Transients in Power Systems
Transients in Power Systems
Transients in Power Systems
Transients in
Power Systems
KEMA High-Power Laboratory, Arnhem, The Netherlands
Preface xi
3 Travelling Waves 31
3.1 Velocity of Travelling Waves and Characteristic
Impedance 32
3.2 Energy Contents of Travelling Waves 34
3.3 Attenuation and Distortion of Electromagnetic
Waves 36
3.4 The Telegraph Equations 38
3.4.1 The Lossless Line 40
3.4.2 The Distortionless Line 42
3.5 Reflection and Refraction of Travelling Waves 42
viii CONTENTS
4 Circuit Breakers 57
4.1 The Switching Arc 58
4.2 Oil Circuit Breakers 63
4.3 Air-Blast Circuit Breakers 64
4.4 SF6 Circuit Breakers 64
4.5 Vacuum Circuit Breakers 66
4.6 Modelling of the Switching Arc 68
4.7 Arc–Circuit Interaction 74
4.8 References for Further Reading 80
5 Switching Transients 83
5.1 Interrupting Capacitive Currents 84
5.2 Capacitive Inrush Currents 91
5.3 Interrupting Small Inductive Currents 93
5.4 Transformer Inrush Currents 95
5.5 The Short-Line Fault 97
5.6 References for Further Reading 105
Index 205
Preface
The power system is one of the most complex systems designed, built,
and operated by engineers. In modern society, the power system plays an
indispensable role, and a comparable quality of life without a constant
and reliable supply of electricity is almost unthinkable. Because electricity
cannot be stored in large quantities, the operation of the power system has
the constraint of balancing the production of electricity in the connected
power stations and the consumption by the connected loads and of
maintaining constant frequency and constant voltage with the clients.
During normal operation, loads are connected and disconnected. Control
actions are therefore continuously necessary – the power system is never
in a steady state. On a timescale of years, planning of new power
plants, the erection of new transmission lines, or the upgrading from
existing lines to higher voltage levels are important items to consider.
When we look ahead into the future, the main topic is the economical
operation – what is the expected load and what is the most economical
fuel to be used to heat the boilers in the power stations. When the
reliability of the system is analysed with repetitive load-flow calculations,
the timescale is usually hours, yet when the dynamic stability is analysed
to verify whether the system remains stable after a major disturbance,
the power system is studied with an accuracy of seconds. Switching
actions, either to connect or disconnect loads or to switch off faulted
sections after a short-circuit, and disturbances from outside, such as
a lightning stroke on or in the vicinity of a high-voltage transmission
line, make it necessary to examine the power system on an even smaller
timescale, microseconds to milliseconds. We speak in that case of electrical
transients. The time that the electrical transients are present in the system
is short, but during a transient period, the components in the system
are subjected to high current and high-voltage peaks that can cause
considerable damage.
xii PREFACE
This book deals with electrical transients in the power system. Much
has been learned about transient phenomena since the early days of power
system operation. Pioneers in this field were men like Charles Proteus
Steinmetz and Oliver Heaviside who focussed on the understanding of
electrical transients in a more or less general way. They took the analytical
approach, which is restricted to linear circuits. When a circuit becomes
more complex, the application of this method becomes very laborious and
time-consuming. After the Second World War, new tools were developed
and used in studying circuit transient phenomena that were previously
avoided because of their complexity. The transient network analyser
(TNA) was exceptionally useful in studying the behaviour of a large
variety of complex linear and nonlinear circuits. The TNA was a powerful
tool for obtaining solutions to problems involving distributed constants
as well as nonlinear impedances. The use of the analogue TNA resulted
in the publication of much technical literature. In 1951, Harold Peterson
published his book ‘Transients in Power Systems’ with many examples
of TNA studies. Peterson’s book is a practical survey of the particular
phenomena (faults, sudden loss of load, switching surges, and so forth.)
that can cause transients and is based on his practical experience with
the General Electric Company in the USA. A classical book is Reinhold
Ruedenberg’s ‘Transient Performance of Power Systems,’ published in
1950 and based on his earlier work written in German. In addition,
switchgear design is closely related to electrical transient phenomena, and
books from authors such as Biermanns and Slamecka, who wrote from
their experience with the switchgear divisions of AEG and Siemens, are
a historical source for the understanding of transient phenomena and
switchgear development.
When I joined KEMA in 1977, as a test engineer in the famous
‘de Zoeten’ high-power laboratory, I entered the world of short-circuit
testing. The testing of power system equipment according to IEC and
ANSI standards, calculating test circuits, measuring high currents and
high voltages in an electromagnetically hostile environment, and so forth
deepened my knowledge about electrical engineering and about physics.
My first introduction to the subject was Allan Greenwood’s ‘Electrical
Transients in Power Systems.’ Later, I went through many more classical
books and papers, which gave me a good overview of the historical
development of high-voltage circuit breakers. In the fifteen years and more
that I had the pleasure of working at KEMA, I learned a lot from my former
colleagues at the high-power laboratory. Together we designed new test
circuits, developed new measuring equipment and built a computerised
measurement system with transient recorders and computer workstations.
PREFACE xiii
frequency transients that quite often result from switching actions and
that can cause considerable damage to the power system components. An
overview of how the different short-circuit duties are represented in the
IEC and IEEE/ANSI standards is given. In Chapter 7, Lightning-Induced
Transients, the mechanism of lightning is explained, and the chapter
focuses on the impact of lightning strokes on or in the vicinity of transmis-
sion lines and substations. The calculation of electrical transients without
the help of a computer is nowadays hardly unthinkable. The mathemat-
ical formulation and the numerical treatment of power system transients
is shown in Chapter 8, Numerical Simulation of Electrical Transients.
Special attention is given on how to incorporate nonlinear elements, such
as arc models, in transient computer programs such as EMTP, MNA, and
XTrans and the MATLAB Power System Blockset. A demo version of
the XTrans program can be downloaded from http://eps.et.tudelft.nl. The
background of the insulation coordination and the relevant IEC-standards
and IEEE/ANSI-standards together with a brief history of IEC, ANSI,
CIGRE and STL are given in Chapter 9, Insulation Coordination, Stan-
dardisation Bodies, and Standards. Testing of high-voltage circuit breakers
(the proof of the pudding is in the eating) in the high-power laboratory
and the related measurements and measuring equipment are described in
Chapter 10, The Testing of Circuit Breakers.
I am very much obliged to my secretary Tirza Drisi who devotedly
edited the manuscript and to Henk Paling who made the excellent draw-
ings. Pieter Schavemaker put many hours in painstakingly reading the
manuscript to filter out errors and inconsistencies. In Chapter 8, The
Numerical Simulation of Electrical Transients, I fruitfully used the educa-
tional and illustrative examples that Pieter developed for the chapter about
numerical transient calculations in his thesis. In writing Chapter 6, Power
System Transient Recovery Voltages, I received valuable support about
the latest developments in IEC and IEEE/ANSI standards from Henk te
Paske, Test Engineer at KEMA’s high-power laboratory. Martijn Venema
from KEMA supplied the photos.
1
Basic Concepts and Simple
Switching Transients
between the number of the windings of the primary coil and the number
of the windings of the secondary coil. However, for a lightning-induced
voltage wave, the stray capacitance of the windings and the stray capaci-
tance between the primary and secondary coil determine the transformer
ratio. In these two situations, the power transformer has to be modelled
differently!
When we cannot get away with a lumped-element representation,
wherein the inductance represents the magnetic field and the capacitance
represents the electric field and the resistance losses, we have to do the
analysis by using travelling waves. The correct ‘translation’ of the physical
power system and its components into suitable models for the analysis
and calculation of power system transients requires insight into the basic
physical phenomena. Therefore, it requires careful consideration and is
not easy.
A transient occurs in the power system when the network changes from
one steady state into another. This can be, for instance, the case when
lightning hits the ground in the vicinity of a high-voltage transmission
line or when lightning strikes a substation directly. The majority of power
system transients is, however, the result of a switching action. Load-
break switches and disconnectors switch off and switch on parts of the
network under load and no-load conditions. Fuses and circuit breakers
interrupt higher currents and clear short-circuit currents flowing in faulted
parts of the system. The time period when transient voltage and current
oscillations occur is in the range of microseconds to milliseconds. On
this timescale, the presence of a short-circuit current during a system
fault can be regarded as a steady-state situation, wherein the energy
is mainly in the magnetic field, and when the fault current has been
interrupted, the system is transferred into another steady-state situation,
wherein the energy is predominantly in the electric field. The energy
exchange from the magnetic field to the electric field is when the system
is visualised by lumped elements, noticed by transient current and voltage
oscillations.
In this chapter, a few simple switching transients are thoroughly anal-
ysed to acquire a good understanding of the physical processes that play
a key role in the transient time period of a power system. As switching
devices, we make use of the ideal switch. The ideal switch in closed posi-
tion is an ideal conductor (zero resistance) and in open position is an ideal
isolator (infinite resistance). The ideal switch changes from close to open
position instantaneously, and the sinusoidal current is always interrupted
at current zero.
SWITCHING AN LR CIRCUIT 3
di
Emax sin(ωt + ϕ) = Ri + L (1.1)
dt
The switch can close the circuit at any time instant and the phase angle
can have a value between 0 and 2π rad. To find the general solution of
the differential equation, we have to solve the characteristic equation of
L R
E = Emax sin(wt + j)
Ri + Lλi = 0 (1.2)
REmax ωLEmax
A= B=− (1.5)
R2 + ω 2 L2 R2 + ω 2 L 2
The complete solution, which is the sum of the general and particular
solution, is
Before the switch closes (Figure 1.1), the magnetic flux in the inductance
L is equal to zero; this remains so immediately after the instant of closing,
owing to the law of the conservation of flux. Therefore, at t = 0, the
instant of closing, we can write
Emax −1 ωL
C1 + √ sin ϕ − tan =0 (1.8)
R2 + ω 2 L 2 R
SWITCHING AN LR CIRCUIT 5
This gives us the value for C1 ; hence, the complete expression for the
current becomes
−(R/L)t −Emax ωL −1
i(t) = e √ sin ϕ − tan
R2 + ω 2 L 2 R
Emax −1 ωL
+√ sin ωt + ϕ − tan (1.9)
R2 + ω 2 L 2 R
The first part of Equation (1.9) contains the term exp[−(R/L)t] and
damps out. This is called the DC component. The expression between the
brackets is a constant and its value is determined by the instant of closing
of the circuit. For [ϕ − tan−1 (ωL/R)] = 0 or an integer times π , the DC
component is zero, and the current is immediately in the steady state. In
other words, there is no transient oscillation. When the switch closes the
circuit 90° earlier or later, the transient current will reach a maximum
amplitude, as can be seen in Figure 1.2.
The current in Figure 1.2 is called an asymmetrical current. In the case
where no transient oscillation occurs and the current is immediately in
the steady state, we speak of a symmetrical current. The asymmetrical
current can reach a peak value of nearly twice that of the symmetrical
current, depending on the time constant L/R of the supply circuit. This
implies that, for instance, when a circuit breaker closes on a short-circuited
high-voltage circuit, strong dynamic forces will act on the connected bus
bars and lines because of the large current involved.
2.0
DC component
Current in per unit value
1.0
−1.0
0 0.01 0.02 0.03 0.04 0.05 0.06
Time in seconds
Figure 1.2 The shape of a transient current in an inductive circuit depends on the instant
of switching
6 BASIC CONCEPTS AND SIMPLE SWITCHING TRANSIENTS
When the time constant of the supply circuit is rather high, which
is the case for short-circuit faults close to the generator terminals, the
transient and subtransient reactance of the synchronous generator cause
an extra-high first peak of the short-circuit current. After approximately
20 milliseconds, when the influence of the transient and subtransient
reactance is not present any longer, the synchronous reactance reduces
the root-mean-square value (rms value) of the short-circuit current. Under
these circumstances, an alternating current flows without current zeros
for several periods in the case of a fault in one of the phases because of
the large DC component. This current cannot be interrupted because the
current zero necessary for current interruption is lacking.
L C
E i(p) Vc (0)
= pLi(p) − Li(0) + + (1.11)
p pC p
When we look for the initial conditions, it is clear that i(0) = 0 as the
current in the network is zero before the switch closes and because of
the physical law of conservation of the flux. This is the case immediately
after closing of the switch too. In the case of the capacitor, the situation is
not so easy because the capacitor can have an initial voltage, for instance,
because of a trapped charge on a capacitor bank.
Let us assume that there is no charge on the capacitor and therefore
Vc (0) = 0 and let ω02 = 1/LC.
Equation (1.12) becomes
C ω0
i(p) = E (1.13)
L p2 + ω02
When there is a charge present on the capacitor, the current in the Laplace
domain becomes
C ω0
i(p) = [E − Vc (0)] (1.15)
L p2 + ω02
E E p
Vc (p) = − pLi(p) = − [E − Vc (0)] 2 (1.16)
p p p + ω02
Figure 1.4 shows the voltage waveforms for three initial values of the
capacitor voltage. From these voltage waveforms in Figure 1.4, it can
be seen that for Vc (0) = 0 the voltage waveform has what is called a
(1-cosine) shape and that it can reach twice the value of the peak of the
source voltage. For a negative charge, the peak voltage exceeds this value,
as the electric charge cannot change instantly after closing the switch. In
addition, when the characteristic impedance of the circuit has a low value,
for example, in the case of switching a capacitor bank (a large C) and a
strong supply (a small L), the peak of the inrush current after closing the
switch can reach a high value.
200
Vc (0) = 0
−100
Time
Figure 1.4 Voltage across the capacitor for three different initial values of the capacitor
voltage. The DC voltage source has the value E = 100 V
SWITCHING AN RLC CIRCUIT 9
d2 i R di 1
0= 2
+ + i (1.19)
dt L dt LC
The general solution of the homogeneous differential equation is
L C R
E max sin(wt + j)
Emax
i(t) = eαt (C1 + C2 ) + 2
1
R +
2 − ωL
ωC
1
− ωL
× sin ωt + ϕ + tan−1 ωC (1.28)
R
with α = −(R/2L).
3. In the case that (R/2L)2 < 1/LC, the roots λ1 and λ2 in the general
solution (Equation (1.23)) are complex.
λ1 = α + jβ and λ2 = α − jβ with α = −(R/2L) and β = [(1/LC) −
(R/2L)2 ]1/2 and Equation (1.23) can be written as
with
C1 = Re(C1 ) + j Im(C1 )
When we substitute for Re(C1 ) = (k1 /2) and for Im (C1 ) = (−k2 /2), the
expression for the general solution is ih (t) = eαt (k1 cos βt + k2 sin βt). The
12 BASIC CONCEPTS AND SIMPLE SWITCHING TRANSIENTS
In the three cases mentioned herewith (see Figure 1.6), the particular
solution is the same but the general solution is different. The transient
component in the current contains sinusoidal functions with angular
frequency β, which usually differs from 50- or 60-Hz power frequency
of the particular solution, and this is the cause for the irregular shape
of the current. When the DC component exp(αt) = exp[−(R/2L)t] =
exp[−(t/τ )] has damped out with the damping time constant and the
transient part of the current has been reduced to zero, the steady-state
2.0
Oscillating
Critically
damped
Current in per unit value
1.0 Overdamped
0
0.01 0.03 0.04 0.05 0.06
Time in seconds
−1.0
Figure 1.6 Overdamped, critically damped, and oscillating response of an RLC series
π
circuit after closing the switch at maximum supply voltage t = 0, ϕ = +
2
REFERENCES FOR FURTHER READING 13
current lags or leads the voltage of the source. The absolute value of 1/ωC
and ωL in the term tan−1 {[(1/ωC) − (ωL)]/R} determine if the current
is lagging (in a dominant inductive circuit) or leading (in a dominant
capacitive circuit). After a time span of three times the damping time
constant τ = 2L/R, only 5 percent of the initial amplitude of the transient
waveform is present in the network.
It is not necessary that after every change of state, such as after a
switching action, transient oscillations occur in a network. It is very well
possible that the initial conditions and the instant of switching are such
that immediately after closing of the switch the steady-state situation
is present. A good example is the switching of a lossless reactor when
the supply voltage is at the maximum. In practice, however, this rarely
happens, and after switching actions, transient oscillations originate in an
electrical network. Power systems have a high quality factor, that is, a large
L and small R and are designed such that the frequency of the transient
oscillations is much larger than the power frequency; this avoids steady-
state overvoltages because of resonance. However, for higher harmonic
frequencies generated by power electronic equipment, resonance can occur
more easily.
2
Transient Analysis
of Three-Phase Power Systems
Figure 2.1 A set of three unbalanced voltage phasors resolved in three sets of symmetrical
components
SEQUENCE COMPONENTS FOR UNBALANCED NETWORK IMPEDANCES 17
The relation between the set of phasors (Va , Vb , Vc ) and the positive
phasors, negative phasors, and zero phasors is
V 1 1 1 Va0
a
Vb = 1 a2 a Va1 (2.2)
Vc 1 a a2 Va2
or
Vabc = AV 012 (2.3)
The a-operator rotates any phasor quantity by 120° and the inverse
relation of Equation (2.2) can be written as
V 1 1 1 Va
a0
1
Va1 = 1 a a2 Vb (2.4)
3
Va2 1 a2 a Vc
or
V012 = A−1 Vabc (2.5)
Ia Zaa
Zab
Ib Zbb
Zca
Zbc
Ic Zcc
m + − n
Vmn
V Z Zab Zac Ia
mn−a aa
Vmn = Vmn−b = Zba Zbb Zbc Ib (2.7)
Vmn−c Zca Zcb Zcc Ic
with
Zs0 = 13 (Zaa + Zbb + Zcc )
Zs1 = 13 (Zaa + aZbb + a2 Zcc ) (2.11)
Zs2 = 13 (Zaa + a2 Zbb + aZcc )
and
Zm0 = 13 (Zbc + Zca + Zab )
Zm1 = 13 (Zbc + aZca + a2 Zab ) (2.12)
Zm2 = 13 (Zbc + a2 Zca + aZab )
Vmn−1 = (Zs1 − Zm1 )Ia0 + (Zs0 − Zm0 )Ia1 + (Zs2 + 2Zm2 )Ia2 (2.13)
The positive-sequence voltage drop depends not only on Ia1 but also on
Ia2 , and this means that there is a mutual coupling between the sequences.
Further, we can conclude that Zmn−012 is not symmetric; therefore, the
mutual effects are not reciprocal and this is a rather disturbing result.
This is the reason that we prefer to work with the special cases of both
self-impedances and mutual impedances in which the matrix Zmn−012 is
simplified. In many practical cases, the mutual impedances can be neglected
because they are small compared with the self-impedances. The matrix
Zmn−012 , however, is nonsymmetric with respect to Zs and Zm terms
and is therefore not made symmetric by eliminating either the self-terms
or the mutual terms, and because elimination of self-impedance terms
cannot be applied (because of the inherent nature of the power system),
a simplification must be sought in the special case of equal impedance
and symmetric impedance. In many practical power system problems, the
self-impedances or mutual impedances are equal in all the three phases. In
such cases, Equation (2.11) and Equation (2.12) become
and
Zm0 = Zbc , Zm1 = Zm2 = 0 (2.15)
20 TRANSIENT ANALYSIS OF THREE-PHASE POWER SYSTEMS
Ia 0 F0 Ia1 F1 Ia 2 F2
+ + +
Z0 + Z1 Z2
Va 0 Vf Va1 Va 2
−
− − −
N0 N1 N2
Figure 2.3 Sequence networks for the zero, positive, and negative sequences
22 TRANSIENT ANALYSIS OF THREE-PHASE POWER SYSTEMS
a F
+
b
+
c
+
Zf
Va Vb Vc Ia Ib = 0 Ic = 0
− − −
Figure 2.4 Three-phase circuit diagram of an SLG fault at fault point F of the power
system
Ib = Ic = 0 (2.20)
Va = Zf Ia (2.21)
These boundary equations are transformed from the a-b-c system to the
0-1-2 coordinate system:
This implies that all the sequence currents are equal, and when the
boundary equations for the voltage relations of Equation (2.21) are trans-
formed, we find the relation
The fact that the sequence currents are equal implies that the sequence
networks must be connected in series. From Equation (2.24), we note
that the sequence voltages add to 3Zf Ia1 – this requires the addition of
24 TRANSIENT ANALYSIS OF THREE-PHASE POWER SYSTEMS
F1 Ia1
+
+ Z1
Vf Va1
−
−
N1
F2
Z2 +
Va 2 3 Zf
−
N2
F0
Z0 +
Va 0
−
N0
and now that the sequence current relations are known, we can determine
the sequence voltage relations from Equation (2.24).
a F
+
b
+
c
+
Zf Zf Zf
Va Vb Vc Ia Ib Ic
Zg
Ia + Ib + Ic
− − −
Figure 2.6 Three-phase circuit diagram of a TPG fault at fault point F of the power
system
Va = Zf Ia + Zg (Ia + Ib + Ic ) (2.26)
Vb = Zf Ib + Zg (Ia + Ib + Ic ) (2.27)
Vc = Zf Ic + Zg (Ia + Ib + Ic ) (2.28)
The boundary conditions are again transformed from the a-b-c system
to the 0-1-2 coordinate system, and when we write in terms of the
symmetrical components of phase a, we get
F1 Ia1 F2 Ia 2 F0 Ia 0
+ Z1 + Z2 + Z0 +
Vf Va1 Zf Va 2 Zf Va 0 Z f + 3Z g
− − − −
N1 N2 N0
This leads to Ia1 = Ia , Ia0 = Ia2 = 0. The sequence networks are therefore
connected as shown in Figure 2.7. When the fault impedance Zf is small,
or even zero, the TPG fault is in fact a short circuit.
After clearing a three-phase fault, the power system changes from the
steady-state situation, in which the three-phase short-circuit current is
flowing, to the state in which only the power frequency–recovery voltage
is present across the contacts of the circuit breaker. In an inductive circuit,
the change from one steady state to another is always accompanied by a
transient – the transient recovery voltage or TRV. At current zero, the arc
voltage and the arc current extinguish and the TRV oscillates from zero
to the crest of the AC power frequency–recovery voltage (See Chapter 4,
Circuit Breakers). One of the breaker poles clears first and the two last
clearing poles interrupt 90° later – they, in fact, form a single-phase circuit.
Of interest is the value of the AC power frequency–recovery voltage of
the first clearing phase because the TRV oscillates to this value. Let us
consider the situation depicted in Figure 2.8.
The system is grounded by means of a neutral impedance at the star
point of the three-phase delta/wye transformer. When the first pole of the
circuit breaker interrupts the short-circuit current, let us assume that this
is phase a, the other two poles are still arcing and therefore in a conducting
state. In fact, these two last clearing poles interrupt a double line-to-ground
Power
Generator transformer
Transmission line Circuit breaker
Zn TPG
(DLG) fault. The boundary conditions for this DLG fault are
Ia = 0
Vb = (Zf + Zg )Ib + Zg Ic (2.33)
Vc = (Zf + Zg )Ic + Zg Ib
These boundary conditions are transformed from the a-b-c system to the
0-1-2 coordinate system and this results in
Vb − Vc = Zf (Ib − Ic ) (2.38)
or
Va1 − Zf Ia1 = Va2 − Zf Ia2 (2.41)
Because Equation (2.42) and Equation (2.46) are equal, we can collect
terms and write
2Va0 − 2Zf Ia0 − 4Zg Ia0 = Va1 + Va2 − Zf (Ia1 + Ia2 ) − 2Zg (Ia1 + Ia2 )
(2.47)
By using Equation (2.41) and the property that Ia1 + Ia2 = −Ia0 , we find,
after rearranging, that
Vf
Ia1 = (2.49)
(Z2 + Zf ) Z0 + Zf + 3Zg
Z1 + Zf +
Z0 + Z2 + 2Zf + 3Zg
F1 Ia1 F2 Ia 2 F0 Ia 0
+ + + +
Z1 Zf Z2 Zf Z0 Z f + 3Z g
Vf Va1 Va 2 Va 0
− − − −
N1 N2 N0
and
Va1 = Vf − Z1 Ia1 (2.50)
Z2 Z0 /(Z2 + Z0 )
Va = 3Va1 = 3Vf (2.51)
Z2 Z0
Z1 +
(Z2 + Z0 )
The ratio between the voltage across the first clearing pole and the phase
voltage of the undistorted power system is called the first-pole-to-clear
(FPTC) factor
Va Z2 Z0
FPTC = =3 (2.52)
Vf Z1 (Z2 + Z0 ) + Z2 Z0
with an isolated neutral, this is two times the FPTC factor, which is three
times the phase voltage.
For solidly grounded systems, Zn = 0, and the expression for the FPTC
factor becomes
3X0
FPTC = (2.54)
X + 2X0
When X has a larger value then X0 , the FPTC factor is smaller than one.
For cases in which X = X0 , the FPTC factor for solidly grounded systems
is one, and in an undamped situation, the peak of the TRV for the first
clearing pole is twice the crest value of the system phase voltage.
3
Travelling Waves
Power systems are large and complex systems, but for steady-state anal-
ysis, the wavelength of the sinusoidal currents and voltages is still large
compared with the physical dimensions of the network – for 50-Hz power
frequency, the wavelength is 6000 km. For steady-state analysis, a lumped-
element representation is adequate for most cases. However, for transient
analysis, this is no longer the case and the travel time of the electromag-
netic waves has to be taken into account. A lumped representation of, for
instance, an overhead transmission line by means of pi-sections does not
account for the travel time of the electromagnetic waves, as can be easily
seen from Figure 3.1.
When we make a representation of an overhead transmission line or
a high-voltage cable by means of a number of pi-sections, we take the
properties of the electric field in a capacitance and the properties of the
magnetic field in an inductance into account and connect these elements
with lossless wires.
When the switch S is closed, a current flows through the first inductance
L1 and charges the first capacitance C1 . The accumulation of charge on
C1 creates a voltage that causes a current to flow through L2 . This current
charges C2 , a voltage builds up across C2 , a current flows through L3
and so on. This kind of reasoning shows that a disturbance at one end of
the pi-section network is immediately noticeable at the other end of the
network. We know from experience that this is not what happens when a
source is switched on in a transmission line; it takes a certain time before
the current and voltage waves reach the end of the line.
A representation of overhead lines and underground cables by means
of lumped elements is not helpful in making us understand the wave
phenomena because electromagnetic waves have a travel time. Only when
the physical dimensions of a certain part of the power system are small
compared with the wavelength of the transients, the travel time of the
32 TRAVELLING WAVES
u Load
S L1 L2 Ln
u C1 C2 Cn Load
H H
n n
i i
+ + L /m, C /m
S
u u
E u E
− −
0 0
∆x ∆x x
Figure 3.2 Electric and magnetic field around a line segment x of a two-wire transmis-
sion line
Q Cux dx
i = limx→0 = limx→0 = Cu = Cuν (3.1)
t t dt
Because the distance x is covered in the time t, x/t is the velocity at
which the charge travels along the line. The magnetic flux present around
the line segment is = Lxi. If this is substituted in Equation (3.1) and
x is made infinitely small, the expression for the induced electromotive
force emf in the loop enclosed by the two wires over the distance x is
2
dx
emf = limx→0 = LCu = LCuν 2 (3.2)
t dt
1
ν=√ (3.3)
LC
The wave velocity depends only on the geometry of the line and on the
permittivity and the permeability of the surrounding medium.
For a 150-kV overhead transmission line with one conductor per phase,
a conductor radius of r = 25 mm and an equivalent distance between
phases of dm = 5.5 m, the inductance and the capacitance values are
µ
dm
o
L= ln = 1.13 mH/km (3.4)
2π 0.779r
34 TRAVELLING WAVES
2π ε0
C= = 10.3 nF/km (3.5)
dm
ln
r
Cu
i = Cuv = √ (3.6)
LC
We notice that the ratio between the voltage and current wave has a fixed
value
u L
= (3.7)
i C
wu = 12 Cxu(x0 , t0 )2 (3.8)
ENERGY CONTENTS OF TRAVELLING WAVES 35
H
n
i
S + L /m, C /m
u E u
−
0
x0
∆x x
The magnetic energy contents of the loop enclosed by the two wires over
the distance x at the time t0 is
wi = 12 Lxi(x0 , t0 )2 (3.9)
The energy passing through x per second is the power P of the electro-
magnetic wave:
wt wt 1 u(x0 , t0 )2
P= = = Cu(x0 , t0 )2 √ =
t (x/v) LC Z
(3.11)
1
2
= Li(x0 , t0 ) √ = i(x0 , t0 )2 Z
LC
− −
∆x
and because of the relation u(x, t) = Zi(x, t) that exists between the
voltage and the current wave, we can write for the voltage wave
When the voltage and current waves travel along a transmission line with
losses, the amplitude of the waves is exponentially decreased. This is called
attenuation and is caused by the properties of the transmission line.
For overhead transmission lines, G is a very small number and we
can simplify Equation (3.15) for the current and Equation (3.16) for the
voltage wave to
i(x, t) = i(x0 , t0 )e−(Rx/2Z) (3.17)
and
u(x, t) = u(x0 , t0 )e−(Rx/2Z) (3.18)
The attenuation is small for a line with a low resistance and/or a large
characteristic impedance.
38 TRAVELLING WAVES
and
d(t) = i(x, t)Lx (3.20)
Applying Kirchhoff’s voltage law on the loop enclosed by the two wires
over the distance x, we obtain
∂
u(x, t) − u(x + x, t) = −u = i(x, t)Rx + d(t)
∂t
(3.21)
∂
= R+L i(x, t) x
∂t
THE TELEGRAPH EQUATIONS 39
∂u(x, t) ∂i(x, t)
= −L − Ri(x, t) (3.22)
∂x ∂t
Similarly, for the current flowing through G and the current charging C,
Kirchhoff’s current law can be applied:
∂
i(x, t) − i(x + x, t) = −i = u(x, t)Gx + dψ(t)
∂t
∂
= G+C u(x, t)x (3.23)
∂t
∂i(x, t) ∂u(x, t)
= −C − Gu(x, t) (3.24)
∂x ∂t
The negative sign in these equations is caused by the fact that when the
current and voltage waves propagate in the positive x-direction i(x, t)
and u(x, t) will decrease in amplitude for increasing x. To solve the
equations, they are transformed into the Laplace domain by substituting
the Heaviside operator p = ∂/∂t; this leaves us with the partial differential
equations
∂u(x, p)
− = (R + pL)i(x, p) (3.25)
∂x
∂i(x, p)
− = (G + pC)u(x, p) (3.26)
∂x
∂ 2 u(x, p) ∂i(x, p)
2
= −Z = Z Y u(x, p) = γ 2 u(x, p) (3.27)
∂x ∂x
∂ 2 i(x, p) ∂u(x, p)
2
= −Y = Y Z i(x, p) = γ 2 i(x, p) (3.28)
∂x ∂x
1
γ = (RG + (RC + GL)p + LCp2 ) = (p + α)2 − β 2 (3.29)
ν
40 TRAVELLING WAVES
In this expression,
and
1 γx
i(x, t) = − e f1 (t) − e−γ x f2 (t) (3.35)
Z
For the lossless line, the series resistance R and the parallel conduc-
tance G are zero, and the propagation constant and the characteristic
impedance are √
γ = p LC = p/ν (3.36)
and
L
Z = Z0 = (3.37)
C
i(x, t) = − f1 t + − f2 t − (3.43)
Z0 v v
Forward waves
u (x, t )
n
i (x, t ) i (x, t )
+
u (x, t ) = Z 0 i (x, t )
−
Backward waves
u (x, t ) i (x, t )
+
i (x, t )
u (x, t ) = −Z 0 i (x, t )
−
1 αx/v x x
voltage and current waves. But what happens if the wave arrives at a
discontinuity, such as an open circuit or a short-circuit, or at a point
on the line where the characteristic impedance Z = (L/C)1/2 changes, for
instance, when an overhead transmission line is terminated with a cable
or a transformer? Because of the mismatch in characteristic impedance,
an adjustment of the voltage and current waves must occur. At the
discontinuity, a part of the energy is let through and a part of the energy
is reflected and travels back. At the discontinuity, the voltage and current
waves are continuous. In addition, the total amount of energy in the
electromagnetic wave remains constant, if losses are neglected. Figure 3.6
shows the case in which an overhead transmission line is terminated with
an underground cable.
When, for the sake of simplicity, both the overhead line and the
underground cable are assumed to be without losses, R = 0 and G = 0,
then the expressions for the characteristic impedances are
LL
ZL =
CL
and
LC
ZC = (3.46)
CC
The forward wave is called the incident wave and it travels on the overhead
line toward the cable. The positive x-direction in Figure 3.6 is from left
Discontinuity
at x 0
Overhead transmission line
to right and the line-cable joint is at x0 . The incident wave reaches the
discontinuity x0 at t = t0 . In our equations, the incident waves have
subscript 1, the reflected waves have subscript 2, and the waves that are
let through have subscript 3. Voltage and current waves are continuous at
the line–cable joint:
νL (x0 , t) = νC (x0 , t)
and
iL (x0 , t) = iC (x0 , t) (3.47)
From Equation (3.48) and Equation (3.49), for the voltage and current
waves that are let through
2ZC
ν3C (x0 , t) = ν1L (x0 , t) (3.50)
ZL + ZC
2ZL
i3C (x0 , t) = i1L (x0 , t) (3.51)
ZL + ZC
the substation, the opposite occurs – a magnified voltage wave leaves the
substation and travels over the overhead line.
When the incident wave encounters a short-circuit at x0 , which means,
in our example, that ZC = 0, the voltage wave disappears at x0 and
the current wave doubles in amplitude; the wave energy is stored in the
magnetic field only. When the incident wave encounters an open circuit
at x0 , the voltage wave is doubled in amplitude and the incident current
wave is nullified by the reflected current wave; the wave energy is stored
in the electric field only. The voltage doubling that occurs when a voltage
source is switched on an unloaded cable or an unloaded overhead line is
called the Ferranti-effect, named after the British scientist and engineer
Sebastiano Ziani de Ferranti (1864–1930).
When a transmission line is terminated with a load impedance ZL ,
different from the lines characteristic impedance Z0 , the ratio of the
complex amplitudes of the reflected and incident voltage waves at the load
is called the voltage reflection coefficient
ZL − Z0
r= (3.54)
ZL + Z0
HV
LV
r Core
Tank
Winding
w
d h
large compared with its diameter (h 2r); the radius of the winding is
large compared to the gap w between winding and core (r w); and
the gap w between winding and core on itself is large compared with
the thickness t of the winding(w t). The distance between tank and
winding is d, and both the tank and transformer core are assumed to be
ideal for conducting. The core and the transformer tank are grounded.
The boundary conditions provide constraints on the components of the
field vectors as they transit across the boundary between two media.
The tangential components of the electric field intensity vector E and the
magnetic field intensity vector H must be continuous across the boundary
between two physical media – medium 1 and medium 2
The core and the tank wall are assumed to be perfectly conductive; this is
characterised by an infinite conductivity σ = ∞. The infinite conductivity
makes all the fields in the perfect conductor to be zero. This requires
that the tangential component of E1 of medium 1 adjacent to medium
2 (the perfect conductor) must be zero at the boundary. In addition, the
normal component of B1 must be zero at the boundary. The tangential
component of H1 and the normal component of D1 cannot be zero;
otherwise, the resulting fields in medium 1 would be overspecified. The
tangential component of the magnetic field in medium 1, H1 , equals the
surface current density on the interface, and the normal component of
the electric flux D1 equals the surface charge density on the interface
(Figure 3.8). This simplifies the mathematical calculations. When the right
contour and surface is chosen, the capacitance and inductance of the
winding can be calculated and this result can be used to calculate the
characteristic impedance.
When a constant voltage V is present between the winding terminal and
the ground, there is a linear voltage profile along the coil that results in an
inner electric field Einner between coil and core and an outer electric field
Eouter between coil and tank. Gauss’ law for the electric field says that
Medium 1 x y
∈, µ, σ Medium 2
perfect
Et 2 = 0 conductor
Et 1 = 0 σ=∞
E1
B1
Bn 2 = 0
Bn1 = 0
Ht 1 Ht 2 = 0
Ks
+
Dn 1 + Dn 2 = 0
+
+
Figure 3.8 Illustration of the boundary conditions in which one medium is a perfect
conductor
48 TRAVELLING WAVES
the net flux of the electric flux density vector out of a closed surface S is
equivalent to the net positive charge enclosed by that surface. Written in
integral form
Dn · dS = ρ (3.57)
S
When a constant current i flows through the perfectly conductive coil, the
surface current density exists on the inside and outside of the coil, which
is orthogonal to the tangential component of the magnetic field vector H
in the medium between the coil and the core and the medium between the
coil and the tank. The magnetic flux inside the coil and outside the coil
are equal in magnitude but have an opposite sign. The total magnetic flux
in the medium between the coil and the core and the coil and the tank can
be calculated with Ampere’s law
d
H · dl = J · dS + D · dS (3.60)
C S dt S
Because we assume that the current does not change in time, the displace-
ment current is zero, and the magnetic flux is determined by the total
conduction current that penetrates the surface S bounded by the contour
C. When N is the number of windings of the coil, the total magnetic flux
is
4π µNi2π r
= (3.61)
h h
+
w d
the coil is
N 2π r wd
L= = 4π µ N2 (3.62)
i h w+d
Line
Line
T Line
G
CB
Three-Phase
ungrounded fault
Z1 Z=∞
Z1
Z1
Z2 Z=∞
L
Z2
G L
Z2
L uR
+
uS uL
−
Ce
Ce
Ce
uS
Z1 Ce L
uR
Figure 3.11 Simplified network representation seen from the breaker terminals
52 TRAVELLING WAVES
uR 3Z1/2 3L / 2
3 √
uR (t) = L 2Iω 1 − e−Z1 t/L (3.64)
2
u
R /L response
Voltage in per unit
1.0
0.8
0.6
0.4
0.2
0
0 t
Voltage
Time
Figure 3.14 Generalised shape of the TRV formed by local oscillations and reflected
travelling waves
Unloaded
Supply distribution
transformer transformer
L = 3000 m L = 100 m
CB Line Cable CB
When the circuit breaker closes at t = 0, the supply voltage has its
maximum value of one per unit and a voltage wave, with an amplitude
of one per unit, travels along the overhead line. As the voltage wave
reaches the discontinuity where the overhead line is connected to the
cable, the incident wave breaks up into a reflected wave with an amplitude
of (40 − 400)/(40 + 400) = −0.818 per unit and a transmitted wave with
an amplitude of 2 × 40/(40 + 400) = 0.182 per unit. The transmitted
1
1.0
00 a end = 1
5
µs
b end = 2
0.1
10 818
82
−0. 11
0.1
49
31
0.3 13
0.1
1 22
7
0.2 15
0.1
00
22
0.2 17
0.0
82
82
0.1
0.0
20 49
67
0.1
0.0
55
22
0.8 0.1
18 0.0
45
00
−0. 0.1
331 0.0
37
82
−0.
271 0.0
0.0
67 30
0.0
30 74
0.1 0.149 + 0.025
3
5 0.0
8
−0.060 + 0.142
.05
+0
6 69
−0. 6 35
.31
+0
71
0.2 9
.14
+0 X0
22
0.2
40
X
Figure 3.16 Lattice diagram showing the reflections and the refractions of the travelling
waves of the circuit of Figure 3.15. The overhead line is terminated with an underground
cable at X0
THE LATTICE DIAGRAM 55
wave propagates along the cable toward the unloaded transformer. The
unloaded transformer has a very high characteristic impedance compared
to the cable and can thus be treated as an open end; the voltage is doubled
because the reflected wave adds up to the incident voltage wave. The
voltage wave travelling from the unloaded transformer back along the
cable on its turn breaks up at the discontinuity where the overhead line
and the cable are connected into a reflected wave that bounces back to the
unloaded transformer, and a transmitted wave travelling to the source.
This process goes on and on. Because the wave phenomena play their
game on the microseconds timescale, the source voltage can be assumed
to remain constant during the process of reflection and refraction of the
electromagnetic waves. If we make a diagram and put the travel time
along the vertical axis and the distance along the horizontal axis, we
get the so-called lattice diagram (Figure 3.16). In this lattice diagram, we
do the bookkeeping of the reflections and the breaking of the travelling
waves. When we start with an incident wave of one per unit, the refrac-
tion and reflection constants at the discontinuities for the voltage waves
(Equation (3.50) and Equation (3.52)) are also placed along the horizontal
axis to determine the amplitude of the voltage wave at a certain place
along the line and the cable, being the superposition of the amplitudes of
the waves having arrived at that place at that instant.
The amplitudes of the voltages at the supply transformer Ustart , the place
where the line and the cable are connected Udis , and at the terminals of the
2.5
Uend
Udis
2.0
1.818
1.5
U [per unit]
1.0 Ustart
0.5
0
0 10 20 30 40 [µs]
t
Figure 3.17 Shape of the TRV as an addition of the reflections of the travelling waves
56 TRAVELLING WAVES
unloaded transformer Uend can be constructed with the help of the lattice
diagram of Figure 3.16, and the voltage curves are drawn in Figure 3.17.
We see from Figure 3.17 that the voltages are discontinuous in time. In
the shape of the voltage at the receiving end, we recognise the response of
a lumped RC network. This is logical because the distributed capacitance
of the cable is charged and this takes some time. When the circuit breaker
closes, the voltage at the terminals of the unloaded transformer reaches its
steady-state value gradually until the cable is charged.
4
Circuit Breakers
For the analysis of simple switching transients and for carrying out
large system studies, it is often sufficient to model a circuit breaker
as an ideal switch. When studying arc–circuit interaction, wherein, the
influence of the electric arc on the system elements is of importance,
a thorough knowledge about the physical processes between the circuit
breaker contacts is absolutely necessary.
A high-voltage circuit breaker is an indispensable piece of equipment in
the power system. The main task of a circuit breaker is to interrupt fault
currents and to isolate faulted parts of the system. Besides short-circuit
currents, a circuit breaker must also be able to interrupt a wide variety
of other currents at system voltage such as capacitive currents, small
inductive currents, and load currents. We require the following from a
circuit breaker:
• In closed position it is a good conductor;
The electric arc is, except from power semiconductors, the only known
element that is able to change from a conducting to a nonconducting state
in a short period of time. In high-voltage circuit breakers, the electric arc
is a high-pressure arc burning in oil, air, or sulphur hexafluoride (SF6 ).
In medium-voltage breakers more often, the low-pressure arc burning in
vacuum is applied to interrupt the current. The current interruption is
performed by cooling the arc plasma so that the electric arc, which is
58 CIRCUIT BREAKERS
arc is a plasma channel between the breaker contacts formed after a gas
discharge in the extinguishing medium. When a current flows through
a circuit breaker and the contacts of the breaker part, driven by the
mechanism, the magnetic energy stored in the inductances of the power
system forces the current to flow. Just before contact separation, the
breaker contacts touch each other at a very small surface area and
the resulting high current density makes the contact material to melt.
The melting contact material virtually explodes and this leads to a gas
discharge in the surrounding medium either air, oil, or SF6 .
When the molecular kinetic energy exceeds the combination energy,
matter changes from a solid state into a liquid state. When more energy
is added by an increase in temperature and the Van der Waals forces
are overcome, matter changes from a liquid state into a gaseous state. A
further increase in temperature gives the individual molecules so much
energy that they dissociate into separate atoms, and if the energy level is
increased even further, orbital electrons of the atoms dissociate into free
moving electrons, leaving positive ions. This is called the plasma state.
Because of the free electrons and the heavier positive ions in the high-
temperature plasma channel, the plasma channel is highly conducting and
the current continues to flow after contact separation.
Nitrogen, the main component of air, dissociates into separate atoms
(N2 → 2N) at approximately 5000 K and ionises (N → N+ + e) above
8000 K. SF6 dissociates into sulphur atoms and fluorine atoms at approx-
imately 1800 K and ionises at temperatures between 5000 and 6000 K.
For higher temperatures, the conductivity increases rapidly. The thermal
ionisation, as a result of the high temperatures in the electric arc, is caused
by collisions between the fast-moving electrons and photons, the slower-
moving positively charged ions and the neutral atoms. At the same time,
there is also a recombination process when electrons and positively charged
ions recombine to a neutral atom. When there is a thermal equilibrium,
the rate of ionisation is in balance with the rate of recombination.
The relation between the gas pressure P, the temperature T, and the
fraction of the atoms that is ionised f is given by Saha’s equation
f2
P = 3.16 ∗ 10−7 T 5/2 e−eVi /kT (4.1)
1 − f2
1.0
Degree of ionisation f
0.8
Cu H
0.6 Hg N
O
0.4
0.2
0
0 4 12 20 28 x103
Temperature in K
Figure 4.1 Degree of thermal ionisation for some metal vapours and atomic gases
Contraction Contraction
zone zone
Ie (Electrons)
Cathode Arc column Anode
I+ (ions)
Figure 4.2 The arc channel can be divided into an arc column, a cathode, and an anode
region
Voltage
u anode
u column
uarc
u cathode
Gap length x
Nozzle
Moving
contact
Pot
air-blast circuit breakers, the contacts were mounted inside a tank filled
with SF6 gas, and during the current interruption process, the arc was
cooled by compressed SF6 gas from a separate reservoir. The liquefying
temperature of SF6 gas depends on the pressure but lies in the range of
the ambient temperature of the breaker. This means that the SF6 reservoir
should be equipped with a heating element that introduces an extra failure
possibility for the circuit breaker; when the heating element does not
work, the breaker cannot operate.
Therefore the puffer circuit breaker was developed and the so-called
double pressure breaker disappeared from the market. In the puffer circuit
breaker (see Figure 4.5), the opening stroke made by the moving contacts
moves a piston, compressing the gas in the puffer chamber and thus
causing an axial gas flow along the arc channel. The nozzle must be able
to withstand the high temperatures without deterioration and is made
from teflon.
Presently, the SF6 puffer circuit breaker is the breaker type used for the
interruption of the highest short-circuit powers, up to 550 kV–63 kA per
interrupter made by Toshiba.
Puffer circuit breakers require a rather strong operating mechanism
because the SF6 gas has to be compressed. When interrupting large
currents, for instance, in the case of a three-phase fault, the opening speed
of the circuit breaker has a tendency to slow down by the thermally gener-
ated pressure, and the mechanism (often hydraulic or spring mechanisms)
should have enough energy to keep the contacts moving apart. Strong
and reliable operating mechanisms are costly and form a substantial part
of the price of a breaker. For the lower-voltage range, self-blast circuit
breakers are now on the market. Self-blast breakers use the thermal energy
released by the arc to heat the gas and to increase its pressure. After the
Fixed
contact
Moving
contact
Puffer chamber
moving contacts are out of the arcing chamber, the heated gas is released
along the arc to cool it down. The interruption of small currents can be
critical because the developed arcing energy is in that case modest, and
sometimes a small puffer is added to assist in the interrupting process.
In other designs, a coil carrying the current to be interrupted, creates a
magnetic field, which provides a driving force that rotates the arc around
the contacts and thus provides additional cooling. This design is called
the rotating-arc circuit breaker. Both self-blast breakers and rotating
arc breakers can be designed with less powerful (and therefore cheaper)
mechanisms and are of a more compact design than puffer breakers.
Current
Sealed
enclosure
Figure 4.6 Vacuum interrupter with slits in the contacts to bring the arc in a spiralling
motion
VACUUM CIRCUIT BREAKERS 67
Figure 4.7 The use of horse shoe magnets as is done in the HOLEC interrupters (courtesy
of Holec)
• Physical models
• Parameter models
The circuit breaker design engineers work mostly with physical arc models
when designing a new prototype. Physical arc models are based on the
equations of fluid dynamics and obey the laws of thermodynamics in
combination with Maxwell’s equations. They consist of a large number
of differential equations. The arc-plasma is a chemical reaction and, in
addition to the conservation of mass equation, describes the rate equations
of the different chemical reactions. In the case of a local thermodynamic
equilibrium, the rate equations become the equilibrium mass action laws
and that, in the simplified case of the reaction of a monatomic gas, becomes
the Saha equation, describing the degree of ionisation in the gas. Because
the arc-plasma is electrically conducting in the momentum equation also,
terms describing the interaction with magnetic fields, either coming from
outside or generated by the arc current, are taken into account. Because
of the resistive heat dissipation of electric energy calculated with Ohm’s
law, a volumetric heat generation term is part of the energy equation.
MODELLING OF THE SWITCHING ARC 69
∂ρ
+ div (ρu) = 0 (4.2)
∂t
∂u
ρ + ρ(u.grad)u = −grad(p) (4.3)
∂t
Conservation of energy
∂h
ρ σ E2 = div (ρu) + div [K.grad(T)] − R[T, ρ]
+ u.grad(ρh) −
∂t
energy Joule work thermal radiation
change heating performed loss
input by conduction
of energy by flow
mass flow loss
in unit
convection
volume
(4.4)
This last equation is very useful when a qualitative evaluation of the effects
of current increase or arc-channel-diameter reduction is required.
In Black box models, the arc is described by a simple mathematical
equation and gives the relation between the arc conductance and measur-
able parameters such as arc voltage and arc current. These black box
models are not suited to design circuit breaker interrupters but are very
useful to simulate arc-circuit interaction in network studies. Black box
models are based on physical considerations but are, in fact, mathematical
models; the behaviour of the arc rather than the physical processes is of
importance. Usually, black box models consist of one or two differential
equations. Parameter models are a variation on black box models in the
sense that more complex functions and tables are used for the essential
parameters of the black box models.
The classical black box models are the Cassie model and the Mayr
model. Both the Cassie and Mayr equation are a solution of the general
arc equation.
In its general form, the arc conductance is a function of the power
supplied to the plasma channel, the power transported from the plasma
channel by cooling and radiation and time:
iarc 1
g = F(Pin , Pout , t) = = (4.6)
uarc R
with
g = the momentary arc conductance
Pin = the power supplied to the plasma channel
Pout = the power transported from the plasma channel
t = time
iarc = the momentary arc current
uarc = the momentary arc voltage
R = the momentary arc channel resistance
The momentary arc conductance g varies when Pin and Pout are not in
equilibrium. The energy stored in the plasma channel is
t
Q= (Pin − Pout ) dt (4.7)
0
MODELLING OF THE SWITCHING ARC 71
dg dF(Q) dQ
= (4.9)
dt dQ dt
1 dg 1 dF(Q) dQ
= (4.10)
g dt g dQ dt
d[ln(g)] F (Q)
= (Pin − Pout ) (4.11)
dt F(Q)
Q = DQ0 (4.13)
Q
g = F(Q) = g0 (4.14)
Q0
g0
F (Q) = (4.15)
Q0
Q
Pout = DP0 = P0 (4.16)
Q0
Equations (4.14), (4.15), and (4.16) substituted in the general arc equation
(Equation (4.11)) gives us the Cassie equation
d[ln(g)] P0 u2arc
= −1 (4.17)
dt Q0 u20
The quotient Q0 /P0 is called the arc time constant τ and can be calculated
from the homogeneous differential Equation (4.17)
d[ln(g)] P0
=− (4.18)
dt Q0
g = g0 e−t/τ (4.19)
The time constant τ in Equation (4.19) can be interpreted as the arc time
constant parameter with which the arc channel diameter changes.
The Cassie model is well suited for studying the behaviour of the arc
conductance in the high-current time interval when the plasma temperature
is 8000 K or more.
The Mayr model describes the arc conductance around current zero.
Mayr considered the arc channel to be cylindrical with a constant diam-
eter. The arc column loses its energy by radial heat transport, and the
MODELLING OF THE SWITCHING ARC 73
temperature of the arc channel varies more or less exponentially with the
temperature and can be expressed as an approximation of Saha’s equation
(Equation (4.1)):
g = F(Q) = keQ/Q0 (4.20)
At the instant of current zero, the power input uarc iarc in the arc channel
is zero, and the rate of change of the conductance of the arc channel is
dg P0
= −g (4.22)
dt Q0
In this expression Q0 /P0 is the time constant τ of the arc cooling without
thermal input to the arc channel and is called the arc time constant.
The Mayr model is suited for modelling of the arc in the vicinity of
current zero when the temperature of the plasma is below 8000 K.
In 1959, T. E. Browne proposed a composition of the Cassie and Mayr
model. The Browne model uses a Cassie equation for the high current
interval and a Mayr equation for the current zero period.
Other black box models capable of simulating thermal breakdown
of the arc channel are the Avdonin model, the Hochrainer model, the
Kopplin model, the Schwarz model, and the Urbanek model. In the 1980s
W. R. Rutgers developed the KEMA model based on experiments in
KEMA’s research laboratory and in KEMA’s high-power laboratory. The
KEMA model is also capable of simulating a dielectric breakdown of the
gaseous space between the breaker contacts after current interruption. (See
Section 4.7 Arc–circuit interaction) In a hybrid arc model, before current
zero, the conductance of the arc channel is calculated with a modified
Mayr-equation as proposed by Haupt, and in the thermal and dielectric
interval around and after current zero, the arc conductance is calculated
from the concentration of charged particles and their drift velocity.
74 CIRCUIT BREAKERS
L iL
iC iR
Figure 4.8 Simple lumped-element representation of the network connected to the breaker
terminals
du/dt 2000
Rp = = = 112 (4.25)
di/dt 17.8
this capacitance must be charged. The value of this time delay is also
specified in the IEC 60056, and for a rate of rise of 2000 V/µs connected
with the 100 percent duty the delay time is two µs. The value of the time
delay equals the RC-time of Cp and Rp , and in this way the value of the
capacitance Cp can be calculated.
The system voltage and the short-circuit current determine the value of
the inductance L. Figure 4.9 shows the voltage and current traces of the
current interruption process around current zero
In the high-current interval, milliseconds before current zero, the di/dt
flowing through inductance L is
diL (u − uarc )
= (4.26)
dt L
iL
Current, voltage
~ dU
dt
U arc ∆t R
t
UL ∆t C
i arc
~ di L
dt
∆i R
iL
t=0
Figure 4.9 Current and voltage around current zero when a high-voltage circuit breaker
interrupts a short-circuit current
ARC–CIRCUIT INTERACTION 77
iR u0 /Rp
tR = = (4.27)
diL diL /dt
dt
A few microseconds before current zero the arc voltage goes to zero with
a steep duarc /dt and causes a capacitive current iC to flow through Cp .
This capacitive current iC lags the current iL with a time tC ;
iC Cp (duarc /dt)
tC = = (4.28)
diL diL /dt
dt
The presence of the resistor Rp and the capacitance Cp delay the current
zero of the linear decaying arc current with a time t
The parallel resistor, that is, the characteristic impedance of the connected
system, and the parallel capacitor achieve commutation of the arc current
in the time interval around current zero. The current is shunted, first by
Rp and then by Cp , the energy input in the arc is reduced, and the cooling
mechanism has a longer time period to be effective so that the interruption
chance has grown. When the current interruption is successful at current
zero the voltage across Cp is equal to zero and the transient recovery
voltage building up across the breaker contacts charges first capacitance
Cp , which causes the so-called time delay of the TRV waveform. When a
breaker manufacturer wants to increase the interrupting capability of an
interrupter design, he can connect a capacitor parallel to the interrupter in
the case of an SF6 circuit breaker or a resistor when it concerns an air-blast
breaker. For air-blast breakers, a capacitance does not work because of
the rather long time constant of physical processes in the arc channel.
The disadvantage of using a resistor is that an extra interrupter is needed
to switch off the resistor after interruption of the current by the main
interrupter.
A high-voltage circuit breaker must be able to interrupt a variety of
current values at different power factors. Because of the huge electric
power that has to be interrupted in the case of short-circuit currents, it
is understandable that fault-current interruption is an important topic.
78 CIRCUIT BREAKERS
i∆ R arc i∆
Source Cs Ct Load
(R − αRarc ) 1
λ2 + λ+ =0 (4.33)
L LC
Lee, T. H., Physics and Engineering of High Power Switching Devices, Chapters 6, 9,
MIT Press, Cambridge, USA, 1975.
Mayer, H., ‘‘High-voltage circuit breakers,’’ Electra (Jubilee issue), 95–106 (1972).
Mayr, O., ‘‘Beitraege zur Theorie des statischen und dynamischen Lichtbogens’’,
Archiv für Elektrotechnik 37(H12), S588–S608 (1943); Mayr, O., ‘‘Über die
Theorie des Lichtbogens und seine Loeschung’’, ETZ-A 64, S645–S652 (1943).
Möller, K., ‘‘Die Geschichte der Mayrschen Differentialgleichung des Dynamischen
Lichtbogens’’, Tech. Mitt. AEG-Telefunken, Jg. 62(7), S338–S341 (1972).
Nakanishi, K., ed., Switching Phenomena in High-Voltage Circuit Breakers,
Chapter 1, 2, 3, Marcel Dekker, New York, 1991.
Portela, C. M. et al., Circuit breaker behaviour in reactor switching. applicability
and limitations of the concept of chopping number,’’ IEEE T-PWRD 3(3),
1009–1021.
Ravindranath, B. and Chander, M., Power System Protection and Switchgear, Chap-
ters 1, 13, 14, 16, Wiley Eastern, New Delhi, 1977.
Reece, M. P., in C. H., Flurscheim, ed., Power Circuit Breakers: Theory and Design,
Chapters 1, 2, Institution of Electrical Engineers, 1982.
Rieder, W., ‘‘Arc-circuit-interaction near current zero and circuit-breaker testing,’’
IEEE International Symposium on High-Power Testing, 705–713 (1971).
Rieder, W., ‘‘Circuit breakers: physical and engineering problems, Part I: Fundamen-
tals,’’ IEEE Spectrum (July), 35–43 (1970); Part II: Design considerations, IEEE
Spectrum (August), 90–94 (1970); Part III: Arc-medium considerations, IEEE
Spectrum (September), 80–84 (1970).
Rieder, W. und Pratl, J., ‘‘Das Löschen und Wiederzünden von Schaltlichtbögen,’’
Scientia Electrica 11(2), S33–S48 (1965).
Rieder, W., Plasma und Lichtbogen, Chapters 5, 6, Vieweg & Sohn GmbH, Braun-
schweig, Germany, 1967.
Rieder, W., ‘‘The activities of the current zero club,’’ Electra 58, 25–32 (1992).
Reid, W. E., ‘‘Effect of transient recovery voltage (trv) on power circuit interruption’’
IEEE Power System Transient Recovery Voltages, Chapter 1 course # 87TH0176-
8-PWR (1987).
Rizk, F., ‘‘Arc instability and time constant in air-blast circuit breaker,’’ Report 107
CIGRE, 1964.
Ruedenberg, R.,Transient Performance of Electric Power Systems: Phenomena in
Lumped Networks, Chapter 39,, McGraw-Hill, New York, 1950.
Saha, M. N., ‘‘Ionisation in the solar chromosphere,’’ Philos. Mag. 40(472), (1920).
Schavemaker, P. H. and van der Sluis L., ‘‘The influence of the topology of test circuits
on the interrupting performance of circuit breakers,’’ IEEE Transactions on Power
Delivery 18(4), 1822–1828 (1995).
Schneider, J., ‘‘Beschaltungsmasznahmen zur Erhöhung des thermischen Ausschaltver-
mögens von Hochspannungsleistungsschaltern’’ ETZ Archiv 5(H5), S149–S153.
Simon, H. Th., ‘‘Über die Dynamik der Lichtbogenvorgänge und über Lichtbogen-
Hysteresis.’’, Physikalische Zeitschrift 6(10), S297–S319 (1905).
Slamecka, E., Pruefung von Hochspannungs-Leistungsschaltern, Part IV, Chapter 6,
Springer-Verlag, New York, 1966.
Slamecka, E. and Waterschek, W., Schaltvorgaenge in Hoch-und Niederspan-
nungsnetzen, Chapter 3, Siemens Aktiengesellschaft, Berlin, 1972.
Slepian, J., ‘‘Extiction of an A-C. arc,’’ AIEE Trans. 47, 1398–1408 (1928).
82 CIRCUIT BREAKERS
CIGRE Study Committee 13, WG 01, ‘‘Application of black box modelling to circuit
breakers,’’ Elektra 149, 41–71 (1993).
Sorensen, R. W. and Mendenhall, H. E. ’’ Vacuum switching experiments at California
institute of technology’’, AIEE Trans. 45, 1102–1105.
van der Sluis, L., Rutgers, W. R., and Koreman, C. G. A., ‘‘A physical arc model for
the simulation of current zero behaviour of high-voltage circuit breakers,’’ IEEE
T-PWRD 7(2), 1016–1022 (1992).
Transients in Power Systems
Lou van der Sluis
Copyright 2001 John Wiley & Sons Ltd
ISBNs: 0-471-48639-6 (Hardback); 0-470-84618-6 (Electronic)
5
Switching Transients
Ls ic L′
+
Rs
E cos (wt ) Cd C uc
Cs
−
2
e, i
ic
1
Time
Instant of
uc current
interruption
Figure 5.2 Current and voltage traces during the interruption of a capacitive current 1:
supply voltage, 2: voltage on the capacitor, and 3: voltage across the circuit breaker
1
f = √ (5.1)
2π Ls Cs
and the rate of rise of this voltage jump can be calculated with
du diC
= Zs (5.2)
dt dt
in which
Ls
Zs = (5.3)
Cs
and the value of the capacitance is 80 µF. The short-circuit power gives us
the value of the supply inductance:
√
Ls = {(20 kV/ 3)/31.5 kA}/100 π = 1.2 mH (5.6)
When the supply side capacitance Cs is 100 nF, the Ferranti rise is 9.2
percent and the frequency of the transient voltage jump is
1
f = √ = 14.7 kHz (5.7)
2π Ls Cs
1
f = √ (5.8)
2π Ls C
for
Ls L
INTERRUPTING CAPACITIVE CURRENTS 87
High-frequency
discharge current
2E
Voltage Ls
+E C
Supply-side
−E
voltage +3E
Current
+E
Load-side
voltage −E
−5E
+2E
Voltage
across the
breaker Instant of current Time
interruption
−4E
Figure 5.3 Re-ignition of the circuit breaker when interrupting a capacitive current
The smaller the value of Ls , the higher the frequency and larger the
amplitude of the transient current. When at the instant of re-ignition
the value of the voltage on the capacitor C was −E, the voltage is then
+3E at the first zero crossing of the transient current. When the arc
channel extinguishes, the circuit breaker interrupts the oscillating current.
The recovery voltage across the contact gap has now increased to 4E
(the contribution of the supply-side voltage is −E and of the trapped
charge on the capacitor C is +3E). The arcing contacts of the circuit
breaker have parted only a little further and another re-ignition is likely to
occur. The result of this second re-ignition of the extinguishing medium
is an oscillating current with a doubled amplitude, and when the breaker
interrupts the transient current at its first zero crossing again, the voltage
88 SWITCHING TRANSIENTS
Ls L′ L′
E cos(wt ) C C
the current are therefore much higher than in the case of single-capacitor
bank switching.
Switching unloaded high-voltage transmission lines in and out of service
is, in principle, the same as switching a capacitor bank because an unloaded
transmission line has a dominantly capacitive behaviour. In addition, a
small voltage jump occurs because of the Ferranti-rise effect. High-voltage
transmission lines also have travel times and must therefore be represented
by distributed elements instead of lumped elements. When a transmission
line is energised after closing of a switching device, the resulting voltage
wave reflects and causes doubling of the voltage at the open end of the line.
In addition, switching off an unloaded transmission line is, in principle,
similar to switching off a capacitor bank. The Ferranti rise causes a
transient voltage jump that results in a voltage wave travelling along the
line. The interrupting device is stressed by a voltage jump at the supply
side and an oscillation at the line side. The resulting steep voltage jump
across the contacts of the interrupting device can cause, because of the
short arcing time, a re-ignition and prolong the arcing time.
When we analyse three-phase switching of unloaded high-voltage
transmission lines, the capacitive coupling between the phases and the
capacitance to ground have to be taken into account (see Figure 5.5).
When the first phase (e.g. phase 1) has cleared the power-system, voltage
Vi of the neighbouring phases 2 and 3 is coupled into the line-side DC
voltage V1 with a coupling factor k = Vi /V1 . The coupling factor between
the neighbouring phases and phase 1 depends on the ratio of C21 /C1 earth
and C31 /C1 earth . When the second phase clears (let us assume that this is
3
3
C31
C32 C31
2 1 2 1
C21
(a) (b)
Figure 5.5 The coupling effect of the line capacitances when switching unloaded
high-voltage transmission lines (a) phase 1 has been switched off and phases 2 and 3
are still at voltage (b) phases 1 and 2 are switched off and phase 3 is still at voltage
90 SWITCHING TRANSIENTS
phase 2), then phase 3 is still at system voltage and this voltage couples
into the line-side DC voltages of phase 1 and phase 2. The value of the
coupling factor depends on the tower structure and circuit design but
usually has a value between 0.2 and 0.4. For a double circuit, when
two circuits are hanging in the same tower, the coupling factor can be
higher when the neighbouring circuit is in operation. Capacitive coupling
increases the TRV across the contacts of the interrupting device with the
coupling factor (Figure 5.6).
When unloaded high-voltage transmission lines are switched off, a
combination of the voltage jump at the supply side, the transient voltage
oscillation at the supply side, the voltage oscillation at the line side, and the
capacitive coupling with the neighbouring phases can result in a recovery
voltage for the first phase to clear, which can be as high as 2.8 per unit
(Figure 5.7).
When switching unloaded high-voltage cables only the capacitive load
current is interrupted. This current is small compared with the capacitive
current when switching capacitor banks, but large compared with the
capacitive current when switching unloaded high-voltage transmission
lines. The actual construction of the cable is of importance when we want
to study the transient recovery voltage in detail. For three-phase cables
with the three conductors in a lead sheet with an earth screen, the same
coupling effects as with transmission lines occur.
When each conductor has its own earth screen, only the capacitance
to earth plays a role and the resulting TRV has the same shape as the
e
Phase C
t1 t2 t3 Time
Phase B
Phase A
Figure 5.6 Line-side voltages after switching off an unloaded high-voltage transmission
line. The three phases interrupt the capacitive current at t1 , t2 , and t3
CAPACITIVE INRUSH CURRENTS 91
e, i
ic
1
e
t1
Time
uc
Figure 5.7 Current and voltage traces after switching off an unloaded high-voltage
transmission line: 1: voltage jump at the supply side; 2: transient voltage oscillation at the
supply side; 3: voltage across the contacts of the interrupting device; 4: power frequency
voltage coupled in from the neighbouring phase
E cos (wt )
Figure 5.8 Single-line diagram for a substation with a capacitor bank for voltage
regulation
92 SWITCHING TRANSIENTS
Ls
Rs
E cos (wt ) Cd C
Cs
is 4300 A. So the peak of the transient inrush current is more than ten
times the value of the peak of the steady-state capacitive current. These
rather high inrush currents can cause damage to the capacitors of the
capacitor bank. But the switching device can also be damaged. When
a dielectric breakdown occurs just before the contacts touch during the
closing operation (a so-called prestrike), the rather large inrush current can
cause the contact material to meld. Especially in cases in which vacuum
circuit breakers are used to switch the capacitor banks, the contacts might
weld together. A prestrike in a minimum-oil circuit breaker or in an SF6
circuit breaker can do other damages to the breaker as has already been
mentioned in Section 5.1. The suddenly created plasma channel causes a
shock wave and can make, in the case of a minimum-oil circuit breaker,
the complete interrupter to explode or, when SF6 is the extinguishing
medium, can sometimes damage the arcing chamber or nozzle.
Ls
+
Rs
E cos(wt ) Cd uc C L
Cs
−
Figure 5.10 Single-phase lumped element representation of a circuit for small inductive
currents
94 SWITCHING TRANSIENTS
The inductivity L of the load is dominant, which means that the load
makes the current to lag the voltage. The capacitance C of the load is
usually very small: a few nanoFarads for a distribution transformer and
a few picoFarads for an air-core reactor, depending on the design. For a
motor or an electric oven, connected to the supply by means of a cable, the
value of the capacitance C is determined by the length and type of the cable.
When the small inductive current is interrupted, commonly at a short
arcing time, the load capacitance C is charged at
uc = L di/dt (5.11)
which is a value close to the supply voltage, because L Ls . After current
interruption, C discharges itself through L by means of an oscillating
current with frequency
1
f1 = √ (5.12)
2π LC
The frequency of this oscillation can be from a few kiloHertz to several
kiloHertz and thus creates very steep du/dt s at the load side of the
switching device. It is likely that the steep increase of the TRV at the load
will cause a dielectric breakdown of the extinguishing medium between
the narrow contact gap, and a re-ignition occurs. After a re-ignition, the
inductive load is connected with the supply again and the capacitance C
of the load is charged by a current with frequency
1
f2 = √ (5.13)
2π Ls C
and a high-frequency current flows through the arc channel. When the
switching device interrupts this high-frequency current, capacitance C is
charged to a higher value because the di/dt of the re-ignition current
is considerably higher than the power frequency current, which was
interrupted first (Equation (5.11)). The capacitance C will discharge itself
again through the inductivity L. In the mean time, the arcing contacts
have moved further apart and have increased the capability to withstand
dielectric breakdown of the contact gap. However, the voltage stress
caused by the oscillation at the load side has increased as well and another
re-ignition might occur. When this process repeats itself a few times, a
considerable overvoltage can be the result.
When switching small capacitive currents or small inductive currents,
the unwanted result is in both cases the same: high overvoltages (and
possible damage to high-voltage equipment in the vicinity) or even a
flashover on the outside of the switching device. In the latter case, the
high-voltage circuit breaker or the load breaker switch is short-circuited
TRANSFORMER INRUSH CURRENTS 95
out of the system – a dangerous situation. Apart from the current being
leading or lagging, the principle difference between the two switching cases
is that the capacitance C at the load side, in the case of small inductive
current switching, does not remain charged but discharges itself through
the load L.
Bm
2
B, Φ
3
Br
Hm 4 Hc 1 Hc
7 Hm
H, I
5 Bm
10000
5000
Current
−5000
4
2
Flux
0
−2
−4
× 104
2
1
Voltage
0
−1
−2
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Figure 5.12 Transformer inrush current, the flux in the core, and the supply voltage
THE SHORT-LINE FAULT 97
Ls Transmission line
+
+ L line
Rs Z=
C line
E cos(wt ) Cd uL
− Cs
Supply-side voltage
e
E cos (wt )
u L = di ∗L line
dt
Time
Line-side voltage
Figure 5.14 Line-side, supply-side TRVs and the TRV across the breaker contacts
place is τline = 1/ν, where ν is the wave velocity fixed by the transmission
line parameters. After twice the travel time, the wave reflected by the short-
circuit point arrives at the open-line end near the breaker. This continuing
process of reflection results in the triangular-shaped line-side oscillation.
The first excursion of this line-side oscillation has a rate of rise, depending
on the interrupted current level and the distance of the short-circuit point
on the line, between 3 kV/µs and as much as 10 kV/µs. This very steep rate
of rise of the TRV at the line side of the breaker has its impact on the still
very hot gas mixture between the breaker contacts, after the arc plasma
has vanished and the current has been interrupted. In a hot gas, there
are always free electrons present as a result of the temperature and the
extensive radiation in the arcing interval. The rapidly increasing electric
field strength makes the free electrons drift, a so-called post-arc current
flows between the breaker contacts in the still hot arc channel. When the
cooling power of the interrupting process is such that the cooling prevails
over the energy input as a result of the post-arc current and the rapidly
increasing TRV, then a successful current interruption takes place. When
the energy input in the hot gas channel prevails, a thermal re-ignition is
the result and the circuit breaker has failed to interrupt the short-circuit
current.
For circuit breakers there exists a so-called critical line length. The crit-
ical line length depends on the type of extinguishing medium. The critical
line length, for instance, for SF6 circuit breakers is around 93 percent
and for air-blast breakers between 75 and 85 percent (see Figure 5.15).
When the standards speak from a certain percentage of line fault, this
means the reduction of nominal rated short-circuit current in the case
of a bolted terminal fault by the impedance of the line segment at the
load side of the breaker. When for example the 100 percent short-circuit
rating of a circuit breaker is 40 kA and the short-circuited line reduces the
short-circuit current till 36 kA, it is called a 90 percent short-line fault.
The percentage of short-line fault can therefore easily be calculated with
• When the line length increases (and the short-line fault percentage
decreases), the current to be interrupted gets smaller.
• Because the rms value of the current gets smaller, the di/dt becomes
lower
• The di/dt and also the du/dt of the line-side oscillation decreases
because du/dt = Zline di/dt.
τ = 0.4 µs
100
90
80
Cooling power P0 [ kW ]
70
60
50
80 84 88 92 96 100
Percentage short-line fault
Figure 5.15 Cooling power of an SF6 puffer circuit breaker, calculated with a Mayr-arc
model, required for a successful interruption at various short-line fault percentages
THE SHORT-LINE FAULT 101
ua + ub + uc = 0
(5.15)
ue = (Ua + Ub + Uc )/3
Three-phase
high-voltage
Ua transmission line
ua
ub
uc
Ub Uc
ue
Ground
Ua = ua + ue
Ub = ub + ue
Uc = uc + ue (5.16)
Za Za Za
Ua Ua Ua
Zb Zb Zb
Ub Ub Ub
Zc Zc Zc
Uc Uc Uc
Ze Ze Ze
0 0 0
Figure 5.17 Phase and earth components of the characteristic impedances. (a) general;
(b) three-phase fault, first phase to clear; (c) three-phase fault, last phase to clear
THE SHORT-LINE FAULT 103
by the conditions of the earth. On the other hand, the earth component of
the line-to-ground voltages does give rise to currents that return through
earth. The influence of the earth is, by the division in phase and earth
components, clearly visualised, and we can now distinguish between earth
waves and phase waves.
The phase waves are only attenuated by the resistance of the conductor.
The voltage level of the travelling waves is usually below the level where
corona occurs, and attenuation by corona losses does not take place. For
line lengths of a few kilometres, the attenuation is rather small and the
travelling waves can be regarded as undistorted waves, travelling with a
velocity close to the speed of light. The earth waves, however, will be
attenuated by the earth return path and this distorts the wave front. The
distortion appears as a prolongation of the wave front in the order of one
microsecond per kilometre and can therefore not be neglected.
For the first phase to clear and the last phases to clear, the conditions
are different. For the first phase to clear, the characteristic impedances are
connected as depicted in Figure 5.17. From Equation (5.15) it follows that
ua + ub + uc = 0 and ue = 0 (5.17)
Only phase waves are present and for a purely inductive line impedance
the initial amplitudes will have, for the 50- or 60-Hz power frequency, the
values
The rate of change of the current di/dt is the value just before current
interruption and can be simply calculated (see Section 4.7 Arc-circuit
interaction): √
di/dt = 2π fI 2 (5.20)
The time to the first peak of the line-side oscillation is twice the length
of the short-circuited part of the transmission line divided by the wave
velocity. The amplitude of the line-side TRV can be calculated from
the du/dt and the time to peak. For a very accurate calculation of the
amplitude, a correction should be made for the fact that for high-frequency
currents the current has no uniform distribution over the cross-section of
the conductor because of the inner field in the conductor. The correction
factor is, in that case, the ratio between the high-frequency inductance
and the power frequency inductance of the high-voltage transmission line.
Most high-voltage lines are equipped with aluminium core steel reinforced
(ACSR) conductors, wherein the aluminium outer layer carries most
of the current, and therefore the difference between the high-frequency
inductance and the power frequency inductance is very small.
For the last phase to clear in the case of a three-phase fault, the
characteristic impedances are connected as depicted in Figure 5.17 and
phase c is assumed to be the last clearing phase. The same connection of
the characteristic impedances can be applied for a single-phase-to-ground
fault if the two healthy phases are disconnected before the faulty phase.
The sum of the line-to-ground Ua , Ub , and Uc does not equal zero, and
this indicates that both phase waves and earth waves exist. For a purely
inductive line impedance the initial amplitudes will have, for the 50- or
60-Hz power frequency, the values
In phase c, both the phase wave and the earth wave are reflected at the
short-circuit point. Till the reflected waves have travelled back, the breaker
side of the line the voltage of phase c will be constant. The rate of rise
of the recovery voltage for the last clearing phase can be calculated from
Figure 5.17.
du di di
= (Z + Ze ) = (2Z1 + Z0 )/3 (5.22)
dt dt dt
The time to the first peak of the line-side TRV is in this case not
simply twice the length of the high-voltage transmission line from breaker
terminal to short-circuit point divided by the velocity of the travelling
REFERENCES FOR FURTHER READING 105
a
Line-side voltage
0 t
Figure 5.18 The TRV waveform for the last phase to clear (a) is composed from a phase
wave (b) and an earth wave (c)
waves. The phase wave and earth wave have a different velocity because
they travel in a different medium and the earth wave is also distorted. The
initial amplitude of the earth wave is only half the amplitude of that of
the phase wave. Figure 5.18 depicts the influence of the earth wave on the
TRV of the last phase to clear. The delay of the earth wave accounts for
the distortion.
Happoldt, H. and Oeding, D., Elektrische Kraftwerke und Netze, Chapter 17,
Springer-Verlag, Berlin, 1978.
Humphries, M. B. and Dubanton, M., ‘‘Transient Recovery voltage in the short line
fault regime Initial Transient Recovery Voltage,’’ Plenum press, NewYork, 1978.
Kennedy, M. W., in Flurscheim, C. H., ed., Power Circuit Breaker: Theory and
Design, Chapter 3, Peter Peregrinus, London, 1975.
Nakanishi, K., ed., Switching Phenomena in High-Voltage Circuit Breakers,
Chapter 1, Marcel Dekker, New York, 1991.
Peterson, H. A., Transients in Power Systems, Chapter 6, Wiley & Sons, New York,
Chapman & Hall, London, 1951.
Ragaller, K., ed., Current Interruption in High-Voltage Networks, Plenum Press, New
York, 1978.
Ruedenberg, R., Elektrische Wanderwellen, Part II, Chapter 9, Springer-Verlag,
Berlin, 1962; Ruedenberg, R., Electrical Shock Waves in Power Systems, Part
II, Chapter 9, Harvard University Press, Cambridge, Massachusetts, 1962.
Ruedenberg, R., in von H., Dorsch and P., Jacottet, eds., Elektrische schaltvorgaenge,
5th ed., Vol. II, Chapter 13, Springer-Verlag, Berlin, 1974.
Slamecka, E. and Waterschek, W., Schaltvorgaenge in Hoch- und Niederspannungs-
netzen, Chapters 3, 4, Siemens Aktiengesellschaft, Berlin, 1978.
Thoren, B., ‘‘Short-Line Faults,’’ Elteknik 9(2), 25–33 (1966).
Transients in Power Systems
Lou van der Sluis
Copyright 2001 John Wiley & Sons Ltd
ISBNs: 0-471-48639-6 (Hardback); 0-470-84618-6 (Electronic)
6
Power System Transient
Recovery Voltages
were carried out in different countries to determine the TRV across circuit
breakers while clearing short circuits in networks to provide a base for
standardisation of the TRV in national rules for type tests on circuit
breakers. System studies on transient network analysers brought insight
into the frequencies of oscillations and were verified by real tests in
the network [2]. Better understanding of the transient phenomena in the
network has led to improved testing practice in the high-power labora-
tories, more accurate measurement of the current-zero phenomena, and
consequently resulted in more reliable switchgear with a higher inter-
rupting capability. The use of SF6 as extinguishing medium allowed
a leap forward in the short-circuit performance of high-voltage circuit
breakers. But it appeared that until that time, unknown transients
played an important role. It was not sufficient anymore to simulate
the networks with lumped elements, and travelling waves also had to
be taken into account. The short-line fault made its appearance [3], (see
Chapter 5, Switching Transients). During that period of time, the end
of the 1950s, effort was made to represent the TRV oscillations by
standardised waveforms to be able to create TRVs by lumped elements
within the walls of the high-power laboratory. The four-parameter wave-
form was proposed by Hochrainer in 1957 [4], and the possibilities to
create the waveform in the high-power laboratory were investigated by
Baltensperger [5].
In the early nineteen sixties, several network studies were undertaken
in Japan [6] and Europe [7] and attempts were made to better-define the
analogue modelling of the transient phenomena. The tests for TRVs were
specified differently in the various national rules, and Subcommittee 17 on
High-Voltage Switchgear of the International Electrotechnical Commis-
sion (IEC) asked the (Conférence Internationale des Grands Réseaux
Electriques à Haute Tension) Study Committee on Circuit Breakers (named
Study Committee 3 in those days) to promote new extensive investigations
on an international base. A working group, CIGRE working group 3.1,
set up for this purpose in 1959, decided to start a complete investigation
of TRVs associated with the opening of the first pole of a circuit breaker
clearing a three-phase ungrounded fault in some large 245 kV networks.
Two of these networks were fully investigated: the Italian network in its
1962 situation and the French network in its 1965 situation. Some 2000
TRV waveshapes associated with short-circuit currents up to 45 kA were
collected. Based on this collection of TRVs, a classification of current
ratings was proposed [7].
In publication 56-2 (1971) on high-voltage alternating circuit breakers,
IEC recommends characteristic values for simulation of the TRV by the
CHARACTERISTICS OF THE TRANSIENT RECOVERY VOLTAGE 109
The IEC and ANSI/IEEE standard bodies are trying to harmonise their
documents in the future.
Uc Uc
Voltage
Voltage
U1
U′ U′
td t′ t3 Time td t ′ t1 t2 Time
The TRV waveforms for voltage ratings up to 100 kV have a so-called (1-
cosine) waveform because the interrupted short-circuit current is mainly
supplied from substation transformers.
In the older version of the IEC-standard, this duty was named test
duty 1.The fault current is supplied through one transformer only. The
TRV has a (1-cosine) shape and a rather steep rate of rise, ranging
from 5.5 kV/µs at 100 kV to 12.6 kV/µs at 765 kV. The current to be
interrupted is symmetrical resulting in the highest peak voltage across the
breaker terminals.
Former test duty 2. The fault current is 30 percent of the nominal rated
short-circuit current and is supplied by a local source or one or two
overhead lines, eventually with one or two transformers in parallel. The
112 POWER SYSTEM TRANSIENT RECOVERY VOLTAGES
In the older versions of the IEC standard this duty was named test duty 3.
The fault current is associated with 2, 3, or 4 overhead lines, eventually
with transformers in parallel and the TRVs have a rate of rise of 3 kV/µs
which is standardised for breakers with a voltage rating of 100 kV and
above. The tests are also performed with a symmetrical current to get the
highest TRV peak.
Former test duty 4, the rate of rise of the TRV is 2 kV/µs, and by
performing this test duty, the circuit breaker proves that it is able to
interrupt its maximum rated symmetrical short-circuit current with its
rated operating sequence.
There are two alternative rated operating sequences:
a) O–t–CO–t –CO
b) CO–t – CO
With
t = 15 s for circuit breakers not intended for rapid auto reclosing;
O represents an opening operation;
CHARACTERISTICS OF THE TRANSIENT RECOVERY VOLTAGE 113
The O–0.3 s–CO–3 min–CO duty cycle is typical for many high-voltage
circuit breakers in service. The rapid auto reclosing, 300 ms after inter-
ruption of the fault current, is to verify whether the fault was created by
touching lines or the current was injected in the system by a lightning
stroke.
For the majority of the circuit breakers, this is a very severe test because
maximum arc energy is developed in the interrupting chamber of the
breaker. Because of the maximum developed arcing energy, this is also a
severe test for the operating mechanism. For gas-insulated metal-enclosed
switchgear or GIS, the exhaust of gas can influence the capacity to
withstand high voltage between phases and between phases and the tank.
The time constant of the DC component is fixed at a standard value of
45 ms. Depending on the system characteristics, for example, if a circuit
breaker is close to a generator, the time constant may be higher than the
standard value. Recommended values are as follows:
• 120 ms for rated voltages up to and including 52 kV;
The basis for the IEC standards was developed in the 1960s and the
ANSI/IEEE TRV parameters were also fixed in that period of time.
While the requirements are similar, there are significant differences. ANSI
C37.06 defines two TRV envelopes. For circuit breakers rated below
100 kV, Figure 6.2 shows the envelope defined as a (1-cosine) waveform
with a crest value E2 equal to 1.88 times the rated maximum voltage V
of the circuit breaker and the time to peak T2 varies as a function of the
voltage rating.
For circuit breakers rated above 100 kV, the envelope is defined in
terms of an exponential-cosine waveform, as shown in Figure 6.3. The
exponential is defined by an initial rate-of-rise parameter R, √
which
√ has been
established at 2 kV/µs, and a value E1 , which equals 1.3 2/ 3 = 1.06
114 POWER SYSTEM TRANSIENT RECOVERY VOLTAGES
E2
E 2 = 1.88 V rated
Voltage T 2 = Variable
T2
Time
Figure 6.2 IEEE standard TRV limiting curve for breakers rated below 100 kV
E2
E2 = KaKf 2 2
√ 3 V= 1.4 ∗ 1.3 √ 3 V= 1.49 V
E1 E1 = Kf 2 V= 1.3 2 V= 1.06 V
Voltage
√3 √3
R T2 = variable T1 = 2 µs R = 2 k V /µs
0
T1 T2
Time
Figure 6.3 IEEE standard TRV limiting curve for breakers rated 100 kV and above
• Below 100 kV, IEEE uses a (1-cosine) TRV waveform form while
IEC uses a two-parameter envelope. For 100 kV and above voltage
rating, IEEE makes use of an exponential-cosine waveform, whereas
IEC uses a four-parameter envelope.
• The TRV parameters for the IEC envelopes are independent of the
rated interrupting currents, whereas IEEE has different values for
different ratings.
The test duties resemble the test duties of the IEC 60056. For voltages
below 100 kV, the TRV waveform has a (1-cosine) shape. The time to
peak T2 corresponds with 1.137 times t3 (or t3 = 0.88T2 ) from the IEC.
For voltages below 100 kV, ANSI/IEEE assumes an ungrounded system
by means of a first-pole-to-clear factor of 1.5, and the peak value of
the TRV is E2 = 1.88 kV. Expressed in terms of an amplitude factor,
this corresponds to an amplitude factor of Ka = 1.535, whereas the IEC
requires an amplitude factor of 1.4 related with a 100 percent short-circuit
current.
For voltage ratings of 100 kV and above, ANSI/IEEE is also in close
harmony with IEC. For the TRV, the exponential-cosine waveform is
used but the rate-of-rise and the time-delay are as in IEC: 2 kV/µs, and
2 µs, respectively. Further, IEEE Std C37.04-1999 reads: ‘Since most, if
not all, systems operating at 100 kV and above are effectively grounded,
a first-pole-to-clear factor of 1.3 is required.’ At this point, ANSI/IEEE
differs from IEC. For voltage ratings between 100 kV up to and including
170 kV, IEC offers the choice between an isolated neutral, and an earthed
neutral, and depending on the system grounding the first-pole-to-clear
factor is 1.5 and 1.3, respectively.
The amplitude factor of the TRV oscillation is, similar to IEC, 1.4,
and this corresponds
√ √ with the parameters of the √exponential
√ cosine:
E1 = 1.3 ∗ 2/ 3 V = 1.06 V and E2 = 1.3 ∗ 1.4 ∗ 2/ 3 V = 1.49 V.
116 POWER SYSTEM TRANSIENT RECOVERY VOLTAGES
Three-Phase-
G to-
(a) ground fault
+ s l
Z1 V l1 =V l2 =V l0 = 0
E Vl =0
−
Z2 s l V s1 =V s2 =V s0 = 1/3E
V s =1.0E
Z0 s l
(b)
+ Lt
E Ct Vs Vl
−
(c)
2E (no damping)
Vs
E
Vs & Vl Vl
(d)
Figure 6.4 The TRV for a three-phase-to-ground fault: (a) one-line system representation;
(b) connection of the sequence networks; (c) transient network; (d) system voltages
Three-phase
G
ungrounded fault
(a)
Z1 s l V l1 =V l2 = 0 V l0 = −0.5E
E V l = −0.5E
Z0 s l
(b)
+ Lt
E Ct Vs Vl
−
(c)
Vs
V s and V l 0.5E
Vl
(d)
Figure 6.5 The TRV for a three-phase-ungrounded fault: (a) one-line system representa-
tion; (b) connection of the sequence networks; (c) transient network; (d) system voltages
REFERENCES 119
6.3 REFERENCES
6.1 Park, R. H. and Skeats, W. F., ‘‘Circuit breaker recovery voltages, Magnitudes
and Rates of Rise’’, Trans. A.I.E.E., 204–239 (1931).
6.2 Catenacci, G., ‘‘Le frequenze proprie della rete Edison ad A.T.,’’
L’Elettrotrechnica XLIII(3), (1956).
6.3 Pouard, M., ‘‘Nouvelles notions sur les vitesses de rétablissement de la tension
aux bornes de disjoncteurs à haute tension,’’ Bulletin de la Société Française des
Electriciens, 7th series VIII(95), 748–764 (1958).
6.4 von Hochrainer, A., ‘‘Das Vier-Parameter-Verfahren zur Kennzeichnung der
Einschwingspannung in Netzen’’ ETZ 78(Part 19), 689–693 (1957).
6.5 Baltensperger, P., ‘‘Définition de la tension transitoire de rétablissement aux
bornes d’un disjoncteur par quatre paramètres, possibilités des stations d’essais
de court-circuit,’’ Bulletin de l’Association Suisse des Electriciens 3, (1960).
120 POWER SYSTEM TRANSIENT RECOVERY VOLTAGES
6.6 Ozaki, Y., ‘‘Switching surges on high-voltage systems,’’ Central Research Ins-
titute of Electric Power Industry, Tokyo, Japan, 1994.
6.7 Catenacci, G., Paris, L., Couvreux, J. P., and Pouard, M., ‘‘Transient recovery
voltages in French and Italian high-voltage networks,’’ IEEE, PAS 1986(11),
1420–1431 (1967).
6.8 Barret, J. P., ‘‘Développements récents des méthodes d’étude des tensions tran-
sistoires de manoeuvre sur les réseaux à haute tension, ’’ Revue générale de
l’électricité, 441–470 (1965).
6.9 Braun, A. et al., ‘‘Characteristic values of the transient recovery voltage for
different types of short circuits in an extensive 420 KV system,’’ ETZ-A, 97,
489–493 (1976).
6.10 Novotry, V. et al., ‘‘Transient recovery voltages in medium-voltage networks,’’
Electra 88, 49–88 (1983).
6.11 Parrot, P. G., ‘‘A review of transformer TRV conditions,’’ Electra 102, 87–118
(1985).
6.12 Bonfanti, I. et al., ‘‘Transient recovery voltages in medium-voltage networks,’’
Electra 181, 139–151 (1998).
Transients in Power Systems
Lou van der Sluis
Copyright 2001 John Wiley & Sons Ltd
ISBNs: 0-471-48639-6 (Hardback); 0-470-84618-6 (Electronic)
7
Lightning-Induced Transients
In addition, the solar system can be the cause of a blackout of the power
system. Heavy eruptions on the solar surface in a so-called solar spot on
March 10th, 1989, caused a large blackout of the Hydro Quebec grid
in Canada on March 13th. Geomagnetic-induced currents were the cause
of a DC component in the system current, and the power transformers
were brought into saturation. The system current got strongly distorted
and contained, apart from the power frequency ground wave, higher
122 LIGHTNING-INDUCED TRANSIENTS
Thundercloud
Charge Q
E E
100 % Ep
90 %
Voltage
50 %
30 %
tf =1.2 µs Time
tt = 50 µs
are brightly illuminated, and the amplitude of the current is in the range
of 20–100 kA. The characteristic impedance of the lightning current path
lies in the order of the characteristic impedance of overhead transmission
lines ZL = 500 . The stroke current Istroke will divide into three parts.
One part of the stroke current, Itower , will flow through the tower structure
to ground and the remaining part of the stroke current divides equally
and flows in opposite directions through the ground wire, as shown in
Figure 7.3.
Istroke = Itower + Iground wire (7.2)
After the stroke, the tower top has an initial voltage equal to the difference
between the potential E of the thundercloud and the voltage drop over
the lightning path VL . Three voltage waves, all equal to the initial tower
top voltage, Vtower = E − VL , travel away from the struck point – one
voltage wave along the tower structure toward the ground and the other
two in opposite directions from the struck point along the ground wire.
The voltage waves travel with a velocity less than that of light in free
space, and each of the three waves will be refracted and reflected at the
nearest transition points. In the tower structure, there will be reflections
between the tower-footing resistance and the tower top. The voltage waves
travelling on the ground wires will be reflected and further transmitted to
the adjacent towers. This process repeats itself as the transmitted voltage
waves travel along the ground wire.
On the ground wire, a forward and backward current and the therewith-
related voltage waves travel. The waves in the positive x-direction are the
I stroke
I tower
R tower
Figure 7.3 Lightning stroke to a transmission line tower with ground wires
DIRECT LIGHTNING STROKE TO A LINE 127
forward waves and the waves in the negative x-direction form the back-
ward waves. This implies a minus sign for the backward-travelling current
wave, because it propagates in the negative x-direction (see Chapter 3,
Travelling Waves).
For the current in the ground wire, Iforward = −Ibackward = 0.5Iground wire ,
and for the voltage on the ground wire Vtower = 0.5Iground wire Zground wire
Substituted in Equation (7.2)
Vtower 2Vtower
Istroke = + (7.3)
Rtower Zground wire
The voltage and current waves have the same shape as the lightning wave.
V = 0.5IZphase (7.5)
The voltage wave travelling along the line hit by lightning, for instance,
phase a, induces voltage waves on the neighbouring phases also. When
the coupling factor (see Chapter 5, Section 5.1, Interrupting of Capacitive
Currents) between phase a and phase b is Kab then the voltage to the
ground, induced on phase b is
Vb = Kab Va (7.6)
128 LIGHTNING-INDUCED TRANSIENTS
The voltage Vab can cause a flashover between the two phases, if the
separation distance is not large enough, thus introducing a two-phase
short-circuit. When no flashover occurs, Va , Vb , and Vc will travel along
the phase conductors a, b, and c and encounter the insulator strings that
support the phase conductors in the tower, which is at ground potential.
The voltage Va is the highest, and the insulator string of phase a can be
subjected to a flashover.
The voltage waves travel in both directions along the line, away from
the point of strike, and they travel from tower to tower. In the vicinity
of each tower, the phase conductor has a larger capacitance to ground
than at other parts of the line, because of the insulator strings and the
presence of the metal tower structure that brings the ground potential
closer. This tower capacitance reduces the front steepness of the voltage
wave and creates eventually a time delay. After passing a few transmission
line towers, the danger of a flashover reduces considerably.
Travelling along the line, the voltage waves can also encounter a series
capacitor for compensation of long transmission lines, a series reactor used
for telecommunication over the line, or the transmission line connected
to a high-voltage cable before entering a substation. In all cases, there
is a change in characteristic impedance at the transition point and the
relation between the current and voltage wave changes (see Chapter 3,
Section 3.5, Reflection and Refraction of Travelling Waves). Examining
the series connection of a high-voltage transmission line, an impedance,
and a high-voltage cable, as depicted in Figure 7.4, gives already a clear
insight into the voltage profile, arising after the reflection and refraction
of the travelling voltage waves. Let us assume that voltage wave e(t)
is present on the high-voltage transmission line and travels toward A,
where the series impedance is connected. For the ease of calculation,
the voltage function and the lumped and distributed network elements
are transformed to the Laplace domain. The lightning-induced voltage
wave can, for a first-approximation, be described by a single exponential
A B
function
e(t) = Ep e−αt (7.8)
Z(p) + Z2 − Z1
rA = (7.9)
Z(p) + Z2 + Z1
The expression for the current wave travelling through Z(p), and subse-
quently through Z2 is
eA (p) 2e(p)
iB (p) = = (7.11)
Z(p) + Z2 Z(p) + Z2 + Z1
Z2
eB (p) = 2e(p) (7.12)
Z(p) + Z2 + Z1
Substituting for the lightning wave, the simplified expression of the single
exponential function of Equation (7.8), which transformed to the Laplace
domain, becomes e(p) = Ep /(p + α), and for the series impedance between
the line and the cable, an air-core reactor with resistive losses, transformed
to the Laplace domain Z(p) = pL + R, gives for the current through the
air-core reactor
1 1
iB (p) = 2Ep (7.13)
(p + α) (pL + R + Z1 + Z2 )
130 LIGHTNING-INDUCED TRANSIENTS
and transformed back into the timed domain, the expression for the
current is
1
iB (t) = 2Ep {e−αt − e−[(Z1 +Z2 +R)/L]t } (7.14)
Z1 + Z2 + R − αL
eB (t) 1
iB (t) = = 2Ep {1 − e−[(Z1 +Z2 +R)/L]t } (7.15)
Z2 Z1 + Z2 + R
The expression for the voltage at point A (see Figure 7.4) can be calcu-
lated with eA (p) = iB (p){pL + R + Z2 }, and is transformed into the time
domain,
2(R + Z2 ) 2Z1
eA (t) = Ep + Ep e−[(Z1 +Z2 +R)/L]t (7.16)
Z1 + Z2 + R Z1 + Z2 + R
Let us now examine a few possible configurations. First: the situation that
line and cable are coupled directly. This means that in our example the
values for R and L are zero. The voltage at the line–cable intersection is
2Z2
eB (t) = Ep (7.17)
Z1 + Z2
The nominator is smaller than the denominator when Z1 > Z2 , and the
refracted wave, travelling from intersection point B onwards, in that case,
is smaller in amplitude than the incident wave. When we consider 450
as characteristic impedance for the transmission line and 50 for the
cable, the reduction in amplitude is eighty percent. In other words, when
an incoming high-voltage transmission line is connected to a substation
via an underground cable, the cable serves as an amplitude reductor
for incoming lightning-impulse and switching-impulse waves. When the
substation itself is hit by lightning, the opposite occurs; the wave travelling
via the cable to the line is amplified at the line–cable intersection point, in
our example by a factor 1.8.
When the connection between the line and the cable is formed by a
component with a large inductivity, for instance, a current transformer,
the series winding of a phase-shift transformer or a line-carrier reactor, the
DIRECT LIGHTNING STROKE TO A LINE 131
expression for the voltage at point B is, when we neglect resistive losses,
2Z2
eB (t) = Ep {1 − e−[(Z1 +Z2 )/L]t } (7.18)
Z1 + Z2
The voltage waves EA (t) and EB (t) are depicted in Figure 7.5 for the
case Z1 = Z2
A series reactance between two lengths of a transmission line reduces
the wave front of the refracted wave, but for the incoming wave, the
reactance behaves first as an open connection, which results in a reflected
wave with a double amplitude. In addition, the reactance itself experiences
severe voltage stress.
The expression for the voltage eA (t) − eB (t) across the reactance is
From Equation (7.20), we can see that the reactance must be able to
withstand twice the crest value of the incident lightning impulse wave
or should be protected by surge arresters. The highest voltage appears
across the terminals of the series reactance when the lightning wave
reaches the reactance at t =0. After a time duration of more than three
reactance time constants, t > 3L/R, the influence of the reactance on the
waveshape disappears and the situation is similar to direct coupled lines
(Equation (7.17)).
When two transmission lines have a parallel impedance connected at
their point of intersection, as depicted in Figure 7.6, the parallel impedance
eL
2Ep + −
Ep Ep
Z1 A B Z2
Figure 7.5 Reflected and refracted travelling waves along two series connected lines and
a series reactance
132 LIGHTNING-INDUCED TRANSIENTS
2Z2 Z(p)
eA (p) = e(p){1 + rA } = e(p) (7.22)
Z(p)(Z2 + Z1 ) + Z1 Z2
1
Z(p) = (7.23)
pC
2Z2 2Z2 1
eA (p) = e(p) = e(p) (7.24)
(Z1 + Z2 ) + pCZ1 Z2 CZ1 Z2 (Z1 + Z2 )
p+
CZ1 Z2
Ep
e(p) = (7.25)
(p + α)
Ep
A
Z1 Z2
Z(p)
Figure 7.6 Series connection of two transmission lines with a parallel impedance at the
intersection point
DIRECT LIGHTNING STROKE TO A LINE 133
and the expression for the voltage at node A can, after back transformation
into the time domain, be written as
2Z2 1 1
eA (p) = Ep
CZ1 Z2 (p + α) (p + γ )
2Z2 1 1 1
= Ep − (7.26)
CZ1 Z2 γ − α p + α p + γ
2Z2 −αt
eA (t) = Ep e − e−γ t (7.27)
Z1 + Z2 − α(Z1 Z2 )C
with
Z1 + Z2
γ =
CZ1 Z2
Z 2 < Z1
Z1 Z2
Figure 7.7 Series connection of two transmission lines with a parallel capacitor
surge arrester starts to draw current. For the classical surge arrester with
spark gaps, the voltage level needs to surpass a certain voltage level until
the spark gap breaks down and the resistive elements are in connection
with the system. The spark gap serves, in fact, as a voltage-controlled fast
switch and prevents current from flowing through the resistive elements
at normal system voltage.
8
Numerical Simulation of
Electrical Transients
are often more accurate and cheaper than a TNA but not always easy to
use!
The computer programs that were developed first were based on the
techniques to compute the propagation, refraction, and reflection of trav-
elling waves on lossless transmission lines. For each node, the reflection
and refraction coefficients are computed from the values of the char-
acteristic impedances of the connected transmission line segments. The
bookkeeping of the reflected and refracted waves was done and visu-
alised by means of a lattice diagram (see Chapter 3, Travelling Waves),
as was first published by Bewley in 1933. Another fruitful development
was the application of the Bergeron method based on the method that
O. Schnyder in 1929, in Switzerland, and L. Bergeron in 1931, in France,
had developed for solving pressure wave problems in hydraulic piping
systems. The Bergeron method applied to electrical networks represents
lumped network elements, like an L or a C, by short transmission
lines. An inductance becomes a stub lossless line with a character-
istic impedance Z = L/τ and a travel time τ . In the same manner, a
parallel capacitance becomes a stub transmission line with a characteristic
impedance Z = τ/C and a travel time τ . A series capacitor, however, poses
a problem.
Almost all programs for the computation of electrical transients solve
the network equations in the time domain, but some programs apply
the frequency domain, such as the frequency-domain transient program
(FTP), or use the Laplace domain to solve the network equations. A great
advantage of the frequency domain is that the frequency-dependent effects
of high-voltage lines and underground cables are automatically included.
The advantage of using the Laplace transform is that back transformation
into the time domain results in a closed analytical expression. When a
certain time parameter is substituted in that equation, the currents and
voltages can directly be calculated for different circuit parameters. For a
program with a solution directly in the time domain, the computation has
to be repeated when a circuit parameter changes. The Laplace transform
has, however, its drawbacks. The method cannot cope with nonlinear
elements, such as surge arresters and arc models.
In this chapter, four different time domain computer programs are
described:
• the Electromagnetic transient program, based on the Nodal Analysis
from network theory;
L C
k m k m
ik,m (t ) ik,m (t )
k m k m
uk (t ) R = 2L /∆t um (t ) uk (t ) R = ∆t /2C um (t )
Reference Reference
k Lossless line m
ik,m(t ) im,k ( t)
uk (t ) Z ik (t −τ) im (t −τ ) Z um (t )
1 R 2 L 3
iL
ie iC
e(t ) C
R = 2L /∆t
1 2
I L(t − ∆t )
e(t ) /R R I C(t − ∆t ) R = ∆t / 2C
Yu = i − I (8.8)
• The time-step loop is entered and the vector of the right-hand side
of Equation (8.8) is computed after a time step t;
THE ELECTROMAGNETIC TRANSIENT PROGRAM 141
The arc models within the EMTP are implemented by means of the
compensation method. The nonlinear elements are essentially simulated
as current injections, which are superimposed on the linear network, after
a solution without the nonlinear elements has been computed first.
The procedure is as follows: the nonlinear element is open-circuited and
the Thevenin voltage and Thevenin impedance are computed. Now, the
two following equations have to be satisfied. Firstly, the equation of the
linear part of the network (the instantaneous Thevenin equivalent circuit
142 NUMERICAL SIMULATION OF ELECTRICAL TRANSIENTS
current through the inductances, because both are continuous when the
topology of the network changes through, for instance, a switching action.
The set of equations to solve the sample RLC circuit, as shown in
Figure 8.2, formulated by means of the MNA method is described in the
following text.
1 1
− 0 1 0 0
R R u1 (t) 0 0
1
− 1 u2 (t) −1 0 0
0 0 0 iL (t)
R R u3 (t) = 1 0 + 0
0 1 C
u (t)
0 0 0 ie (t) 0 0 e(t)
1 0 0 0 0
iC (t) 0 1 0
0 0 1 0 0
(8.11)
The vector with the state variables, containing the currents through the
inductances and the voltages across the capacitance, can be obtained by
integration of the following equation.
u (t)
1 1 1
0 − 0 0 u 2 (t)
iL (t)
= L L
1 u3 (t) (8.12)
uC (t)
0 0 0 0
ie (t)
C
iC (t)
factorised at each time step, but this is not efficient because of the very
long computation times. To solve Equation (8.13) and Equation (8.14),
the following method is used.
• The right-hand side vector of Equation (8.14) is calculated with the
initial values of x1 .
• the computational step size is not present within the MNA matrix
(see Equation (8.11)). If the step size is adjusted during the calcu-
lation, the MNA matrix does not need to be inverted or factorised
once more (assuming that the MNA matrix is not time-dependent).
The method has one disadvantage:
• The matrices A, B, and C require linearity and it is necessary that the
two sets of unknowns can be defined and solved separately from the
corresponding set of equations. Therefore the MNA method deals
with linear models only. Nonlinear diodes, arc models, and so on,
can be incorporated by means of tricks only.
An arc model can be implemented within the MNA method in various
ways.
The arc model can be modelled as a nonlinear conductance. However,
the MNA matrix becomes time-dependent and should be inverted or
factorised at each time step. This involves long computation times and
is, therefore, not efficient. An alternative can be found by applying the
so-called partial matrix (factor) updating. Now the changes are not made
within the MNA matrix itself, but in the inverted MNA matrix or the
MNA matrix factors. This introduces additional computation time for the
updating, but the MNA matrix has to be inverted or factorised only once.
THE XTRANS PROGRAM 145
Thevenin-equivalent
is
u us Arc
model
Load line
Arc resistance R a
Voltage
us
is
Current
Another approach is the one that Van der Sluis followed to implement
the Mayr arc model in the MNA computer program. The arc model is
treated as a voltage source. This technique is depicted in Figure 8.4. The
linear network to which the arc model is connected can be replaced by its
Thevenin equivalent with its corresponding load line.
Each time step, the arc resistance Ra is computed from the arc models’
differential equation. At the point of intersection of the momentary arc
resistance and the load line, both the network and the arc model have the
same voltage us and current is . If two points of the load line, (u1 , i1 ) and
(u2 , i2 ), are known, the voltage us and current is are computed from the
following equation.
(Ra i1 − u1 )(u2 − u1 ) us
us = u1 + is = (8.15)
u2 − u1 + Ra (i1 − i2 ) Ra
The set of equations to solve the sample RLC circuit as shown in Figure 8.2
formulated by means of the DAEs is described in the following text.
u1 (t) − u2 (t)
ie (t) +
R
u1 (t) − u2 (t)
− iL
R
iL − iC
=0 (8.17)
u1 (t) − e(t)
u2 (t) − u3 (t)
iL (t) −
L
iC (t)
u3 (t) −
C
When using the DAEs, an arc model, such as the Mayr arc model, can be
implemented easily by extending the system of equations as follows:
..
.
i − gu
..
=0 (8.18)
.
1 u 2 g2
g − −g
τ P
The solution of the DAEs [Equation (8.16)] is achieved by using the BDF
method (backward differentiation formulas method). This method was
first proposed by Gear in 1971 and has since then been studied and
developed by mathematicians. The solution is performed in two steps:
THE XTRANS PROGRAM 147
1
k
xn+1 = αi xn+1−i (8.19)
hn
i=0
xn+1 − xn
xn+1 = (8.20)
hn
Jxn+1
i = g(xn+1
i , tn+1 ) (8.22)
xn+1 n+1
i+1 = xi + xn+1
i (8.23)
V L
+
C1 C2 R
6.00× 101
3.25× 101
5.00× 100
Volt
−2.25× 101
−5.00× 101
0 2.0 ×10−3 4.0× 10−3 6.0 × 10−3 8.0 × 10−3 1.0 × 10−2
Time (seconds)
but can also be nonlinear and first-order differential if the BDF method
is used.
As demonstrated, the solution of Equation (8.16) is equal to the solution
of Equation (8.21) when applying the iteration formulas (8.22) and (8.23).
This iteration process communicates with the ‘outside’ world by making
a guess for xn+1 and expecting a new mismatch vector in return. With
the guesses for the terminal voltages and the introduced unknowns of an
element, the mismatches for the introduced equations and the currents at
the terminals are calculated. These currents are used in the main program
to calculate the mismatches corresponding to Kirchhoff’s equations.
The requirements of an element model to work with the BDF solution
method can be summarized as follows:
• The additional equations introduced by an element model must
be algebraic equations or first-order differential equations. These
additional equations must add new information to the set of
equations and
• The element models must be able to return guesses for the currents
at their terminals. The data available for this purpose are the guesses
of the introduced unknowns, the guesses of the node voltages at
the terminals, and the first-order derivatives of these unknowns
(obtained with Equation (8.19)).
Table 8.1 Unknowns, equations, and terminal currents for basic elements
between node a and node b
XTrans program
Library Library
• The block calculates the derivatives of its states based on the current
time, the inputs, and the states; and
u x y
(States)
(Input) (Output)
PSB block library act as links between electrical signals (voltages across
elements and currents flowing through lines) and Simulink blocks (transfer
functions) and vice versa respectively. Before the computation, the system
is initialised; the state-space model of the electric circuit is computed,
and the equivalent Simulink system is built up. The computation itself is
analogous to the previously mentioned Simulink computational process.
The sample RLC circuit, previously used in this chapter, built up in the
PSB is shown in Figure 8.9.
It is important to note that the arrows in the diagram do not indicate
causality, as is the case in the Simulink block diagrams. The state space
model of this sample circuit is described in the following text.
R 1
1
− −
ẋ = 1L L x + L VAC
0 0
C
with
i
x= L (8.26)
uC
MATLAB contains a large number of ODE solvers, some with fixed and
others with variable step size. The ode15s solver, which can be used
to solve stiff systems, is a variable order solver based on the numerical
differentiation formulas (NDF’s). These are more efficient than the BDF’s,
though a couple of the higher-order formulas are somewhat less stable.
The arc models can be modelled as voltage-controlled current sources.
This approach is visualised in Figure 8.10, where both the Mayr arc
model block and the underlying system are shown as implemented in the
arc model blockset, which is freely available on http://eps.et.tudelft.nl.
AC Voltage source
Hit
Step crossing
Figure 8.10 Implementation of the Mayr arc model in Matlab Simulink/Power System
Blockset
The equations of the Mayr arc model have been incorporated by means
of the Simulink DEE (differential equation editor) block, as shown in
Figure 8.11.
Therefore, the following system of equations is solved:
dx(1) u(2) ex(1) u(1)2
d ln g u(2) gu2
= −1
dt τ P
dt = P
−1
(8.27)
τ
y = ex(1) u(1)
i = gu
x(1) the state variable of the differential equation, which is the natural
logarithm of the conductance: ln(g).
x(0) the initial value of the state variable, that is, the initial value of the
arc conductance: g(0).
u(1) the first input of the DEE block which is the arc voltage: u.
u(2) the second input of the DEE block, which represents the contact
separation of the circuit breaker: u(2) = 0 when the contacts are
closed, and u(2) = 1 when the contacts are being opened.
y the output of the DEE block, which is the arc current: i.
g the arc conductance
u the arc voltage
i the arc current
τ the arc time constant
P the cooling power
τ and P are the free parameters of the Mayr arc model, which can be set
by means of the dialogue, as depicted in Figure 8.12, that appears when
the Mayr arc model block is double clicked.
THE MATLAB POWER SYSTEM BLOCKSET 155
The Simulink ‘hit crossing’ block detects when the input, in this case the
current, crosses the zero value. By adjusting the step size, this block ensures
that the simulation detects the zero crossing point. This is important
because the voltage and current zero crossing of the arc, which behaves as
a nonlinear resistance, must be computed accurately.
The Simulink ‘step’ block is used to control the contact separation of
the circuit breaker. A step is made from a value zero to one at the specified
contact separation time. When the contacts are still closed, the following
differential equation is solved:
d ln g
=0 (8.28)
dt
Therefore, the arc model behaves as a conductance with the value g(0).
From the contact separation time on, the Mayr equation is being solved:
d ln g 1 gu2
= −1 (8.29)
dt τ P
156 NUMERICAL SIMULATION OF ELECTRICAL TRANSIENTS
Both the initial value of the arc conductance g(0) and the time at which
the contact separation of the circuit breaker starts are specified by means
of the arc model dialogue, as displayed in Figure 8.12.
9
Insulation Coordination,
Standardisation Bodies, and
Standards
• Sustained AC overvoltages.
India, CHPTL (China High Power Test Laboratory Liaison) from China,
NEFTA (National Electric Test Facility) from South Africa, PALTS (Polish
Association of Laboratories for Testing Switchgear) from Poland, VEIKI
from Hungary, and ZKUSEBNICTVI from the Czech Republic participate
in the technical committees from STL.
The basic aim of STL is the harmonised application of IEC and regional
standards related to IEC standards to the type-testing of electrical trans-
mission and distribution equipment (above 1000 V AC and 1200 V DC)
for which type tests include high-power verification tests.
IEC 60044-1
IEC 60056
IEC 60076-1
IEC standard for power transformers. This part of the standard applies
to three-phase and single-phase power transformers (including auto-
transformers) with the exception of certain categories of small and
STANDARDS FOR HIGH-VOLTAGE ELECTRICAL POWER EQUIPMENT 165
IEC 60076-5
IEC standard for power transformers. This part of the standard identifies
the requirements for power transformers to sustain without damage from
the effects of overcurrents originated by external short circuits. The
requirements apply to transformers as defined in the scope of IEC 60076-1.
IEC 60099-1
IEC standard for surge arresters. This part of the standard applies to
surge-protective devices designed for repeated operation to limit voltage
surges on AC power circuits and to interrupt power-follow current. In
particular, the standard applies to surge arresters consisting of single or
multiple spark gaps in series with one or more nonlinear resistors.
IEC 60129
IEC 60265-1
IEC 60282-1
come within the scope of this standard, but the correct opening of the
striker in combination with the tripping mechanism of the switching device
is outside the scope of this standard; see IEC 60420.
IEC 60282-2
IEC 60289
IEC 60298
IEC 60353
IEC standard for line traps. This standard applies to line traps inserted
into high-voltage AC transmission lines to prevent undue loss of carrier
signal power, typically in the range 30 kHz to 500 kHz, under all power
system conditions, and to minimise interference from carrier signalling
systems on adjacent transmission lines.
IEC 60420
IEC 60427
IEC 60517
IEC 60694
IEC 60726
IEC 61128
IEC 61129
IEC 61233
IEC 61259
IEC 61633
IEC technical report for high-voltage metal-enclosed and dead tank circuit
breakers. This report contains information and recommendations for test
circuits and procedures for type-testing relevant to short-circuit making
and breaking, and switching performance of metal-enclosed and dead tank
circuit breakers. Special test circuits are given and the tests described can
be made, in principle, in both direct and synthetic circuits.
10
Testing of Circuit Breakers
Except for the mechanical and environmental tests, all the tests in the
IEC-60056 type test requirements are designed to prove the interrupting
ability of circuit breakers. The fault current interrupting ability is the basic
function of a circuit breaker and, therefore, circuit breaker testing requires
high short-circuit power in conjunction with both the current and voltage
being variable over a wide range. The testing of circuit breakers can be
carried out either in the actual system or in a simulated test laboratory
situation. System testing offers the advantage that no special investments
are necessary for the testing equipment, and the breakers face the real
fault conditions as they would in service.
However, system testing is not very practical. Apart from interfering
with normal system operation and security, it is difficult to create the
various system conditions as prescribed by the standards. Moreover, it
would not readily supply the facilities for development testing for the
manufacturers. A high-power laboratory provides the possibility to test
circuit breakers conveniently under the simulated system conditions.
Transmission
system
SW T MB MS L
R0
TB
C0
Figure 10.1 One-line diagram of a testing station using the network as a source of supply
Transmission
system
SW
M
MB MS L T
G
R0
TB
C0
D
=
Figure 10.3 KEMA’s High-Power Laboratory, located on the banks of the river Rhine,
in Arnhem, The Netherlands, was commissioned in 1973 and extended with a synthetic
test facility in 1980 (courtesy of KEMA)
(Figure 10.3). The total available short-circuit power in the test bay at
KEMA laboratories is 8400 MVA. Other large generator testing stations
are located at Xian in the Peoples Republic of China, at Bangalore in
India, at Bechovicze (near Prague) in the Czech Republic, and at Milano
in Italy. In addition, the large switchgear manufacturers have generator
testing stations: Siemens in Berlin, Germany, ABB in Ludvika, Sweden
and in Baden, Switzerland, GEC-Alsthom in Villeurbanne, France, and
Schneider Electric in Grenoble, France.
The largest network station in the world is operated by IREQ in
Varennes in Canada. Electricité de France owns a large test facility ‘les
Renardieres’ in Moiret-sur-Loing (near Paris) in France. Les Renardieres is
also equipped with generators and can also use the network for the supply.
Figure 10.4 Two of the four short-circuit generators with their auxiliary machines in
KEMA’s generator room (courtesy of KEMA)
DIRECT TEST CIRCUITS 175
Figure 10.5 Single-phase master breaker. This air-blast breaker has an operating pressure
of 80 bar and is capable of clearing 160 kArms in 7 milliseconds at 15 kV. Each generator
phase has its own master breaker (courtesy of KEMA)
Figure 10.6 Single-phase make switch. Each generator phase has its own make switch,
capable of making 270 kApeak with an accuracy of 3° . The maximum short-time current is
100 kArms for 3 s (courtesy of KEMA)
Figure 10.7 Current limiting reactors for one short-circuit generator. (courtesy of KEMA)
the very sensitive electronic circuits make the measurement setup prone
to electromagnetic disturbances, the great advantage is that once the data
are stored in digital form they can be analysed by the computer and stored
permanently for future study.
Although the overall accuracy of a measurement system cannot be
better than the weakest link, which is usually the transducer, recent
improvements in the instrumentation have made it possible to overcome
many traditional problems. Optical isolators and fiber-optic transmission
make it less cumbersome to eliminate electromagnetic interference in the
rather unfriendly electromagnetic measurement environment of the test
bay in the high-power laboratory.
In the test circuit depicted in Figure 10.9, the test breaker TB is the
high-voltage breaker under test. While performing a break test, TB is in a
closed position. In addition, the master breaker MB is closed, but the make
switch MS is open. After the generator is spun up to the nominal power
frequency and the rotor is excited to a voltage, giving the required rated
testing voltage at the high-voltage side of the short-circuit transformer,
178 TESTING OF CIRCUIT BREAKERS
Ls is
L22 R1
us Cd TB
R22
C22 C1
Figure 10.9 Single-phase test circuit for a short-circuit test on a high-voltage circuit
breaker
DIRECT TEST CIRCUITS 179
×105
3
us
2
is
1
−1
−2
−3
TRV
−4
0 0.01 0.02 0.03
Time [s]
the make switch is closed and the short-circuit current flows through
the TB. When the mechanism of TB receives an opening command, the
breaker contacts move apart and the TB interrupts the current. After
a successful interruption of the short-circuit current, the TB is stressed
by the transient recovery voltage, coming from the oscillation of the
TRV-adjusting elements R1 , C1 , Cd , R22 , L22 , and C22 together with the
inductance formed by the current-limiting reactor LS , the synchronous
reactance of the generator, and the leakage reactance of the short-circuit
transformers. When the transient recovery has damped out, the TB faces
the power frequency–recovery voltage. In Figure 10.10, the current and
voltage traces of a single-phase current interruption are shown.
When the circuit breaker has to perform a make–break test, the breaker
must close in on a short circuit. Before a make–break test, the TB is in open
position and it closes after the MS has closed. Therefore, TB senses the
power frequency supply voltage, and when the contacts of TB close
the short circuit, current flows through the test circuit. In Figure 10.11,
the current and voltage traces of a single-phase make–break test are
shown.
The DC component of the current is determined by the instant of
closing of the TB. The supply circuit is mainly an inductive circuit and
this implies that if TB closes when the supply voltage is at its maximum,
the DC component is zero and the current is called symmetrical. When
TB closes at voltage zero, the current starts with maximum offset and is
called asymmetrical. Because of the ohmic losses in the test circuit, the DC
180 TESTING OF CIRCUIT BREAKERS
×105
4
TRV
3 us
is
2
1
DC component
0
−1
−2
−3
0 0.02 0.04 0.06 0.08 0.10
Time [s]
component damps out with a certain time constant. The time constant of
the test circuit can be adjusted by inserting an additional resistance.
System
voltage Short-circuit
current Recovery
voltage
Time
Arcing
voltage
Figure 10.12 Short-circuit current, arcing voltage, and recovery voltage for a circuit
breaker clearing a system fault
based on the fact that during the interrupting process, the circuit breaker
is stressed by high current and by high voltage at different time periods.
This gives the possibility of using two separate sources of energy: one
source supplying the short-circuit current during the arcing period and
another source supplying the transient recovery voltage and the power
frequency–recovery voltage. The overlap of the current and voltage source
takes place during the so-called interaction interval around the current
zero, where the current interruption takes place. Two different synthetic
test methods, current injection and voltage injection, are employed as
synthetic testing technique.
For the current-injection method, whereby a high-frequency current is
injected into the arc of the test breaker before the short-circuit has its
zero crossing, there are two principal possibilities, namely, parallel and
series current injection. In addition, for the voltage injection method,
whereby a high-frequency voltage is injected across the contacts of the
test breaker after the short-circuit current has been interrupted, a parallel
and a series scheme are possible. For the ultra-high voltage levels, when
the breaker ratings exceed the 765-kV level, a combination of the current-
injection and the voltage-injection circuit is used, the so-called three-circuit
arrangement. The parallel current-injection circuit and the series voltage-
injection test circuit are, at present, widely accepted for the testing of
high-voltage circuit breakers.
The parallel current-injection method was developed by AEG in
Germany. The synthetic test circuit used for the parallel current-injection
test (see Figure 10.13) is often referred to as the Weil-Dobke circuit.
Figure 10.14 illustrates the interrupting current and recovery voltage
of the test breaker in the parallel current-injection test circuit. The
182 TESTING OF CIRCUIT BREAKERS
Ls ig AB iv
Rs R1 L22 L1
us TB Cd Gap
R22
+
Cs C1 C22 C0 u0
−
ig ig + iv
iv
t0 t1 t2 t3
Figure 10.14 Current and recovery voltage in the parallel current-injection scheme
in parallel to the TB, and the current through the TB is the sum of the
short-circuit current supplied by the generator and the injected current.
At time instant t2 , the power frequency short-circuit current ig supplied
by the generator, reaches current zero. Because the driving voltage of the
generator is rather low, it is relatively easy for the AB to interrupt ig , and
the AB separates the high-current circuit from the high-voltage circuit.
When the TB interrupts the injected current at time instant t3 , it is stressed
by the transient recovery voltage resulting from the voltage, oscillating in
the voltage circuit across the test breaker. The spark gap is still conducting
during this time period.
The shape of the current waveform in the interaction interval around
current zero is important because the physical processes in the arc column
have a very short time constant, in the order of a few microseconds. It is,
therefore, necessary to keep the rate of change div /dt, of the interrupted
current just before current zero, the same as the di/dt in the direct
test circuit. In the time period around current zero, there is a very
strong arc–circuit interaction and this time period is very critical for
the circuit breaker; the current is approaching zero and at the same
time the arc voltage is changing. During the high current interval, that
is, before current zero when the circuit breaker is in its arcing phase,
it is necessary that the energy released in the interrupting chamber is
the same as the energy that would be released in a direct test, so that
at the beginning of the interval of interaction when the actual current
interruption takes place, the conditions in the breaker are equal to those
in a direct test. The arcing energy is determined by the current and
arc voltage, and care has to be given to ensure the correct magnitude
and shape of the current. The magnitude of the current is controlled
by the reactance of the current circuit and the driving voltage of the
generator and is influenced by the arc voltage of the TB and the arc
voltage of the AB. According to the IEC-60427 standard for synthetic
testing of high-voltage circuit breakers, the influence of the arc voltage
in a synthetic test circuit is limited to such a condition that the distorted
current peak and the duration of the last current loop shall not be less
than 90 percent of the values specified. This corresponds with a reduction
of approximately 12 percent of the energy input in the last current loop.
This leads to the requirement that for a more or less constant value
of the arc voltage, the driving voltage of the supplying current source
must be at least 12 times the sum of the arc voltages of the TB and
the AB.
Apart from the correct rate of change di/dt of the injected current
(which is determined by the charging voltage u0 of the main capacitor
184 TESTING OF CIRCUIT BREAKERS
Ls
Rs AB Cd R1 L22 L1
Gap
R22
us C0
+
C1 C22 C0
−
Cs TB Cc
ig
ug
t1
t2
Figure 10.16 The current and recovery voltage of the series voltage injection test circuit
is added to the recovery voltage of the current source. The resulting TRV
is supplied by the voltage circuit only. Because the current injected by the
voltage circuit is small, the amount of capacitive energy in the voltage
circuit is relatively low. Because of this, the voltage injection method
can use smaller capacitor values than the current-injection method, and
is therefore more economical to use. During the interaction interval,
however, the circuit parameters of the voltage-injection circuit differ
from the circuit parameters of the direct test circuit. This is not the
case for the current-injection circuit, and it is for this reason that the
186 TESTING OF CIRCUIT BREAKERS
Figure 10.17 Main capacitor bank C0 and main inductance L1 of KEMA’s synthetic test
facility (courtesy of KEMA)
current-injection method is used when the direct test circuit reaches its
testing limits.
Figure 10.18 The spark gap of KEMA’s synthetic test facility can be triggered with an
accuracy of microseconds (courtesy of KEMA)
LR R
L2 C2
C1
In this expression,
f = power frequency
u = phase voltage
i = rated short-circuit breaking current
x = the ratio between the short-line fault current and the nominal
rated terminal fault current.
MEASURING TRANSIENT CURRENTS AND VOLTAGES 189
× 104
0.5
0
Arc Arc
current voltage
−1
−2
SLF
peak
−3
−4
−100 −80 −60 −40 −20 0 20
Time [µs]
Figure 10.20 Transient recovery voltage during the interruption of a 90 percent short-line
fault of a 145 kV/31.5 kA/60 Hz SF6 puffer circuit breaker
Figure 10.21 Artificial line to be used for short-line fault testing in a parallel
current-injection circuit (courtesy of KEMA)
Apart from the transducer (e.g. the shunt for current measurement
and the voltage divider for measuring voltages), the measurement circuit
as a whole needs a careful consideration with respect to grounding and
shielding.
A C
B D
Figure 10.22 Cross-section of a coaxial shunt. When A and B are used as input terminals
for the current C and D serve as output terminals for the measuring signal and vice versa.
R is the calibrated resistance tube
192 TESTING OF CIRCUIT BREAKERS
Figure 10.23 Coaxial shunt for measuring currents up to 100 kArms with a bandwidth of
0–1 MHz (courtesy of KEMA)
0.35
Bandwidth = MHz (10.3)
rise time (µs)
90%
Amplitude
10%
Rise time
Time
Figure 10.24 Transient response of a shunt after applying a step form current pulse
transformer and the stray capacitance of the windings can determine the
cut-off frequency. Low-frequency currents cannot be measured correctly
because of saturating effect in the magnetic core. Because of core satu-
ration, it is not possible to record the DC component in short-circuit
currents. The use of large magnetic cores improves the frequency char-
acteristic of the current transformer considerably, and time constants of
twenty-five seconds can be created, but the increase in the price of such
devices is more than proportional.
One of the reasons that current transformers are used for protection
and measurement applications is that they are able to produce the high
power output needed by electromechanical measuring equipment.
Another way of performing potential-free current measurements is by
the use of Rogowski coils. A Rogowski coil (Figure 10.26) is a uniformly
wound coil of constant cross-sectional area on a nonmagnetic former
shaped into a closed loop around the current conductor and that senses
the magnetic field around a closed path. The simplest example is an air-
cored toroid. From most viewpoints, the Rogowski coil is an ideal current
transducer for measuring transient currents. It gives an isolated current
measurement, it does not saturate with high currents, it has an excellent
bandwidth (typically from 0.1 Hz to over 1 MHz) comparable with the
coaxial shunt, it does not load the circuit, and it is inexpensive to make.
Only the DC component of short-circuit currents cannot be accurately
measured. The principal of the Rogowski coil has been known since 1912.
The coil is placed around the conductors whose currents are to be
measured as shown in Figure 10.25.
194 TESTING OF CIRCUIT BREAKERS
n(t )
d
N
v(t) = − j (10.4)
dt
j=1
The total number of turns of the coil is N, and j is the instantaneous flux
for the jth turn. The coil output voltage is v(t). When l is the core length,
n is the number of windings per unit length, S is the core cross-section,
and B is the magnetic flux density, the total flux ψ is given by
ψ= dl B.n dS (10.5)
If the core has a constant cross-section and the wire is wound perpendic-
ularly on the middle line l with constant density, the coil output voltage
is then
d N
d N
v(t) = −µ0 nS ij (t) = −M ij (t) (10.6)
dt dt
j=1 j=1
Loop made by
wound wire
i (t )
Primary
conductor
Loop made by
internal return
wire
n(t )
Figure 10.27 Rogowski coil with the return wire through the centre of the winding
1 ZL (p)
sinh
U2 n Zg (p)
Ht (p) = n =n
(10.7)
U ZL (p)
sinh
Zg (p)
U2
ZL ZL
U Zg Zg
1 element
n elements
U2
R L R L R2 L2
U Cp Cp Cp
Cg Cg Cg
The transfer function is, therefore, dependent upon the impedances ZL (p)
and Zg (p) and can be discussed only individually, depending on the type
of divider.
Resistor dividers are used for AC and DC voltage measurements, and
an equivalent circuit for such a divider is shown in Figure 10.29.
The stray capacitances Cg to ground, represent the effective electrical
field lines to ground and Cp corresponds to the ever-present stray capaci-
tance of each resistor element. The inductive components L of the resistor
elements have to be taken into account only for low ohmic dividers. The
normalised transfer function of a resistor divider is
1 (R + pL)pCg
sinh
n 1 + (R + pL)pCp
Ht (p) = n
(10.8)
(R + pL)pCg
sinh
1 + (R + pL)pCp
R2
U2 = U R1 = (n − 1)R R2 = R
R1 + R2
The applicability of pure resistor dividers for the measurement of very high
AC voltages is limited because the current flowing through the resistor
elements heats up the resistors and changes the resistor values.
MEASUREMENT SETUP FOR TRANSIENT VOLTAGE 199
Figure 10.30 Mixed divider for measuring TRVs upto 500 kVpeak with a bandwidth of
0–1 MHz (courtesy of KEMA)
recorders and induce noise directly into the amplifiers and the associ-
ated electronic circuits. Quasi-static electric and magnetic fields penetrate
the braid mesh of the coaxial cables. Electric fields directly induce, via
capacitive coupling, voltages on the central conductor and cable shield
currents generate additional noise via the coupling impedance or transfer
impedance of a coaxial cable.
For safety reasons, instrument cabinets are usually grounded and so
are the coaxial shunt and the voltage divider. If we also realise that
the 230-volt AC main supply has a connection with the grounding of
the building and with the grounding of the distribution transformer
of the local utility, we face the problem of multiple grounds. Multiple
grounds introduce ground loops in which electromagnetic fields can induce
voltages that not only affect the original measurement signals, but also
may cause potential differences between different ground terminals inside
the laboratory.
A typical measurement setup for a combined current–voltage measure-
ment (for instance, the recording of the interruption of a short-line fault)
is depicted in Figure 10.31.
The coaxial shunt is grounded and so is the low-voltage part of the
voltage divider. As signal cables, single-screened 50- coaxial cables are
applied. The recording instrument, which can be a classical cathode ray
oscilloscope or a sensitive transient recorder, can be touched during test by
the technicians and is grounded for the same reason. To avoid a connection
with the grounding of the main supply, an isolating transformer is placed
between the recording instruments and the 230-volt AC supply. An
Supply
current
Voltage
divider
Test object
Coax Isolating
transformer
Measurement 230 V~
equipment 50 Hz
Coax
Shunt 1:1
Return
current
Cp
Measurement Cm 230 V ~ 50 Hz
equipment
1:1
Figure 10.32 Supply of 230 volts to recording equipment via an isolating transformer
Supply
current Voltage
divider
Test object
Coax Isolating
transformer
Measurement 230 V~
equipment 50 Hz
Coax
Shunt 1:1
Return
current
Recording
F instrument
Figure 10.34 An EMC cabinet diverts the common mode currents from the shielding of
the measurement cables
Slamecka E., Rieder W., and Lugton W. T., ‘‘Synthetic testing: the present state,’’ IEE
Proc. 125(12), 1376–1380 (1978).
van Houten, M. A., Electromagnetic compatibility in high-voltage engineering, Thesis,
Eindhoven University of Technology, (1990).
van der Linden, W. A., and van der Sluis L., ‘‘A new artificial line for testing
High-Voltage Circuit Breakers,’’ IEEE T-PAS 102(4), 797–803 (1983).
‘‘Synthetic testing of high-voltage alternating current circuit breakers,’’ IEC 427,
International Electrotechnical Commission (IEC), 1989.
Transients in Power Systems
Lou van der Sluis
Copyright 2001 John Wiley & Sons Ltd
ISBNs: 0-471-48639-6 (Hardback); 0-470-84618-6 (Electronic)
Index