Professional Documents
Culture Documents
Study of Insulation Co-Ordination in Electric Power Distribution Systems
Study of Insulation Co-Ordination in Electric Power Distribution Systems
Joshua T. Hurley
12203176
TITLE:
AUTHOR
ABSTRACT
Insulation co-ordination studies should be considered for the design of all medium to extra high voltage electric
power distribution systems. The protection against overvoltage transients is of vital importance for the utility as well
as the customer. This paper focuses on the study of a typical 132/22kV zone substation for a Western Australian
electrical utility.
INDEXING TERMS
TECHNICAL WORK
REPORT PRESENTATION
EXAMINER CO-EXAMINER
Joshua Hurley
23 Sandover Drive
Karrinyup WA 6018
3 November, 2005
Furthermore, I state that this thesis is entirely my own work outside of where
acknowledgement is given and that it has been submitted in accordance with the
Yours Sincerely,
Joshua Hurley
12203176
SYNOPSIS
This report covers the insulation co-ordination study of a typical 132/22kV zone
substation for a Western Australian electrical utility. The focus is on the various
possible lightning strikes and the associated risk of striking a substation. New
configurations to the system are also investigated such that an economic benefit
and/or increased protection may be realised. This should be read in conjunction with
the thesis by Browne (2005) who co-developed the simulation tool for the study with
the author.
i
ACKNOWLEDGEMENTS
• David Browne for working diligently with the author for two whole semesters
over all parts of this report. Both he and the author worked together in
creating the model and producing the subsequent results, however have split
up the thesis as required. All work here should also be attributed to him.
• Syed Islam for the constant supervision and advice throughout the semester.
Also, for enhancing the quality of this thesis by criticising the draft.
• Terence Law, Tony Murphy and Tom Pearcy of Western Power for the
supervision and regular meetings that allowed such progress in the report.
• Stuart Romero for critically reading the thesis as well as providing some very
• Blake Hurley for critically reading the English within the thesis and
• John-Ross Torre for critically reading the thesis and providing feedback.
ii
INDEX
1.0 INTRODUCTION 1
2.1 Definition 5
2.5.3 Shielding 8
2.7.3 Backflash 11
iii
3.0 MODEL-BASED DESIGN 14
3.1 Introduction 14
3.4.3 Cables 18
3.4.4 Busbars 18
3.4.6 Switchgear 19
4.1 Introduction 27
4.3.3 Cables 30
iv
4.3.4 Busbars 31
4.3.6 Switchgear 32
5.1 Probability 34
6.2.4 Backflash 48
v
6.3.2 Worst-case Scenario 51
7.1 Introduction 55
8.0 CONCLUSION 65
9.0 REFERENCES 67
vi
12.0 APPENDIX C – SENSITIVITY ANALYSIS TESTING RESULTS 75
vii
LIST OF FIGURES
viii
Figure 12.1: Modified Surge Arrester Test - Surge Arrester 1 75
ix
LIST OF TABLES
Table 7.6: Safety margins for different positions of the surge arrester...................... 63
x
1.0 INTRODUCTION
system against lightning and switching surges, and the cost of providing the
protection. There are many variables that have to be taken into account for a detailed
study including the elements within the substation, the cost of these elements, the
risk the utility is willing to take into account, and so on. As such the study requires a
versatile tool that will allow the elements to be changed and the risk to be quantified
This project was part of a larger joint project between the author and another Curtin
with a Western Australian electrical utility and Curtin University. The entire scope
While both students worked on all aspects of the study, Browne (2005) focuses on
the detailed development and validation of the model, whereas this thesis focuses on
the application of the model and the actual insulation co-ordination study. As such, to
appreciate this paper to the full extent, Browne’s thesis should also be read in
manufacturers and has not made as detailed a study as the one encountered in this
1
thesis. Because this study is part of an undergraduate project, the utility saves time
and resources in conducting the research and also greater credibility can be given to
the results because of the backing of a university professor as supervisor. The link
between industry and research institute clearly benefits the utility, Curtin University
Matlab Simulink with the SimPowerSystems package was used for the development
of the model because the utility has a license for this software. There are many other
implemented with the correct power system elements (included in the software).
Modules could also be produced for ease of use, such as automatically working out
The model that was produced had to correlate with currently available data however
limited it was for strikes against substations. The models that were produced agreed
current and voltage surges where the level of acceptable risk was determined.
Quantifying the risks was possible due to the probabilities associated with the
orientation of the lines and the keraunic level (number of thunderstrike days in a
year).
2
1.2 Simulation Results
After validation by Browne (2005) the models were used to simulate different
lightning events – shielding failure, backflash and direct strikes. These events were
simulated on a transmission line leading into the 132kV side of the substation and
also along the distribution line outgoing from the 22kV switchyard.
Under shielding the worst-case scenarios of lightning strikes hitting the substation
can be found by using a geometric model of the line. In the case of the unshielded
tests the strikes did not have an upper limit so the magnitude was based on a function
of acceptable risk. In all these scenarios flashovers would almost definitely occur as
they exceed the critical flashover levels to a significant extent, however there is still
the possibility of the strike occurring and propagating through the substation.
Worst-case scenarios for the different lightning events allowed the protection of the
substation to be measured. These risks were quantified with the probability of such
events occurring.
The current configuration of the system was deemed to be protected with a good
Sensitivity analysis was conducted on some of the protective devices to observe the
possible reconfiguration of the system for economic savings and also an increase in
protection. Results show that changes could be made to the basic insulation level, the
surge arrester types or position, and also the length of the earth wire.
3
1.3 Structure of Report
Chapters 4 and 5 consider the necessary data that must be inputted into the
The actual insulation co-ordination study is made in Chapter 6 and details the
important results for the zone substations that the utility uses in the network. The
various strikes are considered to both the 132kV and 22kV sides of the substation.
different surge arrester ratings, changes to the distance surge arresters are from the
transformer on the 132kV side and the effects of an earth wire. These are dealt with
under the worst cases built up from the previous three chapters.
investigations and the usefulness of a detailed study as reported here. Also, further
4
2.0 APPROACHES TO INSULATION CO-ORDINATION
2.1 Definition
voltages which can appear on the system for which the equipment is
intended and taking into account the service environment and the
The key is to attain a balance between the level of protection and the associated costs
involved.
substation is damaged then it is very costly for a utility to replace, and as a result
may cause system down time that will dissatisfy the customer-base due to the loss of
5
2.3 Basic Lightning Impulse Insulation Level
All equipment within a substation is rated with a Basic Lighting Impulse Insulation
Level (BIL). The BIL is defined as the electrical strength of insulation expressed in
terms of the rest value of the standard lightning impulse. In other words the BIL
values define a 10% probability of failure at the rated BIL. Conventional BIL is the
crest value of the lightning that BIL is measured so that the equipment does not
surges. Faults such as phase-phase, phase-earth, etc. as well as other phenomena can
also cause overvoltages but these factors are usually dealt with in power system
protection planning.
Lightning surges, as the name suggests are caused by lightning that hits a
transmission line, distribution line or other element within the power system.
Switching surges are caused by lines switching, for example capacitor banks or loads
6
Switching surges were not to be considered in the study. This was because Australian
Standards 1824.1 specify that switching surges do not need to be examined for
systems less than or equal to 245kV. The reason is that the inductances within the
lines for a system below 245kV generally keeps voltages below any critical values,
experienced throughout the electrical path of the power system. Steady-state voltages
The overvoltages exhibited on the terminals of any piece of equipment should not be
greater than the BIL otherwise the system is not insulation co-ordinated.
changing BIL levels, shielding and surge arresters and are discussed within this
section.
The basic impulse level of equipment can be increased above any possible (or
probable) overvoltage that may be caused by a surge. This option is the most
expensive of all the options and is still at risk of failure if a particularly high
7
magnitude surge hits the network. It is best to take other precautions which are both
more economical and safer. Utilities often adopt a reduced BIL which is a BIL lower
2.5.3 Shielding
shielding is most often employed to protect a system from these problems. For
example, earth wires are strung over substations and transmission lines to absorb a
lightning strike rather than any phase conductors that will transmit the surge to the
Even the best shielding schemes can fail – a lightning strike can still reach a phase
conductor (albeit with a lower probability) and cause serious damage. Protective
devices are used for this purpose. Such devices limit voltages within the system.
Devices such as circuit breakers are generally used for clearing faults. They utilise
other current and/or voltage transformers and relays as well as safety margins and
operating times. Surges are very fast compared to the clearance time of circuit
breakers and will cause damage before a circuit breaker could open.
Surges require other methods to keep overvoltages below the basic impulse levels of
equipment. Surge arresters (also known as surge diverters, surge suppressor, and
lightning or line arresters) have the ability to limit voltages within a network.
8
Surge arresters are placed in parallel and within close proximity of equipment to
protect it. Generally surge diverters are made of metal oxide material (such as ZnO)
which have a nonlinear V-I characteristic, which is a property of the boundary layers
between the crystals. They usually have a rated voltage in which the impedance
within them will drop sharply when voltages at the rated voltage are experienced so
that current is diverted through them keeping the voltage around the crest value.
Surge suppressors have an energy rating which should not be exceeded otherwise a
fault could occur, so this value also should be considered when coordinating the
insulation.
Lightning has been a curiosity to mankind but has only been studied in depth since
the advent of power distribution systems (last 100 years) due to the faults that can be
caused and the advances in technology that can record a lightning surge
(oscilloscopes and such). Many studies have been undertaken on lightning and a
general standard wave shape is accepted as the best representation for a surge which
IEEE Standard 1313.1. This representation generally involves the shape of a double
exponential however a more accurate mathematic model of the lightning stroke has
9
The main properties of a lighting surge include:
1. Maximum current – the peak crest current cause by the voltage impulse.
2. Front time (tf) – the time difference when the current is rising between 30%
3. Tail time (tT) – the time the impulse falls to 50% of the crest value.
Generally surges are denoted as a tf/tT impulse such as the standard 1.2µs/50µs or
on the international electrical standard, IEC 71-1. These standards specify the tests
that should be made on equipment including the properties of the surges that could
be exposed to the system. As specified earlier, switching surges are not considered
within this report. Lightning surges are to be tested with 1.2µs/50µs waveshape. This
10
2.7 Lightning Events
Lightning surges can cause a number of different events to take place including
Of course, multiple events of the following can occur as well as flashovers to other
Shielding failures occur when a lightning flash misses an earth wire (the shield) and
hits the phase conductor instead. There is a theoretical maximum current that will be
2.7.3 Backflash
Backflashovers occur when lightning hits a shield wire and has a high enough
magnitude that it breaks down over the insulation to a phase conductor. This
Without shielding, phase conductors are not protected and lightning will terminate
directly to cause this event. There is no maximum threshold except flashovers will
11
2.8 Brief History
Theory of elements within the substation has been built from the ground up from
capacitor. These representations are modelled by equations and were originally very
The complexities of the networks that need to be examined require a powerful tool to
be studied. Software packages such as Matlab Simulink allow a symbolic view of the
network to be studied and with ease that would have been unimaginable to the early
either the surge or the apparatus within a substation. However, the representations
used are fairly accurate and are necessary to conduct the study.
The simulations also allow for the changing of the substation configuration with ease
The utility does not currently undertake insulation co-ordination studies to its
substations. Some minor studies have been made but none as detailed as the model
12
The utility uses both manufacturer’s recommendations and Australian Standards to
build the network. While these options generally produce a very acceptable level of
this report allows the utility to have economic gains while keeping protection to a
suitable level.
Many other utilities such as ARTC (2005) or Merlin Gerin (Metz-Noblat 1994)
throughout the world use simulations as in this report. Metz-Noblat (1994) includes a
study of the associated risk similar to in this paper. Of course, some utilities without
the expertise sometimes only rely on previous experience and the recommendations
of manufacturers.
All of the options are acceptable but a simulation of a network can give the most
detailed answer for a utility and a greater appreciation of the risk involved, as well as
13
3.0 MODEL-BASED DESIGN
3.1 Introduction
the author is detailed within this chapter. All of the major components are detailed
from choice of software, to the actual layout of the substation, to the simulated
The thesis by Browne (2005) should be read if the explanations within this chapter
are not sufficient for the reader. The thesis gives all of the components in greater
detail, as well as expanded concepts and programming that were introduced into the
There are many different forms of software that are used for power system
simulation such as the SimPowerSystems pack for use with Matlab Simulink,
PSCAD, HYPERSIM, and EMTP. All of these programs allow for the study of
Matlab Simulink with SimPowerSystems was chosen for this project because the
utility already had the software and the software is understood by the employees. The
choice of software also allows a proper design-based project due to the programming
14
This package has the functionality to allow an accurate representation of all the
models used within the substation and thus a new product was not considered. The
differential equations, feedback loops and other tools required for the study of
This substation considered in the modelling will be based on the standard design for
new substations of this size and is therefore the model that is used within this design.
very useful approximation. The model can be easily modified if a more accurate
The actual layout is defined by a single line diagram and appears in Figure 3.1. The
diagram clearly shows the location of surge arresters, current and voltage
connections and of course the power transformers which are the focus of the study.
The design is clearly a two transformer system, with the high voltage side involving
two incoming transmission lines. The low voltage side involves two 22kV indoor
switchboard arrangements with eight outgoing distribution lines and four capacitor
15
Figure 3.1: Single line diagram of the 132/22kV zone substation
16
3.4 Element Representation
All of the elements that are shown in Section 3.3 are accurately modelled for the
simulation. Due to the high frequency surge, elements are represented differently to
the usual low frequency model. Discussions of the different models are included
Both transmission lines and distribution lines are generally modelled using a Pi-
network or other approximation under normal power system studies. These models
however are not accurate enough to represent the voltages on a line when subjected
to a high frequency impulse current such as a lightning surge. This is because this
propagation velocity of the surge. This model allows for reflections to be considered
in the model and thus an accurate account of the voltage magnifications occurring
using lumped losses based on Bergeron’s model (Mathworks 2004) which is also
17
3.4.3 Cables
The cables are modelled the same as the transmission and distribution lines, that is
with a distributed parameter. Cables are basically the same as for the lines but have a
lower inductance and higher capacitance as they are closer together. The electric
Clearly, the biggest difference from overhead lines is that cables are not exposed to
air as they are laid underground meaning that lightning will not hit them.
3.4.4 Busbars
The HV and LV busbars were modelled using surge impedance due to the limited
data available from the utility. Basically, the impedance was split up equally through
the busbar portions using resistors. This form of representation is adequate because it
exhibited by the transformer drop to very low impedance to ground and the
capacitance is therefore the dominant feature. Also, all currents are directional
because currents cannot penetrate due to the inductance in the windings at the high
frequencies.
18
3.4.6 Switchgear
Switchgear data was very limited with a single capacitance for phase to ground given.
Using this data assumptions were made so that the measured capacitance could be
distributed among the eight current transformers and two voltage transformers.
The distribution of the capacitances was done such that the voltage transformer
capacitance was six times greater than the capacitance of the current transformer.
SimPowerSystems includes a default surge arrester model. This model utilises the
rated currents and voltages from surge arrester data. The high frequency of the surge
does not affect the operation of the arresters significantly as they are designed for
The single line diagram is easily translated into a three-phase circuit including all of
the elements described earlier in the previous section. Three phases were included
where all of the single lines exist and equipment was similarly replaced with the
model representations.
The model must be split into two because no data exists on capacitance between the
HV and LV sides of the 132/22kV transformer. If the data for the inter-capacitances
19
existed, the model could be built so that the propagation of the surge would be
examples were found that had a fully propagating network from HV to LV sides and
vice versa which validates the choice made here. This representation is possible due
to the propagating surge only significantly affecting the first side that is hit by the
surge, the other side is exposed but most of the current is lost through the low
The 132kV side with incoming lines and the substation component appear in Figure
3.2 and Figure 3.3 respectively. The components are clearly marked in each case.
One half of the 22kV side appears in Figure 3.4, the actual model does not fit well on
an A4 page but the other half is basically connected at the bus-tie and is the same as
the first half except no lightning surge. Figure 3.5 includes the model of the
switchboard room.
20
Figure 3.2: Simulink model - Incoming transmission line
21
Figure 3.3: Simulink model - 132kV substation layout
22
Figure 3.4: Simulink model - 22kV side of substation
23
Figure 3.5: Simulink model - 22kV switchboard room
24
3.7 Current Surge
The current surge has been produced so that it can simulate any specified tf/tT
waveform. Equation (3.1) was used to produce this flexibility and was derived by
Cvetić et al (1999). A double exponential has often been used in the past but the
characteristics don’t quite match the standard waveshape definition and the high
derivative at the origin can cause singularities in the result as was discovered in the
development stage.
n −t
⎛t ⎞ ( )
Io ⎜ ⎟ e τ 2
i (t ) = ⎝ τ1 ⎠ (3.1)
⎛ ⎞
⎜ ⎟
⎜ ⎟
⎜ −τ1 ⎟
⎜ ⎛ 1 ⎞ ⎟
⎜ ⎛ nτ 2 ⎞⎜⎝ n+1 ⎟⎠ ⎟
⎛ ⎛ t ⎞⎞ ⎜⎜ τ 2 ⎜
⎝ ⎝ τ1 ⎠
⎟ ⎟⎟
⎜1 + ⎜ ⎟ ⎟ e
⎠
⎝ ⎝ τ1 ⎠ ⎠
Io : Maximum current
25
3.8 Corona Effects
The corona effect affects surges to a large degree by decreasing the steepness of the
wavefront. An accurate account of the effects of corona has been inbuilt into the
model.
Corona will lessen the impact of a lightning surge the further a surge travels along a
line. It is caused when a threshold voltage on the line is reached such that the
effective radius of the conductor is increased due to streamers emanating from the
There are other phenomena, such as earth resistivity that also causes attenuation and
26
4.0 DATA ACQUISITION
4.1 Introduction
All of the data that was acquired for the model of the substation has been supplied by
WP (2005a) unless otherwise specified. Data sheets or typical values have been used
The BIL ratings and approximate CFO for all equipment are listed in Table 4.1.
The possible incoming lines for the 132kV side of the substation included Triton
The arrangement of the wires is important and the most common arrangements were
27
Arrangement 1 Arrangement 2
E
B
18.5m
C
16.7m
15.9m
15m
Using the conductor parameters and the orientation of the line allows the sequence
28
Arrangement one is the most common and is therefore used for the insulation co-
ordination studies. It is also shown in Browne (2005) that the possible exposure to
current from a shielding failure is greatest for this arrangement type and thus further
The arrangement of the wires is important and the most common arrangement was
29
Using the conductor parameters and the orientation of the line allows the sequence
Parameters
Positive Sequence Resistance (Ω/km) 119
Zero Sequence Resistance (Ω/km) 743
Positive Sequence Inductance (Ω/km) 1.11
Zero Sequence Inductance (mH/km) 3.09
Positive Sequence Capacitance (nF/km) 10.9
Zero Sequence Capacitance (F/km) 3.89
Table 4.5: Sequence impedance parameters for distribution line
4.3.3 Cables
The cables that are used for the 22kV side of the substation are of type 500mm2 1/C
Cu XLPE Heavy Duty Cu wire SCR, PVC Sheath. Datasheets give the relevant data
Cables are assumed to run for 100m before joining to distribution lines as per WP
(2005a).
30
4.3.4 Busbars
The power transformers, current transformers, and voltage transformer values for the
132kV side were measured from terminal to ground and are summarised in Table 4.7.
Likewise, the values were also measured for the 22kV side of the substation and are
The capacitances given were not measured at high frequency to simulate lightning.
However, the representation is the best that the utility had to offer and is therefore
31
4.3.6 Switchgear
A single capacitor value for each of the 22kV switch rooms was given as 400pF.
This was measured from phase to ground with the other phases connected to ground.
The surge arrester data is based on equipment specifications provided by the utility
Total load outgoing from the substation was given as 24MW and 9MVAR. These
loads are averaged out into the eight distribution lines and therefore the outgoing
32
4.3.9 Capacitor Banks
Capacitor bank loading was given as 5MVAR, 22kV nominal with 4.53% series
reactors.
33
5.0 ASSESSING RISK
5.1 Probability
The probability of an event occurring allows the utility to assess risk in a quantifiable
manner. If a standard level of acceptable risk is decided upon then a company has a
benchmark level of safety and can set a standard cost structure to pay for the required
level of protection. Numerous studies have been made in the past of lightning strike
intensities and the probability of events occurring. Most of these have been
other sources. Also, the probability of the events is inbuilt into the model and
This chapter focuses on the probability of the events occurring and some of the
The keraunic level is defined as the average number of thunderstorm days in a year.
The Bureau of Meteorology conducts numerous studies into lightning intensity and
34
Figure 5.1: Keraunic levels throughout Australia
It is very important to note that if the utility built a system in the north of the state,
then the system would be looking upwards of 80 thunderstorm days per year. Given
that the system is in Perth, there are tradeoffs that are applicable that may not be in
other places. These possibilities may need to be considered in another insulation co-
ordination study.
To quantify the rate and likelihood of lightning surges hitting a transmission line and
thus leading into a substation, the incidence of lightning flashes is needed. The
35
The lightning flash density, Ng is an approximation of past data and is a regression
line (5.1) that is based on the keraunic level which is represented as thunderstorm
The number of lightning flashes is also taken from the aforementioned IEEE
28h 0.6
N (G ) = N g [flashes/100km ⋅ years] (5.2)
10
Flashovers are most highly influenced by the orientation of the lines. However, this
chapter deals with what is known as the geometric model and basically defines an
The geometric model presented here is referenced from Hileman (1999) who in turn
references numerous sources to its validity. This model is the accepted method and is
The model itself is based on a regression line that looks at the striking distance and
the current magnitude. The intensity of current that could strike a line is near
proportional to the distance from the conductors and earthing wire. By drawing arcs
from the lines, it can be seen where a current can protect a line and where it cannot.
36
(a) Unprotected zone (b) Protected zone
Figure 5.2 demonstrates that at a certain current the phase conductors are exposed to
the lightning surge but when the magnitude of the surge increases, the phase
conductors are protected by the shielding wire which suggests an upper maximum to
Equation (5.3) gives the maximum theoretical current, i, that can be exposed to the
phase conductors. DC must be greater than or equal to zero for the current to have a
⎛ ⎛ ⎛ a2 ⎞⎞⎞
⎜ ⎜ ⎜ (h − y ) 1 + ⎟⎟⎟
⎜ (r − y ) 2
⎜ ⎛ a ⎞ 1 −1 ⎜ (h − y )2 ⎟⎟⎟ (5.3)
Dc = rc ⎜ 1 − g 2 − cos ⎜ tan −1 ⎜ ⎟ + sin
rc ⎝h− y⎠ 2 ⎜ 2rc ⎟⎟⎟
⎜ ⎜ ⎜ ⎟⎟⎟
⎜ ⎜ ⎜ ⎟⎟
⎝ ⎝ ⎝ ⎠ ⎠ ⎠⎟
There have been many studies, with Berger one of the major contributors and
The most accepted distribution is the log normal distribution (5.4). Consensus on the
parameters is not entirely established however due to the different data sets that have
37
been used to create them. This simulation uses M equal to 34 and β equal to 0.74
The rate of shielding failures to the line is a product of the geometric model, the
length of exposed line and also the probability density function of lightning
mentioned in section 5.5. Thus the rate of shielding failures is found from (5.5).
I max
SFR = 2 N g L ∫
0
Dc f ( I )dI (5.5)
Backflash rates are also a function of the probability density function of lightning,
with the lightning exceeding the critical current as defined by the CFO (critical
flashover) of the line simulation. Equation (5.6) gives the backflash rate.
BFR = N (G ) P( I c ) (5.6)
Direct strikes are simply based on the distribution of lightning from section 5.5.
38
5.7 Risk of Event
The measure of risk for the event happening in the substation’s lifetime can be found
using the exponential distribution. The statistical time between two occurrences is
equal to,
1
Td = (5.7)
λ
The risk of the event exceeding the magnitude of the lightning current being
Risk = 1 − e( − ts / Td ) (5.8)
39
6.0 INSULATION CO-ORDINATION TESTING
The substation was simulated using the Simulink tool that is defined in prior chapters
of this report. All of the system data that was available and appearing in chapter 4.0
was inputted into the model as well as the keraunic level appearing in Figure 5.1.
As such, with all of the data, the substation is able to be accurately simulated. All of
the possible lightning events need to be considered and tested on the model. Possible
The typical 132/22kV zone substation design must be modelled on both the 132kV
and 22kV sides to prove the validity of the design. It is noted however that in general
practice only the high voltage side is modelled because flashovers almost always
occur on the 22kV side due to the size of the conductors, the spacing and proximity
The most important statistics including the maximum voltage experienced by the
transformers and also the energy absorbed by the surge arresters is summarised in the
results output and is displayed in tabular format within this chapter for each event.
The graphical outputs are displayed only for the shielding failure and thereafter
appear in Appendix B.
40
Based on the output of the model, the substation can be shown to be insulation co-
ordinated or in need of a better design. The main figure that governs this is the safety
margin – a safety coefficient of 20%-30% is the norm for a substation (ARTC 2005).
BIL − Vmax
S= *100 (6.1)
BIL
The 132kV side of the substation includes two incoming transmission lines with
shielding wires of approximately 1.5km in length. As such, the possible events are a
The shielding failure and the backflash can occur in the section of line where the
earth wire exists. Direct strikes occur beyond this point as there is no shielding.
considered. The highest transfer of surge occurs to a single transformer when the
high voltage bus-tie is not connected and the surge occurs within the span of
transmission line closest to the substation. Figure 6.1 shows the connections for the
substation.
41
Figure 6.1: 132kV configuration for simulation
42
Even though this situation is regarded as having a low probability of occurring as
opposed to the high probability of the substation operating at full capacity, the safe
level of operation at this level shows that the substation is even higher protected at
normal operation. The following page shows the setup of the model for this scenario.
The shielding failure has been tested within the final span of the model (80m of
transmission line) and the results are shown in this chapter. As this is the first
simulation, all figures are shown and explained as a representation of the results.
Chapters thereafter will only summarise the results and figures can be referred to in
Appendix B.
The simulation automatically tests the model for the highest current theoretically
possible due to the geometrical model of the line as discussed in Section 5.4.
Running the simulation gives Figures 6.2, 6.3 and 6.4 that give the two surge
43
Surge Arrester 1 of Tx 1 Voltage
2000
Voltage (kV) 0
-2000
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Current
50
Current (kA)
-50
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Absorbed Energy
400
Energy (kJ)
200
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Figure 6.2 shows that for the first surge arrester, the voltage has indeed been limited
to the rated voltage level. The current resembles the lightning surge current which
shows that the lightning current has entered the substation and when it drops to zero
the voltage has dropped to a low enough level such that the surge arrester no longer
diverts energy (also evident from the energy diagram). This figure shows the user
44
Surge Arrester 2 of Tx 1 Voltage
500
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Current
20
Current (kA)
-20
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Absorbed Energy
400
Energy (kJ)
200
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge arrester two also performs like surge arrester one and is clearly very
comparable in attributes except the current and energy is lower. This is expected as
the second surge arrester is exposed to the surge after the first surge arrester has
45
Transformer 1 Voltage (kV)
500
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Transformer 1 Current (kA)
4
2
Current (kA)
-2
-4
0 20 40 60 80 100 120 140 160 180 200
Time (us)
The power transformer experiences a maximum voltage at the beginning of the surge
and then drops slowly for 140µs, followed by significant decreases further after this
The graphical outputs are necessary to show that in the testing range of 200µs, the
transformer has been exposed to the maximum voltage and also the surge arresters
have absorbed all the energy that is possible from the surge. The above figures
validate these showing that the testing time of 200µs is a valid operational time for
the simulation to run and as such results can be used with high reliability.
46
Results can be summarised (Table 6.1) to look at the current that was supplied, the
safety coefficient and the energy to the surge arresters as well as the risk that is
Clearly, the surge leaves the power transformer intact with a large safety margin. The
surge arresters also do not exceed the maximum energy for a fault and have a large
An anomalous result occurs in this case as the risk of this event is zero due to the
geometric model limiting the current to a maximum and simulation runs at that
maximum value. The risk for shielding failure is considering the conditional
47
6.2.4 Backflash
The backflash event has been tested within the final span of the model (80m of
transmission line) and the results are summarised in Table 6.2 with the graphical
outputs in Appendix B.
The simulation automatically tests the model for the highest current theoretically
possible for a backflash due to the geometric model and insulation of the line.
Clearly, the surge leaves the power transformer intact with a large safety margin. The
surge arresters also do not exceed the maximum energy for a fault and have a large
safety factor built into them. Graphs in the appendix show the runtime is sufficient.
As a conclusion, there is an 11% chance that, during the service life of the substation,
exceed 462kV, corresponding to 71% of the defined BIL, i.e. a calculated safety
coefficient of 29%.
48
6.2.5 Direct Strike
Direct strike events can only occur for the 132kV line at 1.5km and beyond because
of the earth wire. This event is thus tested at the 1.5km limit and the results are
At a 1.5km distance, attenuation of the surge due to corona and earthing resistivity
lowers the possibility of a destructive surge. And in fact, due to the very high
possibility of flashovers for any surges over the CFO of the line only a low
The design current used is such that 2 times the CFO voltage is experienced so that a
relative large safety margin of 100 percent above the CFO is considered. The current
2* CFO
I direct = (6.2)
Z tower
49
Clearly, the surge leaves the power transformer intact with a large safety margin. The
surge arresters also do not exceed the maximum energy for a fault and have a large
safety factor built into them. Graphs in the appendix show the runtime is sufficient.
As a conclusion, there is a very high risk of this event occurring (in terms of the
magnitude of the current however this is unlikely due to the extremely high
corresponding to 65% of the defined BIL, i.e. a calculated safety coefficient of 35%.
Shielding
Failure Backflash Direct Strike
Maximum Lightning Current (kA) 15.9 7.26 4.9
Peak Transformer Voltage (kV) 485 462 424
Transformer BIL (kV) 650 650 650
Max Voltage as % of BIL 75 71 65
Safety Coefficient (%) 25 29 35
Surge Arrester Fault Level (kJ) 624 624 624
Surge Arrester 1 (kJ) 341 120 83
Surge Arrester 1 (% of fault) 55 19 13
Surge Arrester 2 (kJ) 211 85 59
Surge Arrester 2 (% of fault) 34 14 9
Lifetime of Substation (years) 30 30 30
Risk of event (%) 0 11 99
Table 6.4: Summary of simulations to 132kV side
The conclusion that can be arrived from the tests to the 132kV side of the substation
is that it is highly protected against strikes. With worst-case events the safety
coefficient is at a very acceptable level of 25%. This shows that changes can be made
to the 132kV side such that the cost may be decreased if the current safety level is
relaxed.
50
6.3 The 22kV Side
The 22kV side of the substation includes eight outgoing distribution lines with no
considered. The highest transfer of surge occurs to a single transformer when the low
voltage bus-tie is not connected, only one distribution line is connected and no
capacitor banks, and the surge occurs within the span of distribution line closest to
the substation. Figure 6.5 shows the connections for the simulation.
the high probability of the substation operating at full capacity, the safe level of
operation at this level shows that the substation is even higher protected at normal
operation. The following page shows the setup of the model for this scenario.
51
Figure 6.5: 22kV configuration used for simulation
52
6.3.3 Direct Strike
The 22kV line is high at risk of flashovers; in fact using equation (6.2) only registers
a current crest of 1.375kA. Generally, surges are run from a minimum of 3kA and
Clearly, the surge leaves the power transformer intact with a large safety margin. The
surge arresters also do not exceed the maximum energy for a fault and have a large
safety factor built into them. Graphs in Appendix B show the runtime is sufficient.
As a conclusion, there is a very high risk of this event occurring (in terms of the
magnitude of the current however these are unlikely due to the extremely high
caused by lightning resulting in the direct event will reach or exceed 113kV,
corresponding to 75% of the defined BIL, i.e. a calculated safety coefficient of 25%.
53
6.3.4 Summary of Protection
The conclusion that can be arrived from the tests to the 22kV side of the substation is
that it is highly protected against strikes. With worst-case events the safety
The simulations on both the 132kV and 22kV sides successfully show that the power
margin of 25% is recorded for both sides of the station with surge arresters not
hitting fault levels for the worst-case scenarios that were tested.
This insulation co-ordination study has shown that the typical 132/22kV zone
investigation into possible changes to the configuration that may incur a loss of some
protection but with an economic windfall. The next chapter deals with this topic of
interest.
54
7.0 SENSITIVITY ANALYSIS
7.1 Introduction
substation focusing mainly on the elements but also on the layout. The purpose of
this section is to consider alternative setups that while still providing protection, also
Economically, there are only a limited number of changes that can be made to a
chapter are the power transformer and the surge arresters. There are other options
such as changes to the cables but these are only discussed later and not researched.
The bulk cost of a substation is attributed to the power transformer(s). In the case of
$700,000 but this depends greatly on the demand for new transformers and the price
of copper.
55
The surge arresters considered in this report are the 132kV overhead line connected
and the 22kV aerial feeder. Clearly, the cost of a surge arrester is insignificant when
The sensitivity analysis utilised, where necessary, the same methods as the original
insulation co-ordination study. The worst case scenario was taken for the
Some of the study did not need new simulations, rather an in-depth discussion of the
ramifications of changing the setup based on the results from the initial study.
The relative cheapness and the high protection already existing on the 22kV side
warrant no further investigation. Also, the power transformer cost is mainly based on
the HV BIL further reducing the warrant to study the 22kV side in further depth.
In terms of cost of the substation, the BIL rating has the most significant impact
when looking at protection. Australian Standards 1824.1 specifies two different BIL
ratings for 132kV power transformers – 550kV and 650kV plus some provisions for
a 450kV rating. The current configuration is 650kV and therefore the other two
56
Transformer BIL (kV) 650 550 450
Maximum Lightning Current (kA) 15.9 15.9 15.9
Peak Transformer Voltage (kV) 485 485 485
Max Voltage as % of BIL 75 88 108
Safety Coefficient (%) 25 12 -8
Table 7.2: Comparison of different BIL ratings
A comparison of the different BIL ratings that are possible shows that the 450kV
BIL does not protect the substation. However, the 550kV BIL has a high enough
safety coefficient to run in the station although this somewhat out of the safety
A typical 132kV transformer for the utility costs approximately $700,000 and even
though the quoting for a 550kV BIL is still under consideration, it is believed a
significant portion of the cost could be reduced. Therefore, the 550kV BIL power
If the surge arrester limiting voltage is modified, then the exposed transient voltages
throughout the substation and in particular the power transformer will change
applicably also.
It is not possible to reduce the surge arrester voltage ratings indefinitely because
surge arresters have a rating known as the MCOV – the maximum continuous
overvoltage. In other words the limiting voltage is a function of the surge arrester
type but the type is limited by the MCOV. The MCOV must be greater than the line-
57
neutral voltage of the line with an added safety margin to take into account the
voltage regulation.
The IEEE Std C62.11-1993 lists numerous MCOV voltages. Of concern in this
report is the MCOV ratings of 96kV and a lower value of 84kV with the former
value being the current surge arrester type. Lower ratings are not investigated
because the minimum MCOV will be breached as in Table 7.3. Of course, there are
intermediate ratings (88kV) that could be tested also but the best results are assumed
58
The simulation was run using the trial arrester with MCOV of 84kV. Results (Table
7.4) show that the trial arrester offers far superior performance with an enhanced
safety margin. The one drawback of this different arrester could be the extra risk in
(greater than 10% above the rms voltage) so any problems should be avoided.
Appendix B shows the graphical results for the test which show the runtime is
sufficient.
The cost difference between the two surge arresters is minimal with a slight saving of
$100-$200 produced from the change. The protective nature of the new device would
be the reason for the change however and not the cost saving.
Another enhancement that could be made is lowering the energy level that the surge
arrester could handle, the large gap between absorbed energy and energy to fault at
present leaves this as an option. Generally, however the ratings are fairly standard by
cost saving would most likely be minimum (if anything at all) and the increased risk
59
7.6 Earth Wire
The shielding wire extending from a substation can substantially increase the cost of
the design due to the extra loading on the towers and the tower designs necessary to
include space for the shielding wire as well as the cost of the conductor. While these
differences may not come close to the cost of a power transformer, they are
The current length of earth wire (approximately 1.5km) is based on surge arrester
The simulations were made by testing a direct strike outside of the earth wire range.
Two scenarios – an earth wire length of 500m and no earth wire at all were compared
with the results of the 1.5km length and results are below in Table 7.5.
60
Quite clearly, the results show that no earth wire at all would be sufficient for
protection of the substation. However, this is not recommended because the tests are
based on the risk of a flashover across the line insulation. On the low probability that
a flashover might occur, such an event could be very destructive without the shield to
provide protection against a maximum crest current based on the geometric model.
Based on the results as well as reasoning of the possible event where a flashover does
not occur, it appears that the current length of earth wires can be shortened to
perhaps two or three spans running to approximately 500m to 800m of length. This
The position of the surge arresters can have a large affect on the voltage that would
be experienced on certain parts of the line. Position is important because it lowers the
length of the line for reflections to build up if the conductor is shorter. In particular,
possible to make it closer, then lower transient voltages would be experienced at the
The testing of the position of the surge arrester was made such that the distance
between the second current transformer and the power transformer were fixed but the
surge arrester was not, which makes 6.6m leeway withstanding physical constraints.
Physical constraints do not allow the surge arrester to be within 1.4m of the
transformer.
61
The model was simulated numerous times for different distances with similar
graphical results in all cases. The shielding failure results in Section 6.2 are
A clear downward trend from the results in Figure 7.1 is seen as the distance is
shortened. Clearly, the closer the transformer is to the transformer the better. This
500
495
490
485
Voltage (kV)
480
475
470
465
460
455
450
0 0.5 1 1.5 2 2.5 3 3.5 4
Distance (m)
Based on the data, it is best to have the surge arrester as close to the transformer as
62
Distance Maximum Safety Margin
(m) Voltage (kV) (%)
1.4 462 29
1.6 464 29
1.7 465 28
1.75 466 28
2 468 28
2.25 471 28
2.5 473 27
2.75 475 27
3 477 27
3.1 483 26
3.2 485 25
3.5 490 25
Table 7.6: Safety margins for different positions of the surge arrester
There are numerous other changes that could be made to the substation design. While
these are not considered in the analysis due to the small difference made and/or
Orientation of the line can enhance the substation in two ways – further shielding and
Further shielding by modifying the angles of the lines in comparison to the shielding
wire can result in exposure to a theoretical lower maximum current which would
reduce the need for a higher BIL rating. This is based on the geometric model
conductors. If these parameters are changed then the inductance and capacitance of
the line changes which may retard the strength of the surge.
The problem with the change in orientation is the cost of the modifications to be
made and to the standard towers that are used not just at the entrance to substations
but to entire lengths of line. A new standard could be created for outgoing of
Changes to the conductors that are used for transmission lines, distribution lines and
within the substation are possible but because most suitable conductors used for this
job have similar parameters. As before the RLC parameters would be changed with
deemed significant compared to the other modifications that were considered within
this study.
64
8.0 CONCLUSION
This project was undertaken by David Browne and the author whom together co-
developed a Matlab Simulink model to run insulation co-ordination studies. The full
Utilising the tool in Simulink, the current configuration of the utility’s standard
132/22kV zone substation has been simulated under a full insulation co-ordination
study. The study shows that under a worst-case scenario the substation is protected
Possible changes to the current configuration were also investigated that could bring
2. Change the surge arresters on the 132kV side to a MCOV level of 84kV.
3. Move the surge arrester closest to the power transformer on the high voltage
4. Reduce the length of the earth wire to two to three spans instead of the
current 1.5km.
As the realm of insulation co-ordination is very large and with the flexibility of the
model that was produced for this project, much of this topic can be explored in
greater depth. There are many additional topics of interest that can be included in
65
Substations of all voltage ratings whether low, high or extra high can be simulated
using the methods described within this report. The model is simply adjusted by
Substations of higher voltage (greater than 245kV) can be simulated for both
Section 2.
An insulation co-ordination tool of greater detail could be built using even higher
order models as long as the equivalent data was obtainable from the utility or
manufacturer. Also, more accurate measurements could be made, for example using
high frequency tests to calculate capacitances for the transformers rather than 50Hz.
Multiple strikes leading into the substation could be investigated. Generally multiple
strikes do not damage the transformer equipment because the average time between
surges is 35ms to 45ms (Hileman 1999) and surge propagation from the former strike
has already diminished by then. However, if a circuit breaker opens then damage can
The scope can also be extended beyond protection towards many other interests such
as harmonics. Using simulations there is no real limit to what can be achieved except
for the data that is available and the processing power of the computer.
66
9.0 REFERENCES
www.artc.com.au/docs/nsw/engineering-standards/pdf/pds17_insulation_
Technology, Perth.
CIGRE Working Group 33.01 (1991). Guide to Procedures for Estimating the
Hileman (1999).
Parameters. IEEE Transactions on Power Delivery, vol. 14, no. 2, pp. 399-404.
67
Hileman, A. (1999). Insulation Coordination for Power Systems, Marcel Dekker,
New York.
IEEE Std 1313.2-1999, IEEE guide for the application of insulation coordination
IEEE Std C66.22-1997, IEEE Guide for the Application of Metal-Oxide Surge
Perth.
68
Western Power Corporation (2005a) Personal communication
69
10.0 APPENDIX A – PROJECT SCHEDULE
The project schedule is shown in Figure 10.1. The schedule was followed from start
70
11.0 APPENDIX B – INSULATION CO-ORDINATION TESTING RESULTS
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Current
20
Current (kA)
-20
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Absorbed Energy
200
Energy (kJ)
100
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Current
10
Current (kA)
-10
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Absorbed Energy
100
Energy (kJ)
50
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
71
Transformer 1 Voltage (kV)
500
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Transformer 1 Current (kA)
4
Current (kA)
-2
0 20 40 60 80 100 120 140 160 180 200
Time (us)
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Current
10
Current (kA)
-10
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Absorbed Energy
100
Energy (kJ)
50
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
72
Surge Arrester 2 of Tx 1 Voltage
500
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Current
5
Current (kA)
-5
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Absorbed Energy
100
Energy (kJ)
50
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Transformer 1 Current (kA)
2
Current (kA)
-1
0 20 40 60 80 100 120 140 160 180 200
Time (us)
73
B3 Direct Strike Event to 22kV Side
-100
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Transformer 1 Current (kA)
0.5
Current (kA)
-0.5
-1
0 20 40 60 80 100 120 140 160 180 200
Time (us)
74
12.0 APPENDIX C – SENSITIVITY ANALYSIS TESTING RESULTS
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Current
50
Current (kA)
-50
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Absorbed Energy
400
Energy (kJ)
200
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Current
20
Current (kA)
-20
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Absorbed Energy
200
Energy (kJ)
100
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
75
Transformer 1 Voltage (kV)
500
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Transformer 1 Current (kA)
5
Current (kA)
-5
0 20 40 60 80 100 120 140 160 180 200
Time (us)
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Current
10
Current (kA)
-10
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Absorbed Energy
100
Energy (kJ)
50
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
76
Surge Arrester 2 of Tx 1 Voltage
500
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Current
5
Current (kA)
-5
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Absorbed Energy
100
Energy (kJ)
50
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Transformer 1 Current (kA)
2
Current (kA)
-1
0 20 40 60 80 100 120 140 160 180 200
Time (us)
77
C2.2 No Earth Wire
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Current
10
Current (kA)
-10
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 1 of Tx 1 Absorbed Energy
100
Energy (kJ)
50
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Current
5
Current (kA)
-5
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Surge Arrester 2 of Tx 1 Absorbed Energy
100
Energy (kJ)
50
0
0 20 40 60 80 100 120 140 160 180 200
Time (us)
78
Transformer 1 Voltage (kV)
500
Voltage (kV)
0
-500
0 20 40 60 80 100 120 140 160 180 200
Time (us)
Transformer 1 Current (kA)
2
Current (kA)
-1
0 20 40 60 80 100 120 140 160 180 200
Time (us)
79