372

VOLTAGE DIP EVALUATION

AND
PREDICTION TOOLS

Task Force
C4.102

February 2009
 TF C4.102

VOLTAGE DIP EVALUATION

AND PREDICTION TOOLS

Members:
Juan A. Martínez-Velasco – Spain (Convenor)
Nick Abi-Samra – USA
MyoT. Aung – United Kingdom
Math Bollen – Sweden
Saša Ž. Djokić – United Kingdom
Nikos Hatziargyriou – Greece
Roberto C. Leborgne – Sweden
Jovica V. Milanović – United Kingdom
Gabriel Olguin – Sweden
Melanie Schilder – South Africa

Copyright © 2009
“Ownership of a CIGRE publication, whether in paper form or on electronic support only infers right of
use for personal purposes. Are prohibited, except if explicitly agreed by CIGRE, total or partial
reproduction of the publication for use other than personal and transfer to a third party; hence
circulation on any intranet or other company network is forbidden”.

Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any
responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties and
conditions are excluded to the maximum extent permitted by law”.

ISBN :978-2-85873-059-9
 TABLE OF CONTENTS
1 Introduction 1
1.1 Power quality 1
1.2 Voltage dips 3
1.3 Scope and content of the report 7

2 Voltage Dip Characterization 8

2.1 Introduction 8
2.2 Residual voltage and duration 8
2.3 Phase angle jump 9
2.4 Other characteristics 10
2.5 Three-phase voltage dips 11

3 Equipment Behaviour 11
3.1 Introduction 11
3.2 Voltage tolerance 13
3.3 Computer and consumer electronics 13
3.4 Sensitive loads 15
3.5 Equipment testing and sensitivity 17

4 Voltage Dip Mitigation 18

4.1 Introduction 18
4.2 Improving the system 19
4.3 Installing mitigation equipment 20
4.4 Improving equipment immunity 20

5 Simulation Tools 21
5.1 Introduction 21
5.2 Calculation of event characteristics 21
5.3 Voltage dip studies 22
5.4 Solution techniques 24
5.5 Comparison of simulation tools 25

6 Modelling for Voltage Dip Calculations 25

6.1 Introduction 25
6.2 Lines and cables 27
6.3 Transformers 28
6.4 Modelling of protection systems 28
 6.5 Modelling of power electronics equipment 32
6.6 Load modelling 34
6.7 Models for bulk-power and distributed generators 37

7 Voltage Dip Evaluation 41

7.1 Introduction 41
7.2 Voltage dip calculation 44
7.3 Stochastic assessment pf voltage dips 48

8 Voltage Dip Indices 73

8.1 Introduction 73
8.2 Indices for event characterization 73
8.3 Indices for site characterization 76
8.4 Indices for system characterization 78
8.5 Voltage dip index evaluation 79

9 Conclusions 95

10 References 97
1. Introduction
1.1 Power quality

Electric power systems are designed to generate electric energy economically and with the
minimum ecological disturbance and to transfer this energy over transmission lines and
distribution networks with the maximum efficiency and reliability for delivery to customers at
virtually fixed voltage and frequency. The optimal level of investment in designing such a
system is to be achieved by means of a trade-off between reliability and costs. Some of the
recent developments in electrical energy sector have led the change in design and operation of
power systems, i.e.:
 Due to the re-regulation of the electricity industry, there is no longer a single vertically
structured system but a number of independent companies linking electricity generation
with end users and competing among themselves for customer. Distribution companies
are under considerable pressure to improve the quality of the service they provide.
Achieving these improvements requires very large investments. If, after having made
these investments, they succeed in providing a better quality of service than their peers
can achieve, the Regulator will reward them with a higher rate of return. On the other
hand, a failure to deliver on promises of better quality of service is likely to result in
severe financial penalties. It is thus essential to the future of these companies to target
investments towards the network improvements that will be most effective. This can only
be achieved through a careful quantification of the effects of faults on service reliability
and power quality and the inclusion of these results in network design. Enhancing
network designs to optimise service quality may therefore involve tradeoffs between
reliability and power quality. Faults in the transmission network can also cause power
quality problems that affect many customers and should therefore be taken into account
in any comprehensive assessment.
 Electricity generation is shifting away from large power stations connected to the
transmission system towards smaller units connected at lower voltage levels (e.g.,
combined-heat-and-power and renewable generation connected to distribution , low
voltage, network).
 In addition to the re-regulation of electrical power industry customer expectations
regarding the quality and reliability of their electricity supply are high and steadily rising.
Outages, interruptions and voltage sags shut down industrial processes and computer
systems, often causing enormous losses in productivity and materials. These disturbances
also annoy and inconvenience residential customers who own an increasing array of
electronic devices and find such problems less acceptable in an electricity supply industry
that is privately owned and operates in a competitive environment. Customers with
important and sensitive loads may soon demand from their distribution companies
detailed data showing the number of outages, interruptions and voltage sags that should
be expected at various locations in the network. Given the probability of sags of different
types, magnitudes and shape, customers may also want to know how many of these sags
will cause their variable speed drives, programmable logic controllers and computers to
trip or fail. Combining these two types of data would indeed allow them to calculate the
expected cost of failures and sags. This information could then be used to negotiate
appropriate connection rates with the distribution companies. It could also guide the
choice of location for new plants, the selection of manufacturing equipment or suggest
the purchase of Custom Power devices.
Several definitions related to the “power quality” concept have been proposed in the past [1],

1
[2]:
 Voltage quality is concerned with deviations of the voltage from the ideal single-
frequency sine wave of constant amplitude and frequency.
 Current quality is the complementary term to voltage quality concerned with the
deviation of the current from the ideal single-frequency sine wave of constant amplitude
and frequency, with the additional requirement that the current sine wave is in phase with
the voltage sine wave.
 Power quality, in light of the above, could be considered as the combination of voltage
and current quality.
 Quality of supply is a broader term and represents a combination of voltage quality and
the non-technical aspects of the interaction between the electrical power network and its
customers.

One of the most descriptive and probably the most inclusive definitions of power quality
would be that the power quality is a multidisciplinary area of electrical engineering, related to
the assessment, analysis, characterisation and quantification of all ranges of mutual
interactions between the utilities and customers, expressed firstly as a measure of
conformance to either declared or accepted basic specifications, and then against the sets of
individual customer requirements, formulated in addition to basic specifications.

Additionally, the term electromagnetic compatibility (EMC) is often used to describe the
interaction between equipment and its electromagnetic environment (e.g. the power system).
Even though the EMC concept is more narrow (limited) than the power quality concept, the
latter is often treated as a subset of the former (e.g., in IEC standards).

Power quality disturbances can be generally divided in two categories: variations and events
[1], [2].
 Variation usually denotes small deviation of some voltage and/or current characteristic
from its nominal or declared value/waveform. All characteristics of voltage and current
are never exactly equal to their nominal or ideal values, since they change during the
normal operation within relatively small ranges, depending on system/load interactions.
These disturbances include: voltage and current magnitude/phase/ frequency variations
(i.e., unbalances), harmonic and interharmonic distortions, flicker, notching, and presence
of high frequency signals and noise. Variations are continuous phenomena that often have
steady-state or periodical character, which can be measured and quantified at any moment
in time. Generally, good design practice and selection of equipment with certain
operating characteristics significantly reduce occurrence of variations. Once their
presence is identified, they can be effectively controlled and mitigated with appropriate
power conditioning equipment. Thus, variations are generally assumed to have less
detrimental effects than events.
 Events are power quality disturbances related to usually large deviations of some voltage
or current characteristics from their nominal or declared values/waveforms. They happen
suddenly, representing brief or temporary occurrence of some phenomena that have
notoriously stochastic nature (e.g., system faults, changes in load structure, etc.). This
type of disturbances includes: voltage sags, voltage swells, undervoltages, overcurrents,
overvoltages, voltage and current transients, and (short) interruptions. These events are
related to the presence of excessively high or low voltages and/or currents which
regardless of their short duration have much more detrimental effects on system/load
operation than in the case of variations.

2
The difference between variations and events is not always obvious, and depends to a large
extent on the way in which the disturbance is measured. Variations can be measured at any
moment in time while events require prolonged monitoring periods and “waiting” for a
voltage or current characteristic to exceed a pre-defined threshold.

1.2 Voltage dips

A voltage dip is a sudden reduction of the rms value of the ac voltage below a specified dip
threshold followed by its recovery after a brief interval [3]. From the early 1970’s, results of
several extensive power quality surveys have identified voltage dips as one of the most
common power quality disturbances, accounting for approximately 65% of the total number
of all recorded events. They cause two serious power quality problems, responsible for
frequent malfunctions of electrical equipment in industrial and commercial installations (e.g.,
adjustable-speed drives, process-control equipment, relays, contactors and computers). The
economic impact of dips is often expressed in millions (or even billions) of funds per year.

Voltage dips are in general a three-phase phenomenon, in which all three phase voltages are
involved (and sometimes even the neutral-to-ground voltage). Figure 1(a) shows a three-phase
voltage dip due to a phase-to-phase fault in an underground cable that develops into a three-
phase fault within two cycles. A more common way of presenting a voltage dip is through the
rms voltages as a function of time; Figure 1(b) shows the rms voltage for the dip in Figure
1(a).

300

200
Phase Voltage in V

100

-100

-200

-300

0 0.05 0.1 0.15 0.2 0.25

Time in Seconds

a) Measured voltages
260
240
220
200
Phase Voltage in V

180
160
140
120
100
80
60
0 0.05 0.1 0.15 0.2 0.25
Time in Seconds

b) RMS voltages
Figure 1. Example of a three-phase voltage dip.

3
1.2.1 Causes of voltage dips

Without any doubt, the major cause of voltage dips are short-circuit faults in power supply
system and inside the customer’s installation. Excessive and adverse weather conditions such
as lightning, wind and ice are often directly correlated with the occurrence of dips. Electrical
degradation due to the ageing and contamination of insulators, animal and tree branch
contacts, as well as accidents related to construction and transportation activities also cause
short-circuit faults that result in voltage dips. Additionally, voltage dips are often caused by
induction motor starting, transformer energizing or load switching. Even though a voltage dip
may not be as damaging to industry as an interruption, there are far more voltage dips than
interruptions so the total damage caused by dips can be sometimes larger then one caused by
interruptions .

Examples of voltage dips obtained by computer simulations are shown in Figures 2 to 4. Note
that the rms voltages shown in these figures are calculated over a one-cycle sliding window.

Figure 2 shows a voltage dip due to motor starting: a rather small sudden drop in voltage,
followed by a gradual recovery. As electrical motors are three-phase balanced loads, the
voltage drops are the same in the three phases.

The voltage dip depicted in Figure 3 also shows a sudden drop followed by a slow recovery.
However, the drops are different in the three phases. This event is due to the energizing of a
large transformer. The inrush current is different in the three phases and associated with large
second and fourth harmonic distortion.

The voltage dip shown in Figure 4 is due to a fault: the voltage drops sharp in two phases, and
recovers sharply a few cycles later. Drop and recovery are associated with fault initiation and
fault clearing, respectively. The event in Figure 4 is associated with different voltage drops in
two phases whereas the rms voltage in the third phase is not affected by the event. A
noticeable characteristic of the event shown in Figure 4 are the sharp drop and recovery of the
voltage associated with fault initiation and fault clearing, respectively. As mentioned
previously, power system faults are the most important cause of voltage dips; they lead to the
most severe drops in voltage and to the majority of equipment trips.

In the above example the voltage during the event is constant and the voltage recovers
immediately. However, some events show multiple dip stages due to developing faults,
(change in fault type over the tie) or due to the operation of protective devices. An example of
a multi-stage voltage dip is shown in Figure 5.

4
 220

215

210
RMS voltage in V

205 Figure 2. Voltage dip due to motor

starting.
200

195

190
0 1 2 3 4 5 6 7 8 9 10 11
Time in Cycles

11

10.9

10.8

10.7
RMS voltage in kV

10.6

10.5 Figure 3. Voltage dip due to

10.4
transformer energizing.
10.3

10.2

10.1

10
0 5 10 15 20 25
Time in Cycles

11.5

11

10.5
RMS voltage in kV

10
Figure 4. Voltage dip due to a fault.
9.5

8.5
0 2 4 6 8 10 12 14 16
Time in Cycles

11

10.5
RMS voltage in kV

10
Figure 5. Multi-stage voltage dip due
9.5
to a fault.

8.5
0 5 10 15 20 25 30 35
Time in Cycles
ELECTRIC POWER ENGINEERING

5
1.2.2 Consequences of voltage dips

The reduction in voltage caused during a voltage dip leads to a reduction in energy-transfer
capability of the system, limits the fault-clearing time in transmission systems or results in
disconnection of distributed generation from the grid. Many modern devices (power-
electronic devices in particular) like computers, process-controllers, and adjustable-speed
drives experience operational problems when the voltage drops below 85% for more than 40
ms. The common feature of power electronics based devices is that they are connected to the
power system through a rectifier that converts ac to dc voltage. A voltage dip on the ac side of
the rectifier leads to a drop in dc voltage, which in turn causes problems with the application
voltages. An additional problem is the large inrush current following the voltage recovery
after the dip (i.e., after fault clearing) which may lead to a damage of the rectifier
components. In the case of three-phase rectifiers the unbalance of the currents through the
rectifier and the ripple in the dc voltage are also potential causes of equipment failure.

1.2.3 Characteristics of voltage dips

Adequate post processing of the sampled/measured voltage waveforms is the essential

requirement for characterisation of voltage dips. This is detailed in IEC 61000-4-30 [4], where
the remaining voltage and the duration are defined as the two main characteristics to quantify
a voltage dip. Both are obtained from the rms voltage as a function of time. The
characterisation of three-phase measurements remains a point of discussion, and the current
practice based on dip characterisation using worst affected phase (minimum magnitude
maximum duration approach) is not entirely satisfactory. (Note: Two-dimensional dip
definition, advocated in all existing standards, usually involves characterisation of a dip in
terms of the minimum of all reduced phase rms voltages during the event and the total
duration of the event, again regarding the durations of the individual dips in all affected
phases.) Having identified and agreed on the relevant characteristics of a voltage dip it is
possible to describe the voltage dip performance of a site or the whole network [5].
Approaches for identifying voltage dip characteristics and developing site- and system-indices
have been proposed in the past. The system voltage dip performance is summarized in a form
of magnitude duration tables in e.g., UNIPEDE Disdip table [6], IEC 61000-2-8 [3], IEEE
Std.493 [7] and IEEE Std.1346 [8] or in the form of system performance indices like, SARFI
(System Average RMS Variation Frequency) [5], [9].

1.2.4 Voltage dip mitigation

The ongoing discussion on voltage-dip mitigation concerns the responsibility sharing between
the customer and the network. In some cases the costs of mitigation equipment are shared or
the specific power-quality contracts define the responsibilities. In general, an agreement has
to be reached between parties involved with respect to what are the “normal dips” and what
are the “abnormal dips”. End-user equipment would then be expected to be immune to normal
dips, whereas abnormal dips should have a small frequency of occurrence [1]. Various
alternative methods can be considered to reduce the number of trips of equipment due to
voltage dips. A list of the most effective methods would include reducing the number of
faults, improving the performance of the protection system (e.g., faster fault clearing),
improving the network design and operation, installing dedicated mitigation equipment at the
point of interface, or improving voltage dip ride-through capabilities of end-user equipment.

6
1.3 Scope and content of the report

1.3.1 Scope of the Working Group

For many years, computer simulations have been recognised as the efficient and time saving
tool in solving various problems. When applied in appropriate software environment, they
offer the following advantages:
 Simple design of a fully controllable “environment” in which particular information of
interest can be easily extracted, without the need for expensive measuring instruments,
test beds and other laboratory equipment.
 Evaluation of the effects of various elements and components on simulated system,
including fast assessment of improvements/degradations of the system performance and
thorough “sensitivity” analysis in order to determine various thresholds, tolerances and
areas of applicability.
 Easy implementation of various mathematical, numerical, optimisation, solution-search
and any other analytical techniques, either inside one “off-the-shelf” program, or by using
appropriate interfaces to other programs.
 Fast processing of large amount of information and efficient data storage, sharing and
handling in related databases.
 Transparent and comprehensive application in training and education purposes.

In the power quality area, computer simulations may provide a convenient means to
characterise, study and evaluate possible solutions to power quality problems, as well as to
estimate or predict various power quality disturbance and their characteristics. Computer
simulations in power quality also can be used as a complementary tool to the (permanent)
monitoring, as a several years are usually necessary before the significant statistical data and
information can be gathered. Adequate simulations can shorten monitoring time substantially,
and provide an “instant” information if, for example, changes are introduced to the simulated
network or if there are no monitoring data available. As a decision tool, simulations can help
in choosing between the different system designs, equipment configurations or operating
strategies, in order to find the best possible solution.

Several simulation tools have been developed and used in the past to predict voltage dip
characteristics and describe voltage dip performance of the power networks. It has been
proven that computer simulations can be useful for characterising the voltage dips and
estimating voltage dip performance at the point of connection of sensitive equipment.
This report is aimed at providing
 a discussion on modelling guidelines for calculation of voltage dips using different
solution methods;
 a review of the calculation/simulation techniques for dip prediction;
 a summary of capabilities of existing simulation tools;
 an overview of approaches used for calculation of voltage dip characteristics.

1.3.2 Content of the report

The report has been organised into two parts:

7
  The first part is aimed at introducing the subject of voltage dips and includes Chapters 2,
3 and 4, which cover voltage dip characterisation, equipment behaviour during voltage
dips and mitigation techniques.
 The second part reviews the different techniques and simulation tools used for assessment
of voltage dip and index calculations. Chapter 5 introduces the different simulation tools
and techniques that can be applied to voltage dip studies. Chapter 6 proposes modelling
guidelines for the representation of different power components in voltage dip
calculations. Finally, Chapters 7 and 8 are dedicated respectively to the stochastic
prediction of voltage dips and the calculation of voltage dip indices.

Although the techniques and tools reviewed in this brochure can be applied to the analysis of
voltage dips due to any of the causes mentioned in the previous sections, the emphasis of this
brochure is on voltage dips caused by faults (short-circuits) at any voltage level of a power
system.

2. Voltage Dip Characterisation

2.1 Introduction

The aim of voltage-dip characterisation is two-fold: it allows for a kind of data compression,
so as to describe the event with a limited number of parameters; it also allows a comparison
between different events, classification of events based on severity: presentation of
performance statistics; comparison with equipment performance; etc. A number of
characteristics have been proposed in the literature, but the main ones in use are the residual
voltage (or “magnitude”) and duration.

In this chapter a brief description of a number of voltage-dip characteristics is given, the

description is neither complete, not mathematically exact. For more details on voltage dip
characterisation see, among others [1] and [5].

2.2 Residual voltage and duration

The only voltage-dip characteristics defined in an international standard are “residual voltage”
and “duration”. According to IEC 61000-4-30 [4] the residual voltage is the lowest one-cycle
rms voltage measured in any of the voltage channels during the event. The duration is the time
during which the one-cycle rms voltage for at least one voltage channel is below a voltage-dip
threshold. The standard document does not prescribe a value for the threshold but a value
equal to 90% of nominal voltage is commonly used. In North-American standards the term
“magnitude” is used as a synonym for residual voltage. Before the publication of IEC 61000-
4-30 the terms magnitude and duration were already in common use, e.g. in IEEE 493 [7] and
IEEE 1159 [10], but without a strict measurement definition.

The “voltage channels” referred to in the IEC standard document may be one, two or three
phase-to-phase, phase-to-ground or phase-to-neutral voltages. In most applications three
channels are used, with either phase-to-phase or phase-to-neutral voltages.

An example is shown in Figure 6. The left-hand figure shows the voltage waveforms for the
three channels. The right-hand figure shown the one-cycle rms voltage calculated every half-
cycle as prescribed in IEC 61000-4-30. The solid line in the right-hand figure indicates the

8
voltage-dip threshold. In this case at least one of the rms voltages is below the threshold for
five consecutive values, resulting in a voltage-dip duration equal to 2.5 cycles. The lowest rms
voltage is equal to 4.81 kV or 79.4% of the nominal voltage.
10 6.4

6.2

6
5
5.8
Voltage [kV]

Voltage [kV]
5.6

0
5.4

5.2

-5 5

4.8

-10 4.6
0 2 4 6 8 10 12 14
0 2 4 6 8 10 12 14 Time [cycles]
Time [cycles]
Figure 6. Example of voltage dip: voltage waveforms (left) and one-cycle rms voltages with voltage-
dip threshold (right).

For voltage dips due to faults, the residual voltage of a voltage dip is related to the location of
the fault; the duration is closely related to the fault-clearing time. Voltage dips due to faults at
distribution level have on average a longer duration than those due to faults at transmission
level. The majority of dips due to faults, at all voltage levels, have a duration less than 200
milliseconds. Longer-duration dips are more common among dips due to faults at distribution
level than among those due to transmission-level faults.

For voltage dips due to motor starting and transformer energizing, the residual voltage is
determined by the ratio between the motor or transformer size and the local fault level. The
duration of dips due to motor starting is determined by the starting time of the motor. Non-
loaded transformers give longer dip duration than loaded transformers.

2.3 Phase-angle jump

It was mentioned in the previous sections that short-circuits cause a drop in voltage magnitude
at the terminals of equipment connected elsewhere in the system. This drop in voltage
magnitude is in many cases associated with a change in the phase angle of the voltage. Two
different causes of this phase-angle jump should be distinguished: the difference in X/R ratio
between the source and the faulted feeder; and the difference in voltage drop in the different
phases. For dips due to three-phase faults only the first cause is present, resulting in a relation
between the phase-angle jump in any of the three phases and the X/R ratio of source and
feeder impedance. For dips due to single-phase and two-phase faults, those relations are rather
complex (see also Section 2.4).

To obtain the phase angle jump for a measured voltage dip, the phase angle of the voltage
during the dip must be compared with the phase angle of the voltage before the dip. The phase
angle of the voltage can be obtained from the voltage zero crossing or from the phase of the
fundamental component of the voltage. In the latter case, care must be taken to prevent
unrealistic values occurring at the beginning and end of the dip. An example of a voltage
waveform and its phase angle versus time are shown in Figure 7. No general method exists
for extracting one value of the phase-angle jump for a voltage dip in the same way as the
definition of the residual voltage referred to in Section 2.2.

9
 15
15
10
10 5
0

Angle [degrees]
5
Voltage [kV]

-5
0
-10
-5 -15
-20
-10
-25
-15 -30
0 5 10 15 20 0 5 10 15 20
Time [Cycles] Time [cycles]

Figure 7. Voltage waveforms (left) and phase-angle versus time (right).

2.4 Other characteristics

In some publications the phase angle at which the voltage dip starts and the phase angle at
which the voltage recovers are used as additional characteristics. These angles are referred to
as “point-on-wave of dip initiation” and “point-on-wave of voltage recovery”; they give a
more accurate picture of the start and end of the fault than using the one-cycle rms voltage.

The point-on-wave of voltage recovery is determined by the X/R ratio of the source
impedance at the location of the circuit breaker that clears the fault. For two-phase-to-ground
and three-phase faults the voltage recovers in two or three stages when the different poles of
the breaker clear the fault. It should be noted that the point-on-wave of voltage recovery
cannot be defined for dips due to motor starting and due to transformer energizing.

Instead of using two characteristics to characterize a voltage dip, residual voltage and
duration, in some cases a voltage dip is characterized through one value only. This is
advantageous in some cases as a first step in quantifying the performance of the supply
through voltage-dip indices [5]. Two examples of single-value characteristics are given in [5]:
the voltage-dip energy being the integral of the square of the voltage drop over time; and the
voltage-dip severity obtained by comparing residual voltage and duration with a reference
curve.

When interpreting voltage-dip characteristics it should further be noticed that a certain

fraction of voltage dips is of more complex character than the examples shown here.
Examples of more complex dips are:
 Dips due to developing faults, e.g. a single-phase fault that develops into a two-phase
fault.
 Dips due to faults in meshed parts of the system where there is a clear difference in fault-
clearing time for the breakers on different side of the faulted line.
 Dips with a heavy motor-load influence where the voltage drops slowly during the fault.
 Dips where the voltage recovers slowly due to the reacceleration of motor load and /or
the saturation of power transformers after voltage recovery.

10
2.5 Three-phase voltage dips

Most voltage-dip recordings contain three voltage channels so that non-symmetrical faults can
cause a different rms voltage in different channels. The standard way of characterisation, see
Section 2.2, is to take the lowest of the three rms voltages. This approach has a number of
disadvantages: single-phase faults in non-solidly-grounded systems give a very low residual
voltage whereas the impact on end-user equipment is small; phase-to-phase connected
monitors give a different residual voltage than phase-to-neutral connected monitors; the
residual voltages may change significantly when the dip propagates through a transformer.

An alternative classification of voltage dips, more suitable for three-phase systems, is

proposed in [11] and described in detail in, among others, [1], [12], [13] and [14]. A number
of alternative descriptions of the classification method are available in the literature, with
different levels of mathematical detail. The method classifies dips into a number of different
types, where the types A, C and D are the most important ones, corresponding to an equal
drop in amplitude in three voltages, a drop in two voltages and a drop in one voltage,
respectively. Instead of the lowest rms voltage, the so-called “characteristic voltage” is used
to quantify the severity of the dip. The characteristic voltage is a complex value being any of
the three complex voltages for type A events, the difference between the complex voltages in
the two affected channels for a type C event, and the complex voltage in the main affected
channel for a type D event.

The (complex) characteristic voltage has an absolute value (or magnitude) and an argument
(or phase). The absolute value corresponds to the residual voltage; the argument corresponds
to the phase-angle jump.

Voltage dips of the three main types (A, C and D) are shown in Figure 8 and Figure 9. The
upper figures present the phasor diagrams where the dashed lines are the pre-fault voltage
phasors forming a balanced three-phase system. The bottom figures show the rms voltage as a
function of time. Note that these are synthetic dips where the phasors are kept constant during
the event; for real measured dips the phasors often vary somewhat as a function of time, but
their general character is in most cases similar to the synthetic events shown in the figures.

For the dips shown in Figure 8, the argument (phase angle) of the complex characteristic
voltage is zero (zero “characteristic phase-angle jump”). This corresponds to the situation
where there is no difference in X/R ratio between the source impedance and the impedance to
the fault. For the dips in Figure 9 a non-zero characteristic phase-angle jump is assumed. For
dips of type C and type D this introduces an additional asymmetry resulting in all three rms
voltages being different.

3. Equipment Behaviour

3.1 Introduction

Generally speaking electrical equipment prefers a constant rms voltage. The other extreme is
no voltage for a longer period of time. In that case the equipment will simply completely stop
operating. Some equipment will stop within 1 s, other equipment can withstand a supply
interruption much longer. For each piece of equipment it is possible to determine how long it
will continue to operate after the supply becomes interrupted. A rather simple test would give

11
 1.1 1.1 1.1

1 1 1

0.9 0.9 0.9

Voltage [pu]

Voltage [pu]
Voltage [pu]

0.8 0.8 0.8

0.7 0.7 0.7

0.6 0.6 0.6

0.5 0.5 0.5

0.4 0.4 0.4

0 2 4 6 8 0 2 4 6 8 0 2 4 6 8
Time [Cycles] Time [Cycles] Time [Cycles]

Figure 8. Voltage dips of type A (left), type C (centre) and type D (right) with zero characteristic
phase-angle jump; phasor diagrams (top) and rms voltage versus time (bottom).

1.1 1.1 1.1

1 1 1

0.9 0.9 0.9

Voltage [pu]

Voltage [pu]
Voltage [pu]

0.8 0.8 0.8

0.7 0.7 0.7

0.6 0.6 0.6

0.5 0.5 0.5

0.4 0.4 0.4

0 2 4 6 8 0 2 4 6 8 0 2 4 6 8
Time [Cycles] Time [Cycles] Time [Cycles]

Figure 9. Voltage dips of type A (left), type C (centre) and type D (right) with non-zero characteristic
phase-angle jump; phasor diagrams (top) and rms voltage versus time (bottom).

the answer. The same test can be done for a voltage of 10% (of nominal), for a voltage of
20%, etc. If the voltage becomes high enough, the equipment will be able to operate on it
indefinitely. Connecting the points obtained by performing these tests, results in the so-called
“voltage-tolerance curve”.

Testing is probably the simplest and the most efficient way for the assessment of equipment
sensitivity to power quality disturbances. Monitoring and computer simulations are two
alternative approaches, but they have some significant drawbacks. Although most closely
related to the actual conditions of the equipment utilisation and operation, monitoring is
extremely time-consuming, as it relies on the recording of the disturbances, which are random
and infrequent events. Computer simulations may provide fast and convenient means to

12
characterise and study various power quality disturbances and their influence on equipment
operation. However, simulations are only as precise and reliable as are the used models of the
equipment and system components. Even the most complex and realistic models cannot
include all factors of influence. From the equipment sensitivity point of view, the results
obtained in both, monitoring and simulations should always be validated against the results
obtained in testing.

3.2 Voltage tolerance

The concept of voltage-tolerance curve was introduced in 1978 by Thomas Key [15]; he
developed a voltage-tolerance curve known as the CBEMA (Computer Business Equipment
Manufacturers Association) curve. This curve was redesigned in 1996 and renamed for its
supporting organization, the Information Technology Industry Council (ITIC). Like the
CBEMA curve, the ITIC curve is recommended as a design target for manufacturers of
computer equipment [16].

IEC 61000-4-11 describes how to obtain voltage tolerance of equipment [17]. The standard
does, however, not mention the term voltage-tolerance curve; instead it defines a number of
preferred magnitudes and durations of dips for which the equipment has to be tested. The
equipment does not need to be tested for all these values, but one or more of the magnitudes
and durations may be chosen. Table I gives an overview of the voltage tolerance of currently
available equipment. The values should be read as follows: A voltage tolerance of ( ms, %)
implies that the equipment can tolerate a zero voltage of  ms and a voltage of % of nominal
indefinitely. Any dip longer than  ms and deeper than % will lead to trip-ping or
malfunction of the equipment. In other words: the equipment voltage-tolerance curve is
rectangular with a “knee” at ( ms, %). These values not necessarily apply to a specific
piece of equipment, they are only meant to illustrate equipment sensitivity to voltage dips. It
is in all cases assumed that a dip with longer duration and lower residual voltage will have a
more severe impact on equipment. This is confirmed, with some minor exceptions, by
practical experience, by laboratory tests and by simulations.

Table I – Voltage-tolerance ranges of various equipment presently in use [8].

Voltage tolerance
Equipment
Upper range Average Lower range
PLC 20 ms, 75% 260 ms, 60% 620 ms, 45%
PLC input card 20 ms, 80% 40 ms, 55% 40 ms, 30%
5 h.p. ac drive 30 ms, 80% 50 ms, 75% 80 ms, 60%
Ac control relay 10 ms, 75% 20 ms, 65% 30 ms, 60%
motor starter 20 ms, 60% 50 ms, 50% 80 ms, 40%
personal computer 30 ms, 80% 50 ms, 60% 70 ms, 50%

3.3 Computers and consumer electronics

The power supply of a computer and of most consumer-electronics equipment normally

consists of a diode rectifier followed by some kind of electronic voltage regulator. The power
supply of all these low-power electronic devices is similar and so is their sensitivity to voltage
dips. A simplified configuration of the power supply of a computer is shown in Figure 10.

13
The capacitor connected to the non-regulated dc bus reduces the voltage ripple at the input of
the voltage regulator. The voltage regulator transforms a non-regulated dc voltage of a few
hundred volts into a regulated dc voltage of the order of 10 V. If the ac voltage drops, the
voltage on the dc side of the rectifier drops. The voltage regulator is able to keep its output
voltage constant over a certain range of input voltage. If the voltage at the dc bus becomes too
low, the regulated dc voltage will also start to drop and ultimately errors will occur in the
digital electronics.
Non-regulated
dc voltage

Vac Regulated
dc voltage

Voltage
controller

Figure 10. Computer power supply.

Figure 11 shows what happens with the dc voltage during a dip. When the rms voltage drops
suddenly, the maximum ac voltage remains less than the dc voltage for the whole cycle. Thus
the capacitor continues to discharge. This discharging goes on for a number of cycles, until
the capacitor voltage drops below the maximum of the ac voltage. After that a new
equilibrium will be reached. It is important to realise that the discharging of the capacitor is
only determined by the load connected to the dc bus, not by the ac voltage. Thus all dips will
cause the same initial decay in dc voltage. But the duration of the decay is determined by the
magnitude of the dip. The deeper the dip the longer it takes before the capacitor has
discharged enough to enable charging from the supply.
1
0.9
0.8

0.7
0.6
Voltage

0.5
0.4
0.3

0.2
0.1

0
0 2 4 6 8 10
Time in cycles

Figure 11. Effect of a voltage dip on dc bus voltage for a single-phase rectifier: absolute value of the
ac voltage (dashed line) and dc bus voltage (solid line) (Adapted from [1]).

Table II gives some values of voltage tolerance. This results in what is called a “rectangular
voltage tolerance curve”, shown in Figure 12.

14
 Table II – Voltage tolerance of computers and consumer-electronics equipment.

Minimal dc bus voltage 5% dc ripple 1% dc ripple

0 5 cycles 25 cycles
50% 4 cycles 19 cycles
70% 2.5 cycles 13 cycles
90% 1 cycle 5 cycles
Magnitude

110%
Minimum steady-state voltage
Vmin

Minimum duration
of zero voltage

tmin Duration

Figure 12. Voltage-tolerance curve of a computer: an example of a rectangular voltage-tolerance

curve.

3.4 Sensitive loads

3.4.1 Adjustable speed ac drives

Figure 13 shows an usual configuration of an ac drive. The three ac voltages are fed to a
three-phase diode rectifier. The output voltage of the rectifier is smoothened by means of a dc
capacitor. The inductance present in some drives aims at smoothening the dc link current and
so reducing the harmonic distortion in the current taken from the supply. The dc voltage is
inverted to an ac voltage of variable frequency and magnitude. The motor speed is controlled
through the magnitude and frequency of the output voltage of the voltage source converter
(VSC). For ac motors, the rotational speed is mainly determined by the frequency of the stator
voltages. Thus by changing the frequency an easy method of speed control is obtained.

Adjustable-speed drives are often very sensitive to voltage dips. Tripping of adjustable-speed
drives occurs due to several phenomena:
 The drive controller or protection will detect the sudden change in operating conditions
and trip the drive to prevent damage to the power electronic components. Tripping of the
drive is mainly on dc bus undervoltage, sometimes on ac bus undervoltage, on dc voltage
ripple, or on missing pulses through the rectifier diodes.
 The increased ac currents during the dip or the post-dip overcurrents charging the dc
capacitor will cause an overcurrent trip or blowing of fuses protecting the power
electronics components. This effect is normally considered in the drive design by setting
the dc bus undervoltage protection such that the drive will trip before a dangerous
overcurrent can occur.

15
 The process driven by the motor could not be able to tolerate the drop in speed or the
torque variations due to the dip.
 During an unbalanced dip, the currents through the rectifier diodes become unbalanced. A
small unbalance in voltage can lead to a large unbalance in current, with one current twice
as large as number and another current zero. The large current may lead to component
damage and tripping of the overcurrent protection.
50 Hz Variable
DC link Frequency
ac
ac dc
IM
dc ac

Control system

Figure 13. Typical ac drive configuration.

Many of the existing drives still trip on dc bus undervoltage. Some of the more modern drives
restart immediately when the voltage comes back; others restart after a certain delay time or
only after a manual restart. The various automatic restart options are only relevant when the
process tolerates a certain level of speed and torque variations.

Several solutions have been developed to mitigate voltage dip effects on AC drives. They
include: disabling the operation of the inverter, followed by automatic restart; installing
additional energy storage; improving the rectifier or improving the inverter.

3.4.2 Adjustable-speed dc drives

DC drives have traditionally been much better suited for adjustable-speed operation than ac
drives. The speed of ac motors is, in first approximation, proportional to the frequency of the
voltage. The speed of dc motors is proportional to the magnitude of the voltage. Magnitude
has been much easier to vary than frequency. Modern dc drives consist of a three-phase
controlled rectifier powering the armature winding, and the single-phase controlled or non-
controlled rectifier for the field winding. The armature circuit seldom contains any
capacitance, as the inductance of the armature is high enough to keep the current constant.
The field circuit is more resistive and thus needs some capacitance to prevent excessive
current and torque ripples. The most sensitive part of the dc drive is the three-phase controlled
rectifier. Most dips are unbalanced and thus associated with a phase-angle jump in at least one
of the phases. The firing-angle control of the rectifier will be affected by this, and might even
notice it as a missing pulse. The most likely reaction of the rectifier is to simply trip the drive.
If the rectifier does not trip, the drop in armature voltage will cause a fast drop in armature
current and thus in torque. Even a small drop in armature voltage can bring the torque down
to zero, leading to a reduction in speed. As dc drives are often used for speed-sensitive
processes, this will in most cases not be tolerated. During three-phase unbalanced dips, the
drop in armature voltage will differ from the drop in field voltage. This can lead to strange
drive behaviour, including overspeed.

16
3.4.3 Directly fed induction motors

A directly-fed induction motor is normally rather insensitive to voltage dips, but there are a
few phenomena that could lead to process interruption due to a dip. Deep dips lead to severe
torque oscillations at dip commencement and when the voltage recovers. These could damage
the motor or interrupt the process. The recovery torque gets more severe when the internal
flux is out of phase with the supply voltage, thus when the dip is associated with a phase-
angle jump. For unbalanced dips the motor is subjected to both positive and negative
sequence voltages at the terminals. The negative sequence voltage causes a torque ripple and a
large negative sequence current. The phase currents are however still smaller than the starting
currents, thus should not lead to process interruption. Many induction motors are protected by
contactors, being motor trips actually due to tripping of the contactor. Using dc contactors
will solve this problem.

3.4.4 Directly fed synchronous motors

Synchronous motor problems are similar to those of induction motors: overcurrents, torque
oscillations and drop in speed. However, a synchronous motor can loose synchronism with
the supply: an induction motor is very likely able to reaccelerate again after the fault, when a
synchronous motor loses synchronism it has to be stopped and the load has to be removed
before it can be brought back to nominal speed again.

3.4.5 Lighting

Most lamps just flicker when a voltage dips occur; users will probably notice it, but it cannot
be considered as something serious. It is different when the lamp completely extinguishes and
takes several minutes to recover. In industrial environments, where a large number of people
are gathered, or with street lighting, this can lead to dangerous situations.

3.5 Equipment testing and sensitivity

Standards related to testing of electrical equipment to voltage dips are almost exclusively
concerned with only two parameters: root mean square (rms) value of voltage magnitude, and
duration. Existing standards consider such a “two-dimensional” characterisation as sufficient,
and seldom mention any other dip/interruption characteristic. Testing of electrical equipment
is always related to a reproduction of a simple rectangular voltage dip. Practical instructions
about quantification and characterisation of non-rectangular dips are not given, nor how to
assess their effects on equipment operation. In actual power systems, however, voltage dips
are seldom rectangular one-stage events.

Other important aspect of existing standard for testing of electrical equipment is that they
completely neglect the presence and possible influence of non-ideal voltage supply
characteristics. Current standards recommend that voltage waveforms used in test should be
ideal sine waves at nominal frequency and they do not consider the influence of the
simultaneous occurrence of several disturbances. Sensitivity of the equipment cannot be fully
and precisely assessed if any of the influential factors is excluded from the analysis, or not
included in tests. Therefore, it can be generally concluded that standards and
recommendations for testing of electrical equipment to voltage dips are inadequate, and
should be improved, extended and reformulated in order to include additional characteristics,
parameters and conditions of influence.

17
Factors that may have an influence on the equipment response to voltage disturbances can be
divided into three following general categories [18]:
1. Voltage supply related electrical characteristics. The sensitivity of the equipment may
be influenced by both, voltage supply characteristics present before the occurrence of
the disturbance, and voltage supply characteristics after the end of the disturbance.
Therefore, voltage supply related characteristics could be further divided into three
sub-categories: pre-disturbance voltage supply characteristics (usually are related to
the voltage magnitude and frequency variations, as well as to the presence of the
harmonics and unbalance), during-disturbance voltage supply characteristics
(magnitude and duration, phase shift during the dip, points on wave of initiation and
ending, shape, type), and post-disturbance voltage supply characteristics (the
individual dip characteristics and separate phenomena that may occur after the initial
disturbance was cleared).
2. Equipment specific electrical characteristics: Equipment can be strongly influenced
by some equipment specific factors. These factors can be divided in two general
categories: Equipment operating/loading conditions, as trivial as the position in
which equipment operates, and Equipment malfunction criteria, that is any condition
that represents degradation/loss of some of equipment functions may be selected as
equipment malfunction criterion.
3. Other, non-electrical characteristics: Usually related to environmental conditions
encompassing both, voltage supply and the equipment itself. The most common
factors from this category are characteristics of the ambient in which both equipment
and power supply system should operate: temperature, humidity, air pressure, altitude,
the presence of vibrations, etc.
All types of equipment are neither influenced by the factors from all three categories, nor by
all factors from the same category. Depending on the equipment type, the effects of some
factors might be so small that they can be neglected during the assessment of equipment
sensitivity. The most reliable approach in deciding what factors can be neglected and what
cannot is the direct testing of equipment. Examples of the voltage dip tolerance of different
sensitive equipment were presented in [19], [20] and [21].

4. Voltage Dip Mitigation

4.1 Introduction

Equipment trip is the main voltage quality problem related to voltage dips. The underlying
event of the equipment trip is a short-circuit fault, which will always cause a voltage dip for
some customers. If the fault takes place in a radial part of the system, the protection system
can also lead to an interruption. Short-duration, shallow dips can be mitigated by improving
equipment tolerance characteristics. Long-duration, deep dips and interruptions can be
avoided by changing structure and/or operation of the power system. For industrial customers,
who do not normally have access to system or equipment improvement, the installation of
additional mitigation equipment is often the only option left to achieve the desired quality of
supply at the system-load interface. Traditional devices include motor-generator sets and
constant voltage or ferroresonant transformers.

18
Mitigation methods, which include solution such as reducing the number of short-circuit
faults or the fault-clearing time, changing the system to obtain less severe events at equipment
terminals, connecting mitigation equipment between sensitive equipment and the supply or
improving the immunity of the equipment [1], have been classified into three groups whose
main advantages and limitations are discussed in the following subsections.

4.2 Improving the system

4.2.1 Reducing the number of faults

Reducing the number of short-circuit faults in a system, not only reduces the dip frequency
but also the frequency of sustained interruptions. It is thus a very effective way of improving
the quality of supply. The solution is unfortunately most of the time not that obvious. A short-
circuit not only leads to a voltage dip or interruption at the customer interface but also causes
damage to utility equipment and plant. Therefore most utilities will already have reduced the
fault frequency as far as economically feasible. Some examples of fault mitigation are:
replacement of overhead lines by underground cables; use of special wires for overhead lines;
implementation of a strict policy of tree trimming; installation of additional shielding wires;
increase of the insulation level; increase of the maintenance and inspection frequencies.

4.2.2 Reducing the fault-clearing time

A reduction of fault-clearing time is achieved by using current-limiting fuses or static circuit

breakers, which are able to clear a fault within one half-cycle. Additionally several types of
fault-current limiters have been proposed which not so much clear the fault, but significantly
reduce the fault current magnitude within one or two cycles. At transmission levels the fault-
clearing time is often limited by transient-stability constraints and is already short, so further
reduction is much more difficult. Some remaining options are the use of faster circuit
breakers, a certain reduction in grading margin and faster back-up protection, such as the use
of inter-tripping for distance protection and breaker-failure protection.

4.2.3 Changing the power system

The severity of an event can be reduced by implementing changes in the supply system. Some
examples are:
 Installation of a generator near the sensitive load. The generators will keep up the voltage
during a remote dip. The reduction in voltage drop is equal to the percentage contribution
of the generator station to the fault current.
 Splitting busses or substations in the supply path to limit the number of feeders in the
exposed area.
 Installation of current-limiting coils at strategic places in the system to increase the
“electrical distance” to the fault. This can make the dip worse for other customers.
 Feeding sensitive equipment from two or more substations. A voltage dip in one
substation will be mitigated by the infeed from the other substations. The more
independent the substations are the more the mitigation effect. The best mitigation effect
is by feeding from two different transmission substations. Introducing the second infeed
increases the number of dips, but reduces their severity.

19
The number of short interruptions can be prevented by connecting less customers to one
recloser (thus by installing more reclosers), or by getting rid of the recloser scheme altogether.
Short as well as long interruptions are considerably reduced in frequency by installing
additional redundancy in the system. The costs for this are only justified for large industrial
and commercial customers. Intermediate solutions reduce the duration of (long) interruptions
by having a level of redundancy available within a certain time.

4.3 Installing mitigation equipment

A standard solution for low-power equipment is the UPS (Uninterruptible Power Supply),
which consists of a diode rectifier followed by an inverter. The energy storage device is
usually a battery block connected to the dc link. Low cost, simple operation and control have
made the UPS the standard solution for low-power equipment like computers. For higher-
power loads, UPS costs associated with conversion losses and maintenance of the batteries
become too high and this solution is not economically feasible.

Custom power devices are a more modern solution based on power electronics [22]. The most
common devices are the static series compensator (SSC) [23] and the static transfer switch
(STS) [24]. A static series compensator is a voltage source converter connected in series on
the distribution feeder, which provides a controllable source, whose voltage adds to the source
voltage to obtain the desired load voltage. Depending on the control strategy, it is possible to
use the additional voltage source to correct supply voltage unbalance, perform load voltage
regulation, compensate for voltage dips and cancel low-order supply voltage harmonics. A
static transfer switch consists of two three-phase static switches, each constituted by two anti-
parallel thyristors per phase. The switch on the primary source is fired regularly, while the
other one is off. In the event of a voltage disturbance, the STS is used to transfer the load from
the preferred source to an alternative healthy source. The load will see a disturbance during
the interval in which the transfer takes place; therefore, it must be completed so quickly that
the duration of the resulting disturbance at the load terminals is short enough not to cause
equipment trips. Other custom power devices, such as the Static Voltage Regulator (SVR)
[25] or a series combination of a STS and a SSC have been used to mitigate voltage dips.

4.4 Improving equipment immunity

It is probably the most effective solution, but as a short-time solution it is often not suitable.
Equipment immunity is very hard to achieve for short interruptions; but it is impossible for
long interruptions. Apart from improving large equipment (drives, process control computers)
a thorough inspection of the immunity of all contactors, relays, sensors, etc. can also
significantly improve the process ride-through. When new equipment is installed, information
about its immunity should be obtained from the manufacturer beforehand. Where possible,
immunity requirements should be included in the equipment specification. Most adjustable-
speed drives have become off-the-shelf equipment where the customer has no influence on the
specifications. Only large industrial equipment is custom-made for a certain application,
which enables the incorporation of voltage-tolerance requirements.

20
5. Simulation Tools

5.1 Introduction

Voltage dip calculations and studies can be performed by means of several types of
simulation tools. The selection of an adequate tool depends on several factors; e.g. goals of
the study, users experience or the accuracy required in calculations. Basically, three groups of
simulation tools can be distinguish:
 custom-made tools, whose development is generally based on commercially available
software packages;
 specialized simulation tools, e.g. a program for short-circuit calculations;
 general purpose simulation packages; e.g. EMTP-like tools.

However, to date no one tool can be used to perform all tasks that can be required in voltage
dip studies; for instance stochastic prediction and voltage dip index calculations. At the end,
users will have to use more than one commercial software package or develop their own
custom-made tool.

The following sections present

 an overview of the capabilities required to obtain the characteristics of a single voltage
dip or to perform voltage dip studies;
 a summary of the main solution techniques implemented in simulation tools that can
be used in the calculation of voltage dip characteristics;
 a comparison of simulation tools that can be used in voltage dip studies.

5.2 Calculation of event characteristics

The different methods used to obtain voltage dip characteristics by means of digital
simulation can be classified into the following three groups:

5.2.1 Fault current calculations

This is the simplest method: it results only in the during-fault voltage. Most power system
analysis packages contain a module for such calculations. The retained voltage is obtained
directly from this method (no further calculations are needed). The duration is equal to the
fault-clearing time. The protection does not need to be modelled, and a fault-clearing time can
be allocated to each fault position.

5.2.2 Time-varying phasors

It is a step further. The generator and load impedances are still represented as a complex
number (a phasor) but are calculated as a function of time. The fault-current calculation from
the previous method is in fact repeated every time step. To obtain the dip duration, protective
devices must be included into the model. The calculation of the residual voltage and the
duration of the dip require some additional steps. The standard methods for calculations of dip
characteristics apply to (measured) waveforms; but it seems reasonable to consider the

21
magnitude and the argument of the complex voltage to be equal to the rms voltage and phase
angle, as would be obtained from measurements.

5.2.3 Time-domain calculations

This is the most complex method, since it solves differential equations that characterize the
transient performance of the system. Simulations performed with time-domain tools can
capture all voltage dip characteristics (magnitude, duration, phase angle jump, point of wave)
with a high accuracy. Since models in most time-domain software packages, e.g. EMTP-like
tools, can be improved as much as required, the accuracy of calculations will depend on the
accuracy of the models implemented or selected by users.

Table III shows a summary of modelling guidelines and capabilities of simulation tools
required to perform each type of calculation.

Table III – Required capabilities for voltage dip calculations.

Calculation Voltage dip Modelling

Software tool capabilities
method characteristics guidelines
Fault-current Retained voltage Steady-state Three-phase phasor calcu-
calculations models lations.
Models adequate for steady
state calculations.
Time-varying Retained voltage Steady-state Three-phase phasor calcu-
phasors Duration models lations.
Models to reproduce the
dynamic behaviour and the
control systems of gene-
rators.
Protective device models.
Time-domain All Time-domain Three-phase calculations.
calculations low-frequency Models to reproduce the
transient models transient behaviour of po-
wer components, protecti-
ve devices and generators.
Options to capture voltage
dip characteristics.
Load modelling options.

5.3 Voltage dip studies

The studies in which voltage dip calculations are involved can be classified as follows:

5.3.1 Single-event simulations

The simulation is aimed at determining characteristics of voltage dips caused at the power
system buses of concern during a transient phenomenon.

5.3.2 Statistical simulations

The goal is to obtain the probability density function of voltage dips and the number of trips
at load buses. A Monte Carlo simulation is the natural approach for stochastic prediction of

22
voltage dips when dip causes are of random nature (e.g., short-circuits at either transmission
or distribution levels); however, other approaches can be successfully used when the number
of random variables is not too high. A Monte Carlo simulation is a numerical procedure
applied to problems involving random variables. The goal of this method is to derive the
performance of a system as a function of some stochastic input variables. Sampling is
repeated until convergence is achieved; the solution converges as the number of samples n →
∞, being zero the rate of the statistical error convergence. A Monte Carlo method has a
convergence which is independent of the phase space dimension; however, the solution has a
slower convergence that a numerical solution.

5.3.3 Sensitivity studies

A sensitivity analysis is performed to evaluate the variation of a variable caused by changes of

one parameter. This type of analysis is especially useful when one or more parameters cannot
be accurately specified. A sensitivity analysis will determine for which range of values a
parameter is of concern.

5.3.4 Calculation of voltage dip indices

Voltage dip indices can be used to indicate the different performance experienced at the
different voltage levels of a power system. They can be used to characterise a single event, a
site or a system. System indices can be used as a basis for system improvement.

Table IV shows a summary of the simulation tool capabilities required to obtain the goals
detailed above.

Table IV – Required capabilities for voltage dip studies.

Study Goals Capabilities

Single-event Obtain some or all characte- See Table III.
calculations ristics of voltage dips at buses
of a power system.
Stochastic Obtain the probability density Capabilities shown in Table III plus
predictions function of some or all charac- a multiple run option (this should
teristics of voltage dips at include all options required to
buses of a power system. perform a Monte Carlo solution) and
Obtain the rate of occurrence post-processing capabilities (i.e.
of different voltage dip charac- those capabilities needed to determi-
teristics. ne probability density functions and
rates of occurrence).
Sensitivity Deduce the effect that some Capabilities shown in Table III plus
analyses system and fault parameter can a multiple run option and post-
have on the voltage dip perfor- processing capabilities.
mance; e.g. the number of trips If the analysis is used to obtain the
at a given load bus. sensitivity of a stochastic assess-
ment, then options to perform a
Monte Carlo method are also
required.
Voltage dip index Obtain index values for either Similar to those required for
calculations sites or the entire system. stochastic predictions.

23
5.4 Solution techniques

Two basic techniques are generally applied in voltage dip calculations, using any of the
simulation tools mentioned above: frequency-domain and time-domain solution technique.
Their main characteristics are summarized in the following paragraphs.

5.4.1 Time-domain solution

Although several approaches have been developed for time-domain simulations, the most
common one, implemented in many time-domain simulation tools, is the scheme developed
by H.W. Dommel; it combines the trapezoidal rule and the Bergeron’s method [26], [27]. The
differential equations of network components are converted into algebraic equations involving
voltages, currents and past values. These algebraic equations are assembled using a nodal
approach
[G][v(t)]  [i(t)][I] (1)
where[G] is the nodal conductance matrix, [v(t)] is the vector of bus voltages, [i(t)] is the
vector of current sources, and [I] is the vector of “history” terms.

Two methods are generally used to solve nonlinear networks: pseudo-nonlinear representation
of power components and compensation. Using the first approach, the conductance matrix is
changed and re-triangularized whenever the solution moves from one straight-line segment to
another. Using compensation, nonlinear elements are represented as current injections that are
superimposed to the solution of the linear network after this solution has been computed.

5.4.2 Frequency-domain solution

Although frequency-domain techniques can be used to obtain time-domain solutions, they are
mostly applied to obtain the ac steady state solution of linear networks. At least, two
approaches can be used in voltage dip calculations
 Bus admittance matrix
The solution for a single frequency is obtained from the following set of equations
[Y][V]  [I] (2)
where elements of [Y], [V] and [I] are complex phasor values. Elements of the matrix
[Y] are calculated at the frequency of concern.
 Bus impedance matrix
The solution for a single frequency is obtained from the following set of equations
[ Z][I]  [V] (3)
where [Z] is the impedance matrix calculated at the frequency of concern. As with the
admittance equations, elements of [Z], [V] and [I] are phasor values.
The impedance matrix is a suitable means for calculating fault currents and voltages. It
is also a very efficient tool for voltage dip studies when the goal is to obtain dip
magnitudes and to analyse the propagation of voltage dips.

An important disadvantage of these solution methods is that they cannot be applied in the
presence of nonlinear components or variable topology circuits, which can produce steady-
state harmonics.

24
Hybrid techniques have already been used, mainly in steady state solutions of systems with
non-linearities and variable-topology converters. In fact, both frequency- and time-domain
solution methods are available in many EMTP-like programs. And even hybrid techniques
have been implemented in some of these tools to obtain steady state solution in power
systems with some specific non-linearities.

5.5 Comparison of simulation tools

Although many simulation tools have been used to date for simulation and analysis of voltage
dips, they can be classified into two main groups whose main characteristics, advantages and
disadvantages are summarized in Table V. In the introduction of this chapter, a third group of
simulation tools has been mentioned; however, software packages for a single task; e.g. short-
circuit calculations, are presently very rare since these tasks are included in most general
purpose simulation tools.

Table V – Comparison of simulation tools for voltage dip studies.

Type of tool Main characteristics Advantages Disadvantages

Custom-made They are based on either a Since these tools are The development of these
simulation high level language, e.g. C++, developed for a very tools can be a very difficult
tool or on a equation-oriented si- specific purpose, they can or even a non-affordable
mulation tool, e.g. MATLAB. be very efficient and fast, task if the goal is too
mainly if they are only ambitious and too many
based on capabilities of sophisticated capabilities
high level languages. have to be developed.
General They are generally circuit These simulation tools Since the scope of applica-
purpose oriented tools with capabilities have many built-in power tion of these tools is too
simulation that allow users to expand the component models, needed broad and the built-in
tool scope of applications. for voltage dip studies and models and capabilities are
Several commercial and non- capabilities for adapting too many, simulations can
commercial packages are pre- the tool to this type of be significantly slow, main-
sently used for voltage dip studies; e.g. a multiple run ly for some sophisticated
studies; they are usually inte- option or capabilities for studies, e.g. stochastic pre-
grated power system simula- the development of cus- diction of voltage dips in
tion tools or EMTP-like tools. tom-made power compo- large transmission and dis-
Their solution techniques are nent models. tribution networks.
based or either a frequency-
domain or on a time-domain
method, or even on a combi-
nation of both methods.

6. Modelling for Voltage Dip Calculations

6.1 Introduction

The representation of the equipment involved in a transient process is usually chosen taking
into account the range of frequencies that are associated to the simulated phenomenon [28],
[29]. In general, transients associated to voltage dip causes can be classified as low frequency
and slow front transients. If only voltage dips caused by faults are simulated, the frequency
range of transients is in general below 5 kHz; therefore, models to be implemented should be
capable of reproducing very accurately transients below that frequency.

25
As discussed in the previous chapter, simulation tools for voltage dip calculation are selected
taking into account the voltage dip characteristics to be predicted: if only the retained voltage
is of concern, a frequency-domain tool and steady -state models can be used; however, if
other characteristics must be estimated, e.g. the voltage dip duration, a more detailed model of
the study zone will be developed and calculations will be usually performed with a time-
domain simulation tool.

Built-in capabilities available in most time-domain simulation tools, e.g. EMTP-like tools,
can be used to reproduce very accurately most transients in power systems. However, an
accurate representation of some components is not easy; e.g. a transformer model may require
the representation of its nonlinear and frequency-dependent behaviour. In addition, the user
can be forced to choose between an accurate model and a feasible model. For instance, a
detailed model of a dynamic voltage restorer (DVR) requires a very small time step size
(about 1 μs) in time-domain simulations and would be time consuming in probabilistic
studies; in those cases a different approach should be considered, e.g. an average steady-state
model based on a voltage-controlled source.

The list of models to be used in voltage dip calculations using a time-domain simulation tool
can be classified as follows:
 conventional (bulk power) and distributed generators,
 power components (transformers, lines, cables, voltage regulators, capacitor banks),
 protective devices (breakers, protective relays, reclosers, fuses),
 mitigation devices (DVR, STATCOM, STS),
 loads.
Monitoring devices, aimed at measuring or estimating voltage dip characteristics, can be also
part of the model, since many time-domain software tools allow users to develop their own
modules to process voltage characteristics during simulation time.

Load modelling is an important and complex issue [30]; although most calculations are
performed by assuming a constant impedance representation or without including any load
model, a representation of the load is crucial in many voltage dip studies; e.g. studies aimed at
estimating voltage dip indices or at analyzing the performance of sensitive equipment, for
which a detailed representation can be made.

Two different approaches can be considered for representing some components: deterministic
or probabilistic; their selection depends on the type of study to be performed. Table VI shows
a summary of modelling guidelines to be used by default for voltage dip simulation when
using a tool based on a time-domain solution. When a stochastic prediction is to be performed
and the test system must be simulated several thousand times, a maximum time-step size is
recommendable, e.g. 100 s, and some simplified models should be used. Note that the table
includes guidelines for network equivalents, which can be needed when the system model
represents only a distribution network.

The following sections summarize modelling guidelines to be considered for the most
important components of a power system. Although the guidelines can be considered when
using any type of simulation tool, they are mainly addressed to users of time-domain
simulation tools.

26
 Table VI – Modelling guidelines for voltage dip studies.

Component Modelling guidelines

Conventional and The representation of any generator will depend on the area of
distributed generators vulnerability of the voltage dip. If the generator is connected to an
affected bus then a detailed model, suitable for simulation of slow
transients (incorporating control units and nonlinear effects) can
be needed; otherwise a constant voltage source behind its transient
reactance can suffice.
Network equivalents The most accurate representation should be deduced from the
frequency response of the transmission system that is feeding the
distribution network; however, a three-phase Thevenin equivalent
model deduced from the short-circuit capacity will be good
enough when simulating low frequency transients, e.g. voltage dip
transients.
Lines and cables Distributed-parameter models should be used to obtain very accu-
rate simulation results; however, lumped-parameter models are
usually acceptable and they should be used in probabilistic studies.
Transformers Saturable models are needed when transformer energization is the
voltage dip cause; however, when the event is caused by a short-
circuit, linear models will produce accurate enough results.
Protection systems Circuit breakers, reclosers and any type of disconnectors can be
represented as ideal switches in low-frequency transients. A non-
linear resistance is needed to represent fuses. Protective relay
models should only incorporate delays and reclosing times. A
more accurate model should also incorporate instrument
transformers.
Power electronics A switching model with semiconductor devices represented as
equipment ideal switches will generally suffice, although averaged or steady-
state models can be also considered. Very rarely more advanced
and detailed models will be needed.
Loads A constant impedance (i.e. a parallel R-L) model can be good
enough in many dip studies; a more accurate load model could
also show voltage dependence, dynamic behaviour and voltage dip
sensitivity. In probabilistic studies, the load model could
incorporate a daily variation and a random nature.

6.2 Lines and cables

Most EMTP-like tools have supporting routines for the calculation of line and cable
parameters [27]. These routines can provide the following models: lumped-parameter
equivalent or nominal pi-circuits, at the specified frequency; a constant distributed-parameter
model, at the specified frequency; frequency-dependent distributed parameter model, fitted
for a given frequency range. A frequency-dependent distributed parameter is the best
representation for a single simulation if the goal is to capture very accurately voltage dip
characteristics. The frequency response of the model should be fitted in a given range, e.g.
below 5 kHz. However, a constant distributed parameter model will suffice for most cases.

If models are to be used in probabilistic studies, distributed-parameter models have two

important drawbacks: they are time consuming and can require a very small time step size

27
when the fault that causes the voltage dip is produced not far from a line/cable terminal. A
lumped parameter model is then recommended. If the goal is to obtain approximated values of
voltage dip characteristics (magnitude, duration, phase jump), then a symmetrical-component
model will suffice.

6.3 Transformers

Transformer modelling is an important issue in voltage dip simulations. Some aspects to be

considered when developing a transformer model are: core and winding configuration,
saturation, eddy current effects [31], [32]. Saturation, as well as core configuration, can be
critical if the voltage dip is caused by a transformer energization. Besides, a very small step
size (less than 10 s) is usually required for simulation of nonlinear transformer models. If a
nonlinear model is required, voltage dip assessment using a stochastic prediction and a time-
domain simulation technique will be time consuming [33]. An advantage to be considered for
transformer representation when only dips caused by faults are simulated is that in most cases
transformers do not saturate during the event, so a linear model can be used. An important
drawback when using nonlinear topology-based models for representing three-phase core
transformers is the lack of standards for determination of parameters [32]. If a linear model is
accurate enough, both a linear matrix representation and a saturable model can be used, but
only the first approach is reliable enough for representing all transformer cores.

Voltage dip characteristics obtained from transformer models with and without including
saturation will be very similar during a fault condition. The later is an important consequence
since linear models can be used to obtain accurate enough voltage dip characteristics.
However, this does not mean that all simulation results will be always the same, irrespectively
of the transformer model, since an inrush transient is always produced when voltage is
recovered; only a nonlinear model can accurately reproduce such transient.

6.4 Modelling of protection systems

The most severe voltage dips in transmission and distribution networks are caused by faults.
Since a fault usually implies the operation of protective devices, voltage dip characteristics
caused by faults depend on the performance of the protection system. Therefore, an accurate
computer model for voltage dip studies must include a good enough representation of
protective devices.

The main principles of present protection systems are very different at transmission and
distribution levels. While overcurrent protection is the most commonly used scheme at
distribution level, several other protection schemes are used at transmission levels.

The following subsections discuss modelling guidelines to be used when representing main
protective devices in voltage dip studies at both transmission and distribution levels. Although
instrument transformers are a very important part of a protection system, their models are not
discussed here. It is assumed, by default, that voltages and currents inputted to protective
relays are not modified by instrument transformers, whose models are assumed to have an
ideal performance. This can be untrue, and the protective relay response, as well as that of the
circuit breaker, can be seriously affected by the behaviour of instrument transformers. For a
discussion about modelling of instrument transformers and their effect on the transient
behaviour of protection systems see reference [29].

28
6.4.1 Modelling of transmission level protective devices

In general, circuit breakers, protective relays and instrument transformers need to be

represented in the model of a protection system at transmission levels. Modelling guidelines
of circuit breakers and protective relays are summarized below. The subject is too wide,
mainly for relays; for more details readers are referred to Chapter 7 of reference [29].

A. Circuit breakers

The separation of the contacts of a circuit breaker causes the generation of an electric arc.
Several levels of model complexity can be considered in transient simulations during opening
operations. If the aim is to represent a device which is controlled by a protective relay and its
model is incorporated to estimate voltage dip durations, the breaker can be represented as an
ideal controlled-switch that opens at first current-zero crossing after the tripping signal is
given. The model may include a current margin parameter for approximate modelling of
possible current chopping.

A breaker operation is not always successful. The way in which this feature is incorporated
into the breaker model depends on the approach used to represent the breaker behaviour
during a transient phenomenon. For instance, unsuccessful operation can be included in an
ideal switch model by means of a random variable whose value will determine if the circuit
breaker opens or remain closed.

B. Protective relays

There are at least two approaches for developing a computer model of a protective relay:
 The first approach consists of including every relay component (electro-mechanical,
electronic or software) into the model. If the model is intended to accurately represent
all aspects of a physical relay, then it may be very complex and would require long
simulation times.
 The second approach is to develop more abstract models, which behave in a similar
manner to the relay within certain bounds. A macro-model can be the best choice in
voltage dip studies.

However, several limitations have to be accounted for when developing a relay model, since
accurate models can be difficult to develop and verify, and manufacturers do not divulge
sensitive information about inner workings of their product. Although some simplifications
can lead to erroneous conclusions, a relay model does not have to represent all performance
specifications. Obviously, the lack of detailed information can be a serious drawback when
attempting to develop a very accurate model.

Relay modelling may be based on phasor models of the power system faults and relay
operating characteristics. Phasor methods lead to the symmetrical component representation
of faulted power systems. This also may lead to relay modelling using only phasor
representation of the relay operating characteristic. Modelling relay using only phasor
methods is generally not sufficient, especially for predicting relay time-of-operation within 1
ms. Only more accurate time-domain models should be used. Figure 14 shows the several
submodels into which a digital or microprocessor-based relay model can be divided into:

29
  the first submodel consists of the input auxiliary transformers (if any) and the anti-alising
low-pass filter,
 next is the analog-to-digital conversor process,
 the third submodel is the detector that estimates fundamental frequency information,
 the relay measuring principles are next,
 finally, a submodel that represents the trip logic.

Figure 14. Schematic diagram of a microprocessor-based relay.

Although several types of relays (electromechanical, electronic, microprocessor-based) may

need to be represented, computational capabilities are usually available in present time-
domain simulation tools.

6.4.2 Modelling of distribution level protection systems

Modelling guidelines for the most common types of protective devices used in distribution
systems (circuit breakers, reclosers and fuses) are provided. The main characteristics of each
device that must be included in time-domain models are discussed. As with any other
component several modelling levels can be considered; if models have to be used in statistical
studies (i.e., stochastic prediction of voltage dips) then they should fulfil some conditions, e.g.
provide accurate enough results with time-step sizes of about 100 s [34].

A. Fuses

Fuse modelling, irrespectively of the type to be represented, has to duplicate the following
stages [35]: current sensing, arc initiation, arc interruption, current interruption. The melting
period, during which temperature rises, begins with the fault and finishes when the fuse melts;
during this stage the current flows without limitation. The melting mechanism of a fuse
depends on the magnitude and the duration of the current, as well as on the electrical
properties of the fuse. This characteristic is shown in the so-called time-current curve
provided by manufacturers. The performance of a fuse is depicted by means of the minimum
melting and the total clearing curves: a fuse has an arcing time, which is the time needed to
interrupt the current after the fuse melts, so the total clearing time curve is deduced by adding
the arcing time to the melting time, see Figure 15.
An expulsion fuse interrupts a fault current at current zero, a current limiting fuse interrupts a
fault current by forcing a current zero. Upon interruption, the operation of a current limiting
fuse results in the insertion of additional impedance and the development of an arc voltage,
when this voltage exceeds the system voltage, the arc is extinguished and the action
accomplished. An expulsion fuse heats to its melting point when the fault occurs; the current
continues to flow in the form of an arc, at zero current the arc is extinguished, being the fuse
subjected to a transient recovery voltage (TRV), whose frequency and magnitude depend on
the operating conditions. One or several arc reignitions can be caused by the TRV; the process
stops only when the dielectric strength build up is faster than that caused by the TRV.

30
 10

Minimum Melting Time

Total Clearing Time

Time [s]
0.1

0.01

0.001
100 1000 10000
Current [A]

Figure 15. Extreme time-current characteristics of a fuse.

Although minimum melting curves are similar for all fuses, the approach chosen for repre-
senting every type should be different. Since the action of a current limiting fuse results in the
sudden insertion of a high value resistance, this type of fuse can be modelled as a variable
resistance, whose value can be assumes to vary according to the following expression:
t  Tm  Rs
R(t)   (4)
t  Tm  
Rs  K  1  e  Pt Tm  
where Rs is the fuse resistance prior to any fault condition, Tm is the melting time, deduced
from the time-current curve, K and P are two factors used to adjust the fuse resistance value in
the post-melting period.

An expulsion fuse can be represented as a switch that opens at the first zero-current. For more
details on modelling of both types of fuses see references [36] - [39].

B. Circuit breakers

The performance of a distribution-level breaker during an opening operation is governed by

the characteristics of the overcurrent relay. There are two types of relays: instantaneous and
time-delay. The time-current characteristic of an overcurrent relay can consist of two sections,
the first one is independent of the current, the second one has an operating time that varies
inversely with current, see Figure 16. Depending on the rate with which the relay operating
time and current are related, the time-overcurrent characteristic can be classified as inverse,
very inverse and extremely inverse.

III
Time

II

Current

Figure 16. Time-current characteristics of overcurrent relays.

31
The time-current characteristic can be represented by the following expression [40], [41]
K (5)
t(I ) 
I / I a n  1
where n is a factor that characterizes each type of relay, K is a factor to distinguish each
member of a family, and Ia is the pickup current; i.e. the smallest value of the current that will
trigger the breaker to operate.

Circuit breaker models can be based on a controlled switch, whose time-current curve is
altered by specifying the parameters mentioned above.

C. Reclosers

A recloser is an overcurrent protective device that can sense and interrupt fault currents as
well as reclose automatically a predefined number of times a feeder. Its operation is similar to
that of a breaker with a reclosing relay. In general, reclosers have less interrupting capability
and cost less than breakers [35]. Recloser operation uses two time-current curves. The first
curve, known as fast or instantaneous, is mainly used to save lateral fuses under temporary
fault conditions. The second curve is known as slow or time-delay, and its main purpose is to
delay recloser tripping, and allow fuses to blow under permanent fault conditions. A recloser
can be set for a number of different operations, although a very common reclosing sequence
has two fast operations followed by two time-delay trips. A recloser model can be based on a
controlled-switch module that should allow users to include two tripping curves (fast and
slow), select the type of time-current characteristic (inverse, very inverse, extremely inverse),
and specify the number of reclosing operations for each characteristic and the duration of each
reclosing interval.

6.5 Modelling of power electronics equipment

Models of power electronics converters in voltage dip studies can be needed to represent
some components of the power delivery system, e.g. FACTS and custom power devices, or
parts of end-use installations, e.g. adjustable-speed drives [42], [43]. In addition, they can be
also part of the interface of distributed generators and storage devices. Figure 17 shows a
scheme of the different units that can be included in the model of a power converter, which
performs as an interface between two systems.

System Power System

1 converter 2

Control
v, i signals v, i

Control
Reference unit
signals

Figure 17. Scheme of a power converter interface.

The representation of power converters and their control units in transient simulations may be
made using different modelling levels, i.e. from a detailed representation of each
semiconductor device to an average representation of the converter without explicitly

32
modelling semiconductor devices at all. In many applications, an approximate converter
model can be used without losing too much accuracy. Although detailed modelling of the
power semiconductor devices can be required to accurately assess switching transients, device
stresses and losses, in many cases “ideal” switch models can be used for modelling the power
semiconductor devices [44], [45]. A summary of the different levels of models for power
converters and their semiconductor devices, taken from reference [45], is presented in the
following paragraphs.

6.5.1 Non-switching model

For many studies, averaged or steady-state models of the converters can be used. The power
semiconductor devices are not modelled explicitly, although the internal device characteristics
can be incorporated into the converter model. Instead, an averaged behavioural model for the
converter, based on terminal characteristics, is developed. The converter is often represented
as either a dependent current source or a dependent voltage source. In some cases, the
converter will be viewed as a current source when seen from one side and a voltage source
when seen from the other direction. These models are typically used for steady state operation
and to study the response of slower converter control schemes for power system dynamic
studies where large simulation time steps are often preferred.

6.5.2 Switching models

The degree of detail in the converter model often depends on the relationship between the
frequency of interest in the simulation results and the switching frequencies in the converter.
Switching models are needed when the frequencies of interest are within an order of
magnitude of the switching frequencies. The control model must now include the gating
circuits and the synchronization scheme. In some cases, a model of the snubber circuits has to
be incorporated; in other cases, the snubber circuit can be ignored, although a numerical
snubber may be needed as part of the switch model when using some transients programs.

The degree of detail depends on the relationship between the time periods of the frequencies
of interest and the switch transition times of the power semiconductor devices. The latter are
usually much shorter than the intervals between converter switching operations. There will be
transients associated with these turn-on/turn-off transitions. If these transients (or transients
with similar frequencies) are of interest, then more detailed device models will be required. In
addition, when the slowest transition time of the power semiconductor device (which is often
at turn-off) approaches the period between switch operations, more detailed device turn-on
and turn-off models are required. Use of these models also requires more detail in modelling
the parasitic inductances and capacitances in the converter.

The converter terminal characteristics are often sufficient for many simulations involving
power converters. In such cases, simpler or aggregate device models can be used for which
the converter can be reduced to a simpler equivalent. In addition, it is sometimes sufficient to
represent a converter made up of many converter modules as a simpler converter. If the
converter is connected to a system where the time scales of the dynamic response of interest
are very long compared to the device turn-on and turn-off times, ideal switch models can be
used. In this case, the power electronic device is assumed to open or close in one time step, as
the simulation progresses (essentially instantaneously as far as the external system is
concerned).

33
The behaviour of an ideal switch device models can be summarized as follows:
 when the device is off, it behaves as an open circuit
 when the device is on, it behaves as a short-circuit
 the device turns on at the next time step after a firing command
 the device turns off at next time step after firing command, or for diodes and thyristors
at the next time step after next current zero crossing
 switch transition time is equal to one simulation time step.
These models can be applied when frequencies of interest are much slower than switch turn-
on and turn-off times, or when converter losses, device voltage stresses and device current
stresses are not important. They are accurate enough when representing power converters in
voltage dip studies.

More detailed device models are required in other circumstances, usually of more interest to
the converter designer. As voltage source converters move to more Custom Power and
distributed generation applications it could become necessary to perform insulation
coordination studies for these applications, especially for stresses experienced by
transformers. These models must include device turn-on/turn-off behaviour and conduction
behaviour while the device is on or off. More detail in other aspects of the switching circuit,
e.g. parasitic inductances and capacitances, wire and lead resistance, snubber circuit
characteristics, and accurate gate circuit models, can be needed.

6.6 Load modelling

Load modelling for dynamic performance analysis has been the subject of a significant effort
during the last years [46] – [51]. Load modelling can be important in voltage dip studies since
the model chosen to represent the demand can have a strong influence on some dip
characteristics. Although not much effort has been dedicated to the representation of the load
in voltage dip studies, some interesting works have been performed and some experience is
already available [52] – [54]. In fact, load models to be used in voltage dip studies can be very
similar to those proposed for transient stability studies, since in both cases slow transients are
to be analyzed. The retained voltage is hardly affected by load conditions if the dip is caused
by a fault. However, other voltage dip characteristics can be strongly influenced by the
approach chosen for representing the load. Important aspects to be considered are listed
below.
 Field measurements have shown that power demand is voltage dependent [55].
 Induction motors constitute a significant percentage of the load in a power system, so the
load model has to incorporate a dynamic behaviour and a frequency dependency.
 The impact of a voltage dip will depend on the percentage of the sensitive equipment that
is connected to the system affected by the disturbance; therefore, acceptability curves
must be also incorporated into the load model [56].
 If the goal is to predict the characteristic of voltage dips (i.e. density functions of
magnitude and duration), the study must be based on a probabilistic approach [30];
therefore the daily variation and the random nature of the load must be also included.

Table VII shows a summary of load models for voltage dip calculations based on the goal of
the study [30].

34
 Table VII – Load models for voltage dip studies.

Goal Solution technique Load model

Calculation of Voltage drop Steady state / Time domain It is generally unimportant,
voltage dip Dip duration Time domain a constant impedance repre-
characteristics Phase jump Steady state / Time domain sentation will suffice
Calculation of the undelivered Time domain The model has to incorpo-
energy rate voltage and frequency
dependency, dynamic beha-
viour and sensitivity to
voltage dips (acceptability
curves)
Stochastic predict- Voltage drop Steady state / Time domain It is generally unimportant,
tion of voltage dip Dip duration Time domain a constant impedance repre-
characteristics (dips Phase jump Steady state / Time domain sentation will suffice
per year)
Stochastic prediction of the Time domain The model has to incorpora-
undelivered energy te voltage and frequency
dependency, dynamic beha-
viour, daily variation (inclu-
ding random variation) and
sensitivity to voltage dips
(acceptability curves)

Models for load representation can be divided into two main groups: deterministic and
probabilistic, whose main characteristics are detailed below.

6.6.1 Deterministic models

A complete load model should include voltage and frequency dependency, dynamic
behaviour and tolerance to voltage dips. Different models incorporating one or several of
these features can be considered.

1. Static model. A power demand that incorporates voltage dependence can be expressed as
follows:
np nq
 np nq
 (6)
S  P0  akV k  jQ0  bkV k   ak  1 ;  bk  1
k 0 k 0  k 0 k 0 
where P0 and Q0 are respectively the rated real and reactive power at nominal voltage, and V
is the p.u. voltage.

Expression (6) assumes that there could be a part of a power demand that is voltage-
independent. In fact, this is the approach implemented in the majority of load flow programs.
By using this approach, the power demand remains the same irrespectively of the values of
bus voltages. This is not a realistic model for voltage dip calculation, as it would mean that
even for very low retained voltages, the demand will be the same as that prior to the dip. A V1
dependence means that the load behaves as a constant current source, while a V2 dependence
means that a load behaves as a constant impedance.

2. Dynamic model. Since a significant percentage of the electric consumption is constituted

by induction motors, the load model has to include a dynamic performance. Although models
for representation of induction motors based on a block diagram have been developed [46],

35
other dynamic loads have to be considered. Figure 18 shows the diagram of a dynamic load
model that can be applied to both real and reactive powers with different parameters [50].
V + P
Np(V) Gp (s)
+

Pt (V)

Figure 18. Diagram of a dynamic model for active power.

Several levels of sophistication have been proposed for the model depicted in the figure; in
this work the block G(s) is represented as a first order model. Time-domain equations of this
model can be written as follows:
dPr (V ) dQr (V )
p  Pr (V )  N p (V ) ; q  Qr (V )  N q (V )
dt dt
np nq
N p (V )  P0  askV k  Pt (V ) ; N q (V )  Q0  bskV k  Qt (V ) (7)
k 0 k 0
np nq
Pt (V )  P0  atkV k ; Qt (V )  Q0  btkV k
k 0 k 0

P  Pr (V )  Pt (V ) ; Q  Qr (V )  Qt (V )
where p and q are recovery time constants, Pr(V) and Qr(V) are respectively the active and
reactive power recoveries, while Pt(V) and Qt(V) are the active and reactive powers of the
dynamic part. Pr(V), Qr(V), Pt(V) and Qt(V) are represented by a polynomial expression such
as that shown in expression (6), but coefficients for each power component are different.

The model described by the above equations is known as exponential recovery model; since
not all dynamic loads can be represented by this model, see for instance [54], other
approaches can be needed.

3. Hybrid model. It is the model that results from a combination of any of the above models,
to which an induction motor load can be also incorporated.

4. Load tripping. Sensitivity or tolerance of equipment to voltage dips can be represented by

acceptability or voltage-tolerance curves. The incorporation of this curve is straightforward:
since either the above approach requires a voltage measurement, this value is used to decide
whether the magnitude and duration of the voltage dip will force a trip or not. Another aspect
to be considered is the tolerance of the equipment to phase jump; if this must be incorporated,
then a phase jump measurement is also required.

6.6.2 Probabilistic models

The daily demand variation will be based on two curves for the active and reactive power,
respectively, and a normal probability density function for each power. The determination of
the real and reactive power at a given bus for every period will be therefore based on the
random generation of three values: the first one will be the fault instant, which is necessary to
obtain the mean value of both active and reactive powers, and two additional random values,
which are necessary to obtain the final values of both powers, considering the standard
deviation for each one. One of the above deterministic models will be used after the values of

36
active and reactive powers have been determined. This approach can be extremely complex if
all parameters to be specified in a deterministic model are assumed variable and random.

Other works on load modelling using an electromagnetic transients program has been
previously reported, see for instance [57] - [60]. Other important issues are the estimation of
load parameters [50], [51], or the effect that load conditions can have on voltage dip
characteristics [60].

6.7 Models for bulk-power and distributed generators

In voltage dip studies a generator connected to a transmission network is usually represented

as an ideal constant-voltage source. This model is acceptable if the generator is not inside the
area of vulnerability (see Section 7.2 of this document). However, if the generator terminals
are inside the area of vulnerability, then the control of the excitation will react and a more
detailed generator model should be used. In voltage dip studies of distribution networks the
electrical distance between faults and generators will be rather short, so detailed models of
distributed energy sources are generally required.

The synchronous generator has been the main generation component for decades; although
this is still the situation in all bulk power systems, many large wind energy farms, where the
induction generator is often used, have been already connected to transmission networks. In
addition, the effect that voltage dips can have on wind energy farms is a very important issue
in modern power systems. The following subsections provide modelling guidelines of the
most common generation technologies; the first one summarizes the main features of a
synchronous generator model when it has to be included in voltage dip calculations; the
second subsection discusses models to be used for distributed generation.

6.7.1 Bulk-power generators

Figure 19 shows the main components of a synchronous generator. In voltage dip studies a
complete representation of both the electrical and the mechanical part of the machine, as well
as the model of voltage control unit, will be generally required. Saturation effects may be also
included in the model of the electrical part; although they do not affect the machine
performance during a fault, they can have a strong influence during the post-fault period,
especially with low retained voltage dip cases.
Mechanical Electrical
variables Synchronous variables
Speed Voltage
generator
control control
(electrical and
unit unit
Power/torques mechanical parts) Field voltage

Stator voltages
and currents

Power system

Figure 19. Schematic diagram of a synchronous generator and its control units.

37
The response of protective relays in transmission networks is usually very short (i.e. faults are
generally cleared in less than two cycles of the operation frequency), and the model of the
speed governor is not required. Very rarely breaker operations are delayed more than 10
cycles; only in those situations, and especially when turbines are equipped with fast
controllers, speed governor models should be included. Detailed models of synchronous
generator and the associated control units during low frequency transients have been
presented in many references, see for instance [27], [61] – [64].

6.7.2 Distributed generation

The production of electrical energy from renewables and other distributed power sources
often requires interfacing of the primary source to the grid or the local load via a power
electronics converter. Typical examples in the field of the renewable energy sources are the
variable speed wind turbines, whose aerodynamic power is first converted to AC power of
varying magnitude and frequency, through a rotating electrical generator (induction or
synchronous), then it is rectified to DC and subsequently inverted to the AC frequency of the
grid. Similar is the situation in the case of photovoltaics, where their DC output power has
again to be inverted to AC for grid-connected operation. Figure 20 shows a schematic
representation of a dispersed generator: a generator model is used for the calculation of the
power injected to the power electronics interface; a static converter model and its control
system are used to calculate the power injected to the grid, as well as the voltage and
frequency applied to the energy source and the common coupling point, respectively.
DC side

Energy DC-AC
Grid
source converter

Figure 20. Dispersed generator configuration interfaced to the grid via power electronics.

Model requirements strongly depend on the application, i.e. on the time scale and nature of
the phenomena to be reproduced. On the other hand, modelling of the energy source is case-
specific and depends on the type of source under consideration (wind turbines, photovoltaics,
fuel cells, etc.). In any case, the developed model is connected and solved together with the
model of the output DC/AC converter, to represent the dynamics of the whole system [65].

A. Generation models

i. Wind energy conversion systems: They comprise several subsystems that are indepen-
dently modelled: the aerodynamic, the generator, the mechanical system and the power
converters, in case of variable speed wind turbines. The mechanical subsystem can be
represented by means of three to six elastically connected masses. The use of at least two
masses is necessary for the representation of the low-speed shaft torsional mode. As for
the generator, it can be based on either a induction or a synchronous machine.
Induction machines can be represented by a fourth order model expressed in the arbitrary
reference frame, while the controller can use either a scalar or a vector control strategy.
Figure 21 shows the principle of a vector controller. Using the coordinate transformation
to the rotating flux reference frame, the decoupled control of the machine torque and flux
levels can be achieved. The control system consists of two major loops, one for the

38
rotating speed of blades and the other for the rotor flux linkage. Indirect field-oriented
voltage control is applied to the generators. The wind speed input signal is led to the wind
speed – blades optimal rotating speed characteristic, which produces the input signal to a
low-pass filter. The output of the low-pass filter is used as input to the torque controller.
For low wind speed, maximum energy efficiency is achieved by tracking the optimal
rotating speed. At high wind speeds the control scheme imposes a constant rotating speed.
The q-axis current component in the field-oriented frame can be obtained from the
electromagnetic torque, while the d-axis current component can be obtained from the
rotor flux. The desired voltage components in the field-oriented frame are obtained from
the model of the induction machine. The main drawback of this control scheme is that it
requires accurate knowledge of the machine parameters. It is commonly used with on-line
parameter adaptive techniques for tuning the value of the parameters used in the indirect
field controller, ensuring in this way successful operation.

Figure 21. Indirect field-oriented voltage controller.

Figure 22 shows a variable speed wind turbine equipped with a synchronous generator.
The variable frequency AC output of the generator is converted to DC by an uncontrolled
diode-rectifier. A boost DC-DC converter interfaces the rectifier to the DC/AC grid-side
converter. The generator rotor is electrically excited. Damper windings can be neglected,
because they hardly affect the grid interaction in power system dynamics simulation, due
to the decoupling effect of the power electronic converter. The generator torque control is
performed by the DC chopper, which regulates the rectifier (and hence the generator)
current through a simple current control loop. The DC current reference is the output of
the speed control loop. Alternatively, it may be determined directly from a torque
reference value, if the turbine operates in the torque control mode.

Figure 22. Wind turbine equipped with synchronous generator.

39
ii. Photovoltaic systems: Models of varying complexity can be used to describe the
behaviour of a PV array. To choose an appropriate model, the most important factor is
accuracy; there is always a trade-off between accuracy and simplicity.
1. Simplified single-diode model
The model inputs are the ambient temperature, solar irradiance; array voltage while the
output of the model is the array current. Voltage will be an input from the MPPT control
scheme embedded in the inverter. When using this model, it is assumed that:
 All the cells are identical and they work with the same irradiance and temperature.
Also, voltage drops in the conductors that interconnect the cells are negligible.
 Short-circuit current is affected only by the irradiance and temperature of the cells.
 Open circuit voltage of the cells depends exclusively on the temperature of the cells.
 Temperature of the solar cells depends exclusively on the irradiance and ambient
temperature.
 Series resistance and diode quality factor of the cells are considered constant for all
operating temperature and irradiance range.

2. Integrated MPPT (Maximum Power Point Track) model

It considers that the PV array is working always at its maximum power point for given
temperature and irradiance conditions. This model is based on the linear variation of a PV
module power with its temperature; its inputs are the ambient temperature and the solar
irradiance, while its output is the maximum array power. When using this model, it is
assumed that:
 All the cells are identical and they work with the same irradiance and temperature.
 There are no lo losses in the array.
 The PV array is always working on its maximum power point for given irradiance and
ambient temperature conditions, but if these conditions change, the model instantaneously
changes its maximum operation point.
 Temperature of the solar cells depends exclusively on the irradiance and ambient
temperature.

iii. Fuel cells: A power generation fuel cell model has to represent three main parts: the fuel
processor, which converts fuels such as natural gas to hydrogen and by product gases; the
power section, which generates the electricity; the power conditioner, which converts dc
power to ac power output and includes current, voltage and frequency control. All the
reactions that occur in the fuel cell have an associated time delay. The chemical response
in the fuel processor is usually slow, as it is associated with the time needed to change the
chemical reaction parameters after a change in the flow of reactants. This dynamic
response function is modelled as a first-order transfer function with a time delay constant,
which is generally short and associated with the speed at which the chemical reaction is
capable of restoring the charge drained by the load. The dynamic response function of the
flow is also modelled as a first-order transfer function with the time delay constant of the
respective element. The outputs of this model provide the potential difference between
anode and cathode and the reaction current. A conventional battery can be connected to
the dc output of the fuel cell to provide fast response to load step increases. The interface

40
 with the grid is made through an inverter that may include a power frequency control loop
for stand-alone operation.

B. DC/AC interface

The control of the active and reactive power flow to the grid is performed by a DC/AC
converter. In the case of a voltage source inverter, the controlled variables are the frequency
ωi and the magnitude Vi of the fundamental component of its AC voltage, which is
synthesized by properly switching on and off its semiconductive elements. A decoupled
regulation of P and Q permits the implementation of the control principle illustrated in Figure
23. The active and reactive power regulation loops are independent but not fully decoupled.
P* and Q* are the set points for the output active and reactive powers, P and Q, whereas the
actual P and Q, along with the power angle δ, are calculated from measurements of the phase
voltages and currents. The determination of the P* and Q* reference values depends on the
specific application and installation considered. A usual practice is to utilize the Q* input in
order to maintain constant output power factor (often unity, hence Q* = 0). Alternatively, Q*
may be varied in order to regulate -or simply support- the bus voltage at (or near to) the
output of the converter, provided that the current rating of the converter permits it.
Active power Power angle
controller controller
P* * *

INVERTER
P 

Reactive power
controller
Q* Vi*

Figure 23. Active and reactive output power control principle.

If the DC/AC converter is current controlled (which requires a relatively high switching
frequency and is not yet common at high power levels), decoupling of the active and reactive
power regulation loops can be achieved by means of the vector control principle. For the
fundamental frequency modelling of the converter system shown in Figure 20, the differential
equations of the DC capacitor and the output inductances are used, along with the power
balance equation of the DC/AC converter.

7. Voltage Dip Evaluation

7.1 Introduction

The stochastic approach is the most suitable way to predict the number and characteristics of
voltage dips in a network [66] – [75]. Though one cannot predict the exact time when the dip
will occur, it is possible to estimate the probability of occurrence of the dip of particular
characteristics at a given location. Stochastic prediction methods are as accurate as the model
and the data used. While the accuracy of the models can easily be influenced and improved if
necessary, the accuracy of the data is often outside of our control. The data involved in

41
voltage dip assessment are power system and component reliability data. Component relia-
bility data can only be obtained through observing the behaviour of the system component
over long period of time and hence they have the same uncertainties as the outcome of power
quality monitoring. Many utilities have records of component failures over several decades.

Faults can be categorised as self-clearing, temporary and permanent. A self-clearing fault

extinguishes itself without any external intervention. A temporary fault is a short-circuit that
is cleared after the faulted component is de-energized and reenergized. A permanent fault is a
short-circuit that persists until repaired. Faults can be observed at the customer’s premises as
long interruptions, short interruptions, dips and swells. Outages occur when permanent faults
take place in the direct path feeding the customer. Short interruptions are the result of
temporary faults cleared by the successful operation of a breaker or recloser. Dips and swells
occur during faults that are not in the direct path supplying the load.

A distinction should be made between fault (a condition that causes a device or a component
to fail; e.g., relay maloperation) and short-circuit (an abnormal connection of low impedance
between two points of different potential). By default, it is assumed that all faults causing
voltage dips are short-circuits. Therefore, fault statistics used in this publication are in fact
short-circuit statistics.

Several random factors are involved in the analysis of voltage dips:

 Fault type. Three phase faults are more severe than single-phase faults, but the later are
much more frequent.
 Fault location. Faults originated in transmission systems cause dips that can be seen tens
of kilometres away, while faults at radial distribution systems have a more local effect.
 Fault initiation angle or point on wave. This uncertainty has an important effect on the
ability of some equipment to tolerate the dip. It affects the transient behaviour of the fault
current, but it does not have much relevance for the counting of events.
 Fault impedance. Solid faults cause more severe dips than impedance faults.
 Fault clearing time. Dips last for the time short-circuit current is allowed to flow
throughout the system. Protection devices allow different settings causing dips of
different duration. A post-fault dip may occur due to motor and transformer recovery.
 Reclosing time. It is common practice to set one or two reclosing attempts in radial
distribution feeder protections. This practice is known as fuse saving and has an important
impact on frequency of dips originated at distribution levels.
 Fault duration. Self-cleared faults cause dips which duration depends on the fault itself,
not on the protection setting.
 Power system modifications. The impedance between the point of observation and the
fault point affects the magnitude of a fault caused dip. However the system is not static
and changes that affect this impedance are continuous.
 Severe weather. The occurrence of faults and dips is notoriously bigger during severe
weather, which can take many forms such as wind, rain, ice or snow.

In stochastic prediction of voltage dips two methods are generally used:

42
 The method of fault positions calculates the characteristics of voltage dips (magnitude,
duration and phase shift) at the equipment terminals for a number of faults spread
throughout the system. Each fault position represents faults in a certain part of the system.
By increasing the number of fault positions, the accuracy of the results can be increased.
The expected number of fault occurrences per year is deduced from the monitored
(historical) data for each fault position. The first step in applying the method is the
selection of the actual fault positions. A random choice of new fault positions may not
necessarily increase the accuracy of the result; it may only increase the computational
effort. Fault positions and the type of fault in particular system can vary depending on
weather conditions and utility maintenance. The main criterion in choosing the fault
positions should be that each fault position represents a number of faults leading to
voltage dips with a similar magnitude and duration at the location of interest.
 The method of critical distances calculates the fault position for a given voltage. By using
some simple analytical expressions (strictly correct for radial systems only) it is possible
to find out in which part of the system a fault would lead to a dip of given characteristics
at desired location. Each fault closer to the load would cause deeper dip. Thus, the
number of dips below a given threshold will be equal to the number of faults closer to the
load than the indicated position.

The method of fault positions and the concept of the area of vulnerability are generally used
to assess and understand the system voltage dip performance. The concept of the area of
vulnerability enables to identify fault locations which will lead to dips deeper than a given
voltage threshold. Voltage dip performance is assessed by performing system fault analysis in
order to determine voltages at a particular bus as a function of fault locations throughout the
system. The identification will yield an area that contains exposed buses and lines and
associated area boundary crossing lines.

Voltage dips can be studied from different perspectives and the model used to describe the
disturbance needs to satisfy the study objectives. To obtain statistics on magnitude of fault
originated voltage dips modelling based on phasors has been considered suitable. It should be
noted that the use of phasors restricts the model to the context of steady state alternating linear
systems. The modelling used in this chapter is intended to obtain the during-fault voltage or
the retained voltage during the fault but not its evolution as function of time. Figure 24 shows
a three-phase unbalanced voltage dip in an 11 kV distribution network and the approximation
considered in this chapter. Depending on the fault type the shape of the rms voltage evolution
will show different behaviours [76].

Approximation
Figure 24. Three-phase voltage dip: Rms
used in this work voltage vs. time.

43
The following sections present the main principles of voltage dip calculations using the
method of fault positions and a frequency-domain technique, as well as its application to the
stochastic prediction of voltage dips in transmission networks. The application of a time-
domain technique to the stochastic prediction of voltage dips in distribution networks will be
presented in the following chapter; that example will be also used to illustrate the calculation
of voltage dip indices.

7.2 Voltage dip calculation

7.2.1 The impedance matrix

An electrical network under short-circuit fault conditions can be considered as a network

supplied by generators with a single load connected at the faulted bus. The pre-fault load
currents can usually be ignored since they are small compared to the fault current. Usually
they are considered by means of the superposition theorem if needed.

The use of the impedance matrix provides a convenient means for calculating fault currents
and voltages. The main advantage of this method is that once the bus impedance matrix is
formed the elements of this matrix can be used directly to calculate the currents and voltages
associated with various types of faults.

Bus and current voltages of a general n-port passive linear system can be related as follows:
v1  z11  i1  z12  i2  ........z1n  in
:
vk  z k 1  i1  z k 2  i2  ........z kn  in (8)
:
vn  z n1  i1  z n 2  i2  ........z nn  in
This equation can be written in matrix notation
V  ZI (9)
where Z is the bus impedance matrix of the network.

The impedance matrix contains, in its diagonal, the driving point impedance of every bus with
respect to the reference bus. The driving point impedance of a bus is the Thevenin’s
equivalent impedance seen into the network from that bus. The diagonal elements of the
impedance matrix allow determining the short-circuit current of every potential fault at buses
of the system. The off-diagonal elements of the impedance matrix are the transfer impedances
between each bus of the system and every other bus with respect to the reference bus. The
transfer impedance gives the voltage at bus k when a current (unitary) is injected at bus j.
Hence the transfer impedance allows determining the during-fault voltages due to the fault
currents.

7.2.2 Voltage dip magnitude calculation

Faults in power systems can be symmetrical and unsymmetrical leading to balanced and
unbalanced dips, respectively. For symmetrical faults only the positive sequence network is
required to analyse the during-fault voltage. However the majority of the faults are single-
phase-to-ground faults requiring the use of symmetrical components in the analysis [77].

44
A. Balanced voltage dips

Consider a network with N buses plus a reference bus, which is named zero and is chosen to
be the common generator bus. According to the superposition theorem, the voltage during a
fault is the pre-fault voltage at that bus plus the change in the voltage due to the fault
vkf  v pref ( k )  vkf (10)
where vpref(k) is the pre-fault voltage at bus k and vkf is the voltage-change at bus k due to the
fault at bus f. Expression (10) can be written as a matrix relation
Vdfv  Vpref  ΔV (11)
where Vdfv is the dip matrix that contains the during fault voltages, Vpref is the pre-fault
voltage matrix and because the pre-fault voltage at bus k is the same for a fault at any bus, the
pre-fault voltage matrix is conformed by N equal columns. V is a matrix containing the
changes in voltage due to faults everywhere. Row k of Vdfv contains the retained voltages at
bus k when faults occur at buses 1, 2, ..k, ..N, while column f of Vdfv contains the retained
voltage at buses 1, 2,...N for a fault at bus f.

During a three phase short-circuit at bus f, the current “injected” to this bus is given
 v pref ( f )
if  (12)
z ff
where vpref(f) is the pre-fault voltage at the faulted bus f, zff is the impedance seen looking into
the network at the faulted bus and the minus sign is due to the direction of the current. Only
positive sequence values are needed to perform the calculations.

Once the “injected” current is known, the change in voltage at any bus k can be calculated
using the transfer impedance zkf between bus k and bus f. Then (12) gives the change in
voltage due to the current expressed by
v
vkf   zkf  pref ( f ) (13)
z ff
The matrix version for this equation is given is the following one
ΔV   Z  inv(diagZ)  Vpref
T
(14)
where inv(diagZ) is the inverse of the matrix containing the diagonal elements of the
impedance matrix and the T over Vpref indicates transposition of the pre-fault voltage matrix.
It should be noted that vkf  vfk because zff  zkk.

Voltage dips caused by a three-phase fault are therefore obtained as follows

v
vkf  v pref ( k )  zkf  pref ( f ) (15)
z ff
The voltage seen at a bus k due to a fault at bus f can be expressed by the following general
equation
Vdfv  Vpref  Z  inv(diagZ)  Vpref
T
(16)
If the pre-fault load can be neglected, then pre-fault voltages can be assumed 1 pu, and (16)
can be written as follows
Vdfv  ones  Z  inv(diagZ)  ones (17)

45
The positive-sequence impedance matrix Z is a diagonal dominant full matrix for a connected
network. This means that every bus in the network is exposed to dips due to faults everywhere
in the network, however the magnitude of the voltage drop depends on the transfer impedance
between the observation bus and the faulted point. In general, the transfer impedance
decreases with the distance between the faulted point and load bus. Hence load points will not
seriously be affected by faults located far away in the system. The voltage change also
depends on the driving impedance at the faulted bus. The driving impedance determines the
weakness of the bus, the stronger the faulted bus the larger the voltage drop at another bus and
vice versa.

A radial distribution system is a particular case for which the above procedure can also be
applied. The during-fault voltage at the point of common coupling (pcc) due to a fault at bus f,
assuming that the pre-fault voltage is 1 pu, can be obtained as follows
z zS zF
v pre ( k )  1  kf  1   (18)
z ff zS  zF zS  zF
where ZS is the source impedance (impedance between the source and bus k) and ZF is the
impedance between the pcc and the fault point (bus f). Expression (18) is suitable to analyse
radial systems and is the base of the method of critical distances [67].

B. Unbalanced voltage dips

Unbalanced voltage dips are caused by unsymmetrical faults requiring the use of symmetrical
components for their analysis. Because of the independence of sequences in symmetrical
systems, expression (16) can be obtained for each sequence network and used to determine
the during-fault voltage for each of the sequence components. Before the fault, bus voltages
only contain a positive-sequence component, thus pre-fault voltage matrices of zero and
negative sequences are null.
z
Vdfv  0  ΔV z (19a)
p
Vdfv  Vpref
p
 ΔV p (19b)
n
Vdfv  0  ΔV n (19c)
Phase voltages can be calculated applying the symmetrical components transformation as
follows
a
Vdfv  Vdfv
z
 Vdfv
p
 Vdfv
n
(20a)
b
Vdfv  Vdfv
z
 a 2  Vdfv
p
 a  Vdfv
n
(20b)
c
Vdfv  Vdfv
z
 a  Vdfv
p
 a 2  Vdfv
n
(20c)
Replacing (19) into (20) gives
a
Vdfv  Vpref
p
 ΔV z  ΔV p  ΔV n (21a)
b
Vdfv  a 2  Vpref
p
 ΔV z  a 2  ΔV p  a  ΔV n (21b)
c
Vdfv  a  Vpref
p
 ΔV z  a  ΔV p  a 2  ΔV n (21c)
Equations (21) provide the retained phase voltages during the fault and are valid regardless of
the fault type.

46
7.2.3 Propagation of voltage dips

The rms voltage depression caused by a fault propagates through the network and is seen as a
voltage dip at remote observation buses. The square matrix Vdfv contains the dips at each bus
of the network due to faults at each one of the buses. The during fault voltage at any bus when
a fault occurs at that bus is contained in the diagonal of Vdfv and is zero for solid three-phase
faults. Off-diagonal elements of Vdfv are the voltage dips at a general bus k due to a fault at a
general position f. Hence, column f contains the during-fault voltages at buses 1, 2…f,…N
during the fault at bus f. This means that the effect, in terms of dips, of a fault at a given bus
of the system is contained in columns of the dip matrix. This information can be graphically
presented on the one-line diagram of the power system and it is called affected area. The dip
matrix can also be read by rows. A given row k identifies that bus in the system and the
potential dips to which the load connected at this bus is exposed due to faults around the
system. This information can also be presented in a graphical way on the one-line diagram
and is called exposed area or area of vulnerability.

A. Affected area

The affected area contains the load buses that present a during fault voltage lower than a
given value due to a fault at a given point. Consider the power system of Figure 25, it has
19 load buses. Its impedance matrix has the following form
 z z1,8 z1,19 
z1,11
 0 1  1, 2 .. 1  1 .. 1 
z 2, 2 z8 , 8 z19,19 
z11,11
 
 z 2,1 z z z 2,19 
1  z 0 .. 1  2,8
z8 , 8
1  2,11
.. 1 
z19,19 
z11,11
 1,1

 : : : : : : : 
z z z z
 1  1  8, 2 .. 1  8,19 
1  8,11
8,1
Vdfv .. .. (22)
 z1,1 z 2, 2 z19,19 
z11,11
 z z 
1  11,1 1  z11, 2 z
.. 1  11,8 0 .. 1  11,19 
 z1,1 z 2, 21 z8 , 8 z19,19 
 : : : : : : : 
 z z z z 
1  19,1 1  19, 2 .. 1  19,8 1  19,11 .. 0 
 z1,1 z 2, 21 z8 , 8 z11,11 
The during-fault voltages for all buses due a fault can be obtained by applying (22). For
instance, column 11 contains the during-fault voltages at each bus when a three-phase
fault occurs at bus 11; that is, this column contains the information to draw the affected
area of the system due to a fault at bus 11. In Figure 25 two affected areas are presented
for a fault at bus 11. The larger one encloses the load buses presenting a dip more severe
than a retained voltage of 0.9 pu. The smaller one contains the load buses that will see a
depression in the voltage more severe than 0.3 pu or equivalently the retained voltage will
be less than 0.7 pu. These affected areas have been built considering faults on buses of the
system. Faults on lines are more frequent, however faults on buses cause more severe dips
in terms of magnitude and therefore are considered for building the affected areas.

B. Exposed area

The exposed area or area of vulnerability encloses the buses and line segments where
faults will cause a dip more severe than a given value. The exposed area is contained in

47
 rows of the voltage-dip matrix and as in the case of the affected area can be graphically
presented on the one-line diagram. Figure 26 presents the exposed area of bus 8; the 0.5
pu exposed area of bus 8 contains buses 8, 9 and 11 and lines connecting them indicating
that faults at these buses and lines will cause a during-fault voltage lower than 0.5 pu.
Similarly, the 0.9 pu exposed area for bus 8 contains all the buses and line segments
where faults will cause a retained voltage lower than 0.9 pu. The exposed area does not
enclose the whole length of lines connecting the surrounded buses, because some faults
on the lines might cause dips less severe than the threshold that defines the exposed area;
that is, part of a line may be actually outside the exposed area, meaning that faults in this
part of the line will lead to dips less severe than the magnitude under consideration. To
clarify, suppose that bus 6 is fed through a relatively long double circuit line from bus 9.
In such a case, faults on any of the lines occurring far from the buses 6 or 9 will be seen
as shallow voltage dips at bus 6. However, faults near bus 6 or bus 9 will be seen as more
severe dips at bus 6. The exposed area for bus 6 (severe dips) would be formed by bus 6,
segments of the lines connecting this bus, bus 9 and segments of the lines connecting bus
9. The central part of the lines between buses 9 and 6 might be outside of the exposed
area.
The exposed area has been built using the original bus impedance matrix; however, a
more precise description is needed when faults occur on lines. In order to simulate faults
on lines additional fictitious buses are needed along the lines. Those fictitious buses are
called fault positions. As more fault positions are used to calculate the exposed areas
more precise is the description of these areas and more accurate the stochastic assessment
of dips, but also bigger the computational effort needed to perform the calculation.
Consider the 0.5 pu exposed area in Figure 26. The border of this area crosses the line
between bus 9 and bus 4. In other words: for a fault at bus 4 the retained voltage at bus 8
is above 0.5 pu; for a fault at bus 9 the retained voltage at bus 8 is below 0.5 pu. To obtain
a more accurate border, the retained voltage at bus 8 should be determined for faults at
different locations on the line between bus 9 and bus 4. Also additional fault positions
should be considered along the other lines connecting these buses. This results in a larger
dip matrix Vdfv. The number of rows of this matrix equals the number of original (or
physical) buses, but the number of columns increases with the number of fault locations.
The exposed area is also the area for which a monitor, installed at a particular bus k, is
able to detect events. For example, if a monitor were installed at bus 8 and the boundary
for dip recording were adjusted to 0.5 pu, then the monitor would be able to see events in
the 0.5 pu exposed area of bus 8. When referring to power-quality monitors the exposed
area is called Monitor Reach Area.

7.3 Stochastic assessment of voltage dips

The method of fault positions is probably the most suitable tool for stochastic prediction of
voltage dips. Basically, short-circuit faults are simulated at a number of different positions
along the lines and at different buses throughout the system. Voltage magnitudes and phase
angles following the faults are calculated for each bus. The dip frequency at each location is
calculated using the information about the fault rate. After taking into account the fault rate
for each type of fault, the expected number of dips is calculated. In practise, faults are
randomly distributed along the line, and they may occur at any location in the power system.
The information about the location of the fault is very important as it will result in voltage
dips of different characteristics. Therefore, the choice of the fault positions may significantly
influence the accuracy of the prediction of the voltage dips.

48
Different approaches have been used for selection of fault positions [71], [78], [79]. In the
majority of the studies, the faults were grouped into discrete types and applied at the middle
of the line, or at each 25% of the line length. This however, may not yield high accuracy of
the prediction result. Alternatively, if the lines are divided into several fault sections (e.g.,
every 2% of the line length) the time of calculation will increase significantly. Faults might
not be distributed uniformly along the line. A particular bus or some part of the line may be
exposed to adverse environmental or weather conditions resulting in greater fault occurrence.
The fault rate at such location will be higher than the fault rate at other locations. In general,
the actual distribution of faults might be uniform, normal, exponential or a combination of
those. Different fault distributions will result in different prediction of voltage dips at different
buses. The predicted number of dips at a given bus may be sufficiently different from the

2
3

4 V k 11 <0.7

5
V k 11 <0.9
9
7 6
8
10 11

13
12 14

15 17

16 18 19

Figure 25. Affected area for a three-phase fault at bus 11 [77].

2
3

4 V 8f <0.5

5 V 8f <0.9
9
7 8 6

10 11

13
12 14

15 17

16 18 19

Figure 26. Exposed area of bus 8 for three-phase faults [77].

49
results obtained by using uniform fault distribution to warrant additional effort in modelling
the actual fault distributions.

Two illustrative studies, based on the method of fault positions, are presented in this section:
1. The main objective of the first study is to illustrate how affected and exposed areas can
be determined considering both balanced and unbalanced dips, and assuming an uniform
distribution of faults in transmission lines.
2. The objectives of the second study are:
 to determine compound areas of vulnerability and to assess the influence of
modelling of different fault distributions along the area boundary crossing lines on
prediction of the number and characteristics of voltage dips;
 to assess the influence of the modelling of fault distribution along the line on
prediction of the number and characteristics of voltage dips.

7.3.1 Example 1

A. Test system
The system used in this work is a simplified model of the National Interconnected System of
Colombia (230 kV and 500 kV), see Figure 27. The system has 87 buses and 164 lines with a
total length of 11651 km. It consists of two 230 kV grids connected by a double-circuit 500
kV line (buses 15, 16, 63 and 68). To calculate the impedance of generation buses, the
maximum short-circuit level at some buses was used. Resistive component of the generator
impedance was neglected. Power transformers interconnecting the 500 kV sector with the 230
kV sector were modelled as delta-wye solidly grounded.

B. Balanced dips
To obtain the stochastic assessment of balanced dips and their characteristics, the positive-
sequence impedance matrix was used to model the system. Line capacitance was neglected.
Three-phase faults were simulated around the system and dips due to them saved in the dip-
matrix. The following cases were analysed:
i. Base case: This simulation considers 87 fault-positions each one coinciding with a bus
position. Faults at lines were represented by the fault position at the nearest bus. The fault
rate for each fault position was calculated taking the quotient between the expected total
number of faults (buses and lines) and the number of fault positions -87 in this case-. This
is a rough estimation but it helps to identify the boundaries of dip frequencies. No
provision was made for different number of lines connected to a bus and for different line
lengths.
ii. Case 1: This simulation considers 251 fault positions with 87 at buses and 164 at lines.
Fault positions on lines were chosen in the middle of each line. Fault rates were assigned
at each fault-position corresponding to the type of fault (bus or line). The fault rate for
faults on lines was determined from the quotient between the expected annual number of
faults on lines and the number of fault positions on lines. No provision was made for
different line lengths.
iii. Case 2: This simulation considers 781 fault positions with 87 faults at buses and 694
faults at lines. Lines were divided so that each line segment was no longer than 15 km.
Fault rates were assigned corresponding to the type of fault. Fault rate for line faults was

50
 determined in the same form as in case 1. In this case the number of fault-positions per
line was dependent on the line length.

 During-fault voltages
The voltage magnitude at each bus and for each fault position was calculated. In Table VIII
the during-fault voltages at a selected number of buses are presented for some fault
positions at lines and buses. This table is part of the voltage-dip matrix, but it has been
transposed to present the results in a more suitable manner. From the table it is easy to note
that substations near to the fault experience severe dips. Consider substation 87, which is in
the last column of the table; when a fault occurs at bus 6 (far away from 87), the voltage at
87 only drops 3%. However when the fault occurs at a nearer place as 37 the voltage drop
is 39%. Also faults on the lines connecting substation 87 cause severe dips at this bus.
76 73 30 23 21
12 79 51
28
77

80
24 86
62

63
52
16
84 85 14
19 22
48 7 75
15 71
29 20
46
6 60
25 47 11
8 35
78 81 65 5
44 59 32
53 49 10
54 43

2 26 68
31 56
3
67 74
27 61

38
40 50
69
17
45 41 4
13 82

70

87
34 39 18
58
1 37 33

72
55 83 42
57
9
64
66

36

Figure 27. National Interconnected System of Colombia - Simplified model [77].

 Affected areas
Figure 28 depicts the affected areas for a fault at bus 67. It is obvious that faults at this bus
affect a larger area of the system. The figure shows that more than half of the buses
experience dips more severe than 95% when a fault occurs at bus 67. Consequently, such a
fault has the potential to cause dip-related equipment tripping for a large number of
customers. In general, the size of these affected areas depends on the short-circuit level at
the fault point, which depends on the generator scheduling. The stronger the bus (higher
short-circuit level) the bigger the influence of a short-circuit fault on the system and the
larger the affected area.

51
 Table VIII – During-fault dip voltages due to three-phase faults [77].

SUBSTATIONS >
--------------------------------- 1 6 17 24 29 37 41 55 59 87
FAULT POSITIONS

6 0.99 0.00 1.00 1.00 0.63 0.98 0.99 0.98 0.88 0.97
17 1.00 1.00 0.00 1.00 1.00 0.99 0.83 0.99 1.00 0.99
24 1.00 1.00 1.00 0.00 1.00 1.00 1.00 1.00 1.00 1.00
29 0.99 0.61 1.00 1.00 0.00 0.99 0.99 0.99 0.94 0.99
37 0.79 0.99 1.00 1.00 0.99 0.00 0.98 0.14 0.99 0.61
41 0.98 0.98 0.82 1.00 0.99 0.95 0.00 0.95 0.99 0.95
55 0.92 0.99 1.00 1.00 1.00 0.63 0.99 0.00 1.00 0.84
59 0.99 0.87 1.00 1.00 0.93 0.99 0.99 0.99 0.00 0.98
87 0.63 0.97 0.99 1.00 0.98 0.30 0.95 0.35 0.98 0.00
50% Line 27_87 0.95 0.99 1.00 1.00 0.99 0.90 0.98 0.91 0.99 0.87
50% Line 57_87 0.74 0.99 0.99 1.00 0.99 0.48 0.97 0.53 0.99 0.50
50% Line 39_41 0.98 0.98 0.83 1.00 0.99 0.96 0.10 0.96 0.99 0.95
50% Line 37_57 0.77 0.99 1.00 1.00 0.99 0.28 0.98 0.37 0.99 0.65
50% Line 37_55 0.88 0.99 1.00 1.00 1.00 0.46 0.99 0.08 1.00 0.78
50% Line 24_86 1.00 1.00 1.00 0.35 1.00 1.00 1.00 1.00 1.00 1.00

 Exposed areas
The exposed area for substation 28 is shown in Figure 29. It should be noted that part of
the transmission lines connecting buses inside the exposed area might be actually outside
the area. In Figure 29 the 95% exposed area encloses all the buses and line segments where
the occurrence of faults cause retained voltage lower than 95% in the load point 28.

 Cumulative dip frequencies

The remaining voltage is used to classify the total number of dips experienced by a given
load. In order to distinguish between severe dips and shallow ones a classification of these
in terms of magnitude is useful. A suitable way to present this information is the
cumulative histogram of voltage dips.

The frequency of occurrence of a particular voltage dip at a given location depends on

several factors, the number of faults occurring in the “electrical neighbourhood” being the
most important one. This “electrical neighbourhood” has been already identified as the
exposed area. Other factors, which do affect the magnitude of the dips and their
distribution, can be identified as system configuration, generation scheduling, number of
lines connecting the load bus, etc.

How often one of those dips occurs at a given bus k depends on the reliability of the bus f,
namely on the fault rate of the fault position f. If the fault rate of the fault position f is f
the particular dip caused by this fault will be seen f times per year. The exposed area
shows that bus k will see dips caused by faults inside this area. The expected number of
faults inside the exposed area is given by the sum of the fault rates of the fault positions
contained in the exposed area. Frequency of dips can be calculated from the dip matrix and
the fault rate corresponding to the fault positions.

52
 76 73 30 23 21
12 79 51
28
77

80
24 86
62

63
52
16
84 85 14
19 22
48 7 75
15 71
29 20
46
6 60
25 47 11
8 35
78 81 65 5
44 59 32
53 49 10
54 43
68
2 26
3
31 50% 56
74
27 61

Bus
38
40
69 67 50
17
45 41 4
13 82

70

87
34 39 18
58 95% 33
1 37

72
55 83 42
57
9
64
66

36

Figure 28. Affected area due to a three-phase fault at bus 67 [77].

76
50% 73 30 23
12 79 51 Bus 21

77
28
80
24 86
62

63 95%
52
16
84 85 14
19 22
48 7 75
15 71
29 20
46
6 60
25 47 11
8 35
78 81 65 5
44 59 32
53 49 10
54 43

2 26 68
31 56
3
67 74
27 61

38
40 50
69 17
45 41 4
13 82

70

87
34 39 18
58
1 37 33

72
55 83 42
57
9
64
66

36

Figure 29. Exposed area (50% and 95%) for bus 28 [77].

53
 The dip-matrices contain more buses than the physical buses because they include several
fault positions on lines. Buses corresponding to physical buses of the system have a fault
rate given by the actual fault rate of the bus. Fault positions on lines have a fault rate that is
a fraction of the actual line fault rate. If more than one fault position is considered on the
line the actual fault rate need to be divided by the number of fault positions in order to
calculate the frequency of the resulting dips. If only one fault position is considered on a
given line, then the actual fault rate is taken.

Once the fault rate for all fault positions is determined, the expected number of dips and
their characteristics can be determined by combining the dip-matrices and the fault rates.
Let  be the vector containing the corresponding fault rate of each Fp fault position.
λ  1.... k ... N ....  f .. Fp  (23)
Each fault will cause a drop in voltage for all buses. However only part of these fault-
caused events will be counted as dips at load buses (vkf < 0.9 pu). A large part of these
events will result in retained voltages above 0.9 pu. The number of total fault-caused
events is the same for all buses, however the resulting during-fault voltages are different
which result in different cumulative histograms. Hence, vector  contains the yearly event
rate (voltage drops) caused by faults.

For the system under study, the expected number of dips can be estimated by combining
the information contained in Table VIII and the fault rate for each fault position. This
analysis was performed for all cases. It is worth recalling that the base case only considers
fault-positions on buses, Case 1 considers 251 fault positions, 87 faults at buses and 164
faults in the middle of each line, while Case 2 considers 781 fault positions, 87 faults at
buses and 694 faults at lines.

Cumulative dip frequencies for substations 11 and 67 are presented in Figure 30 and 31,
respectively. It can be seen that the result depends on the number of fault positions. As
more fault positions are considered, the estimation of the expected number of dips becomes
more accurate; in the limit an infinite number of fault positions would be necessary to get
the probability density function for dips at each bus. The rough estimation obtained from
the base case shows the same shape as the more elaborated cases 1 and 2. This confirms
that the base case is a good starting point. However the base case cannot be seen as an
overestimation or an underestimation of the real expected number of dips. Another
conclusion is that the expected number of dips and their shape notably differs between
buses with different short-circuit levels: while the weak bus 11 shows almost 20 dips, the
strong bus 67 shows 8 dips per year deeper than 85%.

 System statistics
To characterize the system as a whole, dip indices are needed. One approach is to
characterize the system by using the average number of cumulative dip frequencies for a
given magnitude. This average number takes into account the spatial variation around the
network. It is the average of the cumulative expected number of dips -in a given bin- of the
all network buses. It could be interpreted as the expected performance of the average bus.
Another approach is to describe the system performance by means of the 95th percentile.
Again the spatial variation of the performance is taken into account. In this case the index
can be interpreted as the expected cumulative number of dips that is not exceeded by 95%
of the system buses. Table IX shows system statistics for the four simulated cases. From

54
 60.0

50.0 BASE CASE

Events/year
40.0 CASE 1

30.0 CASE 2

20.0

10.0
0.0
0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95
V (pu)

Figure 30. Cumulative dip frequency at bus 11 with maximum generation [77].
60.0

50.0 BASE CASE

CASE 1
Events/year

40.0
CASE 2
30.0
20.0

10.0

0.0
0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95
V (pu)

Figure 31. Cumulative dip frequency at bus 67 with maximum generation [77].

Table IX-A it can be seen that for dips more severe than 90%, the worse site is located in
the area of bus 27. In Table IX-B there is no clarity about the worse site for dips deeper
than 70%, however we should expect the worse site near bus 14, because Case 2 is the
most elaborated simulation. This lack of clarity could be explained by the need of a non-
homogeneous split of lines for fault position purpose. For some lines a 15 km long
segmentation of the line implies a similar short-circuit level along the line, for others this
same 15 km implies an important variation in short-circuit level. Hence some dips are over
or under-classified in terms of severity.
Table IX – System statistics based on balanced dips [77].

A.- Number of dips more severe than 0.9 pu

#Fault
Case # Events Worst site 95th % Average
positions
Base
87 2068 46.86 Bus 69 41.24 23.77
1 251 1787 40.87 Bus 27 38.10 20.53
2 781 1556 44.65 Bus 40 40.55 17.88
B.- Number of dips more severe than 0.7 pu
#Fault
Case # Events Worst site 95th % Average
positions
base
87 913 20.62 Bus 5 20.06 10.49
1 251 623 15.87 Bus 28 13.21 7.16
2 781 540 18.86 Bus 14 15.70 6.21

55
C. Unbalanced dips
The system was modelled by means of the positive and zero sequence impedance matrices.
Negative sequence impedance matrix was taken equal to the positive one. Lines were
modelled as medium-length lines perfectly transposed and their shunt parameters were
considered. Independence of sequence components is assumed, meaning that the different
sequences do not react upon each other. Transformers were modelled as delta-wye solidly
grounded. The simulations are based on Case 2. They consider 781 fault positions with 87
faults at buses and 694 faults at lines. In order to count the number of dips occurring at the
load positions, the following criterion was adopted: each fault-originated event, balanced or
unbalanced, was counted as one single dip. The minimum phase-to-neutral voltage was used
to characterize the resulting dip. During fault voltages were registered for all buses and for
each simulated fault.

Three simulation cases were developed as explained below.

i. Case A. This simulation only considers three phase faults. The fault rate for lines is
0.0134 faults per year-km and is the same for the 164 lines. The fault rate for buses is
0.08 faults per bus-year and is the same for the 87 buses.
ii. Case B. Only single phase to ground faults were simulated. Fault rates as in case A.
iii. Case C. Four types of faults were simulated: single-phase to ground, phase to phase, two
phases to ground and three-phase fault. The probabilities of these faults were assumed to
be 80% single phase, 5% two phases, 10% two phases ground and 5% three phases. Fault
rates were as in Case A.

 Exposed areas
Figure 32 presents the 85% exposed areas for buses 5, 15 and 87 deduced from
simulations corresponding to cases A and B. The lowest phase to ground voltage is used
to characterize the magnitude of the event. Faults on buses or lines inside the areas would
cause a remaining voltage in the worst phase lower than 85% at the indicated buses. The
single-phase fault exposed areas for buses 5 and 87 are almost coincident with the three-
phase fault exposed area, however a bit smaller. The three-phase fault exposed area is
bigger than the single-phase fault exposed area but they do not differ too much when
there is no power transformer inside the area. When the exposed area contains power
transformers its shape changes drastically. This can be explained by the blocking of the
zero-sequence component and the changing in the dip type due to the transformer winding
connections. For an impedance-grounded system the single-phase fault exposed area
would be significantly smaller than for three-phase faults.
The exposed areas have been drawn as closed sets to simplify the drawing, however they
might be isolated sets enclosing the part of the system in which a fault would cause a dip
more severe than a given value. To illustrate this, consider the lines connecting bus 87
and 27. Figure 33 shows the during fault voltage at bus 87 when a moving fault is applied
on the line connecting these buses. Single-phase fault and three-phase faults are shown. It
can be seen that, for both fault types, part of the line does not pertain to the 85% exposed
area of bus 87. The term critical distance is used to describe the length of the line exposed
to faults that might lead to critical dips. Hence, for this line and three-phase faults, the
critical 85%-distance would be approximately the first 40% of line plus the last 10% of it.
Figure 33 corresponds to a line connecting the load bus. The data contained in the
voltage-dip matrix also allows the description of the during fault voltage due to faults on
lines not connecting the load bus. Consider now the line connecting buses 62 and 76.

56
 76 73 30 23 21
12 79 51
28
77

80
24 86
62

63
52
16
84 85 14
19 22
48 7 75
15 71
29 20
6 60 46
25 47 11
8 35
44 59
78
32
81 65 5
53 49 10
54 43

2 26 68
31 56
3
67 74
27 61

38
40 50
69 17
45 41 4
13 82

70
87
34 39 18
58
1 37 33

72
55 83 42
57
9
64
66

36

Figure 32. 85% exposed areas for buses 5, 15 and 87.

Dashed line for single-phase fault caused dips; solid line for balanced dips [77].
3 phase fault 1 phase fault
1.0
0.9
0.8
0.7
Magnitude pu

0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 20 40 60 80 100
% line lenght

Figure 33. During-fault voltage at bus 87 due to faults on line 87-27 [77].
3 phase fault 1 phase fault

1.0
0.9
0.8
Magnitude pu

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 20 40 60 80 100
% line lenght

Figure 34. During-fault voltage at bus15 due to faults on line 76-62 [77].

57
 According to the exposed areas shown in Figure 32, only part of this line is inside the
85% exposed area of bus 15 for single-phase fault. This is confirmed by Figure 34, which
shows the resulting remaining voltage caused by faults along line 76-62. The figure also
shows the effect of the zero-sequence blocking by the transformers; the voltage drop due
to single-phase faults are only about half of the drops due to three-phase faults. These
results confirm that single-phase faults are less severe than three-phases faults, in the
sense that the worst phase shows a higher remaining voltage than the dip due to a three-
phase fault at the same position.
 Dip frequencies
Figure 35 shows the cumulative number of dips at bus 87 due to balanced and unbalanced
faults. As it would be expected the contribution of single-phase faults is dominant.
However the contribution of the different type of faults does not exactly follow the
distribution probability of faults. The exposed area can explain this. As the exposed area
for three phase faults is larger than for any other fault the resulting contribution of this
faults to the total frequency is more than the 5% that could be expected.
Table X shows that the three-phase faults contribute with more than 29% of the total dips
more severe than 0.6 pu at bus 87. It should be noted that this bus has several lines
converging to that bus. For buses at the end of a line (radial source) the distribution
probability of the faults is a reasonable indicator of the contribution of unbalanced dips to
the total number of dips.
3 ph 1 ph 2 ph 2 ph-g
40
Events/year

30
20
10
0
0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

V (pu)

Figure 35. Cumulative balanced and unbalanced dip frequency at bus 87 [77].

Table X – Cumulative dip frequency at Bus 87 [77].

Cumulative dip frequency at Bus 87

Remaining voltage
vkf pu 3 ph 1 ph 2 ph 2 ph-g Total
0.55 0.04 0.13 0.00 0.04 0.21
0.60 0.10 0.13 0.01 0.11 0.34
0.65 0.12 0.67 0.02 0.21 1.02
0.70 0.16 1.70 0.10 0.26 2.22
0.75 0.28 2.06 0.12 0.39 2.85
0.80 0.45 3.20 0.23 0.62 4.50
0.85 0.82 5.55 0.46 1.44 8.26
0.90 1.28 14.02 0.99 2.33 18.62

58
 System statistics
System statistics have been also calculated using asymmetrical faults and different number
of faults positions. Table XI presents some results. An interesting conclusion can be
derived from this table: when only symmetrical faults are considered; the stochastic
assessment of dips overestimates event frequencies. The error is more important for severe
dips. An explanation for this overestimation is, again, the fact that three-phase fault are
more severe, which causes bigger exposed areas.

Table XI – System statistics based on balanced and unbalanced dips [77].

A.- Number of dips more severe than 0.9 pu

#Fault
Case # Events Worst site 95th % Average
positions
C 781 1333 43.30 Bus 27 37.50 15.33
2 781 1556 44.65 Bus 40 40.55 17.88
B.- Number of dips more severe than 0.7 pu
#Fault
Case # Events Worst site 95th % Average
positions
C 781 415 14.11 Bus 36 11.94 4.83
2 781 540 18.86 Bus 14 15.70 6.21

7.3.2 Example 2

A. Test system

The studies were carried out using the power system shown in Figure 36 [80]. The network
consists of four 230-kV transmission system infeeds, a 33-kV (predominantly meshed) sub-
transmission network, and an 11-kV radial distribution network. The network comprised 295
buses and 296 overhead lines and underground cables. It is fed from four substations: two
400/132 kV and two 275/132 kV.

B. The area of vulnerability and associated boundary crossing lines

i. Introduction

Faults in a power system have a random nature and they can occur at any location in the
system. The fault distribution might be uniform, normal, exponential or a combination of
those. By changing distribution parameters, virtually any fault distribution pattern can be
modelled. Different fault distribution patterns will yield different dip prediction results. Once
the area of vulnerability is identified for a particular plant, its boundary may cross some lines.
This results in only a part of such line being inside the area while the other lines in the system
may be completely inside or outside the area, as shown in Figure 37. The lines that are only
partially inside the area of vulnerability are called the boundary crossing lines. The fault
distribution pattern along these lines will cause changes in the expected number of dips at a
bus of interest. The fault distribution along the lines that are completely inside the area of
vulnerability will not alter the predicted results because the whole length of those lines is
considered as a single fault position.

59
 SOURCE G
SOURCE G SOURCE G
248 247
400 278 400 178
277 293 400
298 292
299 SOURCE G 177 297 296
275 276 271
234 259 176
233 400 258 273 254
245 246 241 255
295 294 175
279 280 174
256 251 240
257 173
264 263 172 220
252 270
274 253 171 219
227 226 272 262 267 266 265
260 217 218
60 243 244 242
58 57 231 230
261 249 250
59 232 77 225 215 216 211 212 213 214
62 64 56 237 289 9 87
4 288 287 78
61 5 6 7 8 224 163 162 210
63 55 196 193
238 79 80 81 82 83 84 85
66 65 1 2 24 11 10 235 239
164 161
152 151 165 166 188 195 194
21 291 290 86
68 23 13 12 236 89 88
70 72 3 150 149 159 160 168 167 189 190
20 268 223 91 197
16 14 269
67 69 71 73 179 286 285 92 158 157
18 15 22 148 147 169 191 199 198
75 74 76 25 26 106 103 93
17 19 146 153 155 170 192
228 229 200
33 32 30 29 27 28 45 44 222 107 105 104 102 94 145 154 156 180
109 108 201
35 34 31 49 47 46 48 101 98 96 95 144 143 182 181 203
110 111 202

114 100 99 97 142 141 184 183

36 37 38 39 40 41 42 43 204
120
140 139 186 185 205
50 53 112 113 115 116 117 118 119 208
121 131 187 206
221 51 52 54 122 123

128 127 126 125 124 130 129 132 133 134 135 137 138 207
136 209

BUS 132 KV 33KV 11KV

Figure 36. Single line diagram of the test system with marked bus 229 and line 235-268 [82].

Figure 37. Example of the

area of vulnerability and the
associated boundary crossing
lines [80].

ii. Methodology

The methodology applied in this study is based on the guidelines recommended by the IEEE
Std 493-1997 (Gold Book) [81]. Two different approaches are used to calculate the total
expected number of dips due to the faults inside the area of vulnerability and due to the faults
along the boundary crossing lines. Dip performance within the area is assessed by using
equations (26) to (28). A combination of the method of fault positions and exponential fault
distribution pattern along the boundary crossing lines is used in determining the number of
voltage dips. The fault rates at each location along the boundary crossing lines are calculated
using the exponential distribution pattern and then applied to the corresponding fault
positions. After taking into account calculated fault rates, the expected number of dips due to
the faults along the boundary crossing lines is calculated.
4 n
NSBF    B  BFR (24)
i 1 j 1

4 n
NSLF    L  LL  LFR (25)
i 1 j 1

After including fault occurrence probability, equations (24) and (25) become:
4 n 3
NSBF     B  PO  BFR (26)
i 1 j 1 k 1

60
 4 n 3
NSLF     L  LL  PO  LFR (27)
i 1 j 1 k 1

CNS  NSBF  NSLF (28)

where B = Bus inside the area of vulnerability
L = Line inside the area of vulnerability
LL = Length of the line inside the area of vulnerability
PO = Fault occurrence probability (1 for symmetrical fault; 1/3 for asymmetrical
faults)
BFR = Bus fault rate
LFR = Line fault rate
CNS = Cumulative number of dips
NSBF = Number of dips due to the bus faults
NSLF = Number of dips due to the line faults
i = Type of fault
n = Total number of buses or lines inside the area of vulnerability
j = Bus or line inside the area of vulnerability
k = Number of phases.

iii. System modelling and stochastic analysis

The fault clearing times used in the study are listed in Table XII. Each fault position
represents a single fault at the system bus or at the particular fault location along each line.
Each line is divided into five equal sections as illustrated in Figure 38. The fault rates of 0.08
events per bus per year and 2.98 events per kilometre of the line per year are applied to buses
and lines respectively. Detailed system fault statistics are given in Table XIII and Table XIV.
The realistic pre-fault voltages and phase angles obtained from the load flow study are used in
all calculations. System fault analysis is performed using SIMPOW software. A dedicated
voltage-dip database is created in Microsoft Access 2000. All statistical calculations are
performed in the Visual Basic environment using specially developed software.

1 2 3 4 5 6

1 2 3 4 5 6 Figure 38. Fault positions

along the lines.

Table XII – Typical fault clearing time.

Voltage level (kV) Typical fault clearing time (ms)

132 80
33 150
11 300
Table XIII – System fault statistics for buses.
Bus fault rate (Event Distribution of type
Type of fault Fault rate per bus
per year) of fault
Single line to ground fault 0.08 0.73 0.0584
Double line to ground fault 0.08 0.17 0.0136
Line to line fault 0.08 0.06 0.0048
Three phase fault 0.08 0.04 0.0032

61
 Table XIV – System fault statistics for lines.

Line fault rate Distribution of Fault rate per Fault rate per line-
Type of fault
(Event/100 km/year) type of fault line fault position
Single line to
2.98 0.73 2.175 0.3626
ground fault
Double line to
2.98 0.17 0.507 0.0844
ground fault
Line to line fault 2.98 0.06 0.179 0.0298
Three phase
2.98 0.04 0.119 0.0199
fault

iv. Study procedure

In order to investigate the influence of different fault distribution patterns along the boundary
crossing lines on prediction of number of dips, the following procedure was used:
1. Symmetrical and asymmetrical faults were simulated in each phase separately for
every fault position throughout the system.
2. Voltage dip magnitudes at each system bus and corresponding fault locations were
recorded.
3. A voltage dip database containing system fault analysis data, the associated fault
locations and typical fault clearing times was created.
4. Two customer sites, bus 229 and bus 100, from the 11-kV network were arbitrarily
selected as buses of interest.
5. Area of vulnerability and the associated boundary crossing lines for each type of fault
were determined using the equipment sensitivity level (voltage threshold) as a filtering
criterion in database. Voltage magnitude thresholds of 0.7 p.u., 0.5 p.u. and 0.3 p.u.
were applied to each phase of the bus of interest.
6. Number of dips at buses 229 and 100 due to the faults inside the area of vulnerability
was calculated using equations (26) to (28).
7. Fault rates for each fault position along the boundary crossing lines were determined
using exponential distribution and the number of dips due to these faults was
calculated again.
8. Finally, the influence of modelling of different fault distribution patterns along the
boundary crossing lines was investigated.

v. Case studies

Four different case studies were considered:

a) Minimum case: All faults along the boundary crossing lines occurred outside the area
of vulnerability, resulting in minimum expected number of dips at buses 229 and 100.
b) Maximum case: All faults along the boundary crossing lines occurred inside the area
of vulnerability, resulting in maximum expected number of dips at buses 229 and 100.
c) Exponential 1: The exponential fault distribution was used ( f ( x)  e  x , =2) along
the length of the boundary crossing lines. (exponential increase from the outside of the
vulnerability area, as shown in Figure 39).

62
 d) Exponential 2: The exponential fault distribution was used ( f ( x)  e  x , =2) along
the length of the boundary crossing lines. (exponential increase from the inside of the
vulnerability area, as shown in Figure 39).

vi. Results

a) Area of vulnerability due to three-phase faults

Typically, symmetrical fault results in the same degree of severity of voltage dips in each
phase of a three-phase system. Hence, identifying the area of vulnerability for each phase can
be done easily, and each of them will be identical. Area of vulnerability determined using
voltage threshold of 0.7 p.u. for bus 100 is illustrated in Figure 39.

b) Area of vulnerability due to single line-to-ground faults

The area for bus 229 is illustrated in Figure 40. Since the SLGFs are applied to each phase of
a three-phase system, the dip magnitude in each phase will be different from the dip
magnitudes in the other phases depending on the phase at which the fault occurs and the
transformer winding connections. Therefore, the identified areas of vulnerability by taking
into account effect of faults on each phase separately are likely to be different. Table XV,
however, shows that only faults on two phases affect the number of voltage dips on the study
phase. This means that only two different areas for the sensitive load at the study phase exist.
In some cases only one area of vulnerability exists for a single-phase connected load. This
case, however, is rare.
Table XV shows that higher number of buses and lines fall into the area of vulnerability when
the faulted phase and the study phase coincide. Otherwise, the area due to the faults on other
phases is relatively small. 47 buses and 255 line-fault positions are inside the area when the
study phase coincides with the faulted phase. 255 line-fault positions correspond to 40 lines
being completely within the area of vulnerability and 9 boundary-crossing lines. The other
area of vulnerability due to the faults on other phases includes 2 buses and 17 line-fault
positions corresponding to 1 line completely inside the area of vulnerability and 6 boundary-
crossing lines. The resulting area of vulnerability for the sensitive load on the study phase is
obtained by merging the corresponding areas. For example, the area of vulnerability for the
sensitive load with 0.7 p.u. voltage threshold connected to phase A of Bus 229, as shown in
Figure 40, is the result of combination of the area of vulnerability due to the faults on the
study phase (grey shaded area in Figure 40) and due to the faults on non-study phases (dark
grey shaded area in Figure 40). There are 15 boundary- crossing lines in total associated with
that area. Since the power system is balanced, identified area of vulnerability and associated
boundary crossing lines for phase A are the same for phases B and C.

Table XV – Single line-to-ground faults - Area determined by 0.7 p.u. voltage threshold [80].
Number of buses Total number of line-fault positions
Study phase inside the area inside the area
of Bus 229 Faulted phase Faulted phase
A B C A B C
A 47 2 0 255 17 0
B 0 47 2 0 255 17
C 2 0 47 17 0 155

63
 Figure 39. Area of vulnerability due to three-phase faults for bus 100 [80].

Figure 40. Area of vulnerability due to single line-to-ground faults for bus 229 [80].

c) Area of vulnerability due to double line-to-ground faults

The same as before, sensitive equipment is connected to phase A of bus 229. Consider the
area of vulnerability due to the faults on any two phases of a three-phase system. As shown in
Table XVI, all the 20 buses are within the area if the fault happens on phases B and C. The
number of buses inside the area increases to 112 when the faults occur on phases A and C.
Similarly, 95 and 740 fault positions along the lines fall into the area when the faults occur on
phases B and C and phases A and C, respectively. This means that the number of voltage dips
experienced by a single phase connected load at the bus of interest highly depends on faulted
phases even when the faults occur at the same fault positions. Unlike the area of vulnerability
due to a SLGF, three different areas are identified depending on the faulted phases. Resultant
boundary crossing lines are the result of summation of crossing lines from these areas.

64
 Table XVI – Double line-to-ground faults - Area determined by 0.7 p.u. voltage threshold [80].
Number of buses Total number of line-fault positions
Study phase inside the area inside the area
of Bus 229 Faulted phases Faulted phases
AB BC CA AB BC CA
A 51 20 12 290 95 740
B 112 51 20 740 290 95
C 20 112 51 95 740 290

d) Area of vulnerability due to line-to-line faults

The area of vulnerability caused by the line-to-line faults can be identified using the same
approach. Two different areas appear in this case, as shown in Table XVII.

Table XVII – Line-to-line faults - Area determined by 0.7 p.u. voltage threshold [80].
Number of buses Total number of line-fault positions
Study phase inside the area inside the area
of Bus 229 Faulted phases Faulted phases
AB BC CA AB BC CA
A 37 0 117 201 0 774
B 117 37 0 774 201 0
C 0 117 37 0 774 201

e) The influence of fault distribution along the boundary crossing lines

The influence of different fault distributions along the boundary crossing lines is illustrated in
Figure 41. The figure clearly shows that the number of voltage dips in each range (selected for
the illustration of the effect) is affected by the fault distributions along the boundary crossing
lines. The number of voltage dips with magnitude lower than 0.7 p.u., for example, can vary
between 82.69 and 101.49 (i.e., approx. 23%), and between 112.81 and 129.77 (i.e., approx.
15%), at bus 229 and bus 100 respectively. The major difference in cumulative dip frequency
appears in the case of voltage dips with lower magnitudes. At bus 229, the difference between
the maximum and minimum number of voltage dips with magnitudes lower than 0.7 p.u is
18.8 dips. The difference however, increases to 22.47 dips in the case of voltage dips with
magnitudes lower than 0.3 p.u.
Minimum Case Exponential 1 Exponential 2 Maximum Case
Number of dips

150
100

50
0
<0.3 <0.5 <0.7
Voltage dip magnitude (p.u.)
Figure 41. Cumulative voltage dip frequency for bus 229 (Adapted from [80]).

65
C. The influence of fault distribution on stochastic prediction of voltage dips

i. System modelling

Symmetrical and asymmetrical faults were simulated at various, randomly chosen locations
and at different voltage levels [82]. Realistic pre-fault voltages, obtained from the load flow
study were used in all calculations. The IPSA software package was used for fault
calculations and MATLAB for calculating different fault distribution rates and the expected
number of voltage dips.

ii. Study procedure

In order to assess the influence of different fault distributions on voltage dips the following
procedure was used:
1. Firstly, all types of faults at all buses in the test network were simulated and the area
of vulnerability (voltage drop below 95%) was determined for one particular bus (bus
number 229 at 11 kV level). The selected bus and the single line diagram of the test
network are presented in Figure 36.
2. Secondly, the line (between sending end bus 235 and receiving end bus 268, at 33 kV
voltage level), within the identified area of vulnerability, was selected for the study of
the influence of different fault distributions on voltage dips at bus 229.
3. Thirdly, the line was divided into 50 equal sections, and the voltage magnitude and
phase shift at bus number 229 following the fault lasting 100ms at each position were
calculated for all types of faults (three-phase fault, phase A-to-ground fault, phase A-
to-B fault, phase A-to-B-to-ground fault). Fifty fault positions (one per sections) were
used for all types of faults except for the three-phase faults. For the three-phase faults
only 25 fault positions were used, as the total number of the three-phase faults along
the line was only 25 according to the assumption made. The fault rates at each location
were calculated using different fault distributions (uniform, normal, exponential) along
the line.
4. Finally, after taking into account the fault rates, the expected number of voltage dips
(having different characteristics, i.e., magnitude and phase shift) at bus 229 was
calculated.

iii. Case studies

It was assumed that the total of 1250 faults occur along the line. The number of different
types of faults is given in Table XVIII.

Table XVIII – Number of different faults simulated along the line [82].
Percentage of different type Total number of faults
Fault type
of fault (%) for specific fault type
Three-phase fault 2 25
Single line-to-ground 85 1063
Line-to-line 8 100
Double line-to-ground 5 62

The following case studies were considered:

66
 1. Uniform distribution of faults along the line.
2. Normal distribution of faults with the same mean value and different standard
deviations (three sub-cases).
3. Normal distribution of faults with different mean values and the same standard
deviation (three sub-cases).
4. Exponential distribution of faults with different mean values - exponential decay from
the sending end (bus 235) of the line to the receiving end (bus 268) of the line (three
sub-cases).
5. Exponential distribution of faults with different mean values - exponential decay from
the receiving end (bus 268) of the line to the sending end (bus 235) of the line (three
sub-cases).

The fault distributions used in simulations are illustrated in Figures 42 to 45.

(Note: For case 3 when the mean value of normal distribution was varied, the edges of normal
distribution curve went out of the line length range causing an error (i.e., the fault rate for
each type of fault given by normal distribution was not the same as the total expected number
of faults.) Hence, the fault rate was normalised to get the same total expected number of
faults. The shape of the distribution curve remained the same but related fault rate at each
point became higher.)

iv. Results

a) The influence of different fault distributions

The influence of modelling of different fault distributions along the line on the expected dip
frequency at bus 229 is illustrated in Figures 46. It can be seen that the major difference in dip
frequency appears in all phases for different fault distributions along the line. The majority of
dips will have magnitudes above 0.8 p.u. The number of dips having different magnitudes in
different phases is different. The differences in dip frequency in different voltage ranges and
different phases obtained as a consequence of modelling different fault distributions along the
line varies between a few dips and several hundred dips (e.g. dips in phase B in the range
between 0.7 and 0.9 p.u.). Equipment connected to the phase C will experience the largest
number of dips with magnitudes lower than 0.5 p.u. From the cumulative frequency plots it
has been found that the major difference in cumulative dip frequency appears in phases A and
B for the different fault distributions along the line. This is to be expected as these two phases
were involved in the majority of the faults simulated. Similar differences in phase angle jumps
were observed. The frequency of phase angle jumps in different ranges for the phase A is
illustrated in Figure 47. It can be seen from this figure that the phase angle jump for the
majority of dips is positive, i.e., the voltage during the fault lags the voltage before the fault.

b) The influence of different distribution parameters

The influence of the fault distribution parameters on dip prediction is illustrated in Figures 48
and 49. Figure 48 shows the frequency of voltage dips with different magnitudes obtained
using normal distribution with different parameters. It has been found that the variation in
distribution mean has bigger effect on dip frequency in different voltage ranges than the
variation in standard deviation.

67
 f(x)
 = 0.05 (Case 1)

 = 0.15 (Base Case)

 = 0.5 (Case 2)

 = 0.5 x
Line Length

Figure 42. Normal distribution with the same mean value and different standard deviation [82].

Figure 43. Normal distribution with different mean value and the same standard deviation [82].
f(x)

 = 4 (Case 1)

 = 2 (Base Case)

 = 1 (Case 2)

0 0.5 1 x
Line Length
Figure 44. Exponential distribution with different mean value (exponential decay from sending end to
receiving end) [82].
f(x)

 = 4 (Case 1)

 = 2 (Base Case)

 = 1 (Case 2)

x 1 0.5 0
Line Length
Figure 45. Exponential distribution with different mean value (exponential decay from receiving end
to sending end) [82].

68
 uniform normal exponential 1 exponential 2

Frequency (events/year)
100
90
80
70
60
50
40
30
20
10
0
0.1-0.2 0.3-0.4 0.5-0.6 0.7-0.8 0.9-1.0
Voltage dip magnitude (pu)

a) Phase A
uniform normal exponential 1 exponential 2
Frequency (events/year)

14

12
10

4
2

0
0.1-0.2 0.3-0.4 0.5-0.6 0.7-0.8 0.9-1.0
Voltage dip magnitude (pu)

b) Phase B
uniform normal exponential 1 exponential 2
Frequency (events/year)

140
120

100
80

60
40

20
0
0.1-0.2 0.3-0.4 0.5-0.6 0.7-0.8 0.9-1.0
Voltage dip magnitude (pu)

c) Phase C
Figure 46. Voltage dip frequency at bus 229 for different fault distributions along the line 235-268
(Adapted from [82]).
uniform normal exponential 1 exponential 2
Frequency (events/year)

90
80
70
60
50
40
30
20
10
0
(-30)-(-25) (-15)-(-10) 0-5 15-20 30-35 45-50
Phase angle jump (deg)

Figure 47. Phase angle jump frequency in phase A at bus 229 for different fault distributions along
the line 235-268 (Adapted from [82]).

69
 base case case 1 case 2

Frequency (events/year)
30

25

20

15
10

0
0.1-0.2 0.3-0.4 0.5-0.6 0.7-0.8 0.9-1.0
Voltage dip magnitude (pu)

a) Different standard deviations

base case case 1 case 2
Frequency (events/year)

50
45
40
35
30
25
20
15
10
5
0
0.1-0.2 0.3-0.4 0.5-0.6 0.7-0.8 0.9-1.0
Voltage dip magnitude (pu)

b) Different distribution means

Figure 48. Voltage dip frequency in phase A at bus 229 for normal distribution of faults along the
line 235-268 (Adapted from [82]).
base case case 1 case 2
Frequency (events/year)

12
10

8
6
4

2
0
0.1-0.2 0.3-0.4 0.5-0.6 0.7-0.8 0.9-1.0
Voltage dip magnitude (pu)

a) Decay from sending end to receiving end of the line

base case case 1 case 2
Frequency (events/year)

25

20

15

10

0
0.1-0.2 0.3-0.4 0.5-0.6 0.7-0.8 0.9-1.0
Voltage dip magnitude (pu)

b) Decay from receiving end to sending end of the line

Figure 49. Voltage dip frequency in phase A at bus 229 for exponential distribution of faults along
the line 235-268 – different distribution means (Adapted from [82]).

70
 base case case 1 case 2

Frequency (events/year)
180
160
140
120
100
80
60
40
20
0
(-30)-(-25) (-15)-(-10) 0-5 15-20 30-35 45-50
Phase angle jump (deg)

a) Different standard deviations

base case case 1 case 2
Frequency (events/year)

200
180
160
140
120
100
80
60
40
20
0
(-30)-(-25) (-15)-(-10) 0-5 15-20 30-35 45-50
Phase angle jump (deg)

b) Different distribution means

Figure 50. Frequency of phase angle jump in phase A at bus 229 for normal distribution of faults
along the line 235-268 (Adapted from [82]).
base case case 1 case 2
Frequency (events/year)

45
40
35
30
25
20
15
10
5
0
(-30)-(-25) (-15)-(-10) 0-5 15-20 30-35 45-50
Phase angle jump (deg)

a) Decay from sending end to receiving end of the line

base case case 1 case 2
Frequency (events/year)

140
120
100
80
60
40
20
0
(-30)-(-25) (-15)-(-10) 0-5 15-20 30-35 45-50
Phase angle jump (deg)

b) Decay from receiving end to sending end of the line

Figure 51. Frequency of phase angle jump in phase A at bus 229 for exponential distribution of faults
along the line 235-268 – different distribution means (Adapted from [82]).

71
The influence of the parameters of the exponential distribution is illustrated in Figure 49. It
can be seen from these figures that different distribution parameters result in different dip
frequencies in different voltage ranges. The dependence of dip frequency on distribution
parameters though still very pronounced, is smaller than the dependence on the type of
distribution used. Dependence of dip frequency on the parameters of normal distribution is
bigger than the dependence on the parameters of exponential distribution (for the ranges of
distribution parameters used in simulations). As far as the influence of different distribution
parameters on the phase angle jump is concerned it was found that the phase angle jump in
phase A is the most affected. (This was expected having in mind the types of faults modelled.)
The effect of different distribution parameters on phase angle jump in phase A is illustrated in
Figures 50 and 51.

v. Conclusions

The study has shown that different fault distributions along the line lead to different number
and characteristics of voltage dips even though the total number of faults along the boundary
lines is the same. The difference in the expected number of dips is bigger for the dips with
lower magnitudes. It is however, still very significant even for the dips with higher
magnitudes. The size and the shape of the area of vulnerability are not only dependent on the
sensitivity level of the equipment but also on the relationship between the faulted phase and
the study phase. The change in the size and the shape of the area of vulnerability results in
different number of associated boundary crossing lines and such leads to different estimate of
the number of voltage dips at the bus of interest. In order to achieve higher accuracy of
prediction of number and characteristics of voltage dips at the bus of interest it is important to
take into account the effect of faults on non-study phases when the area of vulnerability is
identified, and to properly model the fault distribution along the boundary crossing lines.

The influence of different fault distributions on frequency of voltage dips within different
voltage ranges can be very high. The influence of parameters of different fault distributions on
number and characteristics of voltage dips is also illustrated. It was found that different
distribution parameters have different effects depending on the type of distribution used. The
sensitivity of the number of dips and dip characteristics to distribution parameter variation is
smaller than to the type of the distributions used. The variation in the results however, is still
very significant. The results emphasise the importance of a proper modelling of fault
distribution along the lines within relevant area of vulnerability. This is particularly important
in the case of the lines in close electrical proximity to the customer bus if the correct
information about the number and characteristics of voltage dips is needed.

The variation in expected number of dips and their characteristics indicates that the proper
modelling of fault distribution along the line is particularly important in cases when the
border of the area of vulnerability is crossing the line. In such cases the method used for
modelling of fault distribution along the line may significantly alter the number and
characteristics of voltage dips expected at the customer bus. Depending on the fault
distribution along the transmission line (in an electrical proximity to the bus of interest) a
single phase equipment connected to different phases will experience different number of
voltage dips having different characteristics. Depending on the sensitivity threshold of the
equipment connected to a particular phase and dip characteristics in that phase, there will be
different number of equipment failures or misoperations in the plant. The cumulative loss in
the revenue that might result due to the voltage dips therefore, should be assessed based on
the information about the voltage dip performance of each phase (A, B or C) separately.

72
8. Voltage Dip Indices
8.1 Introduction

Voltage dip indices can be used to quantify a site or a system performance. The characteristic
exposure that a typical distribution system encounters needs to be quantified in order to guide
manufacturers in the appropriate design of ride-through alternatives for user load equipment.
For many years, the only indices defined to quantify rms variation service quality were the
sustained interruption indices (SAIFI, CAIDI, etc.), which quantify momentary interruption
performance as defined in the IEEE Standard 1366 [83]. On the other hand IEEE Standard
1159, Recommended Practice on Monitoring Electric Power Quality [10], defines magnitude
ranges for dips, swells and interruptions and suggests that the terms dip, swell, and
interruption be preceded by a modifier describing the duration of the event (instantaneous,
momentary, temporary, or sustained). Rms variations are therefore classified by the
magnitude and duration of the disturbances.

The present draft of IEEE Std. P1564 identifies voltage dip indices for customers, vendors
and electric utilities and the method of calculating such indices [5]. The draft proposes a five-
step procedure to give a value to the performance of a power system:
1. Obtain sampled voltages with a certain sampling rate and resolution.
2. Calculate event characteristics from the sampled voltages.
3. Calculate single-event indices from the event characteristics
4. Calculate site indices from single-event indices of all events measured over a period of
time.
5. Calculate system indices from the site indices for all sites within the system.

The first two steps have been introduced in the second chapter of this document. A short
summary is included below. A more detailed description of the other steps is presented in the
subsequent sections.

The following section presents new concepts related to the characterization of a single event.
Indices proposed to characterize a site and a system, respectively, are described in the
subsequent sections. The last section of this chapter includes some test cases on calculations
of voltage dip indices.

Readers interested in this subject will find a good survey on voltage dip indices and other
power quality indices in the report published by the Joint WG CIGREC4.05/CIRED, Power
Quality Indices and Objectives [84].

8.2 Indices for event characterization

8.2.1 Magnitude and duration

According to IEC 61000-4-30, a voltage dip or voltage swell should be characterized by its
duration (the time the rms voltage stays below the threshold) and its retained voltage (the
lowest rms voltage during the event). Instead of retained voltage the depth (the difference
between the retained voltage and a reference voltage) may be used.

73
The choice of a dip threshold is essential for determining the duration of the event. This
choice of threshold is also important for counting events, as events are only counted as
voltage dips when the rms voltage drops below the threshold. Dip threshold can be a
percentage of either nominal or declared voltage, or a percentage of the sliding voltage
reference, which takes into account the actual voltage level prior to the occurrence of a dip.
The user shall declare the reference voltage in use.

Voltage dip envelopes may not be rectangular. As a consequence, for a given voltage dip, the
measured duration is dependent on the selected dip threshold value. The shape of the envelope
may be assessed using several dip thresholds set within the range of voltage dip and voltage
interruption threshold detection. The latter concept also called “Time Below Specified Voltage
Threshold” is presented in more details in reference [85]. Another method for treating non-
rectangular dips is part of IEEE Std 493 [81] and discussed in detail in [69]. In the latter
method, characteristics are no longer determined for each individual event, but the rms
voltage versus time curves are directly used to obtain a so-called “voltage-dip co-ordination
chart”.

An annex to IEC 61000-4-30 mentions other characteristics for voltage dips: phase angle
shift, point-on-wave, three-phase unbalance, missing voltage and distortion during the dip.
The use of additional characteristics and indices may give additional information on the origin
of the event, on the system and on the effect of the dip on equipment. Even though several of
these terms are used in the power-quality literature there is no consistent set of definitions.

IEC 61000-2-8 document also refers to IEC 61000-4-30 for measurement, but introduces a
number of additional recommendations for calculating voltage-dip indices. Recommended
values are 90% and 91% for dip-start threshold and dip-end threshold, respectively, and 10%
for the interruption threshold. Dips involving more than one phase should be designated as a
single event if they overlap in time. Voltage dips with a retained voltage below the
interruption threshold can be classified as momentary interruptions. The duration of the event
would be the amount of time the rms voltage is below the dip threshold. Alternatively the
duration of a momentary interruption can be defined as the amount of time the rms voltage is
below the interruption threshold. The difference between these two definitions is normally
small compared to the duration of momentary interruptions.

Voltage swells can be characterized in the same way as voltage dips. The rms voltage is again
used as a characteristic versus time. The single-event indices are the “duration” and the
“retained voltage”. The duration equals the amount of time the rms voltage is above the swell
threshold. The retained voltage is the highest value of the rms voltage. The recommended
value for the swell threshold is 110% of the declared voltage or of the sliding-reference
voltage.

For multi-channel measurements the rms voltage versus time is calculated for each channel
separately. For single-phase measurements only the voltage magnitude versus time is needed
for further processing. The voltage phase angle versus time is needed with some methods to
calculate characteristics for three-phase measurements.

IEC 61000-4-30 distinguishes between single-channel and multi-channel measurements. For

multi-channel measurements,
 the retained voltage is the lowest of the retained voltage for the individual phases.

74
  the starting instant of the dip is the moment the rms voltage in one of the phases drops
below the dip-starting threshold.
 the ending instant of the dip is the moment all rms voltage have recovered above the
dip-ending threshold.
 the duration is obtained as the time difference between the ending instant and the
starting instance.

The event may end in a different channel as the one in which it started. The same result is
obtained by defining the multi-channel rms voltage at any instant as the lowest of the rms
voltages in the different channels. The duration of the multi-channel event is defined as the
time during which the multi-channel rms voltage is below the threshold. In case only retained
voltage and duration in the different channels are available, the lowest retained voltage and
the longest duration are used to characterize the multi-channel event.

8.2.2 Other voltage dip indices

The IEEE P1564 draft introduces two other single-event indices:

A. The voltage dip energy is defined as:

T  V (t )  2 
VS   1    dt (29)
0  V nom  
where V(t) is the rms voltage during the event and Vnom the nominal voltage. The
integration is taken over the duration of the event; i.e. for all values of the rms voltage
below the threshold. The rms voltage can be expressed in any unit (Volt, kV, per unit) as
long as the nominal voltage is expressed in the same unit. The voltage-dip energy has the
unit of time; it may be expressed in cycles, milliseconds or seconds.

B. The voltage-dip severity is calculated from the retained voltage (in pu) and the duration of
a voltage dip in combination with a reference curve. It is recommended to use the SEMI
curve as a reference, but the method works equally well with other methods.
From the retained voltage V and the event duration d the event voltage-dip severity is
calculated as follows:
1V
Se  (37)
1  Vcurve (d )
where Vcurve(d) is the retained voltage of the reference curve for the same duration. For an
event on the reference curve the voltage-dip severity equals one; for an event above the
curve the index is less than one; for an event below the curve the index is greater than
one. For events with the retained voltage above the voltage-dip threshold (with 90% being
the recommended value) the voltage-dip severity is equal to zero.
A voltage-swell severity can be defined for voltage swells in a similar way, by choosing a
reference curve. It is however recommended treating voltage dips and swells as separate
events. A voltage dip and a voltage swell with the same “severity” will not have a similar
effect on equipment.

For other voltage dip indices see references [86] and [87].

75
8.2.3 Aggregation [5]

The definition of “event” is an important aspect because the total count of “events” would be
very different if three-phase measurements were counted as three single-phase measurements.
An approach would consist on collecting small elemental components of measurements (i.e.,
measurement components) and aggregate them at analysis time. The word “aggregate”
literally refers to the collection of units or parts into a mass or whole. Aggregation refers to
the data reduction technique of collecting many distinct measurement components into a
single aggregate event for the purpose of computing system performance indices. How the
measurements are combined depends on the specific needs.

A. Measurement aggregation: Many monitoring instruments will record one or more phases
during an rms variation. For example, a three-phase voltage dip may result in a meter
recording one measurement for each phase. In conducting measurement aggregation, we
choose to represent the multiple phase measurements as only one measurement. A
common practice is to choose the phase measurement that exhibits the greatest deviation
from nominal voltage.
B. Time aggregation: The time aggregation is counting a single event if there is succession
of events within a short time, generally caused by a single power system event. This is
generally accepted practice in indexing voltage dip events. If the customer equipment is
impacted by a voltage dip event, it is unlikely that the equipment will be up and running
and impacted by a succeeding event during the aggregation time period. The EPRI DPQ
project’s results published in IEEE journals used 60-second aggregation time, but also
explored using 120 seconds and 300 seconds. An example of a longer aggregation time is
seen in the Detroit Edison special manufacturing contracts (SMC) with its automotive
manufacturers. The contracts require bookkeeping of the worst voltage dip in fifteen-
minute aggregation intervals.
C. Spatial aggregation: Spatial aggregation refers to finding the worst voltage dips from
more than one monitoring point. An example where this is necessary is with the Detroit
Edison SMC computations, which need only to determine the lowest voltage dip
measured by all of the meters monitoring the various service entrances of a facility.
Spatial aggregation has also been employed when meters are configured to monitor only a
single phase. In this case, three meters, each monitoring one phase of a feeder, can be
combined to give the voltage dip performance of the bus supplying the feeder. In using
spatial aggregation to reduce the number of rms variation measurements, the
measurements from multiple monitoring instruments are combined into a single
measurement. An example on when this has an application is if one is interested in
computing rms variation indices at a single substation that is monitored at multiple buses.
Another example is if one if computing rms variation indices for a single industrial
facility that is monitored at each service entrance of its supplying feeders.

8.3 Indices for site characterization

Single-event characteristics are used as obtained from events recorded at a given site over a
given period, typically one month or one year, as input to the site indices. When publishing
site indices it is important to indicate which method has been used to obtain the single-event
characteristics: e.g. lowest of the three phase-to-ground voltages; lowest of the three phase-to-
phase voltages; characteristic voltage obtained from three-phase measurements.

76
8.3.1 SARFI indices

SARFI is an acronym that stands for System Average RMS Variation Frequency Index. It is a
power quality index that provides a count or rate of voltage dips, swells, and/or interruptions
for a system. The size of the system is scalable: it can be defined as a single monitoring
location, a single customer service, a feeder, a substation, groups of substations, or for an
entire power delivery system. There are two types of SARFI indices: SARFI-X and SARFI-
Curve.
1) SARFI-X corresponds to a count or rate of voltage dips, swell and/or interruptions below
a voltage threshold. For example, SARFI-90 considers voltage dips and interruptions that
are below 0.90 per unit, or 90% of the reference voltage. SARFI-70 considers voltage
dips and interruptions that are below 0.70 per unit, or 70% of the reference voltage. And
SARFI-110 considers voltage swells that are above 1.1 per unit, or 110% of the reference
voltage. These indices are meant to assess short-duration rms variation events only,
meaning that only those events with durations less than the minimum duration of a
sustained interruption (5 minutes) are included in its computation. For voltage swells
SARFI-X is defined as the number of events per year with a retained voltage exceeding X
percent (for X greater than 100).
2) SARFI-Curve corresponds to a rate of voltage dips below an equipment compatibility
curve. For example SARFI-CBEMA considers voltage dips and interruptions that are
below the lower CBEMA curve. SARFI-ITIC considers voltage dips and interruptions
that are below the lower ITIC curve.

8.3.2 Voltage dip tables

A commonly used method of presenting the performance of a site is by means of a voltage dip
table. The columns of the tables represent ranges of voltage-dip duration, the columns
represent ranges of retained voltage. Each cell in the table gives the number of events with the
corresponding range of retained voltage and duration. Strictly speaking a voltage-dip table is
not an index but a way of presenting a set of indices. Each element in the table can be used as
an index, just like there are multiple choices for the SARFI indices.

In the two examples shown below the retained voltage is expressed in percent, but it may also
be expressed in Volts. The duration should be expressed in milliseconds, seconds or minutes,
with the exception of the first row in case the maximum duration is one cycle.

IEC 61000-4-11 prescribes a number of retained voltage and duration values for testing
equipment against voltage dips and short interruptions. Table XIX shows the recommended
tabulation in which each cell is intended to contain the number of dips of the corresponding
magnitude and duration within a specified period, usually one year.

Table XIX – Voltage dip table based on the values in IEC 61000-4-11 [17].
Retained
<1 cycle 1 cycle-200 ms 0.2-0.5s 0.5-5 s 5s-1 min
voltage
70-80%
40-70%
10-40%
10%

77
IEC 61000-2-8 table: The voltage-dip table as proposed in draft IEC 61000-2-8 is shown in
Table XX. The main difference with the UNIPEDE table is in the higher resolution in the
remaining voltage ranges. Also is an additional duration range added with 250 milliseconds as
a limit.

Table XX – Voltage dip table according to IEC 61000-2-8 [3].

Retained <0.1 s 0.1-0.25 s 0.25-0.5 s 0.5-1 s 1-3 s 3-20 s 20-60 s 1-5 min
voltage
80-90%
70-80%
60-70%
50-60%
40-50%
30-40%
20-30%
10-20%
10%

8.4 Indices for system characterization

System indices are used by the network operator to assess the performance of a whole system;
e.g. they can be used to compare year-to-year performance. The results of such a performance
assessment or comparison can be used as a basis for improvements in the system.

System indices are calculated from the site indices obtained for a number of sites. Two
principally different methods can be distinguished for calculating system indices.
 The system index is defined as a weighted average of the site indices. To determine
weighting factors, system and load information is needed, but often a unity weighting
factor is used for all monitored sites.
 The system index is defined as the value not exceeded for 95% of the sites (the 95
percentile of the site indices). To be able to determine a 95 percentile a reasonable
number of sites has to be monitored, at least 20. When between 10 and 20 sites are
available, the 90 percentile may be used instead. For less than 10 sites, the weighted
average and the maximum value should be reported.

SARFI indices are obtained as the average of the indices for the different sites. They thus give
the “average voltage quality” over the whole system. When using SARFI indices to describe
individual sites, it is possible to give a 95 percentile to characterize the quality of the whole
system. This is however not common practice.

When voltage-dip tables are used, both average values over all sites and 95 percentile values
can be used. When average values are used, weighting of the values may be considered.
Weighing is also possible when using the 95 percentile but less useful unless a very large
number of sites is being monitored. The voltage-dip table for the whole system may contain a
smaller number of cells than for individual sites. For individual sites a certain level of detail is
needed to determine the compatibility of sensitive equipment with the supply. For system
indices a lesser level of detail may be more appropriate as that makes e.g. the study of year-to-
year variations easier.

78
8.5 Voltage dip index evaluation

The calculation of voltage dip indices is straightforward from the results derived with a
stochastic prediction. Not much experience is presently available on the application of indices
to analyze the voltage dip performance of transmission and distribution systems. In addition,
most publications deal only with the count of event frequency (i.e. SARFI indices). The
following subsections present three examples of SARFI calculations. In the first one several
SARFI indices are derived and analyzed for a distribution network from the stochastic
prediction of voltage dips using a time-domain simulation tool. The last two examples use the
calculation of SARFI indices to analyze the impact of distributed generation at transmission
and distribution levels.

8.5.1 Example 1: Calculation of indices for a distribution network

A. Introduction

The aim of this study is to predict the voltage dip performance of a distribution network by
estimating voltage dip indices and using a time-domain simulation tool. The work is based on
the application of a Monte Carlo procedure developed using capabilities of the ATP package
[34], [88], a well known EMTP-like tool.

The tasks needed to obtain voltage dip indices can be summarized as follows: internal
capabilities of the transients program are used for the development of power components and
load models; capabilities of the program are linked to external routines for assessment of
voltage dips using a Monte Carlo method; the output results are post-processed to obtain
voltage dip indices.

The study will be performed by assuming that voltage dips are caused only by faults, there is
no distributed generation in the system, and mitigation devices are not installed.

B. Modelling guidelines

Guidelines used in this case are those presented in Chapter 6. Simulations have been based on
a constant impedance representation of loads, which are only installed at LV buses. A
constant resistance model is used for representing the fault impedance.

C. Test system

Figure 52 shows the diagram of the test system and the characteristic of the protective devices
[89]. The lower voltage side of the substation transformer is grounded by means of a zig-zag
reactor of 75  per phase. This grounding system limits overcurrents caused by single-line-to-
ground faults to 600 A.

D. Stochastic prediction of voltage dips

The procedure for voltage dip assessment is based on the Monte Carlo method and assumes
that dips are due only to faults caused within the distribution network. The test system is
simulated as many times as required to achieve the convergence of the Monte Carlo method.
Every time the system is run, fault characteristics are randomly generated using the following
distributions [34]:

79
  The fault location is selected by generating a uniformly distributed random number,
since it is assumed that the probability is the same for any point of the distribution
system.
 The fault resistance has a normal distribution.
 The initial time of the fault is uniformly distributed within a power frequency period.
 The duration of the fault has also a normal distribution.
 Different probabilities are assumed for each type of fault.

Given the topology, the location of protective devices and the parameters of the test system, it
is worth mentioning the following consequences:
 The retained voltage during a three-phase fault at the secondary of the substation can be
approximated by means of the following expression
RF
V( sag )   V( pre  sag ) (31)
Z S  Z Tr  RF
where Zs and Ztr are respectively the impedances of the high voltage (HV) equivalent and
the substation transformer, RF is the fault resistance, while V(pre-dip) and V(dip) are the
voltages prior and during the fault, respectively.
The impedance of the substation transformer seen from the medium voltage (MV) side is
2.5 . If the mean value of the fault resistance is greater than 5 , which is a realistic
value, the retained voltage will be above 90% of the pre-fault voltage. If the threshold
voltage for dip severity is 90% of the rated voltage, then not many equipment trips should
be caused by three-phase faults.
 It is well known that the severity of a voltage dip caused at the MV side of a Delta-wye
transformer is not the same at the low voltage (LV) side [1]. Figure 53 shows the voltage
dips caused by a single-phase-to-ground fault at the MV and LV sides of a distribution
transformer. Therefore, if customer equipment is installed only at the lower voltage side,
as assumed in this work, the percentage of trips due to single-phase-to-ground faults will
significantly decrease.
 The probability density charts of voltage dips corresponding to both terminals of the same
transformer will be very different. Consequently, the number of severe dips at the
customer installation (LV side) will be significantly reduced, since most equipment trips
will be caused by three-phase and two-phase faults located within a certain distance from
the substation.
 A consequence of protective device operation is that voltage dips will not be always
rectangular, as those shown in Figure 53. The coordination between protective devices
can produce multiple events with different retained voltages. An important aspect is then
the characterization of these events. Figure 54 shows different cases.
According to IEEE P1564 [5], multiple events should be merged in a single event since
the effect on end-user equipment will be the same. Consider the case depicted in Figure
54(a); the retained voltage for the three phases would be zero, while the dip duration
would be the time interval between the instant at which the voltage drops below the
threshold value and the instant at which this value is recovered. Although, the profiles are
different, a similar reasoning would be used with the case depicted in Figure 54(b). The
case shown in Figure 54(c) is different, since the retained voltage is the same for each

80
 7 5
2.8922

14
6

0.99
08
2 . 75
4

84
3

0.60
78
0. 84
F2 F1
1.52

0
1

2.4039

0.598
91

1.1116
F2 P
F1 F3 2
5 1.8625
B 1 .948
0.7265 1.3625 F1
F3
Feeder A 10
73 93
F1 1.
8 96
B 0.7907 1.8743 F2 0.6 3.22 12
F1
81
0.5350
Feeder B
F2
F3
65
1.5650 0. 56

1.4927
13
F2
F1
0.7797
2.7335

14
0.72
17 9
2.6043

11

HV equivalent: 110 kV, 1500 MVA, X/R = 10

Substation transformer: 110/25 kV, 20 MVA, 8%, Yd
Distribution transformers: 25/0.4 kV, 1 MVA, 6%, Dy
Lines: Z1/2 = 0.61 + j0.39, Z0 = 0.76 + j1.56 /km
a) Diagram of the test system. (Non integer quantities indicate the length of line sections, in km),
(B  Breakers, F1, F2, F3  Fuses)
10

Iscmax
Fuse F2
Fuse F1
1
Time [s]

Overcurrent
Relay
Breaker B

Fuse F3B
0.1

Fuse F3A

0.01
10 100 1000 10000
Current [A]

b) Time-current characteristics of protective devices.

Figure 52. Test system [89].

81
 Phase A Phase B Phase C
20

15
Voltage (kV)

a) Voltage dip – MV side (Dy

10
transformer)

0
40 80 120 160 200
Time (ms)

Phase a Phase b Phase c

0,25

0,20
Voltage (kV)

0,15
b) Voltage dip – LV side (Dy
transformer)
0,10

0,05

0,00
40 80 120 160 200
Time (ms)

Figure 53. Voltage dips at both sides of a distribution transformer.

phase during both events; only the dip duration would be affected now, being its value
that which results from the addition of the duration of each event.

E. Voltage dip assessment

The following studies, considering a different coordination between protective devices, were
performed [89]:
i. Protective devices do not operate; that is, the fault condition disappears before any device
could open.
ii. Circuit breakers operate faster than fuses and their relays have one reclose operation,
being the reclosing time 200 ms. The simulations are performed without including fuse
models.
iii. The coordination between overcurrent relays and fuses allows fuses to operate. Curve
labelled F3A in Figure 52(b) is selected for fuses F3. Relays will have one 200 ms reclose
operation.
iv. The same as for the previous study, but allowing feeder relays to have two 200 ms reclose
operations, and selecting fuse curve F3B.

82
 1.5

Rms Voltage (pu)

1.0 c c

b b
0.5

a a

0.0
0 200 400 600 800 1000
Time (ms)

a) Voltage dip caused by a two-phase-to-ground fault

1.5

a, b ,c
Rms Voltage (pu)

1.0 c
b

0.5
a

0.0
0 200 400 600 800 1000
Time (ms)

b) Voltage dip caused by a two-phase-to-ground fault

1.5

c c
Rms Voltage (pu)

1.0
b b

a a
0.5

0.0
0 200 400 600 800 1000
Time (ms)

c) Voltage dip caused by a two-phase fault

Figure 54. Different clusters of voltage dips, measured at the LV side (Adapted from [89]).

The fault characteristics were randomly generated according to the following distributions:
 The fault location was selected by generating a uniformly distributed random number.
 The fault resistance had a normal distribution, with a mean value of 5  and a
standard deviation of 1 , for each faulted phase.
 The initial time of the fault was uniformly distributed between 0.05 and 0.07 s.
 The mean value of the fault duration was varied, and by default the standard deviation
was 10% of the mean value.
 The probabilities of each type of fault were LG = 75%, 2LG = 17%, 3LG = 3%, 2L =
3%, 3L = 2%.

83
In all studies the test system is run 1000 times and the following information is recorded
during every run:
 fault characteristics (location, initiation time, duration, resistance, faulted phases, type
of fault)
 voltage dip characteristics (retained voltage, duration) on every phase of all load
buses.

If it is assumed a probability of occurrence of 12 faults per year and 100 km of overhead lines,
1000 runs are equivalent to analyze the performance of the test system during 214 years.
Figure 55 shows respectively the number of dips per year caused at a MV bus and the severi-
ty of these dips as compared to the ITIC curve. These results were deduced by assuming that
the average fault duration was 600 ms and the standard deviation was 60 ms. Only dips cau-
sed at one phase are shown; since the probability of occurrence of faults is assumed to be the
same at each phase in all lines, the charts will be very similar for each phase of the same bus.
0.3

0.2

0.1
ms

0
s
250m
0-50

16
450m

12 0%
s

80 0%
650m

- 17
s

40 %- - 13
850m
200-

0% %- 90 0% 0%
400-

- 10 50 %
%
600-

%
800-

a) Number of dips per year

200
180
160
140
Voltage (%)

120
100
80
60
40
20
0
0 .0 0 0 1 0 .0 1 1 100
T im e (s )

b) Voltage dip characteristics – ITIC curve

Figure 55. Simulation results – 1000 runs (Bus 6, MV side, phase A – Mean fault duration = 600 ms)
(Adapted from [89]).

84
Voltage tolerance (acceptability) curves can be used to predict equipment maloperation.
Although not much information is available in the literature about equipment performance in
front of three-phase unbalanced dips, acceptability curves for single-phase equipment are
widely used. Voltage dip indices can be easily calculated when only single-phase equipment
is installed. It is important to emphasize that the dips caused at the LV side of a transformer
can be very different from those caused at the MV side. Since it is assumed that only LV
single-phase equipment is installed, only dips caused at the LV side of load transformers will
be analyzed.

Figure 56 shows some simulation results that were derived from the four studies mentioned
above. All these results correspond to the voltage dips caused at the LV side of the load bus 6
and were deduced by assuming that the mean fault duration was 600 ms. The plots show the
number of trips per phase (i.e. the number of voltage dips below the lower ITIC curve) caused
to sensitive equipment, and were obtained after removing events with a retained voltage
between 90 and 110% of the rated voltage.

From these results one can deduce that the number of trips shows an evident dependence with
respect to the design of the protection system and that the quantity of dangerous dips is very
different at each side of a distribution transformer, see Figure 53. The performance can be
improved by including recloser and fuse operations, but the optimal design should also
consider the characteristic curves of protective devices.

Table XXI shows a summary of the main results obtained in this work. The results were
derived after 1000 runs, and using the same random generation of fault characteristics for all
cases. Equipment sensitivity was represented by the ITIC curve and only single-phase
equipment was installed at the LV level. The quantities within parenthesis indicate the
number of trips per year of sensitive equipment, and were calculated by assuming that the test
system is analyzed during 214 years, see above. Note that the quantities shown in the table do
not match those depicted in Figure 56; the quantities shown in figures correspond to one
phase only, while the quantities in the table are the average of the three phases.

One can see that the number of trips can decrease by increasing the fault-clearing time and by
introducing reclose operations. The improvement achieved by increasing the fault-clearing
time is evident after comparing results derived with a mean fault duration of 200 ms: the
lowest number of trips is obtained when protective devices do not operate (i.e. its operation is
delayed), but a significant improvement can be also achieved if they operate and the fault-
clearing time is increased for small fault currents, i.e. for single-phase-to-ground faults. The
second conclusion is deduced after comparing the results obtained with a mean fault duration
of 600 ms: the lowest number of trips is achieved when reclose operations are allowed. Both
results are simultaneously confirmed with the last study, when reclosing is allowed and a fuse
operation is delayed (see curve F3B in Figure 52(b)).

F. Calculation of voltage dip indices

Voltage dip indices can provide a count of event frequency and duration, the undelivered
energy during events, or the cost and severity of the disturbances. Only the first type of index,
i.e. count of events, is analyzed in this test case. The information obtained with the above
procedure is manipulated to obtain the number of trips per year in combination with an
acceptability curve [87].

85
 200
180
160
140
Voltage (%)

120
100
80 a) Protective devices do not operate
60
40
Num ber of
20 trip s : 1 7 4
0
0 .0 0 0 1 0 .0 1 1 100
T im e (s )

200
180
160
140
Voltage (%)

120
100
80 b) Breaker operation (Reclosing
60 interval = 200 ms)
40
N um ber of
20 trip s : 1 3 8 1 2 9 e v e n ts
0
0 .0 0 0 1 0 .0 1 1 100
T im e (s )

200
180
160
140
Voltage (%)

120
100
80 c) Fuse operation (Reclosing
60
O pen phase
d u e to fu s e interval = 200 ms)
o p e ra tio n
40
Num ber of
20 trip s : 1 2 2 6 5 e v e n ts
0
0 .0 0 0 1 0 .0 1 1 100
T im e (s )

200
180
160
140
Voltage (%)

120
100
d) Breaker and fuse operation (Two
80
reclosing intervals)
60
40
Num ber of
20 trip s : 1 0 3 5 0 e v e n ts
0
0 .0 0 0 1 0 .0 1 1 100
T im e (s )

Figure 56. Simulation results (Bus 6, LV side, phase A – Mean fault duration = 600 ms) (Adapted
from [89]).

86
 Table XXI – Average number of equipment trips per phase and year (Bus 6 - LV level) [89].

Protection system Fault duration

200 ms 600 ms 1s
No operation 64 169 176
(0.30) (0.79) (0.82)
Breaker operation 111 133 500
(tR = 200 ms) (0.52) (0.62) (2.34)
Breaker and fuse operation 121 122 159
(tR = 200 ms) (0.56) (0.57) (0.74)
Breaker and fuse operation
82 85 194
(change fuse F3)
(0.38) (0.40) (0.91)
(tR = 200 + 200 ms)

As detailed above, SARFI gives the average number of events (dips, swells, short
interruptions) over the assessment period, usually one year, per customer served. A SARFI
value is obtained by means of the following expression
ns
 Ni
SARFI  i 1
(32)
NT
where ns is the number of events, Ni is the number of customers experiencing an event and NT
is the number of customers served from the section to be assessed.

Since only events caused at the MV distribution level are analyzed, the number of costumers
that will experience an event at a LV load bus is the number of costumers served from that
bus. Therefore, for a single site the SARFI value is the number of dips over the assessment
period. The index for an entire system can be obtained as follows
nn
 N j  SARFI ( j )
j 1
SARFI  (33)
NT
where nn is the number of load buses, Nj is the number of costumers served from bus j, and
SARFI(j) is the value for bus j.

SARFI indices are based on the number of customers served from every load bus; it will be
assumed that this number is the same at every load bus; therefore, expression (33) becomes
nn
 SARFI ( j )
j 1
SARFI  (34)
nn
Figure 57 shows the index values deduced for the entire test system [87]. The main
conclusions derived from the simulation results can be summarized as follows:
 The performance obtained with each protection scenario is not the same for every
SARFIx index and some differences can be easily noted when comparing indices
derived for different sites. SARFI values corresponding to different thresholds can
show different behaviour; for instance, at load bus 6, the best SARFI90 performance is
achieved when all protective devices can operate, while the best SARFI60 performance
corresponds to a different scenario, i.e. when fuses are saved.
 SARFI values as a function of the mean fault duration do not show very significant

87
 4.5

3.5

3
SARFI (event/yr)

2.5

2
a) SARFI90
1.5 No Protective Devices

1 Only Breakers
Breakers & Fuses (1Rec)
0.5 Breakers & Fuses (2Rec)

0
200 300 400 500 600 700 800 900 1000
Mean fault duration (ms)

2.5
No Protective Devices
Only Breakers
2
Breakers & Fuses (1Rec)
Breakers & Fuses (2Rec)
SARFI (event/yr)

1.5

1 b) SARFI60

0.5

0
200 300 400 500 600 700 800 900 1000
Mean fault duration (ms)

3
No Protective Devices
2.5 Only Breakers
Breakers & Fuses (1Rec)
SARFI (event/yr)

Breakers & Fuses (2Rec)

2

1.5
c) SARFI-CBEMA
1

0.5

0
200 300 400 500 600 700 800 900 1000
Mean fault duration (ms)

2.5
No Protective Devices
Only Breakers
2
Breakers & Fuses (1Rec)
Breakers & Fuses (2Rec)
SARFI (event/yr)

1.5

1 d) SARFI-ITIC

0.5

0
200 300 400 500 600 700 800 900 1000
Mean fault duration (ms)

Figure 57. SARFI calculations for the whole system – LV buses (Adapted from [87]).

88
 changes for a given protection system, except when the fault duration is about 1
second or longer.
 ITIC equipment has a better performance that CBEMA equipment; however, when
fuses operate the performance is very similar. Results deduced when loads are repre-
sented by means of ITIC and SEMI curves were the same.

G. Conclusions

This study has presented the application of a procedure for stochastic prediction of voltage
dips to the calculation of voltage dip indices in distribution networks. The results deduced
from the procedure are used to obtain the average number of events experienced by each
customer served over the assessment period. The results presented in these test cases have
shown the influence that the design of the protection system and the equipment vulnerability
can have on these indices.

8.5.2 Example 2: Impact of distributed generation

A. Introduction

The replacement of conventional generation by distributed energy resources (DER), like wind
turbines, leads to a reduction of the fault level, and this reduction will in general lead to an
increased number of dips at the interface between transmission and distribution. For instance,
if wind turbines are sensitive to voltage dips, one single fault may lead to loss of several
thousands of MW of generation. This in turn may trigger stability problems in the grid. A
study presented in [77] compares the dip frequency in a transmission system for a high
generation scheduling and a low generation scheduling. The low generation scheduling had
only half the generation capacity connected to each bus. The result was a significant increase
in the dip frequency, especially for locations close to generator stations.

The impact of DER penetration in both transmission and distribution levels when voltage dips
are caused by faults located at the transmission system is discussed below.

B. Impact at transmission levels

Part of the UK transmission system was modelled to study the impact of a reduction in large
generation connected to the transmission system. The system data was obtained from the 2004
Seven Year Statement [90]. The case study has been based on the 400 kV grid in the North of
England and Wales and the 275 kV grids around Sheffield, around Manchester and around
Newcastle. The remaining part of the 400 kV system was modelled as a number of Thevenin
equivalents. The load was taken equal to the one during “NGC peak” according to the 2004
Seven Year Statement.

A DER increase, equally spread over the system, was assumed, being the reduction in reactive
power was the same as the reduction in active power. This was modelled as a reduction in the
load as seen from the transmission system, which results in a reduction of the amount of
generator units in operation. It was assumed that the oldest units are the most expensive ones
and the first to be taken out of operation. No voltage control problems were observed when
taking generator units out of operation. In fact, the voltage profile of the transmission system
improved due to the reduction in load. The increase in distributed generation was studied in a
five-stage scenario as summarized in Table XXII.

89
 Table XXII – Five-stage scenario of increase in distributed generation.
Total load Load Generation reduction compared to
Stage
reduction remaining previous stage
0 None 100% n.a.
1 490 MW 97.24% 1 unit commissioned in 1966
2 2959 MW 83.33% 5 units commissioned in 1967
3 5399 MW 69.59% 6 units commissioned in 1968
4 7888 MW 55.57% 5 units commissioned in 1969
5 10356 MW 41.67% 5 units commissioned in 1970

The impact of faults at 400 kV level was studied for each of the stages. Only three-phase
faults were assumed. For a given short-circuit position, the during-fault voltage was
calculated for each 275-kV and 400-kV bus in the system. The impact of the fault was
quantified as the sum of all active-power loads that experienced a dip below a certain
threshold.

An example is shown in Table XXIII. The stage-0 loads have been used to calculate the
impact of the fault. This fault is located in the south-eastern part of the network. As expected,
the impact of the fault increases with an increase in DER (i.e. with a decrease in large central
generation). From a customer viewpoint, the threshold values should be translated into
immunity (voltage tolerance, sensitivity) of end-user equipment against voltage dips. For high
DER penetration, end-user equipment also includes generation. When the total amount of
generation tripping due to a fault is large, the system stability may be endangered. Such
dynamic effects have not been considered in this study.

These calculations have been repeated for 17 different fault locations equally distributed over
the 400-kV grid studied. It has been assumed that all faults have an equal probability of
occurrence. The sum (or average) of the impacts of each individual fault is in that case a
measure for the number of dips experienced by the load. This "total impact" is obtained as the
sum of the values as in Table XXIII for each of the 17 fault positions. The results are shown
in Tables XXIV and XXV; the former one gives the absolute value (in MW) whereas the
latter one gives the relative increase compared to the base case [91].

Table XXIII – Example of the impact of a fault.

Stage 70% 75% 80%
0 917 MW 1166 MW 3436 MW
1 917 MW 1166 MW 3436 MW
2 917 MW 2071 MW 3732 MW
3 920 MW 2637 MW 3871 MW
4 1166 MW 3436 MW 5920 MW
5 1166 MW 4128 MW 6344 MW

Table XXIV – Average impacts of all fault positions.

Stage 70% 75% 80%
0 2371 MW 3164 MW 4197 MW
1 2458 MW 3196 MW 4254 MW
2 2728 MW 3416 MW 4581 MW
3 2966 MW 3754 MW 4918 MW
4 3104 MW 4054 MW 5162 MW
5 3283 MW 4389 MW 5574 MW

90
 Table XXV – Relative voltage-dip frequency.
Stage % DER 70% 75% 80%
0 0% 100% 100% 100%
1 2.7% 103% 101% 101%
2 16.7% 115% 108% 109%
3 30.4% 125% 119% 117%
4 44.4% 131% 128% 123%
5 58.3% 138% 139% 133%

The relative increase in dip frequency is shown graphically in Figure 58, where the three
curves correspond to three values of end-user equipment's immunity against voltage dips. In
more standardized terms, the three curves give the relative increase in SARFI-70, SARFI-75
and SARFI-80.

Figure 58. Relative increase in dip frequency with increasing percentage of distributed energy
resources, for three values of the end-user immunity: 70%; 75% and 80% residual voltage.

The values in Tables XXIV and XXV can be used as voltage-dip indices to quantify the
performance of the system. These indices can be used to apply the hosting-capacity approach
when a limit value is defined for what is an acceptable behaviour [92], [93]. There is however
no limit set in any standard on what is an acceptable voltage-dip frequency. An alternative
approach is to use the increase in voltage-dip frequency as an index. This has been done in
Table XXV. An acceptable limit is set as a maximum-permissible increase in average voltage-
dip frequency. If an increase of 25% is allowed, the hosting capacity is around 30%. If an
increase of 50% is allowed, the hosting capacity is at least 60%. Again, the hosting capacity
depends strongly on what are perceived to be acceptable limits. This is a subject that is part of
all system design and operational issues, not just related to distributed energy resources.

Strictly speaking the index in Table XXIV is not a voltage-dip frequency. A voltage-dip
frequency is expressed in events per year, whereas the quantities shown in the table are
expressed in MW. A weighted voltage-dip frequency (SARFI), as defined in [5], is obtained
by multiplying the value in Table XXIV by the total number of faults per year and dividing it
by the total load connected to the system. As the multiplication factors are independent of the
amount of distributed energy resources, the relative increase in Table XXV and Figure 58 also
holds for the weighted voltage-dip frequency.

91
In the system under study, 12 large industrial customers were directly supplied from 14
transmission buses (at 275 and at 400 kV). The increase in dip frequency experienced by
these large industrial customers has been assessed separately. The resulting voltage-dip
indices are given in Table XXVI, which provides the average number of dips per fault
location below the threshold, experienced by the large industrial customers. To obtain a
weighted dip frequency (SARFI) the values should be multiplied by the number of faults per
year (for the total 400-kV grid) and divided by the number of industrial customers (12 in this
case). The relative voltage-dip frequency (compared to the stage-0 value) is plotted in Figure
59. As the multiplication factors are independent of the percentage DER, the same figure
holds for the weighted voltage-dip frequency.

Table XXVI – Average impacts of all faults on large industrial customers.

Stage 70% 75% 80%
0 0.9 1.2 1.7
1 0.9 1.2 1.7
2 0.9 1.2 1.8
3 1.0 1.3 2.1
4 1.0 1.5 2.3
5 1.1 1.6 2.6

Note the difference in weighting factors. In the first case, weighting was based on the total
active power load supplied from the bus during the system peak. In the second case the
weighting was based on the number of large industrial customers supplied from the bus. The
resulting weighted dip frequencies (SARFI values) are obviously different, but their general
trend with increasing penetration of DER is very similar.

Figure 59. Relative voltage-dip frequency for large industrial customers with increasing penetration
of distributed energy resources.

From the comparison of Figures 58 and 59, one can deduce that the increase in SARFI-80 is
larger for the industrial customers than for the average load. The curves in Figure 59 are less
smooth due to the limited number of sites involved (12 versus 26). It can be stated, however,
that the increase in voltage-dip frequency is very similar for the average large industrial
customer as for the average load.

92
Other aspects may also impact the voltage-dip frequency. In parts of the transmission system
with a high concentration of large power stations, fault-current reduction techniques are often
used (e.g. busbar splitting or series reactors); shutdown of those stations will make that these
techniques are no longer needed so that the reduction in fault level will be less dramatic. The
resulting increase in connectivity may, however, lead to an increase in the dip frequency,
despite the fault level remaining the same.

C. Impact at distribution levels

The presence of generator units at distribution level will affect the dip frequency experienced
by the end customers. The effect will depend to a large extend on the kind of interface with
the grid. Units with synchronous machines have the ability to mitigate voltage dips due to
three-phase faults. Induction generators only contribute to a symmetrical fault during the first
one or two cycles. Both synchronous and induction generators have a small negative-sequence
impedance (0.15 - 0.2 per-unit) so that they lead to a reduction of the negative-sequence
voltage during a non-symmetrical fault. Voltage dips become thus more balanced. During a
non-symmetrical fault the decay in the positive-sequence current contribution of an induction
generator is much slower than during a symmetrical fault, so that it can consider this as a
sustained contribution for all but the longest dips. The induction generators behave in this
sense similar to the rest of the load, as discussed in [93].

Power-electronic interfaces can be designed to somewhat mitigate unbalanced voltage dips as

well, but for mitigating balanced dips overrating of the converters is needed. However most
power-electronic interfaces will trip rather quickly during a voltage dip as the currents and/or
voltages will exceed their design ratings. For most voltage-dip studies it may thus be assumed
that power-electronic converters do not contribute anything to the current during the fault.

Consider the simplified system model in Figure 60. During a fault at transmission level, the
synchronous machine can be modelled as a voltage source behind impedance. If the load
influence on the voltage and the influence of the synchronous machine on the voltage at
transmission level are neglected, the model to calculate the voltage as experienced by the end-
customers becomes that on the right.
UHV

Z tr
HV
UMV
SM

MV Z gen

Egen
load
Figure 60. Distribution network with synchronous generator at MV level.

The residual voltage is found from the voltage-divider model. If the back-emf of the
synchronous machine Egen is equal to 1 pu, the following expression for the drop in voltage
during the fault is deduced [91]:

93
 Z gen
1  U MV   1  U HV  (35)
Z gen  Z tr
The voltage experienced by end-customers in case there is no generator connected at MV, is
equal to UHV, so that the reduction in voltage drop is related to the impedance ratio between
the transformer and the synchronous generator.

To estimate the impact on the number of dips experienced by the end-customers, a simplified
relationship between the number of dips and the residual voltage can be used:
V
N dips  N 50 (36)
1V
where N50 is the number of voltage dips below residual voltage; it depends on the system
configuration and the fault frequency.

The above expression was derived from theoretical models and has been confirmed from both
simulations and measurements [1].

Combining (35) and (36) gives the following expression for the number of dips at the MV bus
(thus as experienced by the end-customer):
V
N MV (V )  N 50 (37)
1V 
with
z
V   V  tr 1  V  DER (38)
z gen

S gen
 DER  (39)
Str
where DER is the fraction of distributed energy resources, ztr is the transformer impedance in
per-unit on a base equal to the transformer rating Str and zgen is the per-unit generator
impedance on a base equal to the generator rating Sgen.

An example is given in Figure 61, where a transformer impedance of 0.20 pu and a generator
impedance of 0.25 pu have been used [91]. The increase in DER penetration can result in a
significant reduction in the dip frequency. If the local penetration is higher than the average
for the whole system, the local dip frequency reduces significantly.

The dip frequency for end-customers is however not only determined by dips originating at
distribution level, but also by dips originating at transmission level. For some locations the
later may dominate and the overall power quality will only show a limited influence of the
level of DER penetration. An important conclusion from this discussion is that there is no
general rule on how the dip frequency is impacted. It depends strongly on the transmission
system, on the type of distributed generation and on the immunity of the end-customers
against voltage dips. For customers at a location far away from large power stations, e.g.
many customers in large urban areas, the dip frequency may actually reduce with an
increasing level of local small generation.

94
 100

Relative Dip Frequency [%]

90
80
70
60

50
40
30
20

10
0 50 100 150 200
%DER
Figure 61. Relative voltage-dip frequency as a function of the DER penetration: SARFI-80 (solid
line) and SARFI-70 (dashed).

9. Conclusions
Present software packages for power system analysis are powerful tools that can be used to
perform any type of voltage dip study. Users can take advantage of capabilities implemented
in general purpose simulation tools, e.g. EMTP-like tools, apply capabilities of equation-
oriented tools, e.g. MATLAB, or just use a high level language compiler, e.g. C++, to develop
custom-made models of power equipment, test the performance of mitigation devices, analyze
the effect of new distributed generation technologies, embed new signal processing
algorithms for voltage dip characterisation into simulation tools or calculate voltage dip
indices.

Computer simulations provide an efficient mean to analyze the effect of disturbances and the
behaviour of sensitive equipment, but results are as accurate and reliable as models and data
used in calculations. Since the knowledge on the behaviour of sensitive equipment during
unbalanced dips is rather limited, computer models cannot represent all factors that have some
influence on the equipment response.

In addition, standards are only concerned about rectangular voltage dips, which are
characterised by the magnitude and the duration. Characterisation of non-rectangular dips,
assessment of their effects on equipment operation or the possible influence of non-ideal
voltage supply characteristics are aspects not covered by present standards. Recommendations
for testing of sensitive electrical equipment should be, therefore, extended and reformulated in
order to include all characteristics, parameters and conditions of influence.

Disturbances that can lead to voltage dips, i.e. faults, are random in nature. Fault
characteristics are based on reliability data. Although utilities have records of failures and
their causes, many uncertainties still exist in this field. For example, the probability of
occurrence associated to each type of fault can be specified with certain accuracy, but
accurate statistics about fault durations are very rare or do not exist at all.

95
The selection of the simulation tool and the solution technique can be made once the goals of
the voltage dip study have been fixed. A study aimed at estimating voltage dip indices based
on the calculation of during-fault voltages only could be performed by a frequency-domain
tool, since the calculations will be faster than those derived from the application of a time-
domain tool. However, the use of phasors restricts models to the context of linear systems
running under AC steady state and to the calculation of single-stage events. More realistic and
accurate results can be derived only from the application of a time-domain simulation tool.

Two methods have been developed and applied for stochastic prediction of voltage dips; the
method of fault positions and the method of critical distances. The first one is generally
accepted as the most suitable because the method of critical distances has been usually
applied by considering analytical expressions that are correct for radial systems only.

Strictly speaking the method of fault positions can be seen as a Monte Carlo method.
Although fault positions are in many studies distributed uniformly along lines, the distribution
can follow different probabilities density functions. Obviously, different fault distributions
will result in different prediction of voltage dips at different buses. The differences can justify
additional effort in modelling the actual fault distributions.

The main drawback of a Monte Carlo simulation is the number of iterations (i.e. the sample
size or population) that are needed to obtain accurate enough results. The accuracy increases
with the increase of the sample size. However, this can also be achieved by decreasing the
variance of the sample estimates, without increasing the sample size. Several procedures,
known as variance reduction techniques, have been proposed for that purpose: antithetic
variables, importance sampling, stratification, control variates [94]. For example, when using
stratification or stratified sampling the sample size is first divided into mutually exclusive
sub-populations or strata; then units are selected for each stratum. Several strategies may be
used: proportionate allocation (the number of samples per stratum is proportional to the full
sample size ratio; if 80% of all faults are single-line-to-ground, 80% of the samples are taken
for the corresponding stratum), optimum allocation (each stratum is proportional to the
standard deviation of the sub-population), or equal allocation (equal number of samples per
stratum and results are weighted according to the full population ratio).

A computer-based estimation of a power quality index may be the ultimate goal of a voltage
dip study. As for other simulation results, the estimated index will be as accurate and reliable
as the models and data used in calculations. Not much experience is still available on the
application of voltage dip indices, and it is difficult to predict the usefulness of the indices
discussed in this report. Since these indices are based on values related to voltage dip
characteristics and to the comparison of those values to voltage-tolerance curves, see SARFI
definitions in Section 8.3.1, their limitations are evident. For example, readers can review the
discussions about voltage dip characterisation included in Sections 2.2, 2.3 and 8.2, the
discussion about aggregation included in subsection 8.2.3, or the discussions about testing and
behaviour of sensitive equipment included in Section 3.6. To overcome other limitations, e.g.
those related to costs, other indices have been proposed.

96
10. References
[1] M.H.J. Bollen, Understanding Power Quality Problems. Voltage Dips and Interruptions, IEEE Press,
2000, New York.
[2] M.H.J. Bollen, “What is power quality,” Electric Power Systems Research, vol.66, Issue 1, pp. 6-14, July
2003.
[3] IEC 61000-2-8, “Electromagnetic Compatibility (EMC) – Part 2-8: Environment - Voltage Dips and
Short Interruptions on Public Electric Power Supply Systems with Statistical Measurement Results”,
2002.
[4] IEC 61000-4-30, “Electromagnetic Compatibility (EMC) – Part 4-30: Testing and Measurement
Techniques - Power Quality Measurement Methods” 2003.
[5] IEEE P1564, “Recommended Practice for the Establishment of Voltage dip Indices”, Draft, June 2004.
[6] DISDIP Group, “Voltage Dips and Short Interruptions in Medium Voltage Public Electricity Supply
Systems,” Report from the International Union of Producers and Distributors of Electrical Energy
(UNIPEDE), 1990.
[7] ANSI/IEEE Std. 493-1980, IEEE Recommended Practice For Design of Reliable Industrial and
Commercial Power Systems, 6th Printing, 1989.
[8] IEEE Std. 1346-1998, IEEE Recommended Practice for Evaluating Electric Power System Compatibility
with Electronic Process Equipment.
[9] D.L. Brooks, R.C. Dugan, M. Waclawiak and A. Sundaram, “Indices for assessing utility distribution
system RMS variation performance”, IEEE Trans. on Power Delivery, vol. 13, no. 1, pp. 254-259,
January 1998.
[10] IEEE Std. 1159-2001, IEEE Recommended Practice for Monitoring Electric Power Quality.
[11] M.H.J. Bollen, “Characterisation of voltage sags experienced by three-phase adjustable-speed drives”,
IEEE Trans. on Power Delivery, vol. 12, no. 4, pp. 1666-1671, October 1997.
[12] M.H.J. Bollen and L.D. Zhang, “Different methods for classification of three-phase unbalanced voltage
dips due to faults”, Electric Power Systems Research, vol.66, Issue 1, pp. 59-69, July 2003.
[13] M.H.J. Bollen, “Algorithms for characterizing measured three-phase unbalanced voltage dips”, IEEE
Trans. on Power Delivery, vol. 18, no.3, pp. 937-944, July 2003.
[14] M.H.J. Bollen, Ph. Goossens and A. Robert, “Assessment of voltage dips in HV-networks: deduction of
complex voltages from the measured rms voltages”, IEEE Trans. on Power Delivery, vol. 19, no. 2, pp.
783-790, April 2004.
[15] T.S. Key, “Diagnosing power-quality related computer problems”, IEEE Trans. on Industry Applications,
vol.15, pp. 381-393, 1979.
[16] ITIC (Information Technology Industry Council), “ITIC Curve Application Note”, available at
http://www.itic.org/archives/iticurv.pdf.
[17] IEC 61000-4-11, “Electromagnetic Compatibility (EMC) – Part 4-11: Voltage Dips, Short Interruptions
and Voltage Variations Immunity Tests”, 1994.
[18] J.V. Milanović and S. Ž. Djokić, “Equipment sensitivity to disturbances in voltage supply”, Proc. Int.
Electrical Equipment Conf. JIEEC 2003: The Electric Network of the Future and Distributed Generation,
pp. 3.7/1-3.7/14, Bilbao, Spain, 28-29 October 2003.
[19] S. Ž. Djokić, J. V. Milanović and D.S. Kirschen, “Sensitivity of ac coil contactors to voltage sags, short
interruptions and undervoltage transients”, IEEE Trans. on Power Delivery, vol. 19, no. 3, pp. 1299-1307,
July 2004.
[20] S. Ž. Djokić, K. Stockman, J.V. Milanović, J.J.M. Desmet and R. Belmans, “Sensitivity of ac adjustable
speed drives to voltage sags and short interruptions”, IEEE Trans. on Power Delivery, vol. 20, no. 1, pp.
494-505, January 2005.
[21] S. Ž. Djokić, J.J.M. Desmet, G. Vanalme, J.V. Milanović and K. Stockman, “Sensitivity of personal
computers to voltage sags and short interruptions”, IEEE Trans. on Power Delivery, vol. 20, no. 1, pp.
375-383, January 2005.
[22] A. Sannino, J. Svensson and T. Larsson, “Power-electronic solutions to power-quality problems”, Electric
Power Systems Research, vol.66, Issue 1, pp. 71-82, July 2003.
[23] N. Woodley, L. Morgan, and A. Sundaram, “Experience with an inverter based dynamic voltage
restorer”, IEEE Trans. on Power Delivery, vol. 14, no. 3, pp. 1181-1186, July 1999.
[24] J.W. Schwartzenberg and R.W. De Doncker, “15 kV medium voltage static transfer switch”, IEEE IAS
Annual Meeting 1995, vol. 3, pp.2515–2520.
[25] IEEE P1409 Distribution Custom Power Task Force, “Trial Use Guide for Application of Power
Electronics for Power Quality Improvement on Distribution Systems Rated 1 kV Through 38 kV”, Draft
4, July 14, 2000.

97
[26] H.W. Dommel, “Digital computer solution of electromagnetic transients in single- and multi-phase
networks”, IEEE Trans. on Power Apparatus and Systems, vol. 88, no. 2, pp. 734-741, April 1969.
[27] H.W. Dommel, ElectroMagnetic Transients Program. Reference Manual (EMTP Theory Book),
Bonneville Power Administration, Portland, 1986.
[28] CIGRE Working Group 02 (SC 33), “Guidelines for Representation of Network Elements when
Calculating Transients”, 1990.
[29] A.M. Gole, J.A. Martinez-Velasco, and A.J.F. Keri (Ed.), “Modeling and Analysis of System Transients
Using Digital Programs”, IEEE PES Special Publication, TP-133-0, 1999.
[30] J.A. Martinez and J. Martin-Arnedo, “Voltage sag analysis using an electromagnetic transients program”,
IEEE PES Winter Meeting 2002, January 27-31, New York.
[31] J.A. Martinez and B.A. Mork, “Transformer modeling for low- and mid-frequency transients – A
review”, IEEE Trans. on Power Delivery, vol. 20, no. 2, pp. 1625-1632, April 2005.
[32] J.A. Martinez, R. Walling, B. Mork, J. Martin-Arnedo and D. Durbak, “Parameter determination for
modeling systems transients. Part III: Transformers”, IEEE Trans. on Power Delivery, vol. 20, no. 3, pp.
2051-2062, July 2005.
[33] J.A. Martinez and J. Martin-Arnedo, “Voltage sag stochastic prediction using an electromagnetic
transients program”, IEEE Trans. on Power Delivery, vol. 19, no. 4, pp. 1975-1982, October 2004.
[34] J.A. Martinez and J. Martin-Arnedo, “Voltage sag studies in distribution networks. Part I: system
modeling”, IEEE Trans. on Power Delivery, vol. 21, no. 3, pp. 1670-1678, July 2006.
[35] J.J. Burke, Power Distribution Engineering. Fundamentals and Applications, Marcel Dekker, 1994.
[36] K.L. Leix, Lj.A. Kojovic, M. Marz and G.C. Lampley, “Applying current-limiting fuses to improve
power quality and safety”, IEEE T&D Conf., pp. 636-641, April 1999.
[37] Lj.A. Kojovic, S.P. Hassler, K.L. Leix, C.W. Williams and E:E. Baker, “Comparative analysis of
expulsion and current-limiting fuse operation in distribution systems for improved power quality and
protection”, IEEE Trans. on Power Delivery, vol. 13, no. 3, pp. 863-869, July 1998.
[38] Lj.A. Kojovic and S.P. Hassler, “Application of current-limiting fuses in distribution systems for
improved power quality and protection”, IEEE Trans. on Power Delivery, vol. 12, no. 2, pp. 791-800,
April 1997.
[39] A. Petit, G. St-Jean and G. Fecteau, “Empirical model of a current-limiting fuse using EMTP”, IEEE
Trans. on Power Delivery, vol. 4, no. 1, pp. 335-341, January 1989.
[40] “IEEE Standard Inverse-Time Characteristic Equations for Overcurrent Relays”, IEEE Std C37.112-1996.
[41] IEEE Computer Representation of Overcurrent Relays Characteristics WG, “Computer representation of
overcurrent relay characteristics”, IEEE Trans. on Power Delivery, vol. 4, no. 3, pp. 1659-1667, July
1989.
[42] N. Mohan, T.M. Undeland and W.P. Robbins, Power Electronics: Converters, Applications, and Design,
Second Edition, John Wiley & Sons, 1995, New York.
[43] N.G. Hingorani and L. Gyugyi, Understanding FACTS: Concepts and Technology of Flexible AC
Transmission Systems, IEEE Press, 2000, New York.
[44] IEEE TF on Power Electronics (L. Tang, Chairman), “Guidelines for modeling power electronics in
electric power engineering applications”, IEEE Trans. on Power Delivery, vol. 12, no.1, pp. 505-514,
January 1997.
[45] B. Johnson, H. Hess and J.A. Martínez, “Parameter determination for modeling systems transients. Part
VII: Semiconductors”, IEEE Trans. on Power Delivery, vol. 20, no. 3, pp. 2086-2094, July 2005.
[46] IEEE Task Force on Load Representation for Dynamic Performance, “Load representation for dynamic
performance analysis”, IEEE Trans. on Power Systems, vol. 8, no. 2, pp. 472-482, May 1993.
[47] IEEE Task Force on Load Representation for Dynamic Performance, “Bibliography on load models for
power flow and dynamic performance simulation”, IEEE Trans. on Power Systems, vol. 10, no. 1, pp.
523-538, February 1995.
[48] IEEE Task Force on Load Representation for Dynamic Performance, “Standard load models for power
flow and dynamic performance simulation”, IEEE Trans. on Power Systems, vol. 10, no. 3, pp. 1302-
1313, August 1995.
[49] Ch. J. Lin et al., “Dynamic load models in power systems using the measurement approach”, IEEE Trans.
on Power Systems, vol. 8, no. 1, pp. 309-315, February 1993.
[50] D. Karlsson and D.J. Hill, “Modelling and identification of nonlinear dynamic loads in power systems”,
IEEE Trans. on Power Systems, vol. 9, no. 1, pp. 157-166, February 1994.
[51] K. Morison, H. Hamadani and L. Wang, “Practical issues in load modelling for voltage stability studies”,
IEEE PES General Meeting, July 2003, Toronto.
[52] S. Ihara, K. Tomiyama and M. Tani, “Residential load characteristics observed at KEPCO power
system”, IEEE Trans. on Power Systems, vol. 9, no. 2, pp. 1092-1101, May 1994.

98
[53] K. Tomiyama, J.P. Daniel and S. Ihara, “Modeling air conditioner load for power system studies”, IEEE
Trans. on Power Systems, vol. 13, no. 2, pp. 414-421, May 1998.
[54] K. Tomiyama, S. Ueoka and T. Takano, “Modeling of load during and after system faults based on actual
field data”, IEEE PES General Meeting, July 2003, Toronto.
[55] P.A. Gnadt and J.S. Lawler (Eds.), Automatic Electric Utility Distribu-tion Systems, Prentice Hall, 1990.
[56] J. Key et al., “The design of power acceptability curves”, IEEE Trans. on Power Delivery, vol. 17, no. 3,
pp. 828-833, July 2002.
[57] J.A. Martinez, J. Martin-Arnedo and J.V. Milanović, “Load modeling for voltage dip studies”, IEEE PES
General Meeting, July 2003, Toronto.
[58] B. Khodabackhchian and G.T. Vuong, “Modeling a mixed residential-commercial load for simulation
involving large disturbances”, IEEE Trans. on Power Systems, vol. 12, no. 2, pp. 791-796, May 1997.
[59] M. Reformat, D. Woodford, R. Wachal and N.J. Tarko, “Nonlinear load modelling for simulation in time
domain”, 8th ICHQP, vol. 1, October 1998.
[60] J.V. Milanović, R. Gnativ and K.W.M. Chow, “The influence of loading conditions and network topology
on voltage sags”, 9th ICHQP, October 2000.
[61] J. Arrillaga, C.P. Arnold and B.J. Harker, Computer Modelling of Electrical Power Systems, John Wiley
& Sons, 1983.
[62] P.M. Anderson and A.A. Fouad, Power System Control and Stability, IEEE Press, 1994.
[63] IEEE Std. 1110, “IEEE Guide for Synchronous Generator Modeling Practices in Stability Studies,” 1991.
[64] IEEE Committee Report, “Excitation systems dynamic characteristics,” IEEE Trans. on PAS, vol. 92, pp.
64-75, January-February 1973.
[65] “CIGRE TF 38.01.10 (Convener N. Hatziargyriou), “Modeling New Forms of Generation and Storage”,
Technical Brochure 2002.
[66] L. Conrad, K. Little and C. Grigg, “Predicting and preventing problems associated with remote fault-
clearing voltage dips”, IEEE Trans. on Industry Applications, vol. 27, no.1, pp.167-172, January 1991.
[67] M.H.J. Bollen, “Fast assessment methods for voltage sags in distribution systems”, 30th IEEE IAS Annual
Meeting, vol. 3, pp. 2282 -2289, 1995.
[68] M.H.J. Bollen, “Fast assessment methods for voltage sags in distribution systems”, IEEE Trans. on
Industry Applications, vol. 32, no. 6, pp. 1414-1423, November/December 1996.
[69] L.E. Conrad and M.H.J. Bollen, “Voltage dip co-ordination for reliable plant operation”, IEEE Trans. on
Industry Applications, vol.33, no.6, November 1997, pp.1459-1464.
[70] M.H.J. Bollen, “Method of critical distances for stochastic assessment of voltage sags”. IEE Proc.
Generation, Transmission and Distribution, vol. 145, no. 1, pp. 70-76, January 1998.
[71] M.H. Qader, M.H.J. Bollen, and R.N. Allan, “Stochastic prediction of voltage sags in a large transmission
system,” IEEE Trans. on Industry Applications, vol. 35, no. 1, pp. 152-162, January/February 1999.
[72] S. Omar Faried and S. Aboreshaid, “Stochastic evaluation of voltage sags in series capacitor compensated
radial distribution systems,” IEEE Trans. on Power Delivery, vol. 18, no. 3, pp. 744-750, July 2003.
[73] P. Heine and M. Lehtonen, “Voltage dip distributions caused by power system faults”, IEEE Trans. on
Power Systems, vol. 18, no. 4, pp. 1367-1373, November 2003.
[74] J. Wang, S. Chen, and T.T. Lie, “System voltage sag performance estimation”, IEEE Trans. on Power
Delivery, vol. 20, no. 2, pp. 1738-1747, April 2005.
[75] S. Omar Faried, R. Billington and S. Aboreshaid, “Stochastic evaluation of voltage sag and unbalance in
transmission systems”, IEEE Trans. on Power Delivery, vol. 20, no. 4, pp. 2631-2637, October 2005.
[76] E. Styvaktakis, “Automating Power Quality Analysis”, PhD thesis, Chalmers University of Technology,
Gothenburg (Sweden), 2002.
[77] G. Olguin, “Voltage dip (sag) estimation in power systems based on stochastic assessment and optimal
monitoring”, PhD Thesis, Chalmers University of Technology, Gothenburg, Sweden, 2005.
[78] G. Olguin and M.H.J. Bollen, “Method of fault positions for stochastic prediction of voltage sags: A case
study”, PMAPS Conference 2002, September 22-26, Naples, Italy, pp. 557-562.
[79] P. Heine, P. Pohjanheimo, E. Lakervi and P. Rissanen, “Methods and tools for voltage sag evaluation
implemented in a network information system”, 35th UPEC, Belfast, September 6-8, 2000.
[80] M.T. Aung, J.V.Milanović and C.P. Gupta, “Propagation of asymmetrical sags and the influence of
boundary crossing lines on voltage dip prediction”, IEEE Trans. on Power Delivery, vol.19, no. 4, pp.
1819-1827, October 2004.
[81] IEEE Std. 493-1997, IEEE Recommended Practice for Design of Reliable Industrial and Commercial
Power Systems.
[82] J.V. Milanović, M.T. Aung and C.P. Gupta, “The influence of fault distribution on stochastic prediction
of voltage sags”, IEEE Trans. on Power Delivery, vol.20, no. 1, pp. 278-285, January 2005.
[83] IEEE Std. 1366-2001, Trial Use Guide for Electric Power Distribution Reliability Indices.

99
[84] Joint WG CIGRE C4.07/CIRED, “Power Quality Indices and Objectives”, Technical Brochure 261,
October 2004.
[85] Brooks et al, and Joint WG CC02 CIGRE-CIRED, Recommendations for Tabulating RMS Voltage
Variations Disturbances with Specific Reference to Utility Power Contracts.
[86] G-J. Lee, M.M. Albu and G.T. Heydt, “A power quality index based on equipment sensitivity, cost, and
network vulnerability”, IEEE Trans. on Power Delivery, vol. 19, no. 3, pp. 1504-1510, July 2004.
[87] J.A. Martinez and J. Martin-Arnedo, “Voltage sag studies in distribution networks. Part III: Voltage sag
index calculation”, IEEE Trans. on Power Delivery, vol. 21, no. 3, pp. 1689-1697, July 2006.
[88] J.A. Martinez and J. Martin-Arnedo, “Expanding capabilities of EMTP-like tools: From analysis to
design”, IEEE Trans. on Power Delivery, vol. 18, no. 4, pp. 1569-1571, October 2003.
[89] J.A. Martinez and J. Martin-Arnedo, “Voltage sag studies in distribution networks. Part II: Voltage sag
assessment”, IEEE Trans. on Power Delivery, vol. 21, no. 3, pp. 1679-1688, July 2006.
[90] National Grid Transco, 2004, Seven Year Statement, available at http://www.nationalgrid.com.
[91] M.H.J. Bollen and M. Häger, “Impact of increasing penetration of distributed generation on the number
of voltage dips experienced by end-customers”, 18th CIRED, 6-9 June, Turin, 2005.
[92] C. Schwaegerl, M.H.J. Bollen, K. Karoui and A. Yagmur, “Voltage control in distribution systems as a
limitation of the hosting capacity for distributed energy resources”, 18th CIRED, 6-9 June, Turin, 2005.
[93] M.H.J. Bollen and M. Häger, “Power quality: interactions between distributed energy resources, the grid
and other customers”, 1st Int. Conf on Renewable Energy Sources and Distributed Energy Resources,
2004.
[94] G.J. Anders, Probability Concepts in Electric Power Systems, John Wiley, 1990.

100