series windings.
7
This topic will be addressed in more detail later on.
8
This color code will remain valid for all the waveforms, diagrams and experimental setup cables, during this
work.
Page | 10
SA LA LA SA
SB LB
SC LC n1
U V W Series U US
UL
Series Windings
α
n1
turns
Excitation Windings
n2
n2 Excitation U
turns
‐1<D<0
Advance Position
Figure 9 – Schematic representation and phasor diagram of a single‐core asymmetrical type PST.
2.2.1.2 – The Symmetrical Type
The single‐core symmetrical type is also known as “Standard/Extended‐delta single‐core PST”. This
is due to the shape of its phasor diagram, which is shown in Figure 10. In the symmetrical type, the voltage
magnitude under no‐load conditions is identical in source and load sides, with the phase angle shift
having no effect on them.
In this type of transformer, the commutation between advance and retard modes is done by two
switches per phase. This differs from the single switch in the asymmetrical type due to the fact that the
series windings on this type of PST are in fact split in two.
SA LA LA Series U SA
SB LB
SC LC n1
U V W
UL US
Series Windings
α
n1
turns
Excitation Windings
n2
n2 Excitation U
turns
‐1<D<0
Advance Position
Figure 10 ‐ Schematic representation and phasor diagram of a single‐core symmetrical type PST.
Page | 11
2.2.1.3 – Delta‐Hexagonal
The delta‐hexagonal (DHPST) is a PST with a symmetrical design that fall into the category of
squashed‐delta PSTs. (Figure 11). It is a relatively new scheme whose existence is due to the new
advancements in the OLTC technology. It promises to be much more economical to build and install than
previous designs, thus improving the economic viability of deployment of these devices on the
transmission grid [16].
In a DHPST there is a double OLTC with two three‐phase taps. Each tap moves its three terminals
along the series windings so that a phase‐shift is created between the source and load ends. When the
taps pass each other, the transformer shifts the operation from retard to advance mode or vice‐versa
[12].
SA LA Series U
SB LB LA SA
SC LC
U V W n1
Series Windings
n1 UL α US
turns
Excitation Windings
n2 n2
turns
Excitation U
‐1<D<0
Advance Position
Figure 11 – Schematic representation and phasor diagram of a delta‐hexagonal single‐core PST.
2.2.2 – Two‐Core Design
The two‐core design can be classified as an indirect transformer. This happens because it is in fact
composed by two transformers, each with its own core and associated coils, coupled by a “throat
connection”[7]. One variable tap exciter, excitation or shunt transformer that regulates the amplitude of
the quadrature voltage and one series transformer to inject that same quadrature voltage in the correct
phase. A practical example of a two‐core design PST can be observed in Figure 12. A schematic
representation of a two‐core design PST of the symmetrical type was earlier shown in Figure 7.
Page | 12
Figure 12 – Two‐core PST, 600 MVA with a phase angle shift of ±40⁰, for operation in Slovenia9.
If reactive power flow needs to be independently regulated from the active power flow, voltage and
phase angle control can be combined within a single regulating transformer with two OLTCs operating
independently. One OLTC regulates the quadrature‐phase voltages and the other one regulates the in‐
phase voltages, which are combined at the output [34].
The series transformer primary winding is connected in series with the primary system, between the
source and load terminals. In the case of the symmetrical type, this winding is split into two halves, and
the primary winding of the excitation transformer is connected to the midpoint between these two half‐
windings. Thus, total symmetry between source and load side no‐load voltages is achieved. In the
asymmetrical type case, the primary winding of the excitation transformer is connected directly to the
line or source of the series windings.
The regulating circuit consists of a secondary tapped winding in the excitation transformer and a
delta connected secondary winding in the series transformer. The ratings of these two windings can be
optimized independently of the voltage level of the primary. This provides more freedom for the selection
of the OLTC, which sometimes is a limitation factor for a specific design [25].
It is possible to observe the phasor diagram in for this type of PST in Figure 13.
9
Courtesy of Siemens.
Page | 13
LA SA
n1
EA
UL US
α
α
α
EC EB
Series Transformer Excitation Transformer
Figure 13 ‐ Phasor diagram of the two‐core symmetrical PST.
Depending on the voltage level, rated power, maximum no‐load phase angle shift and tap‐changers
of the PST, the series and excitation units can be distributed as follows:
Two three‐phase transformers within the same tank;
Two three‐phase transformers, placed in two distinct tanks connected by High Voltage (HV) cables
and possibly through oil ducts;
Six single‐phase transformers each with its own separate tank interconnected by HV cables and
possibly through oil ducts [25].
Two and six‐tank designs have advantages in aspects such as transportation, neutral grounding and
a constant zero sequence impedance. They have however some disadvantages in terms of the space
requirement and cost. This is due to the connection between the transformers. If we are considering two‐
core transformers, the connection between them is done on the high‐voltage side, which requires the use
of appropriate high‐voltage bushings. In case of having a separate tank design, due to the physical
properties of air, an associated space between transformers is required in order to prevent voltage
breakdown.
2.2.3 – Special Configurations (Quadrature Boosters)
Occasionally, a different type of PST is used. It is often known as quadrature booster (QB) and even
though it usually has a two‐core design, it may also have a single‐core design [35]. It has an asymmetrical
design and it is in fact a combination of a regulating autotransformer with a series unit. A QB can have a
single‐tank or two‐tank design. One of its advantages is that the output voltage can be adjusted in a four‐
quadrant (magnitude and phase) [13].
Page | 14
One of the main differences between this type of PST and a typical autotransformer is that besides
the ability to work just as autotransformer it can also work as a phase‐shifter. Usually a switch exists in‐
between the OLTC winding that allows switching between the two modes of operation [25].
The schematic for the QB can be found in Appendix A.
This type of design has been reported often in the literature and is, in fact, the only type existing in
Portugal [36]. Currently Portugal has four PSTs under operation in its grid. Two 450 MVA, 400/150/60 kV
installed in a substation in Falagueira and two 450 MVA, 400/150/130 kV in a substation in Pedralva.
2.3 – Selection of the Delta‐Hexagonal Phase‐Shifting Transformer
As previously demonstrated, a large variety of PSTs is documented in the available literature. The
diversity of configurations existent do not allow us to establish general considerations on protection and
diagnosis of faults for this type of transformers. As such, the choice of a particular type of PST for analysis
is required.
Our choice lays on the DHPST. Even though this type of PST has a rather complex design due to its
connections and existence of double tap‐changers, it is one of the most recent PSTs and appears to be
one of the most well‐documented. It has the advantage of having a symmetrical single‐core design, which
makes it easier to implement in the laboratory with the existing resources.
Figure 14 shows the detailed schematic representation of the DHPST used in this work.
SC LC
SB LB
SA LA
iLA iLB iLC
LA SA
iSA iSB iSC
X2 U2
U2 V2 W2
Series Windings
n1 n2 uL α uS n2
iEB iEC iEA
turns
uU1X1
U1 V1 W1
Excitation Windings
X1 Excitation U U1
n2
uW1Z1
n2 Advance Position
uU1X1
uV1Y1
iEA iEB
turns ‐1 < D < 0
iEC
X1 Y1 Z1
Figure 14 ‐ Detailed schematic representation and phasor diagram of the DHPST used in this work
Page | 15
The configuration of this DHPST is based on 6 windings assembled on a single three‐legged core.
These windings are connected in a particular way, in which series and excitation windings are electrically
connected, virtually intersecting each other. No windings are isolated in this structure of a DHPST.
The DHPST is composed by three series windings (U2‐X2, V2‐Y2, W2‐Z2) and three excitation windings (U1‐
X1, V1‐Y1, W1‐Z1).
The OLTC establishes the position of the source and load taps, reversing the phase shift when the
source and load taps intersect in the middle position [2]. The relative tap position between the two taps
is given by D which varies from ‐1 (full tap advance) to 1 (full tap retard), where in the middle position
D=0. The scheme shown in Figure 14 reflects the advance position.
A convention was established in which i S and iL point inwards, into the transformer and iE are the
excitation currents that flow on the EWs from the bottom to the top of the windings.
The phase shift is measured from the load to the source. Usually currents are used for this
measurement, as the angle does not vary depending on the load [25] 10.
10
For no‐load conditions, the phase angle shift is measured using the voltages.
Page | 16
Chapter 3 – Faults, Protections and Diagnostics in
Transformers
3.1 – Introduction
The transformer is one of the most important and costly devices in the power grid [20]. Any defect
or fault will most likely cause misoperation, shortening of its life span and consequently its removal from
service. The study of faults and component failure is important in order to understand and predict
longevity and mean time between failures (MTBF) of specific components.
Several attempts were made to find the causes of failures in faulted components of standard
transformers. However, no studies were entirely successful due to complications on compiling and
analyzing the data due to incomplete or incompatible responses and, lately, due to issues regarding data
access restriction [37]. Consequently, from the analysis of these studies, a certain disparity of results was
found.
Nevertheless, in a reliability study presented in [20], 28 years of data regarding transformer failure
was gathered. 549 transformer stoppages were analyzed for transformers ranging from 150 kVA up to
150 MVA in the period of 1979‐2007. Figure 15 shows the sampled percentage of specific component
failures [20].
TRANSFORMER STOPPAGE VS SPECIFIC COMPONENT FAILURE
Core Cooling System
2% Winding
Bushing CT 7%
1% Undetermined Component
Bushing Winding NLTC
14% 34% Oil (Insulating)
OLTC
Tank and Tank and Accessories
Accessories
7% Bushing
Bushing CT
OLTC
10% Undetermined Core
Oil (Insulating) NLTC Component
4% 10% 11% Cooling System
Figure 15 ‐ Transformer stoppage per specific component failure [20].
Page | 17
Similarly to other transformers, whether they have a more standard design or not, PSTs are not
immune to faults. Faults can take place inside the transformer or at the electrical system level where the
transformer is connected. Therefore, faults can be divided into two categories: external and internal.
Faults such as overvoltages, overloads, under or over frequency and short‐circuits caused by external
applied conditions ought to be considered as external faults.
Internal faults can be subdivided as incipient faults (overheating, overfluxing and overpressure),
characterized by being minor and intermittent, or active faults (winding phase‐to‐phase, phase‐to ground
faults, winding inter‐turn faults, shorted laminations, tank faults and tap‐changer faults) [10, 38].
3.2 – Types of Faults
Faults cause changes in the original electromagnetic circuit of the transformer, and thus disturbs the
original balance [5]. Based on that premise, the detection of those changes allows the implementation of
relays that produce a trip signal whenever a fault is detected, in order to prevent damage or aggravation
of the fault [23]. Certain protections are able to identify specific faults while others only detect the
existence of a fault, without discriminating its type or location. While some faults are easy to identify,
others appear to be very difficult to detect.
In this section several faults are presented along with some of their characteristics. Figure 16
illustrates turn‐to‐turn, phase‐to‐phase and phase‐to‐ground faults.
3.2.1 – Overheating, Overpressure and Tank Faults
Overheating, overpressure and tank faults are types of faults that cannot be detected by voltage and
current measurement, only by temperature, sudden pressure and Buchholz relays. For this reason, these
are beyond the scope of this thesis and will not be discussed any further [10, 21].
Page | 18
3.2.2– Phase‐to‐Ground Faults
Phase‐to‐ground faults happen when there is a short circuit between a phase (or more) and the
ground. This may be caused by improper maneuvering, accident or developed from a previous turn‐to‐
turn fault.
Techniques based on transient zero‐sequence and negative‐sequence current analysis have been
used on regular transformers [25, 39] for detection of phase‐to‐ground faults. On DHPSTs, a simple zero‐
sequence differential current shows a high sensitivity with regard to the existence of ground faults [2].
Also, as none of the winding terminals are grounded, ground faults produce large terminal currents that
are easy to detect. Furthermore, electrical methods that mimic the ampere‐turn equations of the
transformer as well as dissolved gas analysis (DGA) methods are capable of detecting this type of fault [9].
3.2.3 – Phase‐to‐Phase Faults
Phase‐to‐phase faults occur whenever there is a short‐circuit between two or more different phases
in the transformer. Phase‐to‐phase faults are more likely to occur in DHPSTs due to the complex nature
of its tap‐changers because of moving source and load tap‐changers, operation of reverse switching and
complex connection of the coils [9]. These characteristics make the connections, specifically between
source and load taps, more prone to this type of fault.
This type of fault is easily detectable by a sudden appearance of very high currents. Such high current
on the source or load sides of the transformer will trip over‐current relays. Similarly to phase‐to‐ground
faults, these can be more accurately detected with DGA relays as well as with negative‐sequence current
analysis and with electrical methods that mimic the ampere‐turn equations of the transformer.
3.2.4 – Turn‐to‐Turn Faults
Turn‐to‐turn faults represent one of the major causes of failures in modern power transformers.
These faults are often traced to internal insulation failure and, besides representing one of the most costly
faults, are also associated with a longer downtime [19]. If turn‐to‐turn faults are left unattended, they can
quickly progress to more severe faults11.
Turn‐to‐turn faults are caused by defective insulation or, most commonly, due to the degradation of
the insulation caused by ageing and mechanical, thermal and electrical stresses. External faults also
contribute to the degradation of the insulation, especially due to the excessive mechanical radial and
compressive forces acting on the windings [18].
A short‐circuit of a few turns is able to generate a heavy fault current in the shorted turns,
contributing even more to the degradation of the insulation and consequently to the spread of the fault.
Even though a heavy fault current is generated, due to the transformation ratio between the healthy and
11
Otherwise, they can cause arcing within the transformer tank until tripped by a Buchholz relay.
Page | 19
the shorted turns, variations in the transformer terminal currents will be quite small and, as such, typical
differential protections present a limited sensitivity and do not trip for minor turn‐to‐turn faults [24, 25].
3.2.5 – Iron Core Faults
During the bibliography survey, it was not possible to find any material regarding this type of fault in
PSTs. This type of fault, however, is common to all types of transformers. It is caused by insulation
problems in the ferromagnetic material.
The core of transformers is constructed by stacking sheets of thin ferromagnetic material12
laminations. Each lamination is insulated from the others by a thin non‐conducting layer of insulation [32].
The effect of laminations is to confine the induced Eddy or Foucault currents to highly elliptical paths that
are longer and narrower and thus, enclose little flux. This reduces their magnitude and, therefore, losses
and heating in the core.
Often due to ageing and magnetostriction effects, the insulation is degraded. Magnetostriction,
caused by the magnetic flux, produces a small variation of the size of the core material, leading to friction
between the core laminations and vibration of the core. Over time, this friction leads to a loss of the
insulation properties and, thus, a less resistive contact, creating larger induced currents in the core
laminations. At specific locations, where the insulating material is degraded, the larger induced currents
cause hot‐spots. Figure 17, shows the Foucault currents in the core of transformer without any fault and
with an iron core fault (hot‐spot in the core).
Hot‐spot
Figure 17 – Appearance of an hot‐spot due to the degradation of the insulation between the core laminations.
Magnetostrictive deformation has a linear relationship with the square of the amplitude of the
magnetic flux density which, in turn, has a linear relationship with the rated voltage of the transformer
[40].
12
The most common compounds are steel or silicon steel, depending on the application.
Page | 20
3.2.6 – On‐Load Tap‐Changer Faults
In spite of the fact that OLTCs have a relatively long lifetime and high level of reliability, in comparison
with the other transformer components they are less reliable and fail more often. In fact, OLTCs are the
one of the main responsibles for failures in HV power transformers and some studies demonstrate that
these are responsible for 41% to 56% of unplanned transformer unavailability13 .
The use of single or double tap‐changers in a PST, result in a more complex design. Therefore, these
may present a higher failure rate due to internal faults.
The mechanical design of OLTCs makes them vulnerable to several issues, such as loose spring,
contact abrasion, loose base, among others. These problems are an issue because not only they can cause
severe damage to the OLTC but also to the transformer itself [43]. Currently, several methods are being
studied not only to detect a fault but also to identify the specific component of the OLTC where the fault
exists. These techniques commonly rely on dynamic resistance measurement (DRM), power spectral
density of typical vibration signal, temperature monitoring and DGA [41, 43].
3.3 – General Considerations on Protections
For a stable and secure operation, a transformer must operate within the tolerable limits. To ensure
that this happens, a protection system must be employed.
Several protections must always be implemented in a transformer in order to protect it from
different types of faults and in a way that it is possible to have redundancy in case of misoperation of the
first protection scheme. Backup protection and redundancy is compulsory.
One of the most important types of protection is the differential protection (DP). The use of a
properly dimensioned DP provides security against internal and external faults by triggering a trip signal.
Due to the importance of this type of protection a special focus was given to it during this work.
A protection may be able to distinguish between zones and trip only when the fault occurs inside the
designated zone. Each zone in a power system has its own protection, and its boundaries are defined by
the circuit breakers, whose purpose is to isolate the zone in the event of a fault inside it [10].
Transformers of standard winding connections have protection rules which are commonly referred
to as: ratio matching, vector group compensation, and zero‐sequence removal [9]. Most of these
protection rules however, cannot be applied straightforward to some types of transformers of
nonstandard or unusual configuration , such as PSTs [44].
13
This number is indeed related to the fact that the majority of OLTCs have an old design and are more prone
to failure than more recent designs [41, 42] based on vacuum switching technology.
Page | 21
The protection system must meet several requirements. Among them, it must be reliable,
dependable and secure, tripping exclusively when it is required to operate. Furthermore, it must be
sensible and fast in order to isolate the zone even in case of a minimal fault in a very short period of time.
For protection, several kinds of relays14 are available, among those, the most common are:
differential, directional, magnitude, ratio, impedance, time dependent, overcurrent, frequency, among
others [45]. Most relays and protection applied on regular, standard transformers can be directly applied
to PSTs. Differential protection however, as it is usually dependent on the structure of the transformer
and its winding connections, must be specific to PST or has to be adapted with a compromise on
sensitivity.
3.4 – Differential Protection of Transformers
The first differential protection dates back to the end of the 19th century. It measures the currents
flowing into and out of the protected object, and establishes a biased comparison. If any unbalance exists,
the relay will trip the primary and secondary circuit breakers (CBs).
It is suited as a main protection as a result of the fast tripping with absolute selectivity, although a
backup protection function for external faults must be always implemented with an additional time
graded protection (over‐current or distance) [46].
Usually, for transformers with a standard connection, CTs at the end of the primary and secondary
sides are chosen in a way that mimics the transformation ratio of the main transformer (CT ratio
matching), thus making the sum of the overall current from both CTs zero. The vector group of the
transformer also plays a role in this measurement and for that reason, modern current relays provide
software phase‐shift compensation.
Figure 18, represents the typical differential relay connection diagram for a single‐phase
transformer.
14
Relays are electrically operated switches responsible for connecting/disconnecting an electrical component
upon the reception of a trip signal.
Page | 22
CT1 IS IL CT2
I1 I2
I1 Differential Relay I2
IDIFF=I1+I2
Figure 18 ‐ Differential protection scheme for a single‐phase transformer.
IRES k I1 I 2 , (8)
where k is a compensation factor, usually 1 or 0.5 [13]. This comparison is done in order to determine
whether the differential relay will trip or restrain from operation. A minimum pickup signal IP is always
present and is usually established at about 20% to 25% of the differential current characteristic15. This
minimum pickup signal is needed but has the disadvantage of delivering poor sensitivity for minor faults.
The tripping condition is given by:
The above condition states that the relay will trip if the differential current exceeds the sum of the
minimum pickup signal and a certain percentage of the restraining current ( K.IRES ).
Usually, the equations above can be used to plot the relay tripping characteristic. Figure 19 shows
this characteristic with the minimum pickup signal represented by IP and the single‐slope characteristic
(SSC) as well as the tripping and restraining areas. Furthermore, in this figure we also show a second slope
(DSC) which is often used in relaying protection. It provides an increased security for high‐current external
faults while allowing more sensitive operation for low‐current internal faults [47]16.
15
This value is related to the security margins that accounts for excitation currents, tap‐changer position errors
as well as CT and measurements errors.
16
This is done by applying greater restraint for greater currents to accommodate CT saturation errors [13].
Page | 23
500
400
Diff Current (%)
Tripping
300 Area
200
Restraining
100 Area
IP
0
0 100 200 300 400 500 600 700 800
Restrain Current (%)
Figure 19 – Differential relay tripping characteristic.
The differential relay is not, however, immune to some conditions. In the first few cycles after
energizing a transformer, an inrush current is present. An example of this current for a single phase
transformer can be seen in Figure 20.
Figure 20 – Energizing inrush current for a single‐phase transformer obtained from the simulation model17.
The energizing inrush current occurs during the energization of the transformer due to the flux
density which is above the saturation level of the magnetic core. This highly asymmetrical current can
have a magnitude from 2 to 5 times the rated load value of the transformer and has a predominant second
and fifth harmonic component, which is used by classical differential protection schemes to restrain relays
from operating [25, 32]. Other methods for restraining are used such as wave‐shape recognition,
advanced digital signal processing (DSP) tools and artificial intelligence (AI) techniques [13].
17
The simulation model will be presented further ahead in this thesis. The obtained data was obtained for a 2
kVA, 220/127 V, 50 Hz single‐phase transformer.
Page | 24
3.5 – Differential Protection of Phase‐Shifting Transformers
3.5.1 – Introduction
PSTs have both magnetically coupled and electrically connected zones of protection, requiring, for
proper protection, a differential protection different than the ones based on ampere‐turn equations for
standard transformers. Protection of PSTs is not yet well established and sometimes a classical
transformer relay is adopted for PST protection. This results however, in a reduced sensitivity and a need
for testing for each particular application [9]. Also, PSTs are available in different designs and hence,
require different types of protections [29]. All the existing protection schemes always contain DP.
DP of PSTs, in the same way as standard transformer connections, require current matching (due to
CT ratio mismatch), phase and zero‐sequence compensation, and sometimes energize inrush
desensitization but, unlike standard designs, these present some new challenges. Among them is the fact
that PSTs have a variable phase shift between the source and load sides and the dependency of differential
and restraining currents on the tap position [10].
3.5.2.1 – Differential Principle for Electrically Connected Windings
In 1999, IEEE published a report concerning “Protection of phase angle regulating transformers”. In
this report, [29], the authors present a method that takes into account the currents entering and leaving
the windings. By applying the Kirchhoff’s current law and knowing that the sum of all currents should be
equal to zero, or at least a very approximate value, if any major difference is detected this difference is
certainly caused by a fault.
Adapting the equations given to our notation, in which source currents ( IS ) and load currents ( IL )
are considered to enter the series winding, and the excitation current entering the series winding of phase
A is IEB and the one leaving is IEA (refer to Figure 14). For the other phases a shift of the currents index was
done and, as such, the differential currents in the series windings can be established as in (10).
18
Even though the following methods of protection are applicable to DHPSTs, some of them can also be applied
to other types of PST with some or even no adjustments.
Page | 25
The same principle is applied to the excitation windings. As such the differential currents can be
established as in (11), in which IEX is the excitation current entering the excitation winding (EW) “x” and
Even though this is a simple method, it is not capable of detecting any winding or turn‐to‐turn fault
inside the given zone, only faults between zones and/or ground faults. Therefore, it is only recommended
as a backup.
This method was later presented as the “differential principle for electrically connected windings”
(DPECW) in [5, 9].
1 n2
2 n1
D . (12)
iLA i SB 1
iEA iEC 2
Some of the differential protection methods presented by the authors are:
Page | 26
Linear Current Differential Principle (LCDP)
Non‐Linear Current Differential Principle (NLCDP)
Power Differential Principle (PDP)
Phase comparison principle (PCP)
and signals will be presented exclusively for phase A19.
Linear current differential principle (LCDP)
This method is based on the ampere‐turn equations. (13) and (14) present the differential and
restrain currents. This method requires tap position measurement.
D 1 n2
iDIFF _ A iLC iSC D. iLA i SB (13)
2 n1
D 1 n2
iRES_ A iLC i SC D. iLA i SB (14)
2 n1
Non‐Linear current differential principle (NLCDP)
This method is obtained by combining equations (13) and (14) for the LCDP and (12) that provide us
the tap estimation, thus leading to
Power differential principle (PDP)
PDP is based on the power transformer losses. These are composed of a constant value due to iron
losses and a factor proportional to power transfer, related to copper losses20. When a fault occurs, the
power losses of the transformer will exceed the specified threshold. Equations (17) and (18) show the
differential and restrain signals.
FDIFF PS PL (17)
SS SL
FRES . (18)
2
19
As the principle for the phases B and C is the same, the equations can be adapted by simply shifting the
indexes once and twice for phase B and C respectively.
20
Relatively accurate values of these losses can be obtained from the transformer manufacturer [5].
Page | 27
Phase comparison principle (PCP)
This method is based on the current displacement between source, load and excitation currents. A
comparison between currents is done in a way that whenever the “comparator angle limit” 0 exceeds
a certain threshold, a TRIP signal is produced. The trip equation is given by
TRIP : Angle iLA iSB ,iEA iEC LIM . (19)
iDIFF _A i SA iLA
1 0 1 0
iDIFF _B I M(0 ) iSB I M( ) iLB (20)
i base
i SC
base
iLC
DIFF _C
where
3.6 – Diagnosis of Faults in Core and Shell‐Type Transformers
While the main purpose of the conventional transformer DP is to isolate a fault as quickly as possible
in order to limit the extent of the damage, diagnostic tools are able to provide an indication of early stages
Page | 28
of the transformer deterioration. These tools are able to deliver a comprehensive and reliable diagnosis
of power transformers with standard winding connections.
Diagnostic tools can be online or offline and, most of them, do not require an invasive procedure as
invasive procedures tend to be very costly not just due to transformer disassembly and transportation but
also due to costs related to the downtime.
Several methods have been broadly used and several recent studies have been published over the
past few years[23]. These methods include:
Park’s Vector approach
Electric and/or acoustic detection of partial discharges
Determination of the degree of polymerization of paper insulation
Frequency Response Analysis (FRA)
Dielectric loss angle measurement
Dissolved gas analysis (DGA)
Vibration analysis
Excitation current analysis
Winding ratio and/or resistance
Magnetic leakage flux
Thermography
As not to over‐extend this document many of them will be beyond the scope of this thesis and we
will be focusing on the Extended Park’s Vector Approach.
3.6.1 – Extended Park’s Vector Approach
Space vector theory was originally used as a diagnostic method for the detection of AC motors faults.
Its use as a diagnosis method in transformers goes just a few years back.
In [23, 27] the authors use the Extended Park’s Vector Approach (EPVA) of the on‐load exciting
current for winding fault diagnosis. In these publications, an online diagnostic method is described as
capable of early turn‐to‐turn fault detection by on‐load excitation current measurement. This method is
applied to a standard connection transformer with a fixed phase shift and turn ratio.
Park’s Vector (PV) components ( iD and iQ ), as a function of the currents i A , iB and iC , are given by:
2 1 1
iD i A iB iC (22)
3 2 2
2 3 3
iQ iB iC . (23)
3 2 2
Page | 29
is centered at the origin of the coordinates in case of normal conditions, whereas in an abnormal
condition, the shape will diverge from the reference pattern. The shape, orientation and magnitude of
this pattern, under abnormal conditions, allow us to infer the affected phase fault as well the severity of
the fault.
In standard transformers, with a fixed phase shift and transformation ratio, the EPVA is applied to
the differential current between the primary (input) and secondary (output), known as on‐load excitation
current. i S should be nearly equal to iL minus the fixed phase shift, the transformation ratio and the
excitation currents. The phase shift compensation is done by interposing CTs relatively to the transformer
vector group or by software compensation and the transformation ratio is compensated by the sum of
the primary currents with the secondary currents affected by the inverse of the transformation ratio.
From the obtained locus, a Fast‐Fourier Transform (FFT) is applied to the modulus of the Park’s
Vector currents and the spectral measurements of the DC component and the 2nd harmonic are obtained.
By dividing the 2nd harmonic component (100Hz for a 50Hz supply from the grid) by the average:
magnitude iD 2 iQ2
2 f component
SF
, x100% (24)
average iD 2 iQ2
we obtain the Severity Factor (SF) which reflects a relative value of the severity of the asymmetry or
fault21. From the shape of the locus it is also possible to identify the phase in which a fault occurs.
Regarding PSTs, no bibliography was available regarding Park or Extended Park’s Vector Approach,
as such, we postulate that the application of this method for PSTs may also be of interest for its diagnosis.
This phase angle shift is given by α measured from the load side to the source side. To provide the phase
compensation, a shifting matrix given by equation (25), as in [25], was applied.
21
Ideally, under healthy conditions, the EPVA signature FFT
iD 2 iQ 2 will be clear from any spectral component,
thus only a DC value is present in the current PV modulus. However, any real transformer presents a certain kind
of asymmetry [23].
Page | 30
Using the equation above, we create our differential currents for application in the EPVA method as:
iDIFF _ A i SA iLA
iDIFF _B i SB M iLB . (26)
iDIFF _C i SC iLC
load currents shifted by the angle that this way overlaps the source currents.
Applying the PV equations we obtain:
2 1 1
iD _ DIFF iDIFF _ A iDIFF _ B iDIFF _ C (27)
3 2 2
2 3 3
iQ _ DIFF iDIFF _ B iDIFF _ C . (28)
3 2 2
Furthermore, we calculated the modulus of this differential signal and a Fast Fourier Transform was
applied. From this it is possible to obtain the spectrum of the Park’s Vector signature. Analyzing the FFT
and applying equation (24) we obtain the SF [23]. The SF will increase proportionally to the increase of
the fault or asymmetry.
3.6.3 – Active and Reactive Differential Power Analysis
Based on the source and load currents and voltages. We proceeded to do a spectral analysis of the
differential active and reactive power of the PST.
We used (22) and (23) to calculate the d and q components of source and load sides voltages and
reactive powers, hence obtaining
and reactive powers, thus obtaining SFP and SFQ.
Page | 31
Chapter 4 – Simulation Model
Computer simulation plays an important role in the testing and development of new technologies.
Since an actual PST is not available for testing, a standard transformer was adapted and a simulation
model was used in order to test the phase‐shifting capabilities of the available transformer, and validate
the applicability and reliability of the model. The detailed characteristics of the transformer used in the
experimental test are presented in Appendix B.2.
To properly validate the model in which this thesis is based on, we established a comparison between
the waveforms obtained from the simulation model and the ones obtained by data acquisition from the
transformer used in the laboratory as a DHPST.
The next subsection will follow closely the information presented in [28]. This will serve as summary
for better understanding the simulation model as [28] was not published and the access to it is restricted.
4.1 – Developed Model
The base simulation model was developed in [28] and it is based on the duality between electric and
magnetic circuits. The model was built based on the transformer available for the experiment and it
accounts for multiple parameters such as core dimensions, hysteresis and electric and magnetic
parameters providing this way, theoretically, seamless reliability. The detailed parameters of the
transformer can be found in Appendix B.2.
The model has a block type configuration in which each winding of the transformer is represented
by a block. Primary and secondary windings are represented by different blocks and their parameters,
such as resistances, reactances and number of turns can be changed according to the user preference. By
connecting the blocks with wires we can create the connections of the primary and secondary windings
or even between the two. This kind of modular design allows us to implement even tertiary windings or
create virtually any kind of design of transformer.
Likewise, the magnetic core of the transformer is also modular, and, therefore, we are able to create
not just the windings and their configuration as we please, but also to module the magnetic core with
blocks. The blocks themselves, similarly to the windings, allow us to change their parameters such as
dimensions, magnetic parameters and hysteresis. Their connection however, is not as straightforward as
the windings connection because it requires the calculation of the magnetic equations of the circuit.
4.1.1 – Electric Circuit
A basic transformer is composed by 2 electric windings per phase crossed by the same magnetic flux.
The windings have an internal resistance and a leakage inductance. The induced voltage in the winding is
Page | 32
primary and secondary windings are given by:
diP d
uP (t) RP iP LLP NP (31)
dt dt
di S d
uS (t) RS i S LLS NS , (32)
dt dt
represent the number of turns of the windings and the indexes P and S stand for the primary and
secondary windings respectively. d / dt represents the magnetic flux variation in the winding and is the
link between the electric and magnetic circuits of the simulation model.
In order to make the connection between the electric and magnetic circuits, flux variation is read as
a voltage source. As such, the magnetic flux variation was adopted as the control variable and the current
flowing in the primary winding is the state variable.
Two different approaches were followed in [28] for the implementation of the primary and
secondary electric circuits in Simulink. The two approaches are illustrated in the figures below.
RP LLP
dφ
i1 dt Magnetic i1
u1 i1 NP dφ V
dt Core
Figure 21 ‐ Implementation of the primary circuit of the transformer.
RS LLS
i
ZL u2 i2 NS dφ i
Magnetic
dt Core
Figure 22 – Implementation of the secondary circuit of the transformer.
The methodology used reflects the interface between the electric and magnetic models.
For the circuit of the primary winding (Figure 21), we read the voltage drop due to the magnetic
flux variation, which, divided by the number of turns ( NP ) gives us d / dt . This variable is passed to the
magnetic core block which calculates the magnetomotive force needed to establish the flux. Dividing
Page | 33
secondary winding is sent to the magnetic core block.
mmf H .l N . I (33)
In order to obtain a certain flexibility, the circuits of the primary and secondary windings were
implemented using two different blocks named as Current Source Block (CSB) and Voltage Source Block
(VSB) respectively. Figure 23 shows the Simulink implementation of the CSB and VSB.
Figure 23 – Simulink blocks of the primary and secondary: (a) CSB; (b) VSB.
As the CSB is the block responsible for creating the flux, this block must always be present in a
transformer configuration. When there is the need to implement more windings, those should be
implemented as VSB blocks.
The implementation of segmented windings is done by adding a single or more CSB blocks in series
with the block we want to segment. An example of the implementation of a primary and secondary
segmented winding will be presented further ahead.
4.1.2 – Magnetic Circuit
The magnetic circuit allows the introduction of faults and asymmetries22, as it contemplates the
existence of all reluctances along the core. The implementation of the magnetic properties of the
ferromagnetic material is based on the equations and model developed by Tellinen [48].
In this model, the approach chosen for the core implementation was the development of blocks
(bricks) in which, each brick has the characteristics of the magnetic material. Each brick corresponds to a
22
Even though it is in fact possible, no faults or asymmetries were implemented in this work.
Page | 34
currents).
Figure 24 is the representation of a segment of the magnetic circuit of the transformer.
Figure 24 – Representation of a brick of the core [28].
The size and magnetic properties of each brick can be configured by the user. In the developed
model, two types of bricks were selected: one for the rows and another for the columns. The magnetic
properties of these two bricks are the same except for the dimensions, which change according to the
type of brick (row or column). Figure 25 shows the implementation of the model of the magnetic core of
the transformer in Simulink.
The bricks are connected in a way that mimics the meshes of the equivalent magnetic circuit.
Furthermore, each brick contains the equations that represent the hysteresis and the losses due to the
Foucault currents. Due to the nature of these equations, the bricks can use the magnetic flux as a control
variable and the mmf as a state variable or vice‐versa.
Following the analysis of the magnetic circuit for the core type transformer available, the equivalent
magnetic circuit was established as in Figure 26, where C1 , C 2 and C 3 and T1 , T2 and T3
Page | 35
Figure 25 – Implementation of the model of the magnetic core of the transformer in Simulink [28].
Figure 26 – Equivalent magnetic circuit for the core type transformer [28].
The circuit of Figure 26 has 5 independent meshes and 8 magnetic fluxes, from which 3 are control
variables and 5 are unknown. It is therefore possible to determine the equation system as in [28].
Page | 36
4.2 – Model Parameterization and Adjustments
The simulation model had to be adapted in order to simulate a DHPST. The first modification was the
electrical connection between the windings, which differs quietly from the usual delta, wye or zigzag
connections of a standard transformer. Secondly, as the base model contemplates the existence of a
tertiary winding, this also had to be removed. Furthermore, the particular electrical connection that these
transformers have, required the downsizing of the rated power. For the existing configuration the rated
voltage was established for the worst‐case scenario, when D=0. The value of the voltage was established
at 237 V phase‐to‐phase using the simulation model. Beyond this voltage level, the voltage at the
terminals of the SWs and EWs would exceed the nominal voltage of 220 V and 127 V and, consequently,
the transformer would saturate.
Also, in the model from [28], the reactance parameters vary linearly with the number of turns, which,
according to [2] does not seem to be a good approximation of reality. As such, these parameters were
assumed to change with the square number of turns.
The implementation of the DHPST also required two OLTCs per winding on the series windings, due
to the existent taps on the source and load sides.
Figure 27 shows the simulation model for the DHPST.
Figure 27 – Simulink model of the DHPST with a resistive load23.
23
This figure considers the nomenclature used for the DHPST, and, as such, primary and secondary sides are
replaced by source and load, while primary and secondary windings were replaced by SWs and EWs.
Page | 37
4.2.1 ‐ Primary Winding Segmentation
Each SW of the DHPST is composed by two OLTCs, one for the source terminals and another for the
load terminals. Each OLTC moves its terminals up and down the SW to create the phase shift across the
transformer. The two OLTCs are moved alternately in such a way that their tap steps maintain a
symmetrical distance from the midpoint of the series windings24.
The strategy used to simulate the two OLTCs was the creation of distinct segments in each of the
series windings. This is done by varying the number of turns of the upper and lower segments25.
The three main situations are represented below in Figure 28. For the middle tap position (Figure
28(a)), the winding has two VSB segments. For the full tap positions D=1 or D=‐1, (Figure 28(b)), the
winding has a single CSB segment fig (c) and for the regulated tap positions (‐1<D<0 or 0<D<1) (Figure
28(c)) we have two VSB and one CSB segments, totalizing 3 segments.
Regarding the most complicated situation, (Figure 28(c)), the SW is divided into 3 segments.
Segments number 1 and 3 are assumed to be of the same length and segment 2 is given by the total length
of the SW minus the sum of the length of the other two. In this situation the VSB segments are represented
in blue and the CSB segment is represented in light orange.
The schematic of the 3 segment SW used to simulate tap positions for ‐1<D<0 in simulink along with
a detailed description of its blocks can be found in Figure 29.
24
The manufacturer specifies that no more than one step off of the symmetrical positions is allowed to
maintain the health of the transformer [49].
25
And, consequently, the middle segment, as the total number of turns per winding remains the same.
Page | 38
Figure 29 – Detailed schematic of the series winding with 3 segments in the Simulink model.
the upper VSB block. As the upper and lower VSB blocks have the same amount of turns, they have the
same resistance and inductance, and therefore, RX 1 LLX 1 RX 2 LLX 2 . Ri C i represents the snubber
mesh which provides an increased stability to the model.
The impedances vary depending on the D chosen because the number of turns of each block changes.
Being RX LLX the total resistance and inductance of the winding, the values of the resistance and
The D inserted in the model should always be a positive number. For simulating a tap position for
0<D<1 we switch the position of the source and load tap wires.
In order to provide proper turn‐to‐turn fault testing, extra segments were added. FS1 (Faulted
Segment 1) and FS2 (Faulted Segment 2) presented in red, are segments that can be short‐circuited in
order to emulate turn‐to‐turn faults26 in the middle and lower segments respectively. These segments are
also of the VSB type. A representative image of the winding design for this situation on the Simulink model
can be seen in Figure 30.
26
The simulation of turn‐to‐turn faults, is carried out by connecting a short‐circuit resistor between the two
ends of segment block on the simulation model. In the laboratory the resistor is connected between two accessible
taps in the winding and the value of the resistor has to be chosen in such a way that the current flowing on the
shorted segmented does not exceeding the rated segment current.
Page | 39
Figure 30 ‐ Series Winding segmentation with a turn‐to‐turn fault in the Simulink model.
4.2.2 ‐ Secondary Winding Segmentation
Similarly to the SWs, segmentation of the secondary was also done. However, the segmentation on
the EW was only for short‐circuit testing as there are no OLTCs in these windings. The two segments used
were of the CSB type. A simple schematic of the segmentation of the EW is presented in Figure 31, in
which n2 represents the number of turns of the EW and nX , the number of shorted turns.
U2
n2‐nX Healthy
segment
Short‐
nX circuited
segment
X2
Figure 31 ‐ Segmented EW with a turn‐to‐turn fault at the bottom with nX short‐circuited turns.
4.2.3 – Testing of Segmentation
To test the validity of the segmented windings the simulation model was configured as a standard 3‐
phase transformer. We proceeded to obtain currents from two different measurements (Figure 32) while
leaving the secondary in open‐circuit. The first one was the measurement of the excitation current, I0 of
the transformer under healthy conditions (Figure 32(a)). For the second, a short‐circuiting resistance was
added and we measured the primary post fault current, I1 , and the current on the short circuit resistance
, ICC , (Figure 32(b)).
Page | 40
I0 I1
ICC
NP NP
NX
(a) (b)
Figure 32 – Current measurement for the testing of segmentation: (a) Excitation current, I0; (b) Post fault winding
current, I1, and current on the short‐circuit resistance, Icc.
'
Due to the SC resistance there will be a superimposed current flowing through the winding ( ICC ) in
the post‐fault situation when compared to the healthy situation. This current is given by:
ICC
'
ICC , (34)
NP
1
NX
of the non‐short‐circuited and short circuited segments respectively. This phenomena occurs due to the
autotransformer effect [50].
Using Labview and the obtained currents we established a comparison between the short‐circuit
current I1 and the expected current, which, by principle, should to be equal to I0 (pre‐fault excitation
'
current) with a superimposed current ICC . (35) translates this principle.
I1 I0 ICC
'
. (35)
including short‐circuit of top, middle and bottom segments of the segmented winding and also more than
one short‐circuit at a time, all of these returning the same negligible difference between I1 and I0 ICC
'
proving this way the validity of the segmentation of the transformer’s model. Varying the size of the
segments and short‐circuit resistance we obtained the expected voltages and currents. Also, replacing the
non‐segmented winding for the segmented one, bypassing it (by not using the tapped windings), we
obtained equal current and voltage waveforms. From this we can conclude that the segmentation is
working according to theory, hence, applicable to the model.
Page | 41
Chapter 5 – Analysis of Results
5.1 – Simulation Model Validation
5.1.1 – Phase Angle Shift at Full Tap
In [2] the authors present a short‐circuit model for the DHPST. Through the phasor diagram we can
obtain the positive‐sequence impedance and we can further calculate the theoretical full‐tap phase shift
of the DHPST according to:
sin
n1
2 (36)
n2
sin 60
2
5.1.2 ‐ Experimental Results Comparison
In order to validate the results, several experimental tests were done to compare simulation and
experimental results. In the interest of not making an extended and thorough exposition of the results,
only the most significant ones will be shown here. All the simulation values were acquired after 13 seconds
of simulation in order to obtain steady‐state conditions.
As we do not achieve a perfect sinusoid wave from the power grid, a spectral analysis of the grid
waveforms was done and these values were used to implement the voltage waveforms used in the model,
in order to attain a superior accuracy.
In both the model and the physical transformer the following measurements were performed at a
rated voltage of 237 V phase‐to‐phase. The figures in the following measurements represent source, load
and excitation currents and load voltages waveforms for simulation, right column and experimental
results (left column). Source voltage waveforms were not shown here due to the similarity between the
two.
Page | 42
5.1.2.1 ‐ Measurement nº1
The figure below shows a comparison between experimental and simulation results for the DHPST
at full tap retard (D=1) under no‐load conditions.
(a) (b)
Figure 33 – Measurements of source and excitation currents, and load voltages for no‐load conditions at full tap
retard (D=1): (a) experimental results; (b) and simulation results.
By analysis of the load voltages waveform it is possible to observe the similarity between the two in
terms of shape, amplitude and phase. In the source current waveforms there is a slight difference in the
amplitude and in the excitation currents, a minor difference in the shape between the experimental
results and the simulation. This difference occurs due to the fact that the model is still in its experimental
phase and still requires a fine tuning. A few adjustments may be required in order to improve these
results.
5.1.2.2 – Measurement nº2
The next figure shows a comparison between experimental and simulation results for the DHPST for
an advance tap position (D=‐0,3) with a 42Ω/phase resistive load.
Page | 43
(a) (b)
Figure 34 – Measurements of source, load and excitation currents, and load voltages for a 42Ω/phase resistive load
at an advance tap position (D=‐0,3): (a) experimental results; (b) and simulation results.
Through the observation of the waveforms we can observe the similarity between the experimental
and simulation waveforms in terms of phase, shape. The amplitude in the excitation, load and source
currents is slightly lower in the experimental measurements but, in general, the simulation results are
quite acceptable.
Page | 44
5.2 – Protection and Fault Diagnosis Results
For testing the protection and fault diagnosis methods, we used data acquired from the simulation
and from the laboratory. As the simulation model has been previously presented and in order not to
extend this document, the experimental setup and a detailed explanation of the data acquisition system
and measurements can be found in Appendix B.
In order to analyze and provide protection and fault diagnosis, we developed a program using
National Instruments (NI) Labview that is capable of interpreting the offline current and voltage
measurements and provide a simple visual trip signal and fault diagnosis. Furthermore, the program is
able to read and interpret both the simulation files from the Simulink model and the binary file containing
the experimental data measurements. This allow us to easily compare results.
The program comprehend Sub‐VIs that are responsible for providing info and/or send a trip signal in
the shape of a boolean array containing the trip signal, the specific phase and, indication of fault. They
may also contain signals related to the state of operation such as, indication of no‐load, correct phase
sequence and retard or advance phase shift.
The program has modular design and has blocks for measuring the voltages, currents, active power
and reactive power, in RMS and instantaneous values, on both the source and load sides. The program
also contains the estimation of the tap position D by using (12), presented in 3.5.2.2. The value of D is
important for the calculation of the differential currents27.
Furthermore, using the measurements obtained, several relays were implemented in order to
provide protection for overvoltages, overcurrents and overloads and also detect unequal phase
displacement, unequal tap positions (by tap position estimation), incorrect phase sequence and abnormal
losses28.
5.2.1 – Differential Protection Results
The differential protection (DP) implemented is based on the Linear Current Differential Principle
(LCDP), (13), with a definite time relay. It measures the peak amplitude of the differential signal waveform.
If in any phase the differential current amplitude is larger than the threshold established for the worst
case situation (D=0)29, a trip signal is produced. PCP DP was also implemented. However, this method
seemed unreliable when the position of the taps started to move further away from the full tap positions.
27
The predetermined value for D should range from ‐1 to 1. However, there may be an associated offset due
to the fact that in reality a single tap difference may be present [2]. The tap estimator does not work for no‐load
conditions.
28
P should not exceed PPI (pickup for iron losses) which is about 5% of the nominal MVA plus a slope K of
about 0.05 related to the copper losses and additional security margin for CT errors [9].
29
D=0 is the tap position in which the differential currents have the larger amplitude.
Page | 45
In this subsection we considered just the results obtained experimentally for the tap positions of D=‐
0.3 and D=1, at no‐load and on‐load for the cases 1 (no fault), 2 (fault on the EWs) and 3 (fault on the SWs)
in order not to over‐extend this document. Furthermore, for the case of shorted turns, a turn‐to‐turn fault
of 5% of the total number of turns of the winding was considered, using a 0.94 Ω shunting resistor.
The line represented in orange in the figures below establishes the tripping threshold for differential
currents superior to 0.65 APeak or inferior to ‐0.65 APeak. For currents whose instantaneous absolute value
is below 0.65 A the protection will restrain from operation. As the amplitude of the differential currents
varies with the tap position, the value of the threshold was established for the worst case situation, when
D=0, using the simulation model.
5.2.1.1 – Healthy Operation
This case comprehends the transformer working under normal conditions without any fault. Figures
35 and 36 show the differential currents for the tap positions D=‐0.3 and D=1 for this situation. Figure 35
shows the results for no‐load conditions and Figure 36 for on‐load conditions.
(a) (b)
Figure 35 – Differential currents for the no‐load operation of the DHPST in healthy conditions: (a) advanced tap
position (D=‐0.3); (b) full retard tap position (D=1).
(a) (b)
Figure 36 – Differential currents for the on‐load operation, with a 42 Ω/phase resistive load, of the DHPST in
healthy conditions: (a) advanced tap position (D=‐0.3); (b) full retard tap position (D=1).
Page | 46
For the no fault case, as expected, the relay restrains from operation as the differential currents are
below the established threshold. It is possible to observe that the tap position affects indeed the
amplitude of the differential currents on both no‐load and on‐load situations.
5.2.1.2 – Turn‐to‐turn Fault in the Phase A of the Excitation Windings
This case comprehends the transformer working with a turn‐to‐turn fault of about 5% of the turns
shorted in phase A of the EWs. Figure 37 shows the results for no‐load conditions and Figure 38 for on‐
load conditions.
(a) (b)
Figure 37 – Differential currents for the no‐load operation of the DHPST with a turn‐to‐turn fault in the excitation
winding of phase A: (a) advanced tap position (D=‐0.3); (b) full retard tap position (D=1).
(a) (b)
Figure 38 – Differential currents for the on‐load operation, with a 42 Ω/phase resistive load, of the DHPST with a
turn‐to‐turn fault in the excitation winding of phase A: (a) advanced tap position (D=‐0.3); (b) full retard tap
position (D=1).
For this case it is possible to observe that the relay would trip for all the on‐load conditions but not
for all the no‐load conditions. For full‐tap positions, D=1 in this case, at no‐load the relay would not trip.
This occurs due to the dependency of the differential currents on the tap position of the DHPST. The
solution for this problem is to create a variable threshold that depends on the tap position of the DHPST.
Page | 47
It should be noted that for the fault testing a shunt resistance was used in order not to exceed the
rated current on the winding and on real world situations as this resistance is much smaller, the amplitude
of the differential currents would be larger and therefore, it would be easier for the DP to detect the fault.
Other particular aspect that should be noted is that using the differential currents in order to identify
the phase affected by the fault does not seem to be easily achievable. Figure 37 is a good example of this.
Knowing that the affected winding is the EW in phase A, we can observe that the differential current with
larger amplitude for D=‐0.3 is I DIFFA . For D=1 however, I DIFFA is the current with smaller amplitude.
5.2.1.3 – Turn‐to‐turn Fault in the Phase A of the Series Windings
This case comprehends the transformer working with a turn‐to‐turn fault of about 5% of the turns
shorted in the SWs of phase A. Figure 39 shows the results for no‐load conditions and Figure 40 for on‐
load conditions.
(a) (b)
Figure 39 ‐ Differential currents for the no‐load operation of the DHPST with a turn‐to‐turn fault in the series
winding of phase A: (a) advanced tap position (D=‐0.3); (b) full retard tap position (D=1).
(a) (b)
Figure 40 ‐ Differential currents for the on‐load operation, with a 42 Ω/phase resistive load, of the DHPST with a
turn‐to‐turn fault in the series winding of phase A: (a) advanced tap position (D=‐0.3); (b) full retard tap position
(D=1).
This case also demonstrates that the DP is not always sensitive. In Figure 39 the amplitude of the
differential current would barely trip the DP. This is a situation where 5% of the turns are shorted and
Page | 48
since the winding has 220 turns, about 11 turns would be shorted. Once again, without the shorting
resistor, the differential currents would be larger and therefore, easier for the DP to detect the fault.
Comparing the results with the fault in the SWs and in the EWs, it is possible to observe that the
differential currents are higher for the case of faults in the SWs than for a fault on the EWs. This occurs
due to the larger number of shorted turns on the SW (5% of 220) as opposed to the number of shorted
turns on the EW (5% of 127)30.
5.2.2 – Diagnosis of Faults
Based on the new differential current EPVA we obtained the spectral analysis of the PV modulus and
the active and reactive power differentials, PDIFF and QDIFF and their severity factor, SF, SFP and SFQ .
Furthermore, several plots were made in order to identify the angle of the major axis of ellipse of
differential current PV and thus identify the fault location. A 4th order low‐pass Butterworth filter with a
cutoff frequency of 120 Hz was applied to the differential currents PV plots in order to filter higher
frequencies (higher than 120 Hz), which are not important for this study. All the faults were implemented
with 5% of shorted turns in the respective winding.
5.2.2.1 – Healthy Conditions
Figure 41 shows the differential current PV plots, from the data acquired in laboratory, for different
tap positions, ranging from full‐tap advance (D=‐1) to full‐tap retard (D=1), in healthy conditions, at no‐
load and with a 42 Ω/phase resistive load. The elliptical shape of the differential current PV is the
characteristic shape for normal working conditions.
Figure 41 ‐ Experimental results for the Park’s Vector of the differential current regarding multiple tap positions: a)
no‐load; b) 42Ω resistive load per phase.
30
This is proven in Appendix D where we plotted the differential currents PV for the DHPST with equal number
of turns in the EWs and SWs and we short circuited 5% of the windings in each. The differential currents PV was the
same in both cases, proving that having a fault in the EWs or in the SWs affects the transformer in the same way.
Page | 49
Figure 42 presents the simulation results under the same conditions as previously presented. In the
simulation, besides the results for the tap positions of the experimental results, we are also able to
simulate the middle tap position which is impossible to implement in the practical experiment due to
limitations of the accessible tap positions. The detailed existent taps of the windings of the transformer
can be found in Appendix B.2 and the simulated waveforms of the source and load currents and voltages
as well as the excitation currents obtained for D=0 are presented in Appendix D.
Figure 42 ‐ Simulation results for the Park’s Vector of the differential current regarding multiple tap positions: a)
no‐load; b) 42Ω resistive load per phase.
Comparing the laboratory results (Figure 41) with the ones obtained from simulation (Figure 42), it
is possible to visualize the similarity between the angles of the major axis and shapes of the ellipses, thus
attesting the reliability of the simulation data regarding the PV’s results. Furthermore, through attentive
observation, it is possible to notice a minor difference on the overall size of the ellipses. This occurs due
to a slight simulation inaccuracy in the amplitude of the waveforms which can be corrected by precisely
tuning the parameters of the model.
Table 1 shows the differential current PV angles of the major axis of the locus. The orange color in
this table sets the conditions that are not possible to test experimentally due to the non‐availability of the
larger resistive load (16 Ω) and unavailable taps to simulate the middle tap position (D=0).
By observation of the experimental and simulation results of the differential currents PV in Table 1,
we can observe that for normal conditions, the angle of the major axis of the ellipse varies almost linearly
with the tap position and is also dependent on whether the PST is loaded or not. The central tap position
(D=0) at no‐load conditions was established as our baseline. This base angle of the locus, B , was
measured at approximately 118⁰ 31. This angle increases or decreases depending on whether the PST is
31
This result was not obtained experimentally as our transformer does not allow a middle tap position due to
the unavailability of these taps. It was however, obtained by averaging the angle of the obtained taps resulting the
angle for the middle tap position (D=0). This was possible as the taps we considered are symmetrical and the
variation is linear. This result was confirmed with the simulation results where we simulated the middle tap position.
Page | 50
working on retard or advance mode by summing half of the phase shift angle, , between the source and
the load. As the maximum and minimum phase‐shift is approximately ± 77.6⁰. The angle of the locus
should vary between approximately 80⁰ and 158⁰ for full‐advance and full‐retard positions.
Table 1 – Differential current PV angle of the major axis of the locus for different tap positions, under normal
operating conditions at no‐load and on‐load conditions.
Tap Position Angle (deg)
No Load Estimated Load 42 (Ω) Estimated Load 16 (Ω)
D=‐1 81.5 79.6 103.0 101.9
D=‐0.3 104.2 105.0 129.6 126.5
Experimental D=0
D=0.3 129.4 131.7 154.2 153.3
D=1 158.2 158.1 179.3 179.3
D=‐1 81.5 79.5 106.1 101.6 106.5
D=‐0.3 104.1 104.8 130.4 126.1 131.7
Simulation D=0 117.3 118.3 144.8 139.5 146.1
D=0.3 131.1 131.9 157.5 153.0 158.6
D=1 160.8 157.1 181.0 179.2 181.4
Still on Table 1, it is possible to notice that under load conditions the no‐load differential current PV
angle is increased by a fairly constant load factor, LF , of approximately 23.2⁰. This value does not vary
significantly under different resistive load values for normal working conditions. As such we developed
(37), below, to allow us to calculate the predicted differential current Park’s Vector angle of the major
axis of the ellipse, e , for a given α and load or no‐load conditions.
e B LF . (37)
2
Any significant deviation between the obtained angle, and the calculated angle, e , will represent a
fault. Depending on the difference it is possible to extrapolate the affected phase. Table 1 shows the
estimated angles with (37). Due to the lack of major differences between the obtained and estimated
angles, we can conclude that there is no fault.
Table 2 shows the SF, SFP and SFQ for normal working conditions, under no‐load and on‐load
situations for different tap positions. These values can be established as a threshold for this PST as these
reflect its typical asymmetry. Under faulty situations, in general, the obtained results will always be
superior to these.
Page | 51
Table 2 – Experimental results for the SF, SFP and SFQ for different tap positions, under normal operating
conditions at no‐load and on‐load conditions.
Figure 43 ‐ EPVA, PDIFF and QDIFF spectrums for the experimental results obtained at no‐load, under healthy
conditions regarding D=‐0.3 and D=1 tap positions.
As it is possible to observe in the previous figure, the tap position has a major influence in the
amplitude of the spectrums of the EPVA, PDIFF and QDIFF, both in the fundamental frequency and in the 100
Hz (2nd harmonic) component. The closer the tap position is to the middle tap (D=0), the larger the DC and
Page | 52
100 Hz component. As both the DC and 100 Hz components increase, the SF, SFP and SFQ remains relatively
constant as it is possible to observe from Table 2.
Figure 44 presents the EPVA, PDIFF and QDIFF spectrums for the experimental results obtained with 42
Ω/phase resistive load under healthy conditions regarding D=‐0.3 and D=1 tap positions in which the SFs
of Table 2 are based on.
Figure 44 ‐ EPVA, PDIFF and QDIFF spectrums for the experimental results obtained with 42 Ω/phase resistive load
under healthy conditions regarding D=‐0.3 and D=1 tap positions.
Under load conditions the EPVA and QDIFF spectrum amplitudes show minor differences when
compared with the no‐load conditions, both in the DC and 100 Hz components. The PDIFF spectrum
however, shows a difference in the DC component while the 100 Hz component remains similar. As such
SFP will be decreased under load conditions while SF and SFQ will remain relatively constant. This can be
observed in Table 2.
Table 3 shows the simulation results for the SF, SFP and SFQ for different tap positions, under
normal operating conditions at no‐load and on‐load conditions.
Table 3 – Simulation results for the SF, SFP and SFQ for different tap positions, under normal operating conditions at
no‐load and on‐load conditions.
The difference between these values and the values of Table 2 are due to a need of fine tuning the
simulation model as previously stated.
5.2.2.2 – Turn‐to‐turn Fault in the Phase A of the Excitation Windings
Figure 45 ‐ Experimental results for the Park’s Vector of the differential current with a 5% turn‐to‐turn fault in the
excitation winding of phase A for D=‐0.3 and D=1: a) no‐load; b) 42Ω resistive load per phase.
Figure 45, above, shows the differential current PV for a turn‐to‐turn fault on 5% of the turns of the
excitation winding on phase A. It is possible to visualize the elongation of the ellipse, which reflects the
increase of the severity factor, as well as a phase shift regarding the normal working situation. This phase
shift is dependent on the phase affected by the fault. The elongation is dependent on the number of
shorted turns. It is still possible to visualize the relatively constant phase shift between different tap
positions and the phase shift between the no‐load and load conditions, even though the difference is
lessened by the fault32. Table 4 contains the differential current PV angle of the major axis of the locus for
the no‐load and on‐load situations for the 2 tap positions both simulation and experimental for a turn‐to‐
turn fault in the EW of phase A.
Table 4 ‐ Differential current PV angle of the major axis of the locus for different tap positions with a turn‐to‐turn
fault of 5% on the EW of the phase A at no‐load and on‐load conditions.
The results presented in the previous table shows the similarity between the obtained results both
from the simulation and the experimental results. It is possible to observe that the fault diminishes the
value between angles at no‐load and on‐load conditions.
The disparity of results between the obtained and estimated angles at both no‐load and on‐load
conditions, give us the indication of a fault. The red color indicates the clear existence of a fault.
Table 5 presents the SF, SFP and SFQ for the above situation.
Table 5 ‐ SF, SFP and SFQ for different tap positions with a turn‐to‐turn fault of 5% on the EW of the phase A at no‐
load and on‐load conditions.
Figure 46 ‐ EPVA, PDIFF and QDIFF spectrums for the experimental results obtained at no‐load, with a turn‐to‐turn
fault of 5% on the EWA regarding D=‐0.3 and D=1 tap positions.
Page | 55
Figure 46 contains the EPVA, PDIFF and QDIFF spectrums for the experimental results obtained at no‐
load, with a turn‐to‐turn fault of 5% on the EWA. Comparing with Figure 43 which is the same
measurement but in healthy conditions it is possible to observe that for both D=‐0.3 and D=1 amplitude
of the DC component and the 100 Hz component increased in the EPVA and PDIFF. In QDIFF however, the DC
component remained nearly the same while the 100 Hz component nearly doubled. The 100 Hz
component could be a clear indicator of the existence of a fault. We have however, to take into account
the influence of the tap changer. Comparing D=‐0.3 with D=1 it is possible to observe that the tap changer
greatly influences the amplitude of the spectrums.
For the conditions stated above with the exception of the transformer being connected to a resistive
load and in comparison with Figure 44, no further conclusions could be obtained apart from the ones
previously stated and as such the figure will not be shown.
Table 6 ‐ Simulation results for the SF, SFP and SFQ for different tap positions with a turn‐to‐turn fault of 5% on the
EW of the phase A at no‐load and on‐load conditions.
5.2.2.3 – Turn‐to‐turn Fault in the Phase A of the Series Windings
Figure 47 ‐ Experimental results for the Park’s Vector of the differential current with a 5% turn‐to‐turn fault in the
series windings of phase A for D=‐0.3 and D=1: a) no‐load; b) 42Ω resistive load per phase.
Page | 56
Above, in Figure 47, as opposed to Figure 45, the fault is in the series windings instead of the
excitation windings. The angles between the two are very similar in both images, this way, it is not possible
to differentiate between faults on SW and EW. However, it is possible to visualize a larger elongation of
the ellipses, and hence the SF, in the case of the fault on the series windings. This occurs due to the fact
that, even though the percentage of shorted turns is the same on both the SWs and EWs, the SW have a
larger number of turns (220 as opposed to 127), hence the shorted turns are 11 and 6 for the SW and EW
respectively 33.
Table 7 contains the PV angle for the no‐load and on‐load situations for the 2 tap positions both
simulation and experimental for a turn‐to‐turn fault in the SW of the phase A.
Table 7 ‐ Differential current PV angle of the major axis of the locus for different tap positions with a turn‐to‐turn
fault of 5% on the SW of the phase A at no‐load and on‐load conditions.
Table 8 – Experimental results for the SF, SFP and SFQ for different tap positions with a turn‐to‐turn fault of 5% on
the SW of the phase A at no‐load and on‐load conditions.
33
To verify this we simulated a transformer with the same number of turns in the series and excitation windings
and we short‐circuited 5% of the turns in each winding separately. We obtained similar curves, angles and severity
factors proving that both affect the transformer in the same way. For further details refer to Appendix D.
Page | 57
In the table above, it is possible to see that the larger number of turns as compared with Case 2
makes the elongation of the ellipses larger, thus increasing the SF. This is also visible on SFP and SFQ.
Table 9 – Simulation results for the SF, SFP and SFQ for different tap positions with a turn‐to‐turn fault of 5% on the
SW of the phase A at no‐load and on‐load conditions.
5.2.2.4 – Turn‐to‐turn Fault on Phase A, B and C of the Excitation Windings
Figure 48 shows a fault with 5% of the turns shorted on the EWs. In this case the faults are presented
in phases A, B and C in both no‐load and on‐load conditions. The first row represents the no‐load situation
for phases A, B and C with the letters a), b) and c) respectively and the second row represents the no‐load
situation with a 42Ω resistive load per phase with the letters d), e) and f) for phases A, B and C respectively.
It is possible to visualize the angle difference caused by the same fault applied to different phases.
The angle difference between different tap positions is relatively steady but the difference between load
and no‐load decreases due to the fault when comparing with the no fault situation.
For this case, phases B and C were not experimentally tested. Figure 48, shows the simulation results
for the differential current Park’s Vector with a 5% turn‐to‐turn fault in the EWs of phases A, B and C for
D=‐0.3 and D=1 under no‐load and loaded conditions
34
Further explanation can be found in Appendix D.
Page | 58
Figure 48 ‐ Simulation Results for the Park’s Vector of the differential current with a 5% turn‐to‐turn fault in the
excitations windings of phases A, B and C for D=‐0.3 and D=1 under no‐load and loaded conditions.
In the figure above it is possible to visualize the PV angle of the major axis of the locus for the same
fault on different phases. The angle changes drastically depending on the affected phase.
Table 10 shows in more detail the PV angle of the major axis of the locus for the above situation.
Table 10 ‐ Differential current PV angle of the major axis of the locus for different tap positions with a turn‐to‐turn
fault of 5% on different phases of the EW at no‐load and on‐load conditions.
Page | 59
As it is possible to observe from the table above the estimated angle values are nearly the same for
the same conditions of tap position and load. The estimated value is calculated based on (37) and,
depending on which winding is affected, the angles between the estimated and obtained values present
a different variation. For a turn‐to‐turn fault on the EWC the difference is the smallest and for D=1
specifically, the difference is rather small. Therefore, methods based on the SF, SFP and SFQ help in the
increase of the reliability of this method in order to detect a fault.
Table 11 presents the simulation results for the SF, SFP and SFQ when the same fault on the EWs
occurs in different phases (A, B and C). The results presented are for different tap positions with no‐load
and on‐load conditions.
Table 11 ‐ Simulation results for the SF, SFP and SFQ for different tap positions with a turn‐to‐turn fault of 5% on the
EWs of the phases A, B and C at no‐load and on‐load conditions.
Page | 60
Chapter 6 – Conclusions
During this dissertation, the working principles and different types of phase‐shifting transformers
were presented, as well as their applications. Additionally, faults, protections and diagnostic tools for this
type of transformers were also explored. A simulink model of an existing transformer was adapted in
order to match the working principle of a PST and a comparison was established between the theoretical
model and the experimental results. Furthermore, several existent differential protection schemes were
implemented, and a new method, based on the EPVA of the differential currents was developed for the
diagnostic of turn‐to‐turn short‐circuits in the windings
The differential protection methods based on ampere‐turn equations implemented in this thesis did
not prove to be very effective in case of minor turn‐to‐turn faults. Most of the times the detection of this
type of faults is only possible once the fault has already reached a considerable extent. The capability of
fault detection seems to be very dependent on the tap position, and as such, for tap positions near the
full tap, the differential currents are very small and, therefore, are not reliable for the detection of minor
faults. The differential currents depend not only on the tap position but also on the affected phase. As
such, the identification of the fault location using the differential current seemed to be difficult, if not
impossible. In conclusion, differential methods must be implemented but their effectiveness is still not
good enough for minor fault detection and these do not provide reliable information regarding the
affected phase.
The new method developed for fault diagnosis was tested under no‐load and loaded conditions for
different tap positions, both in the simulation and in the laboratory. It proved to be a reliable method for
fault diagnosis, being able to detect even relatively minor turn‐to‐turn faults and identify the
corresponding affected phase.
The EPVA, based on the differential currents, which encompasses the severity factor, can be a good
method of detecting a fault, even though it may not always be accurate on the estimation of its extent.
This happens because the severity factor may vary quitely according to the tap position and phase of the
fault. The severity factor is an important indicator of the fault severity. However, it may depend on the
affected phase. Similarly, the methods based on the active and reactive power spectrum present the same
issues. The new method developed assisted with methods based on the power spectrum of the active and
reactive power differential are very effective in diagnosis of faults.
The proposed fault diagnosis method is able to detect the faulty phase. It is not, however, capable
of detecting if the fault is located in the series or excitation windings. In a real world application this is not
an issue because a turn‐to‐turn fault requires the disassembly of the transformer and all the windings of
the affected phase.
Page | 61
In the simulation results, for the same percentage of shorted turns, turn‐to‐turn faults in the
excitation windings tend to have a smaller severity factor and (active and reactive severity factor) than
the faults in the series windings. This is exclusively due to the difference in the number of shorted turns,
as the number of turns in the SWs and EWs is different.
No previous publications regarding the use of Park’s Vector analysis and EPVA for this type of
transformers could not be found. Thus, it is a unique and accurate electrical method for fault diagnosis in
PSTs. As the presented method is based on the differential current given by the load currents minus the
phase‐shifted source currents, it can also be applied to other types of PSTs, requiring just the calculation
of the phase angle shift.
The model developed in [28] proved to be a powerful tool when dealing with transformers. It allowed
us to test the differential protection and the diagnosis without the need of data acquisition from the
physical transformer and also, testing situations that would otherwise be destructive or dangerous. Also,
the development of the segmented windings was tested in comparison with a non‐segmented winding
and besides presenting similar conditions to the ones of a standard transformer, it also proved to be
reliable for fault testing.
In the future, it would be interesting to implement other types of differential protection and test
for both, minor faults, such as turn‐to‐turn, and severe external faults, such as phase‐to‐phase or phase‐
to‐ground faults. Additionally, it would be interesting to test the presented diagnostic solution with other
types of PSTs. Furthermore, an on‐line diagnostic tool could be developed, in order to allow real‐time fault
diagnosis. This tool could also integrate several types of protections in order to provide not just diagnosis
but also fast action in case of a fault.
Page | 62
References
[1] E. M. Carlini, S. Favuzza, S. E. Giangreco, F. Massaro and C. Quaciari, "Uprating an
overhead line. Italian TSO applications to increase system security", in International
Conference on Renewable Energy Research and Applications (ICRERA), pp. 878‐883, 20‐
23 October 2013.
[2] B. Kasztenny and E. Rosolowski, "Modeling and Protection of Hexagonal Phase‐Shifting
Transformers ‐ Part I: Short‐Circuit Model", IEEE Transactions on Power Delivery, vol. 23,
no. 3, pp. 1343‐1350, 2008.
[3] National Instruments, 1 September 2014. Available
from:http://zone.ni.com/reference/en‐XX/help/373375B‐01/lveptconcepts/ep_grids/.
[4] ABB, Transformers Handbook, 2004.
[5] B. Kasztenny and E. Rosolowski, "Protection of Hexagonal Phase Shifting Transformer", in
IET 9th International Conference on Developments in Power System Protection (DPSP),
pp. 144‐149, 17‐20 March 2008.
[6] J. Grainger and W. Stevenson, Power System Analysis, McGraw‐Hill, 2004.
[7] J. K. Bladow and A. Montoya, "Experiences with parallel EHV phase shifting transformers",
IEEE Transactions on Power Delivery, vol. 6, no. 3, pp. 1096‐1100, 1991.
[8] Nicklas Johansson, "Control of Dynamically Assisted PST's", Licentiate Dissertation, KTH
Royal Institute of Technology, Stockholm, 2008.
[9] B. Kasztenny and E. Rosolowski, "Modeling and Protection of Hexagonal Phase‐Shifting
Transformers ‐ Part II: Protection", IEEE Transactions on Power Delivery, vol. 23, no. 3,
pp. 1351‐1358, 2008.
[10] Umar Naseem Khan, "Modeling and Protection of Phase Shifting Transfomers", Doctoral
Dissertation, University of Western Ontario, Canada, November 2013.
[11] J. Verboomen, D. Van Hertem, P. H. Schavemaker, W. L. Kling and R. Belmans, "Phase
shifting transformers: principles and applications", in International Conference on Future
Power Systems, 18 November 2005.
[12] "IEEE Guide for the Application, Specification, and Testing of Phase‐Shifting
Transformers", IEEE Standard C57.135‐2001, 2012.
[13] Gustav Preininger, Electric Power Transformer Engineering, 1st ed., ed. James H. Harlow,
CRC Press, 2004.
[14] J. C. Jimenez and C. O. Nwankpa, "Circuit model of a phase‐shifting transformer for analog
power flow emulation", in IEEE International Symposium on Circuits and Systems (ISCAS),
pp. 1864‐1867, 15‐18 May 2011.
[15] P. Bresesti, M. Sforna, Vittorio Allegranza, D. Canever and R. Vailati, "Application of phase
shifting transformers for a secure and efficient operation of the interconnection
corridors", in IEEE Power Engineering Society General Meeting, pp. 1192‐1197, 10 June
2004.
[16] M. J. Thompson, H. Miller and J. Burger, "AEP experience with protection of three
delta/hex phase angle regulating transformers", in Power Systems Conference: Advanced
Metering, Protection, Control, Communication, and Distributed Resources, pp. 96‐105,
13‐16 March 2007.
[17] R. Sweeney, G. Stewart, P. O'Donoghue and P. Smith, "The specification, design and
protection of phase shifting transformers for the enhanced interconnection between
Page | 63
Northern Ireland and the Republic of Ireland", in Seventh International Conference on AC‐
DC Power Transmission, pp. 274‐281, 28‐30 November 2001.
[18] Mustafa Lahloub, ABB Red TIE Series Transformer Failure Modes, p. 57, 2013.
[19] W. Bartley, "Analysis of Transformer Failures", International Association of Engineering
Insurers 36th Annual Conference, Stockholm, Sweden, 2003.
[20] C. de Jesus Ribeiro, A. P. Marques, C. H. B. Azevedo, D. C. P. Souza, B. P. Alvarenga and R.
G. Nogueira, "Faults and defects in power transformers ‐ a case study", in Electrical
Insulation Conference, EIC 2009, IEEE, pp. 142‐145, 31st May ‐ 3rd June 2009.
[21] Alstom, Network Protection & Automation Guide, Alstom Grid, 2011.
[22] Z. Gajic, I. Brncic, B. Hillström, F. Mekic and I. Ivankovic, "Sensitive Turn‐To‐Turn Fault
Protection For Power Transformers", in Proc. 32th Annual Western Protective Relay
Conference, 2005.
[23] L. M. R. Oliveira, A. J. M. Cardoso and S. M. A. Cruz, "Power transformers winding fault
diagnosis by the on‐load exciting current Extended Park’s Vector Approach", Electric
Power Systems Research, vol. 81, no. 6, pp. 1206‐1214, June 2011.
[24] Martin J. Heathcote, The J & P Transformer Book, 13th ed., Newnes, 2007.
[25] Zoran Gajic, "Differential Protection for Arbitrary Three‐Phase Power Transformers",
Doctoral Dissertation, Lund University, Sweden, 2008.
[26] Efacec, Autotransformador desfasador na subestação de Pedralva. July 2014. Available
from:http://www.efacec.pt/PresentationLayer/efacec_projecto_00.aspx?area=2&idiom
a=3&local=110.
[27] L. M. R. Oliveira, A. J. M. Cardoso and S. M. A. Cruz, "Transformers On‐Load Exciting
Current Park's Vector Approach as a Tool for Winding Faults Diagnostics", in International
Conference on Electrical Machines (ICEM), Brugge, Belgium, pp. 6, 25‐28 August 2002.
[28] Daniel Machado, "Modelação e simulação de transformadores de potência com núcleo
couraçado", Unpublished Master of Science Dissertation, Universidade de Coimbra, 2013.
[29] "Protection of Phase Angle Regulating Transformers", IEEE Special Publication, 1999.
[30] J. Verboomen, G. Papaefthymiou, W. L. Kling and L. Van der Sluis, "Use of Phase Shifting
Transformers for Minimising Congestion Risk", in 10th International Conference on
Probabilistic Methods Applied to Power Systems (PMAPS) pp. 1‐6, 25‐29 May 2008.
[31] X. P. Zhang, C. Rehtanz and B. Pal, Flexible AC Transmission System: Modeling and Control,
Springer, Germany, 2006.
[32] S. A. Khaparde S. V. Kulkarni, Transformer Engineering: Design and Practice, Marcel
Dekker, 2004.
[33] Robert M. D. Vecchio, Bertrand Poulin, Pierre T. Feghali, Dilipkumar M. Shah and Rajendra
Ahuja, Transformer Design Principles: With Applications to Core‐Form Power
Transformers, 2nd Edition ed., CRC Press, 2010.
[34] John J. Winders, Power Transformers: Principles and Applications, 1st ed., Power
Engineering, CRC Press, 2002.
[35] Entsoe, Phase Shift Transformers Modelling, May 2014.
[36] Rita Chaves Rebelo, "Sistemas de Monitorização de Transformadores de Potência",
Universidade do Porto, 2006.
[37] Stefan Tenbohlen, Transformer Reliability Survey, CIGRE, 2008.
[38] Didik Fauzi Dakhlan, "Modeling of internal faults in three‐phase three‐winding
transformers for differential protection studies", Master Dissertation, Technical
University of Delft, June 2009.
[39] Bingyin Xu, Xue Yongduan, Li Jing and Yu Chen, "Single phase fault detection technique
based on transient current and its application in non‐solid grounded network", in Seventh
Page | 64
International Conference on Developments in Power System Protection (IEE), pp. 141‐
144, 2001.
[40] K. Chen, C. Wang, H. Ma, K. Li, Y. Wu, Y. Zhang, H. Liu, P. Gao and T. Chen, "Fault diagnosis
and preliminary location system and method for transformer core looseness", Google
Patents, 2014.
[41] Juriaan J. Erbrink, "On‐load Tap Changer Diagnosis on High‐ Voltage Power Transformers
using Dynamic Resistance Measurements", Technical University of Delft, Delft, 2011.
[42] Rogier Jongen, Peter Morshuis, Johan Smit, Anton Janssen and Edward Gulski, "A
statistical approach to processing power transformer failure data", in 19th International
Conference on Electricity Distribution, Vienna, 21‐24 May 2007.
[43] Qingmin Li, Tong Zhao, Li Zhang and Jie Lou, "Mechanical Fault Diagnostics of Onload Tap
Changer Within Power Transformers Based on Hidden Markov Model", IEEE Transactions
on Power Delivery, vol. 27, no. 2, pp. 596‐601, 2012.
[44] Bogdan Kasztenny, Michael Thompson and Normann Fischer, "Fundamentals of Short‐
Circuit Protection for Transformers", Schweitzer Engineering Laboratories, Inc, IEEE, 2010.
[45] Bogdan Miedzinski and Daniel Pyda, "Power System Protection", Class Notes, Politechnika
Wroclawska.
[46] Gerhard Ziegler, Numerical Differential Protection: Principles and Applications, 2nd ed.,
Publicis Publishing, 2012.
[47] Bogdan Kasztenny, Gabriel Benmouyal, Héctor J. Altuve and Normann Fischer, Tutorial on
Operating Characteristics of Microprocessor‐Based Multiterminal Line Current Differential
Relay, Schweitzer Engineering Laboratories, Inc, 2011.
[48] J. Tellinen, "A simple scalar model for magnetic hysteresis", IEEE Transactions on
Magnetics, vol. 34, pp. 2200‐2206, 1998.
[49] M. J. Thompson, Hank Miller and John Burger "Innovative power flow regulating tap‐
changer control installed on multiple phase‐shifting transformers", 2008.
[50] S. M. A. Cruz, "Apontamentos de Máquinas Eléctricas I", Universidade de Coimbra, 2012.
Page | 65
Appendix A – Schematic representation of the
Quadrature Booster
Figure 49, shows the schematic representation of a quadrature booster, which is A PST based on the
autotransformer design.
USA USB USC
1 Autotransformer
3 unit
2
N
6 7
Series Unit (SU)
1 – Excitation HV winding 5 – Terciary winding
2 – HV winding 6 – Primary winding of the SU
3 – Excitation regulating winding 7 – Secundary winding of the SU
4 – Phase Shift regulating winding
Figure 49 ‐ Schematic representation of the Quadrature Booster.
Page | 66
Appendix B – Experimental Setup
For the experimental data acquisition a setup was built as in Figure 50. Due to coherence purposes
and easier visualization, the cables of the setup respect the color code used in the previous waveforms
and diagrams35.
Figure 50 – Experimental setup of the DHPST with measurement apparatus.
Experimental setup equipment for offline data acquisition:
3ph standard transformer 220V/127V, 6kVA with 9 taps/winding/phase configured as DHPST
Current transducer measurement boxes w/ 3ph current and voltage measurement
NI CRIO with 3 current + 2 voltages acquisition modules
Autotransformer
Variable Resistive load, configurable from 630 to 42 , rated for 3x1152 W
Oscilloscope, digital Multimeters and a Current clamp for online measurement
Variable resistance for SC testing
B.1 – Test Bench and Data Acquisition System
An autotransformer is connected to the grid and provides regulated voltage to the source side of the
PST36. The load side of the PST connects to the regulated resistive load (shown at the right of the
transformer). Both the source and load sides are located in the SWs and the OLTC is simulated by the
accessible taps existing in these windings. The current transducers (CTr) (measurement boxes) are
35
GREEN for phase A, BLUE for phase B and RED for phase C
36
Further details regarding the characteristics of the transformer, the transducers and the CRIO and its
modules can be found in Appendix B.
Page | 67
connected between the autotransformer and the S‐side (CTr1) for source current ( IS ) measurement,
between the L‐side and the variable load (CTr2) for load currents ( IL ) measurement and between SW and
EW in order to measure excitation currents ( IE ). The CTrs 1,2 and 3 provide the appropriate signal
conditioning needed by the CRIO platform. Voltages are measured directly by the CRIO through
appropriate input modules.
The acquired data was obtained in steady‐state conditions. For each measurement, CRIO acquires
0.8 seconds of data with a sampling frequency of 25 kHz. CRIO then outputs a binary file with 15 rows of
20000 points each, correspondent to the acquired voltages and currents for each specific phase.
B.2 – Transformer
The transformer used on our experiments is a 3‐phase, 3 legged core type transformer, with the
following electric characteristics below:
Nominal power: 6 kVA
Nominal voltage: 220/127 Vrms
Nominal current of the primary and secondary windings: 9.09/15.75 Arms
Number of turns of the primary winding: 220
Number of turns of the secondary winding: 127
Transformation ratio of the transformer: 3
Resistance value: 1.14 Ω
Leakage inductance of the primary winding: 1.33 mH
Each core leg has four windings, in which two of them were modified by the addition of a number of
tappings connected to the coils, allowing for the introduction of different percentages of shorted turns at
a few locations in the winding. The transformer has 4 windings/leg, placed as follows:
127 turns with accessible taps;
127 turns with non‐accessible taps;
220 turns with accessible taps;
220 turns with non‐accessible tap.
Figure 51 shows the transformer used, at the left, and the accessible taps on the series windings for
the phase A at the right.
Page | 68
Figure 51 – Transformer used in the experimental setup and accessible tapped winding detail.
B.3 ‐ CRIO
CRIO is a control and data acquisition system that consists of an embedded controller for
communication and processing, a reconfigurable chassis with a user‐programmable FPGA, I/O modules,
and graphical LabVIEW software for real‐time, Windows and FPGA programming.
Rugged, embedded control and monitoring system
400 MHz industrial real‐time processor for control, data
logging, and analysis
Up to 128 MB DRAM memory, 256 MB of nonvolatile
storage
2M gate, 8‐slot FPGA chassis for custom I/O timing,
control, and processing
Two 10/100BASE‐T Ethernet ports; RS232 serial port for
connection to peripherals
Table 12 – CRIO controller specifications.
The input modules used were three NI 9215 modules for measurement of the conditioned current
signal and two NI 9225 modules for voltage acquisition.
4 simultaneously sampled analog inputs, 100 kS/s
±10 Voltage measurement
16‐bit resolution
Hot‐swappable operation
‐40 to 70 °C operating range
NIST‐traceable calibration
Table 13 – NI 9215 voltage acquisition module specifications.
Page | 69
300 Vrms measurement range
3 simultaneous sampled analog inputs 50 kS/s/ch
24‐bit resolution
Built‐in antialias filters
600 Vrms channel‐to‐channel isolation
Connectors and high‐voltage backshells included
Table 14 – NI 9225 current acquisition module specifications.
B.4 – Transducers
Three transducer boxes were used for conditioning the current signal for measurement by CRIO
modules. These were used for measuring IS , IL and IEXC . These boxes also provide conditioning for voltage
signals, however these were not used.
The gains for current and voltage measurement are 3 and 50 respectively.
Figure 52 shows one of the transducers (measurement box).
Figure 52 – Transducer (measurement box).
Page | 70
Appendix C – Simulation Waveforms for the
DHPST at D=0 under healthy conditions
In this appendix, the simulation waveforms for the middle tap position (D=0) in healthy conditions
are shown below.
Figure 53 shows the waveforms for the no‐load condition and Figure 54 for the on‐load condition.
C.1 – Simulation of the DHPST for the middle‐tap position (D=0), in no‐
load conditions
Figure 53 ‐ Simulation waveforms of the DHPST for the middle‐tap position (D=0) for no‐load conditions under
healthy conditions.
Page | 71
C.2 – Simulation of the DHPST for the middle‐tap position (D=0), with a
42Ω/phase resistive load
Figure 54 ‐ Simulation waveforms of the DHPST for the middle‐tap position (D=0) for on‐load conditions, under
healthy conditions.
Page | 72
Appendix D ‐ Comparison between series and
excitation faults on a transformer with equal
sized series and excitation windings.
In order to demonstrate that faults in the SWs and in the EWs affect the transformer in the same
way we simulated a transformer in our Simulink model with the same number of turns on the SWs and
EWs ( n1 n2 220 ). Using that simulation we simulated a turn‐to‐turn fault on both the SWs and the
EWs with 5% of turns short‐circuited by a resistance. We then plotted the PV and we established a
comparison (Figure 55).
Figure 55 – PV plot of a turn‐to‐turn fault on the SWs and EWs for a transformer with n1= n2=220.
As it is possible to observe in Figure 55 the PV plot of the fault in the SWs (represented in red)
overlaps the plot of the fault on the EWs (represented in blue) for a situation when the transformer has
the same number of windings on the SWs and the EWs. This translates the impossibility to detect the
exact location of the fault by this method as the angle and the SF are nearly the same.
Other methods based on the obtained currents and voltages were tried in order to discriminate the
exact winding (SW or EW) with no viable results. From a practical point of view this is not relevant fact as
the detection of the fault by itself would require de disassembly of the transformer for repair.
Page | 73