Synchrophasor Based Islanding & Open Phase Fault Protection
Synchrophasor Based Islanding & Open Phase Fault Protection
Scholarship@Western
2-24-2020 11:00 AM
Part of the Power and Energy Commons, and the Systems and Communications Commons
Recommended Citation
Jalali, Mansour, "Synchrophasor Based Islanding & Open phase fault Protection in Distribution Systems"
(2020). Electronic Thesis and Dissertation Repository. 6947.
https://ir.lib.uwo.ca/etd/6947
This Dissertation/Thesis is brought to you for free and open access by Scholarship@Western. It has been accepted
for inclusion in Electronic Thesis and Dissertation Repository by an authorized administrator of
Scholarship@Western. For more information, please contact [email protected].
Abstract
With the rapid growth of renewable energy resources, energy efficiency initiatives, electric
vehicles, energy storage, etc., distribution systems are becoming more complex such that
conventional protection, control, and measurement infrastructure – typically concentrated
at the main substation, with little to no access to information along the feeder – cannot
maintain the reliability of the system without some sort of additional protection, control
and measurement functionalities. As an example, a dedicated communication channel for
carrying the transfer trip signal from the substation to the Point of Common Coupling
(PCC) to prevent islanding operation of alternative resources, has been a requirement for
many utilities. In the transformation of the distribution system from a simple radial system
to a bidirectional energy flow network, integration of many intelligent devices and
applications will also be required. Thus, this situation calls for investment in
communication infrastructure, and augmentation of protection, control, and measurement
functionalities.
ii
integration, and infrastructure enhancements that are underway, including “smart grid”
initiatives.
In the first use case of this research, the behavior of the major DG types during islanding
is studied through accurate transient modeling of utility type distribution systems using
PSCAD-EMTDC and MATLAB. The study proposes augmentation of PMU-based
solutions to the current passive islanding protection elements, such as voltage and
frequency, and improving the non-detection zone of the passive elements by adapting their
settings based on normal loading conditions at closest known instant prior to the fault or
islanding occurrence. The solution proposes a system architecture that requires one PMU
at each PCC bus and in the main substation. The communication aspect is based on the IEC
6850-90-5 report, where the PMU can subscribe directly to the data stream of the remote
PMUs such that the need for PDCs in this application is eliminated, yielding better
performance.
In the second use case, an open-phase fault – a major concern for distribution utilities from
safety of public and equipment perspective – has been studied. Clearing the open-phase
fault without identifying the type of fault could result in an attempt by the recloser to
reenergize the downed wire; conversely, an undetected open-phase fault could initiate
ferro-resonance, thereby stressing equipment and increasing the risk to public safety, both
urban and rural. This work discusses comprehensive analysis of symmetrical components
of various types of open-phase faults in the distribution feeder with the presence of
distributed generators (DGs) and proposes the use of phasor measurement data located at
substation and PCC to identify the open-phase fault. The proposed algorithm relies on the
rate of change of the various current and voltage sequence components. In the study
conducted, the utility type feeder and substation are modeled in PSCAD-EMTDC, and
different types of open-phase fault and shunt faults are studied to verify the dependability
and security of proposed algorithm.
iii
Keywords
Synchrophasor, PMU, PDC, DG, IEC 61850-90-5, Islanding, Open-phase fault, Breaking
conductor.
iv
Summary For Lay Audience
v
I dedicate this work to my parents, my kids, Zhina and Alan
vi
Acknowledgments
I thank my examiners Dr. Bala Venkatesh and Dr. Jing Jiang for their valuable inputs on
my work. I thank Dr. Mohammad Dadash-Zadeh for his inputs during the publication of
this research.
I thank my parents and my children for their love and affection. Their selfless love and
sacrifices have been the eternal fuel for my ambition to achieve greater heights in my
academic career. I am thankful to my brothers, sisters, and cousins for their love and
affection; it keeps me motivated to pursue higher aspirations in life.
I extend my special thanks to all my other friends and relatives for their moral support.
I am thankful to all the staff members of the electronics shop and administrative staff of
the Department of Electrical and Computer Engineering for their support at various stages
of my degree programs.
Thank you.
vii
Table of Contents
Abstract ............................................................................................................................... ii
Acknowledgments............................................................................................................. vii
Chapter 1 ........................................................................................................................... 1
1 Introduction .................................................................................................................... 1
Chapter 2 ......................................................................................................................... 12
viii
2.4 Conventional Distribution System Properties ....................................................... 20
Chapter 3 ......................................................................................................................... 45
ix
Chapter 4 ......................................................................................................................... 68
4.2 Islanding................................................................................................................ 69
x
5.2 Review of Open Phase Detection Techniques .................................................... 101
5.5 Single Open Phase Fault Between Two Sources ................................................ 114
5.5.2 Open Phase Fault with the Ground Fault ................................................ 115
5.6 Open Phase Fault at the Same Side of the Sources ............................................. 117
xi
6.2 Contribution of this Work ................................................................................... 143
xii
List of Tables
Table 2-1. IEEE Survey Results for Protection Schemes in Distribution Systems [7]..... 18
Table 3-3. Typical Delay Range for PMU Application [28] ............................................ 62
Table 3-4. Expected Data Requirement for μPMU Application [39] .............................. 65
Table 4-7. Balance of Power Case Study Results with 𝝁PMU ........................................ 96
Table 5-1. Current Rate of Change for Parallel Faults ................................................... 112
Table A1. Conductor Type and Data Used in the Model ............................................... 150
xiii
Table E1. Rate of the Change of Current Symmetrical Components ............................. 165
xiv
List of Figures
Figure 2.2. Typical Two Primary Radial Feeders with Open Loop.................................. 14
Figure 2.4. Network Vault One-line Diagram (Courtesy of Toronto Hydro) ................... 16
Figure 2.10. Induction Wind Turbine Type 1 Short Circuit Profile ................................. 25
Figure 2.12. Induction Wind Turbine Type 2 Short Circuit Profile ................................. 27
xv
Figure 2.18. Wind Turbine Type 4 Reactive Power Contribution .................................... 33
Figure 2.22. Grid Connected Solar Farm Short Circuit Profile ........................................ 36
Figure 2.25. Real and Reactive- Power Control of Grid Imposed VSC ........................... 40
Figure 2.27. IEEE 1547 Grid Support Function History [Courtesy of IEEE 1547 WG] . 42
xvi
Figure 3.11. Conventional Phasor Measurement Communication Architecture [29] ...... 61
Figure 4.1. Typical Primary Distribution Feeder Topology in North America ................ 69
Figure 4.5. None Detection Zone Based Daily Profile Variation ..................................... 75
Figure 4.7. Simplified One Line Diagram of System Under Study .................................. 82
Figure 4.9. Wind Turbine Type 3 Islanding Event with Balance of Power Case 1 .......... 86
Figure 4.10. Wind Turbine Type 3 Islanding Event with Balance of Power Case 2 ........ 87
xvii
Figure 5.1. Conceptual System Architecture of the Proposed Solution ......................... 101
Figure 5.2. Open Phase Fault in Phase “a” at P-Q ......................................................... 103
Figure 5.3. Symmetrical Component Circuit of Open Phase Fault ............................... 104
Figure 5.4. Open Phase Fault with Downed Wire .......................................................... 104
Figure 5.5. Symmetrical Component Circuit of Open Phase with Ground .................... 106
Figure 5.6. Open Phase Fault with Ground at the Source Side....................................... 107
Figure 5.7. Symmetrical Component Circuit of Double Open Phase Fault.................... 108
Figure 5.8. Double open phase fault equivalent circuit .................................................. 109
Figure 5.10. Open Phase Fault Scenario 1-Case 1-Power & Voltage ............................ 121
Figure 5.11. Open Phase Fault Scenario 1-Case 1-Current ............................................ 122
Figure 5.14. Single Open Phase Fault (phase b)- Current .............................................. 127
Figure 5.15. Single Open Phase Fault (phase b)- Current ............................................. 128
Figure 5.16. Single Open Phase Fault (phase b)- Current .............................................. 130
Figure 5.17. Single Open Phase Fault (phase b)- Current .............................................. 131
Figure 5.18. Single Open Phase Fault with Solid Ground .............................................. 133
Figure 5.19. Single Open Phase Fault with Solid Ground .............................................. 134
Figure 5.20. Single Open Phase at the End of Feeder with 4% Load ............................. 136
xviii
Figure 5.21. Single Open Phase at the End of Feeder with 4% Load ............................. 137
Figure 5.22. Single Open Phase at the End of Feeder with 4% Load ............................. 138
Figure C1. Wind Turbine Model Type 3 Used in this Study ......................................... 159
Figure C2. Machine Side Converter Control “dq” Value Transformation .................... 160
Figure D1. Wind Turbine Model Type 4 Used in this Study.......................................... 161
Figure D2. Converters, DC link, Grid, and Machine Side Control I/O .......................... 162
Figure D3. Grid Side control, Identification of Current and Voltage Component ........ 162
Figure D4. Grid side Control, Calculation of Id, and Iq Current .................................... 163
Figure D6. Grid Side, Transformation of Grid Side Voltage ......................................... 164
Figure E1. Three phase Fault (ABCG) Power and Voltage............................................ 166
Figure E3. Two Phase Fault (BC) Power and Voltage ................................................... 168
Figure E5. Two Phase to Ground Fault (BCG) Power and Voltage ............................... 170
Figure E6. Two Phase to Ground fault (BCG) Current .................................................. 171
Figure E7. Ground Fault (BG) Power and Voltage ........................................................ 172
xix
Figure E8. Ground Fault (BG) Current ........................................................................... 173
Figure E9. Energizing Unbalance Load Power and Voltage .......................................... 174
Figure E11. De-energizing Unbalance Load Power and Voltage ................................... 176
Figure E13. Two Open Phase Fault (BC) Power and Voltage (point 2)......................... 178
Figure E14.Two Open Phase Fault (BC) Current (point 2) ............................................ 179
Figure E15. Two Open Phase Fault (BC) Power and Voltage (point 1)......................... 180
Figure E16. Two Open Phase Fault (BC) Current (point 1) ........................................... 181
xx
List of Appendices
xxi
List of Abbreviations
DG Distributed Generation
ES Energy Storage
xxii
IEEE Institute of Electrical and Electronics Engineers
MG Microgrid
xxiii
PTP Precision Time Protocol
PU Per Unit
SOC Second-Of-Century
SV Sampled Value
UG Under Ground
WT Wind Turbine
xxiv
1
Chapter 1
1 Introduction
In this chapter the needs for the modernization of power systems are briefly reviewed.
Specific challenges in the forefront of this modernization in distribution systems is
discussed. The research, objective, motivation and structure of this work is presented.
Traditionally, distribution systems were a simple interface between the end users and rest
of the grid with one-way power flow outward system architecture.
Distribution feeders were not interconnected and despite this simplistic blueprint yet today
this system is facing the following challenges:
With the publication of various sections of IEC61850 standard starting from 2011 and
onward the specification of utilities type protection applications based on communication
technology have been formalized and the journey of transformation of the conventional
application and prospective of new solutions have been the focus of many researchers and
application specialists around the globe. Furthermore, the integration of synchrophasor
data to substation automation domain and possibility of PMU using fast peer-to-peer
communication services to transfer the data provides more possibility to utilize the PMU
data in protection and time critical applications [2]. Background and the state of the art
regarding the phasor measurement technology is provided in chapter 3 of this work.
synchrophasor data in this sector have been studied and investigated. Distribution systems
are installed in a much smaller geographical area compared to the transmission systems
thus, smaller angle differences between the voltage phasors measured by PMUs and much
rapid system impedance changes could be expected [4]. In this regard the U.S. Department
of Energy’s Advanced Research Project Agency funded a US$4 million project to build
one of the most precise synchrophasor instruments ever made, with 100 times the
resolution of traditional transmission‐type PMU’s. The µPMU is ideal for research projects
that need ultra‐precise synchrophasor measurements for investigating stability and
impedance questions on distribution grids and microgrids. The current work however
proposes the use of synchrophasor data in protection scheme, the open phase fault detection
and islanding detection. Further information about background of synchrophasor
technology and accuracy required for this work is provided in chapter 3, 4 and 5 of this
work
protection functionality has started. This is a journey, which will transform the landscape
and allow for protection devices to respond to more complex abnormal situations and shift
their focus from traditionally protecting network components alone to protecting the
network status and operation.
The following properties are defining the different aspects of the protection system
performance.
1) Selectivity: The concept of protection selectivity refers to the capability of the protection
scheme to detect faults on a power system and initiate the opening of switchgear in order
to isolate only the faulty part of the system. Good selectivity will maximize service
continuity and minimize system outages. The protection must thus be discriminative.
2) Sensitivity: Sensitivity refers to the minimum operating level (current, voltage, power,
etc.) of protective devices. A relay designed to operate sensitively will be able to detect a
fault with a very low value. For example, a sensitive ground fault relay can detect a very
small ground fault current.
3) Reliability (Security and Dependability): Security and dependability must be evaluated
when assessing the reliability of a protection scheme. Dependability refers to the ability of
a protection scheme to operate and isolate a fault condition when it is required. Security
refers to the ability of the protection system not to operate during any tolerable conditions
such as overloading, switching actions, recoverable power swings and faults on other parts
of the power system, etc. Failure to operate (loss of dependability) can be extremely
damaging and disruptive. False tripping or over-tripping (loss of security) can result in
multiple contingencies, unnecessarily disconnecting the healthy power apparatus out of
service, and possibly cascading into a widespread blackout. The protection scheme should
offer secure and sensitive operation. It should be secure from false operation, not causing
de-energization of circuits due to load unbalances, inrush currents, cold load pickup,
harmonics, and other transient or steady-state conditions not normally harmful to system
components. The equipment in the protection scheme should exhibit enough sensitivity to
be able to detect all recognized fault conditions.
4) Speed: The function of protection is to isolate faults from the rest of the power system in
a very short time. The time in which the fault should be isolated it is an important property
of the protection system that is considered for power system stability, relay coordination,
and minimization of the damage.
7
More faults occur on the distribution system than in either the transmission system or in the
generating facilities. Distribution systems are widespread and have a relatively high degree of
exposure to the environment. Improvements in distribution circuit performance can be achieved
by design and by minimizing the number and extent of faults with overcurrent protection systems.
Knowledge of the characteristics, i.e., magnitude, duration, and waveform, of distribution faults is
essential when applying protection. Fault current calculation methods are fundamental tools for
the protection designer. Protection applications require computation of three-phase, line-to-
ground, and line-to-line short-circuit currents that are possible within the area of operation of the
device.
a) Why Synchrophasor? Since the blackout of August 2003 in North America, there
has been much focus on synchrophasor measurement application by utilities, power
system regulators, manufacturers, and researchers. IEEE synchrophasor standards
C37.118-1 and C37.118-2 are being updated by specifying the communication
requirements and dynamic performance of the synchrophasor measurement unit
(PMU). IEC 61850 is a standard that facilitates the implementation of
communication technology in protection, control, and measurement in power
system application, and integrates the synchrophasor data stream into its data model
and communication services. This integration provides the possibility to use the
functional and communication infrastructure of substation automation systems to
support and reduce the cost of PMU based applications, especially in a smaller area
such as a distribution system.
Installation of synchrophasor units in the distribution system can serve many new
and old challenges that this system is encountering and requires much more
research in this area.
c) In distribution systems, many overhead lines are built right along roads, streets, and
alleys. Because of equipment aging, and more recently extreme weather condition,
and car accidents involving the distribution overhead pole, the phase conductor(s)
can break and hit the ground creating a hazardous situation for the public. The high
impedance ground fault created by this event cannot be detected by any ground
fault protection element selectively and hence, the protection in the substation may
operate well after its time delay. The uncleared open phase conductor, if the created
ground fault is insignificant, can evolve into ferro-resonance which is another
added risk to the public and equipment. For many years, utilities and protection
manufacturers have worked to develop methods for tripping these hazardous
ground faults as quickly as possible. The method proposed in this work describes a
new way to identify the open phase conductor selectively with and without DG
using the synchrophasor measurement data.
Two use cases in this research are being investigated in a distribution system and with use
of synchrophasor data: Islanding detection use case and open phase fault use case.
Stage I – Islanding Detection Scheme: The first objective of the research is to provide a
synchrophasor-based islanding detection scheme that can improve the current anti-
islanding protection scheme practiced by utilities. In this stage, an Electro-Magnetics
Transient (EMT)-based model has been developed for a utility type system, and major
types of distribution generators have been studied as individual and aggregated machines
using PSCAD/EMTDC, supported through an extensive MATLAB simulation study to
formulate and validate the proposed solution.
Stage II – Open-phase fault detection: The scope was to develop a solution that can
selectively identify an open-phase and falling conductor fault in a primary distribution
system based on the minimum data required from different locations. Customized models
in PSCAD/EMTDC have been developed to study this fault and formulate, examine, and
validate the proposed solution.
1.7 Methodology
With these motivations, exhaustive research work was conducted to investigate the
application of synchrophasor measurement in distribution systems, including islanding
detection and open-phase faults. A few alternative models based on utility field data were
used to develop a reliable EMT-based model in PSCAD-EMTDC to validate the solution.
MATLAB was used for mathematical calculations, result validation or circuit analysis -
e.g., load flow for EMT model.
In Chapter 2, fundamentals of the distribution feeder related to this work are studied. The
integration of distributed energy resources (DER), including the history of different types
of distributed generation (DG), is reviewed and discussed. The modeling aspect of the DGs
is studied. The history of IEEE 1547 in DG operation requirement and unintentional
islanding is reviewed.
In Chapter 6, the conclusion and summary are provided, and further research topics are
suggested. The reference document list is provided, and appendices contain additional
model parameterization to enable reproduction of this work and further simulation.
1.9 Summary
A brief introduction to the research and its importance to the area of power system
protection is provided in this chapter. The challenges and need to modernize the power
11
Chapter 2
Introduction to Distribution Systems
Historically, electrical distribution networks have been the consumer interfaces with the
power plants and transmission systems where electricity, which is normally produced far
away from the center of load, is being transferred, delivered, and consumed. Although
interruption in the distribution services or failure in the distribution equipment directly
affects the end user, and the reliability and quality of the service, compared to the other
sector of energy distribution system it has been less technologically advanced. However,
in recent years with the rapid growth of alternative energy resources and the necessity for
the integration of many new devices such as Distributed Generations (DG), microgrids,
electric vehicles, etc., the distribution system is transforming to be at the forefront of the
renovation of electrical grids. It is very important to note that because of the massive
infrastructure of distribution systems, it is very capital and labor intensive [5], and
therefore, simplification and cost awareness have to be considered as chief characteristics
that will be demanded from researchers and solution providers. In this chapter, the
background of distribution system, with focus on the North American grid and some of the
challenges it is facing relevant to the current work, is presented.
The most common primary feeder configuration used by north American utilities is the
four-wire three-phase power multi-grounded neutral system. There are other types of
feeder configurations as well but generally, radial characteristics are very common between
primaries and secondaries. A distribution primary feeder can come in a variety of shapes
and forms, depending on the geometry of the area that a feeder is covering. For example,
the shape of the area and the layout of the streets will heavily impact the number and size
of the branches and overall form of the feeder. Figure 2.2 shows an arbitrary overhead line
primary feeder with a number of single-phase and three-phase laterals taped off from the
main circuit. As shown in this circuit here, the radial distribution feeder is normally
provided with the possibility to be connected to one or more adjacent feeders through the
open tie. This will improve the reliability of the circuit to be able to supply whole or part
of feeder load by closing the tie with the adjacent feeders. The distribution feeder can be
subjected to accommodate an integration with the DER(s) at one or more Point of Common
Coupling (PCC) which will be determined by utilities along the main circuits. The
14
integration of DGs imposes a new set of functionalities that should be provided by both the
utility and DG owner to maintain the safe operation of the feeder with the new
generation(s).
Figure 2.2. Typical Two Primary Radial Feeders with Open Loop
connection to the secondary grid. The network protector provides the following
functionality:
a. It prevents back feeding of the primary circuit by secondary grid during a primary
fault by tripping the circuit. Network protector is equipped with reverse power
protection element.
b. It trips the circuit and disconnects from the secondary grid when primary feeder is
deenergized.
c. It closes the circuit automatically when the primary feeder is energized.
The schematic of a network vault with two network transformers is presented in Figure 2.4.
The primary switch is used to connect the transformer to the primary feeder as well as
provides the ground at the location of the vault for the primary circuit. The meters are also
located at the secondary of the network transformer.
16
zone protection that can detect a fault within the protected zone with high selectivity and
security. Protective relays with the protected zone defined by their “reach” are referred to
as an open zone. The open zone is not bonded by the measuring devices and operates when
the measured quantity exceeds the pre-set threshold. Correct operation of open-zone
protection heavily relies on the protective element setting. Performance of an open-zone
protection scheme is usually a trade-off between security and dependability.
Bus zone
Bus zone
Transformer
zone
Auto recloser or
sectionalizers zone
Station Feeder
Protection zone
Under Ground
Cable Zone Lateral Zone
Transformer
zone
Reach in protection literatures refers to the locus of the most remote prospective fault
locations for which a specific protective device is capable to detect and clear. All points
electrically inside this frontier are considered to be within the zone of that protection
device. Typical example of protective devices with the clear reach are Distance and over
current relays. Distance relay can provide directional impedance measurement based on
the current and voltage of the network at the relay location. The reach in this relay relatively
stable and immune to the system condition. Overcurrent relay reach however, is a highly
18
variable and the ability of the relay to detect the fault within its reach will expand and
contract with variations in fault types and system conditions.
Table 2-1 presents the result of IEEE Power System Relay committee (PSRC) survey for
the practices on distribution system feeder protection. The results of the survey show by
far the over current relay phase and ground are the main protection schemes implemented
in distribution primary feeders.
Table 2-1. IEEE Survey Results for Protection Schemes in Distribution Systems [7]
Number of
Type of Protection utilities Percentage
responded implemented
Circuit recloser (79) 31 73%
Phase overcurrent (50,51) 42 100%
Ground overcurrent (50N,51N) 41 97%
Negative sequence over current 4 5%
(46)
High impedance Detection device 2 0.05%
(Broken conductor)
Distance (21) 4 5%
Directional over current (67, 67N) 1 0.02%
Figure 2.6 presents the most common protection schemes that have been used in
distribution feeders across the North American grid. The circuit recloser (79) that is
frequently used is another over current based protection that is capable to clear the fault in
its downstream location and is utilized to isolate the transient fault and reenergize the
circuit without the permanent outage. The circuit recloser can provide a very fast or a slow
response to the fault. The number of closing attempts and fast or slow tripping can be
programmed. It is common practice for utilities to use the circuit recloser in the fuse saving
or fuse blown schemes which are intended to save or blown the fuse respectively during
the transient faults. It can be noted that the high impedance ground fault protection which
is used to detect the broken conductor when it comes in touch with the ground is used less
19
than 0.05 % among the utilities that responded to survey. The feeders with integrated DER
are often equipped with protection transfer trip when the main is lost in order to prevent
the DER to supply the feeder consumers when the feeder circuit breaker is open.
Figure 2.7 shows a typical interface protection requirement for the DER owner. The anti-
islanding protection as shown here often is a point to point transfer trip (block 3) plus the
frequency and voltage protection elements (27, 59, and 81) that are installed at the PCC.
The anti- islanding protection can be further equipped with frequency rate of change (81R)
and loss of synchronism or out of step (78) protection elements.
20
d. In the overhead circuit, the laterals tapped off from the main circuit through the
fuses to protect laterals selectively, as shown in Figure 2.2.
e. The simplicity of the design is the main property of the existing system that ties to
the fact that distribution system, because of its size, is a very capital-intensive
business.
f. The anti- islanding protection that is shown in Figure 2.7 often is a point to point
transfer trip.
g. Often, no dedicated protection for an open phase fault is implemented.
Considering the status quo and backbone of electrical distribution system as some of the
properties summarized above, with the direction that has been taken by utilities and
government for production of clean energy, distribution grids with minimum hardware
preparation, is at the forefront of transformation to a new and smart gird. Integration of
rapid growth of DER, new solutions such as Microgrids, and energy storage etc. to support
reliability of system [9], [10] [11] from one side and advancement of communication
technology has provided an opportunity to many researchers and solution providers to
work towards addressing many of distribution system issue as a whole and facilitate the
transformation [12]. In the current work, the use of phasor measurement unit is proposed
to provide an advanced solution for addressing legacy issues, such as open phase fault, and
looking into the detection of DER operation in an unintended islanding operation.
Distributed Generation (DG): Any single or multiple generating units at a single location
owned and/or operated by the distribution utility a merchant entity. This includes Solar and
Wind Turbine generation.
Behind the Meter Generation (BTMG): A generating unit or multiple generating units
at a single location (regardless of ownership), of any nameplate size, on the customer's side
of the retail meter that serve all or part of the customer's retail load with electric energy.
Energy Storage Facility (ES): An energy storage device or multiple devices at a single
location (regardless of ownership), on either the utility side or the customer’s side of the
retail meter. This may include various technologies, including electric vehicle (EV)
charging stations.
Microgrid (MG): An aggregation of multiple DER types behind the customer meter at a
single point of interconnection that has the capability to island.
Figure 2.8 presents major DER categories that have been integrated into the distribution
system so far. Among the different type of DERs, the renewable devices, such as wind and
solar, are the most frequently installed and integrated into the distribution system. In the
current study, wind and solar DGs are considered in the modeling wherever DER presence
has been required.
23
magnetizing field of this machine. This is also the reason why capacitor banks are often
required to support the economical operation of this type of machine. From islanding
operation perspective, type 1 machine cannot support the unintentional islanding and
supply the load alone. The type 1 belongs to the early generation of wind turbine; the size
of this machine is in the range of 10 to 100 KW, and lack of speed regulation makes this
type of machine undesirable with today’s available technology.
Figure 2.10 presents the result of different fault type for wind turbine machine type 1
developed in PSCAD EMTDC. Prior to the short circuit instance at t= 4 sec machines were
supplying 1 per unit load. The time of the fault is an arbitrary one and the asymmetrical
current is not maximized based on the moment of the fault. The generator is shorted at the
collector prior to the point of common coupling transformer. It can be observed that the
type 1 machines are able to contribute a significant fault current to the grid depending on
the time of the short circuit. The contribution of the initial cycle of the fault (asymmetrical
current) can be as high as seven times the rated current and more. As the fault persists, the
contribution decreases in magnitude. By its nature, an induction machines consumes
reactive power both in the generating and motoring operation. The reactive power
consumption increases significantly as the output power increases.
25
LLL
LLG
LL
LG
Sec
In the type 2 wind turbine, shown in Figure 2.11, the induction generator used is a wound
rotor. There are no major differences between type 1 and type 2 turbines. They both have
almost fixed speed control. Type 2 has a better possibility for speed regulation by adding
resistance to the rotor circuit, and the real power curve can be “stretched” to the higher slip
and higher speed ranges. That is to say that the turbine will have to spin faster to create the
same output power, for an added rotor resistance.
26
The short circuit characteristics of a wind turbine type 2 is similar to a type 1. When the
external rotor resistance is not added to the rotor or shorted, the short circuit current is not
different with the squirrel-cage induction generator. Figure 2.12 shows the short circuit
simulation carried out in PSCAD EMTDC for the type 2 wind turbine for different types
of fault. The simulation is carried out with one external rotor resistance. The external
resistance in the rotor circuit will affect the value of the short circuit contribution
negatively. Prior to the short circuit instance at t=4 sec the machine is suppling a rating
value. The moment of short circuit is arbitrary and asymmetrical current is not maximize
based on the instance of the short circuit. It can be noted in this simulation that the type 2
machine can contribute significantly to the short circuit in the grid 4 to 5 per unit in the
initial cycle.
27
IA IB IC
4.0
2.0
PU [KA]
0.0 LLG
-2.0
-4.0
IA IB IC
4.0
2.0
PU [KA]
0.0
LL
-2.0
-4.0
IA IB IC
4.0
2.0
PU [KA]
0.0 LG
-2.0
-4.0
Sec 3.94 3.96 3.98 4.00 4.02 4.04 4.06 4.08 4.10 4.12 ...
...
...
Type 1 and type 2 are often equipped with a smooth starter where the induction machine
is supplied with variable frequency and reduced voltage and to run the machine until bring
the machine up to the rating voltage and speed.
Figure 2.13 shows a simple simulation of induction machine start up in PSCAD. The
recorded graphs are active power, reactive power, speed, and generator terminal voltage.
At the start, the voltage and frequency are gradually increased, and the machine absorbs
active and reactive power. When speed and voltage reach the value of network, the smooth
starter is bypassed, and the wind turbine is directly connected to the grid. From this moment
onward, as shown in this simulation, the induction machine generates real power (P) when
the turbine shaft rotates faster than the grid frequency.
28
contain some kind of a slip or decoupling mechanism between the two systems. In variable-
speed wind turbine Doubly Fed Induction Generator (DFIG), the rotor circuit is fed from a
converter with variable frequency, as shown in Figure 2.14.
Thus, the angular velocity of stator rotary field can be written as follows:
s r
= mech
ps pr
(2.1)
𝑛𝑠 − 𝑛𝑟
𝑠= ,
𝑛𝑠
Where ps and pr denote the number of stator and rotor poles respectively, s is the power
system frequency which is equal to the sum of the angular velocity of mechanical rotation
(mech), and rotor current frequency (r). Depending on the direction of the supply
frequency, this machine can operate in over or under synchronous speed and has much
more flexibility to work over a wider range of wind speeds. For the simplification of power
and torque equations, assuming Ps = Pr = 1, then the following can be stated from the
basic asynchronous machine model:
ωm = (1 − S)ωs
1−s
Pmech = 3|ir |2 ( ) Rr (2.2)
2
30
1−s R′r
Tmech = 3|i′r |2 ( ) =3Ψm |i′r | (2.3)
s ωm
Where Ψ𝑚 stator core magnetizing flux and prim indicates reflection of current and rotor
value to the stator side.
Vs
Ψm = Lm im =
ωs
Unlike type 1 and type 2 wind turbine machines, type 3 can provide reactive power to
participate in voltage regulation and when is connected to the grid. Figure 2.15 shows a
simulation that was carried out with the normal loading condition at t around 1.4 seconds;
the grid voltage is reduced by 6%, and thus, instantly the wind turbine in the absorbing
reactive power condition of (-0.4 PU) is changed to generating plus 0.4 PU to compensate
for loss of reactive power and reduction of voltage.
Similar to what is presented for the earlier wind turbine in this chapter, Figure 2.16 shows
the short circuit simulation for the type 3 machine carried out in PSCAD EMTDS. The
short circuit contribution for three-phase fault is shown to have the shortest decay time
with the peak current around 4 per unit. The phase to-phase -to-ground fault gives about
the same short circuit magnitude as the three-phase faults, but the decay time is longer. The
single line-to-ground fault produces the lowest peak current of about 5 per unit and it also
decays longer than the three-phase fault. From the short circuit waveforms, it can be
recognized that the symmetrical component analysis for the unbalanced short circuit is not
producing the same result as a conventional machine.
LLL
LLG
LL
LG
The complete active power, reactive power, and frequency control of DFIG for this work
is developed and customized in “dq” control using PSCAD. The overview of the PSCAD
model is reported in the Appendix C.
Figure 2.17 shows a type 4 full-fledged back-to-back inverter-based machine. The grid
interface can be an induction or synchronous generator. This type of wind turbine is most
frequently implemented around the world. Type 4 can provide an independent active and
reactive power control loop and therefore, it can participate effectively in the grid feeder
voltage regulation. This type of configuration offers a great deal of flexibility in operation
since there is no direct connection between the generator and the grid. The turbine is
allowed to rotate at its optimal aerodynamic speed, and the power output can still be
adjusted to the grid frequency. The dq control model customized for this study and its
parametrization is reported in the Wind Turbine Model Type 4.
Figure 2.19 presents the short circuit contribution of wind turbine type 4 for a different
type of fault. It can be noted that a short circuit current even for a three-phase fault is
limited to the rated current or a little above the machine rated current. The type 4 machine,
depending on the design of inverter it can support some 10% to 20%. The generator in this
type of machine is not connected directly to the grid therefore, during the fault in the grid
the generator can still be running with the connection to the machine side converter and
power will be delivered by the grid side converter with reduced amount of voltage and
current.
34
PU (KA)
0.0
-0.5
-1.0
-1.5
IA IB IC
1.5
1.0
0.5
LLG
PU(KA)
0.0
-0.5
-1.0
-1.5
IA IB IC
1.5
1.0
0.5
LL
PU(KA)
0.0
-0.5
-1.0
-1.5
IA IB IC
1.5
1.0
0.5
LG
PU(KA)
0.0
-0.5
-1.0
-1.5
Time 2.36 2.38 2.40 2.42 2.44 2.46 2.48 2.50 ...
...
...
For the purpose of this work and study of unintended islanding, an integrated model of the
PV solar in PSCAD presented in Figure 2.21 is considered.
The amount of power that can be taken from a solar cell depends on the operating point of
I_V cure which is maximum at the keen point of this curve as shown in the model. MPPT
(Maximum Power Point Tracking) is a power electronic DC-DC converter implemented to
36
ensure that the PV cell operated at maximum power point. Figure 2.22 presents the short
circuit contribution of PV array for a different type of fault. The contribution of the short
circuit current even for the three-phase fault is limited and in the instance of the fault is
very close to the load. However, the short circuit current if the fault persisted in the next
cycles could reach to 2 PU to 3 PU. The PV similar to type 3 and 4 wind turbines can
supply the grid with reactive power.
0.0 LLL
-1.0
-2.0
-3.0
IA IB IC
4.0
3.0
2.0
1.0
PU [KA]
0.0 LLG
-1.0
-2.0
-3.0
-4.0
IA IB IC
4.0
3.0
2.0
1.0
PU [KA]
0.0
-1.0 LL
-2.0
-3.0
-4.0
IA IB IC
2.5
PU [KA]
LG
-2.5
Sec 4.00 4.10 4.20 ...
...
...
The models developed for DER are based on the space phasor on αβ, and dq frames. A
three-phase positive rotation system can be shown as a single space phasor →f(t).
2π
Fb (t) = fmax cos (ωt + Ɵ0 − ) (2.4)
3
2π
Fc (t) = fmax cos (ωt + Ɵ0 + )
3
2π 4π
2
→(t)
F = [ej0 fa (t) + ej 3 fb (t) + ej 3 fc (t)] (2.5)
3
Where Ɵ0 , is the arbitrary initial angle of the three-phase system with the time origin.
Knowing,
2π 4π
Ej0 + ej 3 + ej 3 = 0
1
Cos θ = (ejθ + e−jθ )
2
Therefore,
→
f(t) = (fmax ejθ0 )ejωt
Figure 2.23 presents the space phasor representation of a 3 phase AC system in the
αβ and dq frame.
→(t)
F = fα + jfβ = (fd + jfq )e−jωt =(fd + jfq )ejρt
38
While 𝛼𝛽 frame is static and is not rotating, 𝑑𝑞 frame is locked with the rotation of space
→(𝑡)
phasor 𝑓 and therefore, the component in these frames is similar to the DC type
quantity. In (2.6) and (2.7), the 𝑑𝑞 quantities from three phase “abc” and vice versa are
calculated respectively. These calculations are known as the Park’s transformation.
2𝜋 2𝜋
cos 𝜃 cos (𝜃 − ) cos (𝜃 + )
𝑓𝑑 3 3 𝑓𝑎
2 2𝜋 2𝜋
[𝑓𝑞 ] = 3 sin 𝜃 sin (𝜃 − ) sin (𝜃 + ) [𝑓𝑏 ] (2.6)
3 3
0 1 1 1 𝑓𝑐
[ 2 2 2 ]
cos 𝜃 sin 𝜃 1
𝑓𝑎 2𝜋 2𝜋 𝑓𝑑
[𝑓𝑏 ] = [ cos (𝜃 − 3 ) sin (𝜃 − ) 1] [ ]
3 𝑓𝑞 (2.7)
𝑓𝑐 2𝜋 2𝜋
cos (𝜃 + ) sin (𝜃 + ) 1 0
3 3
The angle Ɵ(t) in the above transformation is estimated based on the angular velocity of
the grid by Phase Locked Loop (PLL) function. The function block diagram of PLL is
shown in Figure 2.24. Voltage Controlled Oscillator (VCO) in this diagram works as a
resettable integrator between 0 and 360 degrees. It will reset the value of Ɵ(t) when it
reaches 360 degree. In Figure 2.23, if PLL forces 𝑓𝑞 to zero at any given time, the “d”
axis in the “dq” frame will be in the same position of f(t) and therefore, the “dq” frame
39
will be locked to the space phasor rotation which generally represents the desired frequency
that should be measured.
Figure 2.25 shows the conceptual control block diagram of grid-imposed Voltage Source
Converter (VSC) illustrating the basic concept of the control of power models. The DC
source (VDC) in the case of type 3 and type 4 wind turbines is a simplified representation
of machine side convertor and in the case of a solar farm, represents a PV panel. The control
based on the 𝑑𝑞 frame is decoupled, i.e., there is a separate control loop for active and
reactive power. The estimation of grid frequency is an essential part of the conversion of
three phase AC system to the “dq0” stationary axis rotating with the angular frequency of
the grid source voltage Ɵ(t). In addition to “dq0“conversion, Ɵ(t) is used to adjust the
frequency of the grid side converter output. In simple terms, for the type 3 type 4 wind
turbines, and the PV solar farm, if there are no provisions for the islanding operation where
DER can supply the load in an islanded feeder, for the grid-imposed voltage source
converter without presence of an active source, the islanding mode cannot be sustained.
40
Figure 2.25. Real and Reactive- Power Control of Grid Imposed VSC
Figure 2.26 shows the simulations carried out for the generic PLL grid used for this study.
The first graph Ɵ(t) in degrees is the output of VCO, and the second graph is a derivative
𝑑𝜃(𝑡)
of the first graph ( ) without any limitation on the output value. The grid frequency at
𝑑𝑡
𝑡 = 1.5 𝑠𝑒𝑐 changes gradually to 58 Hz and at 𝑡 = 1.6 𝑠𝑒𝑐, the frequency is restored to its
original value of 60 Hz. The simulation is carried out for the grid side PLL used for type 4
machine.
The information related to the actual EMT models for DER used in this study are presented
in the Appendix C and Appendix D.
41
The contributions of DER in the regulation of voltage, reactive power; frequency, and
network inertia are the major changes that can been seen from the earlier version of the
standard. In the early generations of DERs, they were mostly unable to support islanding
operation since there was no capacity to generate var and to regulate the voltage and
frequency without an additional control circuit that supports this operation. With DER
getting the functionality that can support grid operation, IEEE 1547 scope has changed
from focusing only on a distribution network to covering both distribution and bulk energy
systems.
42
Figure 2.27. IEEE 1547 Grid Support Function History [Courtesy of IEEE 1547 WG]
In this regard, the standard has defined DER categories (Cat A, B) for the voltage
capabilities based on reactive power generation as a percentage of the power capacity of
DER for individual as well as aggregated units. The performance of the DER for an
abnormal operating condition have also been categorized (cat I, II, III), where Cat III is
specifically indented for bulk power systems.
Following are the main highlights related to islanding in the latest version of the standard
and have taken into consideration the following:
• Intentional island: one that is planned such that DER can carry a specific load
(e.g., microgrid, emergency/standby power supply). 1547-2018 now addresses
intentional islands.
2.7 Summary
A brief review of the conventional role of a distribution system as a power system interface
to the load center is discussed. It was shown that because of distribution system size, which
makes it a very capital- and labor-intensive business, and simplified functionality, which
was expected from this redial system, the technological disadvantage of the distribution
system infrastructure in comparison to the other sector of power systems was ignored by
utilities and system owners. Communication technology typically has not been part of the
solutions and research that have been offered. It is also discussed that distribution system
became the forefront of the green energy and smart grid initiative around the globe which
is transforming this system to be more technologically advanced, and hence, requires more
research and solutions similar to this work.
presented. The detailed EMT models DERs used in this work are presented in Appendix
D. In the next chapter, synchronized phasor measurement technique, signal processing,
phasor and frequency estimation, possible applications, and optimized system architecture
for the use cases studied in this work are presented.
45
Chapter 3
Phasor Measurements
With the many advancements in communication, hardware processors, and information
technology in the industry, these technologies are finding their way into power system
applications. Although reliability and performance remain the leading requirement for
power system applications for any new technology, in recent years, it can be observed that
the phasor measurement system has become a technique of choice for electric power
system utilities. The phasor measurement is a hardware of choice for this work which is
discussed in this chapter.
The minimum structure that is required for preparing the synchrophasor data for specific
applications consist of the following:
Phasor Measurement Unit (PMU): PMUs is a function or logical device that provides
synchrophasor and system frequency estimates, as well as other optional information such
as calculated megawatts (MW) and megavars (MVAR), sampled measurements, and
Boolean status words. The PMU can provide synchrophasor estimates from one or more
voltage or current waveforms [16]. The PMU can be realized as a stand-alone physical
device or as part of a multifunctional device, such as a protective relay, DFR, or meter. The
number of PMUs and locations where the PMU measures the electrical signal can vary
depending on the application requirement.
The introduction of PMU, which nowadays is heavily standardized in the industry, goes
back to the mid-1980s [17]. Since then, with the advancement in hardware computation
power, communication media, and availability of GPS around the globe (see Figure 3.2),
it became possible to use PMU as a standalone device or as a low-cost integrated function
in protection and control of Intelligent Electronic Devices (IEDs) by utilities in many
control, supervisory, and backup protection applications.
Time reference: The samples obtained by the PMU must be synchronized to a common
timing reference, so the angles of the phasors computed at different locations will be
comparable. As such, the electrical phasor which represents the analogue waveform of
current or voltage at different locations of the power system are synchronized by means of
Global Positioning System (GPS) to the Universal Time Clock (UTC) through
communication.
47
Synchronized clocks, providing precise timestamps for events, and data acquisition
applications on electric power systems. One way to provide precise time values is to use a
dedicated GPS receiver for every single device. But this is a costly solution. Alternatively,
time distribution mechanisms via dedicated buses or, for newer IEDs using Ethernet
connectivity can be implemented in practice. Each time synchronization method has its
own advantages and disadvantages and not all of them are optimal for use in substation
applications. Table 3-1 below, the most common time synchronization methods available
in the industry are compared for typical accuracy, data indication capability, dedicated
cabling requirements, cost effective implementation, and scalability. The methods are
briefly described here:
IRG-B- Inter Range Instrumentation Group time codes, also known as IRIG time codes,
are standard formats for transmitting time information. The original code formats were
described in IRIG document 104-60, and later revised several times over the years, with
the latest version being the IRIG Standard 200-04. IRIG codes B (IRIG-B time-codes) is
the industry standard for distributing synchronized time signals to IEDs. For time code
transmission, IRIG-B requires an external time source, such as a GPS receiver and a
dedicated twisted pair, coaxial cable or - fiber links. Therefore, this is not a low-cost
solution for time synchronization.
communication software. NTP is very robust, widely deployed throughout the Internet, and
well tested over the years and is generally regarded as state of the art in distributed time
synchronization protocols for unreliable networks. It can reduce synchronization offsets to
a few milliseconds over the public Internet and sub-millisecond levels over local area
networks. For best accuracy, the logical connection between the clients and servers should
be as short as possible.
IEEE 1588 V2: The IEEE 1588 Standard defines the Precision Time Protocol (PTP) for
packet-based networked systems. The time synchronization of IEEE 1588 protocol is
achieved by send message between master and slave docks. The clocks in the network are
divided into master and slave. Version 1 of the protocol was initially released in 2002 and
in 2008, was revised as Version 2. The first version does not support transparent clocks or
industry profiles and has larger packets that generate more traffic than the second version.
These two versions are not compatible. [19]
IRIG-B 100 µs x
1PPS 1µs x
Built in 1µs x x
GPS
NTP 1-10 ms x x
Phasor Data Concentrator (PDC): The phasor measurements are real time measurements
streamed by PMUs to the other devices, conventionally to the PDC, that work as a node in
a communication network where synchrophasor data from a few PMUs or other PDCs is
correlated and fed out as a single stream to the higher level PDCs and/or applications. The
PDC correlates synchrophasor data using a time tag to create a system wide measurement
set.
The term ej(ωt) will be suppressed knowing that ω is the frequency of the power system
and therefore, the sinusoidal is commonly shown as the phasor equation of:
Xm Xm
X=( ) eJ∅ = ( ) [Cos∅ + Jsin∅] (3.3)
√2 √2
Under this definition, φ is the offset from a cosine function at the nominal system frequency
synchronized to UTC. A cosine has a maximum at t = 0, so the synchrophasor angle is 0
degrees when the maximum of x(t) occurs at the UTC second rollover (1 PPS time signal),
and –90 degrees when the positive zero crossing occurs at the UTC second rollover (sine
waveform). Figure 3.3 shows the phase angle/UTC time relationship.
In the real world, however, the power system frequency will vary in time, and the current
and voltage are not an ideal sinusoidal waveform. Therefore, a more comprehensive
transformation from a pure sinusoidal signal to phasor representation, as the one discussed
51
earlier from frequency, amplitude, and waveform point of view will be discussed in this
chapter briefly.
Fourier transformation is used to break the measured signal into an alternate representation
characterized by summation of series sine and cosine. Equation (3.4) illustrates how an
arbitrary signal or measurement f(t) can be split into sine and cosine components using
Fourier transformation [21]:
a0 2πkt 2πkt
f(t) = + ∑∞
k=1[(a k cos ( ) + bk sin ( ))] (3.4)
2 T T
T
2 + 2πkt
2
ak = ∫ T f(t) cos ( ) dt, K = 0,1,2, , … (3.5)
T −2 T
T
2 + 2πkt
2
bk = ∫ T f(t) sin ( ) dt, K = 0,1,2, , … (3.6)
T −2 T
Figure 3.4 illustrates the same concept where the squared waveform signal with variable
frequency is transformed into Fourier components. The main components with
fundamental frequency in this transformation, i.e., the red color signal will be considered
if representation of such a signal into the phasor form is required.
52
This concept is illustrated in Figure 3.5 where the analogue waveform has been shown in
real time at intervals {0, T0, 2T0, 3T0,..,nT0, …}, where T0= 1/ f0 (the nominal power
system period) and the sequence corresponding phasor of these measurement are {X0,
X1, X2, X3, … Xn, …}.
53
If the phasor magnitude is constant, the phase angles of the sequence of phasors {X0, X1,
X2, X3, … Xn, …} will change at a constant angular velocity proportional to 2πΔf/T0.
Assuming these values are reported in real time, the phase angle will increase continuously
until it reaches 180 degrees where these would wrap around to –180 degrees, and continue
to increase, as shown in Figure 3.6. It should be noted that in synchrophasors, the angles
commonly reported are from –180 degrees to +180 degrees rather than 0 to 360 degrees.
The phasor of the main frequency component is estimated by use of Discrete Fourier
Transform (DFT) or Fast Fourier Transform (FFT).
The DFT is the equivalent of the continuous Fourier transform for signals known by only
N instants separated by sample time of ∆T. If f(t) is an arbitrary analog (continuous) input
signal which is the source of the data and f[0], f[2], …,f[N-1] are the N samples. The
Fourier transform of signal f(t) can be written as follows:
∞
F(jω) = ∫−∞ f(t)e−jωt dt (3.9)
If we regard each sample as an impulse having area f[k] which is f(k)*∆T, then, (3.9) can
exist only at the sample points:
F(jω) = ∑N−1
K=0 f[K]e
−jωK∆T
(3.10)
55
For fundamental frequency and its harmonic (including the DC components (𝝎=0) of the
original signal),
2π 2π 2π
Ω = 0, ,2 … , (N − 1)
NT NT NT
2π
−j nK
F(n) = ∑N−1
K=0 f[K]e N (n = 0 ∶ N − 1)
In principle, (3.10) can be used for any 𝝎 with only N samples. Sampling data from the
input signal is the start of the process to estimate the phasor. Over the years, several
considerations and techniques have been developed to digitalize the analog signal for
precise representation, which is described briefly in this section.
In the actual devices the high frequency components of the input signal that are not within
the Nyquist limit will be cut off during the signal processing by anti-aliasing filter.
𝑉 = 𝐴∡ φ
In power systems, however, amplitude (A), phase (φ), and angular frequency (𝝎) are time-
based variables. Therefore, the phasors for power systems should be estimated in time-
based processed signals, such as voltage and current, which will be estimated for a window
in a time-based manner and the estimation will be updated, as shown in Figure 3.9.
57
A few numbers of techniques have been developed over the years related to real time
phasor estimation. The short and long windows for phasor estimation are the main
categories that are known in this area. The short windows algorithm, such as Miki and
Mikano, Mann and Morrison, Rockefeller and Urden, and long windows algorithm, such
as DFT, LES, can be noted.
V0 = Vp cos(∅) (3.12)
V−1 = Vp cos(−ω∆T + ∅)
From equation (3.12) and (3.13), real and imaginary parts of the phasor can be calculated,
and accordingly, their magnitude and angle :
sin(ω∆T)
∅ = tan−1 (3.15)
𝑉0
j2π
2 kn
Xk = ∑N−1
n=0 X[n]e N k = 0, . . , N − 1 (3.16)
N
j2π
kn
X[n] = ∑N−1
n=0 X k e N n = 0, . . , N − 1
Φk = tan−1 (Xk )
j2π
2 n
X1 = ∑N−1
n=0 X n e N k = 0, . . , N − 1 (3.18)
N
1 2π 1 2π
X1 = ∑N−1
n=0 X n cos ( n) + j ∑N−1
0 Xn sin ( n) n = 0, . . , N − 1
N N N N
1 2π 2 1 2π 2
A1 = √{ ∑N−1
n=0 X n cos ( n)} + {j ∑N−1
n=0 X n sin ( n)} n = 0, . . , N − 1 (3.19)
N N N N
2π 2π
∅1 = tan−1 (∑N−1
n=0 X n cos ( n) + j ∑N−1
0 Xn sin ( n)) n = 0, . . , N − 1
N N
In recent years, much research has been done to improve phasor estimation and how the
estimation should be updated for the real time application. The windows of data acquisition
must move forward with the objective of using phasor-based measurement for transient
study, fast power system phenomenon, such as protection. In [22], the authors propose an
improved DFT to immunize the accuracy of DFT phasor estimation from the presence of
DC component (short circuits scenarios). In [23], a dynamic current phasor measurement
is presented to deal with the noise in current signals. In [24], a method to estimate the
phasor for off-frequency based on Taylor series is proposed. In [25], [26], and [27] more
studies have been done to present phasor estimation for protection and time critical
application.
reasons behind the estimation of frequency is the fact that if the frequency is not known,
the phasor estimation itself is not accurate.
Zero Crossing: This method monitors timing between the wave from zero crossing in the
negative and positive half wave in order to estimate the frequency of the phasor in the
predetermined time interval.
All message frames start with a 2-byte SYNC word followed by a 2-byte FRAMESIZE
word, a 2-byte IDCODE, a time stamp consisting of a 4-byte Second-Of-Century (SOC),
and 4-byte FRACSEC, which includes a 24-bit FRACSEC integer and an 8-bit Time
Quality flag. All frames are transmitted exactly as described with no delimiters. Figure
3.12 illustrates this frame transmission order. The SYNC word is transmitted first and
CHECK word last.
For time critical applications, time delay and device performance are important.
Table 3-3 presents the typical range of time delay in processing and communication that
can be expected for typical PMU based applications.
62
μs μs
Communication distance 3.4 to 6
km km
In line with hardware advancements that have been made in the last couple of decades in
terms of how the PMU and PDC perform, the communication media, such as LAN (Local
Area Network), Ethernet-based communication with speed of 1.00GB/sec or faster, have
replaced serial type communication in utilities-based applications. In this regard
compliance of Phasor measurement data with IEC IEC61850-90-5 standard created the
possibility of using time critical communication services, such as GOOSE message
(General Object-Oriented Substation Event) and Sample Value (SV), that is defined by this
standard series and managed IEC Technical Committee 57 (TC57) to be available for time
critical application such as protection in substation . IEC/TR 61850-90-5 [30] provides a
way of exchanging synchrophasor data between PMUs, PDCs WAMPAC (Wide Area
Monitoring, Protection, and Control), and between control center applications. The data,
63
to the extent covered in IEEE C37.118-2005, is transported in a way that is compliant with
the concepts of IEC 61850. However, given the primary scope and use cases, this document
also provides routable profiles for IEC 61850-8-1 GOOSE [31] and IEC 61850-9-2 SV
[32] packets which can be used to transfer synchrophasor data using GOOSE (General
Object-Oriented Substation Event) message over WAN (Wide Area Network). To
summarize the rule and relation of IEC and IEEE standard in the context of synchrophasor
measurement and application IEEE C37.118.1 remains the global standard for defining the
measurement technology for synchrophasor while IEEE C37.118.2 is the IEEE protocol to
address current system requirements enabling IEC TR 61850-90-5 to be the basis for a
more scalable, and secure, protocol to meet application requirements.
Conventionally, the distribution systems were assumed to be simple and posed little need
to be observed with granularity in space and time. The measuring sensors implemented in
the feeder almost uniquely were limited to the substation location only. Most of the control
and protection applications have been developed based on the accessibility of
measurements only back at the substation. However, with the transformation of the energy
sector and rapid growth in deployment of distributed energy resources, bi-directional
electricity flows, and new devices, such as electric vehicles, there is a growing interest and
requirement for observation tools along the feeder for significantly impacted distribution
feeders. Therefore, PMU, as an integrated function in IED, can play an important role in
redefining many distribution management system functionalities which were based on
limited data. In the current work, islanding, and open phase fault in the distribution
overhead line are targeted and developed.
The availability of PMU data with the higher resolution can potentially increase the
sensitivity of the proposed solutions however, the amount of data that must be transferred
cannot be handled by the current available market devices and communication
infrastructure. The islanding application in this work is intended to be used by utilities as
backup for the local islanding protection scheme and to replace the transfer trip. Therefor
this application is not intended for Microgrid or small size DG applications (2MW and
above) and as such the high-resolution PMU will not provide its full benefit.
3.8 Summary
The fundamentals of synchronized phasor measurement system architecture were
presented in this chapter. The functionality of phasor measurement unit (PMU), data
67
concentrator, and GPS were discussed, and the history of the early PMU and advancement
of phasor measurement technology was presented. It was shown that along with IEEE
C37.118 restructuring in recent years, the technology as well as the standard, are more
prepared to support the real time application of control and protection.
The flow of analogue signal and its process in PMU hardware was also studied in this
chapter. Various phasor and frequency estimation techniques were looked. The
communication aspect of synchronized phasor measurement unit and phasor data
concentrator was studied. The new communication standard of IEC61850-90-5 and the
possibility of the integration of the synchrophasor data stream with the substation
automation data was presented. The implication of using of IEC61850-90-5 and GOOSE
messages in peer to peer communication between the PMUs and the elimination of PDC,
especially in the distribution systems, was discussed. Measurement was discussed in this
chapter as well. The motivation behind the call for use of synchrophasor data (by this work)
in the distribution protection application using substation automation infrastructure was
presented.
The development of μPMU with high accuracy and high resolution was reviewed, along
with the difficulty of measurement in distribution systems and the accuracy required for
them. The possible conceptual application of high performance synchrophasor is reviewed.
In the next chapter, the islanding detection use case for distribution feeder, including the
background, critical review of the state of the art, issue formulation, development of a
solution, and modeling and test system will be discussed.
68
Chapter 4
Adaptive Islanding Scheme
In this chapter, the proposed islanding detection method for distribution system is
presented.
4.1 Introduction
Utilities’ main concern behind islanding operations in the distribution network is safety
and the lack of adequate infrastructure that can monitor and control the operation of DGs
within the islands in a reliable manner. Thus, in the current stage of distribution
modernization, it is broadly taken for granted that the island is an unregulated power system
that has behaves unpredictably. Voltage and frequency in the islanded area can
significantly deviate from the acceptable range since utilities have no means to curb the
power mismatch between the DG production capacity and the load(s) connected to the
island.
One of the most commonly used passive islanding methods consists of detection of under-
, over-, and rate of frequency variations. The primary concerns and limitations of the local
passive detection methods are that these methods cause an operating region where, in that
specific region, islanding conditions cannot be found or detected in a timely manner. This
region is known as the Non-Detection Zone (NDZ). Similar NDZ regions can be identified
among many passive detection techniques which have been the topic of much research.
There is no single passive method that can be effective in all scenarios and the Power
Systems Community is undecided on what type of islanding detection should be used. For
example, IEEE standards 1547-2003 and 929-2000 specify the performance characteristics
of the islanding detection methods with detailed test circuits that can be used to validate
the method considered. Issues related to the passive islanding techniques can be
summarized as follows:
3) Most of the work in this area has been conducted around low DG penetration or
single DG island detection cases, while higher DG penetration cases can further
enlarge the NDZ and affect the security and dependability of the schemes.
4.2 Islanding
Prior to reviewing islanding detection techniques and their significance, it is appropriate to
define the term islanding. Islanding is a generic term used to describe a scenario in which
a section of a transmission, or distribution network (which contains DG) is separated from
the rest of the grid. This separation is often caused by the action of the protective relays to
clear and isolate the electrical fault. Subsequent to this separation, the DG restarts or
continues to power the loads trapped within the island [40]. Figure 4.1 illustrates a typical
North American distribution feeder with a few DGs, a step-down substation with several
outgoing distribution feeders, and one of the outgoing distribution feeders shown in detail.
An islanding situation occurs, for instance, when recloser C opens while DG1 is still
feeding the load and an island is created as a result of the recloser operation. The utility's
main concern behind the islanding operation in the distribution network is safety and the
lack of adequate infrastructure which can monitor and control the operation of DGs within
the islands in a reliable manner. Thus, in the current stage of distribution modernization, it
is broadly taken for granted that the island is an unregulated power system. Its behavior is
unpredictable and voltage and frequency in the islanded area can significantly deviate from
the acceptable range, since utilities have no means to curb the power mismatch between
DG production capacity and load connected to the island. The main concern for such an
operation among utilities can be listed as follows [41]:
1) The quality of power fed to an islanding portion of feeder may be lower compared
to when the power is supplied by the utility. The range of voltage and frequency in
the islanded portion of the feeder is a main concern since the supply utility is no
longer controlling the voltage and frequency delivered to their customers and any
excursion from expected boundaries of voltage and frequency can cause
considerable damage to customers’ equipment within the island.
2) An islanding operation may also create a hazard for line-workers or the public by
causing a line to remain energized that may have been assumed to be disconnected
from all energy sources.
3) The likelihood of the islanded portion not being in phase with the network voltage
and phases, at the instant when the islanded portion is reconnected through
reclosing or an automation, is a real concern. This can damage the generating
equipment and DGs in the island. Ultimately, such an attempt of restoration may
fail.
4) Islanding may interfere with the manual or automatic restoration of feeder or cause
issues for neighboring customers and can complicate loop operation of feeders.
71
The PMU based islanding proposed by this work is a hybrid solution which does not fit in
one category since it uses combination of techniques, as follows:
• The measurement in this solution is a passive type and active power and
frequency are monitored.
• The measurement process relies on the local PCCs and remote communication
therefore communication is an important part of this solution.
72
Issues related to this method are reported in various literature and can be summarized as
follows:
1) Cost is the major concern here, as the Power Line signal must be transmitted in all
three phases in order to detect single phase islanding.
2) The transmitted signal must be re-attenuated along the way if the distances from
the station to the DG or Point of Common Coupling (PCC), are longer than 15 Km
[46].
3) The reliability of this method, with the presence of inter-circuit harmonic pollution,
is another concern which is solicited in different research work.
The generation profile of the DG varies during the day and in many operational instances,
the islanding scheme will not work. Besides frequency, other power quantities can also be
used to help detect island situations such as:
1) Voltage based detection
2) Power factor (P/Q)
3) Change of active power
4) Change of reactive power
5) Change of total harmonic distortion (THD)
6) Built in inverter-based islanding techniques
Reference [47] proposes the use of reactive power rate of change to detect the islanding.
The method can be useful for relatively when a large amount of DG is integrated to grid
and issue of NDZ has not much of importance. In recent years many hybrid methods based
on combination of the passive techniques mostly integrated into the inverter-based devices
are developed. [48] proposes monitoring of voltage and current magnitude together with
current and voltage THD at PCC.
Many technical problems need to be solved before one can use them with confidence. Some
of the issues related to active methods are reported in various literature [40], [46], and are
as follows:
1) One of the main problems of the active methods is the interference of disturbances
introduced by multiple DGs. Not much research has been conducted on such issues.
2) The type of active islanding detection method which can be deployed is very much
dependent on the type of DG installed in the network. The design of a universal
active method solution which can cover a range of installations is very difficult if
not impossible. Therefore, this scheme generally has low adaptability.
3) Generally active methods can have a negative impact on the grid power quality
compare to the passive methods.
DG2
P2,Q2
PM
U
PG,QG
Grid
DG1
P1,Q1 PM
U
PM
U
PD
C
The islanding event is modeled as a small power perturbation where the primary feeder
losses the grid supply and the feeders are then supplied power by the DGs.
S Laplace operator
The proposed method uses continuous measurement of active and reactive power at the
substation (supply by grid) and the distributed generation (supply by DGs) prior to the
islanding event to determine the power mismatch between the load of the feeder and
generation located at the feeder. Based on this information, the sensitivity of the
aforementioned protection elements that detect the islanding will be adapted accordingly.
If we consider the power loss as a part of the feeder load change, the following can be
stated at the instance of islanding event, 𝐏𝐆𝐫𝐢𝐝 = 𝟎 considering the generator response to
speed change equation (4.1) as follows:
1
∆Tm − ∆Te = ω (4.1)
2HS
Therefore, the system response to load change ( 𝐏𝐆𝐫𝐢𝐝 = 𝟎 ) is determined by the machine
inertia constant.
For the load step load change equal to equal to −ΔPGrid and from (4.1) and (4.8)
1
-(-ΔPGrid ) = ( ) Δω (4.9)
2HS+D
ΔPGrid 2H
∆𝜔ss = , τ= (4.10)
D D
df ∆fss
≈≥ 0.632 (4.11)
dt τ
From (4.10) and (4.11), the final frequency deviation and time of the transient and the
transient change during the islanding event can be estimated. For a group of DGs, the same
can be concluded, except that H must represent the total feeder inertia. For the inverter-
based type DGs, such as type 3, type 4, and PV, where no inertia is connected to the grid,
the voltage and reactive power mismatch must be taken into consideration. Similarly, for
reactive power, following can be stated:
For completely balanced islanding, where prior to the islanding event no power mismatch
is measured, the sensitive change of voltage angle differences with some security measures
are considered. The phase angle differences are used to shrink the NDZ where the
sensitivity of the 27, 59, and 81 elements are not adequate to detect the separation of the
primary feeder when there is no power mismatch.
81
∆φ ∆(∡Vpcc −∡VGrid )
= (4.16)
∆t ∆t
∆P ≈ 0,
∆Q ≈ 0 (4.17)
In addition to the initial static conditions, mentioned in (4.10) and (4.11), to enable the
phase angle supervision, the dynamic conditions of such a supervision must also be taken
into consideration. It is important to note that any electrical fault, load, or capacitor bank
switching may trigger a sensitive phase angle supervision; therefore, at the instance of
islanding detection, the feeder must be mostly free of any switching and changes.
∆I2
≅0 (4.18)
∆t
Where I2 is the negative sequence component for both grid and DGs.
interface and they add to the inertia of the system. It should also be noted that a
mixed generation (inverter based and synchronous machines) is considered in this
category. Inverter-based machines will follow the synchronous machine frequency
response during the islanding. Three cases in this scenario is reported.
b) Scenario 2: In this scenario, the variable speed (inverter based) DGs are being
studied. These units are connected to the grid through the inverter, wind turbine as
follow:
a) Wind turbine type 3: case 1 and case 2
b) Wind turbine type 4: case 1 and case 2
c) PV solar: case 1 and case 2
c) Scenario 3: In this scenario, the effectiveness of proposed solutions in non-
detection zone where the power mismatch is almost zero is considered.
Figure 4-8 presents the results of three cases where islanding is performed with the
different active power mismatch at t=40 sec. It is assumed that for any given simulation
that the generator cannot deliver more power than what it is already providing. The
governor gate valve is set to the maximum of its opening to limit the reaction of frequency
load compensation. The plot presents DG frequency (fDG), DG active power (PDG), grid
active power (PGRID), DG reactive, power (QDG), and grid reactive power (QGrid).
Feeder load is the same for all the three cases simulated while contribution of DGs in
supplying active and reactive power are different.
84
Graphs
case3.inf::Main\Freq_B... case4.inf::Main\Freq_B... case2.inf::Main\Freq_B...
1.05
1.00
HZ [PU] 0.95
0.90
0.85
0.80
case3.inf::Main\PDG case4.inf::Main\PDG case2.inf::Main\PDG
1.10
MW [PU]
1.00
0.90
0.80
case3.inf::Main\PGrid case4.inf::Main\PGrid case2.inf::Main\PGrid
0.20
MW[pu]
0.10
0.00
-0.10
x 50.00 60.00 70.00 80.00 90.00 100.00 110.00 120.00 130.00 ...
...
...
Figure 4.8. Islanding Scenario of Synchronous Machine
Table 4-3 presents the on-line estimation of ∆𝐟𝐬𝐬 and 𝛕 based on (4.11) and power mismatch
𝐝𝐟𝐆
measured during the simulation. The frequency rate of change can be directly calculated
𝐝𝐭
from the value estimated for each simulation and linearization of the change during the
interval of one time constant.
where the ∆fGss is a steady state value of frequency deviation. It should be noted that online
estimation is only valid prior to the islanding event. If the D value is known or estimated
correctly, online estimation will generate an accurate result that can serve to set the
frequency element(s) of islanding detection system. Table 4-3 shows the results of the
frequency deviation and time constant of the frequency settlement for the cases that have
85
been simulated based on the concept of the active power mismatch presented earlier. The
dfG
ratio of estimated based on the power mismatch prior to the islanding and can be
dt
dfG
compared with the actual average values of measured. The results in all the cases shows
dt
that estimation value is more conservative than the actual rate of the change and protection
frequency element which adapted to this estimation, can trip for such a rate of the change
in all the cases.
In any protection adaptive setting, the setting should be changed only within the predefined
limitation that has to be set up based on the actual application data.
H [sec] Inertia 3
86
For the purpose of simulation, it was assumed that pitch control was at its optimal position
and remained unmoved after the initial model activation, that maximum power was
obtained from the wind, and that the wind speed remained constant during the simulations.
Prior to the islanding instant at t =2 sec, the load active power is totally compensated by
the wind turbine and the contribution of the grid for the active load is almost zero. This
should have created a most favorable situation to sustain the islanding operation by wind
turbine. In case 1, the wind turbines are in under-excitation mode and are consuming
reactive power while in case 2 the wind turbine is in over-excitation mode and is generating
the reactive power.
Figure 4.9. Wind Turbine Type 3 Islanding Event with Balance of Power Case 1
87
Figure 4.10. Wind Turbine Type 3 Islanding Event with Balance of Power Case 2
The simulation results show that with the loss of a self-regulated power frequency source
(grid), the phase locked loop control that controls the DG frequency by following the grid
frequency becomes unstable and enters to a self-excited and unstable loop resulting in a
major deviation from the grid frequency. The study also reveals that even a zero-power
88
mismatch can lead to significant power frequency deviation (∆𝑓). A greater power
mismatch than the balance of power also cannot be sustained by a type 3 machine and the
island will be detected up by voltage and frequency elements at PCC or by the machine’s
internal protection.
Among that many cases which have been studied, the following two cases in Figure 4.12
and Figure 4.13 serving as the representative cases. Table 4-5 shows machine data for the
type 4 wind turbine used in this study.
89
H [sec] Inertia 2
For the purposes of simulation, it was assumed that pitch control was at its optimal position
and remained unmoved after initial model activation, that maximum power was obtained
from the wind, and that the wind speed remained constant during the simulations. Figure
4.12 and Figure 4.13 present the results of the simulations for a group of 10 type 4 wind
turbines connected to the feeder at one location. The islanding occurs at t= 2 sec. The power
mismatch in both cases is very small and wind turbines were able to sustain the load after
islanding provided that the Phase Locked Loop (PLL) was designed for islanding
operation. The simulation results show that with the loss of a self-regulated power
frequency source (grid), the phase locked loop control that controls the DG frequency by
following the grid frequency becomes unstable and enters into a self-excited and unstable
loop resulting in a major deviation from the grid frequency. The study also reveals that
even a small or no power mismatch can lead to significant power frequency deviations
(∆𝑓). A greater power mismatches cannot be sustained by a type 4 machine and island it
will be picked up by voltage and frequency elements at PCC or by the machine’s internal
protection. In case 1, the wind turbines were in under-excitation mode, while in case 2 they
are in over-excitation mode. In both cases, the active power mismatch is near zero. The
deviation from the rating frequency is very large and in a few cycles after the instant of
islanding takes place.
90
very similar with very small differences in power mismatches prior to the islanding event.
When solar PV is not prepared for the islanding operation as discussed earlier, in the case
of wind turbine Type 3 and Type 4, the PLL does not sense any independent frequency
after unintentional islanding, and frequency becomes unstable.
∆φ = ∡VGrid − ∡VPCC
This supervision can shrink the NDZ which commonly exists for all passive islanding
detection methods. The high sensitivity of this function is the essence of utilizing this
supervision in NDZ. Cases 1 to 3 are highly balanced scenarios where the voltage angles
differences 2 seconds after the islanding events, change by a very small value.
The simulation results are shown in Table 4-6. The power mismatch is expressed as a
percentage of feeder real time load. The negative value of power mismatch means that the
DG, in addition to supplying the feeder load, is also exporting small amount of power to
95
the grid. With the setting of 2 degrees for the voltage angle supervision between PCC and
substation, the time required to detect the islanding for each case are shown. The results of
this study show that NDZ area, with the given supervision setting, can be reduced to
maximum 1.5% (case 2) of the feeder load. The islanding for case 3, is detected after 2
seconds which does not meet the IEEE 1547 requirement. The phase angle supervision,
like any other sensitive function in the field of protection, may provide good dependability;
however, it always lacks good security performance. The reliable scheme, however, is a
right balance of two properties dependability and security. Therefore, (4.22) and (4.23) are
the security condition that must be considered to activate the supervision function to ensure
that during the transient (fault, switching, etc.) where the phase angle can change this
function is disabled. The following can be formulated as the sensitivity limit in terms of
frequency for the proposed solution and the cases studied:
dv2DG
≅0 (4.22)
dt
dI2DG
≅0 (4.23)
dt
It is important to highlight that the sensitive islanding detection solution proposed in this
section relies on IEEE PMU with a 1-degree phase angle resolution. Thus, if we assume
𝜇PMU or high resolution PMU is used, the limit, proposed in (4.20) and (4.21) can
96
theoretically be reduced. For example, considering 0.01° phase angle resolution and
hypotheses of selecting 0.5 degree as angle supervisions setting, the results of the previous
cases will change to what is shown in
Table 4-7. The more sensitive phase angle measurement and threshold theoretically can
improve the sensitivity bottom line. As an example, case 3 which was not detectable in
previous evaluation within 2 seconds (IEEE 1547 requirement) now it can be detected
within 1.363 seconds.
With lower and more sensitive setting for phase angle measurement, supervision to
maintain the security of the proposed solution will be much harder to maintain if it is not
impossible. It should be noted that (4.20) and (4.21) are not the only precondition for the
activation of phase angle supervision. This condition should be maintained during the
entire islanding detection process. Figure 4.17 shows the impact of the transient three-
phase short circuit on the voltage phase angle measurement at the adjacent feeder at t= 30
sec. As suggested in (4.22) and (4.23) and shown here, negative sequence component of
the DG current and voltage can be used to inhibit the phase islanding detection based on
voltage phase angle. Therefore, it must point out that the sensitive islanding detection
method cannot be reliably used without a period of stabilizing where no fault and switching
operation in the feeder is detected. Furthermore, this condition must be present during the
entire islanding detection process. The phase angle supervisions setting, it may differ from
on feeder to another depend the load minimum feeder and DG size.
97
when a more sensitive set of settings cannot be calculated. The communication system
architecture for the proposed solution is illustrated in Figure 4.6. The utilization of PDC,
particularly for the size of suggested application, using IEC61850-90-5 where the PMUs
can directly use the data stream from each other with GOOSE PDU can be eliminated. This
can further improve the time performance of the proposed solution. This paper presented
some comprehensive islanding scenarios with their relevant analysis of active and reactive
power mismatches and a smooth balance power islanding in the distribution type feeder.
The EMT-based detailed modeling using PSCAD of PV, type 4, and synchronous machine
is used to study the transient of this phenomenon in an actual utility type feeder, load, and
network.
It is important to note that the proposed solution works in conjunction with passive local
islanding detection, i.e., protection elements such as 81 and 27 located at PCC (local anti-
islanding elements). The proposed solution will adapt the setting of frequency elements
from a base setting, considered to be the utility standard setting, to the more sensitive one
based on the power measured during the operation. The power measurement will only be
considered if it has been measured during a no-fault situation. For this reason, the moving
average of the power mismatch between the grid and summation of all the DGs within a
selected time interval must be measured and continuously updated until the fault is
detected. During the fault, the moving average must not be updated and if this fault results
in an islanding event, then the last moving average of the power mismatch must be used
for this solution.
4.12 Summary
In this chapter, the first use case for detecting unintentional islanding operation in
distribution systems with DER using the synchrophasor data was studied.
The issue related to the islanding operation was discussed, and the state-of-the-art detection
methods were critically reviewed. Then, an adaptive detection solution was proposed to
augment the exiting anti islanding protection scheme. The solution was formulated and
presented. The concept of the proposed solution is based on measuring the power mismatch
between the grid and DG in non-critical time and prepare the setting or response in real
time.
The provision of the solution for zero power mismatch and non-detection zone was
analyzed and proposed. The proposed method was further developed to include a feeder
with multiple integrated PCCs.
The mathematical formulation was developed and presented. The test system was
developed by detailed modeling of the utility type distribution feeder and the complex
modeling of aggerated DG type 3, type 4, and PV in EMT using PSCAD/EMTDC software.
The test scenario to examine the reliability of the proposed solution was developed with
emphasis on dependability and security. From the many simulations that were carried out,
selected representative cases were reported and analyzed.
The conceptual implementation for the proposed solution with consideration of the marked
available hardware and software was also proposed.
The next chapter will study the second use case using the synchrophasor data to detect the
open phase fault in the distribution feeder.
100
Chapter 5
Open Phase Fault Detection
In this chapter, the summary of research and analysis leading to the proposal of selective
detection of the open phase fault as well as the results of EMT modeling and simulation
are presented.
5.1 Introduction
With the prospect of integration of many Distributed Energy Resources (DERs) into the
electrical grid, especially in the distribution network, coupled with the concept of smart
grid, there is a necessity for more investment in communication infrastructure to operate
such a system while maintaining the safety and reliability of the grid. Even though the use
of Intelligent Electronic Devices (IEDs) enhances the reliability of protection and control
by improving dependability, i.e., the ability to detect the fault, security, and differentiate a
normal situation from the fault.
A typical distribution network has many miles of overhead lines, still protected by fuses.
The protection relays are installed in the substation and their algorithm and performance
relies mainly on local measurements. In this chapter, in line with the distribution network
transformation and requirement of grid modernization for communication infrastructure,
open phase fault detection method based on phasor measurement data from the substation
and every Point of Common Coupling (PCC) in the feeder (where the DER is connected to
the distribution feeder) is presented.
The open-phase or broken conductor fault is a challenging fault for utilities to detect, as
there is no dedicated protection element to identify this fault. Open phase fault often
coincides with a broken and downed conductor to earth, which is a public safety risk. If
this event develops into a ground fault, (a high impedance fault with a very low current), it
becomes hard to detect. Most importantly, for the distribution feeder with an auto-recloser,
without identifying this type of fault as an open phase, and therefore, blocking the recloser
attempts, the feeder may get reenergized and intensify the risk of electrification to the
public.
101
The undetected open phase during a light load condition can cause ferro resonance and an
increase in transient voltage in the feeder. Figure 5.1 presents a conceptual architecture of
proposed solution which shows a set of three-phase current and voltage phasor data from
a feeder breaker at a substation and point of common couplings (PCC1 and PCC2) obtained
from PMU 1 to PMU3 (Phasor Measurement Unit), respectively.
Primary
Feeder
V, I
Substation
PCC1
Grid
V, I
V ,I
PCC2
The data from the PMUs can be processed in Phasor Data Concentrator (PDC) as shown
here, assuming a conventional C37.118 communication is used. Here, the PMU data stream
needs to be aligned in the PDC before being used by application.
there is no other voltage source in the feeder, is essential for any algorithm development.
In [54], [55], and [56], the impact of open phase fault in distribution and sub-transmission
network, Temporary Over Voltage (TOV) and possibility of resonance circuit in resonantly
grounding system is verified. The open phase fault can be very detrimental to the health of
the equipment. In [57], the vulnerability of a Negative Sequence Pilot Protection (NSPP)
scheme for a very long transmission line is verified and a compensation method based on
the open phase symmetrical component analysis is introduced. In [58], it is proposed that
the zero-sequence voltage be measured by electric field sensors alongside of feeder. The
criteria for detection are relying on the fact that unbalanced voltage after the fault will be
much higher than the normal unbalanced operation of the feeder. This model has been
verified in the field; however, open phase detection in the presence of DG will not have
the same signature, and therefore, a decision-making algorithm may not apply to a feeder
with DG. In [59], the focus is on detecting the high impedance fault in distribution feeder
caused by a broken or downed conductor. The characteristic of this fault with consideration
to the harmonic content and the current waveform is verified. In [60], the rule-based fault
detection method, including the open phase fault, is compared with the Artificial Neural
Network (ANN). In [61] and, [62], the use of computer-based modeling of open phase and
fault analysis has been reported.
PQ Q
Va = VaP − Va ≠ 0
PQ Q
Vb = VbP − Vb = 0
PQ Q
Vc = VcP − Vc = 0
Ia = 0, Ib ≠ 0, Ic ≠ 0
103
PQ
Where Va , Ib , and Ic ,represent the voltage across the P and Q and phase currents
respectively.
P Q
Ia
a
Ib
b
Ic
c
The symmetrical component of the above circuit and group equation can be summarized
as follows:
PQ PQ PQ PQ
V1 = 1/3(Va + aVb + a2 Vc ) (5.1)
PQ PQ PQ PQ
V2 = 1/3(Va + a2 Vb + aVc ) (5.2)
PQ PQ PQ PQ
V0 = 1/3(Va + Vb + Vc )
Therefore,
PQ PQ PQ 1
V1 = V2 = V0 = PQ (5.3)
3Va
(1 + a + a2 ) = 0
I1 = I2 + I0 (5.4)
Considering (5.1) to (5.4) , the equivalent circuit for the open phase fault is shown in
104
Figure 5.3.
P1 P2 P0
Z1
Z2 Z0
Q1 Q2 Q0
P
P Q Q
Ia Ia =Ifg
a
Ib P
Rfg Ib
Q
b
Ic P Ic
Q
The following can be stated for voltage and current in P and Q side.
105
Ia = 0
PQ Q
Vb = VbP − Vb = 0
PQ Q
Vc = VcP − Vc = 0
VbP = a2 Va1
P P
+ aVa2 P
+ Va0
Q Q Q Q
Vb = a2 Va1 + aVa2 + Va0
P P P
VcP = aVa1 + a2 Va2 + Va0
Q Q Q Q
Vc = aVa1 + a2 Va2 + Va0
PQ Q Q Q Q
Vb = VbP − Vb = a2 (Va1
P P
− Va1 ) + a(Va2 P
− Va2 ) + (Va0 − Va0 ) = 0 (5.5)
PQ Q P Q P Q P Q
Vc = VcP − Vc = a(Va1 − Va1 ) + a2 (Va2 − Va2 ) + (Va0 − Va0 ) = 0 (5.6)
We know that
a2 + a + 1 = 0 (5.7)
Therefore, from (5.17), (5.6) and (5.7), the following can be concluded:
P Q P Q P Q
(Va1 − Va1 ) = (Va2 − Va2 ) = (Va0 − Va0 ) (5.8)
Eq. (5.8) can be also developed for the healthy phases “b” and “c”:
PQ Q
Ib = IbP − Ib = 0
PQ Q
Ic = IcP − Ic = 0
PQ Q Q Q Q
Ib = IbP − Ib = a2 (Ia1
P P
− Ia1 ) + a(Ia2 P
− Ia2 ) + (Ia0 − Ia0 ) = 0 (5.9)
106
PQ Q P Q P PQ Q
Ic = IcP − Ic = a(Ia1 − Ia1 ) + a2 (Ia2 − Ia2 ) + (Ia0 − Ia0 ) = 0 (5.10)
Eq. (5.11) similar to (5.16) can be extracted from (5.9) and (5.10).
P Q P Q P Q 1
(Ia1 − Ia1 ) = (Ia2 − Ia2 ) = (Ia0 − Ia0 )= Ifg (5.11)
3
Based on (4.20) and (5.11), the equivalent circuit of the single open phase fault with a
downed wire to ground in the interconnected network (source in both side of the fault) is
shown in Figure 5.5. The zero-sequence component of current caused by the ground fault
at Q is shown by use of an ideal 1:1 ratio interpose transformer connecting the sequence
component circuits together. The ground fault resistance (R fg ) in real cases most likely will
be a high impedance. Eq. (5.11) is representing the ground fault zero sequence component.
I1P
I1P
Q
P1 I1P +I1 1/1
Q1
Z1P Z1
Q
~ V1P V1
Q
Q
~
I1
I2P
I2P Q
I2P +I2
P2 Q2 1/1
Ig0
Z2P Q
Q
Z2
V2P V2 Q
I2 I0P
I0P
Q
I0P +I0
Q0 1/1
P0 3Rfg
Z0P Q Z0
Q
V0P V0 Q
I0
The circuit can be simplified further if we consider the radial system presented in Figure
5.2. Such cases can frequently occur in the distribution feeder and thus, it is worthwhile
that the summary of the analysis is presented as an equivalent circuit with the ground fault
at the source side (bus side), as shown in Figure 5.6.
Q
I1
I1P
Ig Q1
P1
1/1 Q
Z1P Z1
Q
~ V1P V1
I1
Ig
I2P I2
Q
Ig Q2
P2
1/1 Z2P Q
Z2
Q
V2P V2
I2
Q
I0P I0
Ig Q0
P0
3Rfg
Q
1/1 Z0P Z0
V0P Q
V0
I0
Figure 5.6. Open Phase Fault with Ground at the Source Side
A similar relationship between current and voltage can be developed for an open phase and
the downed wire at “Q” side (load side) of the circuit. In this case, the 1/1 ratio interpose
transformer, which is used to represent the zero-sequence current caused by the ground
fault, must be moved to the load side (Q side) of the circuit.
108
P Q
Ia
a
Ib
b
Ic
c
Updating (5.1), (5.3), (5.4) and (5.11) from single open phase to double open phase, the
following can be stated for voltage, current, and sequence components:
PQ Q
V𝑎 = V𝑎P − V𝑎 = 0
PQ Q
Vb = VbP − Vb ≠ 0 (5.12)
PQ Q
Vc = VcP − Vc ≠ 0
PQ PQ PQ PQ PQ PQ
V1 = 1/3(Va + aVb + a2 Vc ) = 1/3(aVb + a2 Vc ) (5.13)
PQ PQ PQ PQ PQ PQ
V2 = 1/3(Va + a2 Vb + aVc ) = 1/3(a2 Vb + aVc ) (5.14)
PQ PQ PQ PQ PQ PQ
V0 = 1/3(Va + Vb + Vc ) = 1/3(Vb + Vc ) (5.15)
PQ PQ PQ 1
V1 + V2 + V0 = (1 + a + a2 )(VbPQ + VbPQ )
3
PQ PQ PQ
V1 + V2 + V0 =0 (5.16)
109
Eq. (5.17) shows the similar relation between the current sequence components presented
on (5.16) for the fault of broken wire in two phases.
PQ PQ PQ
Ia ≠ 0 , Ib = 0, Ic = 0
PQ PQ PQ PQ 1 PQ
I1 = 1/3(Ia + aIb + a2 Ic ) = I𝑎
3
PQ PQ PQ PQ 1 PQ
I2 = 1/3(Ia + a2 Ib + aIc ) = I𝑎
3
PQ PQ PQ PQ 1 PQ
I0 = 1/3(Ia + Ib + Ic ) = I𝑎
3
PQ PQ PQ
I1 = I2 = I0 = (5.17)
Eq. (5.16) and (5.17) shows that the sequence component circuit of double open phase are
in series as shown in Figure 5.8.
P1 P2 P0
Z1 Z2 Z0
~
Q1 Q2 Q0
The double open phase fault with downed wire(s) most likely will extend to a phase-to-
phase short circuit fault, which will be detected with the conventional protection scheme.
110
Where I𝑖af and I𝑖bf represents the sequence current component before and after the fault.
The inequality of the short circuit and (∆V1 ≃ 0 ) shows that the open phase fault is not
generating any short circuit current that can be compared with any type of the parallel
faults and therefore the positive voltage at the PCC and the substation almost remain intact.
It also should be noted the inequality (5.18) has no threshold to control the sensitivity of
111
the detection logic thus, in order to immunize the detection logic against fuse failure or
feeder unbalance, zero and negative sequence components must be greater than user
defined threshold (Min set1, Min set2 or Min set3 ) in order to activate the detection
logic.
Not all the three thresholds in (5.19) is required to control the sensitivity of the open phase
fault detection logic. The voltage at the substation and PCC will have no significant change
for an open phase fault contrary to a parallel type of fault in the feeder. Thus, ∆V1 ≃ 0 is
added to the above-mentioned conditions. When there is more than one source in the
feeder, the inequality equations of (5.19) must be considered for the sources that feeding
the load behind the fault point. In the following section, this concept will be developed
adequately. In this regard, an arbitrary distribution feeder with DG is shown in Figure 5.9.
In order to formulate the detection criterion for the open phase fault, two locations of the
faults at point 1 (fault1) and at point 2 (fault 2) is considered that represent an open fault
between the two sources and a fault on the same side of the sources, respectively. The
location of the fault, with respect to the power sources, is required to develop different set
of conditions that must be recognized at the early stage of the fault detection process. It
must be noted that the proposed criteria (5.18) and (5.19) are current based and therefore,
in order for this algorithm to work, the open phase fault should expose minimum load of
the feeder to the fault. This means a minimum load required to go through the fault point
which can be as low as 2% of the feeder load. The sensitivity detection will be determined
by greater value of (5.19) or the minimum required load being exposed to open phase fault.
Table 5-1, the rate of change of the current symmetrical components for different type of
faults and switching operations is compared with the proposed solution for the open phase
fault. The “↑” sign represents the positive rate of the change and the ↓ sign shoes the
negative rate of the change. The results of many simulations shows that dependability
criteria introduced in (5.18) and (5.19) presents a unique characteristic which can be
discriminated with the open phase fault from all other type of the faults. The representative
simulation for each fault or operation is shown in Appendix E.
No. Fault or switching type ∆I1 ∆I2 ∆I0 ∆V1 Figure No.
∆t ∆t ∆t ∆t
The case 6 of in the above table “de-energizing the unbalance load” in some cases when
the load of the feeder is balanced can leave the same signature as the open phase fault
formulated in (5.18). The constrained introduced in (5.19) must be used to desensitized the
proposed open phase algorithm against the unbalanced load that could occur in normal
operation. This case will be further discussed in the sensitivity section in this chapter.
Table 5-2 presents the variables and their units which is used to analyze cases with more
power sources.
Pload0 ≈ Pload
∆PGrid
≤ −1 (5.23)
∆PDG
When the grid is supplying the load behind the open phase, (5.19) is considered as follows:
115
Psource1 = PGrid + ∑m
i=1 DGi
∆Psource1
≤ −1 (5.26)
∆Psource2
▪ High impedance ground fault caused by broken conductor and downed wire, i.e., from
an overhead tower to the ground, takes some time (in the order of seconds). This
116
The above equations, developed for the point in time that an open phase fault occurs, are
valid when the ground fault is at the grid or DG side, respectively. Similar to what was
described for the open phase fault criteria, the PMUs measuring devices are at the
substation and PCCs. Hence, the above equation cannot be directly verified since the
location of the fault also unknown. However, for the high impedance ground fault which
is expected to be less than 50 A on the primary, reduces the unbalance of the system caused
by interruption of one phase load and thus, the following behavior can be measured by
PMUs at the instant the high impedance ground fault develops from an open phase fault.
∆I1G
( )>0
∆t
∆I2G
( )<0 (5.29)
∆t
∆I0G
[( ∆t
) < 0]
Similar inequality equations can be verified by grid side PMU, if the high impedance
ground fault is supplied from the grid.
117
∆PGrid K
0≤ ≅ (5.33)
∆PDG 1−K
In this case, both grid and DG are supplying the load behind the open phase and therefore,
the inequality equations (5.18) can be considered for both power sources.
118
Psource1 = PGrid + ∑m
i=1 DGi
∆Psource1 K
0≤ ≅ (5.34)
∆Psource2 1−K
by both sources rather than one of them, therefore (5.27) and are
to be updated as follows:
However, as discussed in 5.5.2, the high impedance ground fault reduces the load
unbalance of the system caused by interruption of one phase and thus, the following
behavior can be measured by PMUs at the instant of a high impedance ground fault
develops from an open phase fault.
∆I1G
( )>0
∆t
∆I2G
( )<0 (5.36)
∆t
∆I0G
[( ∆t
) < 0]
∆I1DG
( )>0
∆t
∆I2DG
( )<0 (5.37)
∆t
∆I0DG
[( ∆t
) < 0]
b. Case 2: single open phase with downed wire (high impedance ground)
b. Case 2: single open phase with the downed wire (high impedance ground)
∆PGrid −2.20
= = −1.46 ≤ −1
∆PDG 1.5
The above result reveals that the fault is between the two sources and is being fed by the
grid.
∆I1G ∆I0G
( )<0 >0
∆t ∆t
The validity of the above inequality reveals that the fault is an open phase fed by the grid.
121
∆V1G= ∆V1DG ≃ 0
The rate of positive sequence voltage change shows that no parallel fault is detected
during the simulation.
Figure 5.10. Open Phase Fault Scenario 1-Case 1-Power & Voltage
122
∆PGrid
≤ −1
∆PDG
The above result shows that the fault is between the two sources and is fed by the grid.
∆I1G
( )<0
∆t
∆I2G
( )>0
∆t
∆I0G
>0
∆t
The validity of the above inequality reveals that the fault is an open phase type and confirms
that it is being fed by the grid. At t = 3.00 sec, the high impedance ground fault with a
resistance of 200 ohms has been added to the circuit. The changes in the signature of the
open phase fault with high impedance ground fault is negligible, as expected.
∆V1G= ∆V1DG ≃ 0
The rate of change of positive sequence impedance shows that no parallel fault is detected.
124
∆PGrid
>0
∆PDG
The above result shows that the fault is between the two sources and is fed by the grid.
∆I1G ∆I1DG
( )<0 ( )<0
∆t ∆t
∆I2G ∆I2DG
( )>0 ( )>0
∆t ∆t
∆I0G ∆I0DG
( )>0 ( )>0
∆t ∆t
The validity of above inequality reveals the that fault is open phase and confirms that it is
being fed by grid.
∆V1G= ∆V1DG ≃ 0
The rate of change of the voltage shows that during the simulation no parallel fault is
detected.
127
∆PGrid −0.75
= = 0.61
∆PDG −1.22
∆I1G ∆I1DG
( ) = −0.03 KA ( ) = −0.03 KA
∆t ∆t
∆I2G ∆I2DG
( ) = 0.007 KA ( ) = 0.05 KA
∆t ∆t
∆I0G ∆I0DG
( ) = 0.003 ( ) = 0.007 KA
∆t ∆t
The validity of the above inequality shows that the fault is open phase and confirms that it
is being fed by the grid.
∆V1G= ∆V1DG ≃ 0
The rate of change of the positive sequence voltage shows that there is no parallel fault in
the circuit. After the high impedance ground fault is added to the system, the voltage
remains the same, and small changes or perturbations that are observed in the current
component does not change the inequality criteria of open phase.
130
fault. Figure 5.18 and Figure 5.19 show the results of this simulation. The detection of a
solid ground fault is not an issue for the conventional protection system. After t = 3.0 sec,
the signature of the open phase fault is totally dissolved by the ground fault protection
signature.
The sudden increase in rate of change in all the three-current components I1, I2, I0 make
the fault easily detectable by any simple or conventional over current protection element.
However, it should be noted that the extension of the open phase fault is not a solid ground
fault. The concern about the open phase conductor and its consequences is a high
impedance ground fault which is not reliably detected by conventional protection system.
It is interesting to note that one of the major discriminations between the proposed solution
dIG1
and parallel fault is at the instance of the open phase fault the rate for both grid DG
dt
The sensitivity of the proposed solution is not the same among the phases when the feeder
supplies an unbalance load. The phase with the highest load represents the least sensitive
phase. Figure 5.21. shows the open phase simulated at t=3 sec on the phase with the largest
load. The feeder supplies around 15 MVA (1 PU) and phase “C” carries about 5.2 MVA,
1.3% above the average load per phase and has the highest load. The negative sequence
component (I2) prior to the open phase incident created by an unbalanced load is around
3.2%. The plotted results show that the open phase in phase C can be detected only if the
open phase fault generates more negative sequence current than the unbalance load (3.2%)
to leave the detachable signature described in (5.18).
136
Mva
∆𝐒𝐆𝐫𝐢𝐝 =4.01%
∆𝐈𝟏_𝐆𝐫𝐢𝐝
<𝟎
∆𝐭
I1 [KA]
I2 [KA]
∆𝐈𝟐_𝐆𝐫𝐢𝐝
>𝟎
∆𝐭
I0 [KA]
∆𝐈𝟎_𝐆𝐫𝐢𝐝
>𝟎
∆𝐭
Sec
Figure 5.20. Single Open Phase at the End of Feeder with 4% Load
137
∆I1 = 0.078 PU
∆𝐼1
∆I1 = <0
∆𝑡
∆I2 = 0.013 PU
∆𝐼2
∆I2 = <0 ∆𝐼2
∆𝑡 ∆I2 = >0
∆𝑡
∆𝐼0
∆I0 = >0
∆𝑡
∆I0 =
∆𝐼0
<0 ∆I0 = 0.013 PU
∆𝑡
Figure 5.21. Single Open Phase at the End of Feeder with 4% Load
In regard to security of the proposed solution as previously shown in Table 5-1 if a single-
phase load connected directly to the primary feeder is switched off by the consumer, not
by fault, fuse blown, or feeder operator, this can be seen erroneously as an open phase fault
by the proposed sensitive open phase detection algorithm. Figure 5.22 shows that the
single-phase load (0.02 PU) is turned off in the completely balanced three-phase system
and the dynamic of this switching can be seen by this solution as an open phase. The minim
threshold provision in (5.18) is considered to desensitize the algorithm based on the largest
single phase load installed at the primary feeder. For example, if the largest single-phase
load is 0.02PU installed in the primary
138
Therefore, the actual sensitivity of the solution is equal to the greater value of the maximum
expected unbalance load and the largest installed single-phase load on the primary circuit
(one load not a group of loads). With the consideration of the cases discussed in this section
the proposed solution can protect 90% to 95% of the feeder-load against the open phase
fault if we assume the unbalanced load and the largest installed single-phase load on the
primary circuit are somewhat less than 5% to 10% of the feeder rating.
∆𝐼1
∆I1 = <0
∆𝑡
∆𝐼2
∆I2 = >0
∆𝑡
∆𝐼0
∆I0 = >0
∆𝑡
Figure 5.22. Single Open Phase at the End of Feeder with 4% Load
139
The communication time delay depends on the type of media and can vary from 5-30 ms.
PMU real time performance can take around 10-20 ms, and OPDs logic around 10-20 ms,
considering 3 to 5 sample windows for decision making process. The time performance
limitation must be taken into consideration when this application is used to block the fast
auto-recloser scheme. This chapter presented a novel open phase detection scheme in the
distribution feeder with DGs. The application is based on PMU data that are available in
substation and PCCs. The solution is examined based on analysis carried on
PSCAD/EMDC modeling and calculations.
The open conductor with the ground is modeled as a two-step event which starts with the
breaking the conductor and a time delay for it to develop into a ground fault on the bus
140
side. Detection of open phase fault should be done prior to development of this fault into a
ground fault. Symmetrical component characteristics are used as the chief property of the
open phase identifier. Rate of change, both in amplitude and in angle of both current and
voltage symmetrical components, proved to be reliable metrics to identify the target fault
in a complex feeder with multiple power sources.
5.10 Summary
In this chapter, the second use case to detect the open phase fault using the synchrophasor
data was studied. The vulnerability of distribution systems in an open phase fault and a
critical review of the reach conducted in this area was presented in this chapter. The
problem is formulated by analyzing the single- and two-phase open phase fault, with and
without high impedance ground using symmetrical component analysis technique and
EMT model development. Based on the fault signature, specific measurable criteria were
developed to identify open phase fault by PMUs located at the substation and point of
common coupling. The algorithm was further developed for a feeder with the multiple
integrated DGs. The high impedance ground fault recognition and impact were added to
the existing criteria. Although, it is shown that the high impedance fault impact is
predictable following the open phase fault, the open phase fault signature is still
recognizable with and without the ground fault.
The test system was developed using PSCAD/EMTDC software and a utility type
distribution feeder with the actual parameters was modeled using this software. The test
scenario to examine the reliability of the proposed solution was developed with emphasis
on dependability and security. From the many simulations that were carried out, select
representative results were analyzed and reported. The conceptual implementation for the
proposed solution considering the market available hardware and software was proposed.
The next chapter will summarize the work, present the major conclusion of this work, and
provide suggestions for future research work.
141
Chapter 6
Summary
The communication technology to create a new generation of protection relays that works
not only by sensing an abnormal condition based on local measurements, but also by
receiving information from remote devices, is becoming more and more possible. NASPI
(North American Synchro Phasor Initiative), which is funded by U.S. Department of
Energy, has been focusing on bulk power energy and the wide area network. However, the
deployment in this area so far, has been limited to some non-time critical monitoring
applications. In this initiative, not much attention has been given to distribution systems.
The topology of a typical distribution feeder is very similar to wide area networks but at a
much smaller scale, as it contains many connections and branches. However, no
infrastructure is available to provide information from these nodes and branches that can
be utilized for protection and control system. For many years, the simplicity of the
distribution system as a radial system and a network which is designed to be an interface
to the consumers only permitted that utilities operate this system as it is with little need for
communication and measuring technologies.
With the technological changes that are coming, the distribution system is at the forefront
of smart grid initiatives, DG integration, peak demand management, and the microgrid.
These are transforming the distribution network from simple radial systems to the more
complex bidirectional flow systems which must manage and protect the local generation
and the independent, smaller local grids. In this research, the use of synchrophasor data for
the protection of the distribution network has been investigated and it has been shown that
the investment in this application, coordinated with other aforementioned initiatives, is
underway in the distribution system.
The synchrophasor data can serve many protection and control applications with the same
structure, some of which are recommended as future research objectives in this work. This
work investigates two protection use cases and shows that more work and research in this
area can prepare the distribution system with its transformation to a grid with local
generations.
142
In Chapter 2, the fundamentals of the distribution system feeder structure relevant to this
research was presented. The Principle of protection in distribution and feeder protection
specifics were reviewed. Challenges related to the integration of DG and DG characteristics
were also verified. Wind turbine type 1 to 4 and PV solar DG short circuit and reactive
power production capability were studied. The changes in the regulatory standard, such as
IEEE 1547 related to DG integration and islanding, were reviewed and summarized.
In Chapter 4, the proposed research solution for islanding detection based on phasor
measurement data was presented. The mathematical formulation for the detection
143
algorithm was provided. The test model was developed in PSCAD-EMTDC and MATLAB
based on actual utility feeder data. The mathematical formulation is validated by numerous
simulations on a real-world test model. The concept is generalized by developing separate
models for the mainstream DERs and concept is tested with real world scale utility feeder.
In Chapter 5, the research-based proposed solution for open phase fault detection with and
without a downed wire (ground fault) based on synchrophasor data was presented and
discussed. The mathematical formulation for the detection algorithm was also described.
The computer-based test model was developed in PSCAD-EMTDS and MATLAB based
on utility data. The algorithm was validated with numerous simulations representing many
actual cases.
2. The proposed solution for the first use case - the islanding detection - has
contributed to this subject with the following specific innovative features:
e. The proposed solution does not rely on the static angle of voltage
differences between the PCC and DGs which can be small depending on
the location and size of DGs. It is based on monitoring the change of this
angle when islanding occurs, and the maximum sensitivity solution is
defined based on the IEEE compliance PMU with one degree available in
the marketplace.
3. The proposed solution for the second use case, detecting the open phase fault is
a major contribution to this subject since the selective detection of this fault is
nonexistent. The proposed solution relies on the waveform properties of the open
phase fault to recognize it, and within that framework uses an algorithm based on
the available and measurable data in substations and PCCs. The accuracy of the
measured quantity is not as important as the changes in the quantity. The voltage
phase angle is not used in proposed algorithm. Therefore, in contrast to many
protection schemes, the proposed solution is immune to inaccuracies in
measurement and is not relying on phase angle measurement.
At the time of publication of this work, two journal papers summarizing the two use cases
studied for open phase and islanding detection are under review by the IEEE Power System
Access.
145
1) There is potential to use synchrophasor data to provide real time visibility and state
estimation for the distribution network. [63]
2) There is potential to use synchrophasor data to design the PMU-based adaptive over
current protection system for distribution system with DG system [64].
3) There is potential to identify and optimize the number and location of PMU sensors
to serve control and protection applications of a distribution network.
6) There is potential for fault location application using synchrophasor data both in
primary feeder and in the secondary underground network [65].
146
References
[1] United State Dpartment of Energy, "Smart Grid System Report," United State Department of Energy,
Washington, DC, November 2018.
[2] T. S. Usten, S. M. Fraooq and M. S. Hussain, "Implementing Secure Routable GOOSE and SV
Messages Based on IEC 61850-90-5," IEEE Access, vol. 8, Februray , 2020.
[3] North American Synchrophasor Initiative, "Itemizing and Calculating the Benefits From
Synchrophasor Technology Use," NASPI, October 2015.
[4] P. Pegorano, K. Brady, P. Castelo and a. A. Meier, "Compensation of Systematic Measurement
Errors in a PMU-Based Monitoring System for Electric Distribution Grids," IEEE Transactions on
Instrumentation and Measurement, vol. 68, no. 10, October 2019.
[5] TR-109178, "Distribution Cost Structure Methodology and Generic Data," EPRI, Feb 1998.
[6] T. A. Short, Electric Power Distribution Hand Book, New York: CRC Press LLC, 2004.
[7] IEEE Working Group on Distribution Protection, "Distribution Line Protection Practices Industry
Survey Results," IEEE Power System Relaying Committee, December 2002.
[8] Hydro One Inc., "Distribution Generation Technical Interconnection Requirment 50KV and Below-
DT-10-015 R3," Hydro one Inc, Toronto , 2013.
[9] C. Chen, J. Wang, F. Qiu and D. Zhao, "Resilient Distribution System by Microgrids Formation
After Natural Disasters," IEEE Transaction on Smart Grid, vol. 7, no. 2, March 2016.
[10] H. Farzin, M. F. Firuzabad and M. Moeini-Aghtaie, "Role of Outage Management Strategy in
Reliability Performance of Multi-Microgrid Distribution Systems," IEEE Transactions on Power
Systems, vol. 33, no. 3, May 2018.
[11] M. Khederzadeh and S. Zandi, "Enhancement of Distribution System Restoration Capability in
Single/Multiple Faults by Using Microgrids as a Resiliency Resource," IEEE Systems Journal,, vol.
13, no. 2, June 2019.
[12] R. . F. Arritt and R. C. Dugan, "Distribution System Analysis and the Future Smart Grid," IEEE
Transactions on Industry Applications, vol. 47, no. 6, December 2011.
[13] North American Electric Reliability , "Bulk Electric System Definition Reference Document,"
NERC, April 2014.
[14] North American Electrics Reliablity , "Distributed Energy Resources Connection Modeling and
Reliability Considerations," NERC, February 2017.
[15] IEEE PES WG Wind Plant Collector System Design, "Characteristics of Wind Turbine Generators
for Wind Power Plants," IEEE PES, Power & Energy Society, Aug 2016.
[16] Power System Relaying Committee, "IEEE Standard for Synchrophasor Data Transfer for Power
Systems-IEEE Std C37.118.2™-2011,," IEEE, 2011.
[17] A. G. Phadke, "Sychronized Phasor Measurement a Historical Overview," IEEE, Virginia Tech,
Blacksburg, Virginia, USA, 2002.
[18] M.Paolone, "Synchrophasor from Computation to Implemantaion," in IEEE PES General Meeting,
Vancouver, July 2013.
[19] I. P. s. R. &. C. Commitee, "IEEE Standard Profile for Use of IEEE 1588 Precision Time Protocol in
Power System Applications," IEEE, 2016.
[20] "http://w3.siemens.com," Siemense, 2014. [Online]. Available:
http://w3.siemens.com/mcms/industrial-communication/en/rugged-communication/technology-
highlights/ieee-1588-precision-time-synchronization-solution-for-electric-utilities/pages/i. [Accessed
5 11 2014].
[21] A. Papoulis, The Fourier integraland its application, New York: McGrawHill, 1962.
[22] B. Jafarpisheh, S. M. Madani and S. Jafarpisheh, "Improved DFT-Based Phasor Estimation
Algorithm Using Down-Sampling," IEEE Transactions on Power Delivery, vol. 33, no. 6, December
2018.
147
[23] P. Banerjee and S. C. Srivastava, "An Effective Dynamic Current Phasor Estimator for
Synchrophasor Measurements," IEEE Transactions on Instrumentation and Measurement, vol. 64,
no. 3, March 2015.
[24] S. Vejdan, M. Sanaye-Pasand and P. Malik, "Accurate Dynamic Phasor Estimation Based on the
Signal Model Under Off-Nominal Frequency and Oscillations," IEEE Transactions on Smart Grid,
vol. 8, no. 2, March 2017.
[25] S. Affijulla and P. Tripathy, "Development of Phasor Estimation Algorithm for P-Class PMU
Suitable in Protection Applications," IEEE Transactions on Smart Grid, vol. 9, no. 2, March 2018.
[26] M. D. Zadeh and P. S. Bains, "Enhanced Phasor Estimation Technique for Fault Location in Series-
Compensated Lines," IEEE Transactions on Power delivery, vol. 30, no. 4, August 2015.
[27] J. A. Serna, "Reducing the Error in Phasor Estimates From Phasorlets in Fault Voltage and Current
Signals," IEEE Transactions on Instrumentation and Measurement, vol. 56, no. 3, June 2007.
[28] Power System Society Relaying Committee , "IEEE Standard for Synchrophasor Measurement for
Power Systems-IEEE Std C37.118.1™-2011," IEEE, 2011.
[29] K. E. Martin, G. Brunello, M. G. Adamiak, G. Antonova, M. Begovic , G. Benmouyal, P. D. Bui and
. A. Zahid, "An Overview of the IEEE Standard C37.118.2 Synchrophasor Data Transfer for Power
Systems," IEEE Transactions On Smart Grid, vol. 5, no. 4, JULY 2014.
[30] Working Group 10-TC57, "Communication networks and systems for power utility automation - Part
90-5: Use of IEC 61850 to transmit synchrophasor information according to IEEE C37.118-IEC/TR
61850-90-5 ED. 1.0," IEC, 2012.
[31] Working Group 10-10, IEC Technical Comitte 57 , "Communication Networks and Systems in
Substations-Part 8-1: Specefic Communication Service Mapping (SCSM) Mapping to MMS and to
ISO/IEC 8802-3-Second Editioin," IEC, 2012.
[32] Working Group 10- Technical Committe 57, "IEC 61850-9-2-Communication networks and systems
for power utility automation - Part 9-2: Specific communication service mapping (SCSM) - Sampled
values over ISO/IEC 8802-3," IEC Publication, 2011.
[33] U.S.-Canada Power System Outage Task Force, "Final Report on the on the August 14, 2003
Blackout in the United States and Canada," U.S. Secretary of Energy & Minister of Natural
Resources Canada, April 2004.
[34] UCTE, "Final Report of the 28 September 2003 Blackout in Italy," UCTE, April 2004.
[35] D. L. J. Ree, V. Conteno, J. S. Thorp and A. G. Phadake, "Synchronized Phasor Measurement
Applications in Power Systems," IEEE Transaction on Smart Grid, vol. 1, no. 1, p. 564, April 2010.
[36] P. Yang, Z. Tan, A. Wiesel and A. Nehorai, "Power System State Estimation Using PMUs With
Imperfect Synchronization," IEEE Transactions on Power Systems, vol. 28, no. 4, 2013.
[37] NASPI Engineering Team,Analysis Task, "Integrating Synchrophasor Technology into Power
System," North American Synchro Phasor Initiative (NASPI), 2016.
[38] N. N. A. S. Initiative, "Synchrophasor Monitoring for Distribution Systems," NASPI, January 2018.
[39] A. V. Meier, E. Stewart , A. McEachern and . M. Anderson, "Precision Micro-Synchrophasors for
Distribution Systems," IEEE Transactions on Smart Grid, vol. 8, no. 6, November 2017.
[40] W. Xu and K. Mauch, "An Assessment of Distributed Generation Islanding Detection Methods and
Issues in Canada," CANMET Energy Technology Centre– Varennes CANMET Energy Technology
Center (CETEC), 2004.
[41] M. C. Wrinch, "Negative Sequence Impedance Measurement for Distribuated Islanding Detection,"
PHD Thesis, University of British Columbia, Vancouver, 2008.
[42] A. Etxegarai., P. Eguía. and I. Zamora, "Analysis of Remote Islanding Detection Methods for
Distributed Resources.," in European Association for the Development of Renewable
Energies,Environment and Power Quality (FA4EPQ)- International Conference on Renewabl, Spain,
2011.
[43] A. Etxegarai, P. Eguía and I. Zamora, "Analysis of Remote Islanding Detection Methods for
Distributed Resources," in Renewable Energy and Power Quality (ICREPQ 11), Spain, April 2011.
148
[44] W. Wang, J. Kliber, G. Zhang, W. Xu and B. Howell, "A Power Line Signaling Based Scheme for
Anti-Islanding Protection of Distributed Generator Part II: Field Test Results," IEEE Transactiosn on
Power Delivery, vol. 22, no. 3, pp. 1758-1766, July 2007.
[45] W. Wang., J. Klibe. and W. Xu, "A Scalable Power-Line Signaling- Based Scheme for Islanding
Detection of Distributed Generator," IEEE Transactions on Power Delivery, vol. 24, no. 2, April
2009.
[46] B. Liu, "Advanced ROCOF Protection of Distribution Systems," PhD thesis, University of
Nothingham,, Nothingham, March 2012.
[47] S. NiKolovski, H. R. Baghaee and D. Mlaki´, "Islanding Detection of Synchronous Generator-Based
DGs using Rate of Change of Reactive Power," IEEE System Journal, vol. 13, no. 4, December 2019.
[48] H. R. Baghaee, D. Mlakic, S. Nikolovski and T. Dra, "Anti-Islanding Protection of PV-Based
Microgrids Consisting of PHEVs Using SVMs," IEEE Transaction on Smart Grid, vol. 11, no. 1,
Januery 2020.
[49] Y. Zhang, "Anti-islanding modeling of grid-tied inverters," in IEEE Conference-GE (China)
Research and Development Center, 2014.
[50] S. Patra, S. Agrawal, S. R. Mohanty, V. Agarwal and M. Basu, "ESPRIT based Robust Anti-
Islanding Algorithm for Grid-tied Inverter," in Proceedings of the 2016 IEEE Students’ Technology
Symposium, 2016.
[51] S. Ozturk and I. Cadırcı, "A Generalized and Flexible Control Scheme for Photovoltaic Grid-Tie
Microinverters," IEEE Transactions on Industry Application , vol. 54, no. 1, April 2018.
[52] M. Mitolo , R. Musca and G. Zizzo, "A Cost-Effective Solution for Clearing High-Impedance
Ground Faults in Overhead Low-Voltage Lines," IEEE Transactions on Industry Applications, vol.
55, no. 2, pp. 1208 - 1213, April 2019.
[53] O. Brien , U. William , G. Eric and K. Haes , "Catching Falling Conductors in Midair – Detecting
and Tripping Broken Distribution Circuit Conductors at Protection Speeds," IEEE Conference, pp. 1
- 11, Febaruery 2016.
[54] Aaron Kalyuzhny, "Analysis of Temporary Overvoltages During Open-Phase Faults in Distribution
Networks With Resonant Grounding," IEEE Transactions on Power Delivery, vol. 30, no. 1, pp. 420
- 427, Feberuary 2015.
[55] R. A. Walling , R. K. Hartana and W. J. Ros, "Self-Generated Overvoltages Due to Open-Phasing of
Ungrounded-Wye Delta Transforemr Banks," IEEE Transactions on Power Delivery, vol. 10, no. 1,
pp. 526 - 533, January Jaunuary 1995.
[56] B. Gustavsen and J. Å. Walseth, "A Case of Abnormal Overvoltages in a Petersen Grounded 132-kV
System Caused by Broken Conductor," IEEE Transactions on Power delivery, vol. 18, no. 1, pp. 195
- 200, January 2002.
[57] B. Wang, X. Dong, Z. Bo and A. Klimek, "Negative-Sequence Pilot Protection With Applications in
Open-Phase Transmission Lines," IEEE Transactions on Power Delivery, vol. 25, no. 3, pp. 1306 -
1313, July 2010.
[58] E. C. Senger , W. Kaiser, J. C. Santos and P. S. Burt, "Broken Conductors Protection System Using
Carrier Communication," IEEE Transactions on Power Delivery, vol. 15, no. 2, pp. 525 - 530, April
2000.
[59] C. L. Benner and B. D. Russell, "Practical High-Impedance Fault Detection on Distribution Feeders,"
IEEE Transactions On Industry Applications, vol. 33, no. 3, pp. 635 - 640, June 1997.
[60] J. A. Momoh, L. G. Dias and D. N. Laird, "An implementation of a hybrid intelligent tool for
distribution system fault diagnosis," IEEE Transactions on Power Delivery, vol. 12, no. 2, pp. 1035 -
1040, April 1997.
[61] H. Calhoun, M. T. Bishop, C. H. Eichler and R. E. Lee, "Development and Testing of an Electro-
Mechanical Relay to Detect Fallen Distribution Conductors," IEEE Transactions on Power
Apparatus and Systems, Vols. PAS-101, no. 6, pp. 1643 - 1650, 1982.
149
[62] D. R. Smith , "Digital Simulation of Unbalances Involving Open and Faulted Conductors," IEEE
Transactions On Power Apparatus and Systems, Vols. PAS-89,, no. 8, pp. 1826 - 1835, November
1970.
[63] J. Song, E. Dall’Anese, A. Simone and H. Zhu , "Dynamic Distribution State Estimation Using
Synchrophasor Data," IEEE Transactions on Smart Grid, vol. 11, no. 1, 2020.
[64] M. Ghalei , M. Zanjani , K. Mazlumi and I. Kamwa, "A2018pplication of μPMUs for adaptive
protection of overcurrent relays in microgrids," IET Generation, Transmission & Distribution, vol.
12, no. 18, 2018.
[65] M. Gholami , A. Abbaspour, M. Moeini-Ag, M. Fotuhi-Firuzabad and M. Lehtonen, "Detecting the
Location of Short-Circuit Faults in Active Distribution Network Using PMU Based State
Estimation," IEEE Transactions on Smart Grid, vol. 11, no. 2, 2020.
150
Figure A1 shows the distribution system one-line diagram which is used for Anti-Islanding
and open phase study test system. For the open phase study, the location of DGs and AR
is changed as per cases requirement as per what is described in chapter 5
Table A1 is presents the type of conductor which is used in the model with thire electrical
characteristics.
TCP
IP
Ethernet Link Layer
Ethernet physical Layer (100 Mbit/s)
This communication should be fast, and, for this reason, it cannot pass through all the seven
layers of the OSI communication model. In addition to the periodical data transfer from
the publisher to the subscriber for the GSE service, the GOOSE message will be sent
158
instantly upon the occurrence of any changes in the value or state of GOOSE dataset
members. In order to increase the reliability of GOOSE message in comparison to client-
server type communication model, the GOOSE message is repeated after each trigger of
transmission from the minimum time interval (Tmin). This could be as fast as 2 ms after
the original event and can be increased up to Tmax (order of seconds), which will be set
by the user in the GOOSE control block. The current status of the GOOSE data will be
transmitted continuously every Tmax. This feature could be utilized to define an action for
subscriber for supervising the connectivity of the GOOSE message if no message is
detected by the subscriber after the Tmax interval.
159
Figure D2. Converters, DC link, Grid, and Machine Side Control I/O
Figure D3. Grid Side control, Identification of Current and Voltage Component
163
Additional simulation in this section covers the different types of fault, switching
operation, and double open phase fault. The list simulation and thire results are shown in
Table E1. The different types of fault and switching operation simulation serves to
demonstrate how the proposed algorithm for the open phase fault can be discriminated
from any parallel faults and switching operation. The plus sign means that the ratio is
greater than zero. Thus, the quantity of the current component in the table increases after
the instance of the fault while the negative sign represents a decrease in the quantity after
the fault. The double sign represents the intensity of increase or decrease compared to the
single sign.
No. Fault or switching type ∆I1 ∆I2 ∆I0 ∆V1 Figure No.
∆t ∆t ∆t ∆t
Figure E5. Two Phase to Ground Fault (BCG) Power and Voltage
171
Figure E13 and Figure E14 presents the double open phase fault at the point where both
sources of DG and grid are feeding the fault on the opposite side. The first open phase
occurs at t=2.5 second and the second open phase occurs at 3.5 s. Figure E15 and Figure
E16, however, show the open phase fault at point 2 where the fault is on the same side of
substation and DG. The current sequence component rate of the change at t=2.5 second
and t=3.5 sec, respectively, and for both simulations represent the signature of the open
phase, which is discussed in chapter 5, where sequence positive current decreases while
negative and zero components are increasing.
Figure E13. Two Open Phase Fault (BC) Power and Voltage (point 2)
179
Figure E15. Two Open Phase Fault (BC) Power and Voltage (point 1)
181
Curriculum Vitae
University of Waterloo
Waterloo, Ontario, Canada
Master of Applied Science in Electrical & Computer Engineering
Thesis Title: Contribution of DFIG type wind turbine in primary
Frequency regulation
2008-2011
Recent Publications:
M. Jalali, NERC PRC-005 Best Practices for Compliance, CEATI International -2019