Centralized Architecture of The Distribution Substation Automation

Tampereen teknillinen yliopisto.
Julkaisu 1122
Tampere University of Technology. Publication 1122
Jani Valtari
Centralized Architecture of the Electricity Distribution

Substation Automation - Benefits and Possibilities
Thesis for the degree of Doctor of Science in Technology to be presented with due
permission for public examination and criticism in Sähkötalo Building, Auditorium S4, at
Tampere University of Technology, on the 5th of April 2013, at 12 noon.
Tampereen teknillinen yliopisto - Tampere University of Technology

Tampere 2013
ISBN 978-952-15-3044-9 (printed)
ISBN 978-952-15-3061-6 (PDF)
ISSN 1459-2045
Abstract
Smart grid initiatives around the world show how much the control and protection
of distribution networks is expected to change within the next few years. As passive
networks with unidirectional power flow evolve into active networks with a variety
of different active resources, the requirements for distribution substations will also
change, requiring the utilities to take action. Utilities do not want to undertake con-
tinuous and costly upgrades of the whole protection system, but there is still a clear
need for adapting to new requirements. The need to increase the level of automation
in the distribution system has been clearly recognized both on the vendor side and on
the utility side.
Various concept-level proposals have been presented in order to address the con-
flicting requirements for low life-cycle costs and the rapid uptake of new technology.
The most traditional approach has been to increase the functionality of the bay-level
protection and control IEDs (Intelligent Electronic Devices). This approach has been
sufficient, while CPU capacity has been steadily increasing and the price of new tech-
nology has remained at a reasonable level. The issue in this approach has been the
extensive costs of upgrades. New features have also required substantial changes in
the substation’s entire secondary system, requiring long maintenance breaks.
This PhD thesis investigates how station-level data processing can be utilized to
help in creating a future-proof architecture for the secondary system of a distribution
substation. The needed technology is evaluated, and an overall life-cycle cost anal-
ysis is performed showing the cost benefit of a centralized architecture. The thesis
shows that the larger the substation, the greater the benefits of a centralized archi-
tecture. It also shows how great is the impact of increased reliability. The outage
costs of a network exceed all the other life-cycle costs of the secondary system, and
illustrates how focusing on substation automation is a cost-efficient way to improve
the reliability of the network.
i
The new architecture enables the re-allocation of substation functionality, as both
bay-level and station-level data processing is available. This aspect is researched in
the thesis, and a clustering method based on fuzzy c-means clustering is proposed
for this re-allocation. When the function requires communication, but does not have
strict requirements for response times, station-level implementation can be justified.
Complex functionality requiring additional CPU performance and anticipated up-
dates in the near future are clear indications of station-level functionality.
In addition to re-organizing the existing functionality, the new architecture also
enables the utilization of new features which were not feasible previously. A new
measurement method is proposed which emphasizes this aspect by increasing the
overall sampling frequency of the substation measurements without increasing the
sampling frequency of the individual IEDs. The method is based on Time-Interleaved
technology, where the sampling of all the IEDs in the substation is synchronized.
However, this synchronization is done in such a way that each IED does not take the
measurement at exactly the same time stamp. This is achieved by time-shifting the
sampling in the IEDs by a fraction of the sample time. Merging these measurements
at the station level creates a single sample stream with a high sampling frequency.
The usefulness of the architecture and the new measurement method is tested
with a transient-based fault location method. In earlier studies, transient-based meth-
ods have not been used in bay-level IEDs because of the strict requirements for the
sampling frequency. However, using the measurement method presented in this the-
sis, transient-based algorithms can also be used without increasing the sampling fre-
quency of individual IEDs.
ii
Foreword
This doctoral thesis was written for ABB Oy Medium Voltage Products, and finalized
within the Smart Grids and Energy Markets (SGEM) research program coordinated
by CLEEN Oy, and received funding from the Finnish Funding Agency for Technol-
ogy and Innovation, Tekes. The research was supervised by Prof. Pekka Verho from
Tampere University of Technology, whom I wish to thank for his experienced guid-
ance and excellent advice along my journey. I also wish to thank the pre-examiners of
this thesis, Prof. Jero Ahola from Lappeenranta University of Technology and Prof.
Vladimir Terzija from The University of Manchester, whose valuable comments im-
proved the quality of this thesis significantly. An important person in this process
was also Mr. Adrian Benfield, who did the proofreading, thank you for correcting
my English.
I want to thank my line managers, Petri Hovila and Tomas Karlais from ABB Oy
and Jatta Jussila-Suokas from CLEEN Oy for supporting and encouraging me and
for taking a lot of extra trouble to allow me to write my thesis. Special thanks go to
Tapio Hakola, Antti Hakala-Ranta and Dick Kronman from ABB Oy and Prof. Pertti
Järventausta from Tampere University of Technology for their valuable professional
insights and inspiration - people like you, with long experience yet an enthusiastic
focus on the future have encouraged me greatly. I am also grateful to Erkka Kettunen
for sharing my interest in this topic and for his valuable contribution in the area of
software development. However, the largest debt of gratitude I owe is to my parents
and close relatives for all the support I have received during these years, without
whom, none of this would have been possible.
Writing this foreword reminds me of the master’s thesis I wrote over eight years
ago (is it really that long). I remember how easy that was in comparison to this work
- merely something to do on the side. The process with this doctoral thesis has been
much more challenging - periods of self-doubt and trouble in seeing the big picture
iii
took time to overcome - and I could not have done it without the support of my close
friends. Thank you all for that. I particularly want to thank my housemates in The
Yellow House in Pispala: Anniina, Johanna, Minta, Olli, Rene, Tanja and Tuula. The
warm atmosphere you created in our home gave me the energy to finalize my thesis. I
have recently reread the foreword from my M.Sc. thesis eight years back, and I have
noticed that the advice I wrote then, (I do not know if I wrote it then for others or for
myself), it is still valid: remember to keep the little child alive ;-)
iv
Contents
Abstract i
Foreword iii
List of Figures xi
Nomenclature xiii
1 Introduction 1
1.1 The hypothesis and objectives of the thesis . . . . . . . . . . . . . . 3
1.2 Background of the thesis . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 The role of the author . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Station Architecture 8
2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 An electricity network . . . . . . . . . . . . . . . . . . . . 8
2.1.2 The secondary system of the network . . . . . . . . . . . . 11
2.1.3 Network management processes . . . . . . . . . . . . . . . 13
2.2 New drivers and market trends for energy distribution . . . . . . . 15
2.2.1 The global situation and climate change . . . . . . . . . . . 15
2.2.2 The market situation and foreseeable business trends . . . . 16
2.3 New requirements for distribution substations . . . . . . . . . . . . 18
2.3.1 Advanced fault management . . . . . . . . . . . . . . . . . 18
2.3.2 Efficient operation of the network and support for asset man-
agement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.3 Future-proof technologies, upgradeability . . . . . . . . . . 21
v
2.3.4 Low life-cycle costs . . . . . . . . . . . . . . . . . . . . . 23
2.4 The possible architectures . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.1 Definition of a substation . . . . . . . . . . . . . . . . . . . 24
2.4.2 A proposed architecture and a comparison with other solutions 25
2.4.3 IEC 61850 standard . . . . . . . . . . . . . . . . . . . . . . 27
2.5 The cost-efficiency of the possible architectures . . . . . . . . . . . 31
2.5.1 Life-cycle costing in different scenarios . . . . . . . . . . . 32
2.5.2 Acquisition costs . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.3 Acquisition and renewal costs . . . . . . . . . . . . . . . . 34
2.5.4 Acquisition, renewal and maintenance costs . . . . . . . . . 36
2.5.5 Failure costs . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.5.6 Summary of LCC cost estimates . . . . . . . . . . . . . . . 44
2.5.7 Other benefits of the combined set-up . . . . . . . . . . . . 45
2.6 Details of the proposed architecture . . . . . . . . . . . . . . . . . 46
2.6.1 Important standards . . . . . . . . . . . . . . . . . . . . . . 46
2.6.2 Protection and control IEDs . . . . . . . . . . . . . . . . . 48
2.6.3 Station computer . . . . . . . . . . . . . . . . . . . . . . . 48
2.7 Pilot installation of the selected architecture in Noormarkku . . . . 50
2.8 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3 Station Measurements 53
3.1 Measurement replacement . . . . . . . . . . . . . . . . . . . . . . 53
3.2 Measurement merging . . . . . . . . . . . . . . . . . . . . . . . . 54
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.2.2 Time-Interleaved ADC (TI-ADC) . . . . . . . . . . . . . . 55
3.3 Increasing sampling frequency at the substation level . . . . . . . . 56
3.3.1 General concept . . . . . . . . . . . . . . . . . . . . . . . 56
3.3.2 Voltage measurements . . . . . . . . . . . . . . . . . . . . 57
3.3.3 Current measurements . . . . . . . . . . . . . . . . . . . . 57
3.4 Limitations and error analysis . . . . . . . . . . . . . . . . . . . . 58
3.4.1 TI-ADC model . . . . . . . . . . . . . . . . . . . . . . . . 60
3.4.2 Derivation of the explicit SINAD . . . . . . . . . . . . . . 62
3.4.3 Derivation of the expected SINAD . . . . . . . . . . . . . . 62
3.4.4 SINAD values for the substation . . . . . . . . . . . . . . . 64
3.4.5 Time synchronization requirements . . . . . . . . . . . . . 68
vi
3.4.6 Special considerations related to TI-ADC . . . . . . . . . . 70
3.5 Other existing solutions for increasing the sampling frequency . . . 70
3.6 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4 Station Applications 72
4.1 Functionality in the secondary system of a distribution substation . . 72
4.2 Functionality division . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.3 Functionality division criteria . . . . . . . . . . . . . . . . . . . . . 76
4.3.1 Communication requirements . . . . . . . . . . . . . . . . 76
4.3.2 Response time . . . . . . . . . . . . . . . . . . . . . . . . 77
4.3.3 Utilization frequency . . . . . . . . . . . . . . . . . . . . . 77
4.3.4 Function immaturity . . . . . . . . . . . . . . . . . . . . . 78
4.4 Functionality division method . . . . . . . . . . . . . . . . . . . . 78
4.5 Description of the results . . . . . . . . . . . . . . . . . . . . . . . 83
4.5.1 Unit-level mandatory functions . . . . . . . . . . . . . . . 83
4.5.2 Unit-level optional functions . . . . . . . . . . . . . . . . . 84
4.5.3 Station-level mandatory functions . . . . . . . . . . . . . . 84
4.5.4 Station-level optional functions . . . . . . . . . . . . . . . 85
4.6 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5 Station Application Example 88

5.1 Earth fault in the distribution network . . . . . . . . . . . . . . . . 89
5.1.1 Grounded network . . . . . . . . . . . . . . . . . . . . . . 89
5.1.2 Isolated network . . . . . . . . . . . . . . . . . . . . . . . 90
5.1.3 Compensated network . . . . . . . . . . . . . . . . . . . . 93
5.1.4 Initial transients . . . . . . . . . . . . . . . . . . . . . . . . 96
5.1.5 Measured fault resistances during earth faults . . . . . . . . 99
5.2 Earth fault location methods . . . . . . . . . . . . . . . . . . . . . 101
5.2.1 Earth fault location methods based on initial transients . . . 102
5.2.2 Other algorithms for earth fault location . . . . . . . . . . . 104
5.3 Test results for the impact of sampling frequency on transient-based
earth fault location . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.3.1 Description of the algorithm . . . . . . . . . . . . . . . . . 105
5.3.2 Analyzing the simulation results . . . . . . . . . . . . . . . 110
5.4 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
vii
6 Summary 122
6.1 Contribution of the thesis . . . . . . . . . . . . . . . . . . . . . . . 123
6.2 Evaluation of the thesis . . . . . . . . . . . . . . . . . . . . . . . . 124
6.3 Future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
References 126
viii
List of Figures
1.1 Outline of the thesis. . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1 Typical voltage levels in the electricity network in Finland. . . . . . 9

2.2 Example topology of a double busbar substation. . . . . . . . . . . 10
2.3 Electromechanical, static and numerical relays. . . . . . . . . . . . 13
2.4 Expected increase in energy consumption (Mtoe = million ton of oil
equivalent) and in the share of zero-carbon fuels [IEA, 2009]. . . . . 16
2.5 Targeted decrease in CO2 emissions [IEA, 2009]. . . . . . . . . . . 16
2.6 Present value of a 110 kV network [Jeromin et al., 2009] . . . . . . 21
2.7 Possible Architectures for Distribution Substation Automation. . . . 25
2.8 Two main levels – IED Configurator and System Configurator. Up-
dates possible via IID files[IEC, 2009]. . . . . . . . . . . . . . . . . 28
2.9 Reference model for the information flow in the configuration pro-
cess [IEC, 2009]. . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.10 Modification process [IEC, 2009]. . . . . . . . . . . . . . . . . . . 30
2.11 Acquisition costs for different scenarios. . . . . . . . . . . . . . . . 35
2.12 Acquisition and renewal costs for different scenarios. . . . . . . . . 35
2.13 Acquisition, renewal and maintenance costs for different scenarios. . 36
2.14 Total interruption costs. . . . . . . . . . . . . . . . . . . . . . . . . 40
2.15 An example reliability graph of protection . . . . . . . . . . . . . . 42
2.16 Cost saving with different scenarios. . . . . . . . . . . . . . . . . . 42
2.17 Overall set-up of a centralized protection and control system. . . . . 47
2.18 Separate functionality for centralized protection and monitoring. . . 50
3.1 Time-Interleaved ADC with M channels [Vogel, 2005]. . . . . . . . 55

3.2 Timing diagram of time-interleaved ADC with M channels [Vogel,
2005]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
ix
3.3 Voltage measurements combined from 5 different measurements. . . 58
3.4 Current measurements combined from 5 different measurements. . . 59
3.5 Model of a one-channel ADC [Vogel, 2005]. . . . . . . . . . . . . . 61
3.6 The effect of gain error in SINAD. . . . . . . . . . . . . . . . . . . 65
3.7 The effect of offset error in SINAD. . . . . . . . . . . . . . . . . . 66
3.8 The effect of timing error in SINAD. . . . . . . . . . . . . . . . . . 66
3.9 The combined effect of gain and timing mismatches. . . . . . . . . 67
3.10 Contours of Figure 3.9. . . . . . . . . . . . . . . . . . . . . . . . . 67
3.11 Contours of Figure 3.9 in THD+N. . . . . . . . . . . . . . . . . . . 68
3.12 The effect on SINAD of increasing the number of ADCs. . . . . . . 69
4.1 Function re-allocation results. . . . . . . . . . . . . . . . . . . . . 81

4.2 Membership of different functions to different clusters. . . . . . . . 82
5.1 An earth fault in a grounded network. . . . . . . . . . . . . . . . . 89

5.2 An earth fault in an isolated network. . . . . . . . . . . . . . . . . . 90
5.3 Voltage vectors during an earth fault. . . . . . . . . . . . . . . . . . 91
5.4 An equivalent circuit of an isolated network during an earth fault. . . 92
5.5 An earth fault in a compensated network. . . . . . . . . . . . . . . 93
5.6 An earth fault in a compensated network (vector representation). . . 94
5.7 An equivalent circuit of a compensated network during an earth fault. 95
5.8 A simulated example, an earth fault in phase 1, transient components
visible in phase voltages. . . . . . . . . . . . . . . . . . . . . . . . 97
5.9 Network model and equivalent circuit used for modeling the charge
transient, fault resistance 0 Ω. . . . . . . . . . . . . . . . . . . . . . 98
5.10 Neutral current in all feeders during an earth fault in one feeder, 6
feeders in the simulated example. . . . . . . . . . . . . . . . . . . . 99
5.11 The division of the fault resistances in a compensated network [Hän-
ninen and Lehtonen, 1998]. . . . . . . . . . . . . . . . . . . . . . . 100
5.12 The division of the fault resistances in a isolated network [Hänninen
and Lehtonen, 1998]. . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.13 Flow chart of the differential equation algorithm and corresponding
components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.14 Simulation model used for testing. . . . . . . . . . . . . . . . . . . 111
5.15 Results with sampling frequency of 16 kHz, Rf = 10 Ω. . . . . . . . 112
x
5.16 Results with seven different sample streams, combining to 16 kHz
when processed as in Chapter 3, Rf = 10 Ω. . . . . . . . . . . . . . 112
5.17 Results a sample frequency of 16 kHz / 7 = 2.29 kHz, Rf = 10 Ω. . . 113
5.18 Transient frequency and amplitude with Rf = 10 Ω. . . . . . . . . . 114
5.19 Results with sampling frequency of 16 kHz, Rf = 80 Ω. . . . . . . . 115
5.20 Results with seven different sample streams, combining to 16 kHz
when processed as in Chapter 3, Rf = 80 Ω. . . . . . . . . . . . . . 115
5.21 Results a sample frequency of 16 kHz / 7 = 2.29 kHz, Rf = 80 Ω. . . 116
5.22 Transient frequency and amplitude with Rf = 80 Ω. . . . . . . . . . 117
5.23 Mean errors with different sampling set-ups and fault distances, fs =
20 kHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
16 kHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
10 kHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
xi
Nomenclature
ADC Analogue-to-Digital Converter
AMR Automatic Meter Reading
CB Circuit Breaker
CBM Condition Based Maintenance
CID Configured IED Description
CIS Customer Information System
DFT Discrete Fourier Transform
DG Distributed Generation
DMS Distribution Management System
ENS Energy Not Supplied
FLIR Fault Location, Isolation and power Restoration
GHG Greenhouse Gas
GOOSE Generic Object Oriented Substation Event
GPS Global Positioning System
HE High-End
HV High Voltage
HW Hardware
xii
ICT Information and Communication Technology
IDA Intelligent Distribution Automation research project
IED Intelligent Electronic Device
IID Instantiated IED Description
LCC Life Cycle Costs
LE Low-End
LN Logical Node
LV Low Voltage
MV Medium Voltage
NCC Network Control Center
NIS Network Information System
RCM Reliability Centered Maintenance
RTU Remote Terminal Unit
SAV Sampled Analogue Value
SCADA Supervisory Control And Data Acquisition
SCD Substation Configuration Description
SCL Substation Configuration Language
SED System Exchange Description
SGEM Smart Grids and Energy Markets research program
SINAD SIgnal-to-Noise-And-Distortion ratio
SNTP Simple Network Time Protocol
SW Software
THD+N Total Harmonic Distortion with Noise
TI-ADC Time-Interleaved ADC
xiii
List of Symbols
A Amplitude of a signal
C0 Earth capacitance of power lines
Ceq Equivalent capacitance
Cpp Phase-to-phase capacitance of power lines
f Fundamental frequency of the network
fs Sampling rate
gl Deterministic gain of channel l in a TI-ADC set-up
g˜l Random variable for the gain of channel l in a TI-ADC set-up
ik Current, instantaneous value at time k
ˆ
iCh Amplitude of the charge current transient
Ic Current through earth capacitances (phasor)
If Fault current phasor
Ie Fault current phasor
IL Current through the Petersen coil (phasor)
I0 Neutral current phasor
CLC Life cycle costs: Overall costs
CA Life cycle costs: Acquisition costs
CR Life cycle costs: Renewal costs
CO Life cycle costs: Operation costs
CM Life cycle costs: Maintenance costs
CF Life cycle costs: Failure costs
CCR Life cycle costs: Replacement costs
CP Life cycle costs: Penalty costs
L Inductance of a Petersen coil
Leq Equivalent inductance
LT Phase inductance of a substation transformer
M Number of ADCs in a TI-ADC set-up
ol Deterministic offset of channel l in a TI-ADC set-up
õl Random variable for the offset of channel l in a TI-ADC set-up
PSgr Signal power, dependent on gain and timing deviation
PNgr Signal error power, dependent on gain and timing deviation
PNo Signal error power, dependent on offset
PSgrR Expected signal power, dependent on gain and timing deviation
xiv
PNgrR Expected error power, dependent on gain and timing deviation
PNoR Expected error power, dependent on offset
rl Deterministic relative timing deviation of channel l in a TI-ADC
set-up
r˜l Random variable for the relative timing deviation of channel l in
a TI-ADC set-up
Rf Fault resistance during an earth fault
RLE Leakage resistance of a network
Rp Resistance in parallel with a Petersen coil
∆tl Deterministic absolute timing deviation of channel l in a TI-ADC
set-up
Ts Sampling period
uk Voltage, instantaneous value at time k
U0 Neutral voltage phasor
U1 Voltage phasor in phase 1
U0F Neutral voltage phasor during an earth fault
U1F Voltage phasor in phase 1 during an earth fault
µg Expected value of the gain of one ADC/IED in a TI-ADC set-up
µo Expected value of the offset of one ADC/IED in a TI-ADC set-up
σg Standard deviation from the expected gain in a TI-ADC set-up
σr Standard relative timing deviation of a TI-ADC set-up
σo Standard deviation of the offset in a TI-ADC set-up
Ω Continuous-time angular frequency
ωN Angular frequency of the network
ωC Angular frequency of the charge transient
ω0 Angular frequency an input signal, discrete time
Ω0 Angular frequency an input signal, continuous time
xv
xvi
Chapter 1
Introduction
The importance of a continuous supply of electrical energy has steadily increased

over the past few decades. There have already been numerous instances in which
electricity has been recognized as a critical resource for modern society and nothing
indicates any change in this trend [EC, 2008b]. On the contrary, it has been estimated
that the consumption of electrical energy will grow twice as much as overall energy
consumption [IEA, 2009]. This means that in the future an even bigger proportion
of our energy chain will be based on the production, delivery and consumption of
electricity.
At the same time, increasing requirements are being placed on the energy sec-
tor. One major driver for this is the fight against global warming and the need to
cut down on our CO2 emissions. This calls for new, sustainable, renewable and
environmentally-friendly energy sources, such as wind and solar power. Integrating
these energy sources into the present energy network is challenging, as these produc-
tion units are often smaller and more widely distributed than traditional large power
plants. This creates bi-directional power flow in distribution networks, which re-
quires new protection and control schemes. To emphasize this, networks with a large
share of DG (Distributed Generation) are often called ’active networks’ to clearly
distinguish them from the more traditional ’passive networks’.
In addition to new energy resources, it is equally important to increase the en-
ergy efficiency of the network, as a significant proportion of our energy production
is simply lost in various parts of the energy chain. Distributed and renewable energy
resources often result in improvements in this regard, too. If the distributed energy
production unit is located close to the consumer, there is less need for energy trans-
1
CHAPTER 1. Introduction
mission and therefore lower transmission losses during normal operation. However,
in general there is no ’single factor’ for improving the efficiency of the electricity
supply, as the whole chain needs updating - both the ’primary system’ that delivers
the energy and also the ’secondary system’, which monitors, controls and protects the
primary system. An important part of this secondary system is the automation equip-
ment located in the distribution substations, and this is the main focus of this thesis.
The secondary system in a distribution substation is often referred to as distribution
substation automation, which is reflected in the title of this thesis.
Another phenomenon affecting the energy sector is globalization. The drive to
optimize the process of supplying electrical power is not only due to environmental
factors, but also to financial ones. Competition between electrical supply companies
is increasing at all levels. Although the competition between technology providers
has long been self-evident, recent changes in government policies mean that the same
sense of competition and response to financial imperatives has now spread to the en-
ergy producers, too. In Finland, the energy market was opened to competition as long
as 1995 [EMV, 1995] and the European Union’s target of a single european energy
market [EC, 2008b], has further stimulated competition between energy producers.
Despite the severity of the challenges facing the electricity industry, due to the
speed of recent technological innovations the number of available solutions has also
increased. The exponential growth of ICT (Information and Communication Tech-
nology) has made many new solutions available, and has also provided the infrastruc-
ture for sharing knowledge about these innovations globally, as soon as new discov-
eries are made.
So, the new requirements and the corresponding new technical solutions men-
tioned above have recently spawned a multitude of Smart Grid initiatives around the
world. In addition to the predicted overall transformation of the electricity network
as a whole, these initiatives also indicate how much the control and protection of
the network is expected to change within the next few years. As the passive net-
work with unidirectional power flow evolves into an active network with a variety
of distributed generation units, the requirements for distribution substations will also
change, forcing the utilities to take action. Utilities do not want to undertake con-
tinuous and costly upgrades of the whole protection system, but there is still a clear
need to adapt to new requirements. The need to increase the level of automation in
the distribution network has been clearly recognized both on the vendor side [Heckel,
2009] and on the utility side [Gorgette et al., 2007]. For distribution substations, this
2
1.1. The hypothesis and objectives of the thesis
means increasing the functionality of existing IEDs (Intelligent Electronic Devices)

in the substation, or adding entirely new IEDs to the system.
Various concept-level proposals have been presented in order to address the con-
flicting requirements for low life-cycle costs and the rapid utilization of new technol-
ogy, and these are briefly introduced in this chapter and again, in more detail, later
on in the thesis. The most traditional approach has been to increase the functionality
of the bay-level protection and control IEDs. This approach has been sufficient while
CPU capacity has been steadily increasing and the price of new technology has re-
mained at a reasonable level. The issue in this approach has been the extensive costs
of upgrades. The increased functionality has also required substantial changes to the
substations’ entire secondary systems, requiring long maintenance breaks or detailed
and time-consuming planning for back-up connections.
This doctoral thesis investigates how station-level data processing could be uti-
lized to help create a future-proof architecture for the secondary system of a distri-
bution substation. The main requirements and drivers are described and evaluated,
possible technologies are presented, and the proposed architecture is described in
more detail with reference to the relevant technologies and standards. Life-cycle cost
analyses are also performed on the possible architectures.
Furthermore, the thesis investigates new possibilities for the proposed architec-
ture. The feasibility of new innovations in the measurement chain in the substation
is evaluated and a new method for measurement is proposed. Also applications and
functions which are currently being implemented in bay-level protection and control
IEDs are evaluated and a new scheme of substation functionality allocation is pro-
posed. Finally the overall concept is tested and prototyped, both with simulations
and also in practice.
1.1 The hypothesis and objectives of the thesis

The over-riding hypothesis driving the research is that a more centralized architecture
for the secondary system of a distribution substation will provide the most future-
proof platform for dealing with future requirements, such as active distribution net-
work management or automatic fault management. This hypothesis was approached
using both higher concept-level objectives and lower-level, more detailed objectives.
The main concept-level objective was to identify when and in which cases this
proposed new architecture is more cost-efficient than the currently dominant one.
3
1.2. Background of the thesis
Another important target was to specify how this centralized architecture should be
utilized, i.e. what functionality in a distribution substation benefits from, or even
requires, a more centralized architecture.
In addition to the concept-level objectives, another important objective of the the-
sis was to discover entirely new and more accurate measurement methods or innova-
tive functions, which could not be implemented without this centralized architecture.
The aim was to pinpoint practical examples of new innovative functions for substa-
tions, in order to highlight that this new architecture is not just the "same old system
with lower life-cycle costs" but also an architecture which enables something entirely
new.
1.2 Background of the thesis

The possibilities for centralized functionality at the substation level have been re-
searched at ABB since 2007. Their activities started with the Intelligent Distribution
Automation (IDA) project, conducted in cooperation with the Finnish utility com-
pany, Fortum, and Tampere University of Technology. The IDA project ended with
the pilot installation carried out at the Noormarkku substation in the Fortum distri-
bution network. The author of this thesis joined the IDA project in 2008 and was
the project manager when it ended in 2009, and the main results of the project were
outlined in [Valtari et al., 2009a].
Research into a centralized architecture for electricity distribution substations con-
tinued after the IDA project in the Smart Grids and Energy Markets (SGEM) research
program. This was coordinated by CLEEN Oy and received funding from the Finnish
Funding Agency for Technology and Innovation, Tekes. The research program cov-
ers many aspects of the future Smart Grid, and is planned for the years 2009-2014
with an overall budget of e57 M. The author is currently leading the research activ-
ities within the Work Package "New Substation Solutions" under the research theme
"Future Infrastructure of Power Systems". The author has also been acting as the
overall program manager of SGEM since 2011.
A transient-based earth fault location algorithm was the subject of the author’s
master’s thesis [Valtari, 2004]. The aim of that thesis was to implement and test
an algorithm suitable for a bay-level protection and control IED, but this was not
achieved. The requirements for the algorithm were too stringent, especially for the
sampling frequency of the measurements, and the algorithm was never finalized.
4
1.3. Outline of the thesis
While writing this doctoral thesis, the author has also contributed to other projects
supporting the work presented here, such as the supervision of a master’s thesis which
focused on the end-user engineering process of the environment [Kettunen, 2011].
The author also had a consulting role in a pilot project implementing substation-level
fault detection, isolation and restoration functionality [Manner et al., 2011]. Research
related to the centralized architecture of substations will continue in the future under
the SGEM research program.
1.3 Outline of the thesis
New requirements for the automation of electricity distribution substations and the
proposed new automation architecture are presented and evaluated in Chapter 2. This
chapter presents the main features of the technology and also performs a life-cycle
cost analysis. The measurement chain, and a method for increasing the sampling
frequency of substation measurements, are described in Chapter 3. The possibility
of using measurements from several bays simultaneously raises new possibilities for
the whole measurement chain, and these are described and tested. The new archi-
tecture allows for the re-allocation of functionality between the station level and the
unit level, and this is evaluated in Chapter 4. An example case utilizing the new mea-
surement method from Chapter 3 and a transient-based earth fault location algorithm
is presented in Chapter 5. A summary of the thesis is presented in the concluding
Chapter 6.
This thesis is based on five different publications, which together constitute the
main results of the thesis. Chapter 2 is based on results from [Valtari et al., 2009a]
(republished in [Valtari et al., 2009b]), [Valtari and Verho, 2011a] and [Valtari and
Verho, 2011b]. Chapter 3 includes the results from [Valtari and Verho, 2012] while
Chapter 4 is derived from [Valtari et al., 2010]. In addition, a patent application has
been submitted based on the results of Chapter 3 [Valtari, 2012]. The patent has been
allowed and is about to be granted. The function example presented in Chapter 5
has not been published earlier. The basic principles of the algorithm are presented in
[Valtari, 2004], but the method is re-tested with a new set-up based on a substation-
level implementation.
An outline of the chapters and their relations to publications is presented in Figure
1.1. The logical story line of the thesis is also presented on the left side of the figure.
5
1.4. The role of the author
Chapter Area [Publication]
[Valtari et al., 2009a]

Concept [Valtari et al., 2009b]
Technology [Valtari and Verho, 2011b]

Overall concept and Ch2 Architecture
platform deﬁnition
Evaluation [Valtari and Verho, 2011b]
Engineering
aspects [Valtari and Verho, 2011a]
Thesis
Measurement chain Increasing sampling [Valtari and Verho, 2012]
utilizing the proposed Ch3 Measurements frequency [Valtari, 2012]
platform
Application and function re- Functionality

allocation utilizing the Ch4 Applications re- allocation [Valtari et al., 2010]
proposed platform
An example utilizing all three aspects - proposed Earth Fault

platform from Ch2, measurement methods from Ch5 Example location algorithm
Ch3 and functions from Ch4
Figure 1.1: Outline of the thesis.
1.4 The role of the author
The initial concept for centralized protection and control presented in Chapter 2 was
developed in the early stages of the IDA project by a number of experts: Antti Hakala-
Ranta, Dick Kronman and Tapio Hakola from ABB Finland; Thomas Werner and
Bernhard Deck from ABB Switzerland; and Pekka Vierimaa from Fortum Distribu-
tion Finland. The author built on their work by summarizing the ideas and leading the
pilot installation in the Fortum-owned substation. After the pilot installation in 2009,
the author also took on the leading role in the overall research project. The author
performed the life-cycle cost calculations and extended this evaluation to include the
engineering processes.
The measurement principle presented in Chapter 3 was created and developed by
the author. The example application utilizing the measurement principle presented
in Chapter 5 was originally developed by Seppo Hänninen from VTT, but the author
made the necessary extensions to the algorithm so that this measurement principle
could be used. The simulation model used in section 5.3.2 was developed by Mo-
hammed Abdel-Fattah from Aalto University.
6
1.4. The role of the author
The initial example listing of appropriate station-level functions presented in Chap-

ter 4 was created by Tapio Hakola from ABB Finland at the beginning of the IDA
project. The author continued this work and developed the method used for this clas-
sification in order to achieve a more comprehensive list for guiding further research
activities.
7
Chapter 2
Station Architecture
After the background, presented in section 2.1, section 2.2 investigates the new
drivers and market trends affecting the secondary system of a distribution network.
From these drivers and trends, more detailed requirements for the distribution sub-
stations are derived in section 2.3. After that, possible architectures and on-going
research into the secondary system are presented in section 2.4 and evaluated in sec-
tion 2.5. Finally, the apparently most suitable architecture is considered in more
detail in section 2.6, which also outlines the required technologies and the relevant
standards.
2.1 Background
2.1.1 An electricity network
Power generation and consumption often occur at different geographical locations.

Although the trend nowadays is towards distributed power production, major power
plants are normally situated far away from densely populated areas or large factories.
This is not only because the prerequisites for energy generation might be better there,
but also because of safety and environmental issues. As a result of this, energy gen-
eration and consumption have to be connected together with electricity transmission
and distribution networks [Grigsby, 2000].
For economic and safety reasons, the form of electrical energy is different in dif-
ferent parts of the network. The different parts of a network and their typical voltage
levels in Finland are presented in Figure 2.1. The energy production end of the chain
8
2.1. Background
is marked with G (Generator) and the consumption end with L (Load).
Load
Figure 2.1: Typical voltage levels in the electricity network in Finland.
Energy losses are lower if the voltage is relatively high and the current relatively
low. Therefore, the delivery of large quantities of electrical energy can reasonably be
handled by HV (High Voltage) networks (400 kV ... 110 kV). This part of a network
is called the transmission network. [Elovaara and Laiho, 2004]
The voltage level of the transmission network is too high for electrical devices.
The required isolation with high voltages would make such devices too large and
expensive. Therefore, the voltage has to be lowered before it can be delivered to
customers. Delivery to factories and community centers is normally done with MV
(Medium Voltage) networks (20 kV ... 6 kV), and this part of the network is often
referred to as the distribution network [Lakervi and Holmes, 1996].
Because manufacturing costs and safety risks are decreased by lowering the volt-
age, the energy to households is supplied by the LV (Low Voltage) network (in Fin-
land 400 V phase-to-phase, 230 V phase-to-ground) [Elovaara and Laiho, 2004].
These voltage levels are sufficient as long as the required power is relatively low, so
that the distribution losses due to the load current do not rise too much.
The different parts of the electricity network require different protection and su-
pervision systems, and the transformation of electrical energy from one voltage level
to another also calls for specific devices. All this makes the whole energy distribution
system rather complicated and its control is a challenging task.
Distribution substation
The connection point between the distribution and the (sub)transmission network is
called a distribution substation. The term ’primary substation’ is also often used in
the literature [Lakervi and Holmes, 1996], but distribution substation is somewhat
more common and is also used in a number of standards [IEEE, 2000]. Therefore,
9
2.1. Background
’distribution substation’ is the term that will be used in this thesis. There are other
types of substations, such as secondary substations (the connection point between
distribution and low voltage networks), transmission substations (between two trans-
mission lines) or collector substations (collecting several distribution lines, e.g. near
wind farms).
A distribution substation has the equipment needed for changing the voltage level
(power transformers) and also for controlling the topology of the network via CBs
(Circuit Breakers), disconnectors and earthing switches. In addition, for the sake
of power quality, specific capacitors (if more reactive power generation is needed)
or voltage regulators (for keeping the voltage at the required level regardless of the
consumption) may be included. This ’primary system’ of the distribution substation,
i.e. the system which has a direct effect on the transmitted electrical energy, may also
contain other components such as generator units or earthing coils (for compensated
networks, see Chapter 5) [Lakervi and Holmes, 1996][Lakervi and Partanen, 2008].
Transmission (incoming) feeder lines are connected to distribution (outgoing)
feeder lines via a busbar (or several busbars if the substation is large). An exam-
ple topology for a substation is shown in Figure 2.2, where there are two incoming
feeders (and two power transformers) and two outgoing feeders [ABB Ltd., 2000].
The set-up has a double busbar, but only one circuit breaker per feeder (although
there are three disconnectors: two between the CB and the busbars and one after the
CB).
Figure 2.2: Example topology of a double busbar substation.
A separate automation layer is needed in the substation in order to control the

equipment in this ’primary system’, and this is often referred to as the ’secondary
system’. An overview of the secondary system is described in the following section
10
2.1. Background
2.1.2, but the new requirements and future challenges are presented in more detail in
section 2.3
2.1.2 The secondary system of the network
The primary system of the substation has generally remained quite stable since the
beginning of electrification. New materials and new production technologies have, of
course, changed the look and feel of these primary components, but the fundamental
structures have remained surprisingly unchanged. The life-span of such equipment is
typically between 30 and 50 years [Laine, 2005].
The secondary system, on the other hand, has gone through many revolutionary
changes, and this process is expected to accelerate in the future, as described later on
in section 2.3. This section describes the historical background and current solutions
before looking into new requirements and challenges.
At the very beginning of electrification, the secondary system was virtually non-
existent. There were protection relays, but all the control operations were handled
by personnel working in the substation. With the development of telecommunication
technology, this practice has changed and nowadays all the operations are conducted
from a separate NCC (Network Control Center) .
The number of different software systems in NCCs has increased steadily and the
integration of these systems is an on-going process. Normal control operations are
handled via a SCADA (Supervisory Control And Data Acquisition) system, which
in addition to real-time operation possibilities also shows the real-time status of the
network topology and performance [Grigsby, 2000].
A system called DMS (Distribution Management System) has been developed to
support network operations, and this provides a geographical overview of the network
and improved topology management [Grigsby, 2000]. The calculation engine of the
DMS performs various calculations related to the network status, such as evaluating
the voltage profiles of feeders or locating a fault based on fault currents and the net-
work topology. However, a DMS needs detailed data from the network components,
which can be obtained from the NIS (Network Information System). Practically all
the utilities in Finland also have a separate CIS (Customer Information System) han-
dling the customer data, often integrated with another system which handles customer
calls. [Lakervi and Partanen, 2008]
The installation of AMR (Automatic Meter Reading) meters in customer house-
11
2.1. Background
holds has facilitated a new, extensive source of network status information for the
utilities, and the integration of this meter-reading system into the other NCC soft-
ware systems is currently ongoing in Finland.
The secondary system of distribution substations
Nowadays, communication with a substation is routed through a gateway device

called an RTU (Remote Terminal Unit) [Grigsby, 2000]. The term ’Gateway’ is also
often used for this device, and RTU is in fact a product name of ABB for a particular
gateway device. Nevertheless, the general term selected for this thesis is RTU, as that
is often used in the literature. All the protection and control IEDs can be accessed via
RTUs and all the CBs and disconnectors can be operated via them. Protection and
control IEDs are also essential for monitoring the status of the network - both normal
state values (measurements) and fault state information (alarms and events).
In addition to RTUs and protection and control IEDs, the secondary system also
needs other elements, such as measurement transformers or sensors (voltage and cur-
rent), battery systems for guaranteeing operation during interruptions in the electric-
ity distribution, telecommunication modems and sometimes, nowadays, even video
surveillance cameras. [Lakervi and Partanen, 2008]. However, protection and con-
trol IEDs are the core devices in contemporary secondary systems, and these are
described in more detail below.
Protection and control IEDs
The earliest protection devices were electromechanical relays, which were utilized
at the beginning of the 20th century. Separate devices were needed for every phase
and every function, the functions being limited to only simple overcurrent and earth
fault protection. These devices were very inaccurate and unreliable in operation. On
the other hand, they did not require any external power sources, but could utilize the
power from the power lines [Mörsky, 1993] [Lundqvist, 2010].
Static relays became popular during the 1960s, due to the emergence of transis-
tors and electronics. The number of functions increased, and one device was able
to handle all three phases (although separate devices were still needed for separate
functions). These devices needed their own power supply, but their operation was
more reliable and accurate. Even today, almost half of the protection devices in use
are static relays [Mörsky, 1993] [Lundqvist, 2010].
12
2.1. Background
The development of integrated circuits and microprocessors (and computers) caused

the next revolution in the secondary system at the beginning of the 1980s. The new
protection devices which utilized this technology were called numerical relays, and
these could handle all the protection and control functions for one feeder within one
single device. Self-supervision functionality and support for several communication
protocols were introduced at this point, too. Illustrations of these three types of de-
vices are shown in Figure 2.3
Figure 2.3: Electromechanical, static and numerical relays.
The process of integrating more functionality into single devices (and improving
calculation performance) has led to increasingly advanced numerical relays. Because
of this increased functionality, the name of the device has changed, and nowadays
technology vendors prefer to use the term protection and control terminal or IED,
rather than merely a relay. IED is therefore the term used in this thesis.
2.1.3 Network management processes

The previous sections presented the overall physical structure of electricity networks
and the main components of a secondary system. This section gives a short intro-
duction to the processes associated with managing the network. The main focus is
on those processes which affect the secondary system of a distribution substation -
either in terms of the system architecture or simply those processes which require the
acquisition of data from the substation.
Network planning means long term strategic planning related to the network topol-
ogy or to the components used in the network [Grigsby, 2000]. The time frames
13
2.1. Background
involved can span several decades, as the life-span of these components is typically
between 30 and 50 years [Laine, 2005]. The network should be planned so that it
provides electrical energy to all the customers with a cost-efficient architecture and
topology, and with a protection system that fulfills all the safety and power quality
regulations [Lakervi and Partanen, 2008]. The plan should also take into account
predicted future changes which will affect the network, e.g. changes in legislation,
possible technological advances in the near future, increases or decreases in the lo-
cal population, and also industrial and other intensive energy consumption scenarios.
The challenging task in the network planning process is to select an architecture for
the distribution system that meets both current and future demands. In addition, the
measurement data acquired via the secondary systems at the substations provide valu-
able information for this process.
Network maintenance encompasses all the actions that are performed to main-
tain the network, which in practice means servicing, repairing or replacing differ-
ent network components. These operations should be optimally timed, so that the
network performance stays at its target level at all times, while incurring the lowest
possible maintenance costs [Lakervi and Partanen, 2008]. Traditionally, maintenance
operations have been event-based (reacting to a broken component, corrective main-
tenance) or time-based (maintenance of a component at a specific maintenance in-
terval, periodic maintenance). However, recently there has been increased focus on
CBM (Condition Based Maintenance) and RCM (Reliability Centered Maintenance)
[Angel, 2003]. The idea behind CBM is that the actual condition of all the com-
ponents is monitored constantly, and maintenance operations are triggered when the
condition reference value or operation counter exceeds a defined limit. RCM extends
the approach of CBM by also taking into account the importance of that particular
component to the network, so that maintenance actions and type are defined based on
the component’s criticality. This means that the components are used for the optimal
length of time, but critical components are still maintained before they break down.
If a CBM or RCM approach is desired, it is essential to have accurate measurement
data available via the secondary system of the distribution substation.
The process of network operation encompasses all the activities related to day-to-
day work at the utilities. This involves monitoring and control operations when the
network is in a healthy state in order to guarantee a safe power balance (e.g. changes
in the network topology due to changes in energy production or consumption), or
preparing for other network operations (e.g. back-up connections required for future
14
2.2. New drivers and market trends for energy distribution
maintenance operations on the network). During fault situations, the operation pro-
cess should provide fast fault location and isolation. The energy should be restored
to the healthy part of the network, so that the area affected by the fault is kept at a
minimum. For repairing the fault, the process includes controlling the repair person-
nel and providing them with the necessary information about the fault. Nowadays,
an increasingly important part of this process is also communication with the cus-
tomers. On certain occasions, such as those which trigger major disturbances, many
other aspects may become important too, e.g. collaboration with rescue personnel,
ambulances and other medical assistance, the media, the police, etc. The process of
network operation is immense and it is not possible to cover all possible scenarios in
this brief section. Whatever the case, it is the secondary system of the distribution
substation that is an important part of the process.
2.2 New drivers and market trends for energy distribution
2.2.1 The global situation and climate change
Climate change is one of the great challenges of the 21st century [IPCC, 2011]. Ac-
cording to IPCC, CO2 emissions associated with the provision of energy services are
a major cause of climate change: “Most of the observed increase in global average
temperature since the mid-20th century is very likely due to the observed increase in
anthropogenic GHG (greenhouse gas) concentrations.” [IPCC, 2011].
An obvious consequence of this is that CO2 emissions need to be reduced all
around the world. Global conferences have been arranged, and the EU has already
declared its targets in its 20-20-20 program. By 2020, the EU is committed to reduc-
ing CO2 emissions by 20%, to increasing the utilization of distributed generation to
20% and to improving energy efficiency by 20% [EC, 2008a].
At the same time, energy consumption is expected to increase, as shown in Figure
2.4 [IEA, 2009]. The expected increase in the use of zero-carbon fuels will also
increase the proportion of electrical energy used in total energy consumption. The
over-riding aim is to decrease overall CO2 emissions, as shown in Figure 2.5. This
creates a very challenging scenario. How can emissions be reduced when, at the same
time, energy production will increase?
Addressing these challenges calls for many different measures, such as moving
from petroleum-based transportation to electrical transportation, increasing the share
15
Mtoe 4 500 36% Coal

4 000 32% Oil
3 500 28% Gas
3 000 24% Nuclear
2 500 20% Hydro
2 000 16% Biomass
1 500 12% Other renewables
1 000 8% Share of zero-carbon
fuels (right axis)
500 4%
0 0%
1990 2000 2010 2020 2030
Figure 2.4: Expected increase in energy consumption (Mtoe = million ton of oil
equivalent) and in the share of zero-carbon fuels [IEA, 2009].
42 Abatement
Gt
40 (Mt CO2)
Reference Scenario 2020 2030
38
Efficiency 2 517 7 880
36
End-use 2 284 7 145
34 Power plants 233 735
32 Renewables 680 2 741
Biofuels 57 429
30
Nuclear 493 1 380
28 CCS 102 1 410
450 Scenario
26
2007 2010 2015 2020 2025 2030
Figure 2.5: Targeted decrease in CO2 emissions [IEA, 2009].
of renewable and uncontrollable energy production, increasing energy efficiency with

controllable consumption and energy storages etc. [IPCC, 2011]. A major burden of
this challenge also falls on the electricity networks, which need to be more flexible in
order to implement the new actions described above. This challenge has triggered a
wealth of smart grid initiatives around the world, and is also one of the drivers behind
this thesis.
2.2.2 The market situation and foreseeable business trends
Electricity distribution has traditionally been a business which focuses on long time-
frames and has a conservative approach to technological advances. The primary
16
equipment in the substations has a typical life-span of 30 to 50 years, and upgrades to

the secondary equipment have often had to follow almost the same life cycle. How-
ever, this industry is now undergoing major changes. Many of these changes come
from government legislation, such as the new energy market act which has come into
force in Finland [EMV, 1995] and similar legislation which is going on elsewhere in
Europe. Energy production and distribution have been separated, and energy produc-
tion has been opened up to the free market, allowing customers freedom of choice
with regard to their energy supplier.
According to the Finnish energy market act, energy distribution in Finland is still
a local monopoly, but the business is now closely regulated [EMV, 1995]. The au-
thorities define certain criteria for power quality and for continuity of supply. The
legislation defines the permitted ’fair and reasonable’ profit levels, and the utility can
only affect this by improving the quality of the supply. According to current regu-
lations, customers must be recompensed for interruptions in their electricity supply.
Variations in power quality also need to be monitored. In the regulatory model, the
emphasis is on the quality of the distributed energy. So, higher quality will generate
higher profits for the network operators.
The control of a network is also moving further away from the actual, physical
network. Mergers between many companies have created bigger players in the dis-
tribution business. In addition, communication network technology has developed
rapidly over the past few years, enabling wider communication coverage for network
components - for example, the the penetration rate of mobile subscriptions has in-
creased from 20% to 128% (as a proportion of the population) in Europe within the
past 14 years [GSMA, 2012]. This has all resulted in an increased demand for the
acquisition of data from the larger networks, and also that the data should be pre-
processed before it is viewed by the NCC personnel.
In addition to increasing the amount of automation, utilities are streamlining their

processes in terms of personnel so that they can focus more on their core business.
When a function outside the utility company’s core competence cannot be fully au-
tomated, it is often outsourced. Such outsourcing is currently prevalent in Finland,
where many large utilities have recently outsourced their maintenance and service
activities [Fortum, 2010], their communication network operations [ViolaSystems,
2011] and even some network planning tasks, in order to keep the number of person-
nel in check, despite increases in the size of the network.
17
2.3. New requirements for distribution substations
2.3 New requirements for distribution substations

This section describes the new requirements for distribution substations in more de-
tail.
2.3.1 Advanced fault management

More accurate and selective network protection
As society becomes increasingly dependent on electricity, the requirements for its un-
interrupted distribution have become more stringent. This demand for uninterrupted
distribution is also reflected in the legislation of many other countries. Interruptions
in electricity distribution need to be recorded for statistical analysis and customers
must be compensated for interruptions exceeding a certain duration, which in Fin-
land is currently 12 hours [EMV, 1995].
This must be taken into account in the protection system, so that different types of
faults can be accurately detected. Entirely new protection schemes which utilize more
measurements than are locally available have been, and are being, researched and
proposed. There are many levels to increasing the communication between adjacent
protecting nodes. First, there is the horizontal communication within a substation
[Apostolov and Vandiver, 2011], and after that there is real-time communication with
a remote DG unit, or even a full-scale agent-based protection system [Kauhaniemi
et al., 2011]. The results of all these studies indicate the need for protection schemes
to take into account a larger area than only one feeder, as is currently the norm.
This also requires enhancements to the protection functions, and new, more accu-
rate protection functions are needed. In addition to protection schemes, many new
protection functions have also been researched, and these utilize or require more
measurements than are available from one feeder bay. This includes managing new
fault types, such as high-impedance earth faults [Tengdin, 1996][Abdel-Fattah and
Lehtonen, 2009] [Nikander, 2002].
Fault location and advanced control of distribution networks
Post-fault power restoration and self-healing networks are a common topic in smart
grid scenarios [Mekic et al., 2009][Rasmunssen, 2009][Manner et al., 2011]. When
a fault appears in the distribution network, it should be automatically located and
isolated, and the electricity distribution should automatically be restored to all the
18
healthy parts of the network. Sometimes, distribution networks also need to be oper-
ated in an island mode and controlled from the substation [Oudalov et al., 2011].
In Nordic countries these operations are traditionally conducted from a centralized
NCC. However, this places an additional burden on the NCC, which may become a
bottleneck for the process, especially during major disturbances. For this reason,
much of the ongoing research in this area is targeted at handling these tasks in the
substation, so that NCC personnel only need to react when the local automation at
the substation has failed in FLIR (Fault Location, Isolation and power Restoration) .
2.3.2 Efficient operation of the network and support for asset manage-
ment
Automatic adaptation to changes in topology, production and consumption
In addition to fault situations, the network also needs to respond automatically to

changes which may occur during normal operation. The number of active resources
(DG, energy storages) in the network can vary greatly, and this may require load-
shedding functionality at the station level [Apostolov et al., 2007]. Dynamic load-
response becomes more critical as the proportion of uncontrollable production in-
creases in the network. Therefore, demand response has recently gained a lot of in-
terest [EC, 2008b] - the load must be controllable if the production is not, especially
if no large energy storage facilities are available.
In addition to control operations, changes in topology and active resources also
require adaptations to the protection and monitoring functions. The parametrization
of the functions needs to be fine-tuned in order to adapt to the network status. Both
of these issues require information to be acquired from a larger part of the network
than one single bay [Oudalov et al., 2011].
Wide area monitoring, protection and control; Synchronized Measurement Tech-

nology utilizing Phasor Measurement Units
The availability of low cost, high precision timing resources, such as GPS (Global Po-
sitioning System) has made it possible to acquire phasor measurements from line cur-
rents and voltages from a larger network with highly accurate time stamping [O’Brien
and Deronja, 2012]. This new synchronized measurement technology utilizing Pha-
sor Measurement Units has gained interest in recent research publications, and many
19
new applications benefitting from this technology have been proposed, briefly sum-
marized below according to [Terzija et al., 2011].
• Real-time visualization of power systems
• Design of an advanced, early warning system
• Analysis (ex-post) of the causes of system blackouts
• Benchmarking and validation of system models
• Enhancements in state estimation
• Real-time congestion management
• Real-time angular, voltage, and frequency stability
• Improved damping of inter-area oscillations
• Design of adaptive protection and control systems.
Synchronized Measurement Technology allows more accurate status information

from the network to be gathered than was previously possible. In order to utilize
this data fully, and to facilitate the above-mentioned applications, the electricity dis-
tribution substation must be able to receive and process this stream of synchronized
measurements.
Condition monitoring and asset management of primary equipment
Smart grids need to optimize the utilization of all the network’s resources. This means
that all the network components need to be constantly monitored so that their con-
dition is known, and any required maintenance operations can be properly planned.
Using condition monitoring information for evaluating future maintenance needs is
a common topic in much of the published research. CBM and RCM are a focal
point for many utilities. However, these methods can only be properly utilized if data
collection and processing is available at the station level [Angel, 2003].
Typically, for both transmission and distribution networks, the cost of the sec-
ondary equipment in a substation is marginal when compared to the overall cost of
the whole network. An example calculation made for a 110 kV transmission network
20
shows that the secondary equipment in the substation only accounts for 2% of the to-
tal asset value, see Figure 2.6. Similar results have also been obtained for distribution
networks in Finland [Matikainen, 2011]. Clearly the cost of the secondary equipment
is not the main item affecting the overall cost-efficiency. On the other hand, substa-
tion automation does have a significant impact on the reliability of the network. This
indicates that focusing on substation automation is a cost-efficient way of improving
the reliability of the network.
Figure 2.6: Present value of a 110 kV network [Jeromin et al., 2009]
2.3.3 Future-proof technologies, upgradeability
Multi-vendor platform with open interfaces and multiple data streams
As networks get more complex, more data will flow through substations and more
interested parties need access to this data. Utilities are increasingly interested in out-
sourcing parts of their existing services, and this outsourcing to 3rd party service
providers requires clear and open SW interfaces to the process data of the substation.
21
An example of the integration needs of different SW systems is the AMR infrastruc-

ture, which in the near future will also be used for the fault diagnostics, an addition
to invoicing purposes.
As the functionality in a substation increases, the number of vendors providing
functionality to the substation also increases. This means that, in addition to open
process data interfaces, open interfaces are also needed for the SW platforms utiliz-
ing the data. In the future, the same SW platform may run applications from many
vendors, in the same manner as already seen in less critical environments, such as
personal computers or mobile phones [Johnson et al., 2010].
Cyber-secure firewall of the distribution network
When enhanced communication is used in the distribution network, cyber-security

becomes an essential part of the overall security. A secure product is not in itself suf-
ficient, as potential vulnerabilities may arise from insecure integration into existing
infrastructures [Nartman et al., 2009]. While a substation can form a separate, se-
cured island for energy distribution, it must also provide an information firewall for
parties communicating with the substation and the associated distribution network.
The requirement for open interfaces described above highlights the importance of
information security. In addition to preventing any cyber attacks, it must also protect
customer-sensitive information, e.g. related to personal data, invoicing or energy
consumption. This is especially challenging in an environment which also performs
mission-critical operations.
Upgradeability, engineering and verification processes
Modern IEDs are complex devices and commissioning and updating them is normally
handled by skilled personnel from the IED vendor or by a separate service provider.
Utilities seldom have their own personnel for extensive engineering work, at least in
Finland. There is a clear trend towards increasing the utilization of external service
providers. In many cases, utilities have their own designers for determining suitable
settings for the protection functions, but for other functional engineering, especially
system-level engineering of the substation, their know-how is limited.
The main challenges faced by the utilities are testing and verification. Any new
functionality must be tested at the system level, and this normally requires a deep
understanding of the devices in use. Test sequences also often mean interruptions
22
to the supply of electricity or, alternatively, detailed and time-consuming planning of

back-up connections. Therefore, the tendency is often to avoid implementing updates
if the predicted benefit is low, especially updates that affect engineering at the system
level. Furthermore, the functionality of those parts of the system which have not been
updated also needs to be tested. This all leads to updates being postponed until the
last possible moment, which means that the systems do not benefit from the advanced
functionality of modern IEDs. As stated earlier, focusing on substation automation is
a cost-efficient way of improving the reliability of the network, but this opportunity
is lost if the updating process is too complex and time-consuming.
The lack of generic, system-level engineering tools is often mentioned in technical
reports [Castallenos, 2009]. Although, this is not often an issue for the utility com-
panies themselves, as this service is contracted out to other engineering companies,
there is an obvious need for a generic tool to serve such companies. Although the aim
of standardization is to increase interoperability, very often the IEDs are still selected
from a single vendor, simply because this “reduces the possibility of trouble”. In
practice, using a number of different vendors often means a number of different tool
chains, a number of different philosophies in the engineering process and perhaps
even a number of different terms for the same parameter.
2.3.4 Low life-cycle costs
The above sections have described the new requirements for a substation. The speed
at which these requirements change is also expected to increase, which makes the
life-cycle cost calculations difficult. Currently, the life-span of protection and control
IEDs is presumed to be around 15 to 20 years [Lassila et al., 2002], but in many
scenarios, new requirements for automation are already expected over the next 5
to 10 years [Gorgette et al., 2007]. Although the utilities do not want to undergo
continuous and costly updates and upgrades to the whole protection system, the need
to adapt to new requirements is nevertheless clear.
There are many factors which affect the cost-efficiency of the distribution substa-
tion and the overall life-cycle costs. The most obvious, but, perhaps the least signifi-
cant factor in the long run, is the installation cost. Taking only this factor into account
would lead to a grossly oversimplified view, as new secondary systems would only
appear to decrease the life-cycle costs if the initial installation cost was lower than it
had been before.
23
2.4. The possible architectures
The costs of maintenance and upgrades to a distribution substation can be of the

same order as the original installation costs. As future distribution systems are ex-
pected to be more dynamic than static, the possibility of implementing smooth and
cheap updates is an important factor affecting the life-cycle costs. Advanced condi-
tion monitoring functionality, e.g. focusing on the CB condition or the transformer
temperature would also facilitate planning these activities in advance.
More difficult to estimate, but still the factor which may have the greatest impact
on the overall costs, is the profit that increased reliability may bring to the distribution
network. Currently, utilities in Finland are obliged to pay compensation to customers
whenever there is an interruption in the electricity supply exceeding 12 hours [EMV,
1995], and similar conventions are also in operation elsewhere in Europe. Power
quality is increasing in importance, and in the foreseeable future the quality of the
delivered electrical energy could also carry a price tag.
2.4 The possible architectures
2.4.1 Definition of a substation
The first question to be answered when thinking about new architectures is what ex-
actly constitutes a distribution substation, i.e. how it can be defined? In the Interna-
tional Electrotechnical Vocabulary of IEC 60050 [IEC, 1983] a substation is defined
as follows:
The part of a power system, concentrated in a given place, including mainly the
terminations of transmission or distribution lines switchgear and housing and which
may also include transformers. It generally includes facilities necessary for system
security and control (e.g. the protective devices).
The key point in this definition is the change from transmission to distribution –
a substation is the connection point between different voltage levels. There is the
incoming feeder (normally just one, although there may be a few) from a higher
voltage level (the transmission lines) and several outgoing feeders on a lower voltage
level (the distribution lines).
One aspect that is emphasized throughout this thesis is the data processing func-
tionality of the substation. A substation is not only an ‘energy hub’, but also an
‘information hub’. As it delivers energy to a large network at a certain voltage level,
the substation also monitors and controls the network. The substation is responsible
24
for keeping the network operational and running safely. Furthermore, many of the
control operations emanating from the NCC are focused on the substations.
2.4.2 A proposed architecture and a comparison with other solutions
The question is how to proceed with substation automation so that it is both inter-
operable and able to utilize new algorithms during the life-span of an IED, without
increasing the life-cycle costs or shortening the life-span of the physical device it-
self. At the same time, utilities want to discourage vendors from creating monolithic
secondary systems, which can only be sold and maintained by a single vendor.
Various concept-level proposals for the secondary system of a substation have
been proposed which address the conflicting requirements for low life-cycle costs and
the speedy utilization of new technology. These are presented later on in this chapter.
The most traditional approach has been to increase the functionality of the bay-level
protection and control IEDs, as described above in section 2.1.2. This approach has
been sufficient while CPU capacity has been steadily increasing and the price of
new technology has remained at a reasonable level. The issue in this approach has
been the extensive costs of upgrades. New features have also called for substantial
changes in the substation’s entire secondary system, requiring maintenance breaks or
the time-consuming planning of back-up connections. An overview of this set-up,
along with the two others, is presented in Figure 2.7, where the set-up is described as
‘Decentralized’.
Station
Computer
IED
Figure 2.7: Possible Architectures for Distribution Substation Automation.
25
An alternative approach has been to fully centralize the functionality in a distribu-

tion substation [Volberda et al., 2007] [van Riet et al., 2005] (Described in Figure 2.7
as ‘Centralized’). By moving all the functionality to a centralized station computer,
the life cycle of the bay-level measurement devices has been greatly extended, up
to the lifetime of the primary equipment [Volberda et al., 2007]. Furthermore, the
upgrade measures needed to implement new features have been simplified, because
only the centralized station computer requires updating. This, however, creates a
single point of failure in the substation. When the central station computer is out of
operation, the protection for the whole substation is lost. In practice, fully centralized
solutions would always need a redundant protection system – either a redundant sta-
tion computer [Baldinger et al., 2008] or redundant bay-level protection and control
IEDs – which would increase the overall costs of the substation secondary system.
The same maintenance problem as with a fully decentralized solution also exists.
When an upgrade is needed in an environment where all the functionality resides
in the same computer, the upgrade affects the whole protection system. Therefore,
maintenance breaks and extensive testing are required.
A third approach addresses the challenge by combining these two methods [Valtari
et al., 2009a] (described in Figure 2.7 as ‘Combined’). In this approach, only a part
of the bay-level functionality is moved to a new substation-level centralized station
computer. The functionality is divided so that the most critical and important features,
such as earth fault or overcurrent protection, would remain in the bay-level devices,
thus ensuring network safety in all situations. This forms the backbone of a network
protection system with a long life cycle. The functionality defined for the substation
level would consist of value-added applications and other "nice-to-have" features, for
which a faster update cycle is both necessary and acceptable. This set-up also has a
natural inbuilt back-up scheme, as the bay-level and station-level devices provide a
redundant protection system. The measures for updating the central unit are cheaper
and safer, allowing the smooth utilization of new functions. One hypothesis of this
thesis is that this combined approach will provide the most future-proof platform for
the secondary system of an electricity distribution substation, resulting in the lowest
overall life-cycle costs.
26
2.4.3 IEC 61850 standard
The increasing acceptance of the IEC 61850 standard has made it the ’de-facto’ foun-
dation on which all architectures must be based. Its increasing importance is also re-
flected in the title of the standard. In Edition 1, it was still was called ’The Standard
for Communication networks and systems in substations’ [IEC, 2005], but Edition
2 (not yet fully published) uses a broader term ’The Standard for Communication
Networks and Systems for Power Utility Automation’ [IEC, 2009]. Its use is also
spreading beyond substations, which guarantees that it will continue to be used in
substations. Therefore, the standard is briefly presented here, as it is a common ele-
ment in all the alternative architectures.
Station communication (IEC61850-8-1, -9-2)
The introduction and increasing acceptance of the IEC 61850 standard have made fast
and standardized Ethernet-based communication more available. First, the station
bus (part -8-1 of the standard) allows for the replacement of copper wiring between
IEDs on a horizontal level. Secondly, the process bus (part -9-2) makes the digitized
measurement information from instrument transformers available in a standardized
way for other devices.
The IEC 61850-8-1 station bus utilizing GOOSE messages (Generic Object Ori-
ented Substation Events) is already common in distribution substations, but the IEC
61850-9-2 process bus utilizing SAV messages (Sampled Analogue Value) has so far
only been extensively used in transmission applications. The assumption is that the
process bus also will become more common in distribution substations in the near
future.
Engineering aspects (IEC61850-6)
The recently published second edition of IEC 61850-6 also impacts on the engineer-
ing process [IEC, 2009]. The use of SCL (Substation Configuration Language) is
now more explicit, and in addition, the roles of different engineering tools are now
more accurately defined. Although the “real tool” can play many different roles, the
target of the clarification is clear. The division between vendor and/or IED-specific
tasks and system-level tasks must be clear before interoperability can be guaranteed.
It is not enough that IEDs from different vendors can communicate with each other
27
via GOOSE messages, if it is not possible to configure the two IEDs with a similar
engineering flow. Several different tools at the IED level can be accepted, and this
is also often necessary, but it must be possible to perform the system-level configu-
ration with one single system-level tool. These different levels of engineering tools
are described in the standard as the IED Configurator and the System Configurator.
Their different roles are shown in Figure 2.8.
Figure 2.8: Two main levels – IED Configurator and System Configurator. Updates
possible via IID files[IEC, 2009].
All engineering data is transferred via SCL files, whose context is indicated in
the file extension. SCD (Substation Configuration Description) is the extension for
system-level configuration files and CID (Configured IED Description) the extension
for IED-level configuration files. A new extension, IID (Instantiated IED Descrip-
tion) was introduced in order to further clarify the IED engineering process, and this
is also meant for IED configuration, and thus also shown in Figure 2.8. The difference
between CID and IID is that IID does not have any GOOSE engineering information.
IID is only IED-specific and does not have any information about other equipment
in the substation. IID can be edited entirely in the IED Configurator, whereas CID
requires information from the System Configurator. Figure 2.8 also illustrates how
updating the configuration of a particular IED is made easier by using the IID file
28
– any updates of the configuration do not affect the system-level configuration. A

model of the information flow in the whole engineering process is shown in Figure
2.9.
Figure 2.9 illustrates that the aim of the additions to IEC61850-6 is to make the
configuration process more modular, with more accurately-defined interfaces. The
IID file simplifies making modifications during the engineering process of one sub-
station project, but another new extension, SED (System Exchange Description), was
introduced in order to clarify the interfaces with other projects. This SED file is in-
tended for transferring engineering rights between projects. If one project needs to
update the data flow details of another project, the parts can be checked out from the
project via the SED files and checked in again later, after modifications have been
made.
System specifications
.SSD
(Single line, LNs, …)
IED Capabilities
IED (LN, DO, …)
DB .SED
.ICD System System Exchange System
Configurator Configurator
Associations, relation .IID Instantiated IED

to single line, .SCD .SCD
preconfigured reports
Other IEC 61850
Engineering IED project with interfaces
Workplace Configurator between projects
Engineering
environment
.CID Configured IED
File transfer
SA System
Remote Substation File transfer and
gateway parametrization with
File transfer IEC 61850 services
Local
IED IED IED
Figure 2.9: Reference model for the information flow in the configuration process
[IEC, 2009].
The updated standard also encourages vendors to harmonize their engineering

process in the same way that it has encouraged vendors to harmonize their communi-
cation. On the IED level, vendor-specific solutions are allowed, and also necessary,
but looked at from the system level, IEDs from each vendor must behave similarly.
The introduction of IID files also allows more flexible modifications during the ex-
29
ecution of a project. Within IID files, the configuration details of an individual IED
can be updated without affecting other parts of the substation, thus, for example, re-
ducing the need to “remap” the GOOSE signals every time the IED configuration is
changed. The intended modification process with IID files is shown in Figure 2.10.
Figure 2.10: Modification process [IEC, 2009].
The implications of IEC 61850-6 Edition 2
How do the clarifications in IEC 61850-6 affect the engineering flow? What is the
underlying philosophy and how should vendors address this? The second edition is
a natural continuation of other activities defined in the IEC 61850 standard, whose
main aim is to increase interoperability, i.e. it should be possible to use two IEDs
from different vendors in the same substation without extensive additional work. The
actual functionality can, and in practice also should, be different, but using these
different functions in the same substation should be possible with reasonable effort.
Communication via GOOSE is already possible, but the engineering should now also
be possible over a similar tool chain.
The updated standard implies that it should be possible to do system-level engi-
neering with a GOOSE configuration using an external 3rd party tool. This feature
not only affects the engineering interface, but also the details of the implementation
30
2.5. The cost-efficiency of the possible architectures
of the functionality. The functionality of IEDs should be self-contained in a way

that supports multi-vendor solutions. If, for example, advanced functionality is dis-
tributed to all the feeder IEDs which need to be configured in order to communicate
with each other via GOOSE (aggregated measurement values, intermediate calcula-
tions etc.), this does not, strictly speaking, follow the spirit of the standard. It forces
the utility to order all its IEDs from the same vendor because otherwise it would not
be possible to use the functionality. The standard also implies that one piece of log-
ical functionality should reside in one logical device, which in the standard is also
allocated to one physical device.
Of course, with widely-used and already-standardized functions, such as inter-
locking, this is not an issue – the requirements for the horizontal communication data
are well standardized as is the Ethernet frame of the GOOSE message. Sending inter-
locking information between IEDs of different vendors is feasible, and is actually in
use in substations. The problem becomes more apparent with regard to new, advanced
functionality, where, for example, it is necessary to share intermediate calculation re-
sults between different bays. An example of this functionality is high-impedance
earth fault detection, which utilizes measurements from all bays [Abdel-Fattah and
Lehtonen, 2009].
2.5 The cost-efficiency of the possible architectures

From the electricity distribution companies’ point of view, there are several aspects
to cost-efficiency. As previously stated, the most obvious, although least significant
part in the long run, is the installation cost. If only this aspect were to be taken into
account, it would lead to a highly simplified view of the concept, where centralized
protection on a station computer would not be cost-efficient unless the cost saving
gained through the reduced functionality of the bay-level IEDs exceeded the expen-
diture.
The maintenance and upgrade costs of a traditional bay-level protection system
for a substation can be of the same order of magnitude as the original installation
costs. This is not only due to the investment costs, but also because of the planning
work for back-up connections or possible interruptions to the electricity distribution.
In this case, the possibility of updating the functionality on a station computer during
shorter, or even non-existent power distribution interruptions with the consequent re-
duced need for secondary testing, generates considerable cost savings and can even
31
be a decisive factor when planning future maintenance activities. Condition monitor-

ing functionality in a centralized station computer would also help in planning these
activities in advance. A more detailed analysis of life-cycle costs is presented below.
2.5.1 Life-cycle costing in different scenarios
According to [IEC, 1997] [Hinow et al., 2008] the overall LCC (Life Cycle Costs)
of the equipment in an electricity substation can be broken down into the cost of
acquisition CA , the operation cost CO and the renewal cost CR , see (2.1).
CLC (t) = CA (t) + CO (t) + CR (t) (2.1)
In (2.1) CLC (t) describes the overall life-cycle costs during a time period t. The
operation costs can be further divided into maintenance costs CM (scheduled or
condition-based maintenance) and failure costs CF (unscheduled corrective main-
tenance), see (2.2).
CO (t) = CM (t) + CF (t) (2.2)
The failure cost CF itself is the sum of the component replacement cost CCR and
the penalty cost for undelivered energy CP , see (2.3).
CF (t) = CCR (t) + CP (t) (2.3)
In this section, the estimates for the life-cycle cost of the proposed set-up are
compared with other scenarios. Three different scenarios are evaluated:
A) All functionality is decentralized in IEDs and the IEDs need to be HE (high-end)
(HE) relays.
B) All functionality is centralized within a station computer (two station computers
are needed, since a back-up computer is required) and the measurements are retrieved
from merging units.
C) The set-up proposed in this thesis, i.e. a combined solution with a station computer
and IEDs. The IEDs can be LE (low-end) relays, since high-end functionality is
provided by the station computer
These three scenarios are evaluated with different cost factors: acquisition costs,
renewal costs, and maintenance costs. In addition, failure costs are calculated for the
different fault cases. The main purpose is not to derive full life-cycle cost calculations
32
for each scenario, but to indicate the areas where the life-cycle costs differ between
scenarios. The calculations are done for a 40-year period, in which the value of
many of the cost components will change, e.g. the interest rate on investments, the
acquisition cost of the equipment, the customer outage costs, the consumer price
index and the value of money in general. These changes are not taken into account
in the calculations, since they have a similar effect on all the scenarios and it can be
assumed that they will not result in any additional differences between the scenarios.
The cost components also include factors which are impossible to fully estimate
accurately over a 40-year period. An example of this is the acquisition costs of the
equipment presented in the next section. It is impossible to estimate the price of
the substation automation equipment over 40 years accurately. This is particularly
true if the concepts behind this thesis become a reality, i.e. if the substation automa-
tion architecture moves towards a more centralized architecture. Such a fundamental
change in automation architecture will also eventually change the pricing structure of
IED vendors, and may well bring about new competition in this business area. There-
fore the analysis is performed based on the current cost structure, without making any
presumptions about future price developments.
2.5.2 Acquisition costs
In order to obtain acquisition costs, it is necessary to make rough estimates of unit

prices (both for acquisition and renewal) and of the expected intervals for renewal,
and these are shown in Table 2.1. The prices are based on a survey made for the
Finnish Energy Market Authorities [Pahkala et al., 2010], which is aimed at assess-
ing reasonable pricing for electricity distribution and transmission network services.
As indicated above, the analysis is performed based on the current cost structure,
without making any presumptions about future price developments. In the survey,
the protection and automation devices for one feeder (including planning, installa-
tion and wiring) had the unit price of e5900, and the basic automation infrastructure
(including planning, RTU, communication modems and installations) had a unit price
of e13,900.
As the prices were average total prices, some additional estimations were needed.
Enquiries to IED vendors resulted in using the total automation price of one feeder
for the HE relay, and half of that for the LE relay. The most difficult item to evaluate
was the station computer, as it is an item which is not yet on the market. In order
33
Table 2.1: Price and Renewal Interval for Different Equipment

Acquisition / ke Renewal / ke Ren.Interval / a
High-end relays 6 6 15
Low-end relays 3 3 20
Merging units 1 1 20
Station computer 20 3 2
Back-up computer 20 20 20
not to give too positive a picture of the centralized set-up, an estimate of two times
the price of the RTU (assumed to be e10,000 out of total of e13,900) was used, as
shown in Table 2.1. The life-span of the IEDs is estimated to be 15-20 years [Lassila
et al., 2002], of which two alternatives the shorter 15-year period was selected for
the HE relay and the longer 20-year one for the LE relay. Furthermore, no separate
prices were given for merging units, so an estimate of one third of the price of the LE
relay was used to describe the limited functionality of a merging unit. The life cycle
was assumed to be the same as with LE relays, because for both items the life cycle
of the electronics is the limiting factor.
The renewal price is the same as the acquisition price for all the items except the
station computer. The station computer is expected to be updated more frequently,
and in these cases most of the renewal actions consist only of software updates. In
a fully centralized set-up, the second station computer is only needed for back-up
functionality, and therefore the same update interval is estimated as for the low-end
relays.
When using the unit prices from Table 2.1, acquisition costs for different scenarios
are as shown in Figure 2.11. Based on Figure 2.11 one can conclude that using only
bay-level high-end relays is the most cost-effective solution in most of the cases.
When a substation has more than seven feeders, the combined set-up is the most
cost-effective. When the number of feeders exceeds 10, having a fully centralized
set-up yields the lowest acquisition costs.
2.5.3 Acquisition and renewal costs

If the renewal costs are also taken into account, the situation changes. High-end
relays have a shorter life cycle, since all the functionality required for the substation
needs to be realized through the IEDs. This means a faster upgrade cycle has to be
34
Acquisition costs
80
70
60
Costs / k€
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12
No of Bays
HE Relays LE Relays and Station Computer MU and Two Station Computers
Figure 2.11: Acquisition costs for different scenarios.
applied than in the other scenarios (15 years). The life cycle of the low-end relays and
the back-up station computer is estimated to be the same (20 years). The life cycle
of the station computer is set to as low as two years, in order to capture the dynamic
nature of any future substation. On the other hand, these upgrades are mainly SW
updates, which mean inexpensive renewals. Using these parameters, the overall costs
for a 40-year period would be as shown in Figure 2.12.
Acquisition and renewal costs

300
250
200
Costs / k€
150
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12
No of Bays
Figure 2.12: Acquisition and renewal costs for different scenarios.
This addition already makes the scenario with a station computer and low-end
relays more cost-effective, when the substation has only six feeders.
35
2.5.4 Acquisition, renewal and maintenance costs
The next item to be included in the life-cycle cost is the scheduled maintenance. The
secondary system always needs scheduled testing, regardless of the scenario. These
maintenance activities are not taken into account individually, but instead only those
activities that differ between the scenarios are calculated, such as the maintenance
activities during the renewal. The previous calculations have only taken into account
the cost of equipment during the renewal. However, a renewal also requires the util-
ities to take other actions not directly related to the equipment, such as project plan-
ning, gathering quotations, vendor selection, planning temporary network topologies
during the renewal, the execution of switching operations for back-up connections,
etc. Accurate statistics for these costs were not available, but according to estimates
from Finnish utilities, these operations require roughly one and a half man-months
per substation, which in Finland amounts to approx. e15k in additional costs.
These maintenance costs have the same interval as the primary protection of the
station, which in this example means 15 years for a fully decentralized set-up and
20 years for both of the centralized set-ups. Taking these additions into account, the
costs for the secondary system are as shown in Figure 2.13.
Acquisition, renewal and maintenance costs

350
300
250
Costs / k€
200
150
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12
No of Bays
Figure 2.13: Acquisition, renewal and maintenance costs for different scenarios.
This addition makes the centralized set-up even more economically viable, in that
a centralized set-up can be justified even for a substation with only five feeders.
36
Case example: maintenance cost of the main transformer
In addition to the maintenance costs of the secondary system itself, the functionality
in the substation also affects the maintenance costs of other substation components.
As mentioned in section 2.3, a substation’s functionality needs to support the utili-
ties’ asset management procedures. Asset management as a whole is a large topic,
so instead of a full analysis, one example case for a substation main transformer is
presented.
The main transformer of a substation is the single most expensive component
of the substation, with 19% asset value share of the whole network [Jeromin et al.,
2009]. According to [Pahkala et al., 2010] the cost varies from e246 k (for a 6 MVA
transformer) to e900 k (for a 100 MVA transformer) with main transformers from
110kV to 10/20kV. Because of these costs, condition monitoring of the transformer
is an important aspect of asset management. For example, the rated temperature of
the transformer must not be exceeded, because that would shorten the lifetime of
the transformer, leading to premature maintenance. On the other hand, too large a
safety margin would mean that the transformer would be oversized. If the improved
functionality in the protection and automation system were to allow, say, the use of
a 16 MVA transformer instead of a 20 MVA one, this would instantly yield a saving
of e40 k in the investment phase. Furthermore, an improvement in automation can
result in postponing the replacement of the transformer, due to, for example, the lower
operation temperature of the transformer, which would yield yet further savings.
2.5.5 Failure costs
Although it is more difficult to estimate, the factor which can have the largest impact
on the overall costs is the cost saving gained by increasing the reliability of the dis-
tribution network. When the substation automation functionality is up-to-date, it can
be assumed that the overall performance will be better than it was before the update,
i.e. the protection would be more selective (less unwanted trips), the fault situations
would be cleared faster, the areas affected by the faults would be smaller, and the
maintenance requirements would be more accurately known, etc.
However, this factor is very difficult to estimate. What is the reliability of a 10-
year old secondary system, compared to a 1-year old system? How often does the
secondary system malfunction? Furthermore, what would the answer to these two
questions be in 10 years time? For distribution networks in particular, there are no
37
accurate statistics for these factors as the utilities have not had the resources to collect
and analyze the statistics on secondary system failures. The financial consequences
of faulty operation can also vary greatly. One unwanted trip on an outgoing feeder
might not have any real effect on the ENS (Energy Not Supplied). On the other hand,
one missing operation on the outgoing feeder protection can trigger the protection on
the incoming feeder, causing an interruption to the whole substation.
To better understand the costs of a failure, an assessment of the average outage

times in Finnish distribution networks has to be carried out. The financial conse-
quences can then be evaluated based on these average values, and according to spe-
cific cases of use.
Customer outage costs in Finland
The importance of an uninterrupted electricity supply has increased steadily over the
years, and is well illustrated by the estimated customer outage costs. Over the past 10
years these costs have doubled, or in some cases even tripled [Partanen et al., 2006].
Nevertheless, it would be wrong to assume that this trend will continue in exactly the
same way. Recent studies have shown that it looks more likely that the differences
between the different customer groups will grow. For many consumers, an unin-
terruptible electricity supply is not critical, and therefore the increase in customer
outage costs can be expected to stay within reasonable bounds. On the other hand, in
specific industry and service areas the outage costs may increase considerably. The
most recent results from the evaluation of customer outage costs in Finland for differ-
ent customer groups are shown in Table 2.2 (at 2005 prices) [Partanen et al., 2006].
The weighted average values in Table 2.2 are the ones used in the Finnish regulation
model [EMV, 2011] (also at 2005 prices). The consumer price index maintained by
the Official Statistics of Finland (OSF) indicates an increase of 13.6% for the values
in Table 2.2 in terms of 2011 prices [OSF, 2011], but for the sake of clarity, this is
not taken into account in these example cases.
38
Table 2.2: Customer Outage Costs in Finland for Different Customer Groups [Parta-
nen et al., 2006] [EMV, 2011]
Unplanned Planned
e / kW e / kWh e / kW e / kWh
Households 0.36 4.29 0.19 2.21
Agriculture 0.45 9.38 0.23 4.80
Industry 3.52 24.45 1.38 11.47
Public 1.89 15.08 1.33 7.35
Services 2.65 29.89 0.22 22.82
Weighted average 1.1 11.00 0.5 6.8
Average outage times in Finnish distribution networks
The Finnish Energy Industries authority (Energiateollisuus ry, ET) publishes inter-
ruption statistics every year. The results from the past six years and the calculated
average values are shown in Table 2.3 [Energiateollisuus, 2005-2010], and include
both long interruptions (permanent faults) and short interruptions (faults cleared by
auto-reclosing).
Table 2.3: Amount of customer outages in Finland 2005-2010 [Energiateollisuus,

2005-2010]
2005 2006 2007 2008 2009 2010 Average
Time (h)
Short Interruptions (h) 0,12 0,05 0,05 0,06 0,11 0,07 0,08
Long Interruptions (h) 2,36 1,57 1,15 1,56 0,92 3,22 1,80
Planned Interruptions (h) 0,48 0,31 0,3 0,31 0,3 0,36 0,34
All Interruptions (h) 2,96 1,93 1,5 1,94 1,3 3,61 2,21
Amount (pcs)
Short Interruptions (pcs) 10,02 3,54 4,45 4,94 3,89 5,98 5,47
Long Interruptions (pcs) 3,52 1,92 1,66 1,8 1,32 2,45 2,11
Planned Interruptions (pcs) 0,55 0,25 0,26 0,22 0,24 0,31 0,31
All Interruptions (pcs) 15,19 5,71 6,38 6,96 5,43 8,72 8,07
39
Using the average outage values from Table 2.3 and the weighted average values
from Table 2.2, the overall outage costs for a 40-year period are as shown in Figure
2.14, assuming that the outage costs and interruption times remain similar during that
period. In these calculations, 0.81 MW was used for the power of one feeder, i.e. the
average power of feeders in Finland, this also being calculated from the statistics of
ET. When this chart is compared to the other costs calculated in Figure 2.13, we can
see the proportion of outage costs in comparison with the overall costs. An interesting
way to view to these charts is to calculate the ratio between the costs, i.e. only 2.4% of
all the interruption costs equals all the costs of the ’High-End relays’-scenario shown
in Figure 2.13. Remember that the aim of this calculation was not to derive exact cost
shares, but to to indicate how much larger the outage costs are in comparison to the
other costs.
Interruption Costs
14000
12000
10000
Costs / k€
8000
6000
4000
2000
0
1 2 3 4 5 6 7 8 9 10 11 12
No of Bays
Short Interruptions Long Interruptions All Interruptions
Figure 2.14: Total interruption costs.
The reliability of the protection system
An attempt was made to estimate the cost of protection failures in Norway [Kjølle
et al., 2005]. The paper presents the key results gained from a study of incorrect
operation of protection and control systems at voltage levels of 1-420 kV in Nor-
way, consisting mainly of incorrect and missing operations. The statistics for the
period 1999-2003 show that incorrect or unwanted operation is a major fault type,
40
and that the relative number of faults and their contribution to the ENS increases as
the voltage level rises. The study estimated that of all the ENS due to faults in the
distribution network, approximately 5% were due to failures in the protection system,
and approximately 1.5% of all the operations were faulty. 46% of all the failures were
unwanted operations and 6% were missing operations; the remaining 48% of failures
were not accurately specified. Unfortunately, the detailed results for the distribution
networks were insufficient, and only gave an overview of the percentage of faulty
operations.
The outage statistics of ET include some analyses of the causes of the fault, but
unfortunately those statistics are also limited [Energiateollisuus, 2005-2010]. Be-
tween the years 2005 and 2010, on average over 11% of all faults were due to faulty
construction or faulty operations. However, no detailed division into faulty relay op-
erations was provided. Furthermore, on average as much as 22.5% of the faults were
attributed to ’unknown cause’.
Because of these deficiencies in the statistics, it was not possible to carry out an
accurate evaluation of the cost of faulty relay behavior. Instead, a few scenarios were
evaluated in order to show the cost impact under different assumptions. The first
reference point was set to this moment; a certain percentage of all interruptions at
a given moment are unnecessary and faulty. Then, assuming that the reliability of
the protection system will improve in the future, a reliability graph was created, as
shown in Figure 2.15.
In Figure 2.15, the reliability for year zero was set as the reference point. The blue
line in the figure represents the estimated maximum achievable reliability. In this
case, it is assumed that after 15 years the reliability of the protection will be higher
than it is today. The red line in the figure represents the change in the reliability of
the decentralized set-up. After installation, the reliability starts decreasing for several
reasons, e.g. the equipment is aging or the parameterization becomes outdated due
to changes in the network. This is shown in the figure as a decline, based on the
assumption that without upgrade measures a larger proportion of all operations would
be faulty after 15 years. The aim of Figure 2.15 is to illustrate the benefits of a
centralized set-up. The functionality is constantly near the state-of-the-art level due
to frequent updates, but the maintenance costs are kept at a minimum.
How much the reliability might improve within 15 years (and how much the re-
liability of the decentralized set-up might decrease) is not known, due to insufficient
statistics. This is also the reason that there are no numeric values on the Y axis in
41
Reliability due to protection
0 5 10 15 20 25
Years
Maximum achievable reliability Decentralized setup Centralized setup
Figure 2.15: An example reliability graph of protection
Figure 2.15. Instead, different scenarios were evaluated, and the results of this are
shown in Figure 2.16.
Cost saving with increased reliability

0
1 2 3 4 5 6 7 8 9 10 11 12
-50
Costs / k€
-100
-150
-200
-250
No of Bays
0.5% unit impr. 1% unit impr. 2% unit impr.
Figure 2.16: Cost saving with different scenarios.
42
The curves in Figure 2.16 represent three different scenarios. The zero level in the
figure indicates the reference case, in which the reliability of the protection devices
remains the same during the whole 40-year period. The ’0.5% unit imp.’ scenario
means that after 15 years there are 0.5% fewer unnecessary trips per unit. In other
words, if the amount of unnecessary trips were 5% today, after 40 years it would be
4.5%. Similarly ’1% unit impr.’ describes a scenario with 1% unit improvement, and
’2% unit impr.’ a 2% unit improvement. The cost saving in this last scenario would
already be on a par with all the other costs shown in Figure 2.13.
Case example 1: Unnecessary tripping of one feeder
The first example case to be evaluated was the cost of an interruption in one feeder.
The outage costs were calculated for when one feeder experiences an unnecessary
trip and customers have an interruption in the energy supply due to a malfunction in
the protection. The results for one single interruption with different outage times and
for different customer groups are shown in Table 2.4. The first five cases assume that
all the customers of the particular feeder belong to the same customer group, and the
last, the 6th case, is a weighted average feeder [EMV, 2011].
Table 2.4: Cost of one interruption in one feeder (in ke)

Outage cost (ke)
Case Customer group 1min 6 min 60 min 120 min
1 Households 0,35 0,64 3,78 7,27
2 Agriculture 0,49 1,13 7,99 15,62
3 Industry 3,19 4,85 22,74 42,63
4 Public 1,74 2,76 13,80 26,06
5 Services 2,56 4,59 26,46 50,76
6 Weighted Average 1,04 1,79 9,84 18,78
Table 2.4 illustrates the differences between different customer groups. If one
feeder line contains only household customers, the outage costs are negligible, but
the cost for industrial customers is from five to nine times higher.
43
Case example 2: Missing trip in one feeder, causing an unnecessary trip for the
whole substation
The second example deals with a missing operation. If the protection in one feeder
fails, and the back-up protection in the incoming feeder trips instead, all the other
healthy feeders experience an unnecessary trip. According to ET statistics [Energia-
teollisuus, 2005-2010] the average size of a Finnish substation is ten feeders, which
means that nine of them would be tripped unnecessarily. The outage costs in this case
are shown in Table 2.5, again with six different cases.
Table 2.5: Cost of one interruption in the whole substation (in ke)
Outage cost (ke)
Case Outage time(min) 1 6 60 120
1 Households 3,16 5,77 34,03 65,43
2 Agriculture 4,44 10,16 71,94 140,58
3 Industry 28,74 43,65 204,69 383,63
4 Public 15,67 24,87 124,19 234,55
5 Services 23,04 41,27 238,14 245,88
6 Weighted Average 9,39 16,10 88,55 169,05
The total costs derived for different scenarios in Figure 2.13 were in the range
of e200 - 250 k for a substation with 10 feeders. One single 60 minute outage in a
substation with only industrial customers would cost the same. On the other hand,
a substation in a residential area could experience six similar events before the other
costs in Figure 2.13 were exceeded.
2.5.6 Summary of LCC cost estimates

The currently prevalent approach of using only bay-level IEDs will only be econom-
ically justifiable in the future in very small substations. In contrast, according to the
estimates presented earlier in this chapter, a substation with more than five feeder
bays should utilize centralized station computers. The calculations presented here
very much depend on the initial costs given in Table 2.1. Depending on the initial
cost estimates, the size of a substation for which the centralized set-up becomes more
cost-effective can vary. Nevertheless, it is clear that the larger the substation, the
more likely a centralized station computer will decrease life-cycle costs.
44
However, as stated earlier, the costs incurred through failure in the delivery of
electricity are much higher than all other the costs combined, at least in Finland
with the outage costs that apply to Finnish networks (presented in Table 2.2). This
confirms the main idea behind this thesis, i.e. investing in substation automation is a
cost-efficient way of improving the reliability of the network. Of course, this applies
to all the evaluated architectures, and not just to the set-up proposed here, which is
based on a centralized station computer.
2.5.7 Other benefits of the combined set-up
As shown earlier, upgrading functionality in IEDs is a problematic task. It demands

updates to the whole protection system of the substation. Updates are expensive,
not only because of new installations and new equipment, but also because of the
required manual engineering, commissioning and testing work. If the functionality
utilizes measurements from more than one bay, an implementation at the IED level
is only possible when all the IEDs are from the same vendor.
According to the IEC 61850 standard, and also from the engineering point of
view, it is beneficial that system-level engineering and functionality do not have un-
necessary IED-level vendor-specific features, such as algorithm-level communication
based on intermediate calculations. This also means that unnecessary GOOSE engi-
neering at the substation level should be avoided. In addition to the extra engineering
work, testing and commissioning in such environments is a complex task, not to men-
tion the documentation and maintenance issues. When the new functionality can be
encapsulated within one logical device, the amount of system-level communication
engineering is reduced. Testing, maintenance and documentation at the system level
is also reduced compared with a situation where the functionality includes system-
level communication engineering.
The architecture proposed in this thesis, which is based on centralized station-
level protection devices and the IEC 61850-9-2 process bus, provides a framework
for these new logical devices. When all the measurements are available at the station
level via the IEC61850-9-2, it is feasible to implement new functionality by hooking
in to the process data via the process bus. Ideally, system-level configuration can be
avoided entirely. One example of such a station function which is already on the mar-
ket is the high impedance earth fault protection. There are solutions which require
measurements (and separate wiring) to all the feeder bays in order to guarantee de-
45
2.6. Details of the proposed architecture
tection of high-impedance earth faults [Tengdin, 1996][Abdel-Fattah and Lehtonen,

2009] [Nikander, 2002].
This approach reduces the complexity of bay-level IEDs and the bay-level IED
engineering tools. Functionality can be divided into logical devices, which is also the
target of the standard. These logical devices could then be acquired from different
vendors and added to the substation as they all share the same interface – measure-
ment data via SAV messages as defined in IEC 61850-9-2 and other communication
via GOOSE messages as defined in IEC61850-8-1.
The engineering flow could be adjusted so that functional engineering would be
directed via the IID files introduced by IEC 61850-6 Edition 2. When a function
resides in one IED/Computer/Logical device, it can be updated and tested without
affecting the system-level configuration. Functionality requiring changes to the full
system configuration should be avoided as it incurs high costs during the upgrade and
commissioning.
In practice, this approach moves some of the implementation from the GOOSE
signals to the logical device’s internal logic. This implementation can be vendor-
specific, whereas GOOSE should not be. The proposed architecture enables new
advanced functionality to be added as a single module at the station level – such a
module could be either new equipment or even a new module in a centralized SW
platform, as envisioned by the Swedish utility, Vattenfall [Johnson et al., 2010].
2.6 Details of the proposed architecture

This section describes the elements of the architecture investigated in this thesis,
and referred to in earlier sections as the ’combined set-up’.This set-up is based on
centralized protection and control functionality, which complements but does not
replace the functionality of the bay-level protection and control IEDs. The overall
structure of the proposed architecture is presented in Figure 2.17.
2.6.1 Important standards
IEC 61850 modeling and station communication
As already described in section 2.4.3, nowadays IEC 61850 has become the ’de-facto’
standard. The recent additions and clarifications, which are also mentioned in the
same section, have made it clear that future IEDs need to fully support IEC 61850. In
46
Figure 2.17: Overall set-up of a centralized protection and control system.
most respects this is already the case, and does not need any further comment. One
aspect not yet used in distribution automation is the process bus specified by IEC
61850-9-2, yet for the reasons presented in this thesis, in the future it is more than
likely that this process bus will be used.
IEEE 1588 time synchronization
With Ethernet-based technology it is possible to achieve software-based time syn-

chronization with an accuracy of 1 ms quite easily, and without any help from HW.
This is also what the IEC 61850 standard refers to as the basic time synchronization
accuracy class (T1) [IEC, 2005].
An older and more common protocol is the SNTP (Simple Network Time Proto-
col) , which is suitable for local substation synchronization in relatively small sys-
tems Ferrari et al. [2011]. However, if the SNTP server is behind multiple Ethernet
nodes, the latency increases, which reduces the accuracy of the time synchronization.
Therefore, SNTP is not an ideal solution for system-wide implementation. Normally
47
a GPS or equivalent time synchronization resource is required in every substation.

IEEE 1588 [IEEE, 2009] deals with these issues and makes it possible to achieve a
time synchronization accuracy of 1 µs. This is required if an IEC 61850 process bus
is used. IEEE 1588 provides a cost-efficient and accurate method for system-wide
time synchronization, where the network devices are able to correct the node delays
into time synchronization frames. The application notes from technology vendors
promise even greater accuracy of 10 ns [Texas Instruments Inc., 2007].
2.6.2 Protection and control IEDs
Using this concept, the protection and control IEDs are still seen as important compo-
nents of the secondary system. They handle the time-critical basic protection func-
tions, and they also communicate with a centralized station computer. The main
idea is that the protection and control IEDs are ’freed’ to perform their original and
mission-critical operations. By removing the non-critical functions from the bay-
level devices the life cycle of the devices can be maintained at the current level. Tra-
ditionally, the life-span of these devices has been 15-20 years [Lassila et al., 2002],
largely due to the lifetime of the electronics used in the device. The aim is to retain
the same life-span and to avoid the need for early updates for application or functional
reasons. A more detailed description of the ’time-critical basic protection’ function-
ality which would still be needed in bay-level devices will be presented in Chapter
4.
One important feature of the bay-level devices is that they support the relevant
and most important standards, so that it is feasible to integrate new devices and/or
functions later on. These standards have been presented earlier in this chapter, and
for the architecture investigated in this thesis the most important ones are IEC 61850
and IEEE 1588.
2.6.3 Station computer
In the proposed architecture, the centralized station computer handles all the ad-
vanced functionality. As the primary protection is covered by bay-level IEDs, the
functionality in the station computer can be updated at will without affecting the
safety of the network, allowing fast and smooth updates.
Another important factor is that with small (less than 5 feeders) substations, the
48
most economically viable solution is to use only bay-level devices. The concept pro-
posed here allows this, and the centralized station computer can also be added later,
provided that the bay-level devices that are used support the required standards. This
means that future updates can be installed without the need to renew the whole sec-
ondary system, and cost-efficient migration scenarios can be utilized, guaranteeing
optimal utilization of investments made today.
Unlike the IEDs, the station computer is not directly connected to the measure-
ment devices. The IEDs handle the measurements and send them on to the station
computer (as well as using the measurements themselves for protection purposes).
When these measurements (and other relevant data, such as control commands and
status information) are sent according to IEC 61850, it is purely Ethernet-based. This
is an important benefit, as it enables the use of standard industrial PC technology
as a base, and provides for better economies of scale than the dedicated designs for
embedded systems which are currently utilized in protection and control IEDs.
The functionality in the station computer can be broadly divided into two cate-
gories. First, there is the protection and control functionality, which needs real-time
process data and directly affects network safety. This functionality has strict require-
ments for reliability and cyber security.
The second category is offline functionality, which can operate based on historical
information. This functionality only indirectly affects the network’s safety via, e.g.
the condition monitoring and fault analysis functions. In this functionality, open
interfaces can be provided for 3rd party functions, and multi-vendor SW platforms
are possible. This categorization is presented in Figure 2.18. Here, the ’Data Storage’
describes the data to which external partners may have access, and a database symbol
highlights the fact that the data does not directly affect the control and protection of
the network.
It should be noted that, currently, protection and control devices hold many func-
tions with different ’functional life cycles’. Many basic protection functions can be
expected to remain the same throughout the life cycle of the IED, whereas, say, a new
fault location algorithm may be superseded by a new improved version as early as the
following year. The aim of this concept is to locate those functions that have differ-
ent life cycles in different places, thus retaining a long life cycle for the IEDs while
simultaneously allowing a shorter life cycle for the station computer. This approach
will be studied further in Chapter 4.
49
2.7. Pilot installation of the selected architecture in Noormarkku
Figure 2.18: Separate functionality for centralized protection and monitoring.
2.7 Pilot installation of the selected architecture in Noor-

markku
A pilot installation was implemented in the substation located at Noormarkku village

in Finland. This was done in cooperation with Fortum and ABB in 2008, during the
IDA project mentioned in section 1.2. The purpose was to have a ’proof-of-concept’
type of pilot, which would show that time-critical protection functionality can be
achieved with the architecture proposed in this thesis. The pilot installation consisted
of the following items:
• Bay-level protection and control IEDs supporting both the station bus defined
in IEC 61850-8-1, and the process bus defined in IEC 61850-9-2, but not sup-
porting IEEE 1588. There were 8 feeder bays in the pilot installation.
• One station computer, utilizing standard PC technology and supporting IEC

61850-8-1, IEC 61850-9-2 and also IEEE 1588
• IEEE 1588 time syncronization master
• The required wirings from voltage and current transformers to the bay-level
devices
50
2.8. Chapter summary
• Communication between the station computer and the bay-level devices real-
ized with optical fibers
The functionality in the station computer consisted of overcurrent protection, and

also control logics for the CBs with interlocking functionality. The time synchro-
nization was not fully tested, as the bay-level devices did not support IEEE 1588.
Furthermore, the devices were installed next to the existing protection relays. i.e. the
trip signal from the IEDs used in the pilot was not connected to the trip circuit. What
the pilot clearly showed was that the proposed architecture can be implemented with
the existing technology. During the 1-month pilot period there were no real fault
cases, so it was not possible to verify the system’s operation during fault situations.
The project ended after the piloting period.
2.8 Chapter summary

Because of the increasing use of distributed generation, and other active resources,
distribution substations require more intelligence. As the significance of these “Smart
Grid impacts” is not yet known, the secondary system needs a faster upgrade cycle
than it has done previously. Any future architecture for the automation of a distribu-
tion substation needs to enable this without increasing the life-cycle costs. To address
these life-cycle cost challenges, the architecture proposed here utilizes a centralized
station computer and new communication and time synchronization standards.
The currently prevalent approach of using only bay-level IEDs will only be eco-
nomically justifiable in the future for substations with a maximum of five feeders. In
contrast, the calculations presented in this thesis show that a substation with more
than five feeder bays should already utilize centralized station computers. However,
the costs of undelivered electricity outweigh all the other costs combined, at least in
Finland with the outage costs calculated for Finnish networks. This further empha-
sizes the importance of upgradeability for the substation automation system.
Both centralized substation functionality and the second edition of IEC61850-6
support a more modular view of the secondary system. This modularity should be uti-
lized, because the possibility of upgrading functionality without affecting the system-
level communication engineering reduces the cost of an upgrade and increases the
cost-efficiency of the distribution network, thanks to the optimal and fast deployment
of new state-of-the-art functionality.
51
The calculations presented here very much depend on the initial costs given in
Table 2.1, and especially on the estimates for the reliability of the protection system.
More detailed results cannot be provided without further research into the reliability
of the protection system. Depending on the given values, the size of a substation
for which the centralized set-up becomes more cost-effective can vary. However, the
main point is clear, i.e. the larger the substation, the more likely it is that a centralized
station computer will reduce the life-cycle costs.
The requirements presented in section 2.3 may also compel the implementation of
faster renewal cycles on the decentralized set-up than the estimated 15 years, making
the benefits of centralization even more persuasive.
52
Chapter 3
Station Measurements
This chapter shows how station-level measurements can be used to achieve additional
measurement accuracy. Chapter 2 mainly focused on showing how the existing pro-
tection system could be implemented in a more cost-efficient way with a different
architecture. This chapter takes another approach. It seeks out what more can be
done with the proposed substation automation architecture, i.e. what features can be
implemented in the proposed architecture which can not be implemented with the
currently dominant architecture, utilizing only bay-level IEDs.
3.1 Measurement replacement
When all the current measurements are locally available, a missing current measure-
ment can be calculated with the help of Kirchhoff’s current law. This can be useful if,
for example, one of the measurement transformers is out of order and measurements
are not available.
Another option is to save costs by removing unnecessary measurement transform-
ers or sensors entirely, and sharing the measurement data from another IED. Nor-
mally, with current measurements, this is not feasible, as for safety reasons each
feeder must have its own current measurements available. But for voltage measure-
ments this is a possible option - phase voltages measured from the busbar could be
delivered to all the feeder bays with IEC 61850-9-2, which would yield cost savings
as only one measurement chain would be needed [Starck et al., 2012].
However, in this thesis these issues are not studied further, and instead, the main
53
3.2. Measurement merging
focus is on measurement merging, i.e. what can be achieved by merging the mea-
surement streams from different devices.
3.2 Measurement merging
3.2.1 Introduction
Modern protection and control IEDs in a distribution network have a fairly moderate
sampling frequency, typically in the range of 1- 2 kHz [ABB Ltd, 2010]. This is
because most of the functionality in the IED is based on phasor measurements, cal-
culated from the nominal frequency components of the phase currents and voltages.
A higher sampling frequency does not bring any benefits when calculating the com-
ponents of current and voltage measurement signals with a nominal frequency of 50
Hz or 60 Hz.
Much research has been done regarding transients in power systems. Whenever
a change occurs in the status of the network, e.g. due to a fault, there is also always
a transient phenomenon involved. The generated transients normally have a higher
frequency than the system frequency, between 100 Hz and several kHz [Lehtonen and
Hakola, 1996]. According to the Nyquist-Shannon sampling theorem, a 1 kHz signal
should be detectable with a 2 kHz sampling rate [Phadke and Thorp, 1990], but this
is not fully applicable for transients. The transients last for just a few milliseconds
and therefore the number of data points is low, especially if the sampling frequency
is low.
Methods based on transient measurement have rarely been used in IEDs. The ad-
ditional unit costs of the IEDs with higher sampling frequencies (and the additional
CPU performance it requires) have been considered too high compared with the ben-
efits they can provide. Furthermore, there has not been great customer demand for
transient-based methods, as in most cases, the standard measuring methods based on
fundamental frequency components have been sufficient.
This thesis introduces a method for increasing the sampling frequency of current
and voltage measurements in an artificial way, by processing multiple measurements
on a lower sampling frequency.
54
3.2. Measurement merging
3.2.2 Time-Interleaved ADC (TI-ADC)
Acquiring high sampling frequency with low cost ADCs (Analogue-to-Digital Con-
verters) is a constant target in signal processing [Vogel, 2005]. Time-interleaved
technology approaches this target by using multiple ADCs, which have the same
sampling frequency but different sampling phases. The architecture is shown in Fig-
ure 3.1 [Vogel, 2005], where each ADC has a sampling frequency of fs / M, where
M is the number of ADCs and fs is the sampling frequency of the combination. The
timing diagram of the system is shown in Figure 3.2, where the sampling period of
the combination is Ts , but the sampling period of an individual ADC is M Ts .
Figure 3.1: Time-Interleaved ADC with M channels [Vogel, 2005].
The concept itself is not new [Black and Hodges, 1980], but technological chal-
lenges have limited the number of actual implementations to just a few. The main
issue is that any mismatch between the individual ADCs introduces spurious compo-
nents into the spectrum, thus degrading the SINAD (SIgnal-to-Noise-And-Distortion
ratio)[Vogel, 2005]. These mismatches consist of differences in gain, offset and tim-
ing, where timing refers to the deterministic deviation between the ideal and the real
sampling time, resulting from an inaccurate synchronization of the ADCs. Stochas-
tic timing deviation from each ideal sampling point is a separate error source, often
referred to as the timing jitter.
Both the individual and combined effects of these mismatches have been exten-
sively studied in various scientific publications [Vogel, 2005] [Soudan and Farrell,
55
3.3. Increasing sampling frequency at the substation level
Figure 3.2: Timing diagram of time-interleaved ADC with M channels [Vogel, 2005].
2008]. Several different compensation methods have also been investigated for cor-
recting these errors, also presented in [Vogel, 2005]. A more detailed error analysis
suitable for protection and control IEDs is presented later on in this thesis.
The advancements in substation automation presented earlier offer the option of
utilizing the TI-ADC concept in distribution substations, and this is explained in more
detail below.
3.3 Increasing sampling frequency at the substation level
3.3.1 General concept
The concept presented here is based on a synchronized sampling of all IEDs. But the
synchronization is not done in such a way that each ADC performs the measurement
at exactly the same time stamp. The synchronization is done by time shifting the
sampling in the IEDs by a fraction of the sample time.
If, say, five IEDs with a 2 kHz sampling frequency measure the same phase volt-
age, the IEDs are synchronized so that samples from the same time actually have a
time difference of 1 s / 2000 / 5 = 100 µs. This means that, over a period of one sec-
ond, one would get 10,000 samples, all of which have different time stamps, instead
of 10,000 samples from only 2000 different time stamps (5 separate samples from
each time stamp). The effect this has on the time synchronization of the IED will be
presented later on, in section 3.4.5. The method is not dependent on a continuous
external time synchronization signal (e.g. via GPS), but the requirement is that the
56
3.3. Increasing sampling frequency at the substation level
IEDs are locally synchronized to each other.

So in theory, the sampling frequency can be increased to the sampling frequency
of one IED times the number of IEDs in the substation. In practice, the case is not
quite that good, since the sample stream of each IED passes through an anti-alias
filter which reduces the bandwidth of the resulting sample stream. This aspect will
be discussed in more detail in the following sections.
The method, known as Time-Interleaved ADC, has been used in ADC design, but
only within one measurement unit. All the ADCs have been located on the same PCB
(Printed Circuit Board), and thus, viewed from outside the combination appears as
one single ADC. Currently the applications of the concept have focused on telecom-
munications and software radio [Soudan and Farrell, 2008]. The concept has not so
far been used in this context, i.e. within a substation and between separate IEDs.
Other methods for artificially increasing the sampling frequency by using multiple
lower frequency measurements are presented later on, in section 3.5.
3.3.2 Voltage measurements
Voltage measurements are simpler since the same voltage is measured in each bay.
As the same quantity is available in each IED, the samples can simply be merged
as shown in Figure 3.3, which presents an example substation with five feeders, one
incoming and four outgoing.
Currently, the most common case is that each feeder only measures the neutral
voltage (for directional protection functions), so only the sampling frequency of that
quantity can be increased. But it is more than likely that full three-phase voltage
measurements in each bay will become more common, as new cheaper sensor tech-
nologies emerge and as the scope of DG increases. When there is DG in the network,
it may also feed a fault current and maintain line voltage even after a circuit breaker
has been opened (in fact, this functionality may even be desired during an intended
island operation). Therefore, measuring phase voltages behind the circuit breaker
(CB) of each feeder may soon be an actual requirement.
3.3.3 Current measurements
The same procedure is not possible with current measurements, since each current
is only measured once. However, as long as measurements are available from each
feeder, this limitation can be partly avoided using Kirchhoff’s current law.
57
3.4. Limitations and error analysis
U (V)
t0 t1 t2 t3 t4 t (s)
Feeder 4
Incoming Bay Measurement
Measurement
Feeder 1 Feeder 3
Measurement Measurement
Feeder 2
Measurement
Figure 3.3: Voltage measurements combined from 5 different measurements.
The method tested here uses linear interpolation for calculating virtual measure-
ment points between actual measurements. With this procedure, there is one actual
physical measurement and several virtual interpolated ones from each time stamp. If
the sum of all the samples is zero at any given instant, the residual error caused by
the interpolation can be added to the interpolated values. The procedure is presented
in Figure 3.4, in a substation with five feeders: one incoming and four outgoing.
3.4 Limitations and error analysis
Even if, in theory, 5 IEDs with a 2 kHz sampling frequency could produce a 10 kHz
sample stream, in practice it is not that good. Each measurement chain, even with
identical IEDs and sensors, has a unique measurement error, which is hard to fully
eliminate before the signal is used by the applications. If, for example, one IED
has a small scaling error or offset error in the measurement, it creates an additional
58
I (A)
Incoming Bay
Measurement
t0 t1 t2 t3 t4 t (s)
I (A)
Feeder 1
Measurement
For every time stamp
t (s) one physical
I (A)
measurement and
Feeder 2
several interpolated
Measurement
ones
t (s)
I (A) Sum of all
Feeder 3 measurements should
Measurement be zero, add residual to
all interpolated values
t (s)
I (A)
Feeder 4
Measurement
t (s)
Figure 3.4: Current measurements combined from 5 different measurements.
frequency component equal to the sampling frequency (2 kHz in this example), and
if this can’t be compensated for, it needs to be filtered out with a low-pass filter.
Normally, the sample stream also passes through an anti-aliasing filter in the IED
before it is transmitted. With 2 kHz sampling, the cut-off frequency of the filter is
less than 1 kHz, which means that frequencies higher than that cannot be reproduced,
even if multiple sample streams are available. For the transient analysis this is not
a major issue, as the frequency of the dominant charge transient is normally less
than 1 kHz. The problem with transient-based methods is not the Nyquist-Shannon
59
theorem, but the small number of data points. If it were possible to receive samples
from the IED before anti-aliasing, or after a separate anti-aliasing filter with a higher
cut-off frequency, then in theory, the frequency band of the signal could be increased,
but this has not yet been tested.
For a better understanding of these limitations, calculations need to be done to
assess the effect of different mismatches. The following error analysis is largely
based on [Vogel, 2005], and the equations below use the following notation:
A = amplitude of the input signal
ω0 = angular frequency of the input signal, discrete time
Ω0 = angular frequency of the input signal, continuous time
M = number of ADCs/IEDs
µg = expected value of the gain of one ADC/IED
µo = expected value of the offset of one ADC/IED
σg = standard deviation from the expected gain
σr = standard relative timing deviation of the TI-ADC set-up
σo = standard deviation of the offset
The different measurement ADCs/IEDs are assumed to be of same type, which
means the same design of the measurement electronics, and the same anti-alias filters.
The mismatches are indicated with σg , σr and σo , and they can be caused either by
differences in the anti-alias filter, differences in the other measurement electronics,
or both.
3.4.1 TI-ADC model
The calculations were performed based on the simplified model shown in Figure 3.5,
which describes one channel, l, in a multi-channel TI-ADC. In Figure 3.5, the input
signal x(t) is time-shifted by ∆tl , amplified by gl , added to an offset ol and sampled
with a sampling period of M Ts and a constant time shift of lTs .
With this model, the sampled output of a one-channel ADC is as shown in (3.1),
where δ(t) is the Dirac delta function.
∞
X
yl (t) = (gl x(t − ∆tl ) + ol )δ(t − kM Ts − lTs ) (3.1)
k=−∞
If these parameters were identical for all channels, no mismatch errors would
occur. In practice this can never happen, and for further analysis the Fourier transform
60
Figure 3.5: Model of a one-channel ADC [Vogel, 2005].
of (3.1) for a sinusoidal input signal x(t) = Asin(Ω0 t) and with ω0 = Ω0 Ts , is

presented in (3.2) and (3.3).
∞
"
2π X ΩS
Y (jΩ) = α[k]δ Ω − Ω0 − k
Ts M
k=−∞

∗ ΩS
− α [M − k]δ Ω + Ω0 − k (3.2)
M
#
ΩS
+ β[k]δ Ω − k
M
where
M −1
A X 2π
α[k] = gl e−jω0 rl e−jkl M
j2M
l=0
(3.3)
M −1
1 X
−jkl 2π
β[k] = ol e M
M
l=0
The symbol ∗ in (3.2) denotes the complex conjugate, and ΩS indicates the overall
sampling frequency of the setup. When looking at points where δ(t) = 0, it is
noticeable that some additional spurious peaks are centered at (±Ω0 + kΩS /M ) in
the case of gain and timing mismatches, whereas others are centered at (kΩS /M )
in the case of offset mismatches. Both the coefficient sets α[k] and β[k] are discrete
Fourier transforms (DFTs) of their corresponding channel parameters [Vogel, 2005].
61
3.4.2 Derivation of the explicit SINAD

This section shows the formula for the SINAD, when the explicit values for the
gain/offset/timing errors are known.
SINAD is as shown in (3.4).
PSgr

PS
SIN AD = 10 log10 = 10 log10 gr (3.4)
PN PN + PNo
In (3.4) the following notation is used:
PSgr = Signal power, dependent on gain and timing deviation
PNgr = Signal error power, dependent on gain and timing deviation
PNo = Signal error power, dependent on offset
The sub-calculations of this equation are shown in (3.5):
M −1 2
A2 1 X
PSgr = 2|α[0]|2 = gl e−jω0 rl

2 M

l=0
M −1 M −1
X 1 X
PNo = |β[k]|2 = |ol |2 = |o|2
M
k=0 l=0
M −1
PNgr = 2
X
|α[k]|2 − 2|α[0]|2
k=0
A2 MX−1 M −1
A2 X 2 A2 2
= |gl e−jω0 rl |2 − 2|α[0]|2 = gl − 2|α[0]|2 = g − 2|α[0]|2
2M 2M 2
l=0 l=0
(3.5)
This leads to (3.6) [Vogel, 2005].
M −1 M −1
!
X X
SIN AD = 10 log 10 (2|α[0]|2 ) − 10 log 10 2 |α[k]|2 + |β[k]|2
k=1 k=0
! (3.6)
PSgr
= 10 log 10 A2 2 gr
2 g − PS + PNo
3.4.3 Derivation of the expected SINAD

This section shows the formula for the SINAD, when in stead of explicit values, the
probability distribution of these same errors is known.
62
SINAD is as shown in (3.7), where E describes the expected value of the random
variable and tilde (.̃) the random variable itself.
!
E{PSgrR }

E{PS }
SIN AD = 10 log10 = 10 log10
E{PN } E{PNgrR } + E{PNoR }
! (3.7)
E{PSgrR }
= 10 log10
E{ A2 g̃ 2 − PSgrR } + E{PNoR }
2
In (3.7) the following notation is used:

PSgrR = Expected signal power, dependent on gain and timing deviation
PNgrR = Expected error power, dependent on gain and timing deviation
PNoR = Expected error power, dependent on offset
The sub-calculations of (3.7) are shown in (3.8).
2
A2 1 M −1
E{PSgrR }
X
= g˜l e−jω0 r˜l

2 M

l=0
XM
M −1 −1
A2 X
= E{g˜l g˜m e−jω0 r˜l ejω0 r˜m }
2M 2
l=0 m=0
A2 (3.8)
µ2g ((M − 1)|Φr (ω0 )|2 + 1) + σg2

=
2M
A2 2 A2
E{ g̃ − PSgrR } = (M − 1) µ2g (1 − |Φr (ω0 )|2 ) + σg2

2 2M
M −1
1 X
E{PNoR } = E{|õl |2 } = (µ2o + σo2 )
M
l=0
In (3.8), Φr (ω0 ) is the characteristic function of the random variable r̃, and if
Gaussian distribution is expected, it is as shown in (3.9).
Z 2
jω0 x 2 2
2
dx = e−ω0 σr

|Φr (ω0 )| = fr̃ (x)e
(3.9)
Then the overall SINAD is as shown in (3.10), when the expected value of offset
µo is zero [Vogel, 2005].
63
A2 2

−ω02 σr2 2
SIN AD = 10 log 10 (µ ((M − 1)e + 1) + σg )
2M g
2 (3.10)
A 2 2
− 10 log 10 (M − 1)(µ2g (1 − e−ω0 σr ) + σg2 ) + σo2
2M
3.4.4 SINAD values for the substation

(3.10) was initially constructed for ADC design, where the accuracy requirements
are considered from the point of view of a TI-ADC set-up, rather than for individ-
ual ADCs. As in substation automation the individual ADC of an IED still plays an
important role, modifying the equation in that direction makes the results more de-
scriptive. The standard relative timing deviation in (3.10) is presented as a fraction
of the sample time of the TI-ADC set-up. A more descriptive approach is to use the
fraction of the sample time of an individual ADC, see (3.11).
σr = M σr1 (3.11)
where σr1 indicates the standard relative timing deviation of an individual ADC.
For example, a 1 µs deviation with a sample frequency of 1 kHz per device (sample
time 1 ms) means a value = 0.001, regardless of the number of ADCs used. Fur-
thermore, when investigating the effect of multiple measurements, instead of angular
frequency ω0 , it is better to use a frequency relative to the sampling frequency of one
IED, see (3.12).
a0
ω0 = 2π (3.12)
M
In (3.12) a0 indicates the input signal frequency relative to the sampling frequency
of one IED, e.g. a 1 kHz signal with a 2 kHz sampling frequency means a0 = 0.5.
With these additions SINAD can be calculated with (3.13).
A2 2 2 2 2
SIN AD = 10 log 10 ( (µg ((M − 1)e−4π a0 σr1 + 1) + σg2 ))
2M (3.13)
A2 2 2 2
−10 log10 ( (M − 1)(µ2g (1 − e−4π a0 σr1 ) + σg2 ) + σo2 )
2M
The effect of individual components can be derived from (3.13) by setting the
other deviations to zero, for example, the effect of the timing deviation is clearly
64
visible when both the gain and the offset deviation are set to zero. Such formulas,
simplified from (3.10), are presented in [Vogel, 2005], and the same simplifications
can be made to (3.13), the graphical results of which are presented in Figures 3.6, 3.7
and 3.8 . In these figures, M = 8, µg = 1 and a0 = 0.5.
The effect of gain error (offset and timing error zeros)

110
100
90
80
SINAD (dB)
70
60
50
40
30
20
−3 −2 −1 0 1
10 10 10 10 10
Error in gain (%)
Figure 3.6: The effect of gain error in SINAD.
Figures 3.6, 3.7 and 3.8 provide the maximum possible SINAD, if the error in one
element is known. If the timing error were 1% of the sample time of one IED, the
maximum possible SINAD as shown in Figure 3.8 is around 30 dB, no matter how
identical the measurement chains are in other respects. In the measurement chain of
an IED, the offset error is normally negligible, and its effect can be ignored. The most
interesting issue, therefore, is the combined effect of gain and timing mismatches, and
these are presented in Figures 3.9 and 3.10. As before, M = 8, µg = 1 and a0 = 0.5 in
the figures.
The combined effect is clearly visible in Figure 3.10. If, for example, the gain
accuracy in the IEDs is 0.5%, the best possible SINAD value in the TI-ADC set-up
is around 45 dB, and requires timing accuracy to be 0.1%. With a 2 kHz sampling
frequency per IED this means around 0.5 µs accuracy.
Another view of the accuracy of the TI-ADC concept is to look at the percentage
65
The effect of offset error (gain and timing error zeros)

100
90
80
70
SINAD (dB)
60
50
40
30
20
10
−3 −2 −1 0 1
10 10 10 10 10
Error in offset (%)
Figure 3.7: The effect of offset error in SINAD.
The effect of timing error (offset and gain error zeros)

100
90
80
70
SINAD (dB)
60
50
40
30
20
10
−3 −2 −1 0 1
10 10 10 10 10
Error in g (%)
Figure 3.8: The effect of timing error in SINAD.
66
SINAD with fixed amount of IEDs and measurement signal
80
70
60
SINAD
50
40
−5
10
30
−5
10 −4
10
−4
10
−3
−3 10
10
−2
gain dviation
−2
10 10
time deviation
Figure 3.9: The combined effect of gain and timing mismatches.
SINAD contours with fixed amount of IEDs and measurement signal
40 40 40
45
45
10
−3 50
45
55 50
Time deviation
60
65
−4
10
−4 −3
10 10
Gain deviation
Figure 3.10: Contours of Figure 3.9.
67
value of the Total Harmonic Distortion with Noise (THD+N), which is often viewed
in stead of SINAD. Similar contours as in Figure 3.10 are shown in Figure 3.11.
THD+N contours (in percent) with fixed amount of IEDs and measurement signal
1 1 1
0.7 0.7 1
0.5 0.7
0.5
−1
10 0.3
Time deviation / percent
0.2
0.15 0.3
0.
5
0.1
−2
10
−3
10
−3 −2 −1
10 10 10
Gain deviation / percent
Figure 3.11: Contours of Figure 3.9 in THD+N.
One example from an IED manufacturer shows, that the measurement accuracy of
the IEDs is around 0.5% ABB Ltd. [2009], which is in the same range as is achievable
with the TI-ADC set-up as shown in Figure 3.11. One further issue is the effect of
additional IEDs. What happens to the accuracy when the number of IEDs increases,
as shown in Figure 3.12? In this figure, σg = 0.005, σr1 = 0.001, σo = 0 and µg = 1.
SINAD does deterioriate slightly when the number of ADCs increases, but not
greatly. This indicates that the signal quality would remain at the same level, which
is the most important issue with transient-based algorithms. Power system transients
last only for a few milliseconds, e.g. during an earth fault [Lehtonen and Hakola,
1996], so the main point is that the number of data points can be increased without
having any major effect on the SINAD value.
3.4.5 Time synchronization requirements

According to the calculations presented above, the timing mismatch of samples should
not exceed 0.1% of the sample time, as long as the gain accuracy is around 0.5%,
68
SINAD with fixed gain and timing deviations
50
49
48
47
SINAD
46
45
44
43
0
42
0 5
0.2
10
0.4
0.6 15
0.8
1 20 Amount of ADC
input signal frequency
Figure 3.12: The effect on SINAD of increasing the number of ADCs.
which is the measurement accuracy stated in the IED manufacturers manuals [ABB
Ltd., 2009]. Interestingly, other research based on different assumptions came to the
same conclusion of a 0.1% timing accuracy requirement. If the individual ADC has
16 bit sampling, timing accuracies less than 0.1% degrade the SINAD considerably
[Soudan and Farrell, 2008]. With a sampling frequency of 2 kHz the sample time is
0.5 ms, which provides time accuracy requirements of 0.5 µs with 0.1% accuracy. As
indicated earlier in this thesis, the IEEE 1588 time synchronization protocol fulfills
these requirements, so it can be used for the synchronization of IEDs. In addition,
there are the two following requirements:
• The IED synchronization must also affect ADC sampling times.
• The IED must have an adjustable time offset for ADC synchronization. When
the ADC of one IED is running in sync with the synchronization master (offset
= 0), the other IEDs must have an offset time with multiples of (1/M/fs )
where M is the number of IEDs and fs is the sampling frequency. For example,
if 5 IEDs with 2 kHz sampling frequency are synchronized from the same
synchronization master, one IED must have a timing offset of 1 / 5 / 2000 =
100 µs, and the other offset times are 200 µs, 300 µs and 400 µs.
69
3.5. Other existing solutions for increasing the sampling frequency
3.4.6 Special considerations related to TI-ADC
Many research results related to TI-ADC can be directly applied to the concept pro-
posed in this thesis. A few special items related to substation automation, however,
must be given special consideration. Firstly, the number of ADCs is expected to stay
constant in a TI-ADC set-up, which might not be the case in a substation. In particu-
lar, if the measurement sensor is behind a circuit breaker, a change in network topol-
ogy, e.g. due to a fault, can remove one IED from the measurement chain. Because
of this, centralized logics must be implemented in the substation, which would take
this factor into account. If one IED measurement is missing, new timing offset values
must be calculated and provided to the IEDs, in order for the TI-ADC arrangement
to be able to resume operation with (M-1) measurements. As indicated earlier, this
does not much affect the SINAD of the measurement signal, but the number of data
points for transient analysis is reduced. It should also be noted that if a functionality
were to require measurements from the same moment (e.g. comparing phasor values
from the measurements from different feeder bays), the calculated timing offset value
must also be provided to these functions in order for them to operate properly.
Another consideration is that the existing compensation strategies for TI-ADC
mismatches focus on analyzing the combined high-frequency signal, e.g. an offset
mismatch is visible in the signal as a signal component, which has the same frequency
as the sampling frequency of one ADC [Vogel, 2005]. In a substation, however, the
measurement signal of one IED is needed for protection purposes, and thus it has al-
ready undergone extensive signal processing, such as calculating nominal frequency
50Hz or 60Hz components and higher order power system harmonics [ABB Ltd.,
2009]. The gain mismatch can be eliminated through comparison of the fundamental
frequency phasors of the different IEDs. The same elimination method is also useful
in other respects, as it simultaneously provides a new way to supervise measurement
circuits, which is an important issue for the secondary system of a substation. This
aspect is a topic for future research, but is not investigated further in this thesis.
3.5 Other existing solutions for increasing the sampling fre-

quency
Different sparse sampling methods have also been investigated, especially for power
quality monitoring. The need for acquiring information from higher order harmonics
70
3.6. Chapter Summary
with cheap equipment is a common one. The doctoral thesis of Pekka Koponen
[Koponen, 2002] addresses this issue too, and proposes multiple sampling, but with
different sampling frequencies of the same signal (e.g. 2 kHz and 3 kHz).
Using the same sampling frequency with shifted time intervals seems to be a con-
cept which has not been tested earlier. One reason for this is that processing several
sample streams simultaneously in a substation was not feasible earlier. On the other
hand, a similar concept would also work offline with normal bay-level IEDs, by using
disturbance records downloaded after faults have occurred.
3.6 Chapter Summary

The concept presented in this thesis makes it possible to increase the sampling fre-
quency of current and voltage measurements at the substation level. When the tran-
sient frequency is below the cut-off frequency of the anti-aliasing frequency of one
IED, combining the measurements from the IEDs of several bays increases the num-
ber of measurement data points from the transient period.
The accuracy of time synchronization has a significant impact on the overall mea-
surement accuracy in the proposed method. Furthermore, particularly with transients,
the analogue front end of each IED must have the same characteristics and similar
frequency response. However, these factors shouldn’t limit the use of the method, as
the IEEE 1588 time synchronization specifies a time accuracy which is sufficient for
the method presented in this thesis. In addition, the need to calibrate the measure-
ment chain of each IED as accurately as possible is not confined to the use of this
method.
This concept is further tested with an example functionality for transient-based
earth fault location in Chapter 5. Testing the method further with other transient-
based methods, such as intermittent earth fault detection or power quality measure-
ments, is a topic for future research.
71
Chapter 4
Station Applications
As was pointed out in Chapter 2, the proposed architecture allows the functionality
to be allocated in a new way. Not all the functions which currently reside in the bay-
level protection and control IEDs need be located there any longer, since the proposed
concept enables certain functions to be implemented at the station level. This chapter
evaluates how this functionality should be re-allocated, i.e. which functions should
remain at the bay level and which functions should be moved to the station level. An
example case study, targeted at Finnish electricity distribution networks, is used as
the basis for this evaluation because the functionality needed in a substation is always
case-specific.
4.1 Functionality in the secondary system of a distribution

substation
The first issue to consider is what set of functions needs to be re-allocated in this case.
How are they to be selected? With the existing functionality, the issue is easier, as
one can simply collect a representative set of various functions based on the product
guides of the IED vendors, and make the selections from there.
However, with new functions not yet available in the IEDs, the process is more
difficult, as there is no definitive list of the future functions. Although functional
areas may have already been identified, as discussed in section 2.3, the names and
number of the particular functions might not yet be known. Nevertheless, the items
which were regarded as most important were selected and added to the function set.
The list of functions to be re-allocated in this chapter is based on the LN (Logical
72
4.1. Functionality in the secondary system of a distribution substation
Node) classes defined in IEC 61850-7-4 [IEC, 2005], as this encompasses all the main
function classes in modern IEDs. Only the Protection (P), Protection related (R), and
Control (C) categories of the LN classes were selected, as these are the main point of
interest. Some changes to the LN list were made based on a review of the results from
the IED vendors, and some LNs were removed because they were redundant (e.g.
the Directional Element LN PDIR was removed as the protection logic is normally
modeled with PTOC, and nearly all the LNs related to the disturbance recorder were
removed as RDRE LN already describes the main functionality). In addition, the
PTOC LN used for many different protection functions was split into several different
functions. The derived list was extended with currently ongoing research topics, and
is summarized in Table 4.1. Furthermore, the related ANSI numbering according to
[IEEE, 2008] is presented when applicable.
Table 4.1: Functions selected for re-allocation

Abbr ANSI Function
PDIF 87 Differential protection (busbar)
PDIS 21 Distance protection (first zone)
PHIZ 64 Ground detector, high-impedance earth fault protection
PIOC 50 Instantaneous overcurrent protection
PSDE 67N Sensitive directional earth fault protection (intermittent earth fault)
PTEF 67N Transient earth fault protection
PTOC 51 Time overcurrent protection, overcurrent
EFPTOC 51 Time overcurrent protection, non-directional earth fault (double earth
fault)
DEFPTOC 51 Time overcurrent protection, directional earth fault
DISPTOC 51 Time overcurrent protection, phase discontinuity
PTOF 81 Overfrequency protection
PTOV 59 Overvoltage protection
PTTR 49 Thermal overload
PTUC 37 Undercurrent protection
PTUV 27 Undervoltage protection
PTUF 81 Underfrequency protection
RDRE Disturbance recorder
RBRF 50BF Breaker failure
RFLO Fault locator
RREC 79 Auto-reclosing
continued on the next page
73
4.2. Functionality division
continued from the previous page

Abbr ANSI Function
CILO 3 Interlocking
CBCSWI Switch controller (CB)
DISCSWI Switch controller (DIS)
flir Post-fault network restoration, and dynamic reconfiguration of network
topology, self-healing networks (IEC61850: separate logical devices
including several control LNs)
cbm Condition monitoring, Asset Management, CBM/RCM (IEC61850: ad-
ditions made to IEC 61850 after Edition 1, e.g. CB condition monitor-
ing LN SCBR)
dgoper island operation, LOM, FRT (IEC61850: separate logical devices in-
cluding several control LNs)
loadshed load-shedding (IEC61850: separate logical devices including several
control LNs)
adapt Automatic recalculation of protection parameters based on topology
and DG changes, adaptation of protection (IEC61850: Generic process
control LN defined, GAPC)
selfsup Station-level self-supervision (IEC61850: IED-level self-supervision
modeled with LN SSYS)
cybersec Cyber-security (IEC61850: Generic security LN defined, GSAL)
reporting Statistics, regulation reports (IEC61850: Simple statistics report LN
available, MSTA. More extensive reports not available)
The abbreviations in Table 4.1 are LN class names with their functionality defined
in the IEC 61850, in capital letters. To separate different PTOC functions from each
other, different prefixes are used for different functions. Ongoing research topics not
accurately defined in IEC 61850 are abbreviated with lowercase letters for ease of
identification. In these cases, possible ways of modeling these applications in IEC
61850 are shown in the table in parentheses. These abbreviations are used later on in
this chapter when summarizing the results.
4.2 Functionality division
Function re-allocation is approached from two different perspectives. The first divi-
sion is based on importance. First, there are the mandatory functions, such as primary
protection and control. These functions are critical for the operation of the network,
and are therefore normally always present when a suitable protection device is avail-
able. These functions also normally need back-up functionality, in case the primary
74
4.2. Functionality division
function fails to operate.
At other end of this ’importance’ axis are the optional functions, such as mon-
itoring and analysis applications. They are not vital for the safe operation of the
network, but they affect its cost-efficiency. If these functions fail, fault situations
might last longer (e.g. through incorrect fault location) or they might occur more fre-
quently (e.g. due to insufficient condition monitoring resulting in flawed estimates of
maintenance needs). Inadequate monitoring functionality can also impact on profits
due to non-optimal utilization of the network resources. Normally, these functions
do not need to be backed up.
The other division is based on the location of the function. In the evaluated archi-
tecture, functions can be located either at the unit level, or the station level. In the
currently dominant secondary architecture only unit-level devices are available. On
the other hand, in a totally centralized solution there is only one location available -
the station level. The need to position functions on the station-unit axis only exists in
the combined approach investigated in this thesis.
These two ways of dividing the functions create four different categories, which
are presented below in Table 4.2. The four categories have different requirements –
and different execution environments are suitable for different categories. In general,
unit-level functions should have longer life cycles than station-level functions, and
mandatory functions place stricter requirements on the operating system running the
functions. There are more unit-level devices, and they are more closely connected to
the electricity distribution process, which causes longer and more costly maintenance
breaks.
Table 4.2: Functional categories

Unit level Station level
Optional functions Unit level optional Station level optional
Mandatory functions Unit level mandatory Station level mandatory
Careful attention should be paid to these categories when designing the functional
content of the secondary system of a distribution substation. If a function is imple-
mented in the wrong category/environment it can impose updates to the automation
equipment earlier than would otherwise be needed.
75
4.3. Functionality division criteria
4.3 Functionality division criteria
Finding an optimal set-up for distribution automation is a complex issue. Even when
the scope of the functionality is limited to the distribution substations, it is still a dif-
ficult problem. The criteria generally regarded as being most important are presented
below.
4.3.1 Communication requirements
The most obvious indication of station-level functionality is the communication re-

quirement. If the functionality requires horizontal and/or vertical communication, in
other words, if information needs to be exchanged between several units, it is best
to implement the functionality at the station level. Whatever the case, the function
will not operate properly if there are communication problems, so implementing the
function at the station level will not make the system less reliable that it would be if
this functionality were to be implemented at the unit level.
In general, data should be processed at the lowest level possible so that fast and
reliable response can be guaranteed. This ’lowest level’ depends on the functional
requirements and the available computational and communication capabilities. An
apposite example of this approach can be found in the human nervous system. When
fast response is needed (e.g. when a human touches a hot surface), an immediate
command is issued at the spinal cord. Sending the signal from the sensors to the
brain and processing it there would create unnecessary delay when the appropriate
reaction is unambiguous (i.e. to remove the hand from the hot surface). When more
information is needed from other senses, (e.g. the surface is not very hot and it
is better to look at it first), the information needs to be gathered in the brain and
processed there.
The functions listed in Table 4.1 were evaluated based on the communication
requirements. The method selected for this case evaluation was to use the number of
nodes communicating with each other. For ’all nodes in the substation’ a value of 10
was used, because this represents the average size of a Finnish substation (10 feeders).
If several different solutions exist for the function, (both stand-alone algorithms as
well as algorithms requiring communication), an average value of 5 was used.
76
4.3. Functionality division criteria
4.3.2 Response time
The desired response time for the function indicates whether the function should re-
side at the unit level or the station level. In general, the faster the function should
operate, the closer it should be to the process. Another way to express this is to say
that functionality can be centralized to the extent that communication delays do not
add too much latency and uncertainty to the operation. Functions requiring shorter
response times are more often unit-level mandatory functions, whereas those func-
tions which do not require such a short response time, or have no response time
requirement, point toward either optional or station-level functionality, or both. The
distinction between ’short’ and ’long’ response times is application-specific.
The time limits for the response of the different functions can be gathered from
research papers or from the utility requirements. For this thesis, information was
gathered by interviewing protection experts from various Finnish utilities (Fortum,
Vattenfall and Helsinki Energy) and asking for their default operation time settings
for various functions. The average default operation times gathered from these inter-
views were used to define the expected speed of operation.
4.3.3 Utilization frequency
Utilization frequency means how often these functions are used in the real-time op-
eration of the distribution network. One approach to measuring utilization frequency
is to use the statistics gathered from disturbances in the power system. When a
particular fault situation is common in a distribution network, protection against it
becomes essential and makes the function mandatory. According to the outage statis-
tics for Finnish distribution networks [Energiateollisuus, 2005-2010], over 80% of
the faults in electricity distribution networks occur in medium voltage networks. Of
these faults, nearly 50% are short-circuit faults, and nearly 40% are earth faults.
Another approach to defining the utilization frequency is to the refer to technical
specifications of the IEDs, and this was the approach selected in this thesis. Those
functions which are almost always present in protection and control IEDs are manda-
tory, whereas those functions which are only included in high-end IEDs are usually
optional. To this end, a total of 124 different feeder protection IEDs from a num-
ber of different vendors (ABB, Alstom, Arcteq, Areva, Basler Electric, CEE, Cooper
Industries, General Electric, Mikro, Nari-relays Electric Co., Protecta, Reyrolle, Ro-
con, SEL, Schneider Electric, Siemens, Sprecher, Thytronic, Toshiba, VAMP and
77
4.4. Functionality division method
Woodward) were evaluated in terms of their available functions.
4.3.4 Function immaturity
Function immaturity indicates how often a function can be expected to be upgraded.

If the function is stable and is expected to have a long life cycle, it is reasonable
to locate it at the unit level where updates are more costly than at the station level.
If, on the other hand, there is extensive research going on in that ares, or changes
are expected in the requirements for the function, either through legislation or from
the business environment, the function should be located at the station level, where
updating is easier.
The method used in this case was to categorize the functions according to the
technology generation from which they were first introduced, i.e. from electro-
mechanical relays, static relays or numerical relays. The source of information used
here was the historical timeline of the relays from ABB, derived from [Lundqvist,
2010]. This described the milestones of one relay manufacturer and gives an over-
all view of the progress of the industry, showing the decade in which a particular
functionality was first introduced. In addition, functions proposed in recent research
articles were allocated to this decade to illustrate the fact that development is still
going on.
4.4 Functionality division method
It is difficult to express the four above-mentioned criteria accurately in a numerical

form. Furthermore, the aim of the exercise, i.e. the re-allocation of the available
functions, cannot be explicitly defined, as in practice their implementation always
depends on the requirements of the substation in question. For these reasons, the
tool most appropriate for such an analysis is fuzzy logic [Klir and Yuan, 1995]. This
provides a means for handling inaccurate or ambiguous data and using them for fur-
ther analysis. First, the fuzzy logic rule-set for function allocation had to be defined,
which is described below.
• If the function requires communication, does not have strict response-time re-
quirements and is not very mature, it belongs to the station-level functions.
Otherwise, it belongs to the unit-level functions
78
• If the function has strict response-time requirements and high utilization fre-
quency, it is a mandatory function. Otherwise it is an optional function.
After defining the rule-set, the required data must be put into a numerical form.
The methods used were those mentioned above, in section 4.3, and they are summa-
rized below:
• Communication: number of communicating nodes
• Response time: average response times used by Finnish utilities
• Utilization frequency: number of evaluated relays having the functionality
• Function maturity: decade in which the functionality was first introduced
Of course, there are other ways for defining numerical values for the defined cri-
teria. However, these were the methods found suitable for this case study, which
focuses on Finnish electricity distribution networks, and the resulting numerical val-
ues are presented in Table 4.3.
Function Communication Response time Utilization frequency Function maturity

PDIF 10 13ms 7 1940
PDIS 1 20ms 10 1940
PHIZ 5 55s 16 1960
PIOC 1 133ms 120 1900
PSDE 5 2.5s 24 1940
PTEF 1 270ms 13 1970
PTOC 1 700ms 116 1910
PTOF 1 5s 55 1920
EFPTOC 5 300ms 124 1910
DEFPTOC 5 400ms 68 1920
DISPTOC 5 55s 44 1960
PTOV 1 3.25s 57 1920
PTTR 1 30s 73 1910
PTUC 1 10s 36 1920
PTUF 1 5s 55 1920
PTUV 1 6s 61 1920
RDRE 5 60s 103 1980
RBRF 1 100ms 103 1960
continued on the next page
79
continued from the previous page

Function Communication Response time Utilization frequency Function maturity
RFLO 5 60s 29 1980
RREC 1 300ms 73 1920
CILO 10 1s 56 1920
CBCSWI 1 50ms 94 1900
DISCSWI 1 1s 13 1900
flir 10 60s 1 2000
cbm 5 15min 59 2000
dgoper 10 700ms 1 2000
loadshed 10 5s 7 2000
adapt 10 15min 1 2010
selfsup 5 700ms 10 2000
cybersec 1 700ms 30 2000
reporting 5 15min 10 2010
Table 4.3: Numeric values for each function
It was not possible to directly calculate the average response times for the se-
lected active research topics (’flir’, ’cbm’, ’dgoper’, ’loadshed’, ’adapt’, ’selfsup’,
’cybersec’ and ’reporting’ in Table 4.3), since their use in Finnish utilities is cur-
rently rare. Instead, the related response times from existing functions were used.
The research topics abbreviated to ’dgoper’, ’selfsup’ and ’cybersec’ affect the pri-
mary, time-critical protection functionality, so in these cases the response time for
PTOC was used. A power imbalance situation requiring load-shedding functional-
ity is normally detected from frequency variations, so in this case the response time
of PTUF was used. With functions affecting fault management (RREC, RFLO and
flir) the total operating time of the auto-recloser was used (normally, in Finland, two
shots are used with the auto-recloser; the average time for the second shot being 60
seconds). With the other reporting and monitoring related functions, an average time
of 15min was used, which is currently the granularity in the nordic energy market.
These numeric values cannot be processed as raw figures. Instead,they need to be
normalized to a value between 0 and 1. In fuzzy logic, this process is normally called
fuzzification. This normalization allows different criteria to be combined and facili-
tates the unit/station and mandatory/optional categorization targeted in this research.
Logarithmic values were used with the response times, in order to better differentiate
between very short and very long response times. After normalization, the source
data was categorized as shown in Table 4.4.
80
Table 4.4: Normalization rules applied to different criteria

Criteria Value 0 Value 1
Local measurements are Communicating with all nodes
Communication
enough in the substation
Response time Shortest of all Longest of all
Utilization frequency Not present in any IED Present in all IEDs
Used in relays since beginning New function, under active de-
Function immaturity
of 1900 velopment
The resulting membership values of station-level functions and optional functions

are presented in Figure 4.1.
All Functions
1
adapt
reporting
0.9
flir
PHIZ
0.8
RFLO cbm
loadshed
0.7 DISPTOC
dgoper
DISCSWI PTUC
PSDE selfsup
0.6
PTEF
Optional function
PTOF cybersec
PTUF PTTR
PTUV
PTOV
0.5
PDIS PDIF CILO RDRE
0.4
DEFPTOC
RREC
0.3
PTOC
0.2
CBCSWI RBRF
EFPTOC
PIOC
0.1 Unit level mandatory functions
Station level mandatory functions
Unit level optional functions
Station level optional functions
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Station level implementation beneficial
Figure 4.1: Function re-allocation results.
In Figure 4.1, the X-axis presents the membership of a station-level function

group. A value of 100% (value 1.0) means that the function clearly belongs to the
station level, and a value of 0% (value 0.0) means that the function clearly belongs to
81
the unit level. Similarly, the Y-axis presents the membership of Optional functions;
100% indicates an optional function and 0% a mandatory function. An average of the
individual criteria values was used as the AND-operator of the rule-set. The figure
shows that there is a clear correlation between the two viewpoints. i.e station-level
functions are more often also optional functions.
A fuzzy c-means clustering method [Miyamoto et al., 2008] was used to identify
the center points for each of the four main categories used in this example. Having
selected the number of clusters to be four, the identified cluster centers with their
related effect-areas in the defined xy-plane are illustrated in Figure 4.2.
Unit level mandatory functions Station level mandatory functions

1 1
0.8 0.8
Optional function
Optional function
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Station level implementation beneficial Station level implementation beneficial
Unit level optional functions Station level optional functions

1 1
0.8 0.8
Optional function
Optional function
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Station level implementation beneficial Station level implementation beneficial
Figure 4.2: Membership of different functions to different clusters.
The final step in this fuzzy logic operation is the defuzzification - allocating func-
tions to one and only one category based on their fuzzy membership values. This was
done by calculating the degree of membership that each function had to each cate-
gory, and selecting the category in which the membership had the highest value. The
derived results are also presented in Figure 4.1, where a different marker represents a
different category.
82
4.5. Description of the results
It must be noted that the clusters for ’Unit-level optional’ and ’Station-level manda-
tory’ are very close to each other, so depending on the case and on the available
architecture, some functions can reside in either of those two categories. Another
way to look at the situation is that when only a small number of functions belonging
to either the ’Unit-level optional’ or ’Station-level mandatory’ categories are needed
in a particular substation, the most cost-efficient solution would be to allocate them
all to the bay level, and leave the station-level equipment out entirely. But when
the requirements for the substation increase and station-level secondary equipment
is needed, it could be beneficial to also include ’Unit-level optional’ functions in the
station level, and keep the unit level as simple as possible (and with as long a life
cycle as possible).
4.5 Description of the results

4.5.1 Unit-level mandatory functions
This category is the most traditional category in substations. The functionality in

this category should be selected so that the most important features of the unit are
secured at all stages. These functions should not rely on external communication, so
that safety is guaranteed even if communication is lost. For smaller stations with less
stringent requirements, this category might be the only one needed. This category
alone should fulfill the most critical requirements. The functions belonging to this
category are presented in Figure 4.1, and briefly described below.
• Protection
– Overcurrent protection, instantaneous and delayed (PIOC, PTOC)

– Earth fault protection: Non-directional (EFPTOC)
• Control
– Circuit breaker control and operation (CBCSWI)

– Auto-recloser (RREC)
• Supervision
– Breaker failure protection (RBRF)
83
4.5.2 Unit-level optional functions
Because optional functionality is rarely essential for network safety, this category is
not always needed at the unit level, although an extensive library of these functions
is available in modern protection and control IEDs. This includes functions which do
improve the protection by, for example making it more selective or more accurate, but
the network safety in general can also be guaranteed with only mandatory functions.
This may also include functions which are mandatory for a limited number of special
cases, but are not always mandatory.
The functions belonging to the unit-level optional functions category are presented
in Figure 4.1 and briefly described below.
• Protection
– Over/under frequency and voltage protection (PTOF, PTUF, PTOV, PTUV)

– Over/under power protection, thermal overload protections (PTTR)
– Earth fault protection: Directional (DEFPTOC)
– Transient earth fault protection (PTEF)
– Distance protection (PDIS)
– Undercurrent protection (PTUC)
• Control
– Disconnector operation (DISCSWI)
4.5.3 Station-level mandatory functions
Currently, this category is not normally present at all in substations. All such func-
tionality resides at the unit level (feeder bays, transformers, generators, etc.) and the
station-level equipment is only used as a gateway for accessing these unit-level IEDs.
The drawback with this approach is the update cycle, as has already been mentioned.
The purpose of this category is to complement the functions at the unit level,
rather than to replace them. When the unit-level protection can operate without the
station-level protection, station-level functionality can be updated without interrup-
tions. This will also extend the life cycle of the unit-level functions, as most of the
update measures are carried out at the station level.
84
The functions belonging to station-level mandatory functions are presented in Fig-

ure 4.1, and briefly described below.
• Protection
– Differential protection - bus bar (PDIF)

– Sensitive directional earth fault protection - intermittent faults (PSDE)
– Phase discontinuity protection (DISPTOC)
– Protection against faults with low fault current magnitude: e.g. high
impedance earth faults [Tengdin, 1996] [Nikander, 2002] (PHIZ)
– Islanding operation and Loss-of-Mains protection when islanding is not
allowed [Rintamäki and Kauhaniemi, 2009] (dgoper)
• Control
– Interlocking (CILO)
• Other
– Fault locator (RFLO)

– Disturbance recorder (RDRE)
– Cyber security aspects [Nartman et al., 2009] (cybersec)
– Station-level self-supervision (selfsup)
As noted earlier, this category is very close to the ’unit-level optional’ functions
category and in situations where both unit- and station-level devices are available
many protection functions from that category could also be allocated to this category.
4.5.4 Station-level optional functions

The functions belonging to station-level optional functions are presented in Figure
4.1, and consist mainly of functions which are not essential for the safety of the
substation, but important for increasing the efficiency of the network and fast reaction
to fault situations. This category also contains functions which are not yet mandatory,
but can be expected to be mandatory soon. These functions address issues which
are not yet common in the network (at least not in Finland), but which will gain in
importance if the future challenges outlined in section 2.3 become a reality. These
items are listed briefly below, and have been explained in more detail in section 2.3.
85
• Post-fault power restoration [Mekic et al., 2009] and self-healing networks in

general [Rasmunssen, 2009][Manner et al., 2011] (flir)
• Load shedding [Apostolov et al., 2007] (loadshed)
• Automatic recalculation of protection parameters based on topology and DG

changes, adaptation of protection [Oudalov et al., 2011] (adapt)
• Advanced condition monitoring and Asset management support, CBM/RCM

[Angel, 2003] (cbm)
• Reporting support (reporting)
4.6 Chapter summary

The functional division presented here focuses on increasing the life cycle of the
substation automation system by utilizing station-level data processing. Initially, the
station-level functionality criteria are defined. When the function requires commu-
nication, but does not have strict requirements for response times, station-level im-
plementation can be justified. Complex functionality requiring additional CPU per-
formance and anticipated updates in the near future are also clear indications for
station-level functionality. Based on the defined criteria, an example case of both
existing functionality and current research topics was divided between the unit level
and the station level.
The difficulty in the method is the numerical representation of the criteria, which
are also case-specific, depending on the substation and the network. Furthermore,
new algorithms can change the location of a particular function entirely (e.g. the in-
troduction of a new, more accurate method, which may need more communication
than traditional methods). Numerical values for, e.g. ’communication requirement’
or ’function maturity’ are very difficult to determine. For this reason, fuzzy logic
methods and fuzzy c-means clustering were used, which provide a suitable toolset for
handling these types of attributes. The focus of this chapter is more on the method
than on the final categorization. In this thesis, the method was applied to Finnish
electricity distribution substations. When similar categorization is needed for a par-
ticular substation and/or application area, a new list of functions can be defined with
new numeric values for the criteria. With this updated data, new categorizations can
be obtained using the method presented here.
86
A future addition to this method would be a definition of the financial benefit of

each function, such as the effect of the function on life-cycle costs. This could be
added as an additional criterion, with a rule ’if a function has high financial benefit,
it is mandatory’, but it could also be added as a third axis to the derived plane. This
would help in visualizing the implementation roadmap for adding new functionality
to a substation, highlighting those functions with the highest financial benefit, and
showing them in the appropriate category. The challenge in this is in evaluating the
financial benefit of each function. The life-cycle cost evaluations in section 2.5.5
have already shown the problem of insufficient statistics, the relevant different fault
cases, and their root causes. Therefore, a more detailed analysis of fault cases and
their root causes is also a future research topic. If the more detailed fault analysis
proposed in the summary part of Chapter 2 could be conducted, the same data could
be used here, connecting the results from Chapters 2 and 4 together more completely.
87
Chapter 5
Station Application Example
This chapter presents an example case which utilizes all the main aspects of the thesis.
The centralized architecture presented in Chapter 2 is tested with a new function
proposed for the station level in Chapter 4. In order to gain as much benefit from
the new architecture as possible, the measurement method presented in Chapter 3 is
utilized in order to increase the sampling frequency of the measurements.
An earth fault occurs when one or several phases become galvanically connected
to the ground. It is very common in MV distribution networks, where almost half of
the faults are single-phase earth faults. Early indication and location of these faults is
important, since it can prevent new faults from developing from the original one. If
a single-phase earth fault is not detected, it can develop into a two-phase earth fault
and/or a short circuit, and can be hazardous to people.
All this makes the reliable detection and location of earth faults a very important
feature of the secondary system of the distribution network, as was indicated in the
list presented in the previous chapter. In Chapter 4, FLIR functionality (Fault Lo-
cation, Isolation and Power Restoration) was identified as a functionality that would
benefit from the centralized architecture. One such algorithm has already been eval-
uated [Valtari, 2004], but this yielded insufficient results. Due to the possibility of
increasing the sampling frequency for the current and voltage measurements with the
method presented in Chapter 3, this algorithm is worthy of reassessment, and this is
done in this chapter.
88
5.1. Earth fault in the distribution network
5.1 Earth fault in the distribution network
5.1.1 Grounded network
If the neutral point of the network is connected to the ground, the network is grounded.
Figure 5.1 presents an equivalent circuit to the earth fault in a grounded three-phase
network:
Figure 5.1: An earth fault in a grounded network.
The inductances shown in the figure represent the windings of a voltage trans-
former (only secondary coils in the figure). The phase voltages are marked with U1
, U2 and U3 , and the earth capacitances with C0 . The resistance Rf connected to
one of the phases represents an earth fault, and the fault current flowing through the
star point of the network is marked with If . The current flowing through the earth
capacitances can be regarded as negligible in comparison with the current through
the star point.
Detecting and locating earth faults in grounded networks is fairly easy, since the
fault current during the fault is always large. When the potential of the neutral point
is fixed to the earth potential, the voltage across the fault resistance will remain equal
to phase-to-earth voltage. This keeps the fault current high and the faulted part of the
network has to be disconnected for safety reasons. In principle, this fault corresponds
to a normal short circuit between two phases.
Earth faults in grounded networks are usually detected with overcurrent relays,
and the distance can be estimated from their magnitude of the fault current. Grounded
89
networks are practically nonexistent in Finland due to the grounding conditions, al-
though they are relatively common in few other countries such as in the UK.
5.1.2 Isolated network

When the neutral point of the network is not connected, but left floating, the network
is isolated. In Finland, most of the MV networks are isolated, largely due to the low
conductivity of the ground [Lehtonen and Hakola, 1996]. The Figure 5.2 shows an
earth fault in an isolated network:
Figure 5.2: An earth fault in an isolated network.
The above figure is similar to Figure 5.1. The only difference is that the neutral
point of the network is left floating. This causes a neutral voltage between the star
point and ground, which is marked in the figure as U0 . The fault current If flows
through the earth capacitances (C0 ).
An earth fault in an isolated network does not necessarily cause any disturbance
to the power supply. During an earth fault, the potential of the neutral point moves
towards the phase voltage. The earth fault lowers the voltage in the faulted phase and
raises the voltages in the other phases. Because the voltage across the fault location
decreases at the same time, the fault current is not always large, especially when the
network is large or its fault resistance is high. The phase-to-phase voltages remain
the same during the fault. The voltage vectors during an earth fault are presented in
Figure 5.3 .
The phase voltages during normal operation are marked with voltage vectors U1 ,
U2 and U3 . During an earth fault, the potential of the neutral point changes, which
90
Neutral point
(Ground potential
during normal
operation)
Figure 5.3: Voltage vectors during an earth fault.
causes a neutral voltage, U0F . Phase voltages during an earth fault are marked with
U1F , U2F and U3F . The voltage vector U0F forms a circle as a function of fault
resistance. This is mainly because the voltage vector U0F caused by the capacitive
current is perpendicular to the voltage vector U1F , mainly caused by the resistive
current.
Although keeping the network isolated reduces the number of interruptions in the
distribution of electricity, it also puts more stress on the network. Voltages in healthy
phases increase during an earth fault, which can cause more faults in the immediate
future, such as a two-phase earth fault or a short circuit fault. This is also one reason
why isolated networks are not so popular outside Scandinavia. [Nikander, 2002]
Equivalent circuit [Lehtonen and Hakola, 1996]
An equivalent circuit of an isolated network during an earth fault in phase 1 is de-

picted in Figure 5.4. When an earth fault occurs, the fault current flows through the
earth capacitances (C0 ), and the fault resistance (Rf ), because there is no other route
for the current. During the fault, the earth capacitances are connected in parallel so
they can be added together. If the fault resistance is zero, the fault current (Ie ) can
be calculated from (5.1), where ωN = 2πf is the angular frequency of the network.
91
Figure 5.4: An equivalent circuit of an isolated network during an earth fault.
The amplitudes of the vectors U1 and Ie are marked with scalars U1 and Ie . The same
notation is used throughout the thesis.
Ie = 3ωN C0 U1 (5.1)
If the fault resistance differs from zero, the fault current is reduced. The reduced
fault current (Ie ) can be calculated with (5.2).
Ie
If = q (5.2)
2
1 + ( UIe1 Rf )
The neutral voltage (U0 ) measured by the protection and control IED is the volt-
age If caused when it flows through earth capacitances, which is presented in (5.3).
Therefore, the ratio between the amplitudes of the neutral voltage and the phase volt-
age during an earth fault can be calculated with (5.4).
1
U0 = If (5.3)
3ωN C0
U0 1
=p (5.4)
U1 1 + (3ωN C0 Rf )2
(5.4) shows the same thing as Figure 5.3: when the fault resistance is very small
the neutral voltage can rise as high as U1 . That also raises the voltages in the healthy
phases, where the phase voltages can rise to phase-to-phase voltage. The highest
voltage value of the healthy phases is about 1.05 times the phase-to-phase voltage,
which is achieved when the fault resistance is about 37 percent of the impedance of
the earth capacitances (see vector U2F in Figure 5.3). [Nikander, 2002]
92
Because the fault current is not necessarily high during earth faults, it is also diffi-
cult to detect and locate. Detection with overcurrent relays is not normally possible.
The most reliable way to recognize an earth fault is by monitoring the neutral voltage
U0 [Nikander, 2002]. However monitoring the absolute value of U0 is not sufficient.
There is always some asymmetry in the network, and sometimes with high resistance
faults an earth fault can even reduce U0 . That is because an earth fault changes
the symmetry of the system, and by chance this change can be "towards symmetry".
Sometimes the change can also be greater in its angle than in its amplitude. The best
method is, therefore, to monitor the change in U0 .
5.1.3 Compensated network
The network is compensated if the neutral point of the network is connected to the
ground via inductance. The main idea of this is to compensate the imaginary part
of the fault current caused by earth capacitance in power lines. Compensation is
normally done with so-called Petersen coils, and it can be either centralized (at one
point) or distributed to many points in the network [Mörsky, 1993]. Sometimes, a
resistor is also connected in parallel with the coil. Figure 5.5 presents an earth fault
in a compensated network.
Figure 5.5: An earth fault in a compensated network.
As with the previous figures, U0 , U1 , U2 and U3 represent the neutral voltage and
the phase voltages respectively. C0 is the phase-to-earth capacitance and If the fault
93
current caused by fault resistance Rf . The inductance of the Petersen coil is marked
with L , and the resistance connected in parallel with it, Rp .
When an earth fault occurs, the inductance connected to the neutral point com-
pensates the capacitive current due to the earth capacitances (phase-difference is 180
degrees). The fault current decreases considerably, which decreases protective earth-
ing costs. The change in voltages during an earth fault is shown in Figure 5.6.
Neutral point
(Ground potential
during normal
operation)
Figure 5.6: An earth fault in a compensated network (vector representation).
This figure is similar to Figure 5.3. Voltages during symmetrical operation are
marked with U1 , U2 and U3 , and the voltages during an earth fault with U1F , U2F
and U3F , respectively. The current that causes neutral voltage U0F is no longer
entirely capacitive, and the angle between U0F and U1F is greater than 90 degrees.
Therefore the potential of the star point doesn’t form a circle, but a curve closer to an
ellipsoid.
If the Petersen coil fully compensates the capacitive current, the network is said
to be self-extinguishing. If an earth fault is only temporary, the network will ex-
tinguish a possible electrical arc automatically. The Petersen coil also smooths the
changes in the neutral voltage, which prevents the fault recurring after the arc has
been extinguished. Often, total compensation is not desirable, in which case detect-
ing permanent earth faults becomes very difficult. An efficient compensation also
increases the asymmetry during normal operation. In most cases, the coil is tuned
so that the network is under- or over-compensated. Then, the fault current remains
94
large enough to be detected. Sometimes, a resistor is also connected in parallel with

the Petersen coil, in order to increase the fault current during permanent earth faults.
Such a resistor was also depicted in Figure 5.5. [Mörsky, 1993]
Equivalent circuit [Lehtonen and Hakola, 1996]
Figure 5.7: An equivalent circuit of a compensated network during an earth fault.
An equivalent circuit of a compensated network during an earth fault in phase

1 is depicted in Figure 5.7. In this case, the fault current has many routes. The
capacitive current (Ic ) flows through the earth capacitances (3C0 , because all three
are connected in parallel), and the inductive current (IL ) through the Petersen coil
(L ). If the compensation is total, IL + Ic = 0 and the fault current (If ) only has a
resistive component. The resistive component consists of the fault resistance (Rf ),
the leakage resistance of the coils and cables (RLE ) and a possible resistor connected
in parallel with the Petersen coils (Rp ). From the equivalent circuit the amplitude of
the fault current (If ) can be calculated with (5.5).
q
2 (3ω C − 1 2
U1 1 + RLE N 0 ωN L )
If = q (5.5)
1
(Rf + RLE )2 + Rf2 RLE
2 (3ω C −
N 0 ωN L )
2
In the equation, ωN represents the angular frequency of the network. If the net-
work is self-extinguishing, the inductive current fully compensates for the capacitive
current, and (5.5) can be simplified to (5.6).
U1
If = (5.6)
RLE + Rf
The amplitude of the neutral voltage (U0 ) affecting across the parallel-connected
capacitances, inductances and resistances can be calculated with (5.7).
95
If
I0 = q (5.7)
1 1
( RLE )2 + (3ωN C0 − ωN L )
2
As with previous equations, this is also simplified if the compensation is total.

The total reactance drops to zero, and the ratio between the neutral voltage and the
phase voltage can be calculated with (5.8).
U0 RLE
= (5.8)
U1 RLE − Rf
For the equations (5.5) to (5.8) above, it was assumed that there was no resistor in
parallel with the Petersen coil (Rp = 0). If a resistor is used, its effect can be taken
into account by replacing RLE by the parallel coupling of Rp and RLE . As with
isolated networks, the amplitude of the neutral voltage (U0 ) can rise as high as the
phase voltage (U1 ).
5.1.4 Initial transients

When an earth fault occurs in an isolated or compensated network, there is always a
fast transition, i.e. a transient. When the voltages on the phases change, the charge
in them has to change too. During the transient, there is a current flowing from
one phase to another which compensates for the change in the network. When the
charge of the supply lines reaches the new equilibrium, the transient slowly decays.
Utilization of these transients in protection and control IEDs has been rare, due to
the high 10kHz sampling frequency requirements for the detection of the transients
[Lehtonen, 1992].
The voltage transient due to an earth fault is shown in Figure 5.8. It also shows,
how the voltages change during a low resistance earth fault - the voltage of the faulted
phase drops close to zero whereas the voltages of the healthy phases rise to phase-to-
phase voltage.
The transient consists mainly of two components - which are called the discharge
component and the charge component. In addition, there are interline compensating
components, whose function is to equalize the voltages of parallel lines at their sub-
station terminals. In compensated networks there is also a decaying DC-component
(Direct Current), which is due to the Petersen coil. [Lehtonen, 1992]
The discharge component results from the decrement of the voltage in the faulted
phase. When the voltage in the faulted phase decreases, the energy stored in the
96
Figure 5.8: A simulated example, an earth fault in phase 1, transient components

visible in phase voltages.
capacitances of the cabling has to decrease as well. Simultaneously, the voltages

in the healthy phases rise, and these cables can reserve more energy. This appears
in the phase voltage as a higher-frequency charge component. Because the charge
component flows through a transformer, its frequency is much lower than that of
the discharge component. The frequency of the discharge component is normally
500...2500 Hz and that of the charge component 100...800 Hz [Lehtonen and Hakola,
1996]. The amplitude of the charge component dominates the transient, which makes
it more suitable for relaying purposes.
The charge transient can be modeled with the equivalent circuit presented in Fig-
ure 5.9. When an earth fault occurs at the substation, the angular frequency of the
charge transient (ωC ) can be calculated with (5.9), (5.10) and (5.11) [Hänninen,
2001]:
1 1
ωC = p =p (5.9)
Leq Ceq 3LT (C0 + Cpp )
Leq = 1.5LT (5.10)
97
Ceq = 2(C0 + Cpp ) (5.11)
Figure 5.9: Network model and equivalent circuit used for modeling the charge tran-
sient, fault resistance 0 Ω.
Ceq and Leq correspond to the respective equivalent capacitance and inductance
of the network. During an earth fault, the phase-to-phase capacitances (Cpp ) and the
earth capacitances (C0 ) are connected in parallel, as are two of the transformer phase
inductances (LT ). This causes the equivalent quantities to be as shown in (5.10) and
(5.11).
The amplitude of the transient is also dependent on the phase of the fundamental
frequency. If the fault occurs during the fundamental maximum, the transient ampli-
ˆ ) is largest and can be calculated with (5.12) [Lehtonen and Hakola, 1996],
tude (iCh
ˆ = Ceq ωC
iCh Ie (5.12)
3C0 ωN
where ωN corresponds to the angular frequency of the network, and Ie to the am-
plitude of the uncompensated, steady-state earth fault current. The fault resistance in
(5.12) is zero. At values close to fundamental zero, (the amplitude of the fundamental
50Hz or 60Hz component is zero) the amplitude is smallest, and in fact sometimes
the transient might not be visible at all. Normally, the fault occurs during the funda-
mental maximum (i.e. amplitude of the fundamental 50Hz or 60Hz component has
its maximum value), which is also the best case for calculations.
As noted from (5.12), the transient amplitude is linearly dependent on the transient
frequency. In practice, it can be up to 10-15 times the fundamental component of the
98
earth fault current Ie . The fault resistance also has a significant influence on the
charge transient. This not only has an effect on the earth fault current (Ie ), but it
also dampens the transient. The limit when the transient becomes overdamped in
overhead lines is considered to be 50 - 200 Ω [Lehtonen and Hakola, 1996]. The
greater the distance to the fault location, the greater the equivalent inductance will
be, which lowers the transient frequency.
When an earth fault occurs in larger networks, the transients do not only flow from
the faulty phase to the healthy ones, but also from the faulty feeder to the healthy
feeders. The total change in the charge of a feeder is clearly visible in the neutral
(summation) current I0 of the feeder. An example is shown in Figure 5.10, which is
a simulation result from the model presented later on, in section 5.3.2
Neutral current in all feeders
Faulted feeder
Healthy feeder 1
Healthy feeder 2
Healthy feeder 3
40 Healthy feeder 4
Healthy feeder 5
20
Current / A
−20
−40
−60
0.104 0.106 0.108 0.11 0.112 0.114 0.116 0.118 0.12 0.122
Time / s
Figure 5.10: Neutral current in all feeders during an earth fault in one feeder, 6
feeders in the simulated example.
5.1.5 Measured fault resistances during earth faults
Fault impedance is normally purely resistive [Lehtonen and Hakola, 1996]. The fault
reactance can thus be neglected, and only the fault resistance need be considered. As
99
explained in the previous sub-section, the charge transient normally becomes over-
damped when the fault resistance is 50 - 200 Ω [Lehtonen and Hakola, 1996]. This
means that earth contact has to be direct.
The resistance of a tree depends on many factors: the soil, humidity, season, type
of tree, etc. Fault resistance measurements on an experimental 12 kV bare conductor
line with an earthed neutral are reported in [Aro, 1993]. In summer, the resistance of
unseasoned wood with a moisture-free surface was about 20 kΩ. In the spring, the
resistance was 36...58 kΩ, and in winter, with frosty air and frozen soil, it was 2..3
MΩ. These values are much larger than 200 Ω, so in practice, charge or discharge
transients during earth faults caused by fallen trees are too damped for relaying pur-
poses.
Fault resistance is almost never zero, especially with overhead lines. In reference
[Hänninen and Lehtonen, 1998], a study was made in order to determine most normal
fault resistances in real-life situations. Figures 5.11 and 5.12 present the diagrams of
the different fault resistances evaluated in the study.
Figure 5.11: The division of the fault resistances in a compensated network [Hänni-
nen and Lehtonen, 1998].
The study shows that there are clearly two categories of fault resistances. In the
first category, the resistance is below a few hundred ohms. When the resistance is
below a few hundred ohms, there is an actual galvanic connection to the ground, and
the faulted part has to be disconnected from the network. In the other category, the
fault resistance is several kilo-ohms. These situations correspond to cases when, say,
100
5.2. Earth fault location methods
Figure 5.12: The division of the fault resistances in a isolated network [Hänninen and
Lehtonen, 1998].
a tree forms the contact. The fault current is normally low, and normal operation is
not threatened. These faults can, however, develop into low-resistance faults and/or
short circuits if they are not detected in time.
5.2 Earth fault location methods

The indication and location of earth faults has proved to be a very challenging task.
Fault resistances can vary greatly, as can the electrical capabilities of supply lines.
Currents and voltages also vary considerably due to changes in the weather, or in the
load. Connecting or disconnecting different electrical equipment can cause changes
in electrical quantities that are very similar to earth faults.
In an isolated or compensated network, a single-phase earth fault always disturbs
the symmetry of the network. The neutral voltage then changes abruptly, and this is
normally the best method of recognizing an earth fault. Analyzing the changes that
occur in current and voltage measurements helps in determining the fault location.
The main objective has been to calculate the impedance between the IED and fault
location. Cables have a known impedance per kilometer. If the impedance to the fault
location can be estimated accurately, that data can be used to derive the distance to
the fault.
The fault resistance can also vary according to the nature of the fault. If the cable
101
breaks, and the supply end of the cable touches the ground, the fault resistance can be
close to zero. On the other hand, if the supply end stays in the air, the fault resistance
can be several MΩ. The resistance of a tree can also be several kΩ. In practice, the
fault resistance is never zero.
Therefore, merely calculating the fault resistance is not an adequate method for
determining the distance to the fault location. Because the fault reactance is nor-
mally close to zero (fault impedance is normally only resistive), a much more reliable
method is to calculate the inductance of the network.
5.2.1 Earth fault location methods based on initial transients
Much research has focused on calculating the distance from initial transients at the
occurrence of an earth fault. As stated above, the current charge transient of the fault
has a relatively low frequency and high amplitude in comparison with the other tran-
sient components, which makes it eminently suitable for earth fault location methods.
The goal has been to estimate the inductance of the faulted phase from the transient.
The distance can then be calculated from the inductance, assuming that the induc-
tance per kilometer is known. The equations presented in section 5.1.4 were for
those cases when the fault occurs at the substation. When the fault occurs further
away in the network, the inductance of the line must also be taken into consideration.
Connecting a Petersen coil to the neutral point of a voltage transformer does not
affect the amplitude of the charge transient, since the transient frequency is signifi-
cantly higher than the fundamental frequency. The inductive reactance of a Petersen
coil is therefore fairly large in the frequency area of the charge transient and it can
be compared with isolation [Lehtonen, 1992]. For this reason, the transient method
works on both isolated and compensated networks.
A method based on current and voltage transients measured from the incoming
feeder was studied in [Hänninen and Lehtonen, 2002b]. Calculation of the distance
from the transient took up a fairly large amount of the processor time. Several filters
and a DFT (Discrete Fourier Transform) analysis had to be used before the result was
achieved. The algorithm also used a large amount of memory, since the data from
a relatively long period has to be processed. Furthermore, the references implied
that the sampling frequency has to be relatively high, > 10 kHz for a reliable earth
fault location, which is not possible with present-day protection and control IEDs.
New earth fault location algorithms which have been proposed require even higher
102
sampling frequencies, up to 100 kHz [Ma et al., 2010].

Using sampling frequencies of less than 10 kHz creates an additional error in the
estimates of the distance of the fault [Valtari, 2004]. The frequency of the charge
transient is only 100...800 Hz, so based on the Nyquist-Shannon theorem, 2 kHz
should be sufficient for the calculations [Phadke and Thorp, 1990]. The theorem
does not fully apply to this case, since the measured transient is so short that only a
few data points are used for the whole analysis [Lehtonen and Hakola, 1996].
Other drawbacks with the presented algorithm are based on the amount of data
[Hänninen and Lehtonen, 2002b]. Since the whole analysis is made with the data
from the beginning of the fault, there is only one dataset available from the fault
transient. If the transient is not suitable for analysis because of other, concurrent
changes in the voltages, the fault location can not be determined. Fruthermore, the
transient becomes overdamped when the fault resistance is 50...200 Ω [Lehtonen and
Hakola, 1996].
Two different methods have been explored and tested for the calculation in [Hän-
ninen and Lehtonen, 2002a]. The first one, called "the differential equation method",
calculates the inductance in the time domain with the help of a trapezoidal rule
[Schegner, 1989]. Another method has been studied, which uses wavelet transforma-
tion instead of the trapezoidal rule [Hänninen et al., 1999]. This method calculates
the inductance in the frequency domain. The results with the differential equations
have been slightly better, especially in an isolated network, which makes this a more
recommendable implementation. The method for increasing the sampling frequency
presented in Chapter 3 also presents the possibility of testing the performance of the
algorithm without a 10 kHz sampling frequency, and this will be presented later on
in this chapter.
Transient-based methods have also been used to indicate the faulted feeder in a
substation. As described earlier in this chapter, in the case of a larger substation an
earth fault causes the overall charges of different feeders to change, which can be
observed from the neutral current. A method proposed in [Abdel-Fattah and Lehto-
nen, 2009] uses this phenomenon for detecting the faulted feeder. The neutral current
transient flowing from the faulted feeder towards the substation equals the sum of
the neutral currents of all the other feeders flowing away from the substation. A
combination of a transient-based earth fault location algorithm with a faulted feeder,
and faulted phase detection, provides complete earth fault location functionality in a
substation, provided that the fault resistance is low enough.
103
5.2.2 Other algorithms for earth fault location
As stated, the algorithm presented above, which is based on the initial charge tran-
sient, suffers from the fact that it requires a sufficient number of samples from the du-
ration of the transient. If the inductance could be calculated from the network states
(before and after the fault) instead of the transition between them, accuracy could
be greatly improved. This would allow the algorithm to repeat calculations and thus
increase the accuracy of fault location by averaging the various consequent values.
Promising results have been achieved, but fault location, especially in compensated
networks, is still a challenging issue [Hänninen and Lehtonen, 2002a] [Wahlroos and
Altonen, 2011].
MV networks are normally meshed in order to ensure energy distribution in fault
situations. However, normally a network has radial operation, because it is simpler
to control. Operating MV networks with a meshed rather than a radial topology
would enable the utilization of new methods for earth fault location [Nikander, 2002].
Comparing the fault currents from two IEDs connected to the same faulted phase
would provide more information about the fault. If the fault is between the IEDs, the
ratio of the fault currents is inversely proportional to the ratio of the fault distances.
There are also other methods for calculating the fault distance using the data from
two IEDs. However, one drawback to this is that connecting a faulted part of the
network into a closed ring might give rise to other dangerous situations, even if it
were only for a short time.
The use of artificial neural networks (ANN) is also an interesting area for earth
fault location algorithms [Eberl et al., 2000]. The objective is to teach the system the
difference between the healthy state and the faulted state of a power line, and the dif-
ferences in electrical quantities when the distance to the fault location changes. How-
ever, with current technology, the training period of the algorithm is long. Further-
more, the solution is not generic, but dependent on the environment. The algorithm
has to be constantly adapted to new environments, which decreases the independence
of the algorithm. Therefore, current ANN technology is not yet advanced enough for
this purpose, although the situation may well change in the future.
Some algorithms use a model of the network to calculate the fault location [Saha
et al., 2001]. Given adequate information about the network topology, it might be
easier to determine the location of a fault. However, the need for a network topology
increases the implementation costs and makes the IEDs less independent.
104
5.3. Test results for the impact of sampling frequency on transient-based earth fault
location
The use of specific cable radar is not, in itself, another algorithm, but is rather a
separate device which calculates the distance to the fault location. It uses specific
impulses, which it sends to the faulted power line. When the pulse reaches the fault
location, there is a sudden change in the impedance. From this point a part of the
signal impulse is reflected back to the radar. Analyzing the time interval between the
transmitted and the returned impulse helps in determining the fault location. Cable
radar is the most accurate method for detecting the fault. The drawback with this
method is that it needs a separate device for the analysis. Furthermore, voltages
have to be disconnected while the measurements are taken. In urban areas, where
the cables are mostly underground, this method is still the only reasonable choice.
Digging up part of a cable is such an expensive project that fault location has to be
precise. [Mörsky, 1993]
5.3 Test results for the impact of sampling frequency on

transient-based earth fault location
In order to test the performance of the method for increasing the sampling frequency
of current and voltage measurements (as proposed in Chapter 3), the method was
applied to a transient-based earth fault location algorithm. The method is presented
below, in section 5.3.1 and the results are analyzed in section 5.3.2.
5.3.1 Description of the algorithm
The algorithm studied here is based on the research carried out during a Tekes re-
search project in 2002 [Hänninen and Lehtonen, 2002b], and initially tested in [Val-
tari, 2004]. The basic concept of the algorithm comes from previous studies [Scheg-
ner, 1989], and only small modifications have been made. Using the trapezoidal rule
[Phadke and Thorp, 1990] to solve the differential equation, and a current correction
algorithm before the differential equation improved the accuracy. The algorithm uses
voltage and current values measured from one feeder, so one algorithm per substation
is sufficient, which is in line with the results in Chapter 3. In order to determine in
which feeder the fault occurs, the location algorithm presented in [Abdel-Fattah and
Lehtonen, 2009] was used. The inputs and outputs of the algorithm are presented in
Table 5.1. The flow chart of the algorithm is presented in Figure 5.13.
In Figure 5.13 the components of the function were also introduced. The name
105
location
Table 5.1: Inputs and outputs of the earth fault location algorithm
Type Signal Description
U1, U2, U3, I1, I2, I3 Phase voltages and currents
U0 Neutral Voltage
Limit Trigger level of neutral voltage
Inputs
Cpp Phase-to-phase capacitance
Ltf Phase inductance of the voltage transformer
Lkm Inductance of the power lines per kilometer
LEFA Low resistance earth fault alarm
Outputs Distance Distance to the fault location
Deviation Deviation of the calculated distance
of each component is written next to the corresponding flow chart element in italics,
and the different components are explained more accurately below.
DetectTransient_A
The component DetectTransient_A keeps a buffer of phase voltages and currents

from the previous 30 ms. Every 5 ms it checks the buffer and searches for a possible
earth fault. The fault is detected by monitoring the neutral voltage. If the network
has compensation capacitors, their influence has to be reduced in the data. After de-
tecting a fault, the indication and the corresponding voltage and current buffers are
fed to the next component.
Here, the faulted feeder is also detected with the method presented in [Abdel-
Fattah and Lehtonen, 2009]. The average value of the neutral current for each feeder
is calculated over the time-span of the first transient half cycle (from the beginning
of the transient to the first zero crossing), see (5.13).
PN
k=1 i0,k
I0 = (5.13)
N
where:
ir,k is the instantaneous neutral current at sample k, and
N is the number of samples in the transient window.
The determination of the faulted feeder is then made by calculating the K value
according to (5.14)
106
location
Figure 5.13: Flow chart of the differential equation algorithm and corresponding
components.
|I0 − I0,others |
K= ∗ 100% (5.14)
I0,all
where
107
location
P
I0,others = I0 (for other feeders) and
P
I0,all = |I0 | (for all feeders).
This calculation gives K=100% for the faulted feeder and a value close to 0% for
all the other feeders.
FilterTransient_A
FilterTransient_A removes the fundamental frequency from the transient with a comb
filter; see (5.15) [Lehtonen, 1992]. It also determines the beginning of the transient
before delivering data arrays to the next component. The most accurate analysis can
be made from the first transient wavelength, so finding this is essential.
g(t) = f (t) − f (t + T ) (5.15)
In (5.15) f(t) is the original and g(t) the filtered signal. T is the period of funda-
mental frequency.
Fourier_A
Fourier_A performs a DFT analysis of the transient. Its purpose is to determine the
transient frequency, so that higher frequencies can be filtered out. It calculates the
amplitudes of the frequency components between the frequencies of 100 Hz and 1
kHz, and delivers them to the next component.
CutoffFrequency_A
After receiving the data from the Fourier analysis, CutoffFrequency_A calculates
which frequency component has the largest amplitude. This frequency component is
the transient frequency. The cut-off frequency for the following low-pass filter is set
at 50 Hz higher than the transient frequency. [Hänninen and Lehtonen, 2002b]
BesselFilter_A
BesselFilter_A includes a second degree Bessel low-pass filter. Since the most accu-
rate analysis can be made from the charge transient, the discharge transient and other
high-frequency components are removed with this filter.
108
location
CurrentCorrection_A
When the calculations are made with the charge component, the effect of the trans-
former has to be taken into account. Part of the charge component flows through
the transformer windings and part of it is delivered via phase-to-phase capacitances.
These changes have to be compensated before the distance to the fault location can
be calculated. This compensation is made with CurrentCorrection_A.
DifAlg_A
DifAlg_A includes the differential equation, which finally calculates the inductance
of the faulted line, and through that the distance to the fault location. The calculation
is done with the help of the trapezoidal rule, (5.16) [Phadke and Thorp, 1990]:
∆t (ik+1 + ik )(uk+2 + uk+1 ) − (ik+2 + ik+1 )(uk+1 + uk )

L= [ ] (5.16)
2 (ik+1 + ik )(ik+2 − ik+1 ) − (ik+2 + ik+1 )(ik+1 − ik )
In the rule, L is the inductance of the faulted phase, t the time interval between
two consecutive samples, and ik and uk the respective current and voltage values at
instant k. This rule can be used for all situations where the current and voltage fulfill
the following differential equation of the first order (5.17).
di(t)
u(t) = Ri(t) + L (5.17)
dt
The above trapezoidal rule calculates local estimates from only three samples. As
a result there will be an array with many different inductance values. The most likely
real inductance value is determined with local variance [Schegner, 1989]. The algo-
rithm calculates the variance of a certain number of consecutive inductance values.
If the variance is small, the inductance value only changes a little. The period during
which the smallest variance occurs is considered to be the best for calculations. The
average inductance value of that period is the most-likely real value.
After that, both the distance and the distance deviation can be calculated from
the inductance and the variance. For these calculations, the algorithm needs only the
approximated inductance per km of the transmission cable.
109
location
5.3.2 Analyzing the simulation results
The algorithm was tested with simulated data. The model and the results from the
different cases are presented in this section. The main focus of these cases was to
evaluate the performance of the method for increasing current and voltage measure-
ments presented in Chapter 3. The tests were performed in a simulation model with
seven feeders (one incoming, six outgoing) under three different scenarios:
A) Single measurement with the full sampling frequency
B) Seven different sample streams, merged to full sampling frequency, as pre-
sented in Chapter 3. The voltage measurements are merged according to section
3.3.2 and the current measurements according to section 3.3.3.
C) Single measurement with one seventh of a full sampling frequency
The anti-alias filters of the individual IEDs were not taken into account in the sim-
ulation, as the target was only to focus on the effect of different sampling frequencies,
and especially on the differences between cases B) and C) which would both have the
same anti-alias filter.
Fault location accuracy is defined in [IEEE, 2004] according to (5.18), and is also
often expressed as a percentage value instead of a per unit one.
(instrument reading − exact distance to the f ault)

e(error) = (5.18)
total line length
The required accuracy for fault location depends greatly on the type of network.
For a rural overhead-line network, the main purpose is to locate the right control zone,
i.e. which section of the network should be disconnected. Normally, an accuracy of
10% of the line length is considered sufficient for this [Wahlroos and Altonen, 2011],
[Manner et al., 2011].
Simulation model
The simulation model used to test the algorithm and the measurement method was
created with the alternative transient program ATP/EMTP, which is a popular simula-
tion software package mainly intended for transient analysis applications. The model
is presented below in Figure 5.14
More detailed information about the model is presented below, and the faults were
simulated in Feeder 1.
110
location
Healthy Feeders
110/20 kV, ∆/Y

Substation
Transformer

Faulted Feeder
Measuring Fault
Point Point

Figure 5.14: Simulation model used for testing.
• Substation transformers: 10 MVA, 110/20 kV. Six outgoing feeders, of total

length = 244 km
• Feeder 1: 40 km, LV transformers (20/0.4 kV)= 1 MVA, Load = 900 kW
• Feeder 2: 33 km, LV transformers (20/0.4 kV)= 0.8 MVA, Load = 720 kW
• Feeder 6: 42 km, LV transformers (20/0.4 kV)= 1 MVA, Load = 900 kW
Isolated network, Rf = 10 Ω, sampling frequency varied
Figures 5.15 to 5.17 present the error of the algorithm, as well as the calculated
deviation of the network set-up with three different sampling rates. In this set-up, the
fault resistance (Rf ) was 10 Ω and the network had an isolated neutral. The network
had six outgoing feeders in addition to the incoming one. The fault distance was
varied from 0.8 km to 39.2 km.
111
location
Calculation error and reported deviation

(km) 3.5
Error
Deviation
2.5
2
Error / km
1.5
0.5
0
0 5 10 15 20 25 30 35 40
Distance to fault location / km (km)
Figure 5.15: Results with sampling frequency of 16 kHz, Rf = 10 Ω.

(km) 5
Error
Deviation
4.5
3.5
3
Error / km
2.5
1.5
0.5
0
0 5 10 15 20 25 30 35 40
Figure 5.16: Results with seven different sample streams, combining to 16 kHz when
processed as in Chapter 3, Rf = 10 Ω.
112
location

(km) 12
Error
Deviation
10
8
Error / km
0
0 5 10 15 20 25 30 35 40
Figure 5.17: Results a sample frequency of 16 kHz / 7 = 2.29 kHz, Rf = 10 Ω.
The results do not reveal any large differences between the different sampling ar-
rangements. As was expected, although the arrangement with a single, low sampling
rate IED without measurement merging in Figure 5.17 does not perform as well as
the method with measurement merging in Figure 5.16, or with the one single high
sampling frequency measurement in Figure 5.15, the difference is not great. All the
cases fulfill the criteria of 10% accuracy (4km in a 40km network), when the fault
distance is over 5km.
The individual sampling frequency of 2.29 kHz does not correspond to a real case
of protection and control IEDs, where the sampling frequency is a round number
such as 2 kHz (which would result in a total sampling frequency of 7 * 2 kHz = 14
kHz). An overall sampling frequency of 16 kHz instead of 14 kHz was used in order
simplify the simulation arrangements (16 kHz means a sample time of 62.5 µs).
Frequency and amplitude of the charge transient, Rf = 10 Ω
One interesting point is to investigate how the transient behaves with different fault
distances, because that is the only source of information in the algorithm. Viewing
the frequency and amplitude of the transient with a function of fault distance gives us
some idea about the behavior. Now we can also compare it to the theory from section
113
location
5.1. The derived transient frequencies and corresponding amplitudes of the Fourier
component with different fault distances are presented in Figure 5.18.
Frequency and amplitude of the charge transient current
(Hz) 1000 100 (A)
Frequency
Amplitude
Amplitude / A
f / Hz
500 50
0 0
0 5 10 15 20 25 30 35 40
Distance to earth fault / km (km)
Figure 5.18: Transient frequency and amplitude with Rf = 10 Ω.
Figure 5.18 supports the theory. When the distance between the fault location and
the IED increases, the inductance also increases. This reduces both the frequency
and the amplitude of the charge transient. The transient frequency is in the frequency
area predicted by the references,and the amplitude varies from 90 A to 40 A.
Isolated network, Rf = 80 Ω, sampling frequency varied
Figures 5.19 to 5.21 present the same information when the fault resistance is in-
creased to 80 Ω.
Here, the differences between the different sampling arrangements is clear. When
the fault resistance increases, the transient becomes smaller and the analysis must be
made from much smaller current and voltage variations. In this case, only the results
with one single high frequency measurement are below the required 10% (or 4 km),
see Figure 5.19. The results from the measurement merging method in Figure 5.20
are below 10% when the fault distance is between 8 km and 28 km. At the beginning
of the feeder, the error in the algorithm is 30% and at the end of the feeder it is 15%
of the feeder length. The results from the single low sampling frequency, Figure 5.21,
are the worst, the maximum error being over 150%.
114
location

(km) 7
Error
Deviation
4
Error / km
0
0 5 10 15 20 25 30 35 40
Figure 5.19: Results with sampling frequency of 16 kHz, Rf = 80 Ω.

(km) 25
Error
Deviation
20
15
Error / km
10
0
0 5 10 15 20 25 30 35 40
Figure 5.20: Results with seven different sample streams, combining to 16 kHz when
processed as in Chapter 3, Rf = 80 Ω.
115
location

(km) 250
Error
Deviation
200
150
Error / km
100
50
0
0 5 10 15 20 25 30 35 40
Figure 5.21: Results a sample frequency of 16 kHz / 7 = 2.29 kHz, Rf = 80 Ω.
Frequency and amplitude of the charge transient, Rf = 80 Ω
As stated in section 5.1.4, the greater the distance to the fault location, the greater the
equivalent inductance will be, which lowers the transient frequency. Furthermore,
the transient amplitude is linearly dependent on the transient frequency. The fault
resistance also dampens the transient, and with 80 Ω fault resistance it is only around
30-40 A, which is one major reason for the errors of over 150% of the line length in
Figure 5.21
Overall performance
The tests were repeated with different sampling frequencies. The aim was to compare
three different arrangements. First, a case where the measurements are received from
a single measurement device with a high sampling frequency was evaluated. Then the
method presented in Chapter 3 was tried out, where the same sampling frequency was
derived from multiple lower sampling frequency devices (7 in this case). Finally, for
comparison, the measurement from one such device (without measurement merging)
was used for fault distance calculation. These three cases with two different fault
resistances resulted in six different cases, which are shown in Table 5.2.
116
location
Frequency and amplitude of the charge transient current

(Hz) 800 50 (A)
Frequency
Amplitude
600 40
Amplitude / A
f / Hz
400 30
200 20
0 5 10 15 20 25 30 35 40
Distance to earth fault / km (km)
Figure 5.22: Transient frequency and amplitude with Rf = 80 Ω.
Table 5.2: Different measurement cases
Abbr Function
1 Rf = 10 Ω, single measurement with the full sampling fre-
quency
2 Rf = 10 Ω, seven different sample streams, merged to full
sampling frequency (as in Chapter 3)
3 Rf = 10 Ω, single measurement with the one seventh of a
full sampling frequency
4 Rf = 80 Ω, single measurement with the full sampling fre-
quency
5 Rf = 80 Ω, seven different sample streams, merged to full
sampling frequency (as in Chapter 3)
6 Rf = 80 Ω, single measurement with the one seventh of a
full sampling frequency
In order to make a more detailed evaluation, these six cases were repeated with
three different sampling frequencies, 20 kHz (one IED 20 kHz / 7 = 2.86 kHz), 16
kHz (one IED 16 kHz / 7 = 2.29 kHz) and 10 kHz (one IED 10 kHz / 7 = 1.43
kHz). The fault distance has an influence on the transient, and therefore also on the
117
location
accuracy of the estimation of the distance to the fault. Therefore, each network set-up
was divided into three parts, with different distances to the fault location. The mean
errors of the calculations with different sampling arrangements and fault distances
are presented in Figures 5.23 to 5.25.
Error with different network setups and different fault distances

60
Distance 1−13km
Distance 14−26km
Distance 27−40km
50
40
Mean error / km
30
20
10
0
1 2 3 4 5 6
Network setup
Figure 5.23: Mean errors with different sampling set-ups and fault distances, fs = 20
kHz.
With a sampling frequency of 20 kHz, the differences between the different cases
are not major when the fault resistance is 10 Ω (Cases 1-3 in Figure 5.23). All
three sampling frequency set-ups have a fault location error below 10% of the feeder
length. When the fault resistance increases to 80 Ω, the deficiencies of the lower sam-
pling frequencies become more apparent (Cases 4-6 in Figure 5.23). With one single
low sampling frequency measurement (Case 6 in Figure 5.23), the average error in
fault location with fault distances from 1-13 km is nearly 150% of the feeder length.
The other two cases still fulfill the 10% error requirement for the fault location, so
fault location estimation within 10% error limits was achieved with IEDs operating
with less than 3 Khz sampling frequency, when they were synchronized according to
the concept presented in Chapter 3.
When the overall sampling frequency decreases, the performance of the measure-
ment method proposed in Chapter 3 becomes more apparent. With an overall sam-
118
location

70
Distance 1−13km
Distance 14−26km
Distance 27−40km
60
50
Mean error / km
40
30
20
10
0
1 2 3 4 5 6
Network setup
kHz.
40 Distance 1−13km
Distance 14−26km
Distance 27−40km
35
30
25
Mean error / km
20
15
10
0
1 2 3 4 5 6
Network setup
kHz.
119
pling frequency of 16 kHz, fault distances can no longer be calculated without merg-
ing the measurements from different bays (see Figure 5.24). Case 6, with one single
lower sampling frequency measurement introduces errors of over 100% in fault loca-
tion estimation. The measurement merging method does not fulfill the 10% criteria
either, as when the fault distance is below 14 km, the error is around 15%.
The simulation data shows that the measurement method presented in Chapter 3
also makes it possible to use transient-based methods when the sampling frequency
of one IED is below 3 Khz, remembering that the sampling frequency limit earlier
proposed was 10 kHz [Hänninen and Lehtonen, 2002a] [Abdel-Fattah and Lehtonen,
2009]. This was apparent from the voltage measurement, as the TI-ADC set-up from
ADC design principles can be directly applied. However, it also brings benefits to
the current measurements, which are clearly visible in this fault distance calculation
algorithm, dependent on accurate current measurements.
The tests in this thesis were only performed on a simulation model. Unfortunately,
the time frame of this thesis did not allow for a pilot implementation of the proposed
measurement method to be performed on an IED, and neither was it possible to con-
duct field tests with the IEDs. Practical field tests always yield less satisfactory results
than simulations, which are made with ideal signals. Re-evaluating the simulation re-
sults against field test results is a future research topic.
New fault location algorithms requiring even higher sampling frequencies, up to
100 kHz [Ma et al., 2010], have not been tested in this thesis. The research in this
thesis was done utilizing the sampling frequency available with present-day IEDs,
which is below 10kHz. A possible future research topic would be to test the latest
methods (e.g. [Ma et al., 2010]) for increasing the sampling frequency with the set-
up presented in Chapter 3, by utilizing, for example, multiple 10 kHz measurements,
assuming that such IEDs will become available in the future.
5.4 Chapter summary
The set-up presented in the thesis can be used for transient-based, distance-to-fault
calculation during low ohmic faults (below 100 Ω). This method can even be applied
to modern protection and control IEDs with sampling frequencies below 3kHz, if
they are synchronized according to the TI-ADC methodology.
This chapter also demonstrated the combination of the topics presented in this
120
thesis, i.e. the new, centralized, station-level data processing, re-allocation of func-
tionality from the bay-level devices and the measurement method for increasing the
accuracy of the measurements. After this initial and encouraging example, more re-
search can be carried out with other functions.
121
Chapter 6
Summary
This thesis focused on the new challenges facing electricity distribution substations.
New legislative requirements, changes in the business landscape or simply the more
stringent day-to-day needs to improve processes have increased the stress on life-
cycle costing. It seems very likely that the functional life cycle of bay-level protection
and control IEDs is getting shorter. Even though the physical device itself might
have a relatively long life cycle of 15-20 years, future roadmaps presented in various
professional publications often have 5-10 year implementation steps. This means that
the utilities must pay special attention to the overall architecture of the secondary
system of a distribution substation, so that future updates can be carried out in a
cost-efficient manner.
The centralized architecture evaluated in this thesis provides a basis for further
development of the secondary system. It combines the bay-level devices with a target
life-span of 15-20 years with a station-level computer with 2-year upgrade interval.
The proposed architecture allows the primary functionality in the bay-level devices
to remain unchanged, while enabling updates to be made to the station computer.
The calculations show that the larger the substation, the more likely it is that the
architecture presented here will be the most cost-efficient solution.
This centralized architecture can be implemented by utilizing the existing power
utility standards. IEC 61850 provides the means for modeling the environment, and
also for handling the communication. The process bus IEC 61850-9-2 allows all the
station measurements to be available locally, and the latencies are short enough even
for protection functionality, as long as the time synchronization is properly handled,
with, for example, IEEE 1588. The recent additions to IEC 61850 with regard to the
122
6.1. Contribution of the thesis
engineering processes also support the centralized architecture.

The introduction of a station-level protection and control device also calls for a
rethink of the functionality in the substation. As the architecture allows having both
fixed, invariant functions, and flexible, adaptive functions, the functions required in
the substation need to be reallocated. The functionality classification method pro-
posed in this thesis takes a new view of a substation’s functionality and provides the
means for utilizing the new architecture in an optimal way.
In addition to re-allocating the old functionality, the new architecture also enables
the utilization of totally new methods and functions. One such method, related to the
measurement chain in a substation, is presented and tested here. When the measure-
ments of the substation are synchronized in the way proposed here, the total sampling
frequency of the station-level measurements can be increased.
All these aspects were tested on the platform of a transient-based earth fault loca-
tion algorithm. Earlier studies had shown that transient-based methods are not feasi-
ble in bay-level IEDs, because the sampling frequency was too low. When the new
measurement method was applied at the station level, the simulation results showed
that transient-based methods are feasible, without the need to increase sampling fre-
quency at the IED level. The architecture is also tested out in one practical, real-life
pilot project.
6.1 Contribution of the thesis

The main contributions of the thesis can be divided into two categories - top/concept-
level contributions and feature/application-specific contributions.
The concept-level contributions of the thesis are:
• New architectures for the secondary system of a distribution substation are

summarized and evaluated. The most critical elements affecting the overall
life-cycle costs are identified. A method is proposed for determining the op-
timal architecture for the automation of a substation, depending on its size.
According to the results of this study, a substation with more than 5 feeders
should utilize the centralized architecture proposed in this thesis.
• A method is provided for dividing the functionality of a substation, utilizing

both substation-level and bay-level devices. Based on this method, the func-
tional scope of both levels is presented, indicating which functions should be
123
6.2. Evaluation of the thesis
implemented at the station level, and which at the bay level. The method is ap-
plied to a set of functions commonly used in a Finnish distribution substation.
The feature and application-level contributions of the thesis are:
• A new way of handling substation-wide measurements is provided, which can

be used to increase the sampling frequency of the measurement chain with-
out increasing the sampling frequency of the individual IEDs. The proposed
method can be implemented with existing technology, and with the existing
time synchronization method of IEEE 1588. A patent application has also
been submitted for the principle [Valtari, 2012]. The patent has been allowed
and is about to be granted.
• A transient-based earth fault location algorithm has been enhanced, utilizing

the proposed measurement method. With this enhanced measurement method
it was possible to achieve 10 % fault location accuracy with multiple IED mea-
surements with 3 kHz sampling frequency, which compares well with the 10
kHz required in previous research.
6.2 Evaluation of the thesis

The strength of this thesis lies in the way it addresses many different aspects of the
centralized architecture. As well as evaluating the top/concept-level issues, it also
gives detailed specifications for issues such as measurement methods and individ-
ual functions. The measurement method, for which a patent application has been
submitted, gives a clear indication of how the new architecture creates an innovative
environment for new ideas on many levels. Furthermore, the way the measurement
method was directly applied to a practical fault location algorithm adds value to the
results.
The deficiencies in the thesis are in the evaluation of the financial impact of in-
dividual functions. The available statistics were not detailed enough to carry out an
accurate financial analysis. Chapter 2 showed that the value of the increased relia-
bility is more important than other elements in the life-cycle cost, but there were no
statistics available for evaluating the financial benefit of a particular function. For
this reason, the results from both Chapters 2 and 4 are more important in terms of the
methods used to derive the results, rather than the actual values themselves.
124
6.3. Future research
6.3 Future research

The research into substation-level centralized functionality will continue after the
submission of this thesis, particularly at the application level. The thesis proved that
the architecture is cost-efficient and provides an environment which allows updates
on a 2-year cycle while keeping the life-cycle costs below those of a set-up consisting
of only bay-level IEDs. Future research will now focus on the applications allocated
to the station level, as discussed in Chapter 4. During 2011 a laboratory environ-
ment was set up in cooperation with Tampere University of Technology, including
an RTDS R
realtime simulation environment modeling the network and sending all
the data via IEC 61850 station and process buses. The first applications to be re-
searched in this laboratory environment are algorithms for detecting high-impedance
earth faults and double earth faults. In fact, the first results of tests conducted on
high-impedance earth fault detection have already been published [Nikander et al.,
2012]. The next target is also to conduct field tests with pilot installations, in order
to confirm the results from both [Nikander et al., 2012] and from Chapters 3 and 5 of
this thesis.
125
References
ABB Ltd. Teknisiä tietoja ja taulokoita. ABB Ltd., 2000. in Finnish.
ABB Ltd. Relion Protection and Control 615 series Technical Manual, 2.0 edition,
2009.
ABB Ltd. Relion Protection and Control 615 series Technical Manual, 3.0 edition,
2010.
M.F. Abdel-Fattah and M. Lehtonen. A Novel Transient Current-Based Differential

Algorithm for Earth Fault Detection in Medium Voltage Distribution Networks. In
International Conference on Power Systems Transients (IPST), 2009.
G.G. Angel. Maintenance Strategies for M.V. and H.V. Substations. In The 17th In-
ternational Conference and Exhibition on Electricity Distribution (CIRED), 2003.
A. Apostolov and B. Vandiver. IEC 61850 GOOSE applications to distribution pro-

tection schemes . In 64th Annual Conference for Protective Relay Engineers, 2011.
A. Apostolov, D. Tholomier, and A. Edwards. Advanced Load-Shedding Functions

in Distribution Protection Relays. In The 19th International Conference and Exhi-
bition on Electricity Distribution (CIRED), 2007.
M. Aro. Pas-johtoon syntyneen vian ilmaiseminen sähköasemalla. Technical report,

Helsinki University of Technology, Espoo, Finland, 1993. in Finnish.
F. Baldinger, T. Jansen, M. van Riet, F. Volberda, F. van Erp, M. Dorgelo, and W. van
Buijtenen. Advanced secondary technology to evoke the power of simplicity. In
The 9th International Conference on Developments in Power System Protection
(DPSP), 2008.
126
REFERENCES REFERENCES
W.C. Black and D.A. Hodges. Time-interleaved Converter Arrays. IEEE Journal of
Solid-State Circuits, SC-15(6):1022–1029, 1980.
J. Castallenos. Engineering guidelines for IEC61850 based substation automation

systems. In Cigre Working Group B5.12, 2009.
G. Eberl, S. Hänninen, M. Lehtonen, and P. Schegner. Comparison of Artificial

Neural Networks and Conventional Algorithms in Ground Fault Distance Compu-
tation. In IEEE PES Winter Meeting, The Institute of Electrical and Electronics
Engineers Power Engineering Society Winter Meeting, Singapore, volume 3, pages
1991–1996, 2000.
EC. Communication from the Commission to the European Parliament, the Council,
the European Economic and Social Committee and the Committee of the Regions
- 20 20 by 2020 - Europe’s climate change opportunity. European Commission,
2008a.
EC. Communication from the Commission to the European Parliament, the Council,
the European Economic and Social Committee and the Committee of the Regions -
Energy 2020, A strategy for competitive, sustainable and secure energy. European
Commission, 2008b.
J. Elovaara and Y. Laiho. Sähkölaitostekniikan perusteet. Otatieto Oy, Helsinki,

Finland, 2004. in Finnish.
EMV. Electricity Market Act - Sähkömarkkinalaki - 386/1995. Energy Market Au-

thority, Finland, 1995. in Finnish.
EMV. Sähkön jakeluverkkotoiminnan ja suurjännitteisen jakeluverkkotoiminnan hin-

noittelun kohtuullisuuden valvontamenetelmien suuntaviivat vuosille 2012-2015.
Energy Market Authority, Finland, 2011. in Finnish.
Energiateollisuus. Keskeytystilasto 2005-2010 (Outage statistics 2005-2010). Finnish

Energy Industries, Energiateollisuus ry, Helsinki, 2005-2010. in Finnish.
P. Ferrari, A. Flammini, S. Rinaldi, and G. Prytz. Time Synchronization Concerns

in Substation Automation System. In IEEE International Workshop on Applied
Measurements for Power Systems (AMPS), 2011.
127
Fortum. Fortum ulkoistaa sähköverkkopalvelunsa Eltel Networksille ja Empow-

erille Suomessa, October 2010. URL http://www.tekniikkatalous.
fi/energia/fortum+ulkoistaa+sahkoverkkopalvelunsa+
suomessa/a514251. in Finnish.
F. Gorgette, O. Devaux, and J-L. Fraisse. Possible Roadmaps for New Requirements
for French Distribution Control and Automation. In The 19th International Con-
ference and Exhibition on Electricity Distribution (CIRED), 2007.
L. L. Grigsby. The Electric Power Engineering Handbook. CRC Press, 2000.
GSMA. European Mobile Industry Observatory 2011. The GSM Association, 2012.
S. Hänninen. Single phase earth faults in high impedance grounded networks - Char-
acteristics, indication and location. PhD thesis, Helsinki University of Technol-
ogy, Finland, 2001.
S. Hänninen and M. Lehtonen. Characteristics of Earth Faults in Electrical Distribu-

tion Networks with High Impedance Earthing. Electric Power Systems Research,
44(3):155–161, 1998.
S. Hänninen and M. Lehtonen. Earth Fault Distance Computation with Fundamental

Frequency Signals Based on Measurements in Substation Supply Bay. Techni-
cal report, VTT Technical Research Centre of Finland, VTT research notes 2153,
2002a.
S. Hänninen and M. Lehtonen. Maasulkujen etäisyyden laskenta syöttökennomit-

tauksista. Technical report, VTT Technical Research Centre of Finland, TESLA
Report 57/2002, 2002b. in Finnish.
S. Hänninen, M. Lehtonen, T. Hakola, and R. Rantanen. Comparison of Wavelet and

Differential Equation Algorithms in Earth Fault Distance Computation. In 13th
Power Systems Computations Conference, Trondheim, Norway, volume 2, pages
801–807, 1999.
J. Heckel. Smart Substation and Feeder Automation for a Smart Distribution Grid.
In The 20th International Conference and Exhibition on Electricity Distribution
(CIRED), 2009.
128
M. Hinow, M. Waldron, L. Müller, and K. Pohlink H. Aeschbach and. Substation

Life Cycle Cost Management Supported by Stochastic Optimization Algorithm.
In 42nd General Session of CIGRE (International Council on Large Electric Sys-
tems), 2008.
IEA. World Energy Outlook. International Energy Agency, 2009. Available at:
http://www.worldenergyoutlook.org/.
IEC. International Electrotechnical Vocabulary, IEC Standard IEC 60050- 605-01-

01. The International Electrotechnical Commission, 1983.
IEC. Dependability Management. Part 3 Application guide - Section 3: Life Cycle

Costing, IEC Standard 60300-3-6. The International Electrotechnical Commis-
sion, 1997.
IEC. Standard for Communication Networks and Systems in Substations, IEC Stan-
dard 61850. The International Electrotechnical Commission, 2005.
IEC. Standard for Communication Networks and Systems for Power Utility Automa-
tion, IEC Standard 61850-6 Edition 2. The International Electrotechnical Com-
mission, 2009.
IEEE. IEEE Guide for Safety in AC Substation Grounding, IEEE Standard 80-2000.
IEEE Power Engineering Society, 2000.
IEEE. IEEE Guide for Determining Fault Location on AC Transmission and Dis-
tribution Lines, IEEE Standard C37.114-2004. IEEE Power Engineering Society,
2004.
IEEE. IEEE Standard Electrical Power System Device Function Numbers, Acronyms
and Contact Designations C37.2-2008. IEEE Power Engineering Society, 2008.
IEEE. Standard for a Precision Clock Synchronization Protocol for Networked Mea-
surement and Control Systems, IEEE Standard 1588. The Institute of Electrical
and Electronics Engineers, 2009.
IPCC. IPCC Special Report on Renewable Energy Sources and Climate Change
Mitigation. Cambridge University Press, Cambridge, United Kingdom and New
York, NY, USA, 2011.
129
I. Jeromin, J. Backes, G. Balzer, and R. Huber. Life Cycle Cost Analysis of Trans-
mission and Distribution Systems. In The 20th International Conference and Ex-
hibition on Electricity Distribution (CIRED), 2009.
A. Johnson, J.E. Söderström, P. Norberg, and A. Fogelberg. Standard Platform for

integrated soft protection and control. In IEEE PES Conference on Innovative
Smart Grid Technologies (ISGT 2010), 2010.
K. Kauhaniemi, S. Voima, and H. Laaksonen. Adaptive Relay Protection Concept

for Smart Grids. In 17th Power Systems Computation Conference (PSCC), 2011.
E. Kettunen. Sähköasematietokoneen konfigurointi. Master’s thesis, Tampere Uni-

versity of Technology, 2011. in Finnish.
G.H. Kjølle, J. Heggset, B.T. Hjartsjø, and H. Engen. Protection System Faults 1999-
2003 and the influence on the Reliability of Supply. In Power Tech, IEEE Russia,
2005.
G.J. Klir and B. Yuan. Fuzzy Sets and Fuzzy Logic: Theory and Applications. Pren-
tice Hall, 1995.
P. Koponen. Sparse Sampling Methods for Power Quality Monitoring. PhD thesis,
Tampere University of Technology, 2002.
J. Laine. Sähkönjakeluverkon komponenttien pitoajat. Master’s thesis, Lappeenranta

University of Technology, 2005. in Finnish.
E. Lakervi and E.J. Holmes. Electricity Distribution Network Design. The Institution
of Engineering and Technology, 1996.
E. Lakervi and J. Partanen. Sähkönjakelutekniikka. Otatieto, Yliopistokustannus,

2008. in Finnish.
J. Lassila, S. Viljanen, and J. Partanen. Investoinnit sähkön siirron hinnoittelussa.

Technical report, Lappeenranta University of Technology, 2002. in Finnish.
M. Lehtonen. Transient Analysis for Ground Fault Distance Estimation in Electrical

Distribution Networks. VTT Technical Research Centre of Finland, VTT Publica-
tions 115, 1992.
130
M. Lehtonen and T. Hakola. Neutral Earthing and Power System Protection. Earth-
ing Solutions and Protective Relaying in Medium Voltage Distribution Networks.
ABB Transmit Oy, Vaasa, Finland, 1996.
B. Lundqvist. 100 Years of Relay Protection, the Swedish ABB Relay History. Tech-
nical report, ABB Automation Products, Sweden, 2010.
S. Ma, B. Xu, G. Houlei, B. Zhiqian, and A. Kiimek. An improved differential equa-

tion method for earth fault location in non-effectively earthed system. In The 10th
International Conference on Developments in Power System Protection (DPSP),
2010.
P. Manner, K. Koivuranta, A. Kostiainen, and G. Wiklund. Towards Self-Healing

Power Distribution by Means of the Zone Concept. In The 21st International
Conference and Exhibition on Electricity Distribution (CIRED), 2011.
M. Matikainen. Energiamarkkinaviraston Ajankohtaispäivät 2010, March

2011. URL http://www.energiamarkkinavirasto.fi/files/
Matikainen_Mika_201003.pdf. in Finnish.
F. Mekic, Z. Wang, V. Donde, F. Yang, and J. Stoupis. Distributed Automation for

Back-Feed Network Power Restoration. In The 20th International Conference and
Exhibition on Electricity Distribution (CIRED), 2009.
S. Miyamoto, H. Ichihashi, and K. Honda. Algorithms for Fuzzy Clustering: Methods

in c-Means Clustering with Applications (Studies in Fuzziness and Soft Comput-
ing). Springer, 2008.
J. Mörsky. Relesuojaustekniikka. Otatieto Oy, Helsinki, Finland, 1993. in Finnish.
B. Nartman, T. BrandStetter, and K. Knorr. Cyber Security for Energy Automation

Systems – New Challenges for Vendors. In The 20th International Conference and
Exhibition on Electricity Distribution (CIRED), 2009.
A. Nikander. Novel Methods for Earth Fault Management in Isolated or Compen-

sated Medium Voltage Electricity Distribution Networks. PhD thesis, Tampere
University of Technology, 2002.
A. Nikander, O. Raipala, J. Valtari, and E. Kettunen. Verifying The Method for

Indication of High-Resistance Earth Faults in Developed RTDS Test Environment.
131
In The 10th Nordic Electricity Distribution and Asset Management Conference

(NORDAC), 2012.
J. O’Brien and A. Deronja. Use of Synchrophasor Measurements in Protective Re-

laying Applications. Technical report, IEEE Power System Relaying Committee
Working Group C-14 Report, 2012.
OSF. Consumer price index - appendix table 5. 2005=100, January 2011. URL
http://www.stat.fi/til/khi/index_en.html.
A. Oudalov, D. Ishchenko, J. Stoupis, and R. Cai. Re-Coordination of Protection

Devices in Microgrids. In The Electric Power System of the Future - CIGRE 2011
Bologna Symposium, 2011.
T. Pahkala, M. Hänninen, M. Matikainen, and L. Simola. Sähköverkkokomponent-

tien yksikköhintojen määritys. Technical report, Empower, Energiamarkkinavi-
rasto, 2010. in Finnish.
J. Partanen, J. Lassila, T. Kaipia, M. Matikainen, P. Järventausta, P. Verho, A. Mäki-

nen, K. Kivikko, J. Pylvänäinen, and V-P Nurmi. Sähkönjakeluverkkoon sovel-
tuvat toimitusvarmuuskriteerit ja niiden raja-arvojet sekä sähkönjakelun toimitus-
varmuudelle asetettavien toiminnallisten tavoitteiden kustannusvaikutukset. Tech-
nical report, Lappeenranta University of Technology, 2006. in Finnish.
A.G. Phadke and J.S. Thorp. Computer Relaying for Power Systems. Research Stud-
ies Press Ltd, England, 1990.
K.S. Rasmunssen. A Real Case of Self Healing Distribution Network. In The 20th In-
O. Rintamäki and K. Kauhaniemi. Applying Modern Communication Technology to

Loss-of-Mains Protection. In The 20th International Conference and Exhibition
on Electricity Distribution (CIRED), 2009.
M.M. Saha, F. Provoost, and E. Rosolowski. Fault Location Method for MV Cable
Network. In The 7th International Conference on Developments in Power System
Protection (DPSP), pages 323–326, 2001.
P. Schegner. Digitaler Erdschlussuniversalschutz. Konzept und erste Realisierung.

PhD thesis, Universitaet des Saarlandes, Germany, 1989. in German.
132
M. Soudan and R. Farrell. Impact of Time-interleaved Analog-to-Digital Converter

Mismatch on Digital Receivers. In IEEE International Conference on Electronics,
Circuits and Systems, 2008.
J. Starck, A. Hakala-Ranta, and M. Stefanka. Switchgear optimization Using IEC

61850-9-2 and Non-conventional Measurements. In PAC World Conference, 2012.
J. Tengdin. High impedance fault detection technology. Technical report, IEEE

Power System Relaying Committee Working Group D15, 1996. http://www.pes-
psrc.org/.
V. Terzija, G. Valverde, D. Cai, P. Regulski, V. Madani, J. Fitch, S. Skok, M. Begovic,

and A. Phadke. Wide Area Monitoring, Protection and Control of Future Electric
Power Networks. Proceedings of the IEEE, 99(1):80–93, 2011.
Texas Instruments Inc. Application Note 1728 IEEE 1588 Precision Time Protocol
Time, Synchronization Performance, 2007.
J. Valtari. Development of Earth Fault Location Algorithm in Medium Voltage Net-

work Protection Terminal. Master’s thesis, Tampere University of Technology,
2004.
J. Valtari. Method and System for Measuring Electrical Quantity in Electrical Net-
work. Patent Application WO2012110418 (A1), 2012.
J. Valtari and P. Verho. Efficient Secondary System Configuration Process Utilizing

Centralized Substation Functions. In PAC World Conference, 2011a.
J. Valtari and P. Verho. Requirements and Proposed Solutions for Future Smart Dis-
tribution Substations. Journal of Energy and Power Engineering (JEPE), 5(8):
766–775, 2011b.
J. Valtari and P. Verho. Method to Increase Sampling Frequency by Merging Multiple

Measurements . In IEEE International Workshop on Applied Measurements for
Power Systems (AMPS), 2012.
J. Valtari, P. Verho, A. Hakala-Ranta, and J.Saarinen. Increasing Cost-Efficiency

of Substation Automation Systems by Centralised Protection Functions. In The
20th International Conference and Exhibition on Electricity Distribution (CIRED),
2009a.
133
J. Valtari, P. Verho, A. Hakala-Ranta, and J.Saarinen. Increasing Cost-Efficiency

of Substation Automation Systems by Centralised Protection Functions. Indian
Journal of Power and River Valley Development, 1(1):1–4, 2009b.
J. Valtari, T. Hakola, and P. Verho. Station Level Functionality in Future Smart Sub-
stations. In The 9th Nordic Electricity Distribution and Asset Management Con-
ference (NORDAC), 2010.
M.J.M. van Riet, F.L. Baldinger, and W.M. van Buijtenen. Alternative Approach for
a Total Integrated Secondary Installation in MV Substations Covering All Possible
and Required Functions. In The 18th International Conference and Exhibition on
Electricity Distribution (CIRED), 2005.
ViolaSystems. Case Vattenfall: Automating the Distrobution Network, Octo-

ber 2011. URL http://www.violasystems.com/sites/default/
files/Vattenfall%20case.pdf. in Finnish.
C. Vogel. The Impact of Combined Channel Mismatch Effects in Time-Interleaved

ADCs. IEEE Transactions on Instrumentation and Measurement, 54(1):415–427,
2005.
F. Volberda, M. van Riet, and A. Pikkert. The Power of Simplicity. In The 19th In-
A. Wahlroos and J. Altonen. Advancements in Earth Fault Location in Compensated

MV Networks. In The 21st International Conference and Exhibition on Electricity
Distribution (CIRED), 2011.
134

Centralized Architecture of The Distribution Substation Automation

Uploaded by

Copyright:

Available Formats

Centralized Architecture of The Distribution Substation Automation

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Centralized Architecture of The Distribution Substation Automation

Uploaded by

Copyright:

Available Formats

Tampereen teknillinen yliopisto.

Centralized Architecture of the Electricity Distribution

Tampereen teknillinen yliopisto - Tampere University of Technology

5 Station Application Example 88

1.1 Outline of the thesis. . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1 Typical voltage levels in the electricity network in Finland. . . . . . 9

3.1 Time-Interleaved ADC with M channels [Vogel, 2005]. . . . . . . . 55

4.1 Function re-allocation results. . . . . . . . . . . . . . . . . . . . . 81

5.1 An earth fault in a grounded network. . . . . . . . . . . . . . . . . 89

ADC Analogue-to-Digital Converter

AMR Automatic Meter Reading

CBM Condition Based Maintenance

CID Configured IED Description

CIS Customer Information System

DFT Discrete Fourier Transform

DMS Distribution Management System

ENS Energy Not Supplied

FLIR Fault Location, Isolation and power Restoration

GHG Greenhouse Gas

GOOSE Generic Object Oriented Substation Event

GPS Global Positioning System

IDA Intelligent Distribution Automation research project

IED Intelligent Electronic Device

IID Instantiated IED Description

LCC Life Cycle Costs

NCC Network Control Center

NIS Network Information System

RCM Reliability Centered Maintenance

RTU Remote Terminal Unit

SAV Sampled Analogue Value

SCADA Supervisory Control And Data Acquisition

SCD Substation Configuration Description

SCL Substation Configuration Language

SED System Exchange Description

SGEM Smart Grids and Energy Markets research program

SINAD SIgnal-to-Noise-And-Distortion ratio

SNTP Simple Network Time Protocol

THD+N Total Harmonic Distortion with Noise

TI-ADC Time-Interleaved ADC

The importance of a continuous supply of electrical energy has steadily increased

means increasing the functionality of existing IEDs (Intelligent Electronic Devices)

1.1 The hypothesis and objectives of the thesis

1.2 Background of the thesis

1.3 Outline of the thesis

Chapter Area [Publication]

[Valtari et al., 2009a]

Technology [Valtari and Verho, 2011b]

Application and function re- Functionality

An example utilizing all three aspects - proposed Earth Fault

Figure 1.1: Outline of the thesis.

1.4 The role of the author

The initial example listing of appropriate station-level functions presented in Chap-

2.1.1 An electricity network

Power generation and consumption often occur at different geographical locations.

is marked with G (Generator) and the consumption end with L (Load).

Figure 2.1: Typical voltage levels in the electricity network in Finland.

Figure 2.2: Example topology of a double busbar substation.

A separate automation layer is needed in the substation in order to control the